34
1 ST. KILDA SOAY SHEEP & MOUSE PROJECTS: ANNUAL REPORT 2011 J.G. Pilkington 1 , S.D. Albon 2 , A. Bento 4 , C. Berenos 1 , T. Black 1 , J. Blount 15 , E. Brown 6 , D. Childs 6 , L. Christensen 14 , T.H. Clutton-Brock 3 , T. Coulson 4 , M.J. Crawley 4 , J. Dorrens 1 , P. Ellis 1 , A. Graham 10 , J. Gratten 9 , A. Hayward 6 , L. Harrington 16 , S. Johnston 6 , P. Korsten 1 , L. Kruuk 1 , T. McNeilly 13 , C. Mitchell 15 , B. Morgan 7 , K. Morriss 10 , M. Morrissey 1 , D. Nussey 1 , M. Page 14 , J.M. Pemberton 1 , S. Robertson 1 , C. Selman 14 , J. Slate 6 , I.R. Stevenson 8 , K. Watt 1 , A. Wilson 1 , K. Wilson 5 , R. Zamoyska 12 . 1 Institute of Evolutionary Biology, University of Edinburgh. 2 Macaulay Institute, Aberdeen. 3 Department of Zoology, University of Cambridge. 4 Department of Biological Sciences, Imperial College. 5 Department of Biological Sciences, Lancaster University. 6 Department of Animal and Plant Sciences, University of Sheffield. 7 Institute of Maths and Statistics, University of Kent at Canterbury. 8 Sunadal Data Solutions, Edinburgh. 9 University of Queensland, Australia. 10 Princeton University, USA. 11 Roslin Institute, University of Edinburgh. 12 Institute of Immunology and Infection Research, University of Edinburgh. 13 Moredun Research Institute, Edinburgh. 14 Institute of Biological and Environmental Sciences, University of Aberdeen. 15 Centre for Ecology & Conservation, College of Life & Environmental Sciences, University of Exeter Cornwall Campus. 16 Université de Montréal, Institute de Recherche en Immunologie et en Cancérologie, Montréal, Canada. POPULATION OVERVIEW .................................................................................................................................... 2 REPORTS ON COMPONENT STUDIES .................................................................................................................... 4 Vegetation ..................................................................................................................................................... 4 Weather effects on female August weight...................................................................................................... 6 Biomarkers of ageing in Soay sheep: oxidative stress and telomere length ................................................. 8 Explaining variation in immune responses in an evolutionary context....................................................... 11 Immunological and parasitological research: an update ........................................................................... 13 Changes in gene frequency: genetic drift or natural selection? ................................................................. 15 The population genetics of Soay sheep coat colour: a model-based approach .......................................... 19 Improving the pedigree of Village Bay sheep using superabundant genetic variation ............................... 22 Genetic architecture of body size in Soay sheep ......................................................................................... 24 Population dynamics and genetic structuring of the St Kilda field mouse.................................................. 27 PUBLICATIONS.................................................................................................................................................. 30 ACKNOWLEDGEMENTS.. ................................................................................................................................... 31 APPENDIX A: PERSONNEL NEWS & SCHEDULE OF WORK ................................................................................ 31 CIRCULATION LIST ........................................................................................................................................... 33

Annual Report 2011 Final - Soay sheepsoaysheep.biology.ed.ac.uk/.../files/dnussey/AnnualReport2011.pdf1Institute of Evolutionary Biology, University of Edinburgh. 2Macaulay ... 6 Department

  • Upload
    others

  • View
    0

  • Download
    0

Embed Size (px)

Citation preview

Page 1: Annual Report 2011 Final - Soay sheepsoaysheep.biology.ed.ac.uk/.../files/dnussey/AnnualReport2011.pdf1Institute of Evolutionary Biology, University of Edinburgh. 2Macaulay ... 6 Department

1

ST. KILDA SOAY SHEEP & MOUSE PROJECTS:

ANNUAL REPORT 2011

J.G. Pilkington1, S.D. Albon

2, A. Bento

4, C. Berenos

1, T. Black

1, J.

Blount15

, E. Brown6, D. Childs

6, L. Christensen

14, T.H. Clutton-Brock

3,

T. Coulson4, M.J. Crawley

4, J. Dorrens

1, P. Ellis

1, A. Graham

10, J.

Gratten9, A. Hayward

6, L. Harrington

16, S. Johnston

6, P. Korsten

1, L.

Kruuk1, T. McNeilly

13, C. Mitchell

15, B. Morgan

7, K. Morriss

10, M.

Morrissey1, D. Nussey

1, M. Page

14, J.M. Pemberton

1, S. Robertson

1, C.

Selman14

, J. Slate6, I.R. Stevenson

8, K. Watt

1, A. Wilson

1, K. Wilson

5,

R. Zamoyska12

.

1Institute of Evolutionary Biology, University of Edinburgh.

2Macaulay Institute, Aberdeen.

3Department of Zoology, University of Cambridge.

4Department of Biological Sciences, Imperial College.

5Department of Biological Sciences, Lancaster University.

6Department of Animal and Plant Sciences, University of Sheffield.

7Institute of Maths and Statistics, University of Kent at Canterbury.

8Sunadal Data Solutions, Edinburgh.

9University of Queensland, Australia.

10Princeton University, USA.

11Roslin Institute, University of Edinburgh.

12Institute of Immunology and Infection Research, University of Edinburgh.

13Moredun Research Institute, Edinburgh.

14Institute of Biological and Environmental Sciences, University of Aberdeen.

15Centre for Ecology & Conservation, College of Life & Environmental

Sciences, University of Exeter Cornwall Campus. 16

Université de Montréal, Institute de Recherche en Immunologie et en

Cancérologie, Montréal, Canada.

POPULATION OVERVIEW .................................................................................................................................... 2

REPORTS ON COMPONENT STUDIES .................................................................................................................... 4

Vegetation ..................................................................................................................................................... 4

Weather effects on female August weight...................................................................................................... 6

Biomarkers of ageing in Soay sheep: oxidative stress and telomere length ................................................. 8

Explaining variation in immune responses in an evolutionary context....................................................... 11

Immunological and parasitological research: an update ........................................................................... 13

Changes in gene frequency: genetic drift or natural selection? ................................................................. 15

The population genetics of Soay sheep coat colour: a model-based approach .......................................... 19

Improving the pedigree of Village Bay sheep using superabundant genetic variation............................... 22

Genetic architecture of body size in Soay sheep ......................................................................................... 24

Population dynamics and genetic structuring of the St Kilda field mouse.................................................. 27

PUBLICATIONS.................................................................................................................................................. 30

ACKNOWLEDGEMENTS.. ................................................................................................................................... 31

APPENDIX A: PERSONNEL NEWS & SCHEDULE OF WORK ................................................................................ 31

CIRCULATION LIST ........................................................................................................................................... 33

Page 2: Annual Report 2011 Final - Soay sheepsoaysheep.biology.ed.ac.uk/.../files/dnussey/AnnualReport2011.pdf1Institute of Evolutionary Biology, University of Edinburgh. 2Macaulay ... 6 Department

2

PO P U L A T I O N OV E R V I E W

The sheep population on Hirta entered 2011 at a very high level and, as a result, there

was some mortality but not a population crash. 144 tagged sheep were found dead

within the study area between March and May of 2011. Lambing began on the 10th

of

April, much later than any previous year, with 78% of lambs born surviving (Fig. 1).

0

5

10

15

20

25

30

10/

04/2

011

11/ 0

4/201

1

12/ 0

4/201

1

13/04/2011

14/04/2011

15/04/201

1

16/

04/2

011

17/

04/2

011

18/04/2011

19/04/2011

20/04/201

1

21/04/201

1

22/04/201

1

23/ 0

4/2011

24/ 0

4/2011

25/04

/201

1

26/04

/201

1

27/04/201

1

28/ 0

4/20

11

29/ 0

4/20

11

30/04

/2011

01/05

/201

1

02/05

/201

1

Date

Lam

bs

bo

rn

Figure 1. The temporal distribution of lamb births during 2011.

In December 2011, 887 tagged sheep were believed to be alive on Hirta, of which 649

regularly used the study area, a total decrease of 3.4% using the study area since the

previous year. The age distribution of the population is shown in Fig. 2 and changes

in sheep numbers in the study area over time are shown in Fig. 3.

0

20

40

60

80

100

120

0(BG) 1(BL) 2(B W) 3(BR) 4(BO) 5(BY ) 6(AG) 7(AL) 8(AW) 9(AR) 10(AO) 11(AY) 12(YG) ?(OP)

Age (cohort tag) Males/females

Nu

mb

ers

Figure 2. Age distribution of tagged Soay sheep presumed to be alive at the end of

2011.

Page 3: Annual Report 2011 Final - Soay sheepsoaysheep.biology.ed.ac.uk/.../files/dnussey/AnnualReport2011.pdf1Institute of Evolutionary Biology, University of Edinburgh. 2Macaulay ... 6 Department

3

0

500

1000

1500

2000

2500

1985

1987

1989

1991

1993

1995

1997

1999

2001

2003

2005

2007

2009

2011

Year

Sh

eep

Village bay

Island count

Figure 3. The number of tagged sheep regularly using the study area since 1985.

One whole-island count yielded 2147 tagged and untagged sheep, with the details

displayed in Table 1. The total population had increased by 4.3% since summer 2010

when it was at 2058. This gives a delta (calculated as ln (Nt+1/Nt)) of +0.04. The

whole island counts are also shown in Figure 3, where it can be seen that the 2011

count preserves the pattern that we have never seen two population declines in a row

since 1985.

Table 1. Demographic and geographic distribution of sheep observed during the count

of Hirta on August 17th 2011. Coat colours are DW = dark wild, DS = dark self, LW =

light wild, and LS = light self.

Location Females Males Lambs Total

Conachair/Oiseval

Mullach Bi/Cambir

Ruaival/Village

DW

209

304

291

DS

5

11

13

LW

62

82

106

LS

1

11

2

DW

59

75

94

DS

5

0

2

LW

27

17

21

LS

0

0

0

200

277

270

570

778

799

Total 804 29 250 14 228 7 65 0 750 2147

Page 4: Annual Report 2011 Final - Soay sheepsoaysheep.biology.ed.ac.uk/.../files/dnussey/AnnualReport2011.pdf1Institute of Evolutionary Biology, University of Edinburgh. 2Macaulay ... 6 Department

4

REPORTS ON COMPONENT STUDIES

Vegetation.

Mick Crawley.

Sheep numbers have been consistently high for the last 9 years, and this is reflected in

the vegetation. In 2011 we saw total biomass at its second-lowest ever in both March

and August: 7.2 g per 0.04 m2 in March (long-term mean = 10.7; only March 2009

was lower at 6.8) and 9.3 g per 0.04 m2 in August (long-term mean = 15.6; only

August 2008 was lower at 8.9).

The in-bye grassland on Hirta is the most productive plant community used by the

Soay sheep. At a scale of 10cm or so, the sward consists of a mosaic of gaps and

tussocks. The gaps are closely-grazed, remain bright green all year round, and

typically contain lots of clover (Trifolium repens). The tussocks are much taller (up to

20cm sward height), more lightly grazed, and in winter are made up largely of dead

organic matter (uneaten grass production left over from the previous summer). For

both gaps and tussocks, there is a very strong seasonal cycle in their biomass and

botanical composition. Green biomass is high in summer and low in winter, with

dead organic matter showing the opposite pattern. Because they can grow at very low

temperatures, bryophytes (mosses and liverworts) benefit from reduced light

competition resulting from die-back of the grasses, and peak in abundance in winter.

As is so often the case in studies of climate change, the size of these within-year

fluctuations is large compared to the long-term trends in average plant biomass (Fig.

4) There is a significant downward trend over time in green biomass in tussocks (top

right; p < 0.01) but not in gaps (top left).

Fluctuations in the sheep population are reflected by changes in the distribution,

abundance and botanical composition of gaps and tussocks. In the year following a

population crash, grazing pressure is relatively light, offtake is low, and tussocks are

larger and more numerous. The result is that dead organic matter builds up to a peak

in the winter following the crash. When sheep numbers are high (as they have been

for the last 9 years), then tussocks are rare (especially in winter) and gaps are

numerous and very closely grazed. Figure 5 shows the very clear long-term

downward trend in tussock cover. It will be interesting to see if tussock cover ever

returns to its pre-2005 levels.

Page 5: Annual Report 2011 Final - Soay sheepsoaysheep.biology.ed.ac.uk/.../files/dnussey/AnnualReport2011.pdf1Institute of Evolutionary Biology, University of Edinburgh. 2Macaulay ... 6 Department

5

Figure 4. Time series for March and August (1993 to 2011) green biomass (dry mass

of green grass plus herbs per 0.04m2; top row), dead organic matter (DOM mostly

dead grass; centre row), and bryophytes (mosses plus liverworts; bottom row), in

gaps (left column) and tussocks (right column) from the inbye grassland on Hirta (this

is the Agrostis-Holcus grassland between the Head Dyke and the shore; under the

National Vegetation classification, it is U4 Festuca ovina – Agrostis capillaris –

Galium saxatile grassland: Holcus lanatus – Trifolium repens sub-community). Note

the differences in scale on the y-axes between rows and columns (e.g. there is much

more green biomass and DOM in tussocks than in gaps, and more bryophyte biomass

in gaps than in tussocks).

Page 6: Annual Report 2011 Final - Soay sheepsoaysheep.biology.ed.ac.uk/.../files/dnussey/AnnualReport2011.pdf1Institute of Evolutionary Biology, University of Edinburgh. 2Macaulay ... 6 Department

6

Figure 5. The clearest consequence of increasing sheep numbers (left; whole island

count in August showing more than 39 extra sheep per year on average; r2 = 0.48)

has been a decline in the mean cover of tussocks in the following March (right; p <

0.001). Tussock cover (scored on a scale of 1 to 5; right) is an average score out of 30

across 14 stratified random sampling stations within the Head Dyke each year. A

score of 5 equates to tussock cover of 5/30 = 16.7%. Clearly this linear trend cannot

continue (it predicts negative tussock cover by 2015), and we expect that tussock

cover would increase in the year following a crash (if we ever see one again).

Weather effects on female August weight.

Ana Bento.

In the 2010 report, I wrote about the reconstructed weather time series for St. Kilda

and how it could be used to understand population crashes. The analysis showed that

are no clear environmental drivers that explain the crashes.

This past year, I have been focusing on weather effects on several aspects of sheep

demography and traits. Here, I analyse trends in female August weight for the time

period 1989 to 2010. Body weight is important for both fertility and survival, which

are components of fitness. In each age class we see clear interannual variation in the

distribution of body weight (Fig. 6).

Page 7: Annual Report 2011 Final - Soay sheepsoaysheep.biology.ed.ac.uk/.../files/dnussey/AnnualReport2011.pdf1Institute of Evolutionary Biology, University of Edinburgh. 2Macaulay ... 6 Department

7

Figure 6. Female August weight trends for lambs (left) and adults (age 4+; right).

I set out to look for possible weather drivers that may explain this variation in body

weight against a background of past and current population densities, which have an

effect on food availability. In analysing the time series, despite the clear interannual

variability, significant trends emerge for some of the weather variables. For example,

for the period 1957 to 2010, mean maximum December temperature exhibits both

large variability between years, and a significant positive trend (Fig. 7). Furthermore,

recent data from the Met Office for 2011 shows that December maximum temperature

was the mildest since 2006, and over 5°C warmer than December 2010 (for the UK).

This additional data point would reinforce the already existing significant trend.

Figure 7. December daily maximum temperature for 1957 to 2010. The trend is

positive (p = 0.023). The solid and dashed lines depict the trend in the monthly mean

and median December maximum temperature across the years respectively.

Page 8: Annual Report 2011 Final - Soay sheepsoaysheep.biology.ed.ac.uk/.../files/dnussey/AnnualReport2011.pdf1Institute of Evolutionary Biology, University of Edinburgh. 2Macaulay ... 6 Department

8

I used a machine-learning algorithm, called random forests, to select the best

candidate weather variables affecting August weight for each of the age classes. I

collapsed females age four and above into one age class. I then regressed these

weather variables and past and current density dependence against body weight. As

was the case last year, the picture that emerges is complex (Table 2).

Table 2. Weather effects on the female August weight of different age classes.

Age class Significant weather variables Effect

Lambs Previous population density

March rain

Negative

Positive

Yearlings Population density two years prior

Previous population density

February rain

Previous summer rain

Spring day degrees

Negative

Negative

Positive

Positive

Negative

Age 2 Current population density Negative

Age 3 Previous population density

Current population density

Spring day degrees

Previous December wind

Negative

Negative

Positive

Negative

Age 4+ Previous population density

Previous December maximum temperature

March maximum temperature

Negative

Positive

Positive

Different weather variables affect weight at each age class. What this means is that

any climatic change is therefore likely to have complicated, and sometimes opposing,

effects on the dynamics of different demographic groups of the population. It is

interesting to point out that at age two, females appear not to be affected by weather

conditions. Furthermore, birth effects seem to be forgotten by this age too, since there

are no effects of weather or density in the year of birth at this age.

This analysis shows that the traditional way of integrating weather variables into

discrete calendar time periods may be insufficient to explain weather effects on a

population. We are now identifying critical time windows during which climatic

drivers affect traits, such as weight, to better understand the impact of climate change

and variability on population dynamics.

Biomarkers of ageing in Soay sheep: oxidative stress and telomere length.

Louise Christensen, Jen Dorrens, Colin Selman, Melissa Page, Jon Blount, Chris

Mitchell, Lea Harrington, Dan Nussey, and Kathryn Watt.

Ageing, or senescence, is the physiological deterioration with advancing age that is

accompanied by declining survival and reproductive performance in most species.

Studies of wild mammals and birds regularly document age-related declines in

survival and/or reproduction, and find among individual variation in the onset and rate

of the decline. However, the causes and consequences of this variation remain poorly

understood. Senescence is a physiological process, and may vary independently of

chronological age among individuals, so it is important to develop biomarkers that

Page 9: Annual Report 2011 Final - Soay sheepsoaysheep.biology.ed.ac.uk/.../files/dnussey/AnnualReport2011.pdf1Institute of Evolutionary Biology, University of Edinburgh. 2Macaulay ... 6 Department

9

can give some insight into the underlying process of physiological deterioration for

use in longitudinal studies of wild animals. Markers of oxidative stress and telomere

length are two prime candidate biomarkers of ageing that could give insight into the

processes of cellular damage accumulation underpinning senescence at the whole

organism level. Few detailed studies of wild vertebrates exist that have fully

investigated how and why such biomarkers might vary among individuals and within

individuals across their lifespan, and in turn test their capacity to predict future

survival and reproductive performance of individuals. Collaborative projects,

involving researchers from Aberdeen, Edinburgh, Exeter and Montreal are currently

under way to optimise robust measures of oxidative stress and telomere length using

blood samples from Soay sheep, and then determine the causes and life history

consequences of variation in these biomarkers.

Reactive oxygen species (ROS) are produced during normal metabolism but can also

cause damage to crucial cellular components, such as lipids, proteins and DNA,

unless their effects are neutralised by a variety of antioxidant molecules circulating in

our bodies. The accumulation of such oxidative damage to cells by ROS was

proposed as a key mechanism explaining why we age, although this hypothesis is not

standing up to close scrutiny from researchers working on short-lived organisms in

the lab. Increased oxidative damage has also been proposed as an important cost of

investment in metabolically expensive activities like growth and reproduction.

Studying oxidative processes in the field is challenging, as it requires the

measurement of highly unstable molecules and samples need to be kept at ultra-low

temperatures and measured fairly soon after collection. Using liquid nitrogen vapour

shippers and our new -80°C freezer on St Kilda (Fig. 8A) – transported with the kind

help of Angus Campbell of Kilda Cruises and Qinetiq Movement team – we were able

to bring back around 400 plasma samples kept at ultra-low temperatures from the

2010 and 2011 August catch-ups. Jon Blount’s (Exeter) and Colin Selman’s

(Aberdeen) labs have measured markers of oxidative damage to lipid

(malondealdehyde) and protein (carbonyls) molecules, and two measures of

antioxidant protection. As part of her honours project with Colin Selman at Aberdeen,

Louise Christensen has begun analysing this data. Louise has found interesting

associations among the markers measured in 2010, documenting negative associations

between one of the antioxidant markers (superoxide dismutase) and both markers of

damage (Fig. 8B), as well as positive associations among the two antioxidant markers.

However, the protein and lipid damage markers were not correlated, and there was

little evidence for consistent patterns of variation with age or sex among the different

markers. Further analysis on the full 2010 and 2011 data should illuminate similarities

and differences among markers in their associations with age and life history, and

follow-up sample collection and lab work in future will establish a longitudinal data

series to investigate within-individual variation in oxidative stress.

Telomeres are repetitive sequences of DNA that are found at the tips of the

chromosomes of many organisms, including all mammals. At each cell division, a

small amount of DNA at the chromosome ends is lost as the cell replicates its DNA.

Telomeres act to protect coding DNA from this attrition and are thus important for the

maintenance of cell replicative potential and the integrity of chromosomes within the

cell nucleus. Therefore, in the absence of telomere lengthening or repair mechanisms,

telomeres shorten at each cell division and are also thought to further shorten in

response to oxidative stress. Assays of telomere length from samples of cells in

Page 10: Annual Report 2011 Final - Soay sheepsoaysheep.biology.ed.ac.uk/.../files/dnussey/AnnualReport2011.pdf1Institute of Evolutionary Biology, University of Edinburgh. 2Macaulay ... 6 Department

10

humans, typically from white cells in the blood, suggest associations with life stress

and subsequent survival and health. Telomere length has also recently emerged as a

biomarker of some promise in ecological and evolutionary studies. We plan to use the

large freezer bank of stored white cell fractions (buffy coat) from Soay sheep

collected by Josephine Pemberton, Jill Pilkington and the respective August catch

teams going back at least a decade, to investigate variation in telomere length in a

wild mammal for the first time. Jen Dorrens has been working to optimise and

validate two different types of telomere length assay. First, telomere restriction

fragment analysis (TRF), often seen as the ‘gold standard’ measurement of telomere

length, is a low-throughput technique that measures the size range of the terminal

telomere restriction fragment present in a sample. An early successful TRF gel using

Soay samples from August 2011 suggests interesting patterns of change in the

variance of telomere lengths with age (Fig. 8C). The second technique is quantitative

real-time PCR (qPCR), a much higher-throughput and rapid technique which provides

an average measurement of telomere length relative to a single-copy reference gene.

So far, Jen has identified a suitable reference gene and has initiated a collaboration

with Systems Biology in Edinburgh to adapt high-throughput robotics to perform

qPCR assays on the sheep samples. Our plan is to complete optimisation of both

techniques, validate the qPCR assays against TRF assays run on approximately 30

Soay samples, and to utilise this qPCR assay to measure telomere lengths

longitudinally across the lives of many Soays. We hope to assess how genetic and

environmental variation generates among- and within-individual variation in telomere

length, and test whether telomere lengths predict subsequent fitness of sheep

independent of their age.

Figure 8. (A) The new -80°C freezer, installed on Kilda and used to collect samples

for oxidative stress in August 2011; (B) A negative association between a major

antioxidant molecule (SOD, superoxide dismutase) and a measure of damage to

proteins (“PC Concentration”: Protein carbonyls) from samples collected in August

2010; (C) A telomere restriction fragment analysis gel using white blood samples

collected in August 2011. The smears under the numbered lanes are telomeres from

different animals, with the darkest parts reflecting the ‘average’ telomere lengths in

the sample. The bands in the lanes either side are size ladders. Lanes 1-3 are females

7 years or older, lanes 4-7 are lambs.

(C)

(B)

(A)

Page 11: Annual Report 2011 Final - Soay sheepsoaysheep.biology.ed.ac.uk/.../files/dnussey/AnnualReport2011.pdf1Institute of Evolutionary Biology, University of Edinburgh. 2Macaulay ... 6 Department

11

Explaining variation in immune responses in an evolutionary context.

Adam Hayward, Andrea Graham, Jill Pilkington, Kathryn Watt and Andrea Graham.

In recent years, there has been increased interest in investigating the relationships

between the strength of immune responses and indicators of fitness, such as survival,

reproduction, and body condition. Knowing how immunity is related to fitness will

increase our understanding of how variation in immune responses between individuals

is maintained by natural selection. This has led to increased study of immunity in wild

populations, though ecological studies often make several key assumptions which

have been shown to be incorrect. For instance, higher immune responses do not

always lead to lower parasite levels, and lower parasite levels do not always maximise

health or fitness. Therefore, in order to understand how variation in immunity is

associated with fitness, measurement of immunity, parasite levels and fitness at the

same time is crucial. We used data collected from St Kilda from 1997 to 2007 to

examine associations between levels of an antibody produced in response to the most

important parasite (as a measure of immunity), faecal egg counts (FEC) of intestinal

worm eggs (as a measure of parasite burden), and body weight (as a measure of

condition or fitness).

Figure 9. Associations between measures of fitness (Weight), parasite burden (faecal

egg count, FEC), and immune response (Antibody) in A) lambs, and B) adult females.

The plots show schematic relationships from the results of linear mixed-effects

models.

We began by estimating the correlations between the three traits, in data collected

from lambs and adult females; the results are shown schematically in Figure 9. In

Page 12: Annual Report 2011 Final - Soay sheepsoaysheep.biology.ed.ac.uk/.../files/dnussey/AnnualReport2011.pdf1Institute of Evolutionary Biology, University of Edinburgh. 2Macaulay ... 6 Department

12

lambs, we found that sheep with higher body weight had lower faecal egg counts of

parasites (FEC), suggesting that heavier sheep had fewer parasites. We also found

some indication that heavier sheep produced higher antibody responses, though the

association was not as strong as it was between body weight and FEC. This suggested

that sheep with higher antibody responses should have lower FEC, but this was not

the case when the data were analysed. The analysis of adult females showed the same

results: heavier sheep had lower parasite levels, made stronger antibody responses

(though the association was stronger than in lambs), but there was no relationship

between antibody levels and FEC.

Next, we attempted to establish the genetic basis of the traits. Figure 10 shows the

variance in each of the three traits measured in lambs and adult females, and divides

this variance into contributions from genetic factors, temporal variation, and maternal

effects. As in previous studies, we found high genetic variance for body weight, but

low genetic variance for FEC. We also found, for the first time, high genetic variance

for the antibody response to the most important parasite. The genetic variance in the

trait indicates how strongly the trait can responds to natural selection - if there is high

genetic variance for a trait, and strong selection on it, the trait is expected to evolve.

Figure 10. The contribution of genetic variance to the total phenotypic variance in

each of body weight, faecal egg count (FEC), and antibody levels in lambs and adult

females. The total variation in each trait is divided into contributions from: annual

variation (purple); genetic variation (red-brown); non-genetic individual differences

(orange); maternal effects (yellow); variation between dates of antibody assays

(green); unexplained ‘residual’ or ‘environmental’ variation (grey). The proportion

of genetic variance is of interest since it is the heritability of the trait - e.g. the

antibody has a heritability of 30% in lambs and 20% in adults.

Page 13: Annual Report 2011 Final - Soay sheepsoaysheep.biology.ed.ac.uk/.../files/dnussey/AnnualReport2011.pdf1Institute of Evolutionary Biology, University of Edinburgh. 2Macaulay ... 6 Department

13

We next tested the basis of the correlations between the traits, by splitting them into

their genetic and environmental components. In lambs, we found that there was no

genetic basis for either the correlation between body weight and FEC, or for the

correlation between body weight and the antibody. In adult females, we found that the

association between body weight and FEC was due to differences between individuals

- those that were heavy had low parasite levels because of individual differences: for

instance they may have different grazing behaviour, causing them to avoid parasites

and feed on the most nutritious areas of pasture. The association between body weight

and antibody levels was not due to genetic effects, but to environmental sources.

This study is the first attempt to characterize the relationships between parasite

burden, a specific immune response, and a measure of fitness in a natural population.

The results illustrate the complexity of studying the immune system in such

populations, even when a large longitudinal data set is available. It also highlights the

fact that higher measures of immune responses are not necessarily associated with

lower parasite burdens. This is likely to be due in part to the complexity of the

immune response to infection - at early stages, higher antibody responses may be

associated with higher parasite levels in an attempt to counter them; late in infection,

high antibody levels may have reduced parasite numbers to a low level, leading to no

overall relationship. Alternatively, the strength of the immune response may not be a

function of the level of infection, but of individual condition - such a hypothesis is

consistent with the results described here. The results also highlight the need for a

multivariate approach for characterising immune responses in natural populations: by

measuring multiple aspects of immune phenotype simultaneously, richer insight into

the causes of variation in immune responses will be gained.

Immunological and parasitological research: an update

Dan Nussey, Kathryn Watt, Jill Pilkington, Josephine Pemberton, Rose Zamoyska,

Tom McNeilly, Adam Hayward, Andrea Graham, and Kathleen Morriss.

Individuals vary considerably in their susceptibility to infections and the fitness

consequences of those infections in natural populations. The ecological and

evolutionary drivers of this variation are likely to be underpinned by differences in

immune responses and in exposure to parasites associated with genes, environment,

age and sex. Previous work on the Soay sheep population has shown that the sheep

are exposed to a diversity of helminth and protozoan parasites and that levels of

infection vary with age and sex, while high parasite burdens are associated with lower

weight and survival prospects. We have also found that individuals with a high level

of a particular measure of immunity (anti-nuclear antibodies) have improved over-

winter survival during crashes but reduced fecundity, suggesting a trade-off that could

maintain variation in immunity. In August 2010 we preserved white blood cells from

a small number of ewes and assessed the proportions of different kinds of T cells,

which are a crucial aspect of the vertebrate immune response. Work in lab mice,

domestic mammals and humans suggests profound changes in the abundance of

different sorts of T cells as individuals age, and we found broadly similar patterns in

wild Soay sheep. Over the last 12 months, these findings have been followed up in

our labs at Edinburgh University, the Moredun Research Institute and Princeton

University.

Page 14: Annual Report 2011 Final - Soay sheepsoaysheep.biology.ed.ac.uk/.../files/dnussey/AnnualReport2011.pdf1Institute of Evolutionary Biology, University of Edinburgh. 2Macaulay ... 6 Department

14

Antibodies and fitness: We previously documented a positive association between

over-winter survival of a crash and levels of anti-nuclear antibodies (ANA) in the

plasma of adult Soay ewes. We are currently following up this finding, to try to

understand what kinds of antibodies are responsible for improved survival. ANA are

antibodies that bind to material commonly found in mammalian cells, and are

associated with both a normal, healthy immune response (e.g., indicative of the extent

of B cell stimulation and general immune responsiveness) but also, at high levels,

with some forms of autoimmune disease. We first hypothesized that the association of

ANA with adult survival might be driven by antibodies specific to Teladorsagia

circumcincta, the strongyle nematode species most prevalent and virulent in adults in

the population. We measured titres of the antibody in the 1997-2007 sample set but

found that T circumcincta-specific antibody did not explain the survival result. In

Edinburgh, we are currently optimising a variety of other antibody assays (including

total antibody levels, antibodies to a novel antigen, and different antibody isotypes)

and plan to measure these in ewe samples from pre-crash August catches and

determine which, if any, best explain the association between ANA and survival. We

also found a positive association between ANA in reproductive ewes and the

probability of their offspring surviving the first few months of life. Antibodies against

T. circumcincta have now been measured from 1997-2007 for lambs at birth and at 4-

5 months of age, to examine associations between lamb immunity, growth and

survival in relation to the antibody titres of their mothers. The associations are

complex, but we are finding that maternal antibody transfer affects survival of the first

year of life, via effects on growth and independent of birth weight. Finally, in

Princeton, we are developing plasma assays to quantify protein malnutrition. These

will allow us to assess the extent to which immunity and nutrition synergize to

determine whether an individual survives a crash.

T cell phenotype and function: In August 2011, we followed-up exciting results from

the 2010 pilot study which suggested variation in the proportions of different T cell

types with age in females. We separated and preserved white blood cells from around

200 sheep caught in August and are currently using flow cytometry to assess the same

T cell types as previously in a much larger sample. We hope soon to be able to use

this data to examine fitness correlates of T cell phenotypes, and compare these

measures among sexes and also individuals experiencing different levels of infection.

We also cryo-preserved white blood cells from around 150 individuals caught in

August 2011, and these samples are in long-term storage in liquid nitrogen in

Edinburgh. Preliminary examination of these samples suggests that the cryo-preserved

cells have remained viable and capable of proliferating following stimulation in the

laboratory. Our plans are to optimise proliferation assays using these cells, and to

examine the proliferative and cytokine response, and test whether these measures

relate to the levels and types of infection experienced by individuals. For example,

nematode infections are classically thought to elicit so-called Th2-type responses

from T cells, while protozoan parasites (such as Coccidea) elicit Th1-type responses.

We hope to test whether sheep with high nematode burdens but low Coccidea burdens

show stronger Th2 responses, and vice versa, and relate individuals’ infection history,

T cell function and subsequent survival and reproductive fitness.

Microparasite survey: To understand how immunity affects fitness in the sheep, we

are seeking comprehensive information on the infections that they carry. Helminths,

coccidia, and ectoparasitic infections have been well characterized over the years, but

Page 15: Annual Report 2011 Final - Soay sheepsoaysheep.biology.ed.ac.uk/.../files/dnussey/AnnualReport2011.pdf1Institute of Evolutionary Biology, University of Edinburgh. 2Macaulay ... 6 Department

15

whether those are the only infections circulating among the sheep is unknown. In

2000, Ken Wilson spearheaded a small survey to assess the seroprevalence of

microparasites (bacteria and viruses) that are common in sheep on the Scottish

mainland (see Annual Report from 2001): Mycobacterium paratuberculosis,

Chlamydophila abortus (formerly known as Chlamydia psittaci), Mycoplasma

ovipneumoniae, and the pestivirus that causes Border disease. Ken sent 50 plasma

samples to a testing service at Scottish Agricultural College (SAC) and found no

evidence that these microparasites affect the St Kilda population: only one animal was

weakly positive for one of these antigens (C. abortus), and with such a low titre that it

is likely to have been a false positive. In 1986, Josephine Pemberton submitted

samples for a similar serosurvey for Maedi-Visna virus, with similarly negative

results. We decided to undertake a larger serosurvey (in terms of the number of sheep

as well as years tested) for these microparasites plus Leptospira spp., parainfluenza

virus, and orfpoxvirus (with advice and cooperation of Dr. Franz Brulisauer and

colleagues at SAC). Analysis of 400 adults suggests that prevalences are indeed very

low, though positive responses to Leptospira antigens were detectable in 7% of

samples. We are currently verifying this result with a serosurvey of 350 yearling

sheep, the age class most likely to reflect exposure to leptospirosis. We are also

planning to test 2010-11 blood samples for ovine herpesvirus (OHV-2) by PCR. We

hope to have results to report by mid-2012.

Changes in gene frequency: genetic drift or natural selection?

Jon Slate, Jake Gratten and Susan Johnston.

In previous reports we have described the discovery of the genes responsible for the

coat colour, coat pattern and horn type polymorphisms observed in Soay sheep. The

genes are called TYRP1, ASIP and RXFP2 respectively. Because we have obtained

DNA profiles of over 1000 sheep at all three genes, it is possible to ask whether (and

by how much) the frequency of the different variants (alleles) at these genes has

changed over the course of the long-term study? Identifying a significant temporal

change in allele frequency is exciting, because it could mean that the gene is

responding to natural selection (i.e. we are observing microevolution). However,

natural selection is not the only evolutionary process that can lead to changes in allele

frequency. Every generation, allele frequencies can change by random sampling

events in a process known as genetic drift. Genetic drift affects all genes, but in a less

predictable way than natural selection does. Furthermore, genetic drift is strongest in

small, isolated populations and therefore could be important in Soay sheep. We were

keen to establish whether temporal trends in allele frequency at TYRP1, ASIP and

RXFP2 could be explained by genetic drift alone or were large enough to be due to

natural selection.

To test the two alternatives, we utilised a simulation method known as gene dropping.

The idea behind gene dropping is very simple. Using the exact same pedigree as the

Soay sheep pedigree that has been reconstructed by 25 years worth of DNA profiling,

we simulated the inheritance of genes with similar starting allele frequencies as the

genes of interest. We ensured that the inheritance pattern was ‘random’ with respect

to which allele was passed from parent to offspring, and we counted allele frequencies

each year (see Fig. 11). This process mimics genetic drift in that Mendelian

inheritance is random, and there is no selection on the gene. By doing this many times

Page 16: Annual Report 2011 Final - Soay sheepsoaysheep.biology.ed.ac.uk/.../files/dnussey/AnnualReport2011.pdf1Institute of Evolutionary Biology, University of Edinburgh. 2Macaulay ... 6 Department

16

1. Assign alleles in Gen 0 using frequencies at start of study

2. Offspring in Gen 1 inherit an allele at random from parents

3. Individuals in Gen t, with unknown parents, are allocated alleles based on

frequency in Gen t-1.

4. The process is continued until everyone has been assigned a genotype

5. Perform regression of allele frequency against year

6. Repeat 10000 times to obtain distribution of regression coefficients if no

selection

(e.g. 10,000), and recording the change in gene frequency over time it is possible to

determine the amount of change in allele frequency expected by drift alone. By

comparing the distribution of allele frequency changes at the simulated genes with

that seen at our three genes of interest we can ask whether the observed results are

within the boundaries expected by genetic drift. If TYRP1, ASIP or RXFP2 show

greater changes in allele frequency than expected by chance, then genetic drift is

unlikely to be the only mechanism involved, and instead there is evidence of

microevolution due to natural selection.

Figure 11. Cartoon illustrating the gene-dropping simulation.

Coat colour (TYRP1)

The light coat colour has

increased in frequency over

the course of the long term

study (regression of light

colour frequency on year, r2

= 0.33; p < 0.01). However,

the gene dropping

simulations show that

genetic drift alone could

generate a change as great

as this with probability

0.13. Therefore, there is no

overwhelming evidence

that the trend is attributable

to anything other than

genetic drift.

Figure 12. Temporal changes in the light coat colour (L) and the distribution of gene-

dropping predicted regression coefficients (R). The observed temporal change is

within the distribution expected by drift alone.

Page 17: Annual Report 2011 Final - Soay sheepsoaysheep.biology.ed.ac.uk/.../files/dnussey/AnnualReport2011.pdf1Institute of Evolutionary Biology, University of Edinburgh. 2Macaulay ... 6 Department

17

Coat pattern (ASIP)

The frequency of the self coat pattern has declined over the course of the long-term

study (linear regression: r2 = 0.30, F1,22 = 9.38, p = 0.006), but again gene-dropping

shows the trend is within the limits expected by genetic drift alone (p = 0.21).

However, gene-dropping did reveal one unusual pattern at ASIP. Over the course of

the study the frequency of individuals that are heterozygous has increased, going from

a deficit (relative to the number

expected, given allele frequencies) to

an excess. This trend is stronger than

expected by drift (p = 0.019). The

reason for this trend is not certain, but

it could be because homozygous self

individuals have low juvenile survival

but high reproductive success (among

those that do survive). Although self

homozygous adults are rare, they tend

to have a greater than average number

of offspring, and most of those

offspring are heterozygous. Note

however, that this explanation, which

is a form of frequency-dependent

selection, can only result in an excess

of heterozygous individuals

ephemerally, and once the allele

frequencies reach equilibrium, the

pattern should disappear.

Figure 13. Temporal changes and gene-dropping results for the self coat pattern (L)

and the frequency of ASIP heterozygotes (R) at ASIP.

Page 18: Annual Report 2011 Final - Soay sheepsoaysheep.biology.ed.ac.uk/.../files/dnussey/AnnualReport2011.pdf1Institute of Evolutionary Biology, University of Edinburgh. 2Macaulay ... 6 Department

18

Horn type (RXFP2)

The polymorphism in horns type (normal or scurred in males; normal, scurred or

polled in females) is caused by unknown mutations in the RXFP2 gene. Two genetic

markers within RXFP2 are good, but imperfect, predictors of horn type. At both of

these markers, the alleles associated with

the polled allele (which causes scurs in

males and polledness in females) have

increased in frequency. In both

instances, the temporal trends are

stronger than expected by genetic drift,

and therefore it seems likely that RXFP2

is undergoing microevolutionary

changes driven by natural selection. The

precise mechanism for this is unknown

although there is evidence that

heterozygous males, who have relatively

small, but normal, horns have relatively

high survival and

reproductive success

compared to males

with either

homozygous

genotype (Susan

Johnston, PhD thesis

2010). If so, it is

expected that the

temporal change in

allele frequency will

level off and RXFP2

will be in Hardy-

Weinberg Equilibrium, once the polymorphism becomes balanced and reaches

equilibrium frequency.

Figure 14. Temporal changes (top) and the distribution of gene-dropping predicted

regression coefficients (bottom) at RXFP2 markers SNP10 and SNP11

Summary

Gene-dropping is a useful procedure for determining what magnitude of change in

allele frequency can occur by genetic drift alone. Some quite strong trends can be

generated by drift, and so some of the more striking changes in gene frequency, such

as that seen in TYRP1 (the coat colour gene), need not be caused by natural selection.

Of course, gene-dropping only provides clues, not a definitive answer, into what is

causing evolutionary change, because genetic drift and natural selection are not

mutually exclusive. However, in the absence of temporal trends that cannot be

explained by genetic drift, it is hard to make a case that gene frequencies have

changed as a result of natural (or sexual) selection.

P = 0.035 P =0.001

Page 19: Annual Report 2011 Final - Soay sheepsoaysheep.biology.ed.ac.uk/.../files/dnussey/AnnualReport2011.pdf1Institute of Evolutionary Biology, University of Edinburgh. 2Macaulay ... 6 Department

19

The population genetics of Soay sheep coat colour: a model-based approach.

Dylan Childs and Jon Slate.

Understanding the role that natural selection plays in shaping observed patterns of

genetic variation remains an important challenge for evolutionary biologists. The

development of effective molecular genetics tools has made it possible to identify

many of the genes associated with adaptively significant traits in natural populations.

For example, recent work with Soay sheep has identified loci associated with

morphological traits such as coat colour and pattern, as well as horn morphology in

males and females. By analyzing the association between the genotype of an

individual and components of their fitness, we can begin to explain the observed

patterns. However, such relationships can be difficult to synthesize when genotype

effects are sex-specific, or environmentally dependent.

We set out to develop a framework for studying the population genetics of

morphological traits in Soay sheep using a recently developed population modelling

tool, the general Integral Projection Model (IPM). The IPM has gained prominence in

recent years as an efficient tool for modelling the population dynamics of

demographically structured populations, where individual performance varies as a

consequence of variables such as size, age, sex, and of course, genotype. Such models

are valuable because they can incorporate a great deal of information about Soay

sheep demography in order to generate precise quantitative predictions. Most

importantly, with such a model in hand, it becomes possible to perform in silico

experiments to understand how different factors influence the dynamics of genes

within the St Kilda population.

-2 -1 0 1 2

0.0

0.2

0.4

0.6

0.8

1.0

density environment (standardised)

P(s

urv

ival)

F

M

-2 -1 0 1 2

0.5

1.0

1.5

2.0

2.5

density environment (standardised)

rela

tive fecun

dity (

lam

bs s

ired)

Figure 15. (left panel) Adult survival probability as a function of density,

sex, and genotype. Dashed and continuous lines: males and females,

respectively. Dark, grey, and light lines: GG, GT, TT genotypes,

respectively. (right panel) Relative male reproductive success as a

function of density and genotype. Dark, grey, and light lines: GG, GT, TT

genotypes, respectively.

Page 20: Annual Report 2011 Final - Soay sheepsoaysheep.biology.ed.ac.uk/.../files/dnussey/AnnualReport2011.pdf1Institute of Evolutionary Biology, University of Edinburgh. 2Macaulay ... 6 Department

20

Our first “proof of concept” model has been constructed to understand the

polymorphism in coat colour. Coat colour in Soay sheep is determined by a single

locus with two alleles, a dominant “wild type” dark coat (labeled G) and light coat

allele (labeled T). The frequency of the G allele has fluctuated over the course of the

St Kilda study, but little directional change has occurred. Gene-dropping studies (see

previous report) indicate that the observed variation is consistent with random genetic

drift, suggesting that fitness differences between genotypes are weak or nonexistent.

However, such studies do not exclude the possibility that the selection regime has

changed over the course of the study, or that the polymorphism is actively maintained

via a classical mechanism such as heterozygote advantage. We set out to examine

these possibilities.

In order to construct the model we statistically characterised the relationship between

individual state (genotype, sex, age and size) and demographic rates (probability of

survival, reproductive success, offspring size and growth). We compared a range of

different models for the effect of genotype and found that those in which the coat

colour alleles impact demographic performance additively generally performed best.

The fitted models also include terms that capture variation among years (“year

effects”) and total population density. Interestingly, we found clear evidence for an

interaction between genotype and population density (Fig. 15). Such interactions are

interesting because they favour the maintenance of a genetic polymorphism.

With a complete set of demographic functions in hand, it is straightforward to

construct an IPM for the St Kilda population. The resultant model (not shown)

describes the stochastic, density dependent dynamics of the size-age-sex-genotype

distribution. By running the model for many decades we were able to establish

whether the polymorphism is indeed stable and if not, to estimate the rate at which the

successful allele is expected to reach fixation. The model predicts that over the long

term, the wild type dark colour allele will move toward fixation at a rate of

approximately 2% per year. Preliminary analysis suggests that this is largely due to

increased reproductive success of males carrying at least one copy of the dark coat

allele (GG and GT) relative to the light homozygotes (TT).

The density of Soay sheep on the island of Hirta has been increasing over the course

of the study, and the presence of genotype-by-density interactions in the demographic

models suggests that this change may have altered selection on the polymorphism. In

order to understand whether the relative fitness of the two alleles has remained

approximately constant over the course of the study we analysed a slightly modified

version of the model. The statistical functions describing the demographic rates were

refitted with an additional term capturing temporal trends in vital rates. We then took

each year of the study in turn and constructed a “year-specific” model for the

population. However, note that each of the resultant models is still stochastic since it

incorporates temporal variation that cannot be explained by the trend effect.

We then determined whether the polymorphism was stable in each year assuming the

mean environment did not change by calculating the estimated growth rate (fitness) of

each allele when rare (i.e. T invading a homozygous GG population, and vice versa)

in the environment defined by that year. These quantities – the marginal growth rates

– are used to establish conditions for a “protected polymorphism”, corresponding to

the case where each allele has positive fitness when rare. The absence of a protected

Page 21: Annual Report 2011 Final - Soay sheepsoaysheep.biology.ed.ac.uk/.../files/dnussey/AnnualReport2011.pdf1Institute of Evolutionary Biology, University of Edinburgh. 2Macaulay ... 6 Department

21

polymorphism implies that the fittest allele will eventually reach fixation. By plotting

the marginal growth rate of each allele as a function of each year-specific model, we

were able to establish how the selective environment has changed over time (Fig. 16).

This analysis showed that the selective regime operating on the coat colour allele has

reversed over the course of the study, with selection for the light coat colour occurring

early on in the study.

1990 1995 2000 2005

-0.1

0-0

.05

0.0

00

.05

0.1

0

Environment (year type)

Ma

rgin

al G

row

th R

ate

T invading G

G invading T

Figure 16. Marginal growth rates (MGR) of each coat colour allele as a

function of the “year-type”. When one allele has a positive MGR and the

other a negative MGR, the former is predicted to reach fixation. The

environment favoured the light allele until ~1991. The model predicts that

the dark coat allele will eventually reach fixation.

These simple in silico experiments indicate that selection on the coat colour alleles

has shifted over the course of the study, and that the polymorphism is probably not

neutral. Though it initially appears contradictory, this result is not entirely at odds

with gene-dropping studies in Soay sheep, as the predicted reversal of selection in

such a short time may well generate changes in allele frequencies that resemble

genetic drift. In future we aim to validate the predictive properties of the model (by

hindcasting the observed genotype frequency changes, for example) and use

perturbation analyses to establish which aspects of the model most influence our

predictions. Ultimately, the same baseline model can be used to understand selection

on those other morphological traits (e.g. coat pattern, horn type) with well-

characterised genetics.

Page 22: Annual Report 2011 Final - Soay sheepsoaysheep.biology.ed.ac.uk/.../files/dnussey/AnnualReport2011.pdf1Institute of Evolutionary Biology, University of Edinburgh. 2Macaulay ... 6 Department

22

Improving the pedigree of Village Bay sheep using superabundant genetic

variation.

Josephine Pemberton, Camillo Berenos, Phil Ellis and Michael Morrissey

The pedigree of sheep in Village Bay is one of the largest wild pedigrees derived

using genetics in the world, especially among populations in which many individual

males compete to fertilise females. The paternity data and pedigree have been used to

investigate male lifetime reproductive success, inbreeding and its consequences, the

inheritance of visible traits like coat colour and horns and the inheritance of more

subtly-varying traits like body size.

However, the current pedigree is by no means perfect. On the maternal side, some

ewes are known to lamb, but their lamb is never identified, while other ewes are never

seen to have lambed, even though they are prime-aged and it is likely they did.

Meanwhile, dead lambs are found and sampled, but our genetic system to date (a

panel of 18 microsatellite markers) was not powerful enough to make all the

connections between them and their mothers. Our paternity identification system

draws power from knowing the identity of the mother of each lamb, and is under-

powered relative to the paternity identification task at hand (often 2-300 candidate

fathers per lamb), meaning that for some lambs it does not find a statistically

supported father, and for a few it almost certainly identifies a wrong father.

In a new project starting in late 2010, we are genotyping as many as possible of the

archived Soay samples with a sheep chip which probes about 51,000 places (loci) in

the sheep genome for a single nucleotide polymorphism (SNP) known to exist in

sheep. About 38,000 of these loci are polymorphic in the Soays, giving

overwhelming power to determine parentage. Indeed, 38,000 loci is altogether too

much information, and in this report we work with a panel of 348 loci chosen to be

strongly polymorphic (minor allele frequency � 0.4) and independent of each other

(linkage disequilibrium R2 � 0.0099).

As an exercise, we compared the results of paternity analysis using microsatellites

versus SNPs for the 2009 cohort of lambs, which was the last cohort analysed with

microsatellites. We compared the previously ‘best’ pedigree generated by Michael

Morrissey using microsatellites with one generated using the SNPs. In both cases we

used the parentage inference software MasterBayes (written by Jarrod Hadfield).

In 2009, according to our database, 250 individual lambs were born and identified, of

which 214 had a known mother and 36 no known mother. Meanwhile 91 ewes, of

which 46 were older than yearlings, were alive but did not have a known lamb. 213 of

the identified lambs have been genotyped at microsatellites, and of these 163 or 77%

have a father identified by microsatellites.

Of the 2009 lambs, 209 have so far been genotyped at the SNPs. We first checked

whether we could confirm the field-identified mothers and find mothers for lambs

with no recorded mothers. The results are shown in Table 3. Although not all lambs

or their mothers have yet been SNP genotyped, where they both have, 100% of the

mothers were correct. This is a tribute to both the field identification system and the

record-keeping and sampling labeling procedures in place. Even better, 14 new

mother-lamb relationships were established, involving 12 ewes which had no

Page 23: Annual Report 2011 Final - Soay sheepsoaysheep.biology.ed.ac.uk/.../files/dnussey/AnnualReport2011.pdf1Institute of Evolutionary Biology, University of Edinburgh. 2Macaulay ... 6 Department

23

recorded lamb in that year and two which were recorded with a singleton but actually

had twins, of which one (the newly attributed one) died. These 14 lambs consisted of

seven which were found dead soon after birth and seven which were caught in

summer catches after the time at which they could be mothered-up. When applied

across the whole dataset, these adjustments will change ewe breeding records and

estimates of ewe fecundity and lifetime breeding success.

Table 3. Outcome of maternity analysis of 2009 cohort using 384 SNP loci and

MasterBayes.

Category No of lambs % of genotyped

lambs

Database mother confirmed 161 77.0

SNP genotyped but no mother found. BUT database mother

not yet SNP genotyped

29 13.9

Mother unknown in database but found 14 6.7

Mother unknown in database and not found (yet) 5 2.4

Total genotyped at SNPs 209 100

Next we tested the paternity of the 2009 lambs. Of the 209 SNP genotyped lambs a

father was found for 169 or 81%. This may not seem like a drastic improvement over

the 77% reported above, but two points should be noted. First, not all candidate

fathers have yet been genotyped at the SNPs, so this proportion is likely to rise.

Second, the individual confidence of these paternities is in every case 100 rather than

averaging 98% as in the microsatellite paternities.

Finally, we inspected differences between the two paternity analyses. 199 lambs have

been genotyped by both methods and can be compared. Results are shown in Table 4.

Given that not all candidate sires have yet been SNP-genotyped, the results are most

encouraging. Not only do the SNPs identify a father for an additional 20% of lambs,

but there is every chance they will confirm another 10% of paternities once all

candidate sires are genotyped. It looks as if 95% or more of all lambs may have a

known sire once all the genotypes are in.

Table 4. Comparison of paternity analysis of 2009 cohort by the microsatellite and

SNP methods.

Category No of lambs % of genotyped

lambs

Same sire called 114 57.3

Sire called by SNPs but not microsatellites 40 20.1

Sire called by microsatellites but not by SNPs – in every

case the microsatellite sire has not been SNP genotyped

yet

19 9.6

Neither method calls a sire 17 8.5

Different sire called 9 4.5

Total genotyped lambs 199 100

Page 24: Annual Report 2011 Final - Soay sheepsoaysheep.biology.ed.ac.uk/.../files/dnussey/AnnualReport2011.pdf1Institute of Evolutionary Biology, University of Edinburgh. 2Macaulay ... 6 Department

24

In summary, once the SNP genotyping and parentage analysis is complete, it is going

to greatly increase the completeness, accuracy and confidence of the pedigree. In

particular, the enhanced power to identify mothers will help to resolve the current lack

of information on the maternity of the 2001 lambs, which we were unable to tag at

birth due to foot-and-mouth restrictions in that year.

Genetic architecture of body size in Soay sheep.

Camillo Berenos, Phil Ellis and Josephine Pemberton.

Body size has a strong genetic basis in Soay sheep, as genes are estimated to explain

nearly 50% of the variation in body size. Previous work showing positive selection on

body size suggests that it should be increasing in size over time, but in fact it has

decreased during the course of the study. Ecological processes can explain much of

this change, hence it is uncertain if the sheep have responded genetically to selection.

Ideally we would like to 1) find genomic regions that underlie variation in body size

(called Quantitative Trait Loci or QTL) 2) study selection on these regions and 3)

track allele frequencies of variants at QTL over time.

To investigate this issue, we are genotyping the Soay sample collection at 38,000

variable DNA markers called single nucleotide polymorphisms, (SNP), which are

uniformly distributed across the genome. Here we present a preliminary analysis of a

body size trait (adult hindleg length) at three hierarchical levels, First, we examine

how much of the phenotypic variance can be explained by all screened SNP. Second

we partition phenotypic variance between chromosomes. Finally, we examine

whether we can identify specific genomic locations that explain significant amounts

of variation in body size.

The total variance explained by all SNPs is very similar to the variance explained

using a pedigree-based relationship (Fig. 17), and both methods attribute 45-48% of

phenotypic variance to genes. More than 90% of genetic variance (as estimated using

pedigree relatedness) can be explained by the scored SNPs (Fig. 17). This is in

contrast with results from human genetics studies, where only 50% of additive genetic

variance can be explained by SNPs. This can partly be explained by the higher

amounts of linkage disequilibrium in the Soay sheep, due to possible admixture

events and a historically low effective population size.

Page 25: Annual Report 2011 Final - Soay sheepsoaysheep.biology.ed.ac.uk/.../files/dnussey/AnnualReport2011.pdf1Institute of Evolutionary Biology, University of Edinburgh. 2Macaulay ... 6 Department

25

Figure 17. Comparison of proportion of phenotypic variance explained for adult

hindleg length between models using either pedigree relatedness (dark bars; N = 798

individuals) or realised relatedness (calculated using 38,000 SNPs, light bars;

(N=598 individuals). VA, (additive genetic variance), calculated using pedigree

relatedness reflects heritability, while VA calculated using realised relatedness

reflects the proportion of phenotypic variance explained by all SNPs. Other estimated

variance components are VPE (Permanent environment, among-individual variance

attributed to repeated measures); Capyear, (Year of capture) and Birthyear, (Year of

Birth). Estimates obtained are very similar between the two models, and most of the

heritable variation can be explained by SNPs.

Several chromosomes contribute to variation in body size, while other chromosomes

explain very little (Fig. 18). The sum of variance explained for all chromosome is

50% which is very similar to the variance estimated using all SNPs (Fig. 17).

Figure 18. Variance in adult hindleg length explained by each chromosome. There

are substantial differences between chromosomes in how much they contribute to

phenotypic variation, as several chromosomes explain no variation, while others

explain up to 7 percent of total phenotypic variance (N = 598 sheep).

Page 26: Annual Report 2011 Final - Soay sheepsoaysheep.biology.ed.ac.uk/.../files/dnussey/AnnualReport2011.pdf1Institute of Evolutionary Biology, University of Edinburgh. 2Macaulay ... 6 Department

26

We also examined whether specific SNPs showed significant association with hindleg

length using genome-wide association (GWAS). Some regions show SNPs which

suggest a possible association with adult hindleg length, but no single SNP reached

genome-wide significance after correcting for multiple testing (Fig. 19). A reassuring

finding is that the cluster of SNP suggestive of an association with adult hindleg

length which appears at the end of chromosome 16 seems to confirm previous

findings obtained with the same panel of markers, but using a highly selected sample

of Soays.

Figure 19. Results of genome-wide association of adult hindleg length. (N = 598

sheep).

In summary, we have found that SNPs explain most of the genetic variance in a

quantitative trait, and that this is likely due to the joint effect of many genes with

(relatively) small effect. The results presented here are highly preliminary, as

approximately half the projected number of individuals have been successfully

genotyped thus far. As our available sample size increases we expect more power to

detect statistical significance of chromosome contributions to trait variation and

associations between SNP genotypes and trait values. Despite the lack of statistical

power, our results are very promising, and among the first of their kind in any natural

population.

Page 27: Annual Report 2011 Final - Soay sheepsoaysheep.biology.ed.ac.uk/.../files/dnussey/AnnualReport2011.pdf1Institute of Evolutionary Biology, University of Edinburgh. 2Macaulay ... 6 Department

27

Population dynamics and genetic structuring of the St Kilda field mouse.

Tom Black and Shaun Robertson.

An island subspecies endemic to the St Kilda archipelago, the field mouse Apodemus

sylvaticus hirtensis is being studied in order to provide a comprehensive overview of

its ecology and evolutionary history. Intensive and ongoing live trapping over the last

two years at three sites (Village Bay, Glen Bay and Carn Mor) on Hirta has provided

data and samples that can be used to examine the population dynamics, breeding

ecology, diet, size and genetics of these mice. Some preliminary findings are

discussed below.

Population dynamics:

A core part of the project is to determine the density of mice living at each site, and

how this varies over time. Intensive live trapping of individually marked mice at three

sites has provided data with several possibilities for estimating population densities.

However, this is made complex by the fact that animals are wont to move into and out

of a trapping area. Traditional mark-recapture methods can measure the abundance of

mice within an area, but defining the limits of that area can be problematic. To

overcome this, we have used spatially explicit capture-recapture (SECR) analyses

which provide pure estimates of density without the need to define an effective

trapping area.

The population of St Kilda field mice has fluctuated over the last two years in a

manner typical of small mammals in seasonal habitats (Fig. 20), in which densities are

greatest in late autumn, with as many as 55 mice/ha following summer breeding,

followed by a decline in numbers during winter and spring to between 2 and 10

mice/ha. There are also significant differences between years and between the three

trapping sites, with Carn Mor standing out as having more pronounced swings in

population size. It is hypothesized that this may be due to greater summer food

availability provided by the large numbers of seabirds nesting at the site, but results

from the dietary analysis are not yet available to confirm this.

The models underlying the SECR analysis have also highlighted some quirks of

mouse behaviour, such as a behavioural response to trapping in which individuals are

more likely to be caught when they have encountered a trap already (‘trap happy’

mice), and that range size varies according to season and sex, largely as a result of

increased ranging behaviour in males during the mating season.

Page 28: Annual Report 2011 Final - Soay sheepsoaysheep.biology.ed.ac.uk/.../files/dnussey/AnnualReport2011.pdf1Institute of Evolutionary Biology, University of Edinburgh. 2Macaulay ... 6 Department

28

Gle

n B

ay

0

10

20

30

40

50

60

70

80

90 N

ov-0

9

Jan-

10

Mar

-10

May

-10

Jul-1

0

Sep-1

0

Nov

-10

Jan-

11

Mar

-11

May

-11

Jul-1

1

Sep-1

1

Nov

-11

Number caught (solid line)

Density (mice/ha, dashed line)

Ca

rn M

or

0

10

20

30

40

50

60

70

80

90 N

ov-0

9

Jan-

10

Mar

-10

May

-10

Jul-1

0

Sep-1

0

Nov

-10

Jan-

11

Mar

-11

May

-11

Jul-1

1

Sep-1

1

Nov

-11

Number caught (solid line)

Density (mice/ha, dashed line)

F

igu

re 2

0.

Abundance a

nd d

ensi

ty o

f m

ice o

ver

tim

e,

for

thre

e t

rappin

g s

ites

(Vil

lage B

ay,

Gle

n B

ay &

Carn

Mor)

. G

raphs

show

num

ber

of

indiv

idual

mic

e caught

per

trappin

g s

ess

ion (

soli

d l

ine),

alo

ng w

ith S

EC

R -

reve

ale

d d

ensi

ty e

stim

ate

s (d

ash

ed l

ine)

and 9

5%

confi

den

ce

inte

rvals

(dott

ed l

ine).

Densi

ty e

stim

ate

s are

base

d o

n d

ata

fro

m N

ovem

ber

2009 u

nti

l June

2011 o

nly

.

Villa

ge

Ba

y

0

10

20

30

40

50

60

70

80

90 N

ov-0

9

Jan-

10

Mar

-10

May

-10

Jul-1

0

Sep-1

0

Nov

-10

Jan-

11

Mar

-11

May

-11

Jul-1

1

Sep-1

1

Nov

-11

Number caught (solid line)

Density (mice/ha, dashed line)

Page 29: Annual Report 2011 Final - Soay sheepsoaysheep.biology.ed.ac.uk/.../files/dnussey/AnnualReport2011.pdf1Institute of Evolutionary Biology, University of Edinburgh. 2Macaulay ... 6 Department

Population genetic structuring

Knowing the genetic structure of a population can be important for conservation

planning and provides an insight into the broader ecology of a species. Field mouse

populations have been known to demonstrate considerable genetic subdivision over

relatively small spatial scales despite their propensity for dispersal, and this is also

true of the St Kilda mice. Genotyping of a total of around a hundred individuals from

Dun and the three trapping sites on Hirta at six microsatellite loci found a high level

of genetic differentiation between the two islands (Fig. 21), suggesting that there is

little or no gene flow across the narrow sea gap between them. Furthermore, there is

some evidence of genetic subdivision within the Hirta population; mice from Carn

Mor appear to be divergent from those at the Village and Glen Bay sites. Man-made

obstacles have been cited as a likely cause of genetic structuring in mainland field

mice, but here it seems that the precipitous terrain of St Kilda may also be sufficient

to prevent the population from homogenising.

Figure 21. Estimated population structure of mice from Dun and Hirta, including

prior population sampling information. According to the population genetics software

STRUCTURE, three genetic clusters are present, each represented by a different

colour. Each individual is represented by a vertical line coloured according to the

probability of assignment to each cluster. Individuals are arranged by sample

location (1, Village Bay; 2, Glen Bay; 3, Carn Mor; 4, Dun).

The other major finding of this preliminary genetics work is that there is clearly a

very low level of genetic diversity in the St Kilda field mouse population, presumably

as a result of it being founded by a small number of individuals. In fact, with a

maximum of 6 alleles (mean 3.2) at any one locus it is possible that a single pregnant

female could have founded the entire population, given that multiple paternity litters

are common in field mice.

Page 30: Annual Report 2011 Final - Soay sheepsoaysheep.biology.ed.ac.uk/.../files/dnussey/AnnualReport2011.pdf1Institute of Evolutionary Biology, University of Edinburgh. 2Macaulay ... 6 Department

30

Publications

Childs, D. Z., Coulson, T. N., Pemberton J., Clutton-Brock T. & Rees, M. (2011)

Predicting trait values and measuring selection in complex life histories:

reproductive allocation decisions in Soay sheep. Ecology Letters 14, 985-992.

Colchero F. & Clark, J. S. (2011) Bayesian inference on age-specific survival for

censored and truncated data. Journal of Animal Ecology, 81, 139-149.

Di Fonzo, M. M. I., Pelletier, F., Clutton-Brock, T. H., Pemberton, J. M., & Coulson,

T. (2011) The population growth consequences of variation in individual

heterozygosity. PloS one 6:e19667.

Hayward, A. D., Wilson, A. J., Pilkington, J. G., Clutton-Brock, T. H., Pemberton, J.

M., & Kruuk, L. E. B. (2011) Natural selection on a measure of parasite

resistance varies across ages and environmental conditions in a wild mammal.

Journal of Evolutionary Biology 24:1664-1676.

Johnston, S. E., McEwan, J. C., Pickering, N. K., Kijas, J. W., Beraldi, D., Pilkington,

J. G., Pemberton, J. M., & Slate, J. (2011) Genome-wide association mapping

identifies the genetic basis of discrete and quantitative variation in sexual

weaponry in a wild sheep population. Molecular Ecology 20: 2555-2566.

Nussey, D. H., Watt, K., Pilkington, J. G., Zamoyska, R. & McNeilly, T. N. (2012)

Age-related variation in immunity in a wild mammal population. Aging Cell

11, 178-180.

Preston, B. T., Stevenson, I. R., Lincoln, G. A., Monfort, S. L., Pilkington, J. G., &

Wilson, K. (2012) Testes size, testosterone production and reproductive

behaviour in a natural mammalian mating system. Journal of Animal Ecology:

296-305.

In press:

Morrissey. M. B., Parker, D., Korsten, P., Clutton-Brock, T. H., Pemberton, J. M.,

Kruuk, L. E. B., & Wilson, A. J. The prediction of adaptive evolution:

empirical application of the secondary theorem of selection and comparison to

the breeder’s equation. Evolution.

Page 31: Annual Report 2011 Final - Soay sheepsoaysheep.biology.ed.ac.uk/.../files/dnussey/AnnualReport2011.pdf1Institute of Evolutionary Biology, University of Edinburgh. 2Macaulay ... 6 Department

31

A C K N O W L E D G E M E N T S

We are grateful to the National Trust for Scotland and to Scottish Natural Heritage for

permission to work on St Kilda, and for their assistance in many aspects of the work.

The project would not be possible without the generous assistance and support of MOD,

QinetiQ and E.S.S. staff stationed on St Kilda and Benbecula and servicing the island.

We are particularly grateful to Susan Bain, the Western Isles Manager for the NTS, Paul

Sharman the NTS Ranger for St. Kilda, to Ian McHardy and Carol Knott the

Archaeologists on the island, and to Gina Prior the Seabird and Marine Ranger.

We are also grateful for the help of volunteers without whom the fieldwork for 2011

would not have been possible: Gebre Asefa, Billy Craigens, Dean Dowden, Katie

Hatton, Rebecca Hewitt, Linda Kehoe, Deborah Leigh, Ruari Macleod, Oliver Moore,

Anne O’Callaghan and Federico Tettamanti. Thank you.

Our research is supported by grants and studentships from the Natural Environment

Research Council, the Biotechnology and Biological Sciences Research Council, the

Royal Society and the European Research Council.

AP P E N D I X A: PE R S O N N E L NEWS & SC H E D U L E O F WO R K

Personnel News

Jill Pilkington was awarded an MBE for her services to Science. This was conferred

by HRH Prince Charles on November 18th

at Buckingham Palace.

In late 2010 Camillo Berenos and Phil Ellis joined the Project, working on the

genomics of Soay sheep with Josephine Pemberton, funded by the European Research

Council.

Jen Dorrens has joined the project as a post-graduate research assistant, and will be

studying telomere length as part of a project in collaboration with Lea Harringon, Dan

Nussey and Josephine Pemberton.

Louise Christensen has completed her undergraduate honours thesis at the University

of Aberdeen investigating oxidative stress and ageing, as part of an ongoing

collaboration involving Colin Selman, Jon Blount and Dan Nussey.

Schedule of work on St Kilda

Winter - Spring

Jill Pilkington monitored mortality during February and with volunteers, throughout

lambing. During this period, detailed data were collected on individual sheep found

dead, and samples were taken for genetic and parasitological study.

Page 32: Annual Report 2011 Final - Soay sheepsoaysheep.biology.ed.ac.uk/.../files/dnussey/AnnualReport2011.pdf1Institute of Evolutionary Biology, University of Edinburgh. 2Macaulay ... 6 Department

32

From March 18th

until May 6th

, Jill Pilkington and 3 volunteers carried out ten

population censuses and tagged and sampled lambs, with assistance during the peak

of lambing from Dan Nussey. 270 lambs were born to 260 ewes; these figures include

10 sets of twins (7 ewes held both lambs, 2 lost one lamb and 1 ewe lost both lambs).

161 lambs (80 male and 81 female) were caught and tagged; a further 59 lambs died

before any tagging attempt. Mick Crawley and two assistants collected vegetation data.

Tom Black and one volunteer conducted trapping sessions in March and May on Hirta.

Summer

Jill Pilkington and two volunteers returned to Hirta on July 12th to carry out ten

population censuses, conduct mortality searches (yielding 11 tagged dead animals),

and prepare for the main catch-up of study area sheep. The catch-up took place from

August 5th

– 19th, was led by Josephine Pemberton, and conducted by a team of 11

additional project members and volunteers. 288 sheep were caught and processed, of

which 100 were lambs (50 males and 50 females), 23 were yearlings (5 males and 18

females), 44 were adult males, and 121 were adult females. All animals were weighed

and measured to monitor growth, and sampled for parasite and genetic analyses. 25

Sheep were retagged because of damaged or missing tags. 17 previously untagged

lambs and 1 yearling were caught and processed. Mick Crawley and two assistants

collected vegetation data. Jill Pilkington and two volunteers remained on Hirta until

3rd

September to complete parasite counts and pasture larvae counts.

Tom Black and one volunteer conducted trapping sessions in August/September.

Autumn

From October 18th

to December 6th

Jill Pilkington, Alastair Wilson and one volunteer

carried out ten population censuses, monitored the mating period, capturing and

processing 3 incoming tups and 1 resident tup and ewes. 44 previously darted, non-

resident tups were seen in the study area during this rut. 4 dead sheep were found.

Tom Black and one volunteer conducted trapping sessions in November/December.

Page 33: Annual Report 2011 Final - Soay sheepsoaysheep.biology.ed.ac.uk/.../files/dnussey/AnnualReport2011.pdf1Institute of Evolutionary Biology, University of Edinburgh. 2Macaulay ... 6 Department

33

C I R C U L A T I O N L I S T - (Please advise J.Pilkington of any changes or additions)

Prof. S. Albon Macaulay Institute, Craigiebuckler, Aberdeen, AB15 8QH.

Ms. S. Bain NTS, Balnain House, 40 Huntly St., Inverness, IV3 5HR.

Dr. D. Bancroft GPC AG, Lochhamer Str. 29D-82152, Munich, Germany.

Mr. A. Bennett NTS, Balnain House, 40 Huntly St., Inverness, IV3 5HR.

Ms. A. Bento Dept. Biological Sciences, Imperial College, Silwood Park, Ascot, SL5 7PY.

Dr D. Beraldi Roslin Institute Edinburgh Univ., Roslin, Edinburgh EH25 9PS.

Dr. C. Berenos Inst. Evol. Biol., Edinburgh Univ., West Mains Rd, Edinburgh EH9 3JT.

Mr. T. Black Inst. Evol. Biol., Edinburgh Univ., West Mains Rd, Edinburgh EH9 3JT.

Mr. J. Blount Centre for Ecology and Conservation, Univ. Exeter, Cornwall Campus, TR10 9EZ

Ms. E. Brown Dept. of Animal and Plant Sciences, Univ. Sheffield, S10 2TN.

Dr. D. Childs Dept. of Animal and Plant Sciences, Univ. Sheffield, S10 2TN.

Dr. D. Clements Royal (Dick) School of Veterinary Sciences, Edinburgh Univ., Easter Bush, EH25 9RG

Prof. T. Clutton-Brock Dept. Zoology, Cambridge Univ., Downing St., CB2 3EJ.

Dr. D. Coltman Dept. Biol. Sci., Univ. Alberta, Edmonton AB, T6G 2E9, Canada.

Dr. T. Coulson Dept. Biological Sciences, Imperial College, Silwood Park, Ascot, SL5 7PY.

Dr. B. Craig Wildlife, Ecology and Management Group, Central Sc. Lab., York, YO41 1LZ.

Prof. M. Crawley Dept. Biological Sciences, Imperial College, Silwood Park, Ascot, SL5 7PY.

Dr. S. Davies SNH, Fraser Darling House, 9 Culduthel Road, IV2 4AG.

Ms. J. Dorrens Inst. Evol. Biol., Edinburgh Univ., West Mains Rd, Edinburgh EH9 3JT.

Mr. P. Ellis Inst. Evol. Biol., Edinburgh Univ., West Mains Rd, Edinburgh EH9 3JT.

Dr. T. Ezard Dept. Biological Sciences, Imperial College, Silwood Park, Ascot, SL5 7PY.

Ms. J. Ferguson SNH, Stilligarry, South Uist, HS8 5RS.

Dr. P. Feulner Westfälische Wilhelms Univ., Inst. Evol. and Biodiv., Hüfferstrasse,

148149 Münster, Germany.

Dr. A. Graham Dept. Ecol. Evol. Biol., Guyot Hall, Princeton Univ., NJ 08544 2016, U.S.A. Dr. J. Gratten Queensland Inst. Med. Res., PO Royal Brisbane Hospital, Q4029, Australia.

Prof. B. Grenfell Dept. Ecol. Evol. Biol., Guyot Hall, Princeton Univ., NJ 08544 2016, U.S.A. Dr. F. Gulland TMMC, Marin Headlands, Sausalito, CA 94965, USA.

Dr. J. Hadfield Inst. Evol. Biol., Edinburgh Univ., West Mains Rd, Edinburgh, EH9 3JT.

Ms. J. Harden NTS, Balnain House, 40 Huntly St., Inverness, IV3 5HR.

Mr. A. Hayward Inst. Evol. Biol., Edinburgh Univ., West Mains Rd, Edinburgh, EH9 3JT.

Prof. L. Harrington Université de Montréal, Institute de Recherche en Immunologie et en Cancérologie,

Montréal, Canada

Prof. A. Illius Inst. Evol. Biol., Edinburgh Univ., West Mains Rd, Edinburgh, EH9 3JT.

Dr. S. Johnston Dept. of Animal and Plant Sciences, Univ. Sheffield, S10 2TN.

Dr. O. Jones Inst. Zoology, ZSL, Regent’s Park, London NW1 4RY..

Dr. P. Korsten Inst. Evol. Biol., Edinburgh Univ., West Mains Rd, Edinburgh, EH9 3JT.

Dr. L. Kruuk Inst. Evol. Biol., Edinburgh Univ., West Mains Rd, Edinburgh, EH9 3JT.

Dr. G. Lincoln MRC Centre for Rep. Biol., 49 Little France Cres., Edinburgh, EH3 9EW.

Mr. J. Love The Watchers Cottage, Snishival, South Uist, HS8 5RW.

Dr. R Luxmoore NTS, 28 Charlotte Square, Edinburgh, EH2 4DU.

Dr. A. MacColl School of Biology, Univ. of Nottingham, NG7 2RD.

Mr. D. MacLennan SNH, 17 Frances St., Stornoway. Lewis, Outer Hebrides.

Prof. J Matthews Moredun Research Institute, Edinburgh.

Dr T. McNeilly Moredun Research Institute, Edinburgh.

Mr. A. McRae Queensland Inst. Med. Res., PO Royal Brisbane Hospital, Q4029, Australia.

Dr. R. Mellanby Royal (Dick) School of Veterinary Sciences, Edinburgh Univ., Easter Bush, EH25 9RG

Dr. J. Milner Hogskolen i Hedmark, Evenstad, NO2480, Koppang, Norway.

Prof. B. Morgan Inst. Maths.& Stats., Univ. Kent., Canterbury, Kent, CT2 7NF.

Ms. K. Morriss Dept. Ecol. Evol. Biol., Guyot Hall, Princeton Univ., NJ 08544 2016, U.S.A. Dr. M. Morrissey Inst. Evol. Biol., Edinburgh Univ., West Mains Rd, Edinburgh, EH9 3JT.

Mr. S. Murray Craigie Dhu, Cardney, Dunkeld, Perthshire, PH8 0EY.

Dr A. Nisbet Moredun Research Institute, Edinburgh.

Dr. D. Nussey Inst. Evol. Biol., Edinburgh Univ., West Mains Rd, Edinburgh EH9 3JT.

Dr. A. Ozgul Dept. Biological Sciences, Imperial College, Silwood Park, Ascot, SL5 7PY.

Prof. S. Paterson School of Biological Sciences, Univ. of Liverpool, L69 7ZB.

Dr. F. Pelletier Dept. Biologie, Univ. of Sherbrooke, Quebec, Canada, J1K 2RI.

Prof. J. Pemberton Inst. Evol. Biol., Edinburgh Univ., West Mains Rd, Edinburgh EH9 3JT.

Mrs J. Pilkington Inst. Evol. Biol., Edinburgh Univ., West Mains Rd, Edinburgh EH9 3JT.

Dr. B. Preston Max Planck Inst. Evol. Anthropology, 04103 Leipzig, Germany.

Dr. G. Prior Macaulay Institute, Craigiebuckler, Aberdeen, AB15 8QH.

Dr. M Rees Dept. of Animal and Plant Sciences, Univ. Sheffield, S10 2TN.

Mr. S. Robertson Inst. Evol. Biol., Edinburgh Univ., West Mains Rd, Edinburgh, EH9 3JT.

Page 34: Annual Report 2011 Final - Soay sheepsoaysheep.biology.ed.ac.uk/.../files/dnussey/AnnualReport2011.pdf1Institute of Evolutionary Biology, University of Edinburgh. 2Macaulay ... 6 Department

34

Dr. M. Robinson Dept. of Animal and Plant Sciences, Univ. Sheffield, S10 2TN.

Dr. P. Scott Royal (Dick) School of Veterinary Sciences, Edinburgh Univ., Easter Bush, EH25 9RG

Dr. C. Selman Inst. Of Biological and Environmental Sciences. Univ., Aberdeen, AB24 2TZ

Prof. J. Slate Dept. of Animal and Plant Sciences, Univ. Sheffield, S10 2TN.

Dr. R. Stevens Dept. of Archaeology, University of Cambridge, Downing St., CB2 3ER.

Dr. I. Stevenson Sunadal Data Solutions, Midlothian, Innovation Centre, Roslin, EH25 9RE.

Dr. G. Tavecchia Imedea-CSIC/UIB, c. M. Marques 21, 07190 – Esporles, Mallorca, Spain.

Dr. L. Tempest 7 Mandrake Road, London, SW17 7PZ.

Dr. P. Visscher Queensland Inst. Med. Res., PO Royal Brisbane Hospital, Q4029, Australia.

Dr. S. Votier Sch. of Biomed. & Biol. Sci., Davy Building, Plymouth, Devon, PL4 8AA.

Ms. K. Watt Inst. Evol. Biol., Edinburgh Univ., West Mains Rd, Edinburgh EH9 3JT.

Dr. A. Wilson Inst. Evol. Biol., Edinburgh Univ., West Mains Rd, Edinburgh EH9 3JT.

Dr. K. Wilson Dept. of Biological Sciences, Lancaster University, LA1 4YQ.

Prof R. Zamoyska Institute of Immunological and Infection Research, University of Edinburgh