View
1
Download
0
Category
Preview:
Citation preview
An Evaluation of Population Restoration and Monitoring Techniques for Freshwater
Mussels in the Upper Clinch River, Virginia, and Refinement of Culture Methods for
Laboratory-Propagated Juveniles
Caitlin S. Carey
Thesis submitted to the faculty of the Virginia Polytechnic Institute and
State University in partial fulfillment of the requirements for the degree of
Master of Science
in
Fisheries and Wildlife
Jess W. Jones, Chair
Eric M. Hallerman
Marcella J. Kelly
September 27th
, 2013
Blacksburg, Virginia
Keywords: Freshwater Mussels, Epioblasma capsaeformis, Population Restoration and
Monitoring, Mark-Recapture, Culturing, Temperature
Chapter 3 © 2013 by Taylor & Francis
All other material © 2013 by Caitlin S. Carey
An Evaluation of Population Restoration and Monitoring Techniques for Freshwater
Mussels in the Upper Clinch River, Virginia, and Refinement of Culture Methods for
Laboratory-Propagated Juveniles
by
Caitlin S. Carey
ABSTRACT
From 2006–2011, four population reintroduction techniques were applied to three sites
within a reach of the upper Clinch River in Virginia designated suitable for population
restoration of the federally endangered oyster mussel (Epioblasma capsaeformis). These
techniques were: 1) translocation of adults (Site 1), 2) release of laboratory-propagated sub-
adults (Site 1), 3) release of 8-week old laboratory-propagated juveniles (Site 2), and 4) release
of stream-side infested host fishes (Site 3). Demographic data were collected in 2011 and 2012
by systematic quadrat and capture-mark-recapture sampling to assess reintroduction success,
evaluate reintroduction techniques, and compare survey approaches for monitoring freshwater
mussels. Estimates of abundance and density of translocated adults ranged from 450–577
individuals and 0.09–0.11/m2 in 2011, and 371–645 individuals and 0.07–0.13/m
2 in 2012.
Estimates of abundance and density of laboratory-propagated sub-adults ranged from 1,678–
1,943 individuals and 0.33–0.38/m2 in 2011, and 1,389–1,700 individuals and 0.27–0.33/m
2 in
2012. Additionally, three recruits were collected at Site 1. No E. capsaeformis were collected at
Sites 2 and 3. Capture-mark-recapture sampling produced similar mean point estimates as
systematic quadrat sampling, but with typically more precision. My results indicated that the
release of larger individuals (>10 mm) is the most effective technique for restoring populations
of E. capsaeformis, and that systematic quadrat and capture-mark-recapture sampling have
iii
useful applications in population monitoring that are dependent on project objectives. Systematic
quadrat sampling is recommended when the objective is to simply estimate and detect trends in
population size for species of moderate to larger densities (>0.2/m2). Capture-mark-recapture
sampling should be used when objectives include assessing a reintroduced population of
endangered species or at low density, obtaining precise estimates of population demographic
parameters, or estimating population size for established species of low to moderate density
(0.1–0.2/m2).
The ability to grow endangered juveniles to larger sizes in captivity requires improving
grow-out culture methods of laboratory-propagated individuals. A laboratory experiment was
conducted to investigate the effects of temperature (20–28°C) on growth and survival of
laboratory-propagated juveniles of the Cumberlandian combshell (Epioblasma brevidens), E.
capsaeformis, and the wavyrayed lampmussel (Lampsilis fasciola) in captivity. Results indicated
that 26°C is the optimum temperature to maximize growth of laboratory-propagated juveniles in
small water-recirculating aquaculture systems. Growing endangered juveniles to larger sizes will
improve survival in captivity and after release into the wild. As a result, hatcheries can reduce
the time that juveniles spend in captivity and thus increase their overall production and enhance
the likelihood of success of mussel population recovery efforts by federal and state agencies, and
other partners.
iv
ACKNOWLEDGEMENTS
I would like to thank my advisor, Dr. Jess Jones, for the opportunity and resources to lead
this research and for his continued support, patience, and guidance throughout this project. Your
passion for freshwater mussels and curiosity for knowledge has inspired me. I thank my
committee members, Drs. Eric Hallerman and Marcella Kelly, for their support and for providing
me with valuable advice during the planning stages and on my thesis manuscript. I am very
grateful to Dr. Hallerman for his financial support; through helping fund my tuition and the re-
opening of the aquaculture facility, you have given me more time to focus on my project as well
as a place to conduct my culture research. Thank you Dr. Kelly for all of the population
dynamics modeling and statistics expertise you have provided me over the years; your teaching
gave rise to my enjoyment of and pursuit of more knowledge in statistics and modeling
biological data. I would like to thank Bob Butler of the USFWS for all the support, time, and
guidance he has given to this project. Your editorial review of my thesis and publication, helping
me conduct field surveys, financial support of my temperature experiment, and advice the past
three years have been invaluable.
Field work was cooperatively conducted by personnel from USFWS, U.S. Geological
Survey, Virginia Department of Game and Inland Fisheries, The Nature Conservancy, Tennessee
Wildlife Resources Agency, and Virginia Polytechnic Institute and State University. Financial
support for this project was provided by the USFWS Gloucester,Virginia and Asheville, North
Carolina, Field Offices.
I’d like to thank Braven Beaty and The Nature Conservancy, Abingdon, Virginia for their
involvement in my field project and for allowing me easy access to Cleveland Islands, and to
Mike Pinder, Amanda Duncan, Megan Bradley, and Joe Ferraro from VDGIF for providing field
v
and laboratory assistance, release data, laboratory-propagated juveniles, and for their all around
support for this project. I would also like to thank everyone who contributed to this project—
without whom, this project could not have been completed. Thank you Andrew Phipps, Kasey
Ewing, Tim Lane, Jen Rogers, Amanda Graumann, Morgan Brizendine, Lee Stephens, Brian
Parks, Dan Hua, Matt Johnson, Shane Hanlon, Brett Ostby, Steve Ahlstedt, Gale Heffinger , Man
Tang, and the many other personnel from USFWS, U.S. Geological Survey, VDGIF, The Nature
Conservancy, Tennessee Wildlife Resources Agency, and Virginia Polytechnic Institute and
State University—for assisting me in the laboratory and conducting fieldwork in the Clinch
River. I would like to thank my fellow graduate students in the Department of Fish and Wildlife
Conservation, particularly Bonnie Myers, Laci Love, Shannon White, Jen Rogers, for their
support, friendship, and always good times.
Most importantly, I would like to thank my Mom, Dad, family and friends for their love
and encouragement. And finally, to my Papa—thank you for our countless early morning trips
into the estuary to crab, fish, and bond—you introduced me to my passion for working in aquatic
ecosystems.
vi
ATTRIBUTION
Several colleagues aided in the editorial review of one of my chapters presented as part of this
thesis, and subsequently published in the North American Journal of Aquaculture. A brief
description of their contributions is included here.
Chapter 3: Determining optimum temperature for growth and survival of laboratory-propagated
juveniles of two federally endangered species, Cumberlandian combshell (Epioblasma
brevidens) and oyster mussel (Epioblasma capsaeformis), and one non-listed species, wavyrayed
lampmussel (Lampsilis fasciola)
Chapter 3 has been published in the North American Journal of Aquaculture
Jess Jones, PhD, is currently a Restoration Biologist with the U.S. Fish and Wildlife Service,
stationed in the Department of Fisheries and Wildlife at Virginia Tech. Dr. Jones is a co-author
on this paper, and edited this article for publication.
Eric Hallerman, PhD, is currently a professor in the Department of Fisheries and Wildlife at
Virginia Tech. Dr. Hallerman is a co-author on this paper, and edited this article for publication.
Robert Butler, M.S., is currently a Biologist with the U.S. Fish and Wildlife Service in the
Asheville Field Office, North Carolina. Butler is a co-author on this paper, and edited this article
for publication.
vii
TABLE OF CONTENTS
Chapter 1. Restoring the Endangered Oyster Mussel (Epioblasma capsaeformis) to the
Upper Clinch River, Virginia: An Evaluation of Population Reintroduction Techniques…… 1
Abstract ………………………………………………………………………………….. 2
Introduction ………………………………………………………………………………….. 4
Methods ………………………………………………………………………………….. 7
Study Area……………………………………………………………………... 7
Predicted Estimates of Population Parameters (All Sites)……………………... 9
Habitat Measurements…………………………………………………………. 11
Quadrat Sampling……………………………………………………………… 12
Estimation of Population Parameters…………………………………………... 17
Results ………………………………………………………………………………….. 21
Site 1: Translocated Adults and Release of Laboratory-Propagated Sub-Adults 21
Site 2: Release of 8-week Old Laboratory-Propagated Juveniles 25
Site 3: Release of Stream-Side Infested Host Fish 26
Discussion ………………………………………………………………………………….. 28
Literature Cited ………………………………………………………………………………….. 41
Appendix A: Age-Class Categories and Matrices……………………………………………. 75
Appendix B: Sample Size Requirements (Statistical Analyses)……………………………... 92
Appendix C: Species List…………………………………………………………………….. 96
viii
Chapter 2. Evaluation of Systematic Quadrat and Capture-Mark-Recapture Survey
Techniques: Monitoring a Reintroduced Population of Oyster Mussels (Epioblasma
capsaeformis) in the Upper Clinch River, Virginia……………………………………………... 97
Abstract ………………………………………………………………………………….. 98
Introduction ………………………………………………………………………………….. 100
Methods ………………………………………………………………………………….. 103
Study Area……………………………………………………………………... 103
Epioblamsa capsaeformis Translocations and Releases……………………….. 104
Habitat Measurements…………………………………………………………. 105
Quadrat Sampling……………………………………………………………… 105
Capture-Mark-Recapture Sampling……………………………………………. 107
Comparisons of Sampling Methods and Population Size Estimates…………... 118
Growth………………………………………………………………………….. 118
Results ………………………………………………………………………………….. 119
Quadrat Sampling………………………………………………………………. 119
Capture-Mark-Recapture……………………………………………………….. 120
Comparisons of Sampling Methods and Population Size Estimates………… 126
Growth………………………………………………………………………….. 127
Discussion ………………………………………………………………………………….. 128
Literature Cited ………………………………………………………………………………….. 138
Appendix A: Cormack-Jolly-Seber Diagram and Program MARK Input Formatting……….. 156
Appendix B: Species List…………………………………………………………………….. 158
ix
Chapter 3. Determining Optimum Temperature for Growth and Survival of Laboratory-
Propagated Juveniles of Two Federally Endangered Species, Cumberlandian Combshell
(Epioblasma brevidens) and Oyster Mussel (Epioblasma capsaeformis), and One Non-Listed
Species, Wavyrayed Lampmussel (Lampsilis fasciola)………………………………………….. 159
Abstract ………………………………………………………………………………….. 160
Introduction ………………………………………………………………………………….. 162
Methods ………………………………………………………………………………….. 163
Gravid Mussel Collection……………………………………………………… 163
Host Fish Collection and Care…………………………………………………. 164
Infestation with Mussel Glochidia and Juvenile Mussel Collection…………… 164
Test Conditions………………………………………………………………… 166
Experimental Design and Statistical Analyses…………………………………. 167
Results ………………………………………………………………………………….. 168
Epioblasma brevidens………………………………………………………….. 168
Epioblasma capsaeformis……………………………………………………… 169
Lampsilis fasciola……………………………………………………………… 170
Algal Concentrations and Water Quality………………………………………. 171
Discussion ………………………………………………………………………………….. 172
Literature Cited ………………………………………………………………………………….. 180
Appendix A: Detailed Statistical Results……………………………………………………... 192
Appendix B: Species Comparisons within Temperature Treatments………………………… 198
x
List of Tables
Chapter 1. Restoring the Endangered Oyster Mussel (Epioblasma capsaeformis) to the Upper
Clinch River, Virginia: An Evaluation of Population Reintroduction Techniques:
Table 1.
Species and numbers of native host fishes infested with E. capsaeformis
glochidia and released in the upper Clinch River, Virginia at Artrip (Site 3)
each year from 2007–2010.
49
Table 2. Numbers released, predicted abundance ( ) and survival (proportion of
released individuals that survived) of translocated adults and laboratory-
propagated sub-adults in the left-descending channel of Cleveland Islands,
Virginia (Site 1) by release year in 2011 and 2012.
50
Table 3. Survey sample size requirements to estimate predicted abundance and density
levels with a desired precision of 15% of the estimate (CV = SE/mean) in the
left-descending channel (Site 1) and right-descending channel (Site 2) of
Cleveland Islands, and at Artrip (Site 3) in the upper Clinch River, Virginia.
Abundance and density values represent number of surviving individuals
predicted from the Leslie matrix (i.e., reproductive values were not included
in projections).
51
Table 4. Sample size, proportion of area covered by quadrats, person-hours of
sampling effort, number of E. capsaeformis collected, and precision
(CV=SE/mean) of systematic sampling collections conducted in 2011 and
2012, sorted by reintroduction method, in the left-descending channel (Site 1)
and right-descending channel (Site 2) of Cleveland Islands, and at Artrip (Site
3) in the upper Clinch River, Virginia.
52
xi
Table 5. Estimated mean and standard errors (SE) of abundance and density with lower
and upper 95% confidence intervals for translocated adults, released
laboratory-propagated sub-adults, and newly recruited E. capsaeformis in the
left-descending channel of Cleveland Islands, Virginia (Site 1) in 2011 (n=388
quadrats) and 2012 (n=347 quadrats) using systematic quadrat sampling.
53
Appendix B: Sample Size Requirements (Statistical Analyses)
Table B. 1. Estimated number of samples (0.25-m2 quadrats) required to reach a desired
sampling precision assuming a predicted density of the target species.
92
Table B. 2. Sample size requirements (per group) to detect various effect sizes (d =
)
between two years or sites for assorted combinations of power (1-β) and
significance level (ɑ). Effect sizes 0.2, 0.5, and 0.8 are characterized as small,
medium, and large as defined in Cunningham et al. (2007). With a 0.16
sampling variance (σ = 0.4) for E. capsaeformis, detecting effect sizes of
0.0625 and 0.8 are proportionate to 0.025/m2 and 0.32/m
2 differences in
density between two groups.
93
Appendix C: Species List
Table C. 1. Species collected in the upper Clinch River, Virginia at each sampling site in
2011 and 2012.
96
Chapter 2. Evaluation of Systematic Quadrat and Capture-Mark-Recapture Survey Techniques:
Monitoring a Reintroduced Population of Oyster Mussels (Epioblasma capsaeformis) in the
Upper Clinch River, Virginia:
Table 1. Top models, model used for median ĉ GOF test, descriptions, and model 145
xii
summary statistics for E. capsaeformis at Cleveland Islands, Virginia in 2011
and 2012 using closed-capture models in Program MARK. Summary statistics
in bold indicate that the model was used (top model or in model averaging) to
describe the data set in that year.
Table 2. Population size and density estimates for E. capsaeformis, A. pectorosa, and
M. conradicus at Cleveland Islands, Virginia from closed-capture modeling in
Program MARK.
146
Table 3. Top models, model used for median ĉ GOF test, descriptions, and model
summary statistics for A. pectorosa at Cleveland Islands, Virginia in 2011 and
2012 from closed-capture modeling in Program MARK. Summary statistics in
bold indicate that the model was used (top model or in model averaging) to
describe the data set in that year.
147
Table 4. Top models, model used for median ĉ GOF test, descriptions, and model
summary statistics for M. conradicus at Cleveland Islands, Virginia in 2011
and 2012 from closed-capture modeling in Program MARK. Summary
statistics in bold indicate that the model was used (top model or in model
averaging) to describe the data set in that year.
148
Table 5. Contrasts of population size estimates between systematic quadrat and CMR
sampling methods, and between 2011 and 2012, for E. capsaeformis, A.
pectorosa and M. conradicus at Cleveland Islands, Virginia. Effect sizes are
defined as the mean difference in population size.
149
Table 6. Top models, descriptions, and model summary statistics for E. capsaeformis
at Cleveland Islands, Virginia in 2011 and 2012 from open-capture modeling
150
xiii
(Cormack-Jolly-Seber) in Program MARK. Summary statistics in bold
indicate that the model was used to describe the data set.
Table 7. Summary of pros, cons, and recommendations regarding systematic quadrat
and capture-mark-recapture sampling approaches to monitoring freshwater
mussels.
151
Appendix B: Species List
Table B. 1. Species collected in the upper Clinch River at Cleveland Islands, Virginia
using systematic quadrat and capture-mark-recapture (CMR) sampling in
2011 and 2012.
158
Chapter 3. Determining Optimum Temperature for Growth and Survival of Laboratory-
Propagated Juveniles of Two Federally Endangered Species, Cumberlandian Combshell
(Epioblasma brevidens) and Oyster Mussel (Epioblasma capsaeformis), and One Non-Listed
Species, Wavyrayed Lampmussel (Lampsilis fasciola):
Table 1. Summary of gravid female mussel, host-fish collection, and host-fish
infestation methods at the Freshwater Mollusk Conservation Center (FMCC)
and Aquatic Wildlife Conservation Center (AWCC) in 2011 used to produce
juveniles in this study. All gravid females were collected from the lower
Clinch River, Tennessee.
186
Table 2. Experimental items and test conditions for culture temperature tests of E.
brevidens, E. capsaeformis, and L. fasciola juveniles at the FMCC, November
2011–April 2012.
187
Table 3. Final growth and survival (mean ± SE) of E. brevidens (EB), E. capsaeformis 188
xiv
(EC), and L. fasciola (LF) juveniles cultured in one of five temperature
treatments. Values followed by different subscripts are significant (p<0.05);
z–w indicate differences in temperatures within a species, and v–t differences
among species within a temperature treatment. The final sampling event
occurred at 138, 138, and 141 days for EB, EC, and LF juveniles,
respectively.
Appendix A: Detailed Statistical Results
Table A. 1. Summary of growth (mm) and survival (%) ANOVA of fixed effects for E.
brevidens, E. capsaeformis and L. fasciola.
192
Table A. 2. Summary of growth and survival slicing of the F-test for treatments by
sampling event (time=days since start of experiment) for E. brevidens, E.
capsaeformis and L. fasciola.
193
Table A. 3. Contrasts of differences between treatment means for final growth and
survival estimates of E. brevidens at the last sampling event (day 138) with
95% confidence intervals. Effect size for growth is in millimeters (mm).
Effect size for survival (%) data has been arc-sine transformed to achieve
normality.
194
Table A. 4. Contrasts of differences between treatment means of final growth and survival
estimates of E. capsaeformis at the last sampling event (day 138) with 95%
confidence intervals. Effect size for growth is in millimeters (mm). Effect size
for survival (%) data has been arc-sine transformed.
195
Table A. 5. Contrasts of differences between treatment means for final growth and
survival estimates of L. fasciola at the last sampling event (day 141) with 95%
196
xv
confidence intervals. Effect size for growth is in millimeters (mm). Effect
size for survival (%) data has been arc-sine transformed.
Table A. 6. Analysis of variance summary for algae concentrations (µm3/mL) within
buckets (EUs) among treatment temperatures.
197
Appendix B: Species Comparisons within Temperature Treatments
Table B. 1. Comparing E. brevidens, E. capsaeformis and L. fasciola growth (mm) within
temperature treatments. Summary of fixed effects for 20, 22, 24, 26, and 28°C.
199
Table B. 2. Comparing E. brevidens, E, capsaeformis and L. fasciola survival (%) within
temperature treatments. Summary of fixed effects for 20, 22, 24, 26, and 28°C.
200
xvi
List of Figures
Chapter 1. Restoring the Endangered Oyster Mussel (Epioblasma capsaeformis) to the Upper
Clinch River, Virginia: An Evaluation of Population Reintroduction Techniques:
Figure 1. Topographic map of 19.3-km designated population restoration reach for
Epioblasma capsaeformis in the upper Clinch River from Nash Ford to Carbo,
VA (yellow circles=towns) and locations of study sites (red stars).
54
Figure 2. Aerial view of translocation and release sites of E. capsaeformis (red stars)
and sampling areas (yellow polygons) in the left-descending channel (Site 1)
and right-descending channel (Site 2) of Cleveland Islands, VA.
55
Figure 3. Aerial view of release site of stream-side infested host fishes (red star) and
sampling area (yellow polygon) in the left-descending channel of Artrip,
Virginia (Site 3). Black polygon in river represents intermittent island.
56
Figure 4. Numbers of translocated adults and sex proportions per translocation year: A)
initially released for each year, and B) predicted to survive in 2011, and C)
2012 at Site 1 based on matrix transition probabilities presented in Jones et al.
(2012).
57
Figure 5. Numbers of laboratory-propagated sub-adults per each release year: A)
initially released for each year, B) predicted to survive in 2011, and C) 2012 at
Site 1 based on matrix transition probabilities presented in Jones et al. (2012).
58
Figure 6. Number of 8-week old laboratory-propagated juveniles per release year: A)
initially released for each year, B) predicted to survive to 2011, and C) 2012 at
Site 2, assuming 100% of the released juveniles successfully settled into
suitable substrate at the site based on matrix transition probabilities presented
59
xvii
in Jones et al. (2012).
Figure 7. Predicted numbers of excysted juveniles from each stream-side infestation of
host fishes: A) initially released for each year, and B) predicted to survive in
2011, and C) 2012 at Site 3 based on matrix transition probabilities presented
in Jones et al. (2012).
60
Figure 8. Estimated population sizes and densities (±95% CI) of translocated adults,
released laboratory-propagated sub-adults, and newly recruited E.
capsaeformis in the left-descending channel at Cleveland Islands (Site 1) in
2011 and 2012 using systematic quadrat sampling
61
Figure 9. Age-frequencies and sex-ratio distributions of translocated adult (sexed) and
laboratory-propagated sub-adult (unsexed) E. capsaeformis in: A) 2011 and
B) 2012 observed at Site 1 using systematic quadrat sampling. N = total
number of mussels collected in quadrat samples.
62
Figure 10. Observed length-class frequency distributions and sex-ratios of translocated
adult (sexed) and laboratory-propagated sub-adult (unsexed) E. capsaeformis
in: A) 2011 and B) 2012 at Site 1 using systematic quadrat sampling. N = total
number of mussels collected in quadrat samples.
63
Figure 11. Predicted length-class frequency distributions and sex-ratios for: A) 2011
without, B) with laboratory-propagated sub-adults, C) 2012 without, and D)
with laboratory-propagated sub-adults at Site 1.
64
Figure 12. Predicted survival estimates of: A) translocated adults (T) and laboratory-
propagated sub-adults (P) at Site 1, B) 8-week old laboratory-propagated
juveniles (J) at Site 2, and C) juveniles from stream-side infestations of host
65
xviii
fishes at Site 3 by release year over time.
Figure 13. Predicted abundance and density of translocated adults and released
laboratory-propagated sub-adults at Site 1 over time.
66
Figure 14. Predicted age-frequency and sex ratio distributions for translocated adult and
laboratory-propagated sub-adult E. capsaeformis surviving in: A) 2006, B)
2007, C) 2008, and D) 2009 at Site 1.
67
Figure 15. Predicted age-frequency and sex ratio distributions for translocated and
released E. capsaeformis surviving in: A) 2010 without laboratory-propagated
sub-adults (LPSA), B) 2010 with LPSA, C) 2011 without LPSA, D) 2011
with LPSA, E) 2012 without LPSA, and F) 2012 with LPSA at Site 1.
Predicted recruitment was not included in histograms.
68
Figure 16. Predicted abundance and density of 8-week old laboratory-propagated
juveniles at Site 2 over time, assuming 100% and 50% scenarios of the
released juveniles successfully settling into suitable substrate at the site.
69
Figure 17. Predicted age-frequency distributions of released 8-week old laboratory-
propagated juveniles from 2005–2012 at Site 2.
70
Figure 18. Predicted length-class frequency distributions of released 8-week old
laboratory-propagated juveniles at Site 2 in: A) 2011 and B) 2012 assuming a
1:1 sex ratio.
71
Figure 19. Predicted abundances and densities of juveniles released from stream-side
infested host fishes at Site 3 over time, under two scenarios (100% and 50%)
of the estimated average 22 viable juveniles excysted per infested host fish
successfully settled into suitable substrate at the site.
72
xix
Figure 20. Predicted age-frequency distributions of juveniles released from stream-side
infested host fishes from 2007–2012 at Site 3.
73
Figure 21. Predicted length-class frequency distributions of juveniles released from
stream-side infested host fishes to A) 2011 and B) 2012 at Site 3 assuming a
1:1 sex ratio.
74
Appendix A: Age-Class Categories and Matrices
Figure A. 1. Age-class categories, corresponding age, corresponding size ranges by sex,
and associated growing seasons for Epioblasma capsaeformis. Age 0–1 year
olds are referred to as age class 1 and represent newly transformed juveniles
during their first growing season. Predicted length-at-age based on estimated
von Bertalanffy growth curves presented in Jones et al. (2011).
75
Figure A. 2. Leslie matrix (L) of E. capsaeformis survival probabilities referenced in this
study analyses (from Jones et al. 2012).
76
Figure A. 3. Vector format for translocated adults, laboratory-propagated sub-adults, 8-
week old laboratory-propagated juveniles, and juveniles from stream-side
infestations released per sampling site.
77
Figure A. 4. Number and cohort structure at time of release of translocated adults (T) and
released laboratory-propagated sub-adults (P) released per year at Site 1 in
vector format.
78
Figure A. 5. Two scenarios (100% and 50%) representing the predicted number of 8-week
old laboratory-propagated juveniles (J) released per year that successfully
settled into suitable substrate at Site 2.
79
Figure A. 6. Two scenarios (100% and 50%) representing the predicted number of viable 80
xx
juveniles released from each stream-side infestation effort (I) that successfully
settled into suitable substrate after excystment from host fishes at Site 3.
Figure A. 7. Male and female age-class specific survival rates used in analyses (Leslie
matrix).
81
Figure A. 8. Example of how numbers, cohort structure, and lengths of E. capsaeformis
released per year were projected 1 to 6 years into the future depending on time
of translocation or release. Population vectors are provided to predict survival,
cohort structure, and length-frequency distribution of the population in 2011
and 2012.
82
Figure A. 9. Population projection vectors displaying the total number and cohort structure
of individuals from each release effort predicted to survive in 2011 and 2012
at Site 1 (T=translocated adults, P=laboratory-propagated sub-adults).
83
Figure A. 10. Predicted cohort structure and population size (N) of all translocated adults (T)
in 2011 and 2012 at Site 1.
85
Figure A. 11. Predicted cohort structure and population size (N) of laboratory-propagated
sub-adults (P) in 2011 and 2012 at Site 1.
86
Figure A. 12. Predicted cohort structure and population size (N) of all E. capsaeformis
(translocated adults and laboratory-propagated sub-adults) in 2011 and 2012
at Site 1.
87
Figure A. 13. Projected surviving number and cohort structure of 8-week old laboratory-
propagated juveniles (J) from each release effort in 2011 and 2012 at Site 2,
assuming 100% of the released juveniles successfully settlement into suitable
substrate at the site.
88
xxi
Figure A. 14. Predicted cohort structure and population size (N) of released 8-week old
laboratory-propagated juveniles (J) in 2011 and 2012 at Site 2, assuming
100% of the released juveniles successfully settlement into suitable substrate
at the site.
89
Figure A. 15. Projected surviving number and cohort structure of juveniles released from
stream-side infested host fishes (I) from each release effort in 2011 and 2012
at Site 3, assuming 100% of the estimated average 22 viable juveniles
excysted per infested host fish successfully settled into suitable substrate at
the site.
90
Figure A. 16. Predicted cohort structure and population size (N) of juveniles released from
stream-side infested host fishes (I) in 2011 and 2012 at Site 3, assuming 100%
of the estimated average 22 viable juveniles excysted per infested host fish
successfully settled into suitable substrate at the site.
91
Appendix B: Sample Size Requirements (Statistical Analyses)
Figure B. 1. Effect size to detect as a function of power and total sample size for A–D
levels of significance (0.05, 0.10, 0.15, and 0.20) using a two-tailed t-test for
mean differences between two independent groups.
94
Figure B. 2. Significance level (ɑ) as a function of power and total sample size for A–E
effect sizes (0.1, 0.2, 0.3, 0.4, 0.5) using a two-tailed t-test for mean
differences between two independent groups.
95
xxii
Chapter 2. Evaluation of Systematic Quadrat and Capture-Mark-Recapture Survey Techniques:
Monitoring a Reintroduced Population of Oyster Mussels (Epioblasma capsaeformis) in the
Upper Clinch River, Virginia:
Figure 1. Comparison of capture-mark-recapture and systematic quadrat population size
estimates and associated 95% confidence intervals for: A) translocated adult
and B) laboratory-propagated sub-adult (LPSA) E. capsaeformis, C) A.
pectorosa, and D) M. conradicus at Cleveland Islands, Virginia in 2011 and
2012.
152
Figure 2. Capture (and recapture, p and c) probabilities and associated 95% confidence
intervals for E. capsaeformis per sampling occasion for translocated adults in:
A) 2011 and B) 2012, and released laboratory-propagated sub-adults (LPSA)
in: C) 2011 and D) 2012 at Cleveland Islands, Virginia using a closed-capture
model in Program MARK.
153
Figure 3. Capture (and recapture, p and c) probabilities and associated 95% confidence
intervals for A. pectorosa in: A) 2011 and B) 2012, and M. conradicus in C)
2011 and D) 2012 at Cleveland Islands, Virginia using a closed-capture model
in Program MARK.
154
Figure 4. Epioblasma capsaeformis recapture probabilities (p) and associated 95%
confidence intervals per sampling occasion for translocated adults in: A) 2011
and B) 2012, and released laboratory-propagated sub-adults (LPSA) in: C)
2011 and D) 2012 at Cleveland Islands, Virginia using an open-capture model
(Cormack-Jolly-Seber) in Program MARK.
155
xxiii
Appendix A: Cormack-Jolly-Seber Diagram and Program Mark Input Formatting
Figure A. 1. Cormack-Jolly-Seber open-capture model diagram for E. capsaeformis. Black
numbers in boxes represent encounter occasions; numbers 1–5 represent
2006–2010 annual release events (no searches=p fixed at 0), 11 represents
2011 release event (no search=p fixed at 0) that occurred between 2011 and
2012 capture-mark-recapture sampling, and boxes 6–10 and 12–16 represent
capture-mark-recapture active searches with 5 encounter occasions each in
2011 and 2012 (active searches=p time dependent). Red Phii (φi) values
represent survival probability parameters between successive occasions. Blue
pi’s represent recapture probability parameters during encounter occasions.
Black numbers above arrows represent the time in weeks between occasions.
156
Figure A. 2. A sample of Program MARK input formatting for E. capsaeformis open
population modeling (Cormack-Jolly-Seber model). The first two columns
represent the ID (tag) of an individual and its associated encounter history.
The last two columns represent the group the individual was classified under
(translocated adult or a laboratory-propagated sub-adult).
157
Chapter 3. Determining Optimum Temperature for Growth and Survival of Laboratory-
Propagated Juveniles of Two Federally Endangered Species, Cumberlandian Combshell
(Epioblasma brevidens) and Oyster Mussel (Epioblasma capsaeformis), and One Non-Listed
Species, Wavyrayed Lampmussel (Lampsilis fasciola):
Figure 1. Top view of recirculating downweller bucket culture system and chambers. 189
Figure 2. Mean growth versus time for: (a) E. brevidens, (b) E. capsaeformis, and (c) L. 190
xxiv
fasciola juveniles cultured in one of five temperature treatments. Growth
measurements were taken at 2-week intervals for 20 weeks to provide a total
of 10 sampling events.
Figure 3. Mean survival versus time for: (a) E. brevidens, (b) E. capsaeformis, and (c)
L. fasciola juveniles cultured in one of five temperature treatments. Survival
was assessed at 2-week intervals for 20 weeks to provide a total of 10
sampling events.
191
Appendix B: Species Comparisons within Temperature Treatments
Figure B. 1. Comparisons of E. brevidens, E. capsaeformis and L. fasciola growth (mm) at
each of the 5 temperature treatments (20, 22, 24, 26, and 28°C) over 10
sampling events.
201
Figure B. 2. Comparisons of E. brevidens, E. capsaeformis and L. fasciola survival (%) at
each of the 5 temperature treatments (20, 22, 24, 26, and 28°C) over 10
sampling occasions.
202
1
CHAPTER 1
Restoring the Endangered Oyster Mussel (Epioblasma capsaeformis) to the Upper Clinch
River, Virginia: An Evaluation of Population Reintroduction Techniques
2
ABSTRACT
In 2002, the Virginia Department of Game and Inland Fisheries designated an
approximately 19.3-km reach of the upper Clinch River in Virginia as a reach suitable for
population restoration of the federally endangered oyster mussel (Epioblasma capsaeformis).
From 2006–2011, four population reintroduction techniques were applied to three sites within
this reach, including three at Cleveland Islands (Sites 1 and 2), and one at Artrip (Site 3). These
techniques were: 1) translocation of adults (Site 1, N=1,418), 2) release of laboratory-propagated
sub-adults (Site 1, N=2,851), 3) release of 8-week old laboratory-propagated juveniles (Site 2,
N=9,501), and 4) release of stream-side infested host fishes (Site 3, N=1,116 host fishes, with an
estimated 24,552 newly-metamorphosed juvenile excysted from them). The objective of this
study was to evaluate the success of these four reintroduction strategies via population
monitoring at each release site using systematic 0.25-m2 quadrat sampling. Estimated abundance
and density of translocated adults at Site 1 were 577 (SE=155) individuals and 0.11 individuals
/m2 (SE=0.03) in 2011, and 645 (SE=110) individuals and 0.13 individuals /m
2 (SE=0.02) in
2012. Estimated abundance and density of laboratory-propagated sub-adults at Site 1 were 1,678
(SE=42) individuals and 0.33 individuals /m2 (SE=0.01) in 2011, and 1,700 (SE=229)
individuals and 0.33 individuals /m2 (SE=0.05) in 2012. No E. capsaeformis were collected at
sites where 8-week old laboratory-propagated juveniles (Site 2) and stream-side infested host
fishes (Site 3) were released. These results indicate that the translocation of adults and release of
laboratory-propagated sub-adults are the most effective techniques for restoring populations of E.
capsaeformis. I recommend that management efforts focus on the release of larger individuals
for purposes of augmenting vulnerable or reintroducing extirpated mussel populations.
3
KEYWORDS: Freshwater Mussels, Oyster Mussel, Epioblasma capsaeformis, Endangered
Species, Reintroduction, Population Restoration, Translocation of Adults, Release of Cultured
Individuals, Stream-side Infestation of Host Fishes
4
INTRODUCTION
Many freshwater mussel populations have declined significantly in the last 50 to 100
years, with over 70% of North America’s estimated 300 mussel species listed as extinct,
endangered, threatened, or of special concern (Williams et al. 1993; Neves et al. 1997).
Considered freshwater ecosystem engineers, mussels play vital ecological roles in their ability to
filter large portions of the water column through their gills and modify habitat. They provide
physical habitat and serve as a food source to other animals, supply nutrients to the water column
through nutrient cycling, stabilize substrates, and remove silt and pollutants in aquatic
ecosystems (Spooner and Vaughn 2006; Vaughn et al. 2008; Williams et al. 2008). In recent
years, reintroductions of species into historical habitats where they had become extirpated and
augmentations of extant but generally declining populations have been conducted to recover
imperiled species and mitigate future losses (Haag 2012). In addition to verifying suitable
habitats, restoring populations to previously occupied habitats requires adaptive management—
including assessments to identify the most efficient techniques for reintroducing and monitoring
restored populations. In this study, I evaluated four reintroduction techniques to determine the
most successful approach to reestablish viable populations of the federally endangered oyster
mussel (Epioblasma capsaeformis) in the upper Clinch River, Virginia (VA).
Federal and state recovery plans for listed mussel species have identified translocation of
adults and release of laboratory-propagated juveniles as approaches for increasing viability of
existing populations and for reintroducing species to historically occupied sites, thus facilitating
recovery (U.S. Fish and Wildlife Service [USFWS] 2003, 2004; Virginia Department of Game
and Inland Fisheries [VDGIF] 2010). Epioblasma capsaeformis is endemic to, and was once
widely distributed throughout, the upland portions of the Cumberland and Tennessee River
5
drainages, collectively known as the Cumberlandian Region (USFWS 2004; Williams et al.
2008). Historically, its range extended across six states, but now it occurs only in a few
fragmented stretches in these river drainages in Kentucky (KY), Tennessee (TN) and VA. The
species is considered extirpated from Alabama, Georgia, and North Carolina (NC) (USFWS
2004; Jones et al. 2006a).
The Clinch River is part of the upper Tennessee River drainage, flowing southwest
through southwestern VA into northeastern TN. In 2002, VDGIF designated an approximately
19.3-km reach of the upper Clinch River in VA as suitable for population restoration of E.
capsaeformis (Eckert and Pinder 2010; VDGIF 2010). In collaboration with VDGIF’s Aquatic
Wildlife Conservation Center (AWCC) near Marion, VA, Virginia Tech’s Freshwater Mollusk
Conservation Center (FMCC) has worked to restore E. capsaeformis within this reach from
2006–2012. Cleveland Islands (Clinch River Kilometer [CRKM] 435.8) and Artrip (CRKM
441.9) were chosen as reintroduction sites in the upper Clinch River for the project. The native
population of E. capsaeformis in the upper river has severely declined over the past 50 years to
the point of being essentially undetectable using normal sampling methodologies, if not
extirpated. This decline was due to various anthropogenic impacts, including poorly treated
wastewater effluent from municipal treatment facilities located along the river and many other
factors (Eckert and Pinder 2010; VDGIF 2010). However, over the past 10 to 20 years water
quality has improved, thus allowing Cleveland Islands and Artrip’s mussel and fish fauna to
begin to recover and for the sites to become suitable for restoration of this species (Eckert and
Pinder 2010).
The most recent survey of Cleveland Islands was conducted in 2008 by VDGIF (Eckert
and Pinder 2010). They found twenty-three live mussel species, including seven federally
6
endangered species (E. capsaeformis, shiny pigtoe [Fusconaia cor], fine-rayed pigtoe
[Fusconaia cuneolus], slabside pearlymussel [Pleuronaia dolabelloides], fluted kidneyshell
[Ptychobranchus subtentum], rough rabbitsfoot [Quadrula cylindrica strigillata], and purple
bean [Villosa perpurpurea]). All E. capsaeformis they encountered were tagged—indicating they
were from recent translocations. Prior to the start of reintroductions in 2006, the last E.
capsaeformis observed within this designated population restoration reach were found between
CRKM 436.6 and 439.4 (approximately 0.8–3.6 RKM upstream of Cleveland Islands and 2.1–
5.3 RKM downstream of Artrip) in 1985 by Drs. David Stansbery and Thomas Watters of Ohio
State University (Jones 2004, collection records), and at CRKM 437.1 (1.3 RKM upstream)
during surveys conducted in 1972–1975 by Bates and Dennis (1978). Additionally, no E.
capsaeformis were collected during recent qualitative and quantitative surveys of Artrip
conducted in 2003, 2004, and 2010 (Krstolic et al. 2013; B. Ostby, Virginia Tech, personal
communication).
As of 2010, four reintroduction techniques were applied to this reach by the FMCC and
AWCC: 1) translocation of adults, 2) release of laboratory-propagated sub-adults, 3) release of 8-
week old laboratory-propagated juveniles, and 4) release of stream-side infested host fishes. In
order to determine the success of the four restoration strategies, population monitoring was
conducted at each release site in 2011 and 2012 to estimate abundance and density of E.
capsaeformis. The purpose of my study was to determine which technique was most effective at
restoring populations of E. capsaeformis in the upper Clinch River.
7
METHODS
Study Area
Reintroduction techniques were implemented at three sites within the 19.3-km VDGIF
designated augmentation reach of the upper Clinch River, VA (Figure 1). Study sites were
located within a 6.1-km reach from CRKM 435.8–441.9, in Russell County, near the town of
Cleveland. Three of the reintroduction techniques were implemented at Cleveland Islands
(CRKM 435.8, Sites 1 and 2). At Cleveland Islands, translocated adults and laboratory-
propagated sub-adults were released together in the left descending channel (LDC, Site 1), and 8-
week old laboratory-propagated juveniles were released in the right descending channel (RDC,
Site 2) at Cleveland Islands. Owned by The Nature Conservancy, and cooperatively managed by
VDGIF, Cleveland Islands are characterized by four channels and three islands (Figure 2). In
recent years, flow has declined in the furthest-upstream right descending channel; this channel is
not used or referenced in my study. The three remaining channels are referred to as the right,
middle, and left descending channels. Each channel contains excellent water quality and flow
conditions, stable gravel substrates, and darter fish hosts utilized by E. capsaeformis and other
mussel species.
The fourth technique was implemented further upstream near Artrip (CRKM 441.9, Site
3) where stream-side infested host fishes were released. Artrip also has excellent water quality,
stable substrates, and fish hosts to support reproduction and recruitment of E. capsaeformis. It is
comprised of two main channels separated by a large island (Figure 3). Host fish collection and
releases of artificially infested fish hosts were conducted in the LDC. The LDC is characterized
by a shallow pool and run with stable gravel and sand substrates, followed by riffle habitat
composed of gravel, cobble, and sand.
8
Translocation of Adults and Release of Sub-Adults at Cleveland Islands (Site 1)
Over a five-year period from 2006–2010, adult E. capsaeformis were collected from the
lower Clinch River, TN, and translocated throughout the LDC of Cleveland Islands. Additional
laboratory-propagated sub-adult E. capsaeformis were released into the LDC by the AWCC in
2010 and 2011 (Site 1; Figure 2). Translocated adult and released sub-adult E. capsaeformis
were uniquely tagged, measured for length (mm), and sexed for identification purposes.
Generally, individuals <30 mm in size were not sexed because their shells were not yet sexually
dimorphic (i.e., the marsupial shell expansion characterizing females was undeveloped).
Translocation of adults and releases of laboratory-propagated sub-adults were randomly
distributed throughout the LDC.
Releases of Young Juvenile Mussels at Cleveland Islands (Site 2)
Over a four-year period from 2005–2008, juveniles were released into the RDC of
Cleveland Islands by the FMCC and AWCC (Site 2, Figure 2). Laboratory-propagated juveniles
were approximately 8-weeks old (0.5–1.0 mm) and were released in a 25-m2 area located
approximately 185 m upstream of where the RDC reconnects with the main river channel.
Release of Infested Host Fishes at Artrip (Site 3)
From 2007–2010, stream-side infestations of native fishes with E. capsaeformis glochidia
were conducted at Artrip (Site 3; Figure 3). Each stream-side infestation involved the collection
of native host fishes from Artrip, holding them in water tanks with an air source on site, infesting
them for 50 minutes with E. capsaeformis glochidia, and then returning them to the LDC riffle.
9
The number of juveniles assumed to have successfully transformed after stream-side infestations
was derived using a combination of empirical data from published works (Yeager and Saylor
1995; Jones and Neves 1998, 1999, 2001; Rogers et al. 2001; Jones et al. 2002; Jones 2004;
Liberty et al. 2005; Jones et al. 2006a) and unpublished fish host and mussel fecundity studies
conducted at the FMCC. From analysis of these data, I assumed that an average of 22 viable
juveniles excysted per fish.
Predicted Estimates of Population Parameters (All Sites)
Survival rates from one age-class to the next were obtained from data collected by Jones
and Neves (2011) and presented in Jones et al. (2012). An age-class survival transition Leslie
matrix was used to predict the number of individuals alive in 2011 and 2012 from each release
technique at each site (Appendix A, Figure A. 2). Because the focal objective of this study was to
evaluate the success of each reintroduction technique in terms of successful settlement and
survival of released individuals, and due to uncertainty concerning fecundity of translocated,
laboratory-propagated, and stream-side propagated mussels upon release (i.e., uncertainty of
magnitude of short term impact on reproduction; Sarrazin and Legendre 2000), fecundity values
were not included in my projections of current population size.
Using length measurements at time of release, translocated individuals were aged by
applying von Bertalanffy growth curves of predicted length-at-age for females and males from
the lower Clinch River, TN (Jones and Neves 2011). Laboratory-propagated sub-adults and 8-
week old juveniles were of known age when released. Age 0–1 year olds are referred to as age
class 1 and represent newly transformed juveniles during their first growing season. Age-class
10
categories, corresponding age, corresponding size ranges by sex, and associated growing season
are presented in Appendix A.
Numbers and ages of translocated adults, laboratory-propagated sub-adults, 8-week old
laboratory-propagated juveniles, and juveniles from stream-side infestations released per year
were put in vector format to compute expected survival and abundance over time at the
respective release sites. Using Jones et al. (2012) matrix transition probabilities in combination
with age at release, abundance was projected for 1–6 years from each release to 2011 and 2012
for all released mussels (e.g., abundance of mussels released in 2007 were projected for 4 years
to 2011 and 5 years to 2012). Two scenarios (50% and 100%) of successful settlement of
released 8-week old laboratory-propagated juveniles and excysted juveniles from stream-side
infestations within study sites were used in the matrix transition probabilities to predict survival.
Among the three different age 0–1 survival (i.e., probability of a viable newly-metamorphosed
juvenile surviving to the next year) scenarios (30, 35, and 42%), I fitted a 30% survival rate to
my analyses because Jones et al. (2012) concluded this would correspond to stable population
growth. Survival rates were assumed to be the same for females and males up until age-class 11
(10–11 year olds), where the maximum age was set at 10 and 12 years old for females and males,
respectively (e.g., females typically died after reaching 10 years old; Jones et al. 2012).
Numbers, ages, and lengths of E. capsaeformis released per year were projected for 1–6 years
(dependent on time of translocation or release) to predict survival (proportion of originally
released individuals that survived), abundance, density, cohort structure, and length-frequency
distribution of the population in 2011 and 2012 (Appendix A).
11
Habitat Measurements
The upstream and downstream boundaries of each reintroduction site were determined
prior to systematic quadrat sampling by a qualitative snorkel survey. Observation of live E.
capsaeformis or shells, presence of other mussels, substrate composition, and water depth and
flow were taken into consideration for establishing the upstream and downstream boundaries.
River width was measured at 5-m intervals along the total length of each reintroduction site
using a 100-m measuring tape. Area within each 5-m interval was calculated and summed to
determine the sampling area (m2) for each reintroduction site. Study area was used to determine
required intervals between sampling quadrats, and to convert estimates of abundance to density.
Banks were marked every 20 m with orange marking spray to serve as a location guide during
sampling.
Site 1
The LDC of Cleveland Islands is approximately 125 m in length with an average wetted
width of 14.8 m. The extended boundaries of this reintroduction site are approximately 35 m
upstream and 100 m downstream, with average wetted widths of 16.0 and 28.2 m, upstream and
downstream, respectively. The estimated total sample area at Site 1 was 5,085 m2.
Site 2
The 180-m reach of the RDC was primarily characterized by riffle and shallow run
habitat typically <0.5 m deep with gravel, cobble, and sand substrates. Because of the potential
for 8-week old laboratory-propagated juveniles placed at the upper end of the RDC to drift, the
entire length of the reach downstream from the release point at Site 2 was considered the
12
reintroduction survey site after taking dispersal within the channel into account. However, a 25-
m portion of this reach consisted of a deep run (~1 m deep) with a sand and bedrock bottom.
This habitat type is unsuitable for the oyster mussel and represented an area of low likelihood of
successful settlement and development of the released juvenile mussels. Therefore, this section
was not surveyed so that time and effort were focused on areas with the highest probability of
occurrence. Therefore, the actual survey length was 155 m with an average wetted width of 17.9
m. The estimated total sample area at Site 2 was 2,935 m2.
Site 3
The sampled portion in the LDC of Artrip was approximately 90-m in length with an
average wetted width of 29.6 m. The boundaries were approximately 25 m upstream and 65 m
downstream of the head of the riffle in the LDC. The estimated total sample area was 2,655 m2.
Quadrat Sampling
For the purpose of determining which reintroduction technique was most effective for
restoring populations of E. capsaeformis in this study, I estimated abundance and density of
individuals greater than 1-year old. Quadrat surveys were performed at each of the study sites
using 0.25-m2 sampling frames constructed of welded steel rebars. A systematic sampling design
was used to collect population demographic data on E. capsaeformis. Systematic sampling is a
probability-based survey method for assessing rare or clustered populations, is simple to execute
in the field, and offers effective spatial coverage (Christman 2000; Smith et al. 2001; Strayer and
Smith 2003). In addition, with probability-based sampling, I could estimate the probability that
the species is present at a specified mean density even if the target species were not detected
13
(Green and Young 1993; Strayer and Smith 2003). Estimation of basic population demographic
characteristics by quadrat sampling included species presence, abundance and density,
population growth rate, sex ratios, age-class structure, survival, and evidence of recruitment.
Systematic quadrat surveys were conducted at Site 1 in 2011 and 2012 and at Sites 2 and
3 in 2012. The required number of sampling units (0.25-m2 quadrat samples) needed to estimate
population density (mean mussels/m2) for a given level of precision was calculated using the
formula of Strayer et al. (1997):
where: n = number of quadrats searched,
m = mean number of E. capsaeformis per quadrat, and
CV = coefficient of variation.
As a function of my quadrat sampling unit and the predicted density of E. capsaeformis within a
site, the coefficient of variation (CV=SE/mean) is equivalent to the desired level of relative
precision of the estimate (i.e., 15% precision=true value falls within 15% of estimate) (Downing
and Downing 1992; Strayer et al. 1997). Prior to systematic sampling at Site 1, predicted E.
capsaeformis densities were obtained from the 2008 VDGIF (2010) survey, my 2011 mark-
recapture density data (see Chapter 2), and predicted abundance estimates of translocated adults
and released laboratory- propagated sub-adults (Appendix A). Initial estimations of E.
capsaeformis densities at Sites 2 and 3 were obtained from predicted abundance of total released
individuals to 2011 and 2012. I examined various combinations of target densities (0.01–
1.00/m2) and precision levels (SE/mean=0.05–0.50) to estimate the sample size requirements
needed to estimate density with varying levels of precision (Appendix B). Several scenarios were
taken into consideration to determine a sample size that provided a reasonable precision level
14
(e.g., 10–20% of the estimated density) and was logistically feasible (e.g., excavating 200–300
quadrats per 1–2 days).
A series of power analyses was conducted in GPower 3 to determine the number of
sampling units required to detect various standardized effect (d) sizes (0.06–0.80) between two
groups (i.e., between years or sites) for assorted combinations of Type I (ɑ=0.05–0.20, false
positive) and Type II (β=0.05–0.20, false negative) error rates (Cohen 1988; Cunningham and
McCrum-Gardner 2007; Appendix B). Standardized effect size (d) was defined as the difference
between two group means divided by the standard deviation. Standard deviation values were
pooled from previous quadrat surveys. The pooled variance used for d was 0.16 (σ=0.4). A d of
0.0625 and 0.2 correspond to a 0.1/m2
and 0.32/m2 mean density difference between two groups,
respectively. Considering the tradeoff between sample size and power, sample sizes required to
detect standardized effect sizes below 0.2 were not justifiable in terms of fieldwork efficiency
and costs. These sampling size estimates were combined with those obtained from the Strayer et
al. (1997) formula to justify sample sizes required to detect statistical differences that provided
good overall power and were logistically feasible to conduct at each site.
Quadrats were sampled using a systematic design with a minimum of three random starts.
Quadrats were spaced at regular intervals from each random starting point. Intervals between
quadrats were based on survey area, required sample size, and the number of random starts, as
calculated by the formula in Strayer and Smith (2003):
√
where: d = distance between units,
L = length of study site,
15
W = width of study site,
n = total number of quadrats, and
k = number of random starts.
Two random numbers were generated to determine the starting point at the downstream
boundary of each site for each random start (on the left descending bank at Sites 1 and 2, and the
right descending bank at Site 3). Utilization of a minimum of three random starts (k=3) allowed
for estimation of sampling variance without having to make any assumptions about E.
capsaeformis spatial distribution within the reintroduction sites (i.e., one random start assumes a
random distribution of mussels) (Hedayat and Sinha 1991; Smith et al. 2001; Strayer and Smith
2003). Each set of quadrats sampled within a random start constituted one systematic sample.
Each quadrat was carefully hand excavated to hardpan (approximately 15 cm below
surface) or to underlying bedrock, whichever was contacted first. All mussels sampled were
sexed (if possible), identified to species, measured for length (mm), and categorized as being
observed at the substrate surface or completely buried. It was assumed that all individuals in age
class 2 (i.e., 1 going on 2 years old) and older had a 100% probability of detection within a
quadrat. Due to their small size (<10–15 mm) and in the absence of sieving substrates from
quadrats in my study, juveniles <1 year old (i.e., young of year) were difficult to detect during
sampling and therefore were not included in population size estimations. Any untagged E.
capsaeformis were tagged and recorded, and examined for presence of glue on the shell
(indication of a previous tag), growth annuli (estimation of age), and photographed if determined
to be a new recruit. All mussels and excavated substrate were returned to their original collection
location.
16
Sample Size
Site 1.—Sample units required to estimate E. capsaeformis density at Site 1 with a desired
precision level of 15% of the estimated density ranged from 196–411 quadrats using predicted
density estimates of translocated adults, released laboratory-propagated sub-adults and the
combined total in 2011 and 2012 (Table 3). A minimum target sample size of 360 was chosen to
detect low density levels (≤0.25/m2) with a desired precision level of 15%, and to detect a small
standardized effect (d=0.2) between years with 0.10 significance (ɑ=0.10) and 85% power (1-
β=0.85) (Appendix B). Sampling at Site 1 was conducted with four random starts with a total of
388 and 347 quadrats in 2011 and in 2012, respectively. The area sampled by quadrats ranged
from 87–97 m2 and covered approximately 1.8% of the total sample site area. Quadrats were
flipped 14 times to achieve regular distance intervals of 7 m between sampling units.
Site 2.—Sample units required to estimate density at Site 2 with a desired precision of 15% of the
estimated density ranged from 189–284 quadrats using predicted density estimates of 8-week old
laboratory-propagated juveniles to 2011 and 2012 given 50 and 100% successful settlement into
suitable substrate (Table 3). A minimum target sample size of 191 was chosen to detect low to
moderate density levels (≈0.75/m2) with a desired precision level of 15%, and to detect a small to
moderate standardized effect (d=0.3) between sites with 0.10 significance and 90% power
(Appendix B). Sampling at Site 2 was conducted with five random starts with a total of 210
quadrats in 2012. The area sampled by quadrats was 52.5 m2 and covered approximately 1.8% of
the total sample site area. Quadrats were flipped 17 times to achieve regular distance intervals of
8.5 m between sampling units.
17
Site 3.—Sample units required to estimate density at Site 3 with a desired precision of 15% of the
estimated density ranged from 103–150 quadrats using predicted density estimates of stream-side
infested juveniles to 2011 and 2012, given scenarios of 50 and 100% successful settlement into
suitable substrate within the survey area (Table 3). Even though predicted densities were
moderate to high (>1/m2), a minimum target sample size of 191 was chosen to detect moderate
density levels (≥0.70/m2) with a desired precision level of 15%, and to detect a small to moderate
standardized effect size (d=0.3) between sites with 0.10 significance and 90% power (Appendix
B). Sampling at Site 3 was conducted with three random starts with a total of 194 quadrat units
in 2012. The area sampled by quadrats was 48.5 m2 and covered approximately 1.8% of the total
sample site area. Quadrats were flipped 13 times to achieve regular intervals of 6.5 m between
sampling units.
Estimation of Population Parameters
Abundance and Density
Abundance ( ) was defined as the total number of ≥1 year old E. capsaeformis in the
study area at a particular point in time. This was estimated by multiplying the average count per
systematic sample by the total number of possible systematic samples (M) in the study area
(Seber 1973; Smith et al. 2001; Strayer and Smith 2003):
∑
where: = abundance estimate,
M = number of possible systematic samples,
= count per systematic sample, and
m = number of systematic samples.
18
If a site had three random starts (k=3), there were three systematic samples (m=3). Dependent on
the area (A) of the site, the area of the sampling unit (a=0.25 m2), and the total number of
quadrats sampled (n), the total number of possible systematic samples (M) was calculated
following the formula in Smith et al. (2001):
∑
Variance for abundance was estimated by the formula (Smith et al. 2001; Strayer and Smith
2003):
( )
∑
For normally distributed sample data, the 95% confidence intervals for abundance were
calculated as:
√ ( )
Population density was defined as the total number of E. capsaeformis (>1 year old) per
m2 ). This was estimated by dividing abundance ( ) by the survey area (A) (Strayer and Smith
2003):
Variance for population density was calculated by dividing abundance variance ( ( ) by the
squared area (Strayer and Smith 2003):
( ) ( )
19
For normally distributed sample data, the 95% confidence intervals for density were calculated
as:
√ ( )
Data was assessed for normality. Occasionally, the traditional approach to calculating
confidence intervals utilizing the assumption of a normal distribution has been found to be
inaccurate for mussel population size and density estimations. Based on mussel population
sampling simulations, mussel population sizes (or density) tend to have a positively (right)
skewed distribution (Pooler and Smith, unpublished data, cited by Smith et al. 2001; Strayer and
Smith 2003). If normality tests revealed a departure from normality, data were log-transformed
and 95% confidence intervals were calculated for abundance by using a logarithmic
transformation of the estimate and a delta-method approximation of variance (Seber 1982; Smith
et al. 2001; Strayer and Smith 2003):
(
√ ( )
)
The 95% log-based confidence intervals for population density were calculated as:
(
√ ( )
)
Abundances, population densities, and their associated variances were estimated separately for
translocated adults and laboratory-propagated sub-adults.
Abundance and population density estimates at each reintroduction site, and from 2011–
2012 at Site 1, were compared using mixed-model analysis with each random start within a year
representing on sample. Heterogeneity of the data was assessed using residual plots and Levene’s
20
test for equality of variances (Levene 1960). If heterogeneity of variances was revealed, a
Satterthwaite’s approximation was used to account for unequal variances (Satterthwaite 1946).
The P-values were used to interpret whether a statistically significant difference (ɑ=0.05) in
abundance and density existed among reintroduction sites, or from 2011–2012 at Site 1.
However, the appropriate question was not just whether abundances or densities were different
between groups or years, but rather what was the magnitude of any difference (Gerrodette 1987;
Strayer and Smith 2003). Although P-values alone may confirm that an effect exists, they do not
provide information on the magnitude of the effect, what constitutes an important effect size, or
the precision of the effect-size estimate (Stefano 2004; Nakagawa and Cuthill 2007). It is
important to specify an effect size that is ecologically important a priori to the study. The effect
size for my study was defined as the mean difference in population density (Stefano 2004;
Cunningham and McCrum-Gardner 2007).
An effect that was considered ecologically important was determined a priori at a
magnitude of 0.08 individuals/m2 (i.e., a 0.08 E. capsaeformis/m
2 difference in density
constitutes an ecologically important difference between years or sites). Given a standard
deviation of 0.4 (based on previous survey data), a 0.08 individuals/m2 magnitude of an effect
would correspond with Cohen’s small standardized effect size of d=0.2. This magnitude was
judged acceptable based on the sample size required and the feasibility of conducting a field
survey to detect an effect change of this magnitude. Unpaired t-tests were used to calculate effect
sizes (unstandardized effect size=mean difference) and associated 95% confidence intervals
between sites and years. Results were used to provide statistical and biological inference to
whether density estimates differed. Analyses were conducted using SAS software (SAS Institute,
Inc., Cary, North Carolina, version 9.2).
21
RESULTS
Site 1: Translocated Adults and Release of Laboratory-Propagated Sub-Adults
Summary of Translocations and Releases 2006–2011
Over a five-year period from 2006–2010, a total of 1,418 adult E. capsaeformis were
collected from the lower Clinch River, TN, and translocated along the LDC of Cleveland Islands
(Site 1, Figure 4). An additional 2,501 and 350 laboratory-propagated sub-adult E. capsaeformis
were released into the LDC by the AWCC in 2010 and 2011, respectively (Figure 5). Sex ratio of
translocated adults was approximately 1:1. At the time of translocation, female adults ranged
from 23–47 mm and averaged 36 mm in size. Similarly, translocated male adults ranged from
19–47 mm and averaged 33 mm in size. Laboratory-propagated sub-adults were approximately
1–2 years old (age class 2), ranged from 11–31 mm, and averaged 21 mm in size.
Population Parameter Estimates and Sampling Observations in 2011 and 2012
A total of 44 E. capsaeformis were sampled in 2011, comprised of 11 translocated adults,
32 laboratory-propagated sub-adults, and 1 natural recruit. Similarly, 41 individuals were
sampled in 2012 and consisted of 11 translocated adults, 29 laboratory-propagated sub-adults,
and 1 recruit. Observed precision (CV) in number of translocated adults and released laboratory-
propagated sub-adults encountered among the four systematic samples were 0.54 and 0.05 in
2011, and 0.27 and 0.34 in 2012, respectively. Observed precision in total E. capsaeformis
encountered among the four systematic samples was 0.12 in 2011 and 0.26 in 2012.
Approximately 30 person-hours of effort were required to complete sampling of 388 quadrats in
2011 and 347 quadrats in 2012 (Table 4). A total of 440 and 380 individuals representing 20 and
18 species were encountered in 2011 and 2012, respectively (Appendix C).
22
Estimated abundances and densities of translocated adult E. capsaeformis were 577
(SE=155) individuals and 0.11/m2 (SE=0.03) in 2011, and 645 (SE=110) individuals and 0.13/m
2
(SE=0.02) in 2012 (Table 5; Figure 8). Normality tests did not indicate a departure from
normality. Standard deviations of the translocated adult 2012 abundance and density estimates
(625.07 and 0.12) were 1.37 times as large as the 2011 abundance and density standard
deviations (457.02 and 0.09), indicating that the homogeneity of variance assumption was not
violated. A Levene’s test also confirmed no violation of the homogeneity assumption (p=0.62)
and therefore data were not transformed for analysis. The magnitude of abundance and density
differences (effect sizes) between 2011 and 2012 were 68 (SE=190) individuals and 0.02/m2
(SE=0.04). Effect size confidence limits (ɑ=0.05) calculated around abundance [-397, 533] and
density [-0.07, 0.11] contained zero, indicating no significant difference in abundance (p=0.60)
or density (p=0.60) between years.
Estimated abundances and densities of laboratory-propagated sub-adult E. capsaeformis
were 1,678 (SE=42) individuals and 0.33/m2 (SE=0.01) in 2011, and 1,700 (SE=229) individuals
and 0.33/m2 (SE=0.05) in 2012 (Table 5; Figure 8). Normality tests did not indicate a departure
from normality. Standard deviations of the laboratory-propagated sub-adult 2012 abundance and
density estimates (982.80 and 0.19) were 5.38 times as large as the 2011 abundance and density
standard deviations (182.89 and 0.04), indicating a possible violation of the homogeneity of
variance assumption. Levene’s test of equal variances also indicated a potential violation of the
homogeneity of variances assumption (p=0.08); therefore a Satterthwaite’s approximation was -
used. The magnitude of abundance and density differences between 2011 and 2012 were 22
(SE=233) individuals and <0.01/m2 (SE=0.05). Effect size confidence limits (ɑ=0.05) calculated
23
around abundance [-719, 763] and density [-0.16, 0.16] contained zero, indicating no significant
difference in abundance (p=0.93) or density (p=1.00) between years.
Estimated abundance and density of E. capsaeformis recruits were 52 (SE=26)
individuals and 0.01/m2 (SE=0.01) in 2011, and 59 (SE=29) individuals and 0.01/m
2 (SE=0.01)
in 2012 (Table 5; Figure 8). Because plotted data and normality tests indicated a departure from
normality, data were log transformed. Standard deviations of the E. capsaeformis recruit 2012
abundance and density estimates (121 and 0.02) were 1.14 times as large as the 2011 abundance
and density standard deviations (106 and 0.02), indicating no violation of the homogeneity of
variance assumption. A Levene’s test indicated also confirmed no violation of the homogeneity
assumption (p=0.79). The magnitude of differences between 2011 and 2012 were 7 (SE=39)
individuals and <0.01/m2 (SE=0.01). Effect size confidence limits (ɑ=0.05) calculated around
abundance [-88, 102] and density [-0.03, 0.03] contained zero, indicating no significant
differences in abundance (p=0.86) or density (p=1.00) between years.
Age and Length-Frequency Distributions
Estimated ages at Site 1 ranged from 2–12 years (mean=3–4 years old; median=2–3 years
old) in 2011, and from 2–6 years (mean=3–4 years old; median=3–4 years old) in 2012 (Figure
9). Observed lengths ranged from 21.9–41.0 mm (mean=31.5 mm) in 2011 and from 27.0–39.5
mm (mean=32.9 mm) in 2012 (Figure 10). Lengths of the two individual recruits were 29.1 mm
in 2011 and 27.3 mm in 2012. The recruit collected in 2011 was estimated to be 2–3 years old,
and the recruit collected in 2012 was estimated to be 3–4 years old. Lengths from growth annuli
for the 2011 recruit were not recorded. Lengths measured from growth annuli corresponding to
0–1, 1–2, 2–3, and 3–4 years old for the 2012 recruit were 9.7, 16.5, 22.1, and 27.3 mm.
24
Predicted length-frequency distributions of translocated adults in 2011 and 2012 were negatively
(left) skewed. Including laboratory-propagated sub-adults, it was predicted that over half the
population was represented by the 30 and 34 mm length classes in 2011 and 2012, respectively
(Figure 11).
Predicted Estimates of Population Parameters
Predicted survival (the proportion of released individuals that survived) of adult mussels
at Site 1 translocated from 2006–2010 ranged from 29–91% in 2011, and 17–83% in 2012.
Predicted survival of laboratory-propagated sub-adults released from 2010 to 2011 ranged from
95–100% in 2011 and 90–95% in 2012. Predicted survival of all translocated adults and released
laboratory-propagated sub-adults was a function of age and ranged from 29–100% in 2011, and
17–95% in 2012; decreasing over time after mussels were released at the site (Table 2; Figure
12).
The adults predicted to survive, based on survival rates from Jones et al. (2012), from
each translocation effort from 2006–2010 were approximately 58, 107, 148, 320, and 366
individuals in 2011, and 34, 81, 120, 276, and 332 individuals in 2012 (Figure 4). The
laboratory-propagated sub-adults predicted to survive from the 2010 release were 2,376
individuals in 2011 and 2,257 individuals in 2012, and 333 individuals from the 2011 release
were predicted to survive in 2012 (Figure 5). Predicted abundances and densities of translocated
adults were approximately 1,000 individuals and 0.20/m2 in 2011, and 843 individuals and
0.17/m2 individuals in 2012. Predicted abundances and densities of laboratory-propagated sub-
adults were approximately 2,726 individuals and 0.54/m2 in 2011 (including the 350 individuals
released in 2011), and 2,590 individuals and 0.51/m2 in 2012, respectively. Predicted total
25
abundances and densities of E. capsaeformis were approximately 3,726 individuals and 0.73/m2
in 2011, and 3,433 individuals and 0.68/m2 in 2012, respectively (Table 2; Figure 13; Appendix
A).
At the time of release, the average ages of all translocated adults (T) and laboratory-
propagated sub-adults (P) were 4–5 and 1–2 years old, respectively. Predicted age distribution of
translocated adults shifted slightly from 2006–2009, as younger translocated individuals were
introduced on an annual basis (Figure 14). Age distribution was predicted to shift dramatically in
2010 when laboratory-propagated sub-adults were released into the population. Predicted
average age of surviving translocated adults and laboratory-propagated sub-adults were 5–6 and
2–3 years old in 2011, and 6–7 and 3–4 years old in 2012, respectively. Approximate average
age of all surviving released E. capsaeformis in the population was 2–3 years old in 2011 and 3-4
years old in 2012 (Figure 15; Appendix A).
Site 2: Release of 8-week Old Laboratory-Propagated Juveniles
Summary of Releases 2005–2008
Over a four-year period from 2005–2008, a total of 9,501 juveniles (approximately 8-
weeks old and 0.5–1.0 mm) were released into the RDC of Cleveland Islands (Site 2; Figure 6).
Population Parameter Estimates and Sampling Observations in 2011 and 2012
No live or dead E. capsaeformis were collected in 2012 from the RDC. A total of 194
individuals representing 13 other species was encountered (Appendix C). Approximately 16
person-hours of effort were required to complete sampling 210 quadrats (Table 4).
26
Predicted Estimates of Population Parameters
Although no live or dead E. capsaeformis were collected at Site 2 in 2012, given my
methodological assumptions, predicted survival of the 2005 to 2008 released juveniles should
have ranged from 20.8–27.1% in 2011, and 16.6–25.7% in 2012 (Table 2; Figure 12). Because
all individuals were approximately two months of age at release, survival over time would be the
same for each yearly release effort. Juveniles predicted to survive, based on survival rates in
Jones et al. (2012), from each reintroduction effort would have been 632, 390, 938, and 328
individuals to 2011, and 506, 331, 891, and 312 individuals to 2012 (Figure 6). Predicted
abundance and density would have been 2,289 individuals and 0.78/m2 in 2011, and 2,041
individuals and 0.70/m2 in 2012 (Figure 16; Appendix A). Assuming a scenario that only 50% of
the 9,501 juveniles successfully settled into suitable substrate within the survey area after
release, number of individuals predicted to survive to 2011 and 2012 would be half of the
predicted values above. Predicted age classes of individuals would have ranged from 3–4 years
old (mean=4–5 years old) in 2011, and 4–8 years old (mean=5–6 years old) in 2012 (Figure 17).
Assuming a 1:1 sex ratio, predicted average length class would have been 38 mm (±2) in 2011
and 2012 (Figure 18).
Site 3: Release of Stream-Side Infested Host Fish
Summary of Stream-Side Releases 2007–2010
From 2007–2010, eight separate (two per year) stream-side infestations of native fishes
with E. capsaeformis glochidia were conducted at Artrip (Site 3). A total of 1,116 fish were
collected, infested and released into the head of the LDC riffle over this four-year period (Table
27
1). Assuming that an average of 22 viable juveniles excysted per fish within the survey area, I
estimated approximately 24,552 juveniles were released at the site (Figure 7).
Population Parameter Estimates and Sampling Observations in 2011 and 2012
No live or dead E. capsaeformis were collected in 2012 from Artrip. A total of 289
individuals representing 16 other species were encountered (Appendix C). Approximately 14
person-hours of effort were required to complete sampling of 194 quadrats (Table 4).
Predicted Estimates of Population Parameters
Predicted survival of excysted juveniles from host fishes ranged from 26–30% in 2011
and 24–29% in 2012 (Table 2; Figure 12). Because all individuals were of the same age at time
of release (by year), survival as a function of time was the same for each release effort.
Assuming successful transformation, excystment, and settlement into suitable substrate, juveniles
predicted to survive from each release (2007–2010) were 1,205, 1,739, 2,201, and 1,716
individuals to 2011, and 1,145, 1,652, 2,091, and 1,630 individuals to 2012 (Figure 7). Projected
abundance and density would have been 6,861 individuals and 2.58/m2 in 2011, and 6,518
individuals and 2.46/m2 in 2012 (Figure 19; Appendix A). Assuming a scenario of only 50%
success, the number of individuals predicted to survive would be half the previous values.
Under these survival scenarios, the shape of predicted age-frequency distributions would
shift from 2007–2010 as stream-side infested juveniles were introduced on an annual basis. The
distribution would stabilize in 2011, followed by a right shift in 2012. Predicted age classes of
individuals would have ranged from 1–5 years old (mean=2–3 years old) in 2011, and 2–6 years
28
old (mean=3–4 years old) in 2012 (Figure 20). Assuming a 1:1 sex ratio, predicted average
length class would have been 30 mm (±2) in 2011, and 34 mm (±2) in 2012 (Figure 21).
DISCUSSION
The federal recovery plan for E. capsaeformis requires six distinct viable populations in
order to meet delisting criteria to threatened status (USFWS 2004). Currently there are only two
extant populations, each restricted to the upper Tennessee River drainage in the Clinch and
Nolichucky Rivers in eastern TN and southwestern VA. Biologists define a viable population as
a naturally reproducing population that contains enough individuals to maintain genetic diversity
to adapt and respond to environmental changes (Sarrazin and Barbault 1996; USFWS 2004;
Jones et al. 2006b). To meet delisting criteria, recovery plans for listed mussels recommend the
translocation of adults, release of laboratory-propagated individuals, and the release of
artificially-infested host fishes be used as methods to augment existing populations and
reintroduce species to historically occupied sites (USFWS 2003, 2004). Post-release monitoring
of demographic vital rates of reintroduced populations is essential to assessing reintroduction
success, improving method efficiency, providing biologists with data required for effective
management, and evaluating whether down- or delisting criteria have been met (Sarrazin and
Barbault 1996; Sarrazin and Legendre 2000; USFWS 2004; Jones and Neves 2011).
My study has shown that the translocation of adults and release of laboratory-propagated
sub-adults are effective techniques for re-establishing populations of E. capsaeformis among the
reintroduction techniques implemented in this project. Epioblamsa capsaeformis were only
detected at Site 1 where translocated adults and laboratory-propagated sub-adults were released.
Recruitment was also documented at Site 1 during both monitoring years, indicating that natural
29
reproduction and recruitment are occurring. Because no individuals were encountered in any of
the quadrats sampled at Sites 2 and 3, it can be concluded based on the total area sampled by
quadrats that if E. capsaeformis were to exist at these sites, they would occur at a density that
was essentially undetectable (e.g., <0.01/m2).
Although mussels have been reintroduced throughout the United States over the last
century as a management strategy to restore populations (Haag 2012), published literature with
detailed plan objectives, reintroduction methods, post-release monitoring data, and defined
success criteria are limited (Cope and Waller 1995). Of the published studies documenting
reintroductions, implementation concurrent with comparison of techniques is rare, and most
studies involved only translocations. Additionally, there is a general lack of consistency in
monitoring methods used to evaluate success, resulting in highly variable reporting
methodologies and results among studies (Sheehan et al. 1989; Cope and Waller 1995). Cope
and Waller (1995) compiled data from the mussel relocation literature to compare relocation
methods, subsequent monitoring programs, and mortality estimates. Of the 37 relocation projects
they assessed—30% of which were reintroductions to recolonize extirpated populations—only
three were available in the peer-reviewed literature, many lacked detailed explanations of
methods of relocation, monitoring, and assessment approaches, and estimates of mortality varied
greatly among projects. Furthermore, 60% of the projects had ≤1 year or no subsequent
monitoring, thus limiting the amount of data available to accurately evaluate methodologies and
assess success of reintroductions. Likewise, Sarrazin and Legendre (2000) expressed the need for
more detailed data-oriented monitoring studies of reintroduced populations in order to provide
information for future conservation plans.
30
Generally, studies reporting reintroduction efforts conclude with a statement regarding
continued or proposed monitoring of the site; however, they are seldom accompanied by or
followed with post-reintroduction observations or follow-up monitoring reports. Reports
conveying release numbers, collection sources, destination, and species are informative, and can
make conservation efforts appear productive, but they do not disclose anything about the relative
success of reintroduction efforts. As a result of the shortage of both short- and long-term data on
approach-specific reintroduction success, it is difficult for managers to make informed decisions
on which methods to employ for recovery projects (Sarrazin and Legendre 2000). My study
attempted to fill these knowledge gaps by providing a detailed reintroduction and monitoring
methodology, establishing criteria for success, and evaluating effectiveness of each technique.
By performing and reporting post-restoration population monitoring, several projects
have provided insight into the relative success of method-specific restoration efforts. From
1976–1978, almost 3,000 mussels of 16 species were reintroduced to historically occupied sites
in the North Fork Holston River in southwestern VA (Ahlstedt 1979). During subsequent
monitoring, Ahlstedt (1979) documented large variability among results of mussel translocation
efforts, with several reintroduced populations persisting >5 years and others washing out with
flood events soon after translocation (Sheehan et al. 1989). Others have also reported variable to
low recovery rates of translocated individuals (Sheehan et al. 1989), outcomes which contrast
with the generally high post-reintroduction recovery and survival of translocated adults and
released sub-adults observed in this study. However, results of a few projects are in agreement
with the high survival rates of translocated individuals observed in my study. For example,
Layzer and Scott (2006) documented (at 1-year intervals for five years) relatively high survival
of 18 species, and successful recruitment of one species, translocated over four years to the lower
31
French Broad River, TN. Likewise, in 2008, the Kentucky Department of Fish and Wildlife
Resources’ (KDFWR) Center for Mollusk Conservation observed 100% post-reintroduction
survival of 300 individuals—97 of which were E. capsaeformis—three months after
translocation to the Big South Fork Cumberland River, KY (KDFWR 2008). Increased success
in reintroduction and monitoring may reflect differences in use of technical improvements (e.g.,
refined site selection by better understanding species-specific habitat requirements, timing of
release, reduced stress from improved translocation methodologies, refined monitoring
methodologies), as well as more accurate determination of recovery versus survival rates.
Research in controlled propagation of freshwater mussels began over a century ago (Haag
2012). However, only in the past two decades has the technology been refined and used more
routinely for restoring populations, particularly the growing out of cultured individuals to sub-
adult stage, thereby avoiding the high levels of natural mortality among newly transformed
juveniles (Jones et al. 2006b). Correspondingly, there is little information on the immediate
survival of juveniles after settlement or documented success or failure of reintroduction methods
utilizing controlled propagation (Haag 2012). Layzer and Scott (2006) released 801 host fishes
infested with glochidia of pheasantshell (Actinonaias pectorosa), Cumberland moccasinshell
(Medionidus conradicus), and P. subtentum over three years, in addition to adult mussel
translocations of each species in the French Broad River, TN. Similar to results of my study, no
juveniles resulting from stream-side infestations were found during subsequent monitoring
(Layzer and Scott 2006). However, others have been able to observe some evidence of juvenile
mussel survival from infested host fish releases. From 1994–1998, the Tennessee Wildlife
Resource Agency released over 5,000 fish infested with threeridge (Amblema plicata) and
washboard (Megalonaias nervosa) glochidia into Kentucky Lake, TN to augment populations in
32
that reservoir (Hubbs 2000). In an attempt to document that stream-side infestations produced
mussels, they held a portion of the infested fish in cages at a control site. They recovered five
sub-adults two years after release, verifying that some mussels successfully transformed and
settled in cages after the infestation (Hubbs 2000). Since 2001, Higgins eye (Lampsilis higginsii)
glochidia have been infested on host fishes and released at reintroduction sites in the Wisconsin,
Iowa, Wapsipinicon, and Cedar Rivers in Wisconsin and Iowa by the Genoa National Fish
Hatchery (GNFH) (Eckert and Aloisi 2010; N. Eckert, USFWS, GNFH, personal
communication). As of 2013, a few dozen L. higginsii have been detected at sites in the
Wisconsin, Iowa (15 individuals) and Wapsipinicon Rivers (38 individuals), where the species
had been extirpated for many decades (N. Eckert, USFWS, GNFH, personal communication).
Using stream-side infestations, differing survival and recovery rates of transformed individuals
among species across studies may be due to: 1) differences in species-specific infestation
methods (e.g., infestation duration, temperature, degree of infestation, timing of release), 2)
condition (e.g., stress, disease, reproductive condition) and compatibility of host fish, 3)
condition of gravid mussels and maturity of glochidia used for infestation, 4) juvenile dispersal
through down-stream drift, 5) immigration as attached glochidia on host fishes, 6) excystment
over unsuitable habitat, or 7) release site variability (Waller et al. 1985; Roger et al. 2001; Jones
et al. 2005; Layzer and Scott 2006).
Due to recent advancements in controlled propagation over the last decade allowing
facilities to produce quantities of larger juveniles (>10 mm), it has become more feasible to
conduct reintroductions using laboratory-propagated mussels (Jones et al. 2005; Barnhart 2006;
Haag 2012). Research results support the assertion that larger and older individuals have a
significantly increased chance of survival when released in the wild relative to newly-
33
metamorphosed juveniles (Sarrazin and Legendre 2000; Hua et al. 2011). Newly-metamorphosed
juvenile mussels may fall prey to a suite of predators, including hydroids, dragonfly larvae,
dipteran larvae, crayfishes, and especially flatworms (Zimmerman et al. 2003; Klocker and
Strayer 2004). Further, because adult and sub-adults have more general micro-habitat
requirements than young juveniles, fewer factors work against their chances of post-
reintroduction survival, suggesting that release of larger individuals may be a promising
reintroduction method (Cosgrove and Hastie 2001).
For example, from 2009–2010, Virginia Tech’s FMCC released a total of 193 tagged
laboratory-propagated sub-adult Cumberlandian combshell (Epioblasma brevidens) into the
Powell River, TN—50 with passive integrated transponder (PIT) tags. Monitoring of these
individuals revealed high annual survival (>98%) and growth (mean=7.7 mm in 12 months)
(Hua et al. 2011). Additionally, in 2008, KDFWRs’ Center for Mollusk Conservation
documented survival and growth of laboratory-propagated fatmucket (Lampsilis siliquidea) sub-
adults (1.5 years old at time of release) one year after reintroduction to Elkhorn Creek, KY
(KDFWR 2008). Hence, continued and improved reporting and long-term monitoring of
reintroduction efforts contribute to the long-term success of reintroduction efforts.
While unpublished literature documenting reintroduction by controlled propagation is
available, such reports and data are not easily accessible. Data from planned monitoring efforts
are even harder to find outside of personal communications with mussel conservation
practitioners, conference proceedings, and agency reports (Sarrazin and Barbault 1996). The lack
of published studies in the peer-reviewed literature on mussel reintroductions using laboratory-
propagated juveniles (<10 mm) and stream-side infestations of host fishes may be due to short-
term difficulty in detection (i.e., individuals are too small to detect for first few years) and the
34
absence of long-term monitoring programs to document success. Peer-reviewed literature on
reintroduction efforts may also be scarce simply because survival—whether in the form of
encountering surviving individuals or observing consequent natural recruitment—has not
occurred. Because release of laboratory-propagated sub-adults is a relatively new reintroduction
technique, even fewer studies are available to provide data on long-term success (Haag 2012).
Foremost in post-reintroduction project evaluations is the need to clearly identify what
criteria constitute a successful reintroduction (Sarrazin and Barbault 1996). Quantitative and
qualitative criteria provide a reference point for comparisons to other projects. For example, an
ecological criterion of reintroduction success is establishment of a long-term viable population
(Griffith et al. 1989; Fischer and Lindenmayer 2000). A short-term measure of reintroduction
success is documentation of natural recruitment (Cope and Waller 1995; Sarrazin and Barbault
1996). Other reintroduction studies have evaluated success using three objectives: 1) settlement
of released individuals at a release site, 2) survival of individuals after release, and 3) natural
reproduction (i.e., mussels demonstrating natural recruitment) (Teixeira et al. 2007; Matějů et al.
2012).
I considered reintroduction of E. capsaeformis at Site 1 through translocation of adults
and release of laboratory-propagated sub-adults a short-term success because both high post-
reintroduction survival and natural recruitment (i.e., viable young produced from released
individuals) were documented. Abundance and density estimates in 2011 and 2012 obtained
from systematic quadrat sampling were generally lower than estimates predicted using the Leslie
matrix; however, most of the confidence limits for 2011 and 2012 abundance estimates contained
the predicted mean values from the matrix. This outcome signifies that translocated and released
individuals successfully settled into Site 1 and are surviving at rates similar to those reported for
35
E. capsaeformis in the lower Clinch River, TN (Jones et al. 2012). Although 2011 and 2012
density estimates at Site 1 were larger than 2008 quadrat sampling estimates, they do not signify
an increase in population size through natural recruitment. The increase in abundance from 2008
to 2011 and 2012 reflects the outcome of additional reintroduction efforts from 2008–2010.
Nonetheless, based on this project’s reintroduction intensity and time frame, it is too early in the
monitoring program to determine if recruitment is occurring at self-sustaining levels.
Reintroduction efforts at Sites 2 and 3—through release of 8-week old laboratory-
propagated juveniles and stream-side infested host fishes—were not successful in the context of
this study. Given the predicted abundance and size of juveniles at Sites 2 and 3 in 2011 and
2012, they would have been easily detectable using the sampling methods employed in this
study. This signifies that released 8-week old laboratory-propagated juveniles and juveniles from
stream-side infestations may not have successfully settled into Sites 2 and 3 or are not surviving
at rates similar to those reported for E. capsaeformis in the lower Clinch River (Jones et al.
2012). If any individuals did survive from reintroduction techniques employed at Sites 2 and 3,
they occurred at such low densities so as to be nearly undetectable or they dispersed from the
study area through down-stream drift or immigration as glochidia attached to host fishes.
Populations existing at low densities not only affect detectability but can reduce fertilization
success, further reducing the probability of the reintroduced population naturally reproducing
and recruiting at self-sustaining levels (Downing et al. 1993). Factors such as predation, host fish
death before glochidial transformation, settling into unsuitable substrates within the study area,
or unfavorable environmental conditions during excystment (e.g., high flow events) may have
contributed to the apparent failure of reintroductions at Sites 2 and 3 in this study.
36
Although habitat characteristics of the three sites in this study appeared to be very
similar—and no spatial replication of reintroduction techniques was conducted—it could be
hypothesized that subtle differences in habitat among the release sites contributed to the apparent
failures of reintroduction techniques implemented a Sites 2 and 3 in this study. However, given
the diversity and density similarities of native mussel species among the three sites, it is unlikely
that differences in habitat alone, if at all, explain the apparent failures at Sites 2 and 3. The four
reintroduction techniques in this study could not be implemented concurrently at each of the
three sites (to completely remove the habitat factor and replicate) because of the inability to
separate natural recruitment from the multiple releases of 8-week old laboratory-propagated
juveniles or juveniles from stream-side infestations (i.e., no identifiable markings or tags). A
larger-scale (temporally and spatially) experiment with replication is needed to investigate the
effects of habitat characteristics on the success of reintroduction techniques. However, based on
the variability in success of previous releases of newly-metamorphosed juveniles and stream-side
infested host fishes that have been implemented over a diversity of habitats, research supporting
the release of larger individuals (Sarrazin and Legendre 2000; Hua et al. 2011), and the results of
this study, it is in all likelihood that the release of larger individuals has a more reliable (i.e.,
predictable) and accelerated payoff for expediting the recovery of populations.
To improve success of mussel reintroductions, it is important to consider biotic and
abiotic factors that can hinder reintroductions. Several factors can influence reintroduction
success, including species- and size-specific suitability of the destination site (macro- and
microhabitat characteristics), timing of release, host fish presence and density, handling- and
transportation-related stressors, genetic variation among released individuals, condition of
released individuals, and environmental stochasticity (Griffith et al. 1989; Sheehan et al. 1989;
37
Cope and Waller 1995; Sarrazin and Barbault 1996; Fischer and Lindenmayer 2000; Sarrazin
and Legendre 2000; Layzer and Scott 2006; Jones et al. 2012). Additionally, intensity of the
reintroduction effort may be a key to success; as the number of individuals released annually is
increased, natural and stress-related mortality may be offset (Griffith et al. 1989; Sarrazin and
Barbault 1996; Fischer and Lindenmayer 2000; Sarrazin and Legendre 2000; Matějů et al. 2012;
S. Ahlstedt, U. S. Geological Survey, retired, personal communication).
In this study, the potential effects of these limiting factors on reintroduction success were
controlled by: 1) assessing sites for suitable habitat and water quality prior to reintroduction
efforts, 2) conducting reintroductions in late summer and early fall when the reproductive state
of individuals was low, 3) implementing optimal transportation and release protocols to reduce
stress, 4) confirming the presence of host fishes at reintroduction sites, and 5) fostering genetic
diversity. To maintain genetic diversity, translocated adults and those used for host fish stream-
side infestations were collected from multiple source populations within the Clinch River.
Additionally, the production of laboratory-propagated juveniles and sub-adults and the stream-
side infestation of host fish were conducted following controlled propagation policies and
guidelines and permitted by state and federal natural resource agencies (USFWS and National
Oceanic and Atmospheric Administration 2000; Jones et al. 2006b)
The data collected in my study also provided insight into the demographic characteristics
(i.e., age-specific survival rates) of E. capsaeformis. Estimates of abundance and density of
adults at Site 1 were slightly lower than predicted. Assuming the age-based matrix transition
probabilities presented in Jones et al. (2012) represent reasonably accurate year-to-year survival
rates of wild E. capsaeformis, the lower survival observed in this study may reflect higher initial
mortality occurring at time of release as a result of stress-induced mortality from handling (Cope
38
and Waller 1995; Sarrazin and Legendre 2000; Teixeira et al. 2007) or loss of individuals from
the site during high flow events. Additional sources of variability in survival could be attributed
to differences in habitat characteristics (biotic and abiotic) between the upper and lower Clinch
River, or an artifact of detection difficulty at low density rather than representing true differences
in survival rates.
The matrix transition probabilities used to calculate predicted abundance and density at
Sites 1, 2, and 3 were approximations of survival rates using best available data. Jones et al.
(2012) recognized the uncertainty in their Leslie matrix input sources and recommended
additional studies to improve estimates of age-class survival rates for E. capsaeformis. By
following unique individuals through time, I was able to estimate survival based on fates of
individuals captured. Although original aging of uniquely-marked translocated adults was
estimated using predicted age-at-length curves, my study was able to estimate annual survival
rates by combining capture histories with known time since release. Similarly, tagged laboratory-
propagated sub-adults were of known age at release and provided concrete age-specific data for
estimating annual survival rates. Despite the slight differences in survival based on predicted
versus observed data, the results of my study were in general agreement with previous
predictions of Jones et al. (2012) on annual survival for older age-classes—E. capsaeformis
adults and sub-adults exhibit moderate to high annual survival after making it past the vulnerable
age 0–1 stage, respectively, with increasing mortality as individuals approach maximum age.
Assuming introduced E. capsaeformis sub-adults at Site 1 from the 2010 release reached
sexual maturity in 2012 (i.e., based on size, >35 mm), and that environmental conditions have
been favorable for reproduction, it is likely that recruitment from these individuals can be
assessed as early as 2014 (1–2 year-old recruits). In accordance with the recovery plan for E.
39
capsaeformis (USFWS 2004), I suggest Site 1 at Cleveland Islands, VA be monitored every two
years (biennially) beginning in 2014 in order to assess population viability and to obtain
information for understanding age-1 class survival and recruitment rates of E. capsaeformis. An
assessment of genetic diversity should be conducted to further validate reintroduction success
(Jones et al. 2012). Future monitoring at Sites 2 and 3 to confirm if the absence of reintroduced
individuals and subsequent failure of released 8-week old laboratory-propagated juveniles and
stream-side infested fish hosts efforts in this study were accurate assessments of these
reintroduction techniques is not feasible. Since the completion of my study (September 2012),
several releases of laboratory-propagated sub-adult E. capsaeformis have been implemented at
Sites 2 and 3 (October 2012–2013)—making the differentiation between 8-week old laboratory-
propagated juveniles, juveniles from stream-side infestations, laboratory-propagated sub-adults,
and natural recruitment difficult. To invalidate the argument for confounding habitat variability
with reintroduction technique-specific success or failure in this study, and to further evaluate the
efficiency of releasing newly-metamorphosed laboratory-propagated and stream-side infested
host fishes, a spatially replicated experiment should be conducted.
I recommend that management efforts focus on translocation of adults and release of
laboratory-propagated sub-adults for maximizing success of population restoration projects.
Results of my study helped support previous conclusions of moderate to high annual survival
rates of E. capsaeformis adults; however, it may be too early in the monitoring program at
Cleveland Islands to gather sufficient data to assess age 0–1 survival and recruitment rates.
Although catch-curve and shell thin-sectioning analyses are useful techniques for calculating
broad estimates of survival for demographic models, their assumptions are rarely met by natural
populations (Miranda and Bettoli 2007; Jones et al. 2012). Given the chance to follow marked
40
cohorts of known age through time at reintroduction sites presented an opportunity to increase
knowledge of species-specific demographic rates rather than having to rely on catch-curve and
shell thin-sectioning data alone. Future reintroductions should continue to tag reintroduced
individuals in order to improve our understanding of species-specific demographic
characteristics while further refining survival criteria.
41
LITERATURE CITED
Ahlstedt, S. 1979. Recent mollusk transplants into the North Fork Holston River in southwestern
Virginia. Bulletin of the American Malacological Union 1979:21–23.
Barnhart, M.C. 2006. Buckets of muckets: a compact system for rearing juvenile freshwater
mussels. Aquaculture 254:227–233.
Bates, J.M., and S.D. Dennis. 1978. The mussel fauna of the Clinch River, Tennessee and
Virginia. Sterkiana 69–70:3–23.
Center for Mollusk Conservation. 2009. Advances in the propagation of rare and endangered
mussel species. Annual Research Highlights 2008 for the Kentucky Department of Fish
and Wildlife Resources 2:46–48.
Christman, M.C. 2000. A review of quadrat-based sampling of rare, geographically clustered
populations. Journal of Agriculture, Biological, and Environmental Statistics 5:168–201.
Cohen, J. 1988. Statistical Power Analysis for the Behavioral Sciences, 2nd Edition. Lawrence
Erlbaum Associates, Hillsdale, New Jersey.
Cope, G.W., and D.L. Waller. 1995. Evaluation of freshwater mussel relocation as a
conservation and management strategy. Regulated Rivers: Research and Management
11:147–155.
Cosgrove, P.J., and L.C. Hastie. 2001. Conservation of threatened freshwater pearl mussel
populations: river management, mussel translocation and conflict resolution. Biological
Conservation 99:183–190.
Cunningham, J.B., and E. McCrum-Gardner. 2007. Power, effect and sample size using GPower:
practical issues for researchers and members of research ethics committees. Evidence
Based Midwifery 5:132–136.
42
Downing, J.A., and W.L. Downing. 1992. Spatial aggregation, precision, and power in surveys
of freshwater mussel populations. Canadian Journal of Fisheries and Aquatic Sciences
49:985–991.
Downing, J.A., Y. Rochon, M. Pérusse, and H. Harvey. 1993. Spatial aggregation, body size, and
reproductive success in the freshwater mussel Elliptio complanata. Journal of the North
American Benthological Society 12:148–156.
Eckert, N.L., and D. Aloisi. 2010. Lampsilis higginsii and Quadrula fragosa recovery project:
Genoa National Fish Hatchery. U.S. Fish and Wildlife Service, Genoa, Wisconsin. 9 pp.
Eckert, N.L., and M.J. Pinder. 2010. Freshwater mussel survey of Cleveland Island, Clinch
River, Virginia: augmentation monitoring site 2008. Virginia Department of Game and
Inland Fisheries, Richmond. 47 pp.
Fischer, J., and D.B. Lindenmayer. 2000. An assessment of the published results of animal
relocations. Biological Conservation 96:1–11.
Gerrodette, T. 1987. A power analysis for detecting trends. Ecology 68:1364–1372.
Green, R.H., and R.C. Young. 1993. Sampling to detect rare species. Ecological Applications
3:351–356.
Griffith, B., J.M. Scott, J.W. Carpenter, and C. Reed. 1989. Translocation as a species
conservation tool: status and strategy. Science 245:477–480.
Haag, W.R. 2012. North American Freshwater Mussels: Natural History, Ecology, and
Conservation. Cambridge University Press, New York. 505 pp.
Hedayat, A.S., and B.K. Sinha. 1991. Design and Inference in Finite Population Sampling.
Wiley, New York.
43
Hua, D., J. Rogers, J. Jones, and R. Neves. 2011. Propagation, culture, and monitoring of
endangered mussels for population restoration in the Clinch and Powell Rivers,
Tennessee, 2006–2010. Report to the Tennessee Wildlife Resources Agency, Nashville.
44 pp.
Hubbs, D. 2000. Augmentation of natural reproduction by freshwater mussels to sustain shell
harvest. Pages 47–51 in: R.A. Tankersley, , D.I. Warmolts, G.T. Watters, B.J. Armitage,
P.D. Johnson, and R.S. Butler, eds. Freshwater Mollusk Symposia Proceedings. Ohio
Biological Survey, Columbus.
Jones, J.W., and R.J. Neves. 1998. Life history and artificial culture of endangered mussels.
Report to the Tennessee Wildlife Resources Agency, Nashville. 32 pp.
Jones, J.W., and R.J. Neves. 1999. Life history and artificial culture of endangered mussels.
Report to the Tennessee Wildlife Resources Agency, Nashville. 57 pp.
Jones, J.W., and R.J. Neves. 2000. Life history and artificial culture of endangered mussels.
Report to the Tennessee Wildlife Resources Agency, Nashville. 67 pp.
Jones, J.W., and R.J. Neves. 2001. Life history and artificial culture of endangered mussels.
Report to the Tennessee Wildlife Resources Agency, Nashville. 91 pp.
Jones, J.W., R.A. Mair, and R.J. Neves. 2002. Life history and artificial culture of endangered
mussels. Report to the Tennessee Wildlife Resources Agency, Nashville. 80 pp.
Jones, J.W., R.A. Mair, and R.J. Neves. 2005. Factors affecting survival and growth of juvenile
freshwater mussels cultured in recirculating aquaculture systems. North American
Journal of Aquaculture 67:210–220.
Jones, J.W., R.J. Neves, S.A. Ahlstedt, and E.M. Hallerman. 2006a. A holistic approach to
taxonomic evaluation of two closely related endangered freshwater mussel species, the
44
oyster mussel Epioblasma capsaeformis and tan riffleshell Epioblasma florentina walkeri
(Bivalvia: Unionidae). Journal of Molluscan Studies 72:267–283.
Jones, J.W., E.M. Hallerman, and R.J. Neves. 2006b. Genetic management guidelines for captive
propagation of freshwater mussels (Unionidae). Journal of Shellfish Research 25:527–
535.
Jones, J.W., and R.J. Neves. 2011. Influence of life-history variation on demographic responses
of three freshwater mussel species (Bivalvia: Unionidae) in the Clinch River, USA.
Aquatic Conservation: Marine and Freshwater Ecosystems 21:57–73.
Jones, J.W., R.J. Neves, and E.M. Hallerman. 2012. Population performance criteria to evaluate
reintroduction and recovery of two endangered mussel species, Epioblasma brevidens
and Epioblasma capsaeformis (Bivalvia: Unionidae). Walkerana, Journal of the
Freshwater Mollusk Conservation Society 15:27–44.
Klocker, C.A., and D.L. Strayer. 2004. Interactions among an invasive crayfish (Orconectes
rusticus), a native crayfish (Orconectes limosus), and native bivalves (Sphaeriidae and
Unionidae). Northeastern Naturalist 11(2):167–178.
Krstolic, J.L., G.C. Johnson, B., J.K. Ostby and W.F. Henley. 2013. Clinch River: water
chemistry, sediment chemistry, aquatic habitat, geomorphology, and freshwater mussel
population status, 2009–2011, U.S. Geological Survey Data Report 2013. In review.
Layzer, J.B., and E.M. Scott, Jr. 2006. Restoration and colonization of freshwater mussels and
fish in a southeastern United States tailwater. River Research and Applications 22:475–
491.
45
Levene, H. 1960. Tests for equality of variances. Pages 278–292 in: I. Olkin, S.G. Ghurye, W.
Hoeffding , W.G. Madow, and H.B. Mann, eds. Contributions to Probability and
Statistics: Essays in Honor of Harold Hotelling. Stanford University Press, California.
Liberty, A.J., R.A. Mair, and R.J. Neves. 2005. Life history and artificial culture of endangered
mussels. Report to the Tennessee Wildlife Resources Agency, Nashville. 19 pp.
Matějů, J., Š. Říčanová, S. Poláková, M. Ambros, B. Kala, K. Matějů, and L. Kratochvíl. 2012.
Method of releasing and number of animals are determinants for the success of European
ground squirrel (Spermophilus citellus) reintroductions. European Journal of Wildlife
Research 58:473–482.
Miranda, L.E., and P.W. Bettoli. 2007. Mortality. Pages 229–277 in: C.S. Guy and M. L. Brown,
eds. Analysis and Interpretation of Freshwater Fisheries Data. American Fisheries
Society, Bethesda, Maryland.
Nakagawa, S., and I.C. Cuthill. 2007. Effect size, confidence interval and statistical
significance: a practical guide for biologists. Biological Reviews 82:591–605.
Neves, R.J., A.E. Bogan, J.D. Williams, S.A. Ahlstedt, and P.W. Hartfield. 1997. Status of
aquatic mollusks in the southeastern United States: a downward spiral of diversity. Pages
43–85 in: G.W. Benz and D. E. Collins, eds. Aquatic Fauna in Peril: The Southeastern
Perspective. Special publication 1, Southeast Aquatic Research Institute. Lenz Design
and Communications, Decatur, Georgia.
Rogers, S.O., B.T. Watson, and R.J. Neves. 2001. Life history and population biology of the
endangered tan riffleshell (Epioblasma florentina walkeri) (Bivalvia:Unionidae). Journal
of the North American Benthological Society 20:582–594.
46
Sarrazin, F., and R. Barbault. 1996. Reintroduction: challenges and lessons for basic ecology.
Trends in Ecology and Evolution 11:474–478.
Sarrazin, F., and S. Legendre. 2000. Demographic approach to releasing adults versus young in
reintroductions. Conservation Biology 14:488–500.
Seber, G.A.F. 1973. The Estimation of Animal Abundance and Related Parameters, 1st edition.
Charles Griffin and Company, Ltd, London.
Seber, G.A.F. 1982. The Estimation of Animal Abundance and Related Parameters, 2nd edition.
Macmillian, New York.
Sheehan, R.J., R.J. Neves, and H.E. Kitchel. 1989. Fate of freshwater mussels transplanted to
formerly polluted reaches of the Clinch and North Fork Holston Rivers, Virginia. Journal
of Freshwater Ecology 5:139–149.
Smith, D.R., R.F. Villella, and D.P. Lemarié. 2001. Survey protocol for assessment of
endangered freshwater mussels in the Alleghany River, Pennsylvania. Journal of the
North American Benthological Society 20:118–132.
Strayer, D.L., S. Claypool, and S. Sprague. 1997. Assessing unionid populations with quadrats
and timed searches. Pages 163–169 in: K.S. Cummings, A.C. Buchanan, C.A. Mayer,
and T.J. Naimo, eds. Conservation and Management of Freshwater Mussels II: Initiatives
for the Future. Proceedings of an Upper Mississippi River Conservation Committee
Symposium, St. Louis, Missouri.
Satterthwaite, F.W. 1946. An approximate distribution of estimates of variance components.
Biometrics Bulletin 2:110–114.
Spooner, D.E., and C.C. Vaughn. 2006. Context-dependent effects of freshwater mussels on
stream benthic communities. Freshwater Biology 51:1016–1024.
47
Strayer, D.L., and D.R. Smith. 2003. A Guide to Sampling Freshwater Mussel Populations.
American Fisheries Society, Monograph 8, Bethesda, Maryland.
Stefano, J.D. 2004. A confidence interval approach to data analysis. Forest Ecology and
Management 187:173–183.
Teixeira, C.P., C. Schetini de Azevedo, M. Mendl, C. Cipreste, and R.J. Young. 2007. Revisiting
translocation and reintroduction programmes: the importance of considering stress.
Animal Behaviour 73:1–13.
U.S. Fish and Wildlife Service and National Oceanic and Atmospheric Administration. 2000.
Policy regarding controlled propagation of species listed under the Endangered Species
Act. Federal Register 65:183(20 September 2000):56916–56922.
U.S. Fish and Wildlife Service. 2003. Endangered and threatened wildlife and plants; proposed
designation of critical habitat for five endangered mussels in the Tennessee and
Cumberland River basins. Federal Register 68:106(3 November 2003):33234–33282.
U.S. Fish and Wildlife Service. 2004. Recovery plan for Cumberland elktoe (Alasmidonta
atropurpurea), oyster mussel (Epioblasma capsaeformis), Cumberlandian combshell
(Epioblasma brevidens), purple bean (Villosa perpurpurea), and rough rabbitsfoot
(Quadrula cylindrica strigillata). Atlanta, Georgia. 174 pp.
Vaughn, C.C., S.J. Nichols, and D.E. Spooner. 2008. Community and foodweb ecology of
freshwater mussels. Journal of the North American Benthological Society 27:409–423.
Virginia Department of Game and Inland Fisheries. 2010. Virginia freshwater mussel restoration
strategy: upper Tennessee River basin. Bureau of Wildlife Resources, Wildlife Diversity
Division, Nongame and Endangered Wildlife Program, Richmond. 17 pp.
48
Waller, D.L., L.E. Holland, L.G Mitchell, and T.W. Kammer. 1985. Artificial infestation of
largemouth bass and walleye with glochidia of Lampsilis ventricosa (Pelecypoda:
Unionidae). Freshwater Invertebrate Biology 4:152–153.
Williams, J.D., M.L. Warren, Jr., K.S. Cummings, J.L. Harris, and R.J. Neves. 1993.
Conservation status of freshwater mussels of the United States and Canada. Fisheries
18:6–22.
Williams, J.D., A.E. Bogan, and J.T. Garner. 2008. Freshwater mussels of Alabama and the
Mobile Basin in Georgia, Mississippi, and Tennessee. The University of Alabama Press,
Tuscaloosa, Alabama. 908 pp.
Yeager, B.L., and C.F. Saylor. 1995. Fish hosts for four species of freshwater mussels
(Pelecypoda: Unionidae) in the upper Tennessee River drainage. American Midland
Naturalist 133:1–6.
Zimmerman, L.L., R.J. Neves, and D.G. Smith. 2003. Control of predaceous flatworms
Macrostomum sp. in culturing juvenile freshwater mussels. North American Journal of
Aquaculture 65:28–32.
49
Table 1. Species and numbers of native host fishes infested with E. capsaeformis glochidia and
released in the upper Clinch River, Virginia at Artrip (Site 3) each year from 2007–2010.
No. Infested Per Year
Species Common Name 2007 2008 2009 2010 Total
Etheostoma blennioides Greenside darter 18 35 70 18 141
E. camurum Bluebreast darter 0 10 79 30 119
E. denoncourti Golden darter 0 0 2 9 11
E. rufilineatum Redline darter 185 235 164 157 741
E. stigmaeum Speckled darter 0 0 0 1 1
E. zonale Banded darter 3 0 31 36 70
Percina evides Gilt darter 7 12 5 9 33
Total 213 292 351 260 1,116
50
Table 2. Numbers released, predicted abundance ( ) and survival (proportion of released
individuals that survived) of translocated adults and laboratory-propagated sub-adults in the left-
descending channel of Cleveland Islands, Virginia (Site 1) by release year in 2011 and 2012.
Predicted 2011 Predicted 2012
Year No.
Released
Survival
(%)
Survival
(%)
Translocated
adults 2006 201 58 29
34 17
2007 197 107 54
81 41
2008 218 148 68
120 55
2009 401 320 80
276 69
2010 401 366 91
332 83
Subtotal 1,418 1,000 71
843 59
Laboratory-
propagated
sub-adults
2010 2,501 2,376 95
2,257 90
2011 350 350 100
333 95
Subtotal 2,851 2,726 96
2,590 91
Total 4,269 3,726 87
3,433 80
51
Table 3. Survey sample size requirements to estimate predicted abundance and density levels
with a desired precision of 15% of the estimate (CV = SE/mean) in the left-descending channel
(Site 1) and right-descending channel (Site 2) of Cleveland Islands, and at Artrip (Site 3) in the
upper Clinch River, Virginia. Abundance and density values represent number of surviving
individuals predicted from the Leslie matrix (i.e., reproductive values were not included in
projections).
Source of preliminary density estimates
Abundance
(N-hat)
Density
(per m2)
0.25-m2
Sampling
Units Required
2008 VDGIF survey:
Site 1 Left descending channel 585 0.25 338
2011 Capture-mark-recapture survey:
Site 1 Left descending channel 1,569 0.31 303
Leslie matrix transition probabilities to 2011:
Site 1 Translocated adults 1,000 0.20 378
Laboratory-propagated sub-adults 2,726 0.54 228
Total E. capsaeformis 3,726 0.73 196
Site 2 8-week old laboratory-propagated
juveniles
2,289 0.78 189
Site 3 Juveniles from stream-side infested host
fishes
6,861 2.58 103
Leslie matrix transition probabilities to 2012:
Site 1 Translocated adults 843 0.17 411
Laboratory-propagated sub-adults 2,590 0.51 235
Total E. capsaeformis 3,433 0.68 204
Site 2 8-week old laboratory-propagated
juveniles
2,041 0.70 200
Site 3 Juveniles from stream-side infested host
fishes
6,518 2.45 105
52
Table 4. Sample size, proportion of area covered by quadrats, person-hours of sampling effort,
number of E. capsaeformis collected, and precision (CV=SE/mean) of systematic sampling
collections conducted in 2011 and 2012, sorted by reintroduction method, in the left-descending
channel (Site 1) and right-descending channel (Site 2) of Cleveland Islands, and at Artrip (Site 3)
in the upper Clinch River, Virginia.
Site Year Reintroduction
Method
Sample
Size
(0.25-m2)
Area
Covered
(%)
Person-
Hours
No.
Collected
Observed
Precision
(CV)
1 2011 Translocated adults
11 0.54
Laboratory-
propagated sub-adults 33 0.05
Recruits
1 1.00
Total 388 1.9 29 45 0.12
1 2012 Translocated adults
11 0.34
Laboratory-
propagated sub-adults 29 0.27
Recruits
1 1.00
Total 347 1.7 25 41 0.26
2 2012 8-week old
laboratory-propagated
juveniles
210 1.8 16 0 *
3 2012 Juveniles from
stream-side infested
host fishes
194 1.8 14 0 *
* = Precision was not calculated because no E. capsaeformis were collected at Sites 2 and 3.
53
Table 5. Estimated mean and standard errors (SE) of abundance and density with lower and
upper 95% confidence intervals for translocated adults, released laboratory-propagated sub-
adults, and newly recruited E. capsaeformis in the left-descending channel of Cleveland Islands,
Virginia (Site 1) in 2011 (n=388 quadrats) and 2012 (n=347 quadrats) using systematic quadrat
sampling.
2011 2012
Mean SE
95% C.I.
Mean SE
95% C.I.
Lower Upper Lower Upper
Population Size (N-hat)
Translocated Adults 577 155 83 1,070 645 110 295 995
Lab-Propagated Sub-Adults 1,678 42 1,543 1,812 1,700 229 970 2,430
Recruits 52 26 2 1,226 59 29 2 1,375
Total 2,307 134 1,880 2,733 2,403 313 1,409 3,398
Density (per m2)
Translocated Adults 0.11 0.03 0.02 0.21 0.13 0.02 0.06 0.20
Lab-Propagated Sub-Adults 0.33 0.01 0.30 0.36 0.33 0.05 0.19 0.48
Recruits 0.01 0.01 <0.01 0.24 0.01 0.01 <0.01 0.27
Total 0.45 0.03 0.37 0.54 0.47 0.06 0.28 0.67
54
Figure 1. Topographic map of 19.3-km designated population restoration reach for Epioblasma
capsaeformis in the upper Clinch River from Nash Ford to Carbo, Virginia (yellow
circles=towns) and locations of study sites (red stars).
55
Figure 2. Aerial view of translocation and release sites of E. capsaeformis (red stars) and
sampling areas (yellow polygons) in the left-descending channel (Site 1) and right-descending
channel (Site 2) of Cleveland Islands, Virginia.
56
Figure 3. Aerial view of release site of stream-side infested host fishes (red star) and sampling
area (yellow polygon) in the left-descending channel of Artrip, Virginia (Site 3). Black polygon
in river represents intermittent island.
57
Figure 4. Numbers of translocated adults and sex proportions per translocation year: A) initially
released for each year, and B) predicted to survive in 2011, and C) 2012 at Site 1 based on
matrix transition probabilities presented in Jones et al. (2012).
58
Figure 5. Numbers of laboratory-propagated sub-adults per each release year: A) initially
released for each year, B) predicted to survive in 2011, and C) 2012 at Site 1 based on matrix
transition probabilities presented in Jones et al. (2012).
59
Figure 6. Number of 8-week old laboratory-propagated juveniles per release year: A) initially
released for each year, B) predicted to survive to 2011, and C) 2012 at Site 2, assuming 100% of
the released juveniles successfully settled into suitable substrate at the site, based on matrix
transition probabilities presented in Jones et al. (2012).
60
Figure 7. Predicted numbers of excysted juveniles from each stream-side infestation of host
fishes: A) initially released for each year, and B) predicted to survive in 2011, and C) 2012 at
Site 3 based on matrix transition probabilities presented in Jones et al. (2012).
61
Figure 8. Estimated population sizes and densities (±95% CI) of translocated adults, released
laboratory-propagated sub-adults, and newly recruited E. capsaeformis in the left-descending
channel at Cleveland Islands (Site 1) in 2011 and 2012 using systematic quadrat sampling.
62
Figure 9. Age-frequencies and sex-ratio distributions of translocated adult (sexed) and
laboratory-propagated sub-adult (unsexed) E. capsaeformis in: A) 2011 and B) 2012 observed at
Site 1 using systematic quadrat sampling. N = total number of mussels collected in quadrat
samples.
63
Figure 10. Observed length-class frequency distributions and sex-ratios of translocated adult
(sexed) and laboratory-propagated sub-adult (unsexed) E. capsaeformis in: A) 2011 and B) 2012
at Site 1 using systematic quadrat sampling. N = total number of mussels collected in quadrat
samples.
64
Figure 11. Predicted length-class frequency distributions and sex-ratios for: A) 2011 without, B)
with laboratory-propagated sub-adults, C) 2012 without, and D) with laboratory-propagated sub-
adults at Site 1.
65
Figure 12. Predicted survival estimates of: A) translocated adults (T) and laboratory-propagated
sub-adults (P) at Site 1, B) 8-week old laboratory-propagated juveniles (J) at Site 2, and C)
juveniles from stream-side infestations of host fishes at Site 3 by release year over time.
66
Figure 13. Predicted abundance and density of translocated adults and released laboratory-
propagated sub-adults at Site 1 over time.
67
Figure 14. Predicted age-frequency and sex ratio distributions for translocated adult and
laboratory-propagated sub-adult E. capsaeformis surviving in: A) 2006, B) 2007, C) 2008, and
D) 2009 at Site 1.
68
Figure 15. Predicted age-frequency and sex ratio distributions for translocated and released E.
capsaeformis surviving in: A) 2010 without laboratory-propagated sub-adults (LPSA), B) 2010
with LPSA, C) 2011 without LPSA, D) 2011 with LPSA, E) 2012 without LPSA, and F) 2012
with LPSA at Site 1. Predicted recruitment was not included in histograms.
69
Figure 16. Predicted abundance and density of 8-week old laboratory-propagated juveniles at
Site 2 over time, assuming 100% and 50% scenarios of the released juveniles successfully
settling into suitable substrate at the site.
70
Figure 17. Predicted age-frequency distributions of released 8-week old laboratory-propagated
juveniles from 2005–2012 at Site 2.
71
Figure 18. Predicted length-class frequency distributions of released 8-week old laboratory-
propagated juveniles at Site 2 in: A) 2011 and B) 2012 assuming a 1:1 sex ratio.
72
Figure 19. Predicted abundances and densities of juveniles released from stream-side infested
host fishes at Site 3 over time, under two scenarios (100% and 50%) of the estimated average 22
viable juveniles excysted per infested host fish successfully settled into suitable substrate at the
site.
73
Figure 20. Predicted age-frequency distributions of juveniles released from stream-side infested
host fishes from 2007–2012 at Site 3.
74
Figure 21. Predicted length-class frequency distributions of juveniles released from stream-side
infested host fishes to A) 2011 and B) 2012 at Site 3 assuming a 1:1 sex ratio.
75
APPENDIX A: Age-Class Categories and Matrices
Age
Class
Age Female Size
Range (mm)
Male Size
Range (mm)
Growing
Season
1 0–1 0–19.3 0–20.5 1
2 1–2 19.3–26.5 20.5–27.2 2
3 2–3 26.5–32.1 27.2–31.5 3
4 3–4 32.1–36.3 31.5–34.4 4
5 4–5 36.3–39.5 34.3–36.3 5
6 5–6 39.5–41.9 36.3–37.5 6
7 6–7 41.9–43.8 37.5–38.3 7
8 7–8 43.8–45.2 38.3–38.9 8
9 8–9 45.2–46.3 38.9–39.2 9
10 9–10 46.3–47.1 39.2–39.5 10
11 10–11 47.1–47.8 39.5–39.6 11
12 11–12 47.8–48.3 39.6–39.7 12
>12 >12 >48.3 >39.7 >12
Figure A. 1. Age-class categories, corresponding age, corresponding size ranges by sex, and
associated growing seasons for Epioblasma capsaeformis. Age 0–1 year olds are referred to as
age class 1 and represent newly transformed juveniles during their first growing season.
Predicted length-at-age based on estimated von Bertalanffy growth curves presented in Jones et
al. (2011).
76
[
]
Figure A. 2. Leslie matrix (L) of E. capsaeformis survival probabilities referenced in this study
analyses (from Jones et al. 2012).
77
[
]
[
]
[
]
[
]
[
]
where TY = Translocated adults of year Y,
PY = released laboratory-propagated sub-adults of year Y,
JY = released 8-week old laboratory-propagated juveniles of year Y,
IY = released viable juveniles from year Y stream-side infestation,
N = total number of E. capsaeformis,
t = time (year),
TYx(t) = number of translocated adults (year Y) of age x at time t,
PYx(t) = number of laboratory-propagated sub-adults (release year Y) of age x at
time t,
JYx(t) = number of released 8-week old laboratory-propagated juveniles (release
year Y) of age x at time t,
IYx(t) = number of viable juveniles (stream-side infestation year Y) of age x at time
t, and
Nx(t) = number of individuals of age x at time t
Figure A. 3. Vector format for translocated adults, laboratory-propagated sub-adults, 8-week old
laboratory-propagated juveniles, and juveniles from stream-side infestations released per
sampling site.
78
[ ]
[ ]
[ ]
[
]
[
]
[
]
[
]
Figure A. 4. Number and cohort structure at time of release of translocated adults (T) and
released laboratory-propagated sub-adults (P) released per year at Site 1 in vector format.
79
100%
[
]
[
]
[
]
[
]
50%
[
]
[ ]
[
]
[ ]
Figure A. 5. Two scenarios (100% and 50%) representing the predicted number of 8-week old
laboratory-propagated juveniles (J) released per year that successfully settled into suitable
substrate at Site 2.
80
100%
[
]
[
]
[
]
[
]
50%
[
]
[
]
[
]
[
]
Figure A. 6. Two scenarios (100% and 50%) representing the predicted number of viable
juveniles released from each stream-side infestation effort (I) that successfully settled into
suitable substrate after excystment from host fishes at Site 3.
81
[
]
[
]
Figure A. 7. Male and female age-class specific survival rates used in analyses (Leslie matrix).
82
The number projected to be at Site 1 in 2007 after the 2007 yearly reintroduction was
calculated by adding the 2007 translocation vector to the product of the 2006 translocation vector
by the Leslie matrix.
[ ]
[
]
[
]
+
[
]
The ensuing years were calculated by adding the corresponding translocation or release vector (if
applicable) to the product of the previous year’s vector by the Leslie matrix:
[ ]
[ ]
[ ]
[ ]
[ ]
Figure A. 8. Example of how numbers, cohort structure, and lengths of E. capsaeformis released
per year were projected 1 to 6 years into the future depending on time of translocation or release.
Population vectors are provided to predict survival, cohort structure, and length-frequency
distribution of the population in 2011 and 2012.
83
[ ]
[
]
[ ]
[
]
[
]
[
]
[
]
[
]
[
]
[
]
Figure A. 9. Population projection vectors displaying the total number and cohort structure of
individuals from each release effort predicted to survive in 2011 and 2012 at Site 1
(T=translocated adults, P=laboratory-propagated sub-adults).
84
[
]
[
]
[
]
[
]
[
]
Figure A. 9. (continued) Population projection vectors displaying the total number and cohort
structure of individuals from each release effort predicted to survive in 2011 and 2012 at Site 1
(T=translocated adults, P=laboratory-propagated sub-adults).
85
[
]
[
]
Figure A. 10. Predicted cohort structure and population size (N) of all translocated adults (T) in
2011 and 2012 at Site 1.
86
[
]
[
]
Figure A. 11. Predicted cohort structure and population size (N) of laboratory-propagated sub-
adults (P) in 2011 and 2012 at Site 1.
87
[
]
[
]
Figure A. 12. Predicted cohort structure and population size (N) of all E. capsaeformis
(translocated adults and laboratory-propagated sub-adults) in 2011 and 2012 at Site 1.
88
[
]
[
]
[
]
[
]
[
]
[
]
[
]
[
]
Figure A. 13. Projected surviving number and cohort structure of 8-week old laboratory-
propagated juveniles (J) from each release effort in 2011 and 2012 at Site 2, assuming 100% of
the released juveniles successfully settlement into suitable substrate at the site.
89
[
]
[
]
Figure A. 14. Predicted cohort structure and population size (N) of released 8-week old
laboratory-propagated juveniles (J) in 2011 and 2012 at Site 2, assuming 100% of the released
juveniles successfully settlement into suitable substrate at the site.
90
[
]
[
]
[
]
[
]
[
]
[
]
[
]
[
]
Figure A. 15. Projected surviving number and cohort structure of juveniles released from stream-
side infested host fishes (I) from each release effort in 2011 and 2012 at Site 3, assuming 100%
of the estimated average 22 viable juveniles excysted per infested host fish successfully settled
into suitable substrate at the site.
91
[
]
[
]
Figure A. 16. Predicted cohort structure and population size (N) of juveniles released from
stream-side infested host fishes (I) in 2011 and 2012 at Site 3, assuming 100% of the estimated
average 22 viable juveniles excysted per infested host fish successfully settled into suitable
substrate at the site.
92
APPENDIX B: Sample Size Requirements (Statistical Analyses)
Table B. 1. Estimated number of samples (0.25-m2 quadrats) required to reach a desired
sampling precision assuming a predicted density of the target species.
Precision = CV (SE/mean)
0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.50
Den
sity
(per
m2)
0.01 12879 3648 1744 1033 688 494 373 293 236 195
0.05 5668 1605 767 455 303 217 164 129 104 86
0.10 3980 1127 539 319 213 153 115 90 73 60
0.15 3237 917 438 260 173 124 94 74 59 49
0.20 2795 792 378 224 149 107 81 63 51 42
0.25 2494 706 338 200 133 96 72 57 46 38
0.30 2273 644 308 182 121 87 66 52 42 34
0.35 2101 595 284 169 112 81 61 48 39 32
0.40 1963 556 266 157 105 75 57 45 36 30
0.45 1848 523 250 148 99 71 54 42 34 28
0.50 1752 496 237 140 94 67 51 40 32 27
0.55 1668 473 226 134 89 64 48 38 31 25
0.60 1596 452 216 128 85 61 46 36 29 24
0.65 1532 434 207 123 82 59 44 35 28 23
0.70 1475 418 200 118 79 57 43 34 27 22
0.75 1424 403 193 114 76 55 41 32 26 22
0.80 1378 390 187 111 74 53 40 31 25 21
0.85 1336 378 181 107 71 51 39 30 25 20
0.90 1298 368 176 104 69 50 38 29 24 20
0.95 1263 358 171 101 67 48 37 29 23 19
1.00 1230 348 167 99 66 47 36 28 23 19
93
Table B. 2. Sample size requirements (per group) to detect various effect sizes (d =
)
between two years or sites for assorted combinations of power (1-β) and significance level (ɑ).
Effect sizes 0.2, 0.5, and 0.8 are characterized as small, medium, and large as defined in
Cunningham et al. (2007). With a 0.16 sampling variance (σ = 0.4) for E. capsaeformis,
detecting effect sizes of 0.0625 and 0.8 are proportionate to 0.025/m2 and 0.32/m
2 differences in
density between two groups.
Effect Size (d)
(1-β) ɑ 0.0625 0.1 0.2 0.3 0.4 0.5 0.8
0.80 0.05 4020 1571 394 176 100 64 26
0.85 0.05 4598 1797 450 201 114 73 30
0.90 0.05 5381 2103 527 235 133 86 34
0.95 0.05 6655 2600 651 290 164 105 42
0.80 0.10 3436 1238 310 139 78 51 21
0.85 0.10 3995 1439 361 161 91 59 24
0.90 0.10 4759 1714 429 191 109 70 28
0.95 0.10 6013 2166 542 242 136 88 35
0.80 0.15 2891 1041 276 117 66 43 17
0.85 0.15 3406 1227 307 137 78 50 20
0.90 0.15 4114 1482 371 166 94 60 24
0.95 0.15 5286 1904 477 212 120 77 31
0.80 0.20 2502 901 226 101 57 37 15
0.85 0.20 2984 1075 269 120 68 44 18
0.90 0.20 3650 1314 329 147 83 53 21
0.95 0.20 4758 1714 429 191 108 69 28
94
Figure B. 1. Effect size to detect as a function of power and total sample size for A–D levels of
significance (0.05, 0.10, 0.15, and 0.20) using a two-tailed t-test for mean differences between
two independent groups.
95
Figure B. 2. Significance level (ɑ) as a function of power and total sample size for A–E effect
sizes (0.1, 0.2, 0.3, 0.4, 0.5) using a two-tailed t-test for mean differences between two
independent groups.
96
APPENDIX C: Species List
Table C. 1. Species collected in the upper Clinch River, Virginia at each sampling site in 2011
and 2012.
Cleveland Islands Artrip
Species Common name
Site 1
(2011)
Site 1
(2012)
Site 2
(2012)
Site 3
(2012)
Actinonaias ligamentina Mucket
X
Actinonaias pectorosa Pheasantshell X X X X
Cyclonaias tuberculata Purple wartyback X
X
Elliptio dilatata Spike X X X X
Epioblasma brevidensFE
Cumberlandian combshell X X
Epioblasma capsaeformisFE
Oyster mussel X X
Epioblasma triquetraFE
Snuffbox
X
Fusconaia barnesiana Tennessee pigtoe XA X
A X X
Fusconaia corFE
Shiny pigtoe XB X
B X X
Fusconaia cuneolusFE
Fine-rayed pigtoe XB X
B
X
Fusconaia subrotunda Longsolid XA X
A X
Lampsilis fasciola Wavy-rayed lampmussel X X X X
Lampsilis ovata Pocketbook X X
X
Lasmigona costata Flutedshell X X X X
Medionidus conradicus Cumberland moccasinshell X X X X
Pleurobema oviforme Tennessee clubshell XA X
A
X
Pleuronaia dolabelloidesFE
Slabside pearlymussel X X X
Ptychobranchus fasciolaris Kidneyshell X X X X
Ptychobranchus subtentumFE
Fluted kidneyshell X
X
Quadrula cylindrica
strigillataFE
Rough rabbitsfoot X X X
Villosa iris Rainbow X X X X
Villosa vanuxemensis Mountain creekshell X X X A = Fusconaia barnesiana, F. subrotunda, and Pleurobema oviforme individuals were pooled
due to lack of positive identification at this site. B = Fusconaia cor and F. cuneolus individuals were pooled due to lack of positive identification
at this site. FE
= Federally endangered species
97
CHAPTER 2
Evaluation of Systematic Quadrat and Capture-Mark-Recapture Survey Techniques:
Monitoring a Reintroduced Population of Oyster Mussels (Epioblasma capsaeformis) in the
Upper Clinch River, Virginia.
98
ABSTRACT
A total of 1,418 translocated adult and 2,851 laboratory-propagated sub-adult federally
endangered oyster mussels (Epioblasma capsaeformis) were reintroduced from 2006–2011 in the
upper Clinch River at Cleveland Islands, Virginia. The objective of this study was to estimate
population size of this restored population using two survey methods and assess the effectiveness
of these methods to estimate population parameters. Demographic data were collected in 2011
and 2012 by systematic quadrat and capture-mark-recapture sampling. Systematic quadrat
sampling of translocated adult E. capsaeformis estimated population sizes of 577 (SE=155)
individuals in 2011 and 645 (SE=110) individuals in 2012. Systematic quadrat sampling of
laboratory-propagated sub-adult E. capsaeformis estimated population sizes of 1,678 (SE=42)
individuals in 2011 and 1,700 (SE=229) individuals in 2012. With similar point estimates but
more precision than quadrat sampling, capture-mark-recapture sampling produced translocated
adult population size estimates of 451 (SE=97) in 2011 and 370 (SE=80) in 2012. Also similar in
point estimates but with less precision than quadrat sampling, capture-mark-recapture produced
laboratory-propagated sub-adult population size estimates of 1,938 (SE=1,088) in 2011 and
1,390 (SE=611) in 2012. My results indicate that systematic quadrat and capture-mark-recapture
sampling have useful applications in population monitoring, but the choice among them is
dependent on project objectives. I recommend that monitoring projects utilize systematic quadrat
sampling when the objective is to simply estimate and detect trends in population size of species
at moderate to higher densities (>0.2/m2). Capture-mark-recapture sampling should be used
when objectives include assessing a reintroduced population of endangered species, and
obtaining precise population demographic estimates such as survival and recruitment, or
estimating population size for species of low to moderate densities (0.1–0.2/m2).
99
KEYWORDS: Freshwater Mussels, Capture-Mark-Recapture Sampling, Systematic Quadrat
Sampling, Oyster Mussel, Epioblasma capsaeformis, Demographic Data
100
INTRODUCTION
The federal recovery plan for the oyster mussel (Epioblasma capsaeformis) identifies the
quantification of demographic characteristics— such as population size, age-class structure, and
survival rates—as key to assessing species recovery (USFWS 2004). Estimation of demographic
parameters of biological populations is vital to understanding species-specific population
dynamics and, ultimately, determining population viability (USFWS 2004; Jones et al. 2012). In
recent years, reintroductions of freshwater mussel species into historical habitats where they had
become extirpated, and augmentations of extant but generally declining populations, have been
conducted in order to recover imperiled species and to prevent future losses (Haag 2012). These
recovery efforts require post-release monitoring of demographic vital rates in order to assess
restoration success and evaluate whether down- or delisting criteria have been met. Data from
post-release monitoring studies aid biologists in making informed decisions for effective
management (Sarrazin and Barbault 1996; Sarrazin and Legendre 2000; USFWS 2004; Jones
and Neves 2011).
A common methodology that is employed to collect demographic data for population
assessments of mussels is the quantitative quadrat sampling approach. The quadrat method
involves systematically sampling 0.25-m2 or 1-m
2 quadrats using a complete census within each
quadrat at a study site and extrapolating the finding across a wider area. Population parameters
such as species diversity (i.e., richness and evenness), size and density, growth rate, sex ratios,
age-class structure, survival, and evidence of recruitment all can be estimated (see Chapter 1).
From the length data and shells collected during quadrat sampling, catch-curve and shell thin-
sectioning analyses can be used to calculate approximate estimates of survival rates for
demographic models. However, the assumptions of these techniques for survival analyses are
101
rarely met by natural populations (Miranda and Bettoli 2007; Jones et al. 2012). Alternately,
following uniquely marked individuals (or cohorts of known age) through time allows improved
estimates of annual survival rates based on the fates of individuals captured. Although capture-
mark-recapture (CMR) has a long history in wildlife ecology (Lincoln 1930; Young et al. 1952;
Seber 1962; Jolly 1963, 1965; Cormack 1964; Edwards and Eberhardt 1967; Otis et al. 1978)
and is commonly used in studies of many other taxa (Dussart 1991; Karanth 1995; Mowat and
Strobeck 2000; Silver et al. 2004), it has been seldom used to monitor and assess mussel
populations (Pollock et al. 1990; Strayer and Smith 2003; Villella et al. 2004).
In a CMR design, repeated sampling of the population is required. During the first
capture occasion, all individuals captured are uniquely marked, recorded, and released back into
the population. During repeated sampling occasions, all marked individuals encountered are
recorded, and all unmarked individuals captured are uniquely tagged and recorded during each
occasion (Otis et al. 1978). Based on the proportion of marked to unmarked individuals,
population size can be estimated (Petersen 1896; Lincoln 1930; Jolly 1965; Seber 1982; Strayer
and Smith 2003; Villella et al. 2004). Data from individuals monitored over time also can be
used to model and estimate detection (i.e., capture and recapture, or encounter probabilities),
survival rates, and recruitment probabilities for each sampling period (Cormack 1964; Jolly
1965; Pollock 1982; Seber 1982; Pollock et al. 1990; Villella et al. 2004).
Capture-mark-recapture models can be categorized into two broad classes: closed- and
open-population models. Closed-population models are used in CMR designs when capture-
recapture occasions occur over a relatively short period (days to weeks) to ensure that no births
or deaths (demographic closure), and no emigration or immigration (geographic closure) occur
during the census. Open-population models are used over a longer time frame (e.g., years) in
102
order to estimate recruitment, mortality or migration (Seber 1982; Strayer and Smith 2003). A
common assumption of both model types is that all animals–marked and unmarked–are equally
likely to be caught during any sampling occasion (i.e., The Equal Catchability Assumption)
(Seber 1962; Jolly 1963, 1965; Cormack 1964; Seber 1982; Pollock 1982). However, this
assumption is frequently violated in CMR field studies by the inherent variability in capture
probabilities of individuals due to heterogeneity, trap response, temporal emigration, time
effects, and combinations of these and other factors (Pollock 1982; Seber 1982). To cope with
capture variability, models have been developed for closed and open population designs which
allow for parameter estimates that incorporate such factors (Pollock 1975, 1981, 1982; Otis et al.
1978; Seber 1982; White et al. 1982; Pollock and Otto 1983).
The Clinch River is part of the upper Tennessee River drainage, flowing southwest
through southwestern, VA and into northeastern, Tennessee (TN). In 2002, the Virginia
Department of Game and Inland Fisheries (VDGIF) designated approximately 19.3-km of the
upper Clinch River in VA as an population restoration reach for E. capsaeformis (Eckert and
Pinder 2010; VDGIF 2010). In collaboration with VDGIF’s Aquatic Wildlife Conservation
Center (AWCC) near Marion, VA, Virginia Tech’s Freshwater Mollusk Conservation Center
(FMCC) has been working to increase the local population size and viability of E. capsaeformis
within this reach over the last seven years (2006–2012). Cleveland Islands, at Clinch River
kilometer (CRKM) 435.8, was chosen as the population restoration site in the upper Clinch River
for the project.
As of 2011, a total of 1,418 translocated adult and 2,851 laboratory-propagated sub-adult
E. capsaeformis have been released at Cleveland Islands, VA. In order to determine the success
of these reintroduction efforts and assess population viability, population monitoring was
103
conducted in 2011 and 2012 by systematic quadrat sampling (Chapter 1) and CMR. This
population restoration site presented an ecological opportunity to increase knowledge of species-
specific demographic rates because all E. capsaeformis released at this site were uniquely
marked, aged, and sexed, and therefore could be followed through time. The objectives of my
study were to monitor this restored population of E. capsaeformis and evaluate a potentially new
survey technique by estimating population parameters at Cleveland Islands, and to compare and
assess the effectiveness of CMR versus quadrat methods for estimating population parameters.
To further compare CMR population parameter estimates and their precision to those from
traditional quadrat techniques, I also monitored two non-listed, naturally occurring, relatively
common species—the pheasantshell (Actinonaias pectorosa) and the Cumberland moccasinshell
(Medionidus conradicus)—at Cleveland Islands.
METHODS
Study Area
Mussel reintroductions were conducted in the upper Clinch River at Cleveland Islands
(CRKM 435.8) in the left descending channel (LDC). Cleveland Islands is located in Russell
County, near the town of Cleveland, VA. Owned by The Nature Conservancy and cooperatively
managed by VDGIF, Cleveland Islands are characterized by four channels formed by three
islands. The LDC contains excellent water quality, suitable flow conditions, stable gravel
substrates, and darter fish hosts utilized by E. capsaeformis and other mussel species. The most
recent survey of Cleveland Islands was conducted in 2008 by VDGIF (Eckert and Pinder 2010).
They found 23 live mussel species, including 7 federally endangered species—E. capsaeformis,
shiny pigtoe (Fusconaia cor), fine-rayed pigtoe (Fusconaia cuneolus), slabside pearlymussel
104
(Pleuronaia dolabelloides), fluted kidneyshell (Ptychobranchus subtentum), rough rabbitsfoot
(Quadrula cylindrica strigillata), and purple bean (Villosa perpurpurea). All E. capsaeformis
they encountered were tagged—indicating they were from recent translocations.
Epioblasma capsaeformis Translocations and Releases
Over a five-year period from 2006 to 2010, a total of 1,418 adult E. capsaeformis were
collected from the lower Clinch River, Hancock County, TN, and translocated along the LDC of
Cleveland Islands. An additional 2,501 and 350 laboratory-propagated sub-adult E. capsaeformis
were released into the LDC by the AWCC in 2010 and 2011, respectively. Each of these
translocated adult and released sub-adult E. capsaeformis were uniquely tagged (shellfish tag;
Hallprint Inc., Holden Hill, New South Wales, Australia), measured for length (mm), and sexed
for identification purposes. Generally, individuals <25 mm in size were not sexed because their
shells were not yet sexually dimorphic (i.e., the marsupial shell expansions of females were
undeveloped). However, the sex of a few translocated individuals, as small as 19 mm, was
predicted using inference from estimated age (e.g., if female, a 3–4+ year old estimated from
growth annuli should have begun to develop marsupial expansion). Additionally, individuals
were sexed at recapture and compared to original release data to further validate sex.
Sex ratio of translocated adults was approximately 1:1. At the time of translocation,
female adults ranged from 23 to 47 mm and averaged 36 mm in size, and male adults ranged
from 19 to 47 mm and averaged 33 mm. Laboratory-propagated sub-adults were approximately
1–2 years old (age class 2), ranged from 11 to 31 mm and averaged 21 mm in size. Translocation
of adults and releases of laboratory-propagated sub-adults were randomly distributed throughout
an approximately 125 x 15-m reach of the LDC.
105
Habitat Measurements
Upstream and downstream boundaries of the sampling study area were determined prior
to sampling by a preliminary qualitative snorkel survey. Observation of live E. capsaeformis or
shells, presence of other mussels, substrate composition, water depth and flow, and specific
locations of releases were taken into consideration. The LDC of Cleveland Islands is
approximately 125 m in length with an average wetted width of 14.8 m. The extended
boundaries (i.e., where the LDC reconnects with the main channel) of this study site are
approximately 35 m upstream and 100 m downstream of the core 125 m LDC, with average
wetted widths of 16.0 and 28.2 m, upstream and downstream, respectively. The estimated total
sample study area (A) was 5,085 m2. Banks were marked every 20 m with orange marking spray
to serve as a location guide during sampling.
Quadrat Sampling
Field Methods
Population demographic data for A. pectorosa, E. capsaeformis, and M. conradicus from
systematic quadrat sampling were collected in the LDC at Cleveland Islands, VA, following the
methods given in Chapter 1.
Estimation of Population Size and Density
Population sizes ( ) were defined as the total number of ≥1 year olds in the study area at
a particular point in time. This was estimated by multiplying the average count per systematic
106
sample by the total number of possible systematic samples (M) in the study area (Seber
1982; Smith et al. 2001; Strayer and Smith 2003):
∑
where: = abundance estimate,
M = number of possible systematic samples,
= count per systematic sample, and
m = number of systematic samples.
Because the site had four random starts (k=4), there were four systematic samples (m=4).
Dependent on the area (A=5085 m2) of the site, the area of the sampling unit (a=0.25 m
2), and
the total number of quadrats sampled (n), the total number of possible systematic samples (M)
was calculated following the formula in Smith et al. (2001):
∑
Variances for population sizes were estimated by the formula (Smith et al. 2001; Strayer
and Smith 2003):
( )
∑
For normally distributed sample data, the 95% confidence intervals for abundance were
calculated as:
√ ( )
Data was assessed for normality. Occasionally, the traditional approach to calculating
confidence intervals utilizing the assumption of a normal distribution has been found to be
inaccurate for mussel population size and density estimations. Based on mussel population
107
sampling simulations, mussel population sizes (or density) tend to have a positively (right)
skewed distribution (Pooler and Smith, unpublished data, cited by Smith et al. 2001; Strayer and
Smith 2003). If normality tests revealed a departure from normality, data were log-transformed
and 95% confidence intervals were calculated for abundance by using a logarithmic
transformation of the estimate and a delta-method approximation of variance (Seber 1982; Smith
et al. 2001; Strayer and Smith 2003):
(
√ ( )
)
Capture-Mark-Recapture Sampling
Field Methods
According to Otis et al. (1978), closed-population models with capture probabilities
averaging at least 0.10 require 5–10 sampling occasions to obtain reasonable estimates of
population size (Pollock 1982). Closed-population sampling designs would typically have
sampling occasions occurring on days close together (3–7 days apart) to ensure geographic and
demographic closure. This assumption of complete closure is frequently violated in CMR field
studies. Completing sampling within such a restrictive time frame is not always feasible due to
field conditions or labor availability. However, if closed population studies are properly
designed, closed-population CMR model assumptions can be met approximately (Otis et al.
1978).
Due to the large study site and time required to complete one capture occasion in my
study, a closed-population design of five occasions within a four-month period was chosen to
estimate population. This design permitted me to complete one closed-capture study of
108
individual encounter occasion (EO) histories per year (2011 and 2012). Time between
consecutive sampling occasions was set at 2–3 weeks; however, sampling time was dependent on
feasibility of sampling in unfavorable river conditions.
The study area was divided into twelve 20-m transects that extended from bank to bank.
Each transect was given a label (A–L), with A being at the downstream and L being at the
upstream boundary of the reach. Surveying begin at the downstream end and moved upstream
through each transect. Each transect reach was sampled and timed individually, and divided into
equal-width lanes (approximately 1 m wide) that ran parallel to the flow across the width of the
transect. Multiple surveyors lined up systematically and began sampling at the downstream end
of each lane, continued upstream, and stopped at the top of the transect. Surveyors then would
start over at the downstream end of that transect until all lanes within the transect survey area
were completed. This ensured that the entire transect substrate area from bank to bank was
systematically and thoroughly searched by surveyors while snorkeling. Excavation was not
performed during the CMR sampling. All individual mussels seen at the surface were collected
and their location in the substrate marked with surveyor’s flags. After a transect had been
completed, all mussels were identified, measured for length (mm), sexed (if possible), tag
numbers recorded, and placed back into the substrate where they were found. Any untagged E.
capsaeformis found throughout the study period were tagged before returning them to the
substrate.
In case population parameter estimation was influenced by population density, the size of
individuals, or typical species-specific behavior at the substrate surface, I included two additional
species, A. pectorosa and M. conradicus, in the CMR study in order to further compare the point
estimates and precision of CMR relative to quadrat sampling. I choose these two non-listed,
109
naturally occurring, relatively common species because they occur at moderate to high densities
(>0.2/m2) at Cleveland Islands, are characterized by different maximum sizes, and exhibit
different ‘availability for detection’ behavior when at the surface relative to E. capsaeformis.
Average maximum sizes of A. pectorosa, E. capsaeformis, and M. conradicus are 150, 48–55,
and 60 mm, respectively. ‘Availability for detection’ behavior at the substrate surface is defined
as the appearance of siphons or displays at the substrate surface that would influence surveyor
detection ability (e.g., larger siphons=individual more likely to be detected by surveyor). On
average, siphon appearance at the substrate surface of A. pectorosa, E. capsaeformis, and M.
conradicus are large, medium, and small, respectively. Additionally, female E. capsaeformis
possess a specialized behavior in which they display a whitish-blue mantle-lure to attract host
fish in the Spring to early Summer (Jones and Neves 2011). This display is very visible at the
substrate surface, thus increasing its likelihood for detection. All untagged A. pectorosa and M.
conradicus observed were tagged before returning them to the substrate beginning from the first
CMR EO in 2011 up to the fourth EO in 2012. Tagging was not necessary for the fifth EO in
2012 because it was the final sampling event of this study. All individual mussels of any species
seen at the surface during the first CMR sampling event (i.e., first EO) of each year (2011 and
2012) were collected; however, during all subsequent sampling events within each year, only A.
pectorosa, E. capsaeformis, and M. conradicus individuals were collected, measured, recorded,
and tagged if previously untagged.
Complete closure of A. pectorosa, E. capsaeformis, and M. conradicus populations were
assumed for within year sampling. Populations were defined as all individuals that were ≥1 year
old due to the difficultly of observing same-year recruitment (i.e., individuals <10 mm) in the
absence of excavating or sieving substrates, and thereby adhered to the assumption of no ‘births’
110
(i.e., recruits) during the capture period (Negishi and Kayaba 2010). Based on previous
laboratory and field studies, mussel annual mortality was assumed to be minimal (<5%) for
individuals within each population (Hua et al. 2011; Jones et al. 2012; C. Carey, Virginia
Polytechnic Institute and State University, unpublished data). Concerning migration, adult
mussels are sedentary and usually spend their entire lives in the same location with limited
means of dispersal (Vaughn and Taylor 1999). Based on these inferences, and the absence of E.
capsaeformis observations for this area prior to translocations and releases, the likelihood that: 1)
any individuals were recruited into the defined population (>10 mm), 2) mortality was
significant, and 3) that any individuals permanently migrated in or out of the study area within a
four-month period was minimal—thereby allowing the study to adhere to demographic and
geographical closure.
Estimation of Population Parameters
Population Size, Density, and Capture Probabilities—Population size and capture (i.e.,
encounter or detection) probabilities of tagged mussels were modeled within sampling year
(2011 and 2012). Population size ( ) was defined as the number of individuals in a defined area.
The capture probability ( ) was defined as the probability that a mussel is encountered given that
it was alive and in the study area. Collected E. capsaeformis were recorded in two data sets (one
for each year), each having five EOs, two groups (translocated adults and laboratory-propagated
sub-adults), and one covariate (length at capture). Similarly, A. pectorosa and M. conradicus
each had two data sets (one for each year). Actinonaias pectorosa data sets were comprised of
five EOs, one group, and individual length covariate. Medionidus conradicus data sets were
comprised of four and five EOs in 2011 and 2012, one group, and individual length covariates.
111
Individual covariates were added to the analysis to allow me to assess whether capture
probabilities of an individual(s) were a function of length at capture, and if so, whether this
variable applied to all three or just specific species.
I used Program MARK (White and Burnham 1999) to estimate population parameter
estimates with associated standard errors from EO histories of marked individuals within each
survey year. I used the Huggins’ Full Closed Captures with Heterogeneity model, which allows
for effects of time, behavior, and heterogeneity. The Huggins Closed Capture data type also
allows for the incorporation of individual covariates (e.g., length of individual) to be used to
model capture probabilities and to estimate population size. In contrast to the Closed Capture
model (i.e., full likelihood model), the Huggins Closed Capture model does not include
individuals in its likelihood that were never captured. Therefore population size ( ) is
conditioned out of the likelihood in the model (i.e., conditional likelihood model) and is
estimated as a derived parameter (Huggins 1989, 1991; Pledger 2000; White 2008; Cooch and
White 2012). The derived population size ( ) still is defined as the number of individuals in a
defined population, but is estimated, for data with no individual covariates, as
[ ]
where: = total number of individually-marked mussels captured at
least once,
= probability the mussel is captured at time (i) given that it is alive
and in the defined population, and
= probability of not being captured at time (i).
112
Individuals have their own capture probabilities ( ) when using individual covariates in
the analysis. The derived population size ( ) with individual covariates is now estimated by
calculating:
[ ]
for each individual and then taking the sum (White 2008; Cooch and White 2012). A more
complex formula and explanation is discussed by Huggins (1989, 1991).
Program MARK provided top model population size estimates and associated variances.
If two or more models were competing, models were averaged in Program MARK to provide
model averaged population size estimates and associated unconditional variances. The lower and
upper bounds of the confidence intervals for model averaged population sizes were derived as
(Cooch and White 2012):
{ [ (
)]
}
Population densities were calculated by dividing the modeled averaged population sizes
( ) by the sampling study area (A):
113
Variances for population densities were calculated by dividing the unconditional
variances ( ( ) by the squared area:
( )
The 95% log-based confidence intervals for population densities were calculated as
(Cooch and White 2012):
[ ( ( )
)]
Apparent Survival Rates
Apparent survival probabilities (φi) of tagged E. capsaeformis from year to year (2006–
2012) and recapture probabilities (pi) within a sampling year (2011 and 2012) were estimated by
Program MARK using Cormack-Jolly-Seber (CJS) models. Apparent survival (φ) is the
probability of surviving between EOs and being available for recapture given that the individual
has not permanently emigrated from the study area (the CJS model cannot distinguish between
mortality and losses due to permanent emigration) (Pledger et al. 2003; Villella et al. 2004).
Translocated and laboratory-propagated sub-adult E. capsaeformis were recorded in one data
input file with 16 EOs. The data set consisted of two groups, translocated adults and laboratory-
propagated sub-adults. Time intervals between successive EOs were set in weeks. Time intervals
1–5 ranged from 39–54 weeks apart (equivalent to approximately 9–12 months between release
occasions), 6–10 ranged from 1–4.5 weeks apart, 11 was 36 weeks, and 12–15 ranged from 2–
3.5 weeks apart. There were 15 recapture parameters (p2–p16) representing recapture probabilities
p during each occasion i that occurred after the 1st encounter history (2006), and 15 survival
114
parameters (φ1–φ15) representing survival probabilities φ between each occasion i in the data set.
A CJS diagram and data set formatting can be found in Appendix A.
Encounter history occasions 1–5 and 11 represented translocation and release events and
not active CMR surveys (2006–2010 annual release events, and the 2011 release event that
occurred after 2011 CMR sampling). These release events were treated as missing occasions
shown as ‘dots’ in the 1st through 5
th and 11
th EOs, and corresponding recapture probabilities ( )
were fixed at zero (i.e., recapture parameters p2–p5 and p11) to indicate that individuals were not
searched for on those occasions. However, the presence of a ‘1’ in the 1st through 5
th and 11
th
EOs indicated that the individual was released in that year. Encounter occasions 6–10 and 12–16
represented active CMR survey EOs in 2011 and 2012, respectively. All recapture probabilities
corresponding to active surveys in 2011 and 2012 (p6–p10 and p12–p16) were allowed to be time-
dependent. Survival parameters φ6–φ10 and φ12–φ15 were held constant (.) within 2011 and within
2012 because the population was assumed to be closed demographically within years. A time-
dependent constraint (i.e., allowed to vary year to year) was imposed on survival probabilities for
survival parameters φ1–φ5 and φ11—representing survival between initial release and the first
survey year 2011 and between 2011 and 2012 surveys (Appendix A).
To test whether stream discharge influenced recapture probabilities, a constraint of mean
daily discharge was imposed on recapture parameters p6–p10 and p12–p16. Daily mean values of
discharge for the Clinch River were obtained at U.S. Geological Survey gage number 03524000,
which is located 200 m upstream of the study site. As with E. capsaeformis, demographic data
collected for A. pectorosa or M. conradicus within each active sampling year, 2011 and 2012,
were treated as closed. Survival was not be estimated between 2011 and 2012 for A. pectorosa
115
and M. conradicus because with only two years of data, survival and recapture parameters are
not individually identifiable.
.
Program MARK Model Selection: Huggins Closed Capture & Cormack-Jolly-Seber Models
An a priori model candidate set based on species biology as well as selection models
from Otis et al. (1978), including a saturated model, were run in Program MARK. Given the
data, Program MARK uses Akaike’s Information Criterion (AIC) to select the most
parsimonious model—best fit with fewest parameters— to explain the variation in the data.
The AIC is an estimator of the difference between the unknown ‘true’ model that explains the
data and the given approximating model in our candidate set. To optimize precision and fit of the
model to the data, the AIC is calculated using the model likelihood and the number of parameters
in the model. The fit of the model is positively associated with model likelihood, thus as the
model fit increases, the AIC declines. More parameters in a given model indicates greater
uncertainty— thus as the precision decreases, models are penalized and AIC increases. In
otherwords, the model with the lowest AIC in the given candidate set is the ‘best’ model to
balance precision and fit and describe the data (i.e., model nearest to the unknown truth).
Specifically, Program MARK uses the corrected AIC (AICc) to account for small sample sizes.
To account for lack of fit between saturated and general models in a candidate set, the AICc is
adjusted to yield the quasi-likelihood adjusted AIC (QAICc) for model selection (Anderson et al.
2000; Boulanger et al. 2002; Johnson and Omland 2004; Cooch and White 2012).
Models are ranked in Program MARK by lowest–highest corrected or adjusted AIC
(AICc or QAICc, respectively). For each candidate model, the difference in AIC (ΔAIC) between
two models (the model with the lowest AIC and the given model) are provided. When ΔAIC<2
116
between two models, Burnham and Anderson (1998) suggest that it is reasonable to conclude
that both models have approximately equal weight in the data (i.e., as ΔAIC increases, there is
evidence to suggest a real difference between models). For model selection, generally models
with ΔAIC<2 have support, models within 2<ΔAIC<7 have less support, and models ΔAIC>7
have no support (Burnham and Anderson 1998; Anderson et al. 2000; Cooch and White 2012).
For my analysis, I reported any models with a ΔAIC<7 of the most parsimonious model.
Program CAPTURE was used to test for violations of the closure assumption for each
closed capture data set. Because Program CAPTURE is limited to running 2,000 individual
encounter histories, and the number of individual A. pectorosa encountered exceeded this limit, I
chose three random samples of 2,000 A. pectorosa encounter histories among the data sets for
each year (2011 and 2012) in order to test closure. While this closure test is unaffected by
heterogeneity in capture probabilities, it is not appropriate for populations that may exhibit time
or behavior variation in capture probabilities or temporary migration during the study period
(Otis et al. 1978; Stanley and Burnham 1999).
Additional goodness-of-fit (GOF) testing was done to verify that my saturated (or
general) model adequately fits the data (i.e., test underlying model assumptions) and to assess
overdispersion in the data (Boulanger et al. 2002; Cooch and White 2012). Lack of model fit
indicates that the assumptions underlying the model not being met. This is assessed by
measuring overdispersion, or extra-binomial noise, in the data set—the degree to which the data
exhibit greater variability than is predicted by the model (Boulanger et al. 2002; Hinde and
Demétrio 1998; Cooch and White 2012). By measuring overdispersion in the data, a quasi-
likelihood parameter (variation inflation factor, ĉ) can be estimated and lack of fit can be
117
corrected for. An estimate of ĉ=1 indicates the model fits the data, ĉ >1 indicates overdispersion,
and ĉ <1 indicates underdispersion (Cooch and White 2012).
To test lack of fit and produce a quasi-likelihood parameter (ĉ), I ran GOF tests on the
saturated (fully parameterized) model without individual covariates in Program MARK using the
χ2 test statistic that allows for time variation in capture probabilities. I used the median ĉ and
parametric bootstrapping approaches to test the closed-population and CJS models, respectively.
These approaches produced median ĉ values (closed-population model) and bootstrapped
simulation data (CJS model). If the logistic regression for the median ĉ test failed to run on the
fully saturated model, the next-most parameterized model was used. Bootstrapped simulation
data were used to estimate ĉ values by two approaches: 1) taking the ratio of the original data
deviance by the simulated mean deviance, and 2) taking the ratio of the observed ĉ (i.e., observed
deviance divided by the observed deviance degrees of freedom) by the simulated mean ĉ (Cooch
and White 2012). If overdispersion was detected (ĉ >1), the ĉ parameter was adjusted and QAICc
was used for model selection. If underdispersion was detected (ĉ<1), ĉ was left unadjusted at 1.
Likelihood ratio tests were utilized in Program MARK to compare nested models as
needed. Top models were chosen based on parsimony and biology, and top competing models
were then averaged to produce estimates of population size, capture probabilities, and survival
rates. A constraint was imposed on the final p in all models to ensure that it was identifiable (i.e.,
not confounded by final survival or recapture parameter; White 2008). To test the assumption
that survival is approximately 100% within a year (e.g., within 2011 and 2012 sampling
occasions), nested survival models were compared (i.e., time-dependent versus constant and/or
time fixed at zero model). I predicted E. capsaeformis capture probabilities to vary by group
(translocated adults versus un-sexed laboratory-propagated sub-adults), time, and to be positively
118
related to size. Similarly, I expected A. pectorosa and M. conradicus capture probabilities to
have a time effect and relation to size. Based on previous studies (see Chapter 1), I predicted E.
capsaeformis to have relatively high (>80%) annual survival rates.
Comparisons of Sampling Methods and Population Size Estimates
To compare sampling method estimators and between year (2011 to 2012) population
size estimates, unpaired t-tests were used to calculate estimates of the magnitude of difference
(unstandardized effect size=mean difference) and associated 95% confidence intervals. Unequal
variance t-tests were used to handle unequal sample sizes and variances (Satterthwaite 1946;
Welch 1947; Ruxton 2006). Results provided statistical and biological inference to whether the
sampling method estimates significantly differed. Biological importance was assessed based on
the magnitude of effect and in the context of the species. For example, an effect size of 1,000
individuals for E. capsaeformis would be biologically important in this study because it would
cause two very different conclusions to be drawn about the survival of reintroduced individuals
and the effectiveness of population restoration efforts at this site. Alternatively, the same effect
size may not be biologically important for A. pectorosa or M. conradicus, both of which are
established, common species at this site occurring at moderate to high density levels. Analyses
were conducted using SAS software (SAS Institute, Inc., Cary, North Carolina, version 9.2).
Growth
Length data from live individuals and shells of E. capsaeformis were collected during
systematic quadrat and CMR sampling and used to determine absolute growth:
119
where L1 is length at release, L2 is length at recapture, and t2-t1 is time between release and
recapture. The growth parameters asymptotic maximum length (L∞) and growth coefficient (k)
were estimated from absolute growth data using the NLIN procedure is SAS:
( )
where Lt is length at recapture and t is time between release and recapture.
RESULTS
Quadrat Sampling
A total of 44 E. capsaeformis were collected in 2011, comprised of 11 translocated
adults, 32 laboratory-propagated sub-adults, and 1 natural recruit. Similarly, 41 individuals were
collected in 2012, comprised of 11 translocated adults, 29 laboratory-propagated sub-adults, and
1 natural recruit.
Estimated abundances and densities of translocated adult E. capsaeformis were 577
(SE=155) individuals and 0.11/m2 (SE=0.03) in 2011, and 645 (SE=110) individuals and 0.13/m
2
(SE=0.02) in 2012. Estimated abundances and densities of laboratory-propagated sub-adult E.
capsaeformis were 1,678 (SE=42) individuals and 0.33/m2 (SE=0.01) in 2011, and 1,700
(SE=229) individuals and 0.33/m2 (SE=0.05) in 2012. Estimated abundances and densities of E.
capsaeformis recruits were 52 (SE=26) individuals and 0.01/m2 (SE=0.01) in 2011, and 59
(SE=29) individuals and 0.01/m2 (SE=0.01) in 2012. There was no significant difference
between the 2011 and 2012 estimates of abundance and density for translocated adults or
laboratory-propagated sub-adults (Figure 1).
Estimated abundance and density of A. pectorosa were 9,227 (SE=328) individuals and
1.81/m2 (SE=0.07) in 2011, and 7,972 (SE=482) individuals and 1.57/m
2 (SE=0.09) in 2012.
120
Estimated abundance and density of M. conradicus were 4,404 (SE=203) individuals and
0.87/m2 (SE=0.04) in 2011, and 5,158 (SE=349) individuals and 1.01/m
2 (SE=0.07) in 2012
(Figure 1).
Approximately 40 person-hours of effort were required to complete sampling of quadrats
each year. A total of 440 and 380 individuals representing 20 and 18 species were encountered in
2011 and 2012, respectively (Appendix B; Chapter 1).
Capture-Mark-Recapture Sampling
Summary of Encounters in 2011 and 2012
A total of 255 individual E. capsaeformis were collected in 2011, comprised of 144
translocated adults, 110 laboratory-propagated sub-adults, and one natural recruit. The recruit
was 23.8 mm, male, and 2–3 years old; lengths corresponding to 0–1 and 1–2 years old growth
annuli were 8.6 and 15 mm. Likewise, 231 individuals were collected in 2012, comprised of 98
translocated adults and 132 laboratory-propagated sub-adults. Time between consecutive
sampling occasions within each year was approximately 19 days apart. On average, 0.5 min of
search effort was required to survey 1 m2 of substrate surface area, and approximately 172
person-hours of effort were required to complete CMR sampling each year. This included time
spent surveying substrate surface, collecting mussels, and processing (i.e., recording, measuring,
and tagging) all study species (A. pectorosa, E. capsaeformis, and M. conradicus). A total of
9,923 and 5,719 individuals (including approximately 100 shells each year) representing 25
species (some Fusconaia and Pleurobema species were pooled due to lack of positive
identification) were collected in 2011 and 2012, respectively (Appendix B).
121
Closed Capture-Mark-Recapture Modeling: Population Size and Capture Probabilities
Epioblasma capsaeformis 2011—There were 293 captures (n) of 254 uniquely marked [M(t+1)]
E. capsaeformis in 2011, comprised of 144 translocated adults and 110 laboratory-propagated
sub-adults. Of the translocated adults, 83 were female and 61 were male. Of these 254
individuals, 33 were captured 2–3 time times (no individuals were encountered four or more
times). The closure test in CAPTURE was marginally significant (p=0.05), indicating a potential
violation of closure. A reduced model was used (pi[.]p[g*t]=c[g*t] in Table 1) for median ĉ GOF
testing in MARK due to nonsensical standard errors of parameter estimates for the full group x
time model. The median ĉ was 1.07 (SE=0.04), suggesting minimal overdispersion.
The candidate model set consisted of 22 starting (a priori) closed-capture models. I
removed ten of the starting models due to one or more nonsensical detection probabilities,
population size estimates, and associated standard errors that likely resulted from sparse data and
an inability to model certain parameters. Of these remaining 12 models, 2 had ΔQAICc<7.
Greater than 99% of parameter support came from these top two models, with 63% from the
most parsimonious model (Table 1). Model averaged population size and density estimates were
453 (SE=96) and 0.09/m2 (SE=0.02) for translocated adults and 1,915 (SE=1,030) and 0.38/m
2
(SE=0.20) for laboratory-propagated sub-adults (Table 2; Figure 1). There was support that
capture probabilities (p) varied by time and group and were a function of length (Table 1; Figure
2).
Epioblasma capsaeformis 2012—There were 254 captures (n) of 230 uniquely marked [M(t+1)] E.
capsaeformis in 2012, comprised of 98 translocated adults and 132 laboratory-propagated sub-
adults. Of the translocated adults, 66 were female and 32 were male. Of these 230 individuals, 21
122
were captured 2–3 times (no individuals were encountered four or more times). The closure test
in CAPTURE was not significant (p=0.25), indicating the population was closed during the
sampling period (i.e., met closure assumptions). The median ĉ estimated generated in MARK
was 2.78 (SE=0.07), suggesting overdispersion of the data.
The candidate model set consisted of 22 starting (a priori models same as 2011) closed
capture models. I removed ten of the starting models due to one or more nonsensical detection
probabilities, population size estimates, and associated standard errors that likely resulted from
sparse data and an inability to model certain parameters. Of these remaining 12 models, 3 had
ΔQAICc<7 (making up >99% of parameter support), with over 72% from the most parsimonious
model (Table 1). Derived population size and density estimates were 372 (SE=132) and 0.07/m2
(SE=0.03) for translocated adults, and 1,390 (SE=1,018) and 0.27/m2 (SE=0.20) for laboratory-
propagated sub-adults (Table 2; Figure 1). Capture probabilities (p) varied by time and were a
function of length (Table 1; Figure 2).
Actinonaias pectorosa 2011—A total of 3,771 [M(t+1)] A. pectorosa were uniquely marked in
2011, of which 1,641 individuals were captured on multiple occasions. There were 6,140
captures (n) over five encounter occasions. The test for closure procedure in CAPTURE was
significant (p<0.01), indicating that the population is most likely not closed during sampling.
Median ĉ estimate generated in MARK was 1.93 (SE=0.19), indicating moderate overdispersion
of the data.
The candidate model set consisted of 18 starting (a priori models same as E.
capsaeformis but with no group models) closed capture models. I removed two of the starting
models due to nonsensical detection probabilities and associated standard errors that likely
123
resulted from sparse data and an inability to model certain parameters. No models fell within
ΔQAICc<7 of the most parsimonious model, which >97% of parameter support came from the
most parsimonious model (Table 3). Derived population size and density estimates were 6,615
(SE=483) individuals and 1.30/m2 (SE=0.09) (Table 2; Figure 1). Capture probabilities (p) varied
by time and were a function of length (Table 3; Figure 3).
Actinonaias pectorosa 2012—A total of 2,471 [M(t+1)] A. pectorosa were uniquely marked in
2012, of which 722 individuals were captured on multiple occasions. There were 3,372 captures
(n) over the five encounter occasions. The test for closure procedure in CAPTURE was
significant (p<0.01), indicating that the population is most likely not closed during sampling.
Median ĉ estimate in MARK was 0.26 (SE=0.27), indicating underdispersion of the data.
The candidate model set consisted of 18 starting (a priori models same as 2011 A.
pectorosa models) closed capture models. I removed four of the starting models due to one or
more nonsensical detection probabilities, population size estimates, and associated standard
errors. Of these remaining models, three models had ΔAICc<7 (making up >99% of parameter
support), with >49% of parameter support from the most parsimonious model. These three
models were competing models, each within 2 ΔAICc of one another (Table 3). Model averaged
population size and density estimates were 4,729 (SE=203) individuals and 0.93/m2 (SE=0.04)
(Table 2; Figure 1). Capture probabilities (p) varied by time and were a function of length (Table
3; Figure 3).
Medionidus conradicus 2011—A total of 1,366 [M(t+1)] M. conradicus were uniquely marked in
2011, of which 338 individuals were captured on multiple occasions. There were 1,753 captures
124
(n) over the four encounter occasions. The test for closure procedure in CAPTURE was
significant (p=0.03), indicating that the population was likely not closed during sampling.
Median ĉ in MARK was 4.67 (SE=0.17), indicating overdispersion of the data. The median ĉ
estimate likely was large due to sparse data (only four encounter occasions).
The candidate model set consisted of 18 starting (a priori) closed capture models. I
removed eight of the starting models due to one or more nonsensical detection probabilities,
population size estimates, and associated standard errors. The remaining ten models consisted of
three models with a ΔQAICc<7, and two competing models (ΔQAICc<2) with 55.3% of
parameter support from the most parsimonious model, followed by 32.9% for the next best
supporting model (Table 4). Model averaged population size and density estimates were 3,237
(SE=825) individuals and 0.67/m2 (SE=0.16) (Table 2; Figure 1). Capture probabilities (p) varied
as a function of length (Figure 3). There was some support (2<ΔQAICc<7) that capture
probabilities (p) varied over time (Table 4; Figure 3).
Medionidus conradicus 2012—A total of 1,088 [M(t+1)] M. conradicus were uniquely marked in
2012 (or encountered from 2011 and considered a new tag), of which 194 individuals were
captured on multiple occasions. There were 1,304 captures (n) over the five encounter occasions.
The test for closure procedure in CAPTURE was not significant (p=0.23), indicating the
population met closure assumptions. Median ĉ in MARK was 0.60 (SE=0.13), indicating
underdispersion of the data, and ĉ was left unadjusted at 1.0.
The candidate model set consisted of 18 starting (a priori) closed capture models. I
removed eight of the starting models due to one or more nonsensical detection probabilities,
population size estimates, and associated standard errors. Of these remaining ten models, six
125
models had a ΔQAICc<7. The top two models were competing, with 59.8% of parameter support
from the most parsimonious model followed by 25.0% for the next best supporting model (Table
4). Model averaged population size and density estimates were 2,849 (SE=160) individuals and
0.56/m2 (SE=0.03) (Table 2; Figure 1). There was support that capture probabilities (p) varied
over time and as a function of length (Table 4; Figure 3).
Open Capture-Mark-Recapture Modeling: Survival and Recapture Probabilities
Epioblasma capsaeformis—There were a total of 4,841 captures (n) (includes initial releases) of
4,269 uniquely marked [M(t+1)] E. capsaeformis, comprised of 1,418 translocated adults and
2,851 laboratory-propagated sub-adults. Of these 4,269 individuals released, 3,828 were never
recaptured (i.e., not observed after release during CMR sampling in 2011 and 2012). There were
a total of 533 recaptures (i.e., captures after initial release) consisting of 441 unique individuals.
Of these 441 individuals, 365 were recaptured once, 62 were recaptured twice, 12 were
recaptured three times, and 2 were recaptured four times. No individuals were recaptured five or
more times over the 2011 and 2012 CMR sampling events.
Bootstrap parametric and median ĉ GOF tests may not be accurate due to the missing
encounter occasions in my data set. Parametric bootstrap simulations provided ĉ estimates of
0.13 and 0.67, indicating underdispersion. However, because the observed model deviance fell
below all deviances from the simulated data (p=1.00), I concluded that there is an adequate fit of
the model to the data. Median ĉ in MARK was 1.26 (SE=0.01), indicating moderate
overdispersion of the data. The ĉ parameter was adjusted to 1.26 for model selection.
The candidate model set consisted of 24 starting (a priori) CJS models incorporating
group, heterogeneity, time, and gage discharge covariate effects. I removed 20 of the starting
126
models due to one or more nonsensical survival estimates, detection probabilities, and associated
standard errors that likely resulted from sparse recapture data relative to number of individuals
released and an inability to model certain parameters. Additionally, all starting models
incorporating gage discharge did not differ from corresponding models without gage discharge
and therefore were removed from model selection. Of these remaining four models, >99% of
parameter support came from the most parsimonious model. Apparent annual survival estimates
were >99% for translocated adults and laboratory-propagated sub-adults. There was support that
recapture probabilities (p) varied by time and group (Table 6; Figure 4).
Comparisons of Sampling Methods and Population Size Estimates
Statistically (p<0.05) and biologically significant differences were revealed between
estimator methods for A. pectorosa 2011 (p <0.01, effect size=2,612) and 2012 (p =0.01,effect
size=3,243), and M. conradicus 2012 (p =0.01, effect size=2,309) population size estimates.
Statistically significant differences were revealed between estimator methods for laboratory-
propagated sub-adult E. capsaeformis 2011 (p<0.01) and for native M. conradicus 2011 (p
=0.01) population size estimates; however, these differences were not considered biologically
significant at effect sizes of 237 and 1,167 individuals, respectively. No significant differences in
population size estimates were revealed by comparisons of estimator methods (Table 5).
Generally, CMR estimates were more precise than systematic quadrat estimates. The
CMR population size estimates were more precise than systematic quadrat estimates for
translocated adult E. capsaeformis, and for native A. pectorosa and M. conradicus in 2011 and
2012. Neither population size estimators produced consistent precise estimates of laboratory-
127
propagated sub-adult E. capsaeformis, which may be related to their smaller size and subsequent
lower capture probabilities (Table 2; Table 5; Figure 1).
Actinonaias pectorosa population sizes estimated by CMR revealed a statistically
(p<0.01) and biologically significant difference between 2011 and 2012, with an effect size of
1,886 (SE=4) individuals that represented a density decline of 0.37/m2. Actinonaias pectorosa
population size estimated by systematic quadrats revealed a marginally statistical (p=0.08) and
biologically significant difference, with an effect size of 1,255 (SE=583) individuals, that
represented a density change of 0.25/m2 from 2011 to 2012. Statistical significance was detected
between 2011 and 2012 population sizes estimated by CMR for E. capsaeformis (p<0.01) and M.
conradicus (p<0.01). However, these differences were not considered biologically significant at
effect sizes of 81, 525, and 388 individuals for translocated adult and laboratory-propagated sub-
adult E. capsaeformis, and M. conradicus, respectively (Table 5).
None of the other within species comparisons (estimated by either survey method),
showed a significant decline or increase in population size between 2011 and 2012 (Table 5;
Figure 1). Population size of recruits (individuals captured that were neither translocated nor
laboratory-propagated) could not be modeled in Program MARK because of insufficient data
(i.e., only one individual was encountered on one occasion during CMR sampling), therefore no
comparisons to systematic quadrat estimates were made.
Growth
Recaptured E. capsaeformis length data indicate that growth (i.e., length from one year to
the next) of translocated adults and laboratory-propagated sub-adults can be estimated using the
equations:
128
( ) for female translocated adults,
for male translocated adults, and
for unsexed, laboratory-propagated sub-adults.
DISCUSSION
My study has shown that E. capsaeformis population restoration efforts since 2006 were
successful in the upper Clinch River at Cleveland Islands, VA, and that CMR has useful
application in the inference of demographic parameters for mussels. Recruitment was
documented by systematic quadrat and CMR survey methods, indicating that natural
reproduction is occurring for this species. Estimated population parameters were similar between
systematic quadrat sampling and the CMR approach. Although CMR was nearly four times more
time intensive (person-hours effort) than quadrat sampling, I processed nearly three times the
number of mussels per person-hour effort and collected over 5 and 10 times the number of
individual laboratory-propagated sub-adult and translocated adult E. capsaeformis, respectively,
than I did from systematic quadrat sampling. Because CMR resulted in data from more
individuals, it provided more precise estimates of population size and density, and improved
estimates of survival, and growth rates. To my knowledge, this is the first study to
simultaneously conduct systematic quadrat and CMR surveys to estimate population parameters
for mussels and to determine monitoring design effectiveness for each method.
Estimating species-specific demographic vital rates is essential to assessing
reintroduction success, evaluating whether delisting criteria have been met, and for developing
effective management plans (USFWS 2004; Villella et al. 2004; Jones and Neves 2011; Meador
et al. 2011). The common methodology for collecting demographic data is the quantitative
129
quadrat survey, which has been used in the Clinch River since the mid-1970s (Dennis 1985).
Systematic quadrat sampling is a probability-based survey method for assessing rare or clustered
populations, is simple to execute in the field, and offers effective spatial coverage (Christman
2000; Smith et al. 2001; Strayer and Smith 2003). In addition, with probability-based sampling,
the probability that a species is present at a specified mean density even if it were not detected
can be estimated (Green and Young 1993; Strayer and Smith 2003).
Because not all mussels are available at the surface for one quadrat survey point in time
(temporary emigration), population parameter estimates may be biased if quadrat excavation is
not executed (Amyot and Downing 1991; Smith et al. 2001). Even when excavation is applied to
minimize temporary emigration, it is unknown whether excavation disrupts substrate
composition and stability, causes increased mortality, disrupts reproduction, or causes significant
displacement of individuals (i.e., permanent emigration) (Smith et al. 1999). In addition to
possible biological disturbances, excavation can be resource intensive. Excavating or not,
quadrat sampling is often problematic to implement in deep water and high velocity habitats
(Meador et al. 2011). Although useful for estimating and detecting trends in abundance and
density, quadrat sampling provides only broad estimates of species diversity, sex ratios, length-
frequency distributions, growth rates, age-class structure, and periodic survival and recruitment
rates—particularly when target species are at low densities (Table 7).
Less commonly used to assess mussel populations, CMR-based designs track and collect
data on uniquely tagged individuals over time in order to model and estimate demographic
parameters. The assumption of equal catchability of marked and unmarked mussels during any
sampling occasion often is violated due to temporary emigration (i.e., vertical migration) and
leads to biased population parameter estimates. A properly designed and executed CMR study
130
can model the effects of temporary emigration, providing more precise population parameter
estimates, and determine what factors are influencing survival and capture probabilities (Otis et
al. 1978; Pollock 1990; Villella et al. 2004; Meador et al. 2008).
Similar to quadrat surveys, CMR is useful for estimating and detecting trends in
abundance and density, but in addition, it can: 1) offer improved precision of species density
estimates (i.e., low to high), 2) provide information for species’ vital rates (i.e., annual survival,
recruitment) which are difficult to determine with quadrat methods, 3) investigate factors
influencing survival and capture, and 4) validate and improve species-specific demographic
models (Table 7; Cormack 1964; Jolly 1965; Seber 1982; Strayer and Smith 2003; Villella et al.
2004). Improved estimates of population parameters are partly a result of high numbers of
captures and recaptures of individuals, thus increasing sample size. Additionally, testing what
factors are important predictors of capture probabilities provides biologists with guidelines to the
conditions under which specific species are most likely to be encountered at any given time (i.e.,
under what specific temperature, discharge, reproductive condition, etc.), and to creating
efficient monitoring plans (Villella et al. 2004; Meador et al. 2011). Because of the increasing
importance of understanding and monitoring species-specific population dynamics for
conservation management, the use of CMR studies for mussels has been expanding (Villella et
al. 2004; Meador et al. 2011).
Mussels exhibit seasonal (time) and species-specific life requirement patterns of vertical
migration in substrate. Several factors can influence when mussels are available at the substrate
surface— and consequently influence their capture probability—to include age, size (length),
water temperature, stream discharge, day length, habitat type, and reproductive condition
(Amyot and Downing 1991, 1997; Balfour and Smock 1995; Watters et al. 2001; Villella et al.
131
2004; Meador et al. 2011). Some of these studies have shown that larger and older individuals
tend to be more epibenthic than juveniles and smaller individuals, even during warmer months.
Mussel size can positively influence surveyor ability to detect individuals at the substrate surface
simply because larger individuals are easier to detect visually.
A few mussel CMR studies have examined the factors influencing temporary emigration
and capture probabilities (Villella et al. 2004; Meador et al. 2011; B. Watson, VDGIF,
unpublished data). Villella et al. (2004) and Meador et al. (2011) found that vertical migration
patterns varied by species and season, and suggested it was associated with reproductive
behavior (e.g., actively spawning or releasing glochidia). They also determined that capture
probabilities were influenced by body size (i.e., shell length), water temperature, and habitat
type. Capture probabilities may also be affected by environmental sampling conditions, such as
discharge affecting visibility due to turbidity (B. Watson, VDGIF, unpublished data).
In agreement with those of Meador et al. (2011) and Villella et al. (2004), my results
indicate that capture probabilities varied by species and with time, and were positively associated
with shell length. Though my mean capture probability estimates for E. capsaeformis (2–6%)
were lower than those reported for other species during the warmer months by Meador et al. (8–
20%; 2011) and Villella et al. (7–19%; 2004), my mean capture probability estimates for A.
pectorosa (11–24%) and M. conradicus (7–15%) were similar. The lower capture rates of E.
capsaeformis may reflect the difficulty of detecting a species existing at much lower densities
(<0.4/m2) than my other two study species, or be related to species-specific behavior or smaller
body size. Actinonaias pectorosa had the highest capture probabilities, presumably because
individuals are larger. Interestingly, capture probability trends were similar for all three species
within each year. Capture probabilities were higher in mid-June, then declined by mid-July
132
before steadily increasing through late September. Because all three species are long-term
brooders—spawning in late summer and autumn, gravid through winter, and releasing glochidia
the following summer—capture patterns may be related to vertical migration due to reproductive
behavior (Watters et al. 2001; Villella et al. 2004). Because mean capture probabilities were
different within species among years, but trends were similar within years among species,
environmental conditions (e.g., temperature, stream discharge) additionally may have played a
role.
Estimating vital rates and identifying factors influencing mussel survival is important for
developing effective conservation plans. Several factors have been found through CMR studies
that influence annual survival, including age, length, habitat type, and stream discharge (Villella
et al. 2004; Meador et al. 2011). Over a four-year CMR study using a CJS model, Villella et al.
(2004) estimated high annual survival rates for the Eastern elliptio (Elliptio complanata), and
suggested that survival was time- and size-dependent. Using a PIT-tag CMR methodology, Hua
et al. (2011) documented high annual survival (>98%) for the Cumberlandian combshell
(Epioblasma brevidens) in the Powell River, TN. Furthermore, during a one-year Robust Design
CMR study, Meador et al. (2011) reported survival variation among habitat type and a positive
association with shell length. In agreement with these other CMR findings of high adult annual
survival rates (>90%), my results indicated that E. capsaeformis exhibited very high annual
survival (>98%). By understanding these and other factors affecting survival, reintroduction plan
details such as habitat characteristics of the release site, and recommendations regarding age,
length and sex of released individuals can made to optimize annual survival, and ultimately,
restoration success.
133
In addition to providing important insights into the ecological relationships affecting vital
rates and capture probabilities (Villella et al. 2004), CMR studies can be used to validate
previous conclusions on vital rates estimated from other sampling approaches, such as
quantitative quadrats, length-at-age catch-curves or shell thin-sectioning analyses. By following
unique individuals through time, I was able to estimate survival rates based on fates of
individuals captured. Although original aging of uniquely marked translocated adults was
estimated using predicted age-at-length curves, my study was able to estimate annual survival
rates by combining capture histories with known time since release. Similarly, tagged laboratory-
propagated sub-adults were of known age at release and provided concrete age-specific data for
estimating annual survival rates. The results of my study were in agreement with previous
predictions—high annual survival for sub-adult and adult age-classes—estimated using shell
thin-sectioning analyses and length-at-age data of mussels collected from systematic quadrat
surveys in the Clinch River, TN (Jones et al. 2012).
Presently, a von Bertalanffy growth curve (von Bertalanffy 1938) could not accurately be
fitted to my growth data for E. capsaeformis for several reasons: 1) the ages of translocated
adults were estimates based on predicted length-at-age equations in Jones and Neves (2011), 2)
laboratory-propagated sub-adult growth data only represented younger age classes (≤ 3 years
old), and 3) shell thin-sectioning was not conducted. Using the available laboratory-propagated
sub-adult length-at-age data for 1–3 year olds likely would have resulted in biased estimates of
growth parameters. Nevertheless, the collection of more individuals through CMR sampling did
allow me to make inferences about growth of reintroduced individuals using absolute growth
(i.e., directly measured growth). Growth parameters from previously estimated von Bertalanffy
growth curves are not directly comparable to those from my study, but can be used as a base-line
134
for comparison. Jones and Neves (2011) reported higher growth coefficients (k) for naturally
reproduced female (0.27) and male (0.42) E. capsaeformis in the Clinch River, TN than the
translocated adult females (0.17) and males (0.13) evaluated in my study. In contrast to Jones
and Neves (2011), this likely is because my estimated growth parameters do not represent the
growth curve over the lifespan of E. capsaeformis, but represent growth after handling,
transportation, and release into a new habitat (i.e., growth earlier in life is not included). My
estimated growth coefficients are lower and may reflect growth variation because my data
represent only older age classes (≥ 3 years old). Younger age classes (< 3 years old) of E.
capsaeformis exhibit accelerated growth (larger k) before decreasing as individuals approach
older and maximum ages (Jones and Neves 2011); thus, the absence of the first years of growth
data for translocated adults would produce underestimates of growth coefficients.
Stress from a physical disturbance also may have influenced growth after release.
Mussels are known to lay down disturbance rings, which represent brief cessations of growth due
to factors such as handling (Haag and Commens-Carson 2008). Population restoration through
reintroductions or augmentations involves transportation and considerable physical handling; it is
unknown how much this may disrupt short- and long-term growth (Cope and Waller 1995).
Additional sources of variation in estimated growth parameters from my study relative to those
from Jones and Neves (2011) may include differences in site characteristics, such as water
chemistry and temperature (Haag and Rypel 2011), mean annual streamflow (Rypel et al. 2008),
or growth rate determination techniques (i.e., measuring absolute growth of marked individuals
versus shell thin-sectioning ) (Kesler and Downing 1997). Through shell thin-sectioning and
future monitoring at my study site, a complete length-at-age data set for introduced E.
capsaeformis can be compiled and a predicted length-at-age equation can be computed to
135
compare to that of Jones and Neves (2011). Consequently, further data from marked individuals
can be used to test the assumption of shell annuli formation for E. capsaeformis and accuracy
and precision of shell thin-sectioning.
Also of concern is whether sampling and monitoring efforts can cause declines in
abundance and density due to disturbance. It is unknown how much disturbance—through the
excavation of substrate or removal of mussels from substrate for processing—influences
mortality rates or increases displacement of individuals. In this study, CMR indicated a
significant decline in A. pectorosa abundance between 2011 and 2012. This decline was not
revealed by the systematic quadrat survey, presumably due to the larger variation in the quadrat
survey abundance estimator. It is likely that displacement from the site—rather than natural or
induced mortality—was responsible for this decline in abundance. Other studies suggested low
to no mortality from presumed similar levels of handling stress, nor did they reveal related
declines in abundance (Kesler and Downing 1997; Villella et al. 2004). Even though all mussels
were returned to where they were found in the substrate, the average size of A. pectorosa
individuals was larger than those of the other study species and these mussels may have had a
more difficult time re-burrowing into the substrate after handling. This in combination with
surveyors moving about the stream bed and high flow events after surveying may have displaced
some A. pectorosa individuals downstream out of the survey area, resulting in the decline in
abundance noted from the study area.
Despite the large number of mussel population restoration projects that have been
conducted over the last century (Haag 2012), few have determined the long-term success of these
efforts (Cope and Waller 1995). Detecting trends in population size, estimating species-specific
vital rates, identifying factors influencing capture and survival, and long-term monitoring are
136
essential to developing effective conservation plans and to determine long-term success of
reintroduction efforts (Sarrazin and Legendre 2000). By performing and reporting post-
restoration population monitoring, projects can provide insight into the relative success of
method-specific restoration efforts and population viability. Both systematic quadrat and CMR
sampling techniques have useful applications in population monitoring—and towards assessing
population viability—but are dependent on project objectives.
My results indicate that CMR is a more useful method than quadrat sampling for
monitoring population abundance and density trends. Employing CMR methodologies improves
our knowledge and precision of species-specific vital rates, and accounts for incomplete capture
of mussels, both shortcomings of quadrat sampling. However, CMR can be more resource
intensive depending on the scope of the project and sampling design. In addition to considering
project objectives and availability of resources, the selection of an appropriate mussel CMR
design should consider study site area and species distribution. If a study area is large (>2,000
m2), and target species distribution is random, biologists should sample a random sub-set of
strip-transects because it is less resource intensive and still provides precision in population
parameter estimates. If a study area is small (<500 m2), a CMR design surveying the entire
substrate surface area could easily be implemented in a cost-effective manner. Finally, if
monitoring is long-term (≥3 years), a Robust Design CMR study should be implemented to allow
estimation of recruitment—that which is crucial to evaluating population viability (Villella et al.
2004; Jones et al. 2012).
I recommend that monitoring projects utilize systematic quadrat sampling when the
objective is to simply estimate and detect trends in population size for established species, or
restored species, of moderate to larger densities (>0.2/m2). Capture-mark-recapture sampling
137
should be used when objectives include assessing restored populations of species reintroduced or
augmented at low to moderate densities, obtaining precise population demographics (e.g.,
survival and recruitment), or estimating population size for any species occurring at low to
moderate densities (<0.2/m2). Future mussel restoration efforts should continue to tag released
individuals and use CMR to improve our understanding of species-specific demographic
characteristics as well as assess likelihood of success of species restorations. Assuming released
E. capsaeformis laboratory-propagated sub-adults from the 2010 release will reach sexual
maturity in 2012 (i.e., based on size, >35 mm), and that environmental conditions are favorable
for reproduction, it is likely that recruitment from these individuals could be assessed as early as
2014 (1–2 year-old recruits). In accordance with the recovery plan for E. capsaeformis (USFWS
2004), I suggest Cleveland Islands, VA be monitored biennially beginning in 2014 to determine
age-1 class survival and recruitment rates, compute length-at-age equations, and assess
population viability of the species.
138
LITERATURE CITED
Amyot, J.P., and J.A. Downing. 1991. Endo- and epibenthic distribution of the unionid mollusc
Elliptio complanata. Journal of the North American Benthological Society 10:280–285.
Anderson, D.R., K.P. Burnham, and W.L. Thompson. 2000. Null hypothesis testing: problems,
prevalence, and an alternative. The Journal of Wildlife Management 64:912–923.
Balfour, D.L., and L.A. Smock. 1995. Distribution, age structure, and movements of the
freshwater mussel Elliptio complanata (Mollusca: Unionidae) in a headwater stream.
Journal of Freshwater Ecology 10:255–268.
Bates, J.M., and S.D. Dennis. 1978. The mussel fauna of the Clinch River, Tennessee and
Virginia. Sterkiana 69–70:3–23.
Boulanger, J., G.C. White, B.N. McLellan, J. Woods, M. Proctor, and S. Himmer. 2002. A meta-
analysis of grizzly bear DNA mark-recapture projects in British Columbia, Canada.
Ursus 13:137–152.
Burnham, K.P., and D.R. Anderson. 1998. Model Selection and Inference: A Practical
Information-Theoretical Approach. Springer-Verlag, New York.
Christman, M.C. 2000. A review of quadrat-based sampling of rare, geographically clustered
populations. Journal of Agriculture, Biological, and Environmental Statistics 5:168–201.
Cope, G.W., and D.L. Waller. 1995. Evaluation of freshwater mussel relocation as a
conservation and management strategy. Regulated Rivers: Research and Management
11:147–155.
Cooch, E.G., and G.C. White. 2013. Program MARK: a gentle introduction. 12th
edition.
Available: http://www.phidot.org/software/mark/docs/book/ (August 2013).
139
Cormack, R M. 1964. Estimates of survival from the sighting of marked animals. Biometrika
51:429–438.
Dennis, S.D. 1985. Distributional analysis of the freshwater mussel fauna of the Tennessee River
system, with special reference to possible limiting effects of siltation. Report No. 85-2.
Tennessee Wildlife Resources Agency. Nashville. 171 pp.
Dussart, G.B.J. 1991. Mark-recapture experiments with freshwater organisms. Journal of
Biological Education 25:116–118.
Eckert, N.L., and M.J. Pinder. 2010. Freshwater mussel survey of Cleveland Island, Clinch
River, Virginia: augmentation monitoring site 2008. Virginia Department of Game and
Inland Fisheries, Richmond. 47 pp.
Edwards, W.R., and L.L. Eberhardt. 1967. Estimating cottontail abundance from live-trapping
data. The Journal of Wildlife Management 31: 87–96.
Green, R.H., and R.C. Young. 1993. Sampling to detect rare species. Ecological Applications
3:351–356.
Haag, W.R., and A.M. Commens-Carson. 2008. Testing the assumption of annual shell ring
deposition in freshwater mussels. Canadian Journal of Fisheries and Aquatic Sciences
65:493–508.
Haag, R.H., and A.L. Rypel. 2011. Growth and longevity in freshwater mussels: evolutionary
and conservation implications. Biological Reviews 86:225–247.
Hinde, J., and C.G.B. Demétrio. 1998. Overdispersion: models and estimation. Computational
Statistics and Data Analysis 27:151–170.
Hua, D., J. Rogers, J. Jones, and R. Neves. 2011. Propagation, culture, and monitoring of
endangered mussels for population restoration in the Clinch and Powell Rivers,
140
Tennessee, 2006–2010. Report to the Tennessee Wildlife Resources Agency, Nashville.
44 pp.
Huggins, R.M. 1989. On the statistical analysis of capture experiments. Biometrika 76:133–140.
Huggins, R.M. 1991. Some practical aspects of a conditional likelihood approach to capture
experiments. Biometrics 47:725–732.
Johnson, J.B., and K.S. Omland. 2004. Model selection in ecology and evolution. Trends in
Ecology and Evolution 19:101–108.
Jolly, G.M. 1963. Estimation of population parameters from multiple recapture data with both
death and dilution—deterministic model. Biometrika 50:113–128.
Jolly, G.M. 1965. Explicit estimates from capture-recapture data with both death and
immigration-stochastic model. Biometrika 52:225–247.
Jones, J.W. 2004. A holistic approach to taxonomic evaluation of two closely related endangered
freshwater mussel species, the oyster mussel (Epioblasma capsaeformis) and tan
riffleshell (Epioblasma florentina walkeri). Master’s thesis. Virginia Polytechnic Institute
and State University, Blacksburg.
Jones, J.W., R.J. Neves, S.A. Ahlstedt, and E.M. Hallerman. 2006. A holistic approach to
taxonomic evaluation of two closely related endangered freshwater mussel species, the
oyster mussel Epioblasma capsaeformis and tan riffleshell Epioblasma florentina walkeri
(Bivalvia: Unionidae). Journal of Molluscan Studies 72:267–283.
Jones, J.W., and R.J. Neves. 2011. Influence of life-history variation on demographic responses
of three freshwater mussel species (Bivalvia: Unionidae) in the Clinch River, USA.
Aquatic Conservation: Marine and Freshwater Ecosystems 21:57–73.
141
Jones, J.W., R.J. Neves, and E.M. Hallerman. 2012. Population performance criteria to evaluate
reintroduction and recovery of two endangered mussel species, Epioblasma brevidens
and Epioblasma capsaeformis (Bivalvia: Unionidae). Walkerana, Journal of the
Freshwater Mollusk Conservation Society 15:27–44.
Karanth, K.U. 1995. Estimating tiger Panthera tigris populations from camera-trap data using
capture-recapture models. Biological Conservation 71:333–338.
Kesler, D.H., and J.A. Downing. 1997. Internal shell annuli yield inaccurate growth estimates in
freshwater mussels Elliptio complanata and Lampsilis radiata. Freshwater Biology
37:325–332.
Lincoln, F.C. 1930. Calculating waterfowl abundance on the basis of banding returns. U.S.
Department of Agriculture, Biological Survey Bureau. Circular No. 118. 4 pp.
Meador, J.R., J.T. Peterson, and J.M. Wisniewski. 2011. An evaluation of the factors influencing
freshwater mussel capture probability, survival, and temporary emigration in a large
lowland river. Journal of the North American Benthological Society 30:507–521.
Miranda, L.E., and P.W. Bettoli. 2007. Mortality. Pages 229–277 in: C.S. Guy and M. L. Brown,
eds. Analysis and Interpretation of Freshwater Fisheries Data. American Fisheries
Society, Bethesda, Maryland.
Mowat, G., and C. Strobeck. 2000. Estimating population size of grizzly bears using hair
capture, DNA profiling, and mark-recapture analysis. The Journal of Wildlife
Management 64:183–193.
Otis, D.L., K. P. Burnham, G.C. White, and D.R. Anderson. 1978. Statistical inference from
capture data on closed animal population. Wildlife Monographs 62:3–135.
Petersen, C.G.J. 1896. The yearly immigration of young plaice into the Limfjord from the
142
German Sea. Report of the Danish Biological Station for 1895. 6:1–77.
Pledger, S. 2000. Unified maximum likelihood estimates for closed capture-recapture models
using mixtures. Biometrics 56:434–442.
Pledger, S., K.H. Pollock, and J.L. Norris. 2003. Open capture-recapture models with
heterogeneity: I. Cormark-Jolly-Seber model. Biometrics 59:786–794.
Pollock, K.H. 1975. A k-sample tag-recapture model allowing for unequal survival and
catchability. Biometrika 62:577–583.
Pollock, K.H. 1981. Capture-recapture models allowing for age-dependent survival and capture
rates. Biometrika 37:521–529.
Pollock, K.H. 1982. A capture-recapture design robust to unequal probability of capture. The
Journal of Wildlife Management 46:752–757.
Pollock, K.H., and M.C. Otto. 1983. Robust estimation of population size in closed animal
populations from capture-recapture experiments. Biometrics 39:1035–1049.
Pollock, K.H., J.D. Nichols, C. Brownie, and J.E. Hines. 1990. Statistical inference for capture-
recapture experiments. Wildlife Monographs 107:3–97.
Ruxton, G. D. 2006. The unequal variance t-test is an underused alternative to Student’s t-test
and the Mann-Whitney U test. Behavioral Ecology 17:688–690.
Rypel, A.L., W.R. Haag, and R.H. Findlay. 2008. Validation of annual growth rings in
freshwater mussel shells using cross dating. Canadian Journal of Fisheries and Aquatic
Sciences. 65:2224–2232.
Satterthwaite, F.E. 1946. An approximation distribution of estimates of variance components.
Biometrics Bulletin 2:110–114.
Seber, G.A.F. 1962. The multi-sample single recapture census. Biometrika 49:339–350.
143
Seber, G.A.F. 1982. The Estimation of Animal Abundance and Related Parameters, 2nd edition.
MacMillian, New York.
Silver, S.C, L.E.T. Ostro, L.K. Marsh, L. Maffei, A.J. Noss, M.J. Kelly, R.B. Wallace, H.
Gómex, and G. Ayala. 2004. The use of camera traps for estimating jaguar Panthera
onca abundance and density using capture/recapture analysis. Oryx 38:148–154.
Smith, D.R., R.F. Villella, D.P. Lemarié, and S. von Oettingen. 2000. How much excavation is
needed to monitor freshwater mussels? Pages 203–218 in: R.A. Tankersley, D.I.
Warmolts, G.T. Watters, B.J. Armitage, P.D. Johnson, and R.S. Butler, eds. Freshwater
Mollusk Symposia Proceedings. Ohio Biological Survey, Columbus.
Smith, D.R., R.F. Villella, and D.P. Lemarié. 2001. Survey protocol for assessment of
endangered freshwater mussels in the Alleghany River, Pennsylvania. Journal of the
North American Benthological Society 20:118–132.
Stanley, T.R., and K.P. Burnham. 1999. A closure test for time-specific capture-recapture data.
Environmental and Ecological Statistics 6:197–209.
Strayer, D.L., and D.R. Smith. 2003. A Guide to Sampling Freshwater Mussel Populations.
American Fisheries Society, Monograph 8, Bethesda, Maryland.
U.S. Fish and Wildlife Service. 2004. Recovery plan for Cumberland elktoe (Alasmidonta
atropurpurea), oyster mussel (Epioblasma capsaeformis), Cumberlandian combshell
(Epioblasma brevidens), purple bean (Villosa perpurpurea), and rough rabbitsfoot
(Quadrula cylindrica strigillata). Atlanta, Georgia. 174 pp.
Villella, R.F., Smith, D.R., and Lemarie, D.P. 2004. Estimating survival and recruitment in a
freshwater mussel population using mark-recapture techniques. American Midland
Naturalist 151:114–133.
144
Virginia Department of Game and Inland Fisheries. 2010. Virginia freshwater mussel restoration
strategy: upper Tennessee River basin. Bureau of Wildlife Resources, Wildlife Diversity
Division, Nongame and Endangered Wildlife Program, Richmond. 17 pp.
Von Bertalanffy, L. 1938. A quantitative theory of organic growth. Human Biology 10:181–213.
Watters, G.T., S.H. O’Dee, and S. Chordas III. 2001. Patterns of vertical migration in freshwater
mussels (Bivalvia: Unionida). Journal of Freshwater Ecology 16:541–549.
Welch, B.L. 1947. The generalization of ‘Student’s’ problem when several different population
variances are involved. Biometrika 34:28–35.
White, G.C., D.R. Anderson, K.P. Burnham, and D.L. Otis. 1982. Capture-Recapture and
Removal Methods for Sampling Closed Populations. Los Alamos National Laboratory.
235 pp.
White, G.C., and K.P. Burnham. 1999. Program MARK: survival estimation from
populations of marked animals. Bird Study 46 Supplement:120–138.
White, G.C. 2008. Closed population estimation models and their extensions in Program MARK.
Environmental and Ecological Statistics 15:89–99.
Young, H., J. Neess, and J.T. Emlen. 1952. Heterogeneity of trap response in a population of
house mice. The Journal of Wildlife Management 16:169–180.
145
Table 1. Top models, model used for median ĉ GOF test, descriptions, and model summary
statistics for E. capsaeformis at Cleveland Islands, Virginia in 2011 and 2012 using closed-
capture models in Program MARK. Summary statistics in bold indicate that the model was used
(top model or in model averaging) to describe the data set in that year.
Model Description
QAICc
Rank ΔQAICc
AIC
Weight
(wi)
Model
Likelihood
No.
Parameters Deviance
2011
pi(0) p(t)=c(t) +
Length
Temporal variation in
detection probabilities &
length covariate
1 0.00 0.63 1.00 6 970.21
{pi(0)
p(g*t)=c(g*t)+Length}
Temporal variation in
detection probabilities for
each group & length
covariate
2 1.07 0.37 0.59 11 961.14
pi(.) p(h)=c(h) +
Length
Heterogeneity in capture
probabilities & length
covariate
3 10.62 0.00 0.00 3 986.88
pi(0) p(g)=c(g) +
Length
Group specific detection
probabilities & length
covariate
4 12.00 0.00 0.00 3 988.26
A pi(0) p(g*t)=c(g*t) Temporal variation in
detection probabilities for
each group
5 15.54 0.00 0.00 10 977.65
2012
pi(0) p(t)=c(t) +
Length
Temporal variation in
capture/recapture
probabilities & length
covariate
1 0.00 0.73 1.00 6 314.74
pi(0) p(t)=c(t) Temporal variation in
capture/recapture
probabilities
2 3.36 0.14 0.19 5 320.12
pi(0) p(t) c(.) Temporal variation in
capture probabilities,
constant recapture
probabilities with behavior
effect
3 5.02 0.06 0.08 6 319.76
pi(0) p(t) c(t) Temporal variation in
capture/recapture
probabilities with
behavioral effect
4 7.69 0.02 0.02 8 318.38
A pi(0) p(g*t) c(g*t) Temporal variation in
capture/recapture
probabilities for each group
with behavioral effect
12 18.14 0.00 0.00 16 312.48
AModel used for median ĉ GOF test
146
Table 2. Population size and density estimates for E. capsaeformis, A. pectorosa, and M. conradicus at Cleveland Islands, Virginia
from closed-capture modeling in Program MARK.
2011 2012
Mean SE
95% CI
Mean SE
95% CI
Lower Upper Lower Upper
Population Size (N-hat)
Epioblasma capsaeformis
Translocated adults 453 96 314 703
372 132 113 669
Laboratory-propagated sub-adults 1,915 1,030 747 5,220
1,390 1,018 313 5,057
Actinonaias pectorosa* 6,615 483 5,815 7,729
4,729 203 2,366 5,162
Medionidus conradicus 3,237 825 2,186 5,639
2,849 160 2,562 3,192
Density (per m2)
Epioblasma capsaeformis
Translocated adults 0.09 0.02 0.06 0.13
0.07 0.03 0.04 0.15
Laboratory-propagated sub-adults 0.38 0.20 0.13 1.08
0.27 0.20 0.07 1.15
Actinonaias pectorosa* 1.30 0.09 1.13 1.50
0.93 0.04 0.86 1.01
Medionidus conradicus 0.67 0.16 0.39 1.05 0.56 0.03 0.50 0.63
*Statistically significant difference (p<0.05) detected between 2011 and 2012.
147
Table 3. Top models, model used for median ĉ GOF test, descriptions, and model summary
statistics for A. pectorosa at Cleveland Islands, Virginia in 2011 and 2012 from closed-capture
modeling in Program MARK. Summary statistics in bold indicate that the model was used (top
model or in model averaging) to describe the data set in that year.
Model Description
QAICc
Rank ΔQAICc
AIC
Weight
(wi)
Model
Likelihood
No.
Parameters Deviance
2011
pi(.)
p(t*h)=c(t*h)
+Length
Temporal variation and
heterogeneity in
capture/recapture probabilities &
length covariate
1 0.00 0.98 1.00 12 11267.51
pi(.) p(h) c(h)
+Length
Heterogeneity in
capture/recapture probabilities
with behavior effect & length
covariate
2 8.84 0.01 0.01 6 11288.37
pi(.)
p(h)=c(h)
+Length
Heterogeneity in
capture/recapture probabilities &
length covariate
3 9.84 0.01 0.01 4 11293.37
pi(0) p(t) c(.) Temporal variation in capture
probabilities, constant recapture
probabilities with behavior effect
& length covariate
4 13.54 0.00 0.00 7 11291.06
Api(0) p(t)
c(t)
Temporal variation in
capture/recapture probabilities
with behavior effect
13 247.32 0.00 0.00 8 11522.85
2012
pi(0) p(t) c(.)
+ Length
Temporal variation in capture
probabilities, constant recapture
probabilities with behavior effect
& length covariate
1 0 0.49 1 7 12394.24
pi(0) p(t)=c(t)
+Length
Temporal variation in capture
probabilities & length covariate
2 1.01 0.30 0.60 6 12397.24
pi(0) p(t)
c(t)} +Length
Temporal variation in capture
probabilities with behavior effect
& length covariate
3 1.70 0.21 0.43 9 12391.93
pi(.) p(h) c(h)
+Length
Heterogeneity in capture
probabilities with behavior effect
& length covariate
4 138.15 0.00 0.00 6 12534.39
Api(0) p(t)
c(t)
Temporal variation in
capture/recapture probabilities
with behavior effect
9 182.46 0.00 0.00 8 12574.70
AModel used for median ĉ GOF test
148
Table 4. Top models, model used for median ĉ GOF test, descriptions, and model summary
statistics for M. conradicus at Cleveland Islands, Virginia in 2011 and 2012 from closed-capture
modeling in Program MARK. Summary statistics in bold indicate that the model was used (top
model or in model averaging) to describe the data set in that year.
Model
Description
QAICc
Rank ΔQAICc
AIC
Weight
(wi)
Model
Likelihood
No.
Parameters Deviance
2011
pi(0) p(.)=c(.)
+Length
Constant capture/recapture
probabilities & length covariate
1 0 0.55 1.00 2 1237.71
pi(0) p(.) c(.)
+Length
Constant capture and recapture
probabilities with behavior effect
& length covariate
2 1.04 0.33 0.59 3 1236.75
pi(0) p(t)=c(t)
+Length
Temporal variation in capture
probabilities & length covariate
3 4.42 0.06 0.11 5 1236.12
pi(0) p(t) c(.)
+Length
Temporal variation in capture
probabilities, constant recapture
probabilities with behavior effect
& length covariate
4 5.54 0.03 0.06 6 1235.23
Api(0) p(t)c(t) Temporal variation in
capture/recapture probabilities
with behavior effect
10 16.96 0.00 0.00 6 1246.66
2012
pi(0) p(t)=c(t)
+Length
Temporal variation in
capture/recapture probabilities &
length covariate
1 0.00 0.60 1.00 6 4847.33
pi(0) p(t)=c(t) Temporal variation in
capture/recapture probabilities 2 1.74 0.25 0.42 5 4851.08
pi(0) p(t)c(t)
+Length
Temporal variation in
capture/recapture probabilities
with behavior effect & length
covariate
3 4.58 0.06 0.10 9 4845.89
pi(0) p(t) c(.)
+Length
Temporal variation in capture
probabilities, constant recapture
probabilities with behavior effect
& length covariate
4 5.12 0.05 0.08 7 4850.45
Api(0) p(t)c(t) Temporal variation in
capture/recapture probabilities
with behavior effect
5 6.21 0.03 0.04 8 4849.53
AModel used for median ĉ GOF test
149
Table 5. Contrasts of population size estimates between systematic quadrat and CMR sampling
methods, and between 2011 and 2012, for E. capsaeformis, A. pectorosa and M. conradicus at
Cleveland Islands, Virginia. Effect sizes are defined as the mean difference in population size.
Contrasts
Effect
Size Error (SEp)
95% CI
Lower Upper
Population size between sampling methods
Epioblasma capsaeformis
Translocated adults
2011 124 155 -369 617
2012 273 110 -77 623
Laboratory-propagated
sub-adults
2011* 237 51 111 363
2012 310 231 -426 1,046
Actinonaias pectorosa
2011* 2,612 328 1,567 3,657
2012* 3,243 482 1,710 4,776
Medionidus conradicus
2011* 1,167 204 519 1,815
2012* 2,309 349 1,199 3,419
Population size between 2011 and 2012
CMR sampling
Epioblasma capsaeformis
Translocated adults* 81 4 74 88
Laboratory-propagated
sub-adults*
525 35 457 593
Actinonaias pectorosa* 1,886 4 1,879 1,894
Medionidus conradicus* 388 11 366 410
Systematic quadrat sampling
Epioblasma capsaeformis
Translocated adults 68 190 -397 533
Laboratory-propagated
sub-adults
22 233 -720 764
Actinonaias pectorosa 1,255 583 -171 2,681
Medionidus conradicus 754 404 -234 1,742
*Statistically significant difference (p<0.05) detected.
150
Table 6. Top models, descriptions, and model summary statistics for E. capsaeformis at Cleveland Islands, Virginia in 2011 and 2012
from open-capture modeling (Cormack-Jolly-Seber) in Program MARK. Summary statistics in bold indicate that the model was used
to describe the data set.
Model Description
AICc
Rank ΔAICc
AIC
Weight
(wi)
Model
Likelihood
No.
Parameters Deviance
Phi(.) p(g+t)
with
constraints
Constant survival rates. Additive temporal and group
variation in recapture probabilities. Constraints: recapture
probabilties during releases
1 0.00 1.00 1.00 12 179.4058
Phi(.) p(t) with
constraints
Constant survival rates. Temporal variation in recapture
probabilities. Constraints: recapture probabilities during
releases
2 128.07 0.00 0.05 11 309.4906
Phi(.) p(t) with
constraints
Constant survival rates. Groupl variation in recapture
probabilities. Constraints: recapture probabilities during
releases
3 208.68 0.00 0.00 3 406.1427
Phi(.) p(g*t)
with
constraints
Constant survival rates. Temporal and group variation in
recapture probabilities. Constraints: recapture probabilties
during releases
4 323.66 0.00 0.00 20 486.9525
151
Table 7. Summary of pros, cons, and recommendations regarding systematic quadrat and
capture-mark-recapture sampling approaches to monitoring freshwater mussels.
Systematic quadrat sampling Capture-mark-recapture
Pros Offers good estimates of
population size for species that
occur at moderate to high densities
(>0.2/m2)
Offers improved precision of abundance
estimates for species that occur at low
to moderate (≤0.2/m2) densities
Useful for detecting trends in
density
Useful for detecting trends in
abundance and density
Relatively quick, simple, and cost
effective to implement in all sized
study sites
Relatively quick and simple to
implement in small (<500 m2) study
sites
Offers effective spatial coverage Spatial coverage can be customized to
project objectives (e.g., complete
surface area versus random strip-
transect sampling)
Can detect recruitment Can detect recruitment and obtain
reliable estimates of recruitment over
long-term study
Good for follow-up monitoring of
restored populations of moderate
to high densities (>0.2/m2)
Good for monitoring restored
populations regardless of density
Provides more precise estimates of
population demographics (population
size, density, growth rate; survival rates;
sex-ratios; growth; age-class structure)
and detecting species presence than
quadrat sampling
Useful for investigating factors that
influence survival and capture
probabilities
Cons Estimates of population
demographics for species of low to
moderate densities (≤0.2/m2) are
inaccurate and imprecise and
likely inaccurate
Complete study area substrate surface
coverage can be more resource
intensive to conduct than quadrat
sampling in large (>2,000 m2) study
sites
Excavation of quadrat samples is
difficult and time consuming to
implement in deep and high
velocity habitats
Additional costs incurred if using
shellfish tags
Biological disturbance caused by
excavation is unknown
Recommendations Should be conducted when survey
objective is to estimate or detect
trends in population size or density
for species that occur at moderate
to high densities (>0.2/m2)
Should be utilized when it is necessary
to obtain accurate and precise estimates
of population demographics and when
monitoring the status of restored
populations of endangered species and
other species at low densities
Use if study site is large, and
project has limited available
resources
If project has resource constraints and
target species is randomly distributed
within study area, sample random strip-
transects
152
Figure 1. Comparison of capture-mark-recapture and systematic quadrat population size
estimates and associated 95% confidence intervals for: A) translocated adult and B) laboratory-
propagated sub-adult (LPSA) E. capsaeformis, C) A. pectorosa, and D) M. conradicus at
Cleveland Islands, Virginia in 2011 and 2012.
153
Figure 2. Capture (and recapture, p and c) probabilities and associated 95% confidence intervals
for E. capsaeformis per sampling occasion for translocated adults in: A) 2011 and B) 2012, and
released laboratory-propagated sub-adults (LPSA) in: C) 2011 and D) 2012 at Cleveland Islands,
Virginia using a closed-capture model in Program MARK.
154
Figure 3. Capture (and recapture, p and c) probabilities and associated 95% confidence intervals
for A. pectorosa in: A) 2011 and B) 2012, and M. conradicus in C) 2011 and D) 2012 at
Cleveland Islands, Virginia using a closed-capture model in Program MARK.
155
Figure 4. Epioblasma capsaeformis recapture probabilities (p) and associated 95% confidence
intervals per sampling occasion for translocated adults in: A) 2011 and B) 2012, and released
laboratory-propagated sub-adults (LPSA) in: C) 2011 and D) 2012 at Cleveland Islands, Virginia
using an open-capture model (Cormack-Jolly-Seber) in Program MARK.
156
Appendix A: Cormack-Jolly-Seber Diagram and Program MARK Input Formatting
Figure A. 1. Cormack-Jolly-Seber open-capture model diagram for E. capsaeformis. Black
numbers in boxes represent encounter occasions; numbers 1–5 represent 2006–2010 annual
release events (no searches=p fixed at 0), 11 represents 2011 release event (no search=p fixed at
0) that occurred between 2011 and 2012 capture-mark-recapture sampling, and boxes 6–10 and
12–16 represent capture-mark-recapture active searches with 5 encounter occasions each in 2011
and 2012 (active searches=p time dependent). Red Phii (φi) values represent survival probability
parameters between successive occasions. Blue pi’s represent recapture probability parameters
during encounter occasions. Black numbers above arrows represent the time in weeks between
occasions.
157
/*ID Encounter History Trans group LPSA group semicolon*/
/*REDAA691*/ ...1.00100.00000 1 0 ;
/*REDAA692*/ ...1.00000.00010 1 0 ;
/*REDAA693*/ ...1.01000.00000 1 0 ;
/*REDAA694*/ ...1.00000.00000 1 0 ;
/*REDAA695*/ ...1.00000.00000 1 0 ;
/*REDAA696*/ ...1.00000.10000 1 0 ;
/*REDAA697*/ ...1.00000.00000 1 0 ;
/*REDAA698*/ ...1.01110.10000 1 0 ;
/*REDAA699*/ ...1.00000.00000 1 0 ;
Figure A. 2. A sample of Program MARK input formatting for E. capsaeformis open population
modeling (Cormack-Jolly-Seber model). The first two columns represent the ID (tag) of an
individual and its associated encounter history. The last two columns represent the group the
individual was classified under (translocated adult or a laboratory-propagated sub-adult).
158
APPENDIX B: Species List
Table B. 1. Species collected in the upper Clinch River at Cleveland Islands, Virginia using
systematic quadrat and capture-mark-recapture (CMR) sampling in 2011 and 2012.
Species Common name
Quadrats
(2011)
CMR
(2011)
Quadrats
(2012)
CMR
(2012)
Actinonaias ligamentina Mucket
X X X
Actinonaias pectorosa Pheasantshell X X X X
Amblema plicata Threeridge X
Cyclonaias tuberculata Purple wartyback X X
X
Elliptio crassidens Elephantear X
Elliptio dilatata Spike X X X X
Epioblasma brevidensFE
Cumberlandian combshell X X X X
Epioblasma capsaeformisFE
Oyster mussel X X X X
Epioblasma triquetraFE
Snuffbox
X X X
Fusconaia barnesiana Tennessee pigtoe XA X
A X
A X
A
Fusconaia corFE
Shiny pigtoe XB X
B X
B X
B
Fusconaia cuneolusFE
Fine-rayed pigtoe XB X
B X
B X
B
Fusconaia subrotunda Longsolid XA X
A X
A X
A
Lampsilis fasciola Wavy-rayed lampmussel X X X X
Lampsilis ovata Pocketbook X X X X
Lasmigona costata Flutedshell X X X X
Ligumia recta Black sandshell X X
Medionidus conradicus Cumberland moccasinshell X X X X
Plethobasus cyphyusFE
Sheepnose X X
Pleurobema oviforme Tennessee clubshell XA X
A X
A X
A
Pleuronaia dolabelloidesFE
Slabside pearlymussel X X X
Ptychobranchus fasciolaris Kidneyshell X X X X
Ptychobranchus subtentumFE
Fluted kidneyshell X X
X
Quadrula cylindrica
strigillataFE
Rough rabbitsfoot X X X
Villosa iris Rainbow X X X X
Villosa vanuxemensis Mountain creekshell X X X X A = Fusconaia barnesiana, F. subrotunda, and Pleurobema oviforme individuals were pooled
due to lack of positive identification at this site. B = Fusconaia cor and F. cuneolus individuals were pooled due to lack of positive identification
at this site. FE
= Federally endangered species
159
CHAPTER 3
Determining Optimum Temperature for Growth and Survival of Laboratory-Propagated
Juveniles of Two Federally Endangered Species, Cumberlandian Combshell (Epioblasma
brevidens) and Oyster Mussel (Epioblasma capsaeformis), and One Non-Listed Species,
Wavyrayed Lampmussel (Lampsilis fasciola)
Co-authors: J. Jones, E. Hallerman, and R. Butler
This is an Author’s Original Manuscript of an article whose final and definite form, the Version
of Record, has been publish in the North American Journal of Aquaculture, 25 September 2013,
copyright Taylor & Francis, available online at:
http://www.tandfonline.com/doi/pdf/10.1080/15222055.2013.826763#.UpNs1MTihZg
160
ABSTRACT
The effects of temperature on growth and survival of laboratory-propagated juvenile
freshwater mussels of two federally endangered species, the Cumberlandian combshell
(Epioblasma brevidens) and oyster mussel (Epioblasma capsaeformis), and one non-listed
species, the wavyrayed lampmussel (Lampsilis fasciola), were investigated to determine
optimum rearing temperatures for these species in smallwater-recirculating aquaculture systems.
Juveniles 4–5 months old were held in downweller buckets at five temperatures. Growth and
survival of juveniles were evaluated at 2-week intervals for 10 sampling events. At the end of the
20-week experiment, mean growth at 20, 22, 24, 26, and 28°C was, respectively, 0.75, 2.22, 3.27,
4.23, and 4.08 mm for E. brevidens; 1.35, 3.73, 3.81, 4.90, and 4.70 mm for E. capsaeformis;
and 2.09, 3.96, 4.99, 5.13, and 4.87 mm for L. fasciola juveniles. Generally, temperature was
positively correlated with growth of juveniles. Final mean maximum growth occurred at 26°C for
all three species, although no significant differences in growth were detected between 26°C and
28°C. The relationship between temperature and survival of juveniles was less clear. Final
survival was 82.5, 89.0, 91.0, 89.5, and 93.5% for E. brevidens; 73.0, 83.5, 78.0, 78.0, and
68.1% for E. capsaeformis; and 75.0, 89.5, 87.0, 86.5, and 89.5% for L. fasciola juveniles at the
five temperature treatments, respectively. Based on the species used in this study, results indicate
that 26°C is the optimum temperature for maximizing growth of juvenile mussels in downweller
bucket systems. The ability to grow endangered juveniles to larger sizes will improve survival in
captivity and upon release into the wild and will reduce time spent in hatcheries. As a result,
hatcheries can increase their overall production and enhance the likelihood of success of mussel
population recovery efforts by federal and state agencies and other partners.
161
KEYWORDS: Freshwater Mussels, Temperature, Growth, Survival, Laboratory-Propagated
Juveniles, Culturing Methods, Oyster Mussel, Cumberlandian Combshell, Wavyrayed
Lampmussel
162
INTRODUCTION
Because of significant declines of mussel populations in recent decades (Williams et al.
1993; Neves et al. 1997; Neves 1999), and with the culture and release of laboratory-propagated
mussels into the wild being applied as a recovery method (USFWS 2003, 2004; Jones et al.
2005; Jones et al. 2006; Eckert and Pinder 2010), there is a growing need to improve culture
methods, particularly grow-out of propagated juveniles. Water temperature is a vital
environmental parameter affecting growth and survival of juvenile mussels in captivity and in
the wild, also affecting various reproductive processes in adults, such as gametogenesis,
spawning, and larval brooding (Krebs 1972; Hastie et al. 2000; Zimmerman and Neves 2002;
Gosling 2003; Hastie et al. 2003; Zimmerman 2003; Jones et al. 2005; Negishi and Kayaba
2010; Pandolfo et al. 2010b). Temperature affects mussel developmental and physiological
processes, with specific effects on different life stages (Krebs 1972; Negishi and Kayaba 2010;
Pandolfo et al. 2010b). Efforts to propagate and culture mussels in captivity require an
understanding of the environmental factors that influence growth and survival at different life
stages. Thus, defining the optimum temperature for production of laboratory-propagated juvenile
mussels is critical for optimizing propagation and culture success and thereby has important
implications for their conservation.
In the past few years, it has become clear that larger and older laboratory-propagated
juveniles have a significantly increased chance of survival when released in the wild in
comparison to newly-metamorphosed juveniles (Sarrazin and Legendre 2000; Hua et al. 2011).
Although methods have been developed to produce thousands of newly-metamorphosed
juveniles, refinement of culture methods to grow these species to larger sizes is needed. The
ability to grow juveniles of imperiled species to larger sizes improves survival of individuals
163
while captive and upon release to the wild by decreasing the incidence of predation in both
settings. In addition, enhancing grow-out of cultured mussels increases detection probabilities for
subsequent monitoring, and most importantly, improves the likelihood of population recovery
(Zimmerman et al. 2003; Hua et al. 2011).
The purpose of my study was to determine the effect of temperature on the growth and
survival of juvenile (>4 months old and ≥1.5 mm) mussels of two federally endangered species,
Cumberlandian combshell (Epioblasma brevidens) and oyster mussel (Epioblasma
capsaeformis), and one non-listed species, wavyrayed lampmussel (Lampsilis fasciola), in
captivity. The intent of this research was to determine optimum rearing temperatures to
maximize growth and survival of juvenile mussels of these three mussel species in captivity.
METHODS
Gravid Mussel Collection
Juveniles were produced by the Freshwater Mollusk Conservation Center (FMCC) at
Virginia Polytechnic Institute and State University (Virginia Tech) in Blacksburg, and Virginia
Department of Game and Inland Fisheries’ Aquatic Wildlife Conservation Center (AWCC) near
Marion, Virginia (VA) following standard propagation and culture methods for these organisms.
Gravid females of each species were collected in May 2011 by snorkeling and using view scopes
in the lower Clinch River, Hancock County, Tennessee (TN). Gravid individuals were held and
transported to the FMCC and AWCC in coolers containing river water with aeration.
After arriving at the facilities, gravid females were placed in holding systems with
maintained water temperatures of 15°C in order to prevent early glochidial release before
infestation of host fishes could be conducted. The holding system at the FMCC contained 50–80
164
mm of river substrate (pebble, gravel) and water from the facility’s pond; the holding system at
the AWCC contained 50–80 mm of coarse limestone gravel substrate and water sourced from the
South Fork Holston River. Mussels were fed daily with a premixed commercial algae diet
(Nanno 3600 and Shellfish 105 Diet 1800 from Reed Mariculture, Campbell, California).
Host Fish Collection and Care
Based on the results of previous studies (Zales and Neves 1982; Yeager and Saylor
1995), black sculpin Cottus baileyi were used as the host fish for E. brevidens and E.
capsaeformis, and largemouth bass Micropterus salmoides were used as the host for L. fasciola.
Black sculpin were collected using a backpack electrofisher (Model LR24, Smith-Root,
Vancouver, Washington) and largemouth bass were obtained from a regional fish farm in
Arkansas.
Black sculpin were held and transported to each facility in 140-L coolers containing local
stream water. Salt was added to coolers to increase salinity to 0.7‰ in order to reduce fish stress
during transport. Water temperature was maintained at ambient stream levels during
transportation, and dissolved oxygen was maintained using an aerator. Transport time ranged
from 1 to 2 hours. After arrival at culture facilities, fish were acclimated to laboratory conditions,
regarding temperature and salinity, and were quarantined for 2–3 days at a salinity of 3.0‰ prior
to being infested with glochidia.
Infestation with Mussel Glochidia and Juvenile Mussel Collection
Host fish were infested with mussel glochidia following FMCC established non-lethal
laboratory protocols (Zale and Neves 1982; Neves 2004). At the FMCC, 180 black sculpin were
165
separated into groups of 45 fish which were placed into one of four 16-L containers with 3.5-L of
conditioned water at 21°C under continuous aeration. Glochidia from two gravid E.
capsaeformis were mixed into each of the four containers (eight gravid E. capsaeformis in total)
and allowed 45 minutes to attach to host fish. After infestation, host fish were moved into
recirculating aquaculture holding systems. Water quality parameters were monitored bi-weekly
in the host-fish holding systems. Similar host-fish infestation methods were used at AWCC to
produce juvenile mussels.
Once juveniles began to excyst from host fish, tank water was siphoned daily through
300-μm and 150-μm mesh sieves. Collected juveniles were rinsed into a petri dish, counted, and
placed into 18-L downweller bucket culture systems for growth and development (Barnhart
2006; Figure 1). Buckets were filled with 18 L filtered (<5 μm) pond (FMCC) or river (AWCC)
water, bucket water was exchanged once per week, and water temperatures were maintained
between 20 and 24°C. At each water exchange interval, buckets were cleaned and standard water
quality parameters were measured. Juveniles were fed continuously with a premixed commercial
algae diet. Because young juveniles experience a mortality bottleneck (4–8 weeks of age) and
are susceptible to flatworm predation at small sizes (Henley et al. 2001; Jones et al. 2005),
juveniles were cultured for 4–5 months to the desired initial size of 1–2 mm before the culture
experiment was initiated in order to remove any confounding factors. A summary of gravid
mussel collection, captive holding conditions, and host fish infestation protocols are given for
each species in Table 1.
166
Test Conditions
Juvenile mussels were acclimated to 20°C before testing and then allowed to acclimate to
treatment temperatures gradually over a 24-h period. Temperature was controlled by a water bath
surrounding the buckets and held constant (±0.5°C) through the use of heaters or chillers, and
monitored daily using a temperature data logger (Onset Computer Corporation, HOBO Pendant
Logger Model UA-001-08). Water quality in each bucket was conducted bi-weekly for ammonia
(salicylate method, Hach Method 8155), nitrite (diazotization method, Hach Method 8507),
nitrate (cadmium reduction method, Hach Method 8171), dissolved oxygen, pH, and specific
conductivity (YSI Professional Plus Multiparameter Meter). Total hardness (mg of Ca/L as
CaCO3 plus mg of Mg/L as CaCO3) via the titration method (Hach Method 8213) and total
alkalinity (Hach Model AL-AP; mg of phenolphthalein alkalinity/L plus mg total methyl orange
alkalinity/L as CaCO3) were measured on the source water once a week.
Mussels in each bucket were fed 500 mL daily (21 mL/h) of a premixed commercial
algae formula (mean cell concentration , about 1.0–2.0 x 106 μm
3/mL) delivered continuously
from a 1-L water bottle through a drip valve. Eighty water samples were taken randomly from
the buckets over the course of the experiment to quantify the algal cell concentrations using a
Coulter counter (Beckman Coulter, Multisizer 3) located at the AWCC. Algal concentrations
also were measured by a hemocytometer and compared to those from the Coulter counter.
Feeding bottles were cleaned, juvenile mussel holding chambers were rinsed, and bucket water
was completely exchanged once a week. Air bubbles were removed from culture chambers and
pumps, and power sources and water levels were checked daily, as per the FMCC protocols.
Testing of E. brevidens and E. capsaeformis juveniles began 18 November 2011 and
finished 4 April 2012. Testing of L. fasciola juveniles began 22 November 2011 and finished 12
167
April 2012. Mussels in buckets were sampled at 2-week intervals for 20 weeks to provide a total
of 10 sampling events. Random samples of 10 of the 40 juveniles in each chamber were
measured under a microscope (Olympus American, Model SZ40) to assess mean growth (i.e.,
mean length at time t minus initial length). All live individuals and shells within a chamber were
counted to assess survival rates since the start of experiment. Shells of dead mussels were
removed and documented. A summary of test conditions is given in Table 2.
Experimental Design and Statistical Analyses
Five temperature treatments were tested (20, 22, 24, 26, and 28°C), covering the range of
normal (20–24°C) to upper (26–28°C) temperatures that mussels experience in the wild during
the warmer months of the annual growth period in the Clinch River. The test was conducted in
recirculating downweller bucket aquaculture systems, in which each bucket was independent of
others and served as one experimental unit (EU) (Barnhart 2006).
Following a power analysis that determined the appropriate sample size needed to
achieve a minimum of 80% power at α=0.05, each temperature treatment was assigned five
independent downweller buckets that served as replicates. Each bucket contained a total of six
juvenile mussel holding chambers. The three species were tested alongside one another within
buckets, with a single chamber containing juveniles of only one species, while the other three
chambers remained unoccupied (Figure 1). Juveniles of each species were randomly separated
into chambers containing 40 individuals and placed in 1 of the 25 buckets. The EUs were
randomly assigned to one of the five different treatment temperatures. A summary of the
experimental design is given in Table 2.
168
Growth and survival of juveniles for each species were compared among temperature
treatments using a mixed-model analysis with repeated measures. Growth and survival of
juveniles within EUs were treated as random effects. Temperature treatment, time, and
temperature x time interaction were treated as fixed effects. Treatment means were estimated and
compared at each sampling event for further analyses. Survival data (proportion survival) for
each species was arcsine(x)-transformed before being compared among temperature treatments
in order to meet the normality assumption. Additionally, a one-way analysis of variance
(ANOVA) was used to determine whether algal concentrations differed among treatments.
Analyses were conducted using SAS software (SAS institute, Inc., Cary, North Carolina, version
9.2) and were considered significant at the ɑ≤0.05 level. Unless otherwise stated, all significant
results were p<0.01.
RESULTS
Epioblasma brevidens
For the five temperature treatment conditions tested in this study, final growth at 138
days for E. brevidens juveniles ranged from 0.75 to 4.23 mm, at 20°C and 26°C, respectively
(Table 3; Figure 2). Analysis of simple effects (i.e., separating the data by sampling events and
conducting one-way ANOVAs at each time-step) revealed significant differences in growth
between temperature treatments at each of the 10 sampling events. Results of the mixed model
analysis for growth indicated that the fixed effects of temperature, time, and temperature x time
interaction were all significant (Appendix A).
Contrasts of differences in treatment means (effect size ± SE) for final growth (i.e., mean
shell length at final sample minus initial mean shell length) revealed that growth at 20°C was
169
significantly lower than that at all of the other treatment temperatures. Growth at 22°C was
significantly lower than growth at 24, 26, and 28°C, and growth at 24°C was significantly lower
than growth at 26°C and 28°C. No significant difference in juvenile growth was detected
between 26°C and 28°C (p=0.36) (Appendix A).
Epioblasma brevidens juvenile survival ranged from 82.5 to 93.5% (Table 3; Figure 3).
Examination of simple effects for temperature treatments at individual sampling events showed
some significance (p=0.05) of temperature on survival at the fourth (day 54) sampling event;
however, survival was not affected by treatment temperature at any other sampling event.
Survival was not affected by temperature (p=0.13), while the effects of time and temperature x
time interaction were significant. Contrasts of differences in treatment means for final survival
showed that survival was significantly lower at 20°C than at 24°C (p=0.05) and 28°C (p=0.03).
The remaining final survival estimates were not significantly different between other treatment
temperatures (Appendix A).
Epioblasma capsaeformis
Final growth increment at 138 days for E. capsaeformis juveniles ranged from 1.35 to
4.90 mm across all temperature treatments (Table 3; Figure 2). Analysis of simple effects of
temperature at each sampling event revealed significant differences in growth between
temperature treatments at each of the 10 sampling events. Mixed-model analysis for growth
indicated that the effects of temperature, time, and temperature x time interaction were all
significant (Appendix A).
Contrasts of differences in treatment means for final growth revealed that growth at 20°C
was significantly lower than growth at all other treatment temperatures. Significantly lower
170
growth also was observed for the 22°C and 24°C treatments when compared to growth at the
26°C and 28°C treatments. Growth was similar between 22°C and 24°C (p=0.63), and between
26°C and 28°C (p=0.25) (Appendix A).
Substantial mortality was observed in one of the EUs of the 28°C treatment during the
fourth sampling event (day 54), causing it to be a significant outlier for the survival analysis and
violating the homogeneity of variance assumption. It is unlikely that the mortality observed was
caused by temperature because: 1) no other EU within the treatment experienced similar
mortality, 2) mortality occurred in a single early sampling event, and 3) mortality ceased in this
bucket for all further sampling events (sample events 5–10). It is possible that this single
mortality event may have been induced by human error during sampling efforts (e.g., handling
stress). Data from this outlier unit were removed from the mixed-model analysis of survival from
the fourth sampling event forward (day 54 to 138) to reduce model variance and thereby to meet
the assumption of homogeneity of variance. Final survival of E. capsaeformis juveniles ranged
from 68.1 to 83.5% (Table 3; Figure 3). Analysis of simple effects for temperatures by time
under this model revealed no significant differences in survival between any temperature
treatments. Time was significant, whereas the effects of temperature (p=0.16) and temperature x
time interaction (p=0.71) were not significant. Contrasts of differences in treatment means for
final survival uncovered significantly lower survival at 22°C than at 28°C (p=0.03). Final
survival contrasts between all other temperature treatments were not significant (Appendix A).
Lampsilis fasciola
Final growth at 141 days for L. fasciola juveniles ranged from 2.09 to 5.13 mm (Table 3;
Figure 2). Analysis of simple effects for temperatures by time under this model revealed
171
significant differences in growth between temperature treatments at each sampling event. All
fixed effects for growth were significant (Appendix A).
Contrasts of differences in treatment means for final growth revealed that growth at 20°C
was significantly lower than growth at all other temperature treatments. Significantly lower
growth was observed at 22°C than at 24, 26 and 28°C. Growth at 24°C did not differ statistically
from growth at 26°C (p=0.44) and 28°C (p=0.51). No significant differences in growth were
detected between 26°C and 28°C (p=0.15) (Appendix A).
Lampsilis fasciola juvenile final survival ranged from 75.0 to 89.5% (Table 3; Figure 3).
Statistically significant differences in survival between some temperature treatments were
detected at the second (day 29, p=0.04) and sixth (day 86, p=0.03) sampling events. No
significant differences in survival between temperature treatments at other sampling events were
revealed by examination of simple effects under this model. Survival was not affected by
temperature (p=0.10), while time and the temperature x time interaction effects were significant.
Contrasts of differences in treatment means for final survival revealed significantly lower
survival at 20°C than at 22°C (p=0.04) and 28°C (p=0.02). Final survival means were not
significantly different for any of the other temperature treatment comparisons (Appendix A).
Algal Concentrations and Water Quality
Algal cell concentrations within buckets ranged from 1.54 to 2.06 x 106 μm
3/mL
(mean=1.80 x 106 μm
3/mL) and did not differ among temperature treatments (p=0.23)
(Appendix A). Temperatures within treatments did not vary greatly from target temperatures
(±0.2°C). Ammonia, nitrite, and nitrate concentrations within buckets stayed within acceptable
levels, and averaged 0.01 mg/L as NH3, 0.005 mg/L as NO2, and 0.2 mg/L as NO3, respectively.
172
Water in buckets had a mean dissolved oxygen concentration of 7.33 mg/L, pH of 8.46, and
specific conductivity of 393 µS/cm. Total hardness and alkalinity of source pond water ranged
from 193.76 to 209.09 mg/L with (mean=201.42 mg/L) as CaCO3, and 174.76 to 193.68 mg/L
(mean=184.22 mg/L) as CaCO3, respectively.
DISCUSSION
Previous experimental and observational studies have examined the direct effects of
numerous factors affecting growth and survival rates of freshwater bivalves in captivity and the
wild. Factors that have been found to correlate with mussel growth and survival rates include,
but are not limited to, substrate type and size (Hinch et al. 1986; Liberty et al. 2007), flows and
sediment load (Beaty 1997; Zimmerman 2003; Jones et al. 2005; Liberty et al. 2007; Rypel et al.
2008), toxicant exposure (Pandolfo et al. 2010b), mussel density (Hanson et al. 1988; Beaty
1997; Beaty and Neves 2004; Negishi and Kayaba 2009), food availability (Hanlon 2000),
sampling frequency (Beaty 1997; Zimmerman 2003; Liberty et al. 2007), maturity of larvae
(Jones et al. 2005), and temperature (Hanson et al. 1988; Buddensiek 1995; Beaty 1997; Hanlon
2000; Zimmerman and Neves 2002; Zimmerman 2003; Liberty 2004; Hanlon and Neves 2006;
Pandolfo et al. 2010a, 2010b; Negishi and Kayaba 2010). Results of these studies have helped
define requirements for mussel propagation and culture by better understanding factors affecting
growth and survival, and have shown that mussels are useful biological indicators of
environmental change. Providing optimal temperatures for laboratory-propagated mussels is
critical for propagation and culture success.
Due to their small size (<10–20 mm), juvenile mussels are difficult to detect in the wild,
restricting field investigations to the adult life stage and making it difficult to examine effects of
173
temperature and other factors on early life stages (Negishi and Kayaba 2010). Although reports
of growth rates of juveniles from field-based studies are uncommon, researchers have begun to
close this knowledge gap by utilizing laboratory-propagated juveniles for experimental studies,
as I have done here. To my knowledge, no other studies have been published which directly
tested the effects of temperature on the growth and survival of older and larger (>4 months, ≥1.5
mm) laboratory-propagated endangered-species juveniles with the goal of determining an
optimal rearing temperature for maximizing culture success.
I found that temperature had a positive correlation with growth of E. brevidens, E.
capsaeformis, and L. fasciola juveniles, which agreed with conclusions from previous studies
regarding the effect of temperature on juvenile mussel growth (Buddensiek 1995; Beaty 1997;
Hanlon 2000; Hanlon et al. 2006). Further, the positive relationship between temperature and
growth and the magnitude of growth varied between juveniles of these three species. These
observed differences in juvenile mussel growth demonstrated that growth among these species
varies in relation to water temperature. In contrast to previous studies, temperature was neither
positively nor negatively associated with survival (Buddensiek 1995; Beaty 1997). The
relationship between temperature and survival of juveniles was less clear within the time-scale
and temperature treatments of this study. Even though survival did not differ statistically over
time between treatments for all three species of juveniles, a few significant treatment
comparisons between final survivals (at sampling event 10) were detected. Generally, it appeared
that lowest survival occurred at 20°C, although some pairwise comparisons were not significant.
Prior to my study, I set a biologically important effect size for final growth between
temperature treatments at 1 mm. For monitoring release and population success, juveniles are
individually tagged in the laboratory with a shellfish tag (Hallprint Inc., Holden Hill, New South
174
Wales, Australia) before being released into the wild. This tagging procedure requires that
individuals be a minimum size of 10 mm because of the size of the tags (8 x 4 mm oval tag size).
Thus, a difference in 1 mm between individuals can influence how soon juveniles can be tagged
in the laboratory and then released to sites selected for population restoration. In addition to size
influencing when juveniles can be released, survival of overwintering juveniles may be directly
correlated with size, significantly improving the survival of individuals when released to the wild
(Buddensiek 1995; Hanlon 2000; Sarrazin and Legendre 2000; Hanlon and Neves 2007; Hua et
al. 2011). Greater size also will increase detection probability during monitoring efforts to detect
released individuals and enhance the overall likelihood of population recovery success (Hua et
al. 2011).
One goal of my study was to determine the optimum temperature for maximizing growth
of juveniles in captivity. I found that maximum growth in shell length after approximately 4.5
months for E. brevidens, E. capsaeformis, and L. fasciola juveniles occurred at 26°C. However,
growth at 26°C did not differ statistically nor biologically (difference < 1.0 mm) from growth at
28°C. Therefore, differences in final survival rates within species were assessed to make
evaluations between these two temperatures.
While E. brevidens and L. fasciola juveniles experienced highest final survival at 28°C,
E. capsaeformis juveniles had the lowest survival at this temperature treatment. It is not clear
whether high mortality at 28°C was due to approaching an upper thermal limit for E.
capsaeformis juveniles, sampling stress, or factors other than temperature. Sampling procedure
involves handling juveniles to obtain shell measurements and to estimate survival data—both
requiring short-term exposure to air—which can cause stress (Liberty et al. 2007). Several
studies have reported lower mortality in juveniles that were sampled less frequently (Beaty 1997;
175
Zimmerman 2003; Liberty et al. 2007). Considering that differences in survival of juveniles over
time were not significant for the three species in my study, perhaps sampling frequency or other
factors contributed to final mortality rather than temperature alone. With no statistical or
biological difference detected between the 26°C and 28°C in growth and survival within species,
and due to the unknown source of additional mortality at 28°C for E. capsaeformis juveniles, I
incorporated conclusions of previous studies on water temperature relationships into our
assessment of optimum rearing temperature for these species.
Water temperature is one of the most important environmental parameters affecting
growth and survival of juvenile mussels in captivity (Zimmerman 2003; Jones et al. 2005;
Pandolfo et al. 2010a, 2010b). Several laboratory experiments have described the effects of
temperature on growth and survival of freshwater bivalves during early life stages (i.e., newly
transformed juveniles and < 1-year old juveniles) (Buddensiek 1995; Beaty 1997; Hanlon 2000;
Zimmerman 2003; Hanlon and Neves 2007; Pandolfo et al. 2010a, 2010b). Buddensiek (1995)
found that growth rates and mortality of freshwater pearl mussel Margaritifera margaritifera
juveniles were positively correlated with temperature. Similarly, Beaty (1997) reported a positive
relationship between temperature and growth and survival of newly transformed rainbow
mussels Villosa iris. Hanlon (2000) also reported a positive relationship between temperature
and growth in juvenile L. fasciola, but showed seasonal variation in survival that suggested that
temperature is negatively associated with mortality. Hanlon (2000) further suggested that the
relationship between temperature and survival is not always clear, and that studies with opposing
results may be due to resource availability at different experimental scales (i.e., streams are less
likely to be food-limiting at higher temperatures, in contrast to a laboratory-scale experiment).
176
Two other studies examined temperature effects on survival during early life stages and
determined acute lethal temperatures (LT50s) for glochidia and laboratory-propagated juveniles.
The aims of these studies were to determine upper thermal limits of early life stages to provide
insight into any effects that rising maximum water temperatures—due to global climate
change—may have on mussel populations. Pandolfo et al. (2010b) reported acute lethal thermal
tolerances of eight species of glochidia and seven species of juveniles ranging in age from <1 to
8 weeks old. They reported that mean LT50s in 96-h tests were 34.7°C for juveniles, and 31.6°C
in 24-h tests for glochidia. Pandolfo et al. (2010b) concluded that the survival of these early life
stages can decline significantly with small increases in temperature. Dimock and Wright (1993)
also reported acute thermal tolerances for 1-week old juveniles of two freshwater mussel species
and reported LT50s between 31.5°C and 33°C. Because my study goal was to determine optimum
production temperatures, my experiment did not cover the upper temperature ranges (i.e., >30°C)
considered in these studies, suggesting why we likely did not observe a clear relationship
between temperature and survival for E. brevidens, E. capsaeformis, and L. fasciola juveniles.
Temperature has a significant effect on aquatic organism growth and survival rates in
hatchery settings due to its influence on physiological processes such as respiration, filtration,
and excretion rates (Zimmerman and Neves 2002; Spooner 2007; Spooner and Vaughn 2008;
Pandolfo et al. 2010a, 2010b; Fitzgibbon and Battaglene 2012). These metabolic activities of
mussels generally increase with higher water temperatures (i.e., within the natural range)
(Hanlon 2000; Spooner and Vaughn 2008; Vaughn et al. 2008). Typically, oxygen and food
resources can become limiting with increasing water temperature (Hanlon 2000). The availability
of dissolved oxygen in a system is negatively related to temperature, and dependent on the water
system (Hastie et al. 2003). The availability of food in a closed system is limited by the amount
177
of supplemental diet dispensed to individuals exhibiting higher feeding rates in systems cultured
at higher temperatures. Therefore, a combination of increased dissolved oxygen demand and
feeding rates with lower availability of these resources at higher temperatures can strongly
influence growth and survival. In addition, total ammonia concentrations have been shown to
increase with increased excretion rates of mussels due to higher temperatures (Spooner and
Vaughn 2008). Although ammonia toxicity (total ammonia) from increased water temperatures
is negligible between 3°C and 30°C for fish in freshwater systems, early life stages of mussels
are more sensitive to total ammonia concentrations than other aquatic organisms (USEPA 1998,
cited by Randall and Tsui 2002; Wang et al. 2007a, 2007b).
In healthy non-degraded streams, juveniles generally do not face limitation issues with
food and oxygen availability or ammonia toxicity—that which would increase mortality—
because of the continuous influx of freshwater and high turnover rate. Conversely, experiments
that are confined to small recirculating aquaculture systems have a higher likelihood of
encountering (if not managed properly) limited food and oxygen or increased ammonia levels at
higher temperatures in comparison to streams because of their lack of a continuous influx of
freshwater (Hanlon 2000). As a consequence, juveniles may experience increased levels of
mortality. Because of the possible occurrence of food and oxygen limitations and (sub)lethal
ammonia levels in small recirculating systems, researchers have been cautious about culturing
juveniles at higher temperatures. These general temperature relationships were taken into
consideration, even though food quantity was not a limiting factor in our experiment, and my
experimental culture systems did not experience any abnormally low dissolved oxygen or high
total ammonia levels.
178
Based on my analyses of final growth and survival, and previously described temperature
relationships, I believe that the optimal rearing temperature for maximum growth and survival in
captivity is around 26°C for E. brevidens, E. capsaeformis, and L. fasciola juveniles. I believe
my findings can be applied by researchers to improve laboratory culture methods for juveniles of
other species of mussels. Present culture temperatures for juvenile mussels are set based on
research manager discretion and source water temperatures, and sometimes overwintering
juveniles in captivity are held below growing temperatures (i.e., <15°C). Researchers also have
been cautious about culturing juveniles, particularly those of endangered species, at temperatures
consistently exceeding 24°C for concern of increased mortality. My results suggested that a
simulated winter season is not necessary for continued mussel growth or survival. However,
because some biologists believe laboratory-propagated mussels need to experience lower
overwintering temperatures in order to be better-adapted to natural conditions upon release,
further investigation is needed to determine whether long-term survival after release is affected
by the absence of a cooling period in captivity.
Determination of an optimum rearing temperature has clear implications for culturing of
laboratory-propagated juveniles, and ultimately for conservation efforts. The culture and release
of laboratory-propagated juveniles has been identified by federal species recovery plans and
other documents as an approach to increasing the viability of existing populations or
reintroducing species within their historical ranges (Williams et al. 1993; Neves et al. 1997;
Neves 1999; USFWS 2003, 2004). Optimizing temperature to maximize growth and survival of
mussels in hatchery settings reduces the length of time juveniles are held in the laboratory,
allowing biologists to grow endangered juveniles to larger sizes more quickly and maximizing
production levels relative to costs. Decreasing holding time is important because it reduces
179
mortality in captivity (i.e., subjects them to less handling stress) and frees up space in hatcheries,
thereby increasing the overall number of individuals produced for population recovery efforts by
resource managers.
Understanding the relationship between temperature and mussel growth and survival
across all life-stages is important for optimizing propagation and culture success—and by
extension, recovery of imperiled species. My findings support previous conclusions that higher
temperatures increase growth rates but neither supported nor contradicted conclusions on the
relationship between temperature and survival. Upper thermal limits (i.e., >50% mortality over
the duration of this experiment) were not observed for juveniles of any species in our study. This
experiment should be repeated with newly-transformed juveniles to determine whether
temperature affects growth and survival differently for younger and smaller juveniles.
Furthermore, additional testing of growth and survival of juveniles within these temperature
ranges (20–28°C) over a larger temporal scale, and at higher temperature ranges (>28°C), is
needed to reveal a the relationship between temperature and survival and to understand and
predict the potential effects of persistent high water temperatures on mussel populations due to
global climate change (Hastie et al. 2003; Pandolfo et al. 2010b).
180
LITERATURE CITED
Barnhart, M.C. 2006. Buckets of muckets: a compact system for rearing juvenile freshwater
mussels. Aquaculture 254:227–233.
Beaty, B.B. 1997. Development of juvenile culture techniques and testing of potential
biomarkers of environmental stress in freshwater mussels (Bivalvia: Unionidae).
Doctoral dissertation, Virginia Polytechnic Institute and State University, Blacksburg.
Beaty, B.B., and R.J. Neves. 2004. Use of a natural river water flow-through culture system
for rearing juvenile freshwater mussels (Bivalvia: Unionidae) and evaluation of the
effects of substrate size, temperature, and stocking density. American Malacological
Bulletin 19:15–23.
Buddensiek, V. 1995. The culture of juvenile freshwater pearl mussels Margaritifera
margaritifera L. in cages: a contribution to conservation programmes and the knowledge
of habitat requirements. Biological Conservation 74:33–40.
Dimock, R.V., and A.H. Wright. 1993. Sensitivity of juvenile freshwater mussels to hypoxic,
thermal, and acid stress. Journal of the Elisha Mitchell Scientific Society 109:183–192.
Eckert, N.L. and M.J. Pinder. 2010. Freshwater mussel survey of Cleveland Island, Clinch
River, Virginia: augmentation monitoring site–2008. Final Report to the Virginia
Department of Game and Inland Fisheries, Bureau of Wildlife Research, Richmond.
Fitzgibbon, Q.P., and S.C. Battaglene. Effect of water temperature on the development and
energetics of early, mid and late-stage phyllosoma larvae of spiny lobster Sagmariasus
verreauxi. Aquaculture 344–349:153–160.
Gosling, E. 2003. Bivalve molluscs: biology, ecology, and culture. Fishing New Books, Osney
Mead, Oxford, United Kingdom.
181
Hanlon, S.D. 2000. Release of juvenile mussels into a fish hatchery raceway: a comparison of
techniques. Master’s Thesis, Virginia Polytechnic Institute and State University,
Blacksburg.
Hanlon, S.D., and R.J. Neves. 2006. Seasonal growth and mortality of juveniles of Lampsilis
fasciola (Bivalvia: Unionidae) released to a fish hatchery raceway. American
Malacological Bulletin 21:45–49.
Hanson, J.M., W.C. Mackay, and E.E. Prepas. 1988. The effects of water depth and density
on the growth of a unionid clam. Freshwater Biology 19:345–355.
Hastie, L.C., M.R. Young, and P.J. Boon. 2000. Growth characteristics of freshwater pearl
mussels, Margaritifera margaritifera (L.). Freshwater Biology 43:243–256.
Hastie, L.C., P.J. Cosgrove, N. Ellis, and M.J. Gaywood. 2003. The threat of climate change
to freshwater pearl mussel populations. Ambio 32:40–46.
Henley, W.F., L.L. Zimmerman, and R.J. Neves. 2001. Design and evaluation of recirculating
water systems for maintenance and propagation of freshwater mussels. North American
Journal of Aquaculture 63:144–155.
Hinch, S.G., R.C. Bailey, and R.H. Green. 1986. Growth of Lampsilis radiata (Bivalvia:
Unionidae) in sand and mud: a reciprocal transplant experiment. Canadian Journal of
Fisheries and Aquatic Sciences 43:548–552.
Hua, D., J. Rogers, J. Jones, and R. Neves. 2011. Propagation, culture, and monitoring of
endangered mussels for population restoration in the Clinch and Powell Rivers,
Tennessee, 2006–2010. Final Report to the Tennessee Wildlife Resources Agency,
Nashville.
Jones J.W., R.A. Mair, and R.J. Neves. 2005. Factors affecting survival and growth of juvenile
182
freshwater mussels (Bivalvia: Unionidae) cultured in recirculating aquaculture systems.
North American Journal of Aquaculture 67:210–220.
Jones, J.W., E.M. Hallerman, and R. J. Neves. 2006. Genetic management guidelines for
captive propagation of freshwater mussels (Unionidae). Journal of Shellfish Research
25:527–535.
Krebs, C.J. 1972. Ecology: the experimental analysis of distribution and abundance. Harper
and Row Publishers, New York, New York.
Liberty, A.J. 2004. An evaluation of the survival and growth of juvenile and adult freshwater
mussels at the Aquatic Wildlife Conservation Center (AWCC), Marion, Virginia.
Master’s thesis, Virginia Polytechnic Institute and State University, Blacksburg.
Liberty, A.J., B.J. Ostby, and R.J. Neves. 2007. Determining a suitable substrate size and
sampling frequency for rearing juvenile rainbow mussels Villosa iris. North American
Journal of Aquaculture 69:44–52.
Negishi J.N., and Y. Kayaba. 2009. Effects of handling and density on the growth of the
unionid mussel Pronodularia japanensis. Journal of the North American Benthological
Society 28:821–831.
Negishi, J.N., and Y. Kayaba. 2010. Size-specific growth patterns and estimated longevity of
the unionid mussel (Pronodularia japanensis). Ecological Research 25:403–411.
Neves, R.J., A.E. Bogan, J.D. Williams, S.A. Ahlstedt, and P.W. Hartfield. 1997. Status of
aquatic mollusks in the southeastern United States: a downward spiral of diversity. Pages
43–85 in G.W. Benz and D.E. Collins, eds. Aquatic fauna in peril: the
Southeastern perspective. Southeast Aquatic Research Institute, Lenz Design and
Communications, Special publication 1, Decatur, Georgia.
183
Neves, R.J. 1999. Conservation and commerce: management of freshwater mussel (Bivalvia:
Unionidae) resources in the United States. Malacologia 41:461–474.
Neves, R.J. 2004. Propagation of endangered freshwater mussels in North America. Journal of
Conchology Special Publication 3:69-80
Pandolfo, T.J., W.G. Cope, and C. Arellano. 2010a. Thermal tolerance of juvenile freshwater
mussels (Unionidae) under the added stress of copper. Environmental Toxicology and
Chemistry 29:691–699.
Pandolfo, T.J., W.G. Cope, C. Arellano, R.B. Bringolf, M.C. Barnhart, and E. Hammer.
2010b. Upper thermal tolerances of early life stages of freshwater mussels. Journal of the
North American Benthological Society 29:959–969.
Randall, D.J., and T.K.N. Tsui. 2002. Ammonia toxicity in fish. Marine Pollution Bulletin
45:17–23.
Rypel, A.L., W.R. Haag, and R.H. Findlay. 2008. Validation of annual growth rings in
freshwater mussel shells using cross dating. Canadian Journal of Fisheries and Aquatic
Sciences 65:2224–2232.
Sarrazin, F., and S. Legendre. 2000. Demographic approach to releasing adults versus
young in reintroductions. Conservation Biology 14:488–500.
Spooner, D.E. 2007. An integrative approach to understanding mussel community structure:
linking biodiversity, environmental context and physiology. Doctoral dissertation,
University of Oklahoma, Norman.
Spooner, D.E., and C.C. Vaughn. 2008. A trait-based approach to species’ roles in stream
ecosystems: climate change, community structure, and material cycling. Oecologia
158:307–317.
184
USEPA (United States Environmental Protection Agency). 1998. Addendum to ‘‘Ambient
water quality criteria for ammonia––1984’’. National Technical Information Service,
Springfield, Virginia.
USFWS (U.S. Fish and Wildlife Service). 2003. Endangered and threatened wildlife and plants;
proposed designation of critical habitat for five endangered mussels in the Tennessee and
Cumberland River basins. Federal Register 68:106(3 June 2003):33234–33282.
USFWS. 2004. Recovery plan for Cumberland elktoe (Alasmidonta atropurpurea), oyster
mussel (Epioblasma capsaeformis), Cumberlandian combshell (Epioblasma brevidens),
purple bean (Villosa perpurpurea), and rough rabbitsfoot (Quadrula cylindrica
strigillata). U.S. Fish and Wildlife Service, Atlanta, Georgia.
Vaughn, C.C., S.J. Nichols, and D.E. Spooner. 2008. Community and foodweb ecology of
freshwater mussels. Journal of the North American Benthological Society 27:409–423.
Wang, N., C.G. Ingersoll, I.E. Greer, D.K. Hardesty, C.D. Ivey, J.L. Kunz, W.G.
Brumbaugh, F.J. Dwyer, A.D. Roberts, T. Augspurger, C.M. Kane, R.J. Neves, and M.C.
Barnhart. 2007a. Chronic toxicity of copper and ammonia to juvenile freshwater
mussels (Unionidae). Environmental Toxicology and Chemistry 26:2048–2056.
Wang, N., C.G. Ingersoll, D.K. Hardesty, C.D. Ivey, J.L. Kunz, T.W. May, F.J. Dwyer, A.D.
Roberts, T. Augspurger, C.M. Kane, R.J. Neves, and M.C. Barnhart. 2007b. Acute
toxicity of copper, ammonia, and chlorine to glochidia and juveniles of freshwater
mussels (Unionidae). Environmental Toxicology and Chemistry 26:2036–2047.
Williams, J.D., M.L. Warren, Jr., K.S. Cummings, J.L. Harris, and R.J. Neves. 1993.
Conservation status of freshwater mussels of the United States and Canada. Fisheries
18:6–22.
185
Yeager, B.L., and C.F. Saylor. 1995. Fish hosts for four species of freshwater mussels
(Pelecypoda: Unionidae) in the upper Tennessee River drainage. American Midland
Naturalist 133:1–6.
Zale, A.V., and R.J. Neves. 1982. Fish hosts of four species of lampsiline mussels (Mollusca:
Unionidae) in Big Moccasin Creek, Virginia. Canadian Journal of Zoology 60:2535–
2542.
Zimmerman, L.L. 2003. Propagation of juvenile freshwater mussels (Bivalvia: Unionidae) and
assessment of habitat suitability for restoration of mussels in the Clinch River, Virginia.
Master’s thesis, Virginia Polytechnic Institute and State University, Blacksburg.
Zimmerman, L L., and R.J. Neves. 2002. Effects of temperature on duration of viability for
glochidia of freshwater mussels (Bivalve: Unionidae). American Malacological Bulletin
17:31–35.
186
Table 1. Summary of gravid female mussel, host-fish collection, and host-fish infestation
methods at the Freshwater Mollusk Conservation Center (FMCC) and Aquatic Wildlife
Conservation Center (AWCC) in 2011 used to produce juveniles in this study. All gravid females
were collected from the lower Clinch River, Tennessee.
Species
Experiment details Cumberlandian
combshell
Oyster mussel Oyster mussel Wavyrayed
lampmussel
Facility AWCC AWCC FMCC AWCC
Mussel collection month June June May July
Mussel holding system 150-L circular
fiberglass tank
150-L circular
fiberglass tank
300-L living
stream
150-L circular
fiberglass tank
Host fish species Black Sculpin Black Sculpin Black Sculpin Largemouth Bass
Fish collection site Middle Fork
Holston, VA
Middle Fork
Holston, VA
South Fork
Holston, VA
Regional Fish
Farm, AR
Fish holding system AHABa AHAB
a Quarantine tank RPS
b
Infestation month June June May July
No. gravid mussels used 1 4 8 4
No. fish used 64 84 180 78
Infestation temperature (°C) 22–24 22–24 21 22–24
Duration of infestation (mins) 60 60 45 60
Infested fish recirculating
aquaculture holding system
AHABa AHAB
a 76-L Tanks
b RPS
c
Days to first excystment 12 12 13 12
No. provided for experiment 1000 500 500 1000 aAHAB=Aquatic Habitats, Inc. Z-Hab System
bA 2,000-L closed recirculating system made up of twenty 76-L tanks and two sumps
cRPS=Recirculating Propagation System
187
Table 2. Experimental items and test conditions for culture temperature tests of E. brevidens, E.
capsaeformis, and L. fasciola juveniles at the FMCC, November 2011–April 2012.
Experimental items Test conditions
Statistical analysis Randomized, repeated measures
Test system Downweller buckets with six chambers
Test duration 20 weeks
Test bucket volume 18 L
Water renewal Every 7 days
Initial age of juveniles E. brevidens: 4.5 months
E. capsaeformis: 5 months
L. fasciola: 4.5 months
Initial size (mm) of juveniles E. brevidens: 2.2 ± 0.03
E. capsaeformis: 1.5 ± 0.03
L. fasciola: 1.8 ± 0.03
Chambers/species/bucket 1
Juveniles/chamber 40
Buckets (replicates)/ treatment 5
Juveniles/treatment 200
Feeding (each bucket/day) 0.05 mL Nanno 3600: 0.15 mL Shellfish Diet 1800: 500 mL
conditioned water
Algal cell concentration (within bucket) Mean range of 1.0 – 2.0 x 106 um
3/mL
Flow Submersible pumpa, maximum flow=590 L/hour
Test water Pond water filtered to <5 µm, mean hardness=200 mg/L as
CaCO3, alkalinity=184 mg/L as CaCO3
Test temperatures 20, 22, 24, 26, or 28°C
Water quality Bi-weekly testing of ammonia, nitrite, nitrate, dissolved
oxygen, pH, and specific conductivity
Sampling interval (days) 14
Endpoints Growth (mean length at time t minus mean initial length) and
survival (proportion survival) aAquarium Systems Mini-Jet Model MN-606
188
Table 3. Final growth and survival (mean ± SE) of E. brevidens (EB), E. capsaeformis (EC), and
L. fasciola (LF) juveniles cultured in one of five temperature treatments. Values followed by
different subscripts are significant (p<0.05); z–w indicate differences in temperatures within a
species, and v–t differences among species within a temperature treatment. The final sampling
event occurred at 138, 138, and 141 days for EB, EC, and LF juveniles, respectively.
Temperature (°C)
Species 20 22 24 26 28
Growth (mm)
EB 0.75 ± 0.04 z v
2.22 ± 0.13 y v
3.27 ± 0.16 x v
4.23 ± 0.16 w v
4.08 ± 0.11 w v
EC 1.35 ± 0.09 z u
3.73 ± 0.10 y u
3.81 ± 0.14 y u
4.90 ± 0.10 x u
4.70 ± 0.22 x u
LF 2.09 ± 0.12 z t
3.96 ± 0.13 y u
4.99 ± 0.09 x t
5.13 ± 0.14 x u
4.87 ± 0.21 x u
Survival (%)
EB 82.50 ± 3.45 z v
89.00 ± 4.37 zy v
91.00 ± 3.76 y v
89.50 ± 1.70 zy v
93.50 ± 1.70 y v
EC 73.00 ± 2.00 zy v
83.50 ± 3.02 y v
78.00 ± 4.96 zy u
78.00 ± 6.02 zy u
68.13 ± 4.25 z u
LF 75.00 ± 8.44 z v
89.50 ± 3.20 y v
87.00 ± 3.98 zy vu
86.50 ± 3.10 zy vu
89.50 ± 2.89 y v
189
Figure 1. Top view of recirculating downweller bucket culture system and chambers.
190
Figure 2. Mean growth versus time for: (a) E. brevidens, (b) E. capsaeformis, and (c) L. fasciola
juveniles cultured in one of five temperature treatments. Growth measurements were taken at 2-
week intervals for 20 weeks to provide a total of 10 sampling events.
191
Figure 3. Mean survival versus time for: (a) E. brevidens, (b) E. capsaeformis, and (c) L. fasciola
juveniles cultured in one of five temperature treatments. Survival was assessed at 2-week
intervals for 20 weeks to provide a total of 10 sampling events.
192
APPENDIX A: Detailed Statistical Results
Table A. 1. Summary of growth (mm) and survival (%) ANOVA of fixed effects for E.
brevidens, E. capsaeformis and L. fasciola.
Species Fixed Effects Num DF Den DF F-value p-value
Growth
E. brevidens Temperature 4 20 97.07 <0.0001
Time 9 95.2 477.37 <0.0001
Temperature x Time 36 95.2 20.83 <0.0001
E. capsaeformis Temperature 4 20.3 138.94 <0.0001
Time 9 101 370.60 <0.0001
Temperature x Time 36 101 11.98 <0.0001
L. fasciola Temperature 4 20.1 67.71 <0.0001
Time 9 94.3 487.61 <0.0001
Temperature x Time 36 94.3 9.78 <0.0001
Survival
E. brevidens Temperature 4 23.9 1.98 0.1302
Time 9 173 12.14 <0.0001
Temperature x Time 36 173 2.25 0.0003
E. capsaeformis Temperature 4 22.2 1.82 0.1604
Time 9 167 11.24 <0.0001
Temperature x Time 36 167 0.85 0.7111
L. fasciola Temperature 4 23.2 2.20 0.1003
Time 9 173 22.91 <0.0001
Temperature x Time 36 173 2.12 0.0008
193
Table A. 2. Summary of growth and survival slicing of the F-test for treatments by sampling
event (time=days since start of experiment) for E. brevidens, E. capsaeformis and L. fasciola.
Species
Time
Growth Survival
F-value p-value F-value p-value
E. brevidens 12 2.74 0.0379
1.18 0.3345
26 20.04 <0.0001
1.38 0.2584
41 32.73 <0.0001
1.85 0.1395
54 35.84 <0.0001
2.69 0.0453
68 63.18 <0.0001
2.20 0.0870
82 81.04 <0.0001
2.31 0.0755
96 88.86 <0.0001
1.88 0.1335
110 102.60 <0.0001
2.30 0.0760
124 112.21 <0.0001
2.26 0.0801
138 151.20 <0.0001
1.65 0.1819
E. capsaeformis 12 4.32 0.0031
1.15 0.3516
26 13.19 <0.0001
1.66 0.1832
41 29.09 <0.0001
0.83 0.5148
54 44.08 <0.0001
1.40 0.2548
68 71.94 <0.0001
1.89 0.1358
82 82.50 <0.0001
1.79 0.1535
96 90.62 <0.0001
1.62 0.1928
110 112.87 <0.0001
1.66 0.1826
124 127.96 <0.0001
1.64 0.1872
138 135.31 <0.0001
1.63 0.1881
L. fasciola 15 2.53 0.0467
2.03 0.1110
29 10.80 <0.0001
2.75 0.0424
43 10.61 <0.0001
2.27 0.0800
58 25.21 <0.0001
1.87 0.1356
72 29.36 <0.0001
2.50 0.0588
86 29.95 <0.0001
3.07 0.0280
100 44.93 <0.0001
1.49 0.2241
114 49.67 <0.0001
1.14 0.3550
128 90.87 <0.0001
1.50 0.2228
141 96.16 <0.0001
1.84 0.1412
194
Table A. 3. Contrasts of differences between treatment means for final growth and survival
estimates of E. brevidens at the last sampling event (day 138) with 95% confidence intervals.
Effect size for growth is in millimeters (mm). Effect size for survival (%) data has been arc-sine
transformed to achieve normality.
Contrast
Effect
Size
Error
t-value
p-value
95% CI
Lower Upper
Growth Contrasts
22°C - 20°C 1.46 0.17 8.80 <0.0001 1.13 1.80
24°C - 20°C 2.51 0.17 15.12 <0.0001 2.18 2.85
26°C - 20°C 3.48 0.17 20.90 <0.0001 3.14 3.81
28°C - 20°C 3.32 0.17 19.98 <0.0001 2.99 3.66
24°C - 22°C 1.05 0.17 6.32 <0.0001 0.72 1.38
26°C - 22°C 2.01 0.17 12.11 <0.0001 1.68 2.35
28°C - 22°C 1.86 0.17 11.19 <0.0001 1.53 2.19
26°C - 24°C 0.96 0.17 5.79 <0.0001 0.63 1.30
28°C - 24°C 0.81 0.17 4.87 <0.0001 0.48 1.14
28°C - 26°C -0.15 0.17 -0.92 0.3614 -0.49 0.18
Survival Contrasts
22°C - 20°C 0.11 0.08 1.37 0.1791 -0.05 0.26
24°C - 20°C 0.16 0.08 2.07 0.0452 0.00 0.32
26°C - 20°C 0.08 0.08 1.05 0.3015 -0.08 0.24
28°C - 20°C 0.18 0.08 2.28 0.0283 0.02 0.33
24°C - 22°C 0.05 0.08 0.70 0.4867 -0.10 0.21
26°C - 22°C -0.02 0.08 -0.32 0.7498 -0.18 0.13
28°C - 22°C 0.07 0.08 0.91 0.3682 -0.09 0.23
26°C - 24°C -0.08 0.08 -1.02 0.3125 -0.24 0.08
28°C - 24°C 0.02 0.08 0.21 0.8360 -0.14 0.17
28°C - 26°C 0.10 0.08 1.23 0.2255 -0.06 0.25
195
Table A. 4. Contrasts of differences between treatment means of final growth and survival
estimates of E. capsaeformis at the last sampling event (day 138) with 95% confidence intervals.
Effect size for growth is in millimeters (mm). Effect size for survival (%) data has been arc-sine
transformed.
Contrast
Effect
Size
Error
t-value
p-value
95% CI
Lower Upper
Growth Contrasts
22°C - 20°C 2.38 0.17 13.87 <0.0001 2.04 2.72
24°C - 20°C 2.46 0.17 14.35 <0.0001 2.12 2.81
26°C - 20°C 3.55 0.17 20.67 <0.0001 3.21 3.89
28°C - 20°C 3.35 0.17 19.52 <0.0001 3.01 3.69
24°C - 22°C 0.08 0.17 0.48 0.6292 -0.26 0.42
26°C - 22°C 1.17 0.17 6.81 <0.0001 0.83 1.51
28°C - 22°C 0.97 0.17 5.65 <0.0001 0.63 1.31
26°C - 24°C 1.09 0.17 6.32 <0.0001 0.74 1.43
28°C - 24°C 0.89 0.17 5.16 <0.0001 0.55 1.23
28°C - 26°C -0.20 0.17 -1.16 0.2503 -0.54 0.14
Survival Contrasts
22°C - 20°C 0.13 0.07 1.83 0.0762 -0.01 0.28
24°C - 20°C 0.07 0.07 0.92 0.3646 -0.08 0.22
26°C - 20°C 0.01 0.07 0.09 0.9293 -0.14 0.16
28°C - 20°C -0.04 0.08 -0.51 0.6163 -0.19 0.12
24°C - 22°C -0.07 0.07 -0.91 0.3686 -0.22 0.08
26°C - 22°C -0.13 0.07 -1.74 0.0909 -0.28 0.02
28°C - 22°C -0.17 0.08 -2.27 0.0296 -0.33 -0.02
26°C - 24°C -0.06 0.07 -0.83 0.4125 -0.21 0.09
28°C - 24°C -0.11 0.08 -1.39 0.1733 -0.26 0.05
28°C - 26°C -0.04 0.08 -0.59 0.5580 -0.20 0.11
196
Table A. 5. Contrasts of differences between treatment means for final growth and survival
estimates of L. fasciola at the last sampling event (day 141) with 95% confidence intervals.
Effect size for growth is in millimeters (mm). Effect size for survival (%) data has been arc-sine
transformed.
Contrast Effect
Size
Error t-value p-value 95% CI
Lower Upper
Growth Contrasts
22°C - 20°C 1.88 0.18 10.24 <0.0001 1.51 2.24
24°C - 20°C 2.90 0.18 15.84 <0.0001 2.54 3.27
26°C - 20°C 3.05 0.18 16.62 <0.0001 2.68 3.41
28°C - 20°C 2.78 0.18 15.17 <0.0001 2.42 3.14
24°C - 22°C 1.03 0.18 5.60 <0.0001 0.66 1.39
26°C - 22°C 1.17 0.18 6.38 <0.0001 0.81 1.53
28°C - 22°C 0.90 0.18 4.93 <0.0001 0.54 1.27
26°C - 24°C 0.14 0.18 0.78 0.4356 -0.22 0.51
28°C - 24°C -0.12 0.18 -0.67 0.5050 -0.49 0.24
28°C - 26°C -0.27 0.18 -1.45 0.1501 -0.63 0.10
Survival Contrasts
22°C - 20°C 0.18 0.08 2.10 0.0428 0.01 0.35
24°C - 20°C 0.14 0.08 1.68 0.1023 -0.03 0.31
26°C - 20°C 0.08 0.08 0.91 0.3675 -0.09 0.25
28°C - 20°C 0.20 0.08 2.36 0.0235 0.03 0.37
24°C - 22°C -0.04 0.08 -0.42 0.6752 -0.21 0.14
26°C - 22°C -0.10 0.08 -1.19 0.2435 -0.27 0.07
28°C - 22°C 0.02 0.08 0.27 0.7917 -0.15 0.19
26°C - 24°C -0.06 0.08 -0.76 0.4504 -0.24 0.11
28°C - 24°C 0.06 0.08 0.69 0.4955 -0.11 0.23
28°C - 26°C 0.12 0.08 1.45 0.1551 -0.05 0.29
197
Table A. 6. Analysis of variance summary for algae concentrations (µm3/mL) within buckets
(EUs) among treatment temperatures.
Source DF
Sum of
Squares Mean Square F-value p-value
Model 4 7.54 x 1012
1.89 x 1012
1.43 0.232
Error 75 98.82 x 1012
1.32 x 1012
Corrected Total 79 106.36 x 1012
198
APPENDIX B: Species Comparisons within Temperature Treatments
At the 20, 22, and 24°C experimental temperatures, growth differed significantly among
the three species (Table 1). Final mean growth for L. fasciola was significantly larger than E.
brevidens at 20, 22, and 24°C, and was larger than those for E. capsaeformis at 20 and 24°C.
Epioblasma capsaeformis growth was significantly greater than that for E. brevidens at the 20,
22 and 24°C temperatures. Final growth among the three species ranged from 0.63–2.43 mm
with a mean of 1.40 mm at 20°C, 1.86–4.33 mm with a mean of 3.30 mm at 22°C, and 2.83–5.24
mm with a mean of 4.02 mm at 24°C, respectively (Figure 1).
Growth differed significantly among some of the comparisons among the three species at
the 26 and 28°C experimental temperatures (Table 1). Final mean growth for L. fasciola and E.
capsaeformis were significantly larger than those for E. brevidens at 26 and 28°C. Mean growth
for L. fasciola did not differ from that of E. capsaeformis at 26 or 28°C. Final growths among the
three species ranged from 3.79–5.52 mm with a mean of 4.75 mm at 26°C, and 3.41–5.68 mm
with a mean of 4.55 mm at 28°C (Figure 1).
Survival of all species was similar among the 20, 22, and 24°C experimental
temperatures, whereas survival of some species differed significantly at 26 and 28°C (Table 2).
Final survival of E. brevidens was significantly greater than E. capsaeformis at 24, 26, and 28°C.
Survival of L. fasciola was significantly greater than that of E. capsaeformis at 28°C. No
significant differences in survival were found between E. brevidens and L. fasciola at 26 and
28°C, and between L. fasciola and E. capsaeformis at 26°C. Final survival ranged from 51.6–
98.4% with a mean of 76.8% at 20°C, 75.1–100% with a mean of 87.3% at 22°C, 64.2–100%
with a mean of 85.3% at 24°C, 61.3–95.1% with a mean of 84.7% at 26°C, and 54.6–98.2% with
a mean of 83.7% at 28°C (Figure 2).
199
Table B. 1. Comparing E. brevidens, E. capsaeformis and L. fasciola growth (mm) within
temperature treatments. Summary of fixed effects for 20, 22, 24, 26, and 28°C.
Treatment Fixed Effects Num DF Den DF F-value p-value
20°C Species 2 12 18.85 0.0002
Time 9 61.3 129.61 <0.0001
Species x Time 18 61.3 8.82 <0.0001
22°C Species 2 12.7 82.58 <0.0001
Time 9 50.5 187.48 <0.0001
Species x Time 18 50.5 6.83 <0.0001
24°C Species 2 12 28.22 <0.0001
Time 9 58.7 553.16 <0.0001
Species x Time 18 58.7 11.74 <0.0001
26°C Species 2 12 4.58 0.0333
Time 9 58.9 521.98 <0.0001
Species x Time 18 58.9 4.00 <0.0001
28°C Species 2 12.4 6.01 0.0149
Time 9 60.2 159.90 <0.0001
Species x Time 18 60.2 2.49 0.0043
200
Table B. 2. Comparing E. brevidens, E, capsaeformis and L. fasciola survival (%) within
temperature treatments. Summary of fixed effects for 20, 22, 24, 26, and 28°C.
Treatment Fixed Effects Num DF Den DF F-value p-value
20°C Species 2 15.4 3.02 0.0781
Time 9 102 16.24 <0.0001
Species x Time 18 102 2.98 0.0003
22°C Species 2 13.8 1.95 0.1789
Time 9 105 12.10 <.0001
Species x Time 18 105 3.79 <.0001
24°C Species 2 13.1 2.58 0.1136
Time 9 107 8.26 <0.0001
Species x Time 18 107 3.13 0.0001
26°C Species 2 12.5 4.31 0.0377
Time 9 101 7.69 <0.0001
Species x Time 18 101 1.05 0.4177
28°C Species 2 14.6 21.36 <0.0001
Time 9 94.9 4.96 <0.0001
Species x Time 18 94.9 0.81 0.6819
201
Figure B. 1. Comparisons of E. brevidens, E. capsaeformis and L. fasciola growth (mm) at each
of the 5 temperature treatments (20, 22, 24, 26, and 28°C) over 10 sampling events.
202
Figure B. 2. Comparisons of E. brevidens, E. capsaeformis and L. fasciola survival (%) at each
of the 5 temperature treatments (20, 22, 24, 26, and 28°C) over 10 sampling occasions.
Recommended