EXPERIENCES IN WINDERMERE WITH ESTIMATING POPULATION NUMBERS BY TAG-RECAPTURE METHODS
EXPERIMENTATIONS D'EVALUATION DE POPULATIONS AU MOYEN DES METHODES DE MARQUE-RECAPTURE A WINDERMERE

by/par

C. Kipling and E.D. Le Cren
Freshwater Biological Association
The Ferry House, Ambleside
England/Angleterre

ABSTRACT

This paper concerns attempts to estimate population numbers of char (Salvelinus willughbii Gunther), perch (Perca fluviatilis L.) and pike (Esox lucius L.) in Windermere by tag-recapture methods. Satisfactory results were obtained for char and pike, but not for perch.

RESUME

Ce document illustre les essais d'évaluation de populations d'omble (Salvelinus willughbii Günther), de perche fluviatile (Perca fluviatilis L.) et de brochet (Esox lucius L.) au moyen des méthodes de marque-recapture, à Windermere. On a obtenu des résultats satisfaisants pour la perche et le brochet, mais pas pour la perche fluviatile.

CONTENTS

1. CHAR

2. PIKE

3. PERCH

4. CONCLUSIONS

REFERENCES

1. CHAR

The estimation of population numbers of char (Salvelinus willughbii Günther) by tag-recapture methods has been described in detail by Le Cren and Kipling (1963). In these experiments two quite different problems were under consideration. Firstly the pattern of movement within the spawning season was studied and secondly population parameters were estimated from catches in different years. Only the latter problem, the year-to-year (between seasons) population study will be considered here; for this work sub- cutaneous tags and fin cuts were used (Le Cren, 1954). In Windermere some of the char spawn in November in shallow water in the lake or in the river Brathay and others spawn in February in deep water in the lake (Frost, 1965). During the spawning season (which lasts about a month) the char congregate and can be caught quite easily. Individuals that spawn in a particular place return to that place in successive seasons with a very high consistency and each place therefore appears to have a discrete spawning population. The populations of four spawning places were studied out of a total number of at least twelve in the lake. All the fish were caught (or recaught) during their spawning seasons by the Fresh- water Biological Association staff except for a very few which were recaught by anglers.

Weather conditions caused difficulties in the river Brathay. In 1951 it was in flood for a period of twelve days in the middle of the season and no nettings were possible at this time. The catch of males, who stay longer than the females on the site, was not greatly affected, but the catch of females was very much reduced, as presumably most of them came and went during the time netting was impossible. In 1953 the river was in flood for almost the whole of the spawning season, and only one effective netting could be done. Because of this the catches of both males and females were very small and were not used in the calculations.

Char were tagged subcutaneously and given a distinctive fin cut at Brathay and Red Nab in 1950 and 1951, at Low Wray in 1950 and at Holbeck in 1951 and 1952. Table I gives for some of these experiments the numbers of char tagged (shown in parentheses) and the numbers recaptured in subsequent years. The question of homing is not being dealt with in this paper; it is necessary however for our present purpose to mention that all these recaptures were made at the same spawning site as the one at which the fish had originally been tagged. From these data estimates of population numbers have been calculated and some results are shown in Table II. It must be stressed that all these results refer only to adult mature fish, which would be aged at least four years (Frost, 1951) when first captured and tagged; many would be five or more years old.

From the numbers of recaptures of tagged fish and fish which had lost their tags but which were identified by cut fins, estimates of population size and their standard errors were calculated for all four spawning sites by direct proportion using Bailey's modified formulae (Bailey, 1951). Wherever possible these estimates were calculated separately for recaptures after one and more years and for double recaptures, i.e., fish which were recaught both one and two years after tagging. Allowance was made for recruitment of young fish to the spawning populations by using the length-frequency distributions of the total catches in the years after tagging to identify and omit all fish which would have been too small to have formed part of the original population tagged. In doing so the known increments in length put on by tagged fish have been used to estimate growth during this period. In calculations using double recapture allowance was made for fish which had lost their tags, which could not therefore be identified as double recapture, by assuming that 20 percent of all tags were lost (Le Cren, 1954).

Only at Brathay and Red Nab was tagging done on a sufficient scale in two successive years to enable Bailey's triple-catch method to be applied. Population estimates, recruitment rates and survival rates and their standard errors were calculated. Standard errors were found by the Holt-Spicer method (Holt, 1958). For Holbeck, where recaptures were made over a longer period, the combined method of Holt-Spicer was also used to find standard errors.

Further estimates of population size were made graphically by extrapolation of the lines fitted to the logarithms of the numbers of recaptured fish plotted against time. The ratio of the intercept of this line with the vertical at the time of tagging to the number of fish tagged gave an estimate of the proportion of the population caught at the time of tagging; hence, multiplication gave an estimate of the total population at that time.

The population estimates in Table II arrived at by various methods were remarkably consistent, with a few exceptions. It is clear that the spawning population at Brathay was much smaller than the other populations studied, and that numbers declined between 1950 and 1951. This decline was also found at Red Nab, the only other place for which figures for these years were available. This fall in population numbers was almost certainly connected with the recruitment rates, which, as found from length frequency-distributions, were very low for the period 1950 to 1951. The estimates at Red Nab made by the extrapolation method were too low, with small numbers of recaptures this method is very unreliable, and catches at Red Nab are always rather erratic.

The estimates made for Holbeck based on recaptures after two years' absence and on double recaptures were definitely too high, the males more so than the females. All these estimates ultimately depend on the estimated number of recruits present in the population in the second year after tagging. How the spring spawners (Holbeck) grow faster than the autumn spawners and the males grow faster than the females. It is probable that the faster- growing of the recruits had after two years grown larger than the smaller of the tagged fish, and that therefore the method used to separate recruits from the rest of the population was not in this case satisfactory, insufficient allowance being made for recruitment. The males would be the more affected, as they grow faster than the females.

The very large standard errors of some of the population estimates emphasize the unreliability of estimates based on few recaptures, as in some cases only one or two tagged fish were recaught.

In round figures, assuming that the Red Nab and Low Wray Bay populations are equal, the population of the three autumn spawning sites studied was in the winter 1951/52 approximately 2 000 male and 9 000 female char, and of the one spring spawning site 700 male and 1 300 female char. This gives a total of 13 000 char from the four spawning sites; there are probably at least twelve spawning sites in the lake but most of these have not been studied, therefore no accurate estimate of the total number of char in the lake can be made.

2. PIKE

The pike tagging experiments have been described briefly by Kipling and Frost, 1970. The results were used as a check on other methods of population estimation, not as the sole source of information on population numbers. For this present paper further calculations have been done to find out the differences between estimates of the total pike population of Windermere based on tagging methods and those based on methods which require catch data over many years. The majority of tagged pike are recaught in the first year after tagging, so estimates of population numbers by this method can be made much more quickly than by the other methods.

The tags used were loops of monel metal and were punched with a six-figure number for individual identification. In the years from 1949 to 1952 the tags were attached to the opercular bone, but it became apparent from the many fish recaptured with damaged operoular bones that a large proportion of the tags had been torn out. Therefore in 1953 half the tags were attached as before, and the other half were fastened to the upper jaw. The jaw tags proved more satisfactory, and from 1954 onwards all tagging was done on the jaw.

From 1949 to 1952 all the pike for tagging were captured in traps set to catch perch during May and June. From 1953 onwards tagging of pike from perch traps continued, and in addition gillnets were set during the spawning season in March and April specially to catch pike for tagging, and also any pike taken in routine monthly seine nettings carried out from April to September inclusive were tagged and released. The pike tagged on the opercular bones and subsequently recaptured provided useful information on movements and were used to check growth calculation (Frost and Kipling, 1959) but owing to the many lost tags they could not be used satisfactorily for estimates of population numbers.

The calculation of population numbers from data from pike tagged on the jaw was considered to be justified for the following reasons:

1. Tag losses are negligible.

2. The pike are tagged in the spring, and recaught from the following autumn onwards, they therefore have the opportunity to mix randomly with the rest of the population before recapture.

3. Jaw tags do not appear to make the tagged fish more liable to capture than the untagged fish even in gillnets.

4. Although no direct evidence is available on natural mortality it seems unlikely that this differs between the tagged and untagged fish. The tags are small relative to the size of fish, and do not appear to impede them in any way. The growth of tagged fish is similar to that of untagged fish. One or two tagged females are occasionally found dead soon after having been tagged in the spawning season, but untagged female fish are also found dead at this time. Total mortality of tagged fish must be small owing to the high recapture rate; 73 percent of all tagged fish were recaught, up to 89 percent of some batches.

5. All gillnetting has been done by the staff of the Freshwater Biological Association who keep detailed records of the recaptured tags and of the total catches, so there is no loss of information due to incomplete reporting.

The ages of all recaptured fish have been found from the opercular bones, and, by using this information in conjunction with length-frequency distributions of the fish at the time of tagging, estimates have been made of the ages of all fish (whether recaptured or not) at the time of tagging. The sex was recorded at the time of tagging.

With all these advantages it would appear likely that extremely accurate population estimates could be obtained for pike by the tag-recapture method. Certainly estimates made of the numbers in sections of the population agreed closely with those found by Paloheimo's (1958) method, which is based on total numbers captured throughout the life of each year class. For example, estimates were made of the numbers of female pike aged three years in the autumn of 1958 (i.e., from the 1955 year class) in the south basin of the lake. The estimate made by Paloheimo's method was 255 fish. Two separate tag-recapture estimates could be made, from pike tagged after having been caught (a) in perch traps, and (b) in gillnets. The formulae given by Bailey (1951) were used to calculate the estimates and their standard errors. The results were (a) 230 (S.E.70), and (b) 209 (S.E.55). The agreement is good. This example was chosen because in this year sufficient pike were tagged from both perch traps and gillnets, as the 1955 year class was exceptionally abundant.

Comparisons of population estimates for the whole lake for several years for males and females separately are shown in Table III. These particular years have been chosen because Paloheimo estimates were available for comparison and of the pike tagged in the spring at least ten were recaught in the next winter by pike gillnets. The gillnets catch pike of 550 mm and upward; in most years the catch includes males of four years old and females of three years old, the comparisons have therefore been made with the estimates of numbers of fish at these ages and older. In two cases, males in 1954 and females in 1963, the estimates differ significantly, the tag-recapture estimate being much smaller. Further investigation revealed that in 1954 the males of four years old and in 1963 the females of three years old had made particularly poor growth, and had not been vulnerable to the gillnet. However, when the tag-recapture estimates were compared with Paloheimo estimates for males aged five and over in 1954 (290 fish) and females aged four and over in 1963 (523 fish) the agreement was satisfactory.

The results emphasize the difficulties which can arise in interpreting tag-recapture results. The estimates based on recaptures from the gillnets were for pike of 550 mm and over. Growth varies considerably from year to year, and in some years many three-year-old females and four-year-old males were caught, and in some almost none. Therefore the estimates were not consistent from year to year, as in some years these age groups would be included and in some years not, and without the background knowledge of growth and gillnet selection the results could be very misleading.

3. PERCH

Perch were tagged in the north basin of Windermere in 1943/44, 1948/50 and 1953.

The several perch tagging experiments proved to be most disappointing as regards the estimation of population numbers, although they provided valuable information on other aspects of the perch, especially growth and movements.

In 1943/44, 3 103 perch were tagged in the late summer with discs held in place by wire. The recapture rate was low, and only 3 percent were ultimately recaptured, although in these years traps were being fished all round the lakeshore. The discs were easily detached, and were easily visible from a distance, so it is likely that the tag losses and mortality of tagged fish (by predation of pike and other causes) were high. In these circumstances no useful estimates of population numbers could be made.

The tags used in the later experiments, 1948/53, were much more satisfactory. They were loops of monel metal and were attached to the pre-opercular bone (Bardach and Le Cren, 1948), and were inconspicuous and firmly fastened. In 1948/53 9 726 perch were tagged and 21 percent were ultimately recaptured, 17 percent in the same summer in which they were tagged. The main difficulty which prevented satisfactory estimates of population numbers was the unknown difference in the mortality of tagged fish and untagged fish in the time between tagging and recapture. There was definitely some mortality on account of catching and handling the fish when they were tagged. Occasionally some dead tagged fish were found in the lake shortly after release, and almost certainly others died and were not found.

Experiments carried out in a drainable fish pond with tagged and untagged perch showed that mortality could result immediately after tagging, especially if the temperature were high. Variations in the recapture rates for perch tagged and released into the lake on different days could also be ascribed to temperature differences between the days.

Even if the problem of mortality at the time of tagging had been overcome (possibly by strict attention to prevailing temperature conditions), other difficulties would arise in interpreting perch tagging data. The perch are caught in the spawning season, and the trap- entering behaviour is different for males and females, and different from year to year. In 1943/44 and 1948/50 the sex of tagged perch was not recorded. In 1953 it was recorded whenever possible; in the early weeks of the spawning season almost all the fish could be readily sexed, but later when most of the fish were spent the sex could not be ascertained externally, so estimates were made from the sex ratios of other samples of perch which were killed. However, also in this experiment mortality of tagged fish was high and no satisfactory population estimates were made.

4. CONCLUSIONS

Tag-recapture methods have provided satisfactory information on population numbers of char and pike, but not perch. The information on char was limited to certain spawning grounds which were accessible and could conveniently be netted. The information on pike was limited by gillnet selection to female pike aged three years and over and male pike aged four years and over, but in some years the youngest ages were not completely vulnerable to the nets and so erroneous results were obtained from the tag-recapture method. However, estimates of numbers of pike in selected sections of the population were very satisfactory. The main reasons that no satisfactory estimates could be made of the population numbers of perch were the unknown mortality caused by the tagging operation, and the need to treat each sex separately owing to their very different and variable vulnerability to the traps that were used to recapture them.

Our experience with tag-recapture experiments in Windermere has shown that they can provide useful results which could not have readily been obtained in any other way, but the interpretation of the results depends on background knowledge of the biology and behaviour of the fish.

Double recaptures, also recaught in previous year (a) 8, (b) 2, (c) 8, (d) 16.