Monitoring grey seal populations using individual identification
Final Scientific Report on Project MF0707
September 1998
NERC Sea Mammal Research Unit, Gatty Marine Laboratory
University of St Andrews, St Andrews, Fife KY16 8LB
Executive summary
This report presents an analysis of grey seal photo-identification data collected in the North Sea from 1991 to 1996 to estimate local population levels in summer (as distinct from population sizes in autumn/winter estimated by SMRU from pup counts as part of NERC’s responsibilities under the Conservation of Seals Act 1970). Rates of interchange between the main haulout areas and the rate of turnover of the population possessing distinguishing marks were also estimated.
Data were collected by photographing seals on summer haul-outs in the Humber Estuary, Farne Islands, and Forth and Tay Estuaries. In 1993 and 1994, seals were also photographed in the Dornoch Firth and Orkney Islands. These areas include all significant grey seal haul-outs in the North Sea during the summer months.
Acceptable photographs were scanned into a computer and processed to extract a sample of the pelage pattern on the side of the seal’s head to act as unique identifying marks. All extracted images were compared with each other to create a database of so-called capture histories of each recognizable animal.
Analysis of photo-identification data using mark-recapture methods to estimate population size is not straightforward and specific analytical methods were developed for this particular study. They are described in detail in the Scientific Report.
Almost 23,000 images were processed and over 100 million comparisons of photographs were made to create the database of capture histories. The minimum number of seals represented in the database is 1,528. The analysis estimated the number of seals at each main haul-out area and in total. The total population estimates for summer, of approximately 5,500 seals (range 5310-5682), were consistent across all six years except for the last, 1996, for which no supportable explanation can be found. Estimated numbers at each area varied more among years (Humber: range 817-1250; Farnes: 1929-2851; Tay+Forth: 1920-2496). Estimated precision of the population estimates was very good with coefficients of variation of about 5% for the total.
The average total population estimate for the same areas for 1991-1996 based on pup counts during the November-December pupping season is around 9000 seals. This is about 50% larger than the summer population as estimated here. It is not currently known whether this reflects a truly smaller population in summer or whether the summer estimates are biased downwards because of the effects of individual differences in the probability of photographing animals. This difference can be resolved by explicitly modelling such differences and will be important in any future work to assess fish consumption by grey seals in the North Sea.
Rates of interchange between the main haulout areas were calculated as transition probabilities as described in the Scientific Report. The results show that the average probability of resighting a seal in the same area is very high (94-97%) with very low rates of exchange between areas.
The rate of turnover of the population possessing distinguishing marks was estimated at about 15% per annum. This results from a combination of recruitment (to the marked population), death, immigration and emigration.
The results of this project will be used to help formulate and parameterize the models being developed as part of project MF0515.
Introduction
The Natural Environment Research Council (NERC) has a statutory obligation under the Conservation of Seals Act 1970 to provide advice to government on matters related to the management of seal populations around Britain. NERC discharges this responsibility through its Special Committee on Seals (SCOS) which meets annually to formulate advice to government based on research undertaken by the Sea Mammal Research Unit (SMRU). This research centres on the regular monitoring of grey and harbour (common) seal populations throughout Great Britain.
For grey seals, this monitoring takes the form of annual aerial surveys of breeding colonies during the pupping season (October-November). Counts of white-coated pups on aerial photographs taken at intervals during the pupping season allow pup production to be estimated for each breeding colony. From the annual pup production estimates, total population size is estimated from a population model. Trends in population numbers can then be investigated for the main breeding areas: Outer Hebrides, Inner Hebrides, Orkney and the east coast of Great Britain (comprising the Isle of May, Farne Islands and Humber Estuary).
Population numbers and trends calculated under SMRU’s core research programme give good indications of changes over time in the size of the breeding population of grey seals around Britain. But it is the geographical and seasonal distribution and abundance at sea during foraging that is of most relevance when considering the interactions between grey seals and fish stocks. Grey seals are known (from SMRU’s core research and MAFF-funded work under projects MF0503, MF0309 and MF0311) to travel extensively during the majority of the year when they are foraging rather than breeding. Although much foraging occurs close to haul out sites, these sites are not necessarily the same as those used for breeding. It is not possible, therefore, to infer the distribution and abundance of foraging seals from data collected during the breeding season.
To estimate the numbers of foraging grey seals, SMRU has undertaken a six year project on seals associated with east coast haul out sites during summer; the time when they are foraging to build up fat reserves for breeding in autumn. This work was funded in full by MAFF during the first three years and then 50% by MAFF for the second three years. The objectives of the second three-year project (MF0707) were:
- to estimate the number of grey seals associated with haul out sites along the east coast of Britain during the summers of 1994 to 1996;
- to estimate rates of interchange among haul out sites within this area;
- to use the results to inform the modelling work under project MF0311.
Methods
Details of the data collection and analysis methods are given in the attached Technical Report Abundance estimates for grey seals in summer based on photo-identification data. The following is a brief summary.
Data Collection
Data were collected by photographing the natural pelage markings of seals on summer haul-outs in the Humber Estuary, Farne Islands, and Forth and Tay Estuaries. These areas include all significant grey seal haul-outs in the North Sea during the summer months. In 1993 and 1994, seals were also photographed in the Dornoch Firth and Orkney Islands
Photographs were taken of all ‘well-patterned’ seals, that is those that had a distinct black and white pattern certain to be recognisable from a subsequent photograph. Such seals were mostly adult females. The pelage pattern is not the same on both sides of the animal so, whenever possible, photographs were taken of both left and right sides.
Data processing
Acceptable photographs were scanned into a computer and processed to extract a sample of the pelage pattern on the side of the seal’s head to act as unique identifying marks. All extracted images were compared with each other to create a database of capture histories of each recognisable animal. Purpose-written software was written to accomplish this.
Almost 23,000 images were processed to establish a catalogue of identifiable individual grey seals, each with a history of when and where it was photographed over the study period. Over 250 million comparisons are implicit in the process of matching these images but more than half of these were avoided by pre-sorting images obtained during the same photographic session. The rest of the comparisons were performed using specialised software. All potential matches were confirmed by inspecting the pair of original images by eye.
The catalogue holds 6,475 seal identifications represented by at least one adequate photograph taken within the three main haul out area (Isle of May, Farne Islands, Humber estuary). For about 60% of these, at least one photograph also survived the second screening, designed to limit the number of missed matches. The catalogue contains 1,145 seals photographed from the left hand side and 1,528 seals photographed from the right hand side. It is not possible to say exactly how many seals have been photographed from both left and right sides, but the minimum number of seals is at least 1,528.
Data analysis
Analysis of photo-identification data using capture-recapture methods to estimate population size is not straightforward and specific analytical methods were developed for this particular study. In particular, capture-recapture methods make a number of assumptions for the estimates to be valid regarding equal probability of capture, not missing matches between photographs, animal markings not changing, and population closure. Aspects relating to pelage pattern quality, photographic quality, the risk of missing matches and the loss of patterns over time were addressed in the analysis (see Technical Report).
Preliminary analysis showed that the percentage of seals recaptured at any one site was initially high for a few days but then declined and stabilised at about 10% by the time the interval between sampling dates reached two weeks, and stayed at 10% even up to 2.5 years. This is important for two reasons. First, it means that the population mixes sufficiently to allow capture-recapture analyses based on samples two weeks apart to provide unbiased (in this respect) estimates of population size. Second, the fact that the recapture percentage does not decline below 10% even after a period of years indicates that seals have preferences for particular haul out sites. If movements between areas were governed simply by Markov transition probabilities, the recapture percentage would continue to decline over time.
This notion of fidelity to a particular area is reinforced by observing the number of recaptures within and between areas, as shown in Table 1.
Table 1. Number of different animals recaptured within and between main areas
Humber estuary / Farne Islands / Tay/Forth estuariesHumber estuary / 152 / 5 / 1
Farne Islands / 9 / 294 / 33
Tay/Forth estuaries / 1 / 22 / 408
The methods developed were based on a model which assumes population closure. This assumption was not met because of the normal population processes of recruitment, death, immigration and emigration over the six years of the study. Recruitment in this context means recruitment to the population of marked animals, which happens at an as yet unknown age. It was also not met because of the loss of some patterns over time as the pelage of young males seals darkens as they get older. To account for this, the model included a parameter describing the annual turnover rate of the population of marked animals. This parameter was estimated as 15% per annum.
Rates of interchange between the main haul out areas were included in the model as probabilities that a seal photographed in one area would next be photographed in the same or a different area. For ease of analysis, these probabilities (which can also be thought of as representing preferences for certain areas) were assumed to remain constant within a year but allowed to vary among years.
The model finally derived (see Technical Report) was judged to meet the assumptions of the method sufficiently well to provide reliable population estimates.
Results
Table 2 gives the population estimates for each main area for each year of the study.
Table 2. Estimates of the number of grey seals (coefficient of variation derived by simulation in parentheses) associated with the three main haul out sites along the east coast of Britain during summer.
1991 / 1992 / 1993 / 1994 / 1995 / 1996Humber estuary / 885 (.14) / 847 (.17) / 817 (.14) / 1250 (.11) / 1135 (.11) / 2076 (.20)
Farne Islands / 1929 (.08) / 2851 (.07) / 2288 (.09) / 2422 (.11) / 2436 (.07) / 2109 (.09)
Tay/Forth estuaries / 2496 (.08) / 1920 (.06) / 2461 (.09) / 2011 (.06) / 1945 (.06) / 2227 (.07)
Total / 5310 (.05) / 5618 (.05) / 5566 (.06) / 5682 (.06) / 5515 (.04) / 6412 (.08)
The average total estimated number of seals during the period was 5,684.
These estimates include the corrections incorporated for selecting well-patterned seals, for the risk of failing to match photographs and for the turnover rate. The probabilities of locating a match were estimated for each year 1991-1996, respectively, as follows: 0.925, 0.928, 0.930, 0.955, 0.955, 0.930.
Rates of interchange between the main haul out areas (modelled as probabilities that a seal photographed in one area would next be photographed in the same or a different area) averaged over all years of the study are shown in Table 3.
Table 3. Average probability (over all years) that a seal photographed in one area would next be photographed in the same or another area.
Humber estuary / Farne Islands / Tay/Forth estuariesHumber estuary / 0.973 / 0.026 / 0.001
Farne Islands / 0.016 / 0.941 / 0.043
Tay/Forth estuaries / 0.001 / 0.044 / 0.955
The table shows that the average probability of resighting a seal in the same area was very high (94% - 97%) but much lower between the Tay/Forth estuaries and the Farne Islands (~4%) and the Farne Islands and the Humber estuary (~2%). The resighting probabilities between the Tay/Forth and Humber estuaries were very low.
The resighting probabilities within an area shown in Table 3 are averages of within-year resighting probabilities. The high resighting probabilities within each area are thus not inconsistent with the changes in estimated numbers associated with each area shown in Table 2.
Discussion
Summer population estimates
The population estimates show complementary shifts between the Tay/Forth estuaries and the Farne Islands between 1991 and 1993 but the total estimates between 1991 and 1995 remained approximately the same at about 5,500 seals (range 5,310 – 5,682). In 1996 however, there was a large increase in the estimated number of seals associated with the Humber estuary haul out site and a corresponding increase in total estimated population size. A number of checks on the data base were made to search for errors which could have caused this high estimate but none were found. Furthermore, there were no differences in the sampling strategy or any difference in the treatment of the photographs between 1995 and 1996 and between the Humber estuary and other areas.
If the increase in number of animals at the Humber estuary is real, this could have resulted from a large recruitment of young animals to the patterned component of the population in this area between 1995 and 1996. It is not clear why this may have occurred. An alternative explanation is that the seals photographed at the Humber included a high proportion of young males and that a high proportion of them lost their markings when their patterns became indistinct on becoming sexually mature between 1995 and 1996. This seems unlikely and, again, it is not clear why it may have occurred.
Estimated precision of the population estimates was generally very good; coefficients of variation (CVs) were typically of the order of 10% or less for the Tay/Forth estuaries and the Farne Islands, and slightly higher for the Humber estuary. However, precision for the 1996 Humber estuary estimate was lower with a higher CV of 20%. Because of this, the 1996 estimate of 2,076 is not significantly different from the next highest estimate of 1,250 (CV = 11%) for 1994, although it is different from all others for that area.
Comparison of summer and autumn population estimates
The average total breeding population during 1991-1996 estimated from pup counts during the autumn for the Tay/Forth estuaries, Farne Islands and Humber estuary is around 8,000 seals. This is about 40% larger than the average total summer population in the same areas estimated here. The two main possible reasons for this difference are: a proportion of the breeding population goes elsewhere to forage in summer; the summer population estimates are biased downwards because of the effects of individual differences (heterogeneity) in the probability of photographing animals.
- Breeding population foraging outside the study area
Satellite tracking work under projects MF0503, MF0309 and MF0311 has shown that individual grey seals can travel to and forage around areas which are far removed from the site at which they were captured. In particular, some individuals from the Farne Islands and Tay/Forth estuaries have travelled to, and foraged around, the Orkney Islands and beyond.
To explore further the idea that summer numbers are lower because some seals forage outside the study area, the autumn and summer population estimates for the Tay/Forth estuaries, the Farne Islands and the Humber estuary were compared separately. Results are shown in Table 4.
Table 4. Comparison of estimated number of grey seals during the autumn pupping season and during the summer. (Numbers in parentheses for the Humber estuary exclude the 1996 estimate).
Mean pupping season estimate / Mean summer estimate / Ratio of mean pupping to mean summer estimate / Pearson’s correlation among annual estimatesHumber estuary / 787 / 1,168 (986) / 0.67 (0.80) / 0.672 (0.918)
Farne Islands / 3,060 / 2,339 / 1.31 / 0.165
Tay/Forth estuaries / 4,140 / 2,177 / 1.90 / 0.125
Farnes + Tay/Forth / 7,200 / 4,516 / 1.59 / - 0.230
The results show that while the estimated number of seals associated with the Humber estuary is less during the autumn than during summer, the reverse is true for the Farne Islands and, even more so, for the Tay/Forth estuaries. Furthermore, the correlation between annual estimates in the autumn and summer is high (very high if the 1996 estimate is excluded) for the Humber estuary but low for the more northern sites, including the Tay/Forth estuaries and Farne Islands combined.