A CHECK ON TAGGING - RECAPTURE ESTIMATES OF POPULATIONS OF BROWN TROUT (Salmo trutta L.)
VERIFICATION DES ESTIMATIONS MARQUAGE - RECAPTURE DE POPULATIONS DE TRUITE (Salmo trutta L.)

by/par

Kj. W. Jensen

Freshwater Fishery Laboratory, P.O. Box 63, Vollebekk
Norway/Norvège

ABSTRACT

In a Norwegian mountain lake harbouring brown trout yearly experiments in tagging- recapture have been done in the period 1958–73. The fishery has been completely controlled during the period. Population estimates for the years 1958–66 have been obtained by correcting the virtual populations for natural mortality. The resulting population estimates were compared with corresponding Petersen estimates corrected for recruitment. In another mountain lake with similar conditions Petersen estimates based on tagged trout were compared with population estimates obtained from independent calculations of yield-recruitment- mortality.

The Petersen estimates are in good agreement with the other estimates. It is therefore concluded that eventual changes in vulnerability to gillnets caused by Carlin tagging are too small to give serious bias to Petersen estimates of brown trout.

RESUME

Dans un lac de montagne norvégien ayant une population de truite (Salmo trutta L.) des expériences de marquage-recapture furent effectuées pendant la période 1958–73. La pêche fut entièrement contrôlée durant cette période. Des estimations de la population pour les années 1958–66 furent obtenues par la rectification de ces populations pour la mortalité naturelle. Les estimations furent comparées aux estimations Petersen rectifiées pour le recrutement. Dans un autre lac de montagne ayant des conditions similaires des estimations Petersen, basées sur des truites marquées, furent comparées à d'autres estimations de population obtenues au moyen de calculs indépendants du rendement, du recrutement et de la mortalité.

Les estimations Petersen se comparent favorablement aux autres estimations. On en conclut que des changements de vulnérabilité aux filets maillants provoqués par le marquage Carlin sont trop faibles pour influer sérieusement sur les estimations Petersen de la truite.

Lake ∅vre Heimdalsvatn is a small lake (77.5 ha) situated in the central mountains of southern Norway at an altitude of 1 090 metres. The lake harbours only brown trout and a very small population of minnow (Phoxinus phoxinus L.). Since 1958 the fishery in the lake has been completely controlled. In 1958 the lake was very densely populated with small, slow-growing trout. More intensive fishing in the following years reduced the trout population significantly.

The ice usually breaks up in the first half of June; the fishing season begins in the beginning of July and usually ends in the beginning of October. A number of trout were tagged in August, 1958. In the years 1959–73 the tagging was always done in June-July at the beginning of the fishing season. Numbered Carlin tags with double steel thread were used, and the tags were attached below the front of the dorsal fin in the way commonly used in smolt tagging and described by Carlin (1955). Trout for tagging were caught on a chase- net or on hook, and kept under observation in small natural ponds for 12 h or more before being released.

As the number of trout caught and the age composition of the catch is known for the years 1958–71 (scale analysis of the 1972 material not yet completed) virtual populations (Fry, 1949) can be constructed for a sequence of years. These are shown in Table I. The yearly rate of survival, S, in each of the years 1961–64 and 1966–68 was estimated from the tagging data. As the yearly survival and the number of tagged trout caught in each month was known the natural mortality coefficient, M, could be estimated by a method described by Regier (1962). On average the estimated natural mortality was M = 0.31.

From the virtual populations the true abundance of the different year-classes can be estimated by monthly addition of the estimated number of fish killed by natural mortality. The detailed procedure will be published elsewhere; the results are shown in Table II. The author believes these estimates (the v.p. estimates) to be sufficiently correct to serve as a basis for checking population estimates obtained by means of the Petersen mark-recapture method for the same years.

Ricker (1958 p. 86) mentions six conditions that must be met if valid Petersen estimates shall be obtained. In our case no tags were lost during the first fishing season after tagging; tagging and fishing was done all over the lake; all recaptured tages were recognized and reported. A condition that may have been violated was that the tagged specimens could have suffered a higher natural mortality than the untagged fish, but this is always difficult to control. The resulting Petersen estimates would then be too high. As the fishing season corresponds with the growth season, recruitment (growth into vulnerable size) took place, so we must correct the estimates for recruitment. Further, immigration of small trout from the nursery streams was going on during the fishing season. The v.p. estimates include all trout in a year-class which are alive at a certain time regardless of whether they at that time are living in the lake itself or in a nursery stream. For small trout the Petersen estimates can therefore be expected to be lower than the v.p. estimates. Of gravest concern is the possibility of increased vulnerability caused by the tags. Most of the fishing in the lake was done with gillnets, and the tags can be entangled in the nets. Especially for tagged trout smaller than the model length corresponding to a certain mesh size the vulnerability could be increased, and the resulting Petersen estimates tend to be too small.

We shall use the Petersen estimate for 1961 to illustrate the method used to correct for recruitment. The tagging in 1961 took place in the period 28 June-3 July. The length distribution of the tagged fish and the recaptures in 1961 are shown in Table III. Fish smaller than 20 cm when tagged were not recaptured. Apparently the probability of recapture is highest for trout bigger than about 26 centimetres. Because of this we shall exclude all trout that were smaller than 20 cm at the time of tagging and we shall make separate estimates of the length group 20–25 cm and the group 26 cm and more.

The growth season had already begun when the tagging took place. Too few scale samples from July were available to calculate the length increment between annulus completion and tagging, so 10 mm was chosen as a reasonable average value. This means that our estimate shall exclude all trout with back-calculated length smaller than 19 cm, and we shall estimate separately the group 19–24 cm and the group 25 cm or more when the annulus was completed in 1961.

Table IV shows for different age groups the back-calculated lengths when the last annulus was laid down in 1961. These figures were used to divide the total catch in 1961 after 3 July on the two length groups 19–24 cm and 25 cm and bigger. The results are shown in Table V.

In the same way the figures in Table IV were used to divide the v.p. estimate for 1 July 1961 on the two length groups (Table VI).

For the Petersen estimates we shall use the common formula (with Bailey's correction):

The figures are given in Table VII. For (r + 1) smaller than 51 the approximate 95 percent confidence intervals were obtained by treating (r + 1) as a variable in a Poisson distribution. For (r + 1) bigger than 50 the binomial confidence intervals for the ratio were used.

The same methods were used for all the years 1958–66 and the results are shown in Table VIII. As seen from the last column in the table, five of the Petersen estimates (the N_p values) are smaller than the estimates from the virtual population (the N_v values) and eight are bigger. This indicates that in the involved length groups there is no serious increase in the vulnerability of the trout carrying the Carlin tags. In seven of the comparisons the difference between the two kinds of estimates is 10 percent or smaller. The greatest difference is only 26 percent, but the broad confidence intervals warn us that much greater differences could easily have appeared.

In another trout lake in the same district (Lake Olavatn, 2.72 km², altitude 967 m) where the fishery was completely controlled in 1969 and 1970, tagging data were combined with catch figures, data on growth and gillnet selectivity to estimate the equilibrium yield. The estimated yield was in good agreement with the observed catch (Jensen 1972). The model used showed a population of 3 890 trout of age five years or more on 1 July 1969. The corresponding (independent) Petersen estimate based on tagging and recapture was 3 877 with 95 percent confidence interval 2 942–4 812. Again there is good agreement between a Petersen estimate of trout carrying Carlin tags and another independent estimate.

The conclusion that can be drawn from these experiments is that although the vulnerability to gillnets for brown trout may be increased by Carlin-tagging, the change in vulnerability is usually too small to give serious bias to population estimates based on tagging- recapture.

Table VII
Æ. Heimdalsvatn - Petersen estimates of trout population on 3 July 1961

	m	r	c	p	95% conf. int. for p
19–24 cm at annulus	71	22	1 153	3 562	2 374–5 620
25 cm and bigger at annulus	139	70	1 229	2 408	1 904–3 233

Table VIII
Æ. Heimdalsvatn- Petersen and v.p. estimates of trout populations

Date	Length at last annulus	m	r	c	p	95% conf. int. for p	v	p:v
8.8.1958	24 cm and bigger	68	18	1 149	4 116	2 636–6 836	5 001	0.82
9.7.1959	25 cm and bigger	31	7	1 031	3 999	2 030–9 273	3 917	1.02
6.7.1960	25 cm and bigger	18	6	1 382	3 556	1 726–8 859	3 452	1.03
3.7.1961	19–24 cm	71	22	1 153	3 562	2 374–5 620	4 226	0.84
3.7.1961	25 cm and bigger	139	70	1 229	2 408	1 904–3 233	2 263	1.06
3.7.1962	25 cm and bigger	70	35	829	1 614	1 166–2 304	1 525	1.06
8.7.1963	21–24 cm	33	11	702	1 933	1 107–3 742	1 882	1.03
8.7.1963	25 cm and bigger	76	34	460	1 001	720–1 437	793	1.26
1.7.1964	21–24 cm	51	12	853	3 350	1 959–6 294	3 303	1.01
1.7.1964	25 cm and bigger	54	22	510	1 200	800–1 893	1 095	1.10
14.7.1965	25 cm and bigger	123	61	705	1 401	1 088–1 746	1 876	0.75
14.7.1965	20–24 cm	57	10	663	3 441	1 923–6 894	4 104	0.84
14.7.1966	25 cm and bigger	160	62	806	2 050	1 594–2 552	2 293	0.89