Benthic biology of the western Canadian continental shelf

Continental Shelf Research, Vol. 11, Nos 8-10, pp. 737-754, 1991. 0278-4343/91 $3.00 + 0.00 Printed in Great Britain. © 1991 Pergamon Press plc

Benthic biology of the western Canadian continental shelf

RALPH O . BRINKHURST*

(Received 22 January 1990; in revised form 11 June 1990; accepted 22 August 1990)

Abstract--Quantitative benthic studies on the British Columbian continental shelf are virtually non-existent. This paper attempts to review the inadequate data available. Sand and mud substrata on the western Canadian continental shelf of British Columbia off Barkley Sound support equally diverse benthic infaunal associations. Of the 282 taxa found in grab samples 142 were unique to this study, 42 were also present at other deep water sites, and 32 were ubiquitous. Organic carbon and nitrogen Values were lower in sandy substrata than in mud. Reduced oxygen values in deep water resulting from a barotropic eddy present during the summer appear to be responsible for significant differences in the faunal associations between cruises in contrast to the community stability observed in other locations.

I N T R O D U C T I O N

Tins paper considers the structure of macrobenthic infaunal communities of the only investigated site on the continental shelf of British Columbia and relates the findings of the single available quantitative study to similar work done in the northeast Pacific. Some of these studies have been documented in technical reports (this study--BIUNKmrRST, 1987; Alice Arm--BmNKnURST et al . , 1987; Hecate Strait--BuRD and BmNKnURST, 1987; and Boundary Bay--BuilD et al . , 1987). Reference will also be made to unpublished data derived from more recent studies of British Columbian fiords and Vancouver Harbour (Burrard Inlet).

The shelf study included a test of the effect on the benthos of an incursion of water with reduced oxygen concentration. Samples were taken on two cruises, 20 April-7 May and 14-17 September, 1981. Outflow water from a side spur of the Juan de Fuca Canyon, ultimately driven by the Fraser River, interacts with coastal circulation to create a barotropic eddy after the spring transition (FREELAND and DENMAN, 1982) that delivers low-oxygen water onto the shelf in summer. The authors state "We would expect such a periodic low oxygen environment to affect the local benthos".

The study also provided a test bed for new statistical procedures used to establish significance within and between clusters based on the benthos data and environmental variables (NEMEC and BmNKrtURST, 1988a,b).

There is little comparable work on quantitative macrobenthic infauna of the shelf in the northeast Pacific. LEVINGS et al. (1983) reviewed studies of the benthos of the Strait of Georgia and contiguous fiords, and showed that the total of three stations at depths greater than 20 m were sampled by grab, one off Nanaimo (402 m) and two in Burrard Inlet (43 and

* Ocean Ecology Laboratory, Institute of Ocean Sciences, P.O. 6000, Sidney, British Columbia, Canada V8L 4B2.

737

738 R . O . B R I N K H U R S T

4 9 ° 0 0 ' -

4 8 ° 4 0 '

4 8 0 2 0

&,. c> , # .US

'.. / --' / ) r ~ , ,,, ,, ' 9 ) " " - ' / / .c~ r

- \

\

0 10 20 30 eC2 I i I i

Kilometres ec4

q. - . . . . . . . .

f C a p e O ~ / Flattery ~L I

1 2 8 0 0 0 , 1 2 5 ° 0 0 '

Fig. 1. Station locations for the shelf study. Depth contours at 100 and 200 m.

70 m). More intensive work was undertaken in channels (ELLIS , 1971) but most studies are concerned with epilithic communities at depth, the littoral zone, and estuaries. LEVINGS et al. (1985) studied the effects of dredged material disposal in Alberni Inlet. The largest body of work is that by Lie which is reported in a series of publications from 1968 to 1974. The study areas were mostly in Puget Sound, with some work off the Washington coast. Specimen identifications were limited to bivalves, echinoderms, and crustaceans. Samples were taken with a Van Veen grab and are not comparable with those from British Columbia.

S T U D Y S I T E A N D M E T H O D S

The sample grid for the shelf study is shown in Fig. 1. Samples were taken in the general area of the spur canyon (see particularly C1-C2, Fig. 1) in which low oxygen water was drawn up onto the shelf. An attempt was made to minimize the variability of sediment types sampled. Samples were obtained with a Smith-McIntyre grab that samples a surface area of 0.27 m 2 (Fig. 2). Grabs less than 75 % full were rejected. Subsamples were obtained for sediment chemistry. Then the bulk sample was washed through a sluice into a collecting bucket. Two samples were taken at each site.

Samples for particle size analysis were divided by wet sieving through a 63/~m mesh. The fine fractions were analysed with a Model 5000D Micrometrics sedigraph and the coarse fractions with a 2 m long settling tube. For details of the hardware and software techniques see BR1NKHURST (1987).

Samples for per cent total carbon and nitrogen analysis were obtained with a stainless steel trowel and were then frozen. They were subsequently oven-dried at 105°C for 24-26

h, then pulverised. Small aliquots were analysed with a Carlo-Erber model 1106 elemental analyser.

Benthic biology of the western Canadian continental shelf 739

Oxygen content of the water column was determined using the Winkler technique (STRICKLAND and PARSONS, 1972). Full data sets of these analyses are available in H~LL et

al. (1983). A series of cores taken for the shelf study were sampled within the large grab (75 % of the

total sample, Fig. 2). These were passed through 0.25 mm mesh screens. Sediment remaining in the grab after subsampling was screened through 1 mm mesh screens. Results from the cores and the residue were then combined. Studies in shallow water areas (Burrard Inlet, Boundary Bay) employed fine screens (0.3 and 0.5 mm, respectively). In all studies samples were washed within i and 2 h of acquisition. Specimens once screened were washed into plastic containers and fixed with buffered formalin containing Rose Bengal. Sorting was done in the laboratory using stereo microscopes. Organisms were separated into major taxonomic categories and identified by experts (see Acknowledge- ments). Biomass data showed too much variance to use. A OC process was applied. Specimens are deposited in the National Museum of Natural Science, Ottawa.

Data matrices were prepared with the relative abundance of the various taxa expressed as numbers per m 2 of surface. Rare taxa were included within higher taxonomic units. The matrices were analysed by cluster analysis in Q mode to determine groups of similar stations using the complement of the Bray-Curtis coefficient with weighted pair group mean averaging (0 indicates total similarity, 1.0 total dissimilarity). The significance levels associated with the derived clusters were tested using Sigtree. The null hypothesis for Sigtree is that any two clusters are sufficiently alike that they can be considered to represent a common community. Rejection of the hypothesis at the 0.5 level (5%) indicates the alternative interpretation; that the two clusters represent different communi-

1 o

ul

O~

1 I 5 4 . 5 cm I

L e g e n d :

in fauna l c o r e s , s c r e e n e d at 0 . 2 5 m m mesh

~ " m a c r o ' s a m p l e s , s c r e e n e d at 1 . 0 m m mesh

(~ a r e a s used f o r s e d i m e n t c h e m i s t r y a n a l y s i s

Fig. 2. Spatial arrangement of samples taken from each 0.27 m 2 Smith-Mclntyre grab. Once small samples were removed for sediment chemistry analysis, the rest of the material in those areas

was screened (1.0 mm mesh) for benthos.

740 R . O . BRINKHURST

ties. Comparisons between matrices of community and between them and others based on environmental data were made using Comtrel and Comtre2 (NE~tEc and I~r_qr0RST, 1988a,b). Comtrel evaluates benthic data against constants such as depth or location. The null hypothesis is that there is no agreement between :the two clusters except that which is purely coincidental. Comtre2 is used to compare two benthic data sets or a data set with a predetermined matrix. Here the null hypothesis is the opposite; that is, the two clusters are from identiCal populations.

RESULTS

Particle size characteristics

Sediment type is illustrated in Shepard ternary diagrams (Fig. 3), Sediments at all D stationsplus C4 are composed of sand whereas other stations are composed of silt or clay.

Sediment chemistry

The variance of carbon from duplicates was 7% whereas the between grab variance was 74%. Nitrogen values were however extremely variable (BluNr~uRs'r, 1987), The results are expressed as averages in Table 1. Consequently carbon values were used in statistical analyses. These data show, as expected, low carbon and nitrogen values where the seabed is sandy. Samples from C4 have values similar to those for area D.

Other sediment characteristics (chlorophyll a, phaeopigments) varied more between replicates than between stations and so were considered constants throughout the study region.

Oxygen in bottom water

Much of the deeper water had oxygen concentrations below 4 mg L- ~ but only on the second cruise was there an extensivearea with oxygen levels between 2.5 and 1.5 mg L-l (Figs 4 and 5). This area covered stations C1-C2 of the benthic survey.

Benthic fauna--qualitative data

Polychaetes dominated the fauna as a whole with the capiteUid Mediomastus ambiseta as the most numerous organism. This ~ c i e s was found at every station sometimes in

Table 1. Carbon/nitrogen ratios summarized for areas A, B, C and D

Mean C Mean N Cruise Station Samples (+S.D.) (_+S.D.) C/N ratio

1 A 47 2.15 (0.064) I}.296 (0.013) 7.26 1 B 36 1.72 (0.132) 11:341 (0,008) 5.04 1 C 36 1.86 (0.824) (}.296 (0.053) 6.28 1 D 39 0.33 (0.001) 0.057 (0.005) 5.79

2 A 60 2.11 ((}.121) 0.331 (0.051) 6.37 2 B 56 1.94 (0.143) 0.278 (0.005) 6198 2 C 50 1.71 (1A)2) 0.216 (0.015) 7.92 2 D 5i 0.36 (0.009) (t.088 (0.004) 4,09


IO O e / ° A O% D3-A

CRUISE D 2-e Din- A 1

,°°- °"

CLAY

IN N o/_ ~o/_

UISE 2

0%

C L A Y

foo%

Fig. 3. Shepard ternary diagrams for Cruises I and 2.

742 R . O . BRINKHU~ST

4,ooo,- ,, ~ ' ~ / / " l u (aS ~-~",,'~v,"/ IX \ % / . < ~ %~Er"-

/ l l ~ / ) t o ~ " * " ~ J J ~ i\ L/ ~'~_ ear kiey Sound ~ ' ~ m p m \ I\~ 99 ~ ~ V a n c o u v e r

~ i ° ~ ' ~ , ~ ,a,a.d . ,

\ % / t~ #~-/ , / 7

iLW ~JI I ~0 ¢" (" CA4 /..~

) \ % @ ) f ~ l - w , ) ~ ~ " ~ ' L / . g " _ _ J / ; ~ 2 o3, \ +- ,, ~ ] < - " . A s / ' ~ - ~ ~ v

/ "'~ ~ " . , i ) , r ~ ~ . ~ " i - - ~ . i , "

0 10 20 30 *C2 . - - ' ~ I i I i

K i l o m e t r e s i t 4

/ " / Flat ter y'~t

4 8 ° 2 0 ' - ~ " ~ , / - ~ 1 2 6 0 0 0 , 1 2 5 0 0 0 ,

Fig. 4. Bottom water oxygen data, Cruise 1. The area enclosed by the two solid lines had values <4.0 mg L l but >2.5 mg E -i . Data from HiLL et al, (1983).

4 9 ° 0 0 ' -

4 8 ° 4 0 ' -

4 8 ° 2 0

, ~ ~ . . .r , , . Sound p T ~ / \ ~\5 IP~i - - ~ ~ V alns~ ;nU; e V a n c o u v e r

\ <~ f t,, , J - A / / " ~ '~ - - r ~ " ,)A1 \ , v .~ , / / / j

,,, l o i . ' : <<- - - ~ - " . . . . . . . . . . . "<-4 ~ ' % . ,

</ ',, ",.. f} .> " ? ... c q_ _ __ I " . . - ' - . - , . - - "...<--V-.~~ ~ / " - - I "~. ~.><

o ,o ~o ~o . . ~ . . x ~ . . y .,.-.

K i l ome t red "C4 % I : / " ~ - -/- 0 " ' ~ " . . , - / Ca.o~C'--

. . . . . ,' , " [ F l a l l e , y ~ , "'.., t , L/ I - ' ~

1 2 6 ° 0 0 ' 1 2 5 ° 0 0 '

Fig. 5. Bottom water oxygen data, Cruise 2. The outer area enclosed by solid lines as in Fig. 4, the smaller circular area enclosed by a solid line indicates the region in which all the many oxygen values lie between 2.5 and 1,5 mg L -I . The dotted line encloses an area between it and the inner

solid line for which there is no hard data.


quantities of hundreds per grab. Other common species were Galathowenia oculata, Prionospio steenstrupi, Aricidia ramosa and unidentified species of Cirratulidae and Euclymeninae. Several polychaetes were common in the A1-C2 group of stations including Levinsenia gracilis, Cossura soyeri, Euchone incolor, Sternaspis scutata, A ricidia lopezi, Nephtys cornuta and an undescribed Cossura. In the sandy areas the species Spiophanes berkleyorum, Glycera capitata, Sphaerosyllis brandhorsti, Tharyx secondus, Notomastus lineatus and Chaetozone spp. were common.

The mollusc species were less ubiquitous but Axinopsida serricata, Adontorhina cyclia," Yoldia scissurata, Y. thraciaeformis, Macoma elimata and M. carlottensis, as well as the scaphopods Cadulus spp. were present at more than half the stations, usually at densities of less than 200 individuals m -2. Huxleya minuta and A. cyclia were more abundant at D stations than at others. Of the crustaceans, ostracods were abundant everywhere. The following occurred in more than half the samples, usually at densities of less than 120 individuals m-2: the harpacticoids Bradya cf. typica, Typhalamphiacus cf. typhops, an ectinosomatid, the cumaceans Eudorella pacifica and Leucon cf. nasica, the tanaids Cryptocope spp. a leptogonathiid and the amphipods Harpiniopsis spp., Heterophoxus oculatus (A1-C2 samples only) and Ampelisca spp. The cumacean Lamprops serrata was found only at C4-D3.

Only 142 of 314 fully identified species were limited to the shelf (Table 2). Only 32 species were ubiquitous (common to sites 1-4, shown as 1, 2, 3 and 4, Table 3). Another 42 species were common to the shelf and other deep water sites (sites 1, 2 and 3, Table 3), but a surprising number were found on the shelf and also in shallow water sites (27 in 1 and 4, Table 3) or in these places plus Hecate Strait (1, 2 and 4).

Among the echinoderms, the most frequent (although usually less than 40 individuals m -2) were the brittle stars Amphioplus macraspis, A. strongyloplax and Ophiura sarsi, the heart urchin Brisaster latifrons and the holothurian Pentamera pseudocalcigera.

Several species were found throughout the study area regardless of substratum, but others were specific to either silt or sand.

Some non-quantitative information on specimens collected by epibenthic sled or Agassiz trawl is available in the detailed report on this study, as well as all of the species lists and abundance data for the grab sample analyses (BRINKHURST, 1987).

Benthic fauna--quantitative studies

Total abundances of invertebrates collected averaged 3606 individuals m -2 in Cruise 1 and 3492 individuals m-2 in Cruise 2; detailed estimates of abundance and numbers of taxa are presented in Table 2. Cluster analyses and Sigtree/Comtree tests were run on two symetrical matrices using 26 replicate samples and 324 taxa (revised from the 338 taxa reported by BRINKHURST, 1987). The results of the cluster and Sigtree analyses are shown in Figs 6 and 7. There is clearly no difference between station C4-D3 and the rest in regard to either parameter , regardless of the difference in substrate. If a linkage is shown to be inapplicable at a low level of dissimilarity (for instance, the linkage between B 1 and B2 and the A1-A5 group in Fig. 6 at 1%, dissimilarity of 0.425), any linkage at a higher level of dissimilarity with a Sigtree value of greater than 5 % cannot be considered. In this example, the linkage between the whole of the previously mentioned stations and the B3-C2 group, with a value of 7%, must be discounted as the two members of the uppermost subgroup (A1-A5, B1-B2) at a lower level of dissimilarity already belong to significantly different


Table 2. Numbers of identified species unique to grab and trawled samples from the continental shelf and other areas of British Columbia

Survey* Total Polychaeta Crustacea Moltusca Echinodermata Other

I t 142 55 48 29 4 6 1, 25 20 12 3 3 2 - - 1,3 12 5 4 2 I - - 1,4 27 11 8 4 --- 4 I. 2, 3 10 3 1 4 2 - - 1 ,2 ,4 19 13 2 4 -- - - 1, 3, 4 20 13 5 2 . . . . t, 2, 3, 4 32 21 2 8 ! - -

Total 1 (Grabs) 282 133 73 56 l0 10

Total 1 (Sled/trawl) 32 4 15 9 4

Total 2, 3, 4 405 146 103 123 t3 20

Total all sites 719 2823 191 188 27 30

* 1 = Shelf, 2 = Hecate Strait, 3 = several, fiords including Alice Arm, 4 = Boundary Bay and Vancouver Harbour. All data refer to grab samples unless otherwise specified.

tTaxa unique to shelf grab samples. STaxa unique to grab samples from both shelf and Hecate Strait, etc.

Table 3. Species abundance and numbers of species

Abundance per square metre Number of species Number of

station Cruise 1 Cruise 2 Cruise 1 Cruise 2

A1 5360 3554 162 161 A2 4500 2724 140 139 A4 4294 4532 150 169 A5 4132 5720 130 147 BI 4842 3558 169 147 B2 3544 3244 140 136 B3 2554 3500 124 135 C1 1616 3912 1(18 146 C2 1728 2766 102 127 C4 4392 3442 154 170 D1 3666 2940 179 142 D2 3282 3238 159 141 D3 2974 2274 178 119

Average 3606 3492 146 145

Values are averages derived from paired samples of 0.27 m -2 each.


clusters. The statistically significant groupings are shown overlain on the site map in Figs 8, 9. The fact that these clusters form coherent spatial groups already suggests a departure from random clustering.

Several station groups can be identified from the benthic data. One group consists of the D stations, plus C4 in Cruise 1 but not Cruise 2. The linkage between C4 and the D group in Fig. 7 is not significant. This conforms exactly to the distribution of the sandy substrata sampled in terms of both particle size (Fig. 3) and carbon content, where it is obvious that data from C4 are similar to, but not identical with that from the D stations (C5 shown in Fig. 3 was only sampled on Cruise 2 and so the benthic data were not used). There are several clusters of stations with the other substrate type. In Cruise 1, A1-A5 form a homogeneous group but in Cruise 2, A5 leaves the cluster to link with C1. The sediment type sampled at A5 on Cruise 1 was more akin to that sampled at the other A-B stations than it was on Cruise 2 (Fig. 3). In Cruise 2 data there is no agreement in sediment type between A5 and C1 however. The benthic data from Cruise 1 creates two groups among the B 1-C2 stations, but in Cruise 2 data, the A1-4 and B 1-3 groups are homogeneous (at 9 and 13%) but also form a larger homogeneous grouping (at 7%) at a higher level of dissimilarity. Station C1 is associated with A5, as already noted, and C2 is isolated (the linkage of C2 with the whole group A1-C1 in Fig. 7 at 4% requires rejection of the hypothesis that the two groups so linked are homogeneous).

1.0 0.5 O.1 I | I I I I I I | " l

CRUISE 1 F_~29 %

9% i:S%

I I1% ,11% 7%

1!0 I

<1% I7 %

i o ! 7 ~ 0~.s '

t 11%

3%

A1

A4

A2

A5

B1

B2

B3

C1

C2

C4

D1

~27% D2

i 43% o~ ' o!1 03 BRAY-CURTIS DISBIMILARITY INDEX

Fig. 6. Results of cluster analysis of benthic community data, Cruise 1. Percentage values are derived from Sigtree, and indicate the test of the hypothesis that, at any given linkage level, the entities clustered together are derived from the same community, hence any value of <5% (P < 0.05) indicates that there are two distinct clusters that cannot be merged at that linkage level (B1

and B2 do not belong to a cluster with the A stations for example).

746 R . O . BRINKItURST

Fig. 7.

1.0 0.5 0.1 I I I I I I I I I I

CRUISE 2 [ A1 19%

A4

9 ~ 1 1 olo A2 7 %

B1

B2 - - '--127%

4 %

9 %

4 %

C2

C4

D1 ~ 29%

D2 ~!o ' ' o'.r ' o!~ ' o!a ' 05

BRAY-CURTIS DISSIMILARITY INDEX Results of cluster analysis of benthic communi ty data, Cruise 2.

4 9 0 0 0 ~ -

4 8 ° 4 0 ' -

4 8 0 2 0

i , \% o.;?E¢" I I _ i • ~ v . , - ~ ,.S

v 5 . . . . k , . so°n° "l \ \ , ~ i ~ J Vancouver

I , , "~ ~ ". _ ~ I ' - _.. I + - v , t i J)

o io l o 3 o ~ ~ - ~ . , . . / _.--'-" .

' ' ' ' ~ \ ~ "~ r- K i l o m e t r e s ~ . ~ . ~ ' / ~

/ ~ , ~ 1 i

1 2 6 ° 0 0 ' 1 2 5 ° 0 0 '

Fig. 8. Significant clusters identified f rom Fig. 6 overlaid on a m a p to illustrate the value of such a presentat ion in applied biology.


4 9 ° 0 0 ' -

4 8 ° 4 0 '-

4 8 0 2 0

\ / ~ I - . ~ J Vancouver °wwd , . , . . , . .

,.,, oo k .... ,- k_ _ I/d

0 10 20 30 ~ - . - - ' -

' ; . 0 - , , . ' . ' "7 (' Flattery ~L I

1 2 6 o 0 0 ' 1 2 5 ° 0 0 '

Fig. 9. Significant clusters identified from Fig. 7 for comparison with Fig. 8. Note especially changes in relationships of C1 and C2. The change in C4 is not important as it only links with the D

stations at 7% (Fig. 6) and is separated at 3% in Cruise 2 (Fig. 7).

Tests of the difference between the dendrograms based on relative abundance of the taxa suggest that the apparent differences between the two sets of data, as illustrated in Figs 8 and 9, are real. Tests using Comtre2, based on the hypothesis that the two dendrograms are the same, could not be rejected at any linkage level, but testing the alternative hypothesis with Comtre l indicated the two dendrograms are different above linkage 6. The comparisons of linkages involving the position of C4, B3, C1/A5 and C2 yielded probabilities (that the two clusters are not the same) of 0.33-0.38 and so the hypothesis is accepted in each instance (values below 0.05 are required to reject the hypothesis). The stations that show the most change between cruises are A5, B3 and C2. Changes in similarity could be due to shifts in the benthic associations within any of the groups of stations concerned.

Clusters with abiotic factors

Similar clusters can be produced for abiotic factors, and visual !nspection suggests possible comparisons with the benthos data. These were tested for statistical significance using Comtre l . This tests the hypothesis that there is no agreement between the two clusterings except that which is purely coincidental, Comtre2 is now capable of accepting reference trees (based on single variables) in order to test the opposite hypothesis.

Reference trees (clusters) were created based on geographic distance between stations, sediment particle size and sediment carbon content for each cruise (Figs 10-14). These were compared with the abundance data for both cruises. The hypothesis could be rejected (suggesting that the two trees were not based on random clusterings) at the level on the clusters that separated either seven or nine clusters, that is at quite a low level of similarity.

748 R.O. BRINKHURST At these levels, the benthic community data separated A 1 - A 4 (twice including A5 for Cruise 1 data), B1-B2 in Cruise 1 with B3 added for Cruise 2. and D 1 - D 3 in every instance. The other stations usually remained unclustered at this level. The position of the C stations would appear to be affected by some other factor, and we suggest oxygen as a possibility.

The results would seem to suggest: (1) as each station in the benthic data sets (Figs 6 and 7) clusters first with its nearest neighbour in most instances (based on distance between them. Fig. 10), there are continua along gradients within each of the two separate communities which occupy sand and muddysubstrata. (2) there is a significant relationship between the substrate type and the benthic community at the level of separation into sand vs mud substrata, but little agreement between the groupings within the mud substratum station. Particle size and carbon valueswithin the A - C areas are so similar that very slight changes would drastically rearrange the clusters produced, so comparisons of them and the benthos are only meaningful at the level of separation of D stations (usually associated with C4), from the rest in both instances.

1.0 . 9 . 8 .7 . 6 .5 ,4 .3 . 2 1

I I | I I I I ! I , I

| I | I

1.0 ,9 , 8 .7

~ A2 A4

A1

A 5

a2

B 3

C1

C 2

C 4

D1

..... ~ D3 D2

= ! I v I !

• 6 .5 .4 .3 .2 .1

D i s s i m i l a r i t y I n d e x

LATITUDE / LONGITUDE (DISTANCE) CR1 + CR2

Fig. 10. Cluster analysis based on geographic distance between Stations.


1.0 . 9 . 8 .7 . 6 .5 .4 .3 . 2 .1

A5

C1

A1

A4

A2

B3

C2

B1

B2

D1

D2

D3

- ~ - ' 1 C4

R 0

1.0 .9 .8 .7 .6 ,5 .4 ,3 .2 .1

D i s s i m i l a r i t y I n d e x

SEDIMENT PARTICLE SIZE CRUISE 1

Fig. 11. Cluster analysis based on mean particle size data, Cruise 1.

DISCUSSION

The results of this study support the concept that major species associations may be related to the nature of the substrata (the clustering of D stations, the position of C4 and of A5 vs the rest), but not the idea that sandy substrata are likely to support fewer organisms or a lower diversity (Table 2). There is evidence to suggest that the low oxygen conditions known to have been present in the vicinity of critical stations in the time interval between the two cruises may have had an effect on the benthic faunal associations (Figs 4 and 5, and the significant differences between clusters for the two sets of benthic data).

Tests of data from other locations, especially Hecate Strait (BuRD and BRINKHURST,

1987), show considerable stability of benthic communities over time in this area (based on a complete lack of significant differences between the station clusters of the benthic community data from three cruises from June 1985 to January 1986). It seems unlikely that such complex faunal assemblages would change significantly over the short time interval between cruises on the shelf unless there were some causal agent when elsewhere we observe considerable stability. Very large data sets were available for Hecate Strait (300 taxa, 45 or 60 grab samples per cruise). LIE and EVANS (1973) also found considerable long-

750 R . O . BmNKHURST

term stability of species assemblages at four stations in Puget Sound, although species abundances changed.

Clear. statistically meaningful results can be extracted from very large data bases using non-parametric statistical methods. The bootstrap uses the distributional properties of the bootstrap sample to make statistical inferences about the underlying population, whereas our knowledge of the underlying properties of the actual benthic data do not allow too many assumptions to be made. The Sigtree/Comtree package in association with cluster analysis makes possible comparisons between data sets that incorporate not only relative abundance of the fauna and species richness, but also taxonomic identity. Attaching significance values to them reduces the subjectivity of these methods. Other non- parametric techniques can be used to discriminate between various causal factors impli- cated in community change (BuRD et al., 1990), but comparing dusters of biotic and abiotic factors produces diagrams that can be readily appreciated visually,

LIE and KELLEY (1970) used factor analysis to distinguish three communities in Puget Sound. These were described further by LIE and KISKER (1970). Their data matrices excluded some major taxonomic groups, such as the polychaetes, and most scarce species,

1.O , 9 . 8 . 7 , 6 . 5 . 4 ~ . 2 1

L | | I I I I I ! I

B 2

C 2

B 3

A 2

A1

A4 ~

C1

A5

D1

D3

D2

C4

|

1.0

i i t i i ! i l !

,9 ,8 .7 .6 .5 .4 .3 .2 .I

Dissimi lar i ty Index

SEDIMENT PARTICLE SIZE CRUISE 2

Fig. 1:2. Cluster analysis based on mean particle size data. Cruise 2,


1.0 . g . 8 .7 ,6 ,5 .4 .3 ,2 ,1

I I I I I I I I i I

A2

A 4

C2

~ A1 D)

B2

B!

D1

D2

D3

! s I t , I I ' It I

1,0 .9 .8 .7 .6 .5 .4 .3 .2 * l

D i s s i m i l a r i t y I ndex

CARBON CONTENT CRUISE 1

Fig. 13. Cluster analysis based on sediment carbon content, Cruise 1.

and are therefore not fully comparable with IOS studies. Several of the dominant species in the deep water mud-bottom community identified at the A, B and C1-C2 sites on the British Columbia shelf were also abundant in Puget Sound. There were some similarities between the C4 and D area community and the intermediate sand bottom community, but the shallow-water sand-bottom community of LIE and KELLEY was not included within the sampling scheme of the IOS work. Again, comparisons including relative abundance rather than presence/absence have yet to be made.

ELLIS (1971) suggested similarities between the communities in Satellite Channel, British Columbia and the Puget Sound and the Amphiodia-Maldane-Ophiura communities of TrIo~sor~ (1957), but most of this work on parallel communities is based on intuitive pattern recognition rather than statistical evaluation, and differences in screen sizes and degree of taxonomic resolution make rigorous comparisons difficult.

BURD and Blm~ra-IUaST (1984) described the effect of seasonal changes in the depth of the anoxic layer in Saanich Inlet, British Columbia, and the fate of the organisms trapped beneath the upwelled anoxic water in the fall of 1980. The lowest layers of this fiord are usually anoxic, with no resident macrofauna. This is due to the combination of sill height at


1.0 . 9 . 8 ,7 . 6 *5 .4

I I ! I I I I

Fig. 14.

i | i | i i

1.0 ,9 .8 7 .6 ,5

,3 . 2 .1

I I I

C 4

B1

A 2

B 2

8 3

C1

A5

C 2

D1

D 3

D 2

4 . 3 , 2 .1

D i s s i m i l a r i t y Index

CARBON CONTENT CRUISE 2

Cluster analysis based on sediment carbon content. Cruise 2.

the mouth of the fiord, the volume of freshwater discharge within the sill and the seasonal oceanographic circulation outside. These same factors create intermittent oxygen deficits in Howe Sound. These may be drastic enough to kill slow moving fish species and all the sedentary benthic organisms (LEXqNGS. 1980). reductions going down to 0:5 mg L- 1 on the shelf. The prolonged deoxygenation event of 1977 reduced the benthic invertebrate fauna collected by trawls to zero in October 1977. but the somewhat quickerreduction of oxygen m 1978 than in 1979 (about 7 months rather than 12 to go from 2 mg 02 L -1 to less than 0.5 mg 02 L -1) caused a much less severe response in the fauna, which had rapidlyreturned to pre-1977 values in terms of mean numbers per tow by May 1978. After deep water renewal, oxygen values at nearly all depths were greater than 6 mg L -1. Cluster analyses were used to demonstrate changes in the fauna over time.

Cluster analyses were also used by LEVINGS et al. (1985) in both Q and R mode to demonstrate spatial patterns of both stations and species in the fauna of the ocean dumping site in Atberni Inlet, British Columbia. The results of the spatial analysis were presented in the form of contours enclosing the various clusters identified. This work predated the development of the Sigtree/Comtree analyses and so the cluster groupings were identified


by intuition, but this provided the stimulus for the development of these new techniques. The sampling program in Alberni Inlet involved a Smith-Mclntyre grab and screens with mesh sizes as fine as 0.5 mm. Relative abundance of identified benthic species was used in the cluster analysis using the Czekanowski coefficient for spatial pattern, but presence/ absence data for 44 species were used in the species grouping work. The fauna contained 60 polychaete species, 36 crustaceans, nine molluscs and one ascidian among the 106 taxa found. Numbers at each station ranged from 290 m -2 at the dump site to 2920 m -2 at one of the control sites. Axinopsida serricata was dominant at some stations and common throughout, and several species from the central province were the same as those from the shelf (Aricidea lopezi, Eudorella pacifica, Heterophoxus oculatus and Macoma carlottensis). This group, with some additions such as Nephtys cornuta, made up group A of the species cluster (R mode analysis). Overlaying the cluster pattern onto the site map made it very clear that the dump site formed a community of its own, surrounded by a narrow dump site fringe. These data are currently being entered into the IOS database so that they can be re-analysed by the techniques that they helped to generate, and also that R mode analysis can help relate these species groupings to those found elsewhere in British Columbia.

Acknowledgements--I wish to thank B. J. Burd and D. Moore for their extensive help with the statistics and tabulation of the data. B. Burd, R. Forbes, R. D. Kathman and C. D. Levings all critically reviewed the manuscript. The following identified or verified the taxa for the shelf and other IOS studies in addition to the staff of Dobrocky Seatech Ltd: W. C. Austin, S. Byers, H. Jones, N. S. Jones,. R. D. Kathman, P. Lambert, R. G. B. Reid, E. Ruff, A. H. Scheltema, P. Shaw, R. Shimek, C. Staude, I. Williams, G. Wilson and J. Word. I wish to thank C. Amos and other reviewers for their helpful comments.

R E F E R E N C E S

BPJNKHURST R. O. (1987) Distribution and abundance of macrobenthic infauna from the continental shelf off southwestern Vancouver Island, British Columbia. Canadian Technical Report of Hydrography and Ocean Sciences, 85, 86 pp.

BRINKHURST R. O., B. J. BURD and R. D. KATHMAN (1987) Benthic studies in Alice Arm, B.C. During and following cessation of mine tailings disposal, 1982-1986. Canadian Technical Report of Hydrography and Ocean Sciences, 89, 45 pp.

BURD B. J. and R. O. BRINKHURST (1984) The distribution of the galatheid crab Munida quadrispina (Benedict, 1902) in relation to oxygen concentrations in British Columbia fiords. Journal of Experimental Marine Biology and Ecology, 81, 1-20.

BURD B. J. and R. O. BPaNr~URST (1987) Macrobenthic infauna from Hecate Strait, British Columbia. Canadian Technical Report of Hydrography and Ocean Sciences, 88, 125 pp.

BURD B. J., D. MOORE and R. O. BmNKHURST (1987) Distribution and abundance ofa macrobenthic infauna from Boundary and Mud Bays near the British Columbia/U.S. border. Canadian Technical Report of Hydrogra- phy and Ocean Sciences, 84, 34 pp.

BURD B. J., A. F. L. NEMEC and R. O. BItlNKHUItST (1990) The development and application of analytical methods in benthic marine infaunal studies. Advances in Marine Biology, 26, 169-247.

ELLIS D. V. (1971) A review of marine infaunal community studies in the Strait of Georgia and adjacent inlets. Syesis, 4, 3-9:

FREELAr~D H. J. and K. L. DENMAN (1982) A topographically upwelling center off southern Vancouver Island. Journal of Marine Research, 40, 1069-1093.

HILL S., K. DENMArq, D. MACr, AS, H. SEFTON and R. FORBES (1983) Ocean Ecology data report: coastal waters off southwest Vancouver Island spring and summer 1981. Canadian Data Report of Hydrography and Ocean Sciences, 8, 93 pp.

7 5 4 R . O . BRINKHURS1

LEVlNGS C. D. (1980) Benthic biology of a dissolved oxygen deficiency event in Howe Sound. B.C, In: Fiord oceanography, H. J. FREELAND. D. M. FARMER and C. D. LEVINGS. editors. Plenum Publishing Corpor- ation. New York, pp. 515-522.

LEVlNGS C. D. , R. E. FOREMAN and V. J. TUNNICUFFE (1983) Review of the benthos of the Strait of Georgia and contiguous fiords. Canadian Journal o f Fisheries and Aquatic Sciences. 40, 1120-1141.

LEVINGS C. D . . E, P. ANDERSON and G. W. O'CONr~ELL (1985) Biological effects o f dredged-material disposal in Alberni Inlet. In: Wastes in the ocean. Vol. 6. B. KETCHUM. J. CAPUZZO. W. BURX, I. DUEDALL. P. PARK and D. KESTER. editors. John Wiley, New York.

LIE U. and J. C. KELLEY (1970) Benthic infauna communit ies off the coast of Washington and in Puget Sound: identification and distribution of communit ies. Journal o f the Fisheries Research Board o f Canada, 27.621- 651.

LIE U. and D. S. KISKE~ (1970) Species composition and structure of benthic infauna commumties off the coast of Washington. Journal of the Fisheries Research Board of Canada. 27. 2273-2285.

LIE U. and R. A. EVANS (1973) long-term variability in the structure of subtidal benthic c o m m u n m e s in Puget Sound. Washington. USA. Marine Biology, 21. 122-126.

NEMEC A. F, L. and R. O. BRINKHURST (1988a) Using the bootstrap to assess statistical significance in the cluster analysis of species abundance data. Canadian Journal of Fisheries and Aquatic Sciences. 45. 965-970.

NEMEC A. F. L. and R. O. BR1NKHURST (1988b) The Fowlkes-Mallows statistic and pattern recognition in cluster analysis. Canadian Journal o f Fisheries and Aquatic Sciences. 45. 971-975.

STRICKLAND J. D. H. and T. R. PARSONS (1972) A practical manual of seawater anab, sis. Bulletin of the Fisheries Research Board o f Canada 167.2nd edn.. 310 pp.

THORSON G. ( 19571 Bottom communit ies. Memoir of the Geological Society of America, 67.461-534.

Documents

Benthic biology of the western Canadian continental shelf