return to Contents

original text of the thesis:
Population dynamics of the Gyrinid beetle Gyrinus marinus Gyll (Coleoptera)
With special reference to its dispersal activities (1987)

CHAPTER V. SURVIVAL OF ADULTS

SUMMARY
Survival chances have been estimated for populations of the waterbeetle Gyrinus marinus Gyll. from capture-recapture sequences in 1974 - 1978. The average weekly survival chance is estimated as about 0.9247 in spring, decreasing to 0.7701 in autumn; males survive longer than females. Survival shows little variation in time or space (variation coefficient v.c.< 0.1). Survival chance possibly decreases with age. Freshly emerged tenerals apparently have a lower survival chance than older beetles. The survival chance from hibernation (October - April) is about 0.33, with a greater variation between years (coefficient of variation v.c.= 0.7) than between populations (v.c.= 0.4).

1. INTRODUCTION
1.1. Loss of individuals in a population can be due to mortality or emigration. Losses by flight from Gyrinus marinus populations are small (chapter VI Dispersal by flight). The study of Gyrinus populations in isolated pools therefore gives estimates of real survival rates, while the study of open populations affords estimates of 'survival' rates that are the result both of mortality and of emigration by swimming. A comparison of open and closed populations thus permits an estimate of population loss due to emigration.
1.2. The survival chance of individuals may be dependent on their age. Recently emerged tenerals have only small food reserves (small fat bodies) and empty stomachs. This may entail a lower survival chance for tenerals than for older beetles. Succeeding in finding food shortly after emerging may be essential to survival in the first weeks after emerging.
1.3. Survival chance may decrease as a beetle gets older. In spring all beetles are of about the same age (having emerged the previous autumn) and if survival depends on age than the average survival chance for the population as a whole will decrease during spring. In summer, however, the population is composed of beetles of different ages, since new beetles emerge every week. The average survival chance for the population as a whole should therefore be rather constant in summer. The same will be the case if survival depends mainly on predation or food supply.

2. METHODS
2.1. This study has been carried out since 1974 in an area of pools and brooks in the Northern part of the Netherlands (Fig III-3). Some pools are closed, others are connected with each other by ditches.
The population size in the closed pools is mainly affected by death and recruitment; in the open pools it is also influenced by immigration and emigration. Real survival chance - due to the death of individuals - can therefore only be estimated in the closed pools. Throughout this paper closed and open populations are considered separately.
2.2. During daytime Gyrinus (whirligig) beetles live in groups on the surface of the water along particular parts of the banks of a pool, where they are easy to see and to catch. After sunset the beetles swarm around for several hours.
Three generations are present: in spring the generation of the hibernated beetles reproduces, in summer (July, August) there is a reproducing summer generation and in autumn (September, October) the hibernating generation emerges. (For more features of the natural history of Gyrinus marinus see Chapter IV Reproduction).
2.3. The beetles were marked individually by means of pin-pricks (see Chapter VI Dispersal by flight for more details). The beetles have to be caught, carried to the laboratory and the marks have to be read off with a binoculair. When some hundred beetles are caught in a single sample it takes one to three days before they can be released again. In general, a population can thus be sampled at intervals of about 10 days.
2.4. When frequent recaptures of marked individuals are available several methods can be used to estimate the survival chance and/or population size during the period of sampling. In some methods survival chance is assumed to be constant (Lack 1943, Fisher and Ford 1947, Bailey 1951). The identical methods of Jolly (1965) and Seber (1965) only assume equal capture chances of all individuals throughout the sample period. For the method of Manly and Parr (1968) it is sufficient when individuals have an equal capture chance per sample. Because of the rather high capture chances (see below) we are able to use a method which is based on estimating the average period that a beetle has been present in the population (TS method). This mean time of presence corresponds with a mean survival chance per individual of 0.5 over that period. This is the most direct method of estimation and also the most detailed, because the survival chances are given per date of release. A disadvantage is that the method gives a minimum estimate since an individual will usually survive for some time after the last recapture. This method gives more reliable - for higher - estimates of the survival chance than the methods of Jolly, of Fisher and Ford or of Manly and Parr (See Appendix for a comparison of the three methods).

3. RESULTS

3.1. The reliability of the samples
3.1.1. The reliability of the estimates of survival depends on the reliability of the samples and the recapture chance per individual. The chance (estimated according to Manly and Parr 1968) that a beetle was captured in a sample varied between 0.2 in open and large pools and more than 0.4 in small closed pools and in ditches. Males have a greater capture chance than females (Wilcoxon test: n=90, z=2.41, P<0.02).
3.1.2. In small, closed pools and ditches the average chance of recapture of marked beetles is more than 0.7. In some cases beetles are recaptured on the average more than once. In most other situations 10 - 30 % of the beetles is recaptured. There is no significant difference in this recapture chance between males and females (Wilcoxon test: n=18, T=62, P>0.05).

3.2. Survival chances of males and females in open and isolated populations
3.2.1. The survival chances of males and females can be compared per population and for each generation with the Wilcoxon matched pair signed-rank test (Table V-1). The average survival chance per week varies in isolated populations between 0.9247 for males in spring and 0.7701 for females in autumn; on the average 8 of each 100 male beetles is lost per week in spring and about 23 female beetles per week in autumn. This corresponds to a monthly survival chance of 0.7123 in spring and of 0.3224 in autumn. Except in the spring, the survival chance of males and females does not differ in open populations. But in isolated populations males have a better survival per generation than females. The Friedman test indicates that the survival chance in isolated populations decreases from the spring generation to the autumn generation. However, the Wilcoxon and Wilcox test (providing multiple comparisons of samples, cf. Sachs 1982) shows that in general there are no significant differences between the spring and the summer generations. In open populations a significant decrease in survival is found only for males. The lower survival chances in autumn may be due to a lower capture chance as hibernation approaches.
3.2.2. Note the low variation coefficients v.c. (=st.dev/mean), which in most cases are far below 0.1. The different habitats show hardly any overall differences in the conditions that determine survival chance. The same low v.c.-values are found when the survival chances of the same population in different years are compared: mean v.c. = 0.02 - 0.08 (for sexes, generations and open or isolated populations seperately). No differences appear between the variation coefficients of open and of isolated populations.
3.2.3. As could be expected, in most cases the survival in isolated populations is significantly greater than that in open populations, with the exception of males in the spring (Mann-Whitney U-test Table V-1). The difference between the survival chance in open and in isolated populations gives some indication of the rate of loss in open populations due to swim-emigration (Table V-2). This emigration is highest in spring (16 resp. 27 % of the loss due to emigration), but decreases to 5 - 7 % in summer, and is very small in autumn (0 - 1.6 %). Swim-emigration will be treated more extensively in Chapter VII Dispersal by swimming.

3.3. Survival chance in relation to age
3.3.1. The supposition that freshly emerged tenerals should have a lower survival chance than older beetles is tested in four populations by comparing the frequency of recapturing of tenerals released with a very soft elytra (probably younger than 1 week) with that of beetles with harder elytra at release (Table V-3). If the recaptures are tested per release date per pool a signigicant difference was found for males but not for females (U-test on sample data: n1= 17, n2 = 51; males: U = 315.5, z = 1.72, P<0.05; females: U = 410.5, z = 0.33, P>0.05). However, if testing the total results the beetles ap-peared to be better recaptured as they were growing older (X2-test: males X2 = 92.97, P<0.001; females: X2 = 30.30, P<0.001). The fact that these results are not, of barely, significant may indicate that the assumed lower survival chance for tenerals only applies to the first days after emerging and that by the time of first capture most young tenerals had already found some food.
3.3.2. All beetles present in the spring did emerged the previous autumn, and thus grew older together. If their survival chance is age-dependent the time spent (TS) in the population will decrease with age until week 25 (third week of June = last week before the tenerals of the next generation will appear) the age-composition of the population is rather homogeneous. We have seen (Table V-1) that a decrease in survival rate occurs from April to June, which seems to indicate an age-dependent survival chance, but the decrease continues in summer and autumn, indicating that there is probably some relation with the time of year. We therefore analyse changes in the rate of survival during a period of 12 weeks after release, irrespective of the date of release. All records of such 12-week periods from isolated populations are used in a cumulative frequency distribution of the times spent in the population (see Fig V-1). A period of 12 weeks is chosen because this approximately coincides with a single generation and gives an optimal combination of the period and the number of records. The survival chance can be expressed as (a) the survival chance from week w-1 to week w (e.g. for males in spring, for w=3 -> 4, Q4 = 1164/1326 = 0.8778), or (b) a mean survival chance per week from the period w = 0 - t: Q = Qt(1/t) (e.g. for males in spring Q4= (1164/1726)(1/4)= 0.6744(1/4) = 0.9062). Spearman tests on the decrease of the weekly Q-values of both males and females from 1 to 12 weeks after the first release in spring do not give significant r-values, i.e. survival chances in spring do not decrease significantly with age. In summer no decrease in survival chance is found with the increasing number of weeks after release (Table V-4).

3.4. Survival during hibernation
3.4.1. Survival during hibernation can be considered as an independent factor in the population dynamics of this species. Beetles hibernate under water at the bottom of a pool and between roots of waterplants at the banks. The beginning and end of the hibernation period depend on the weather but this runs about from end-October to mid-April.
3.4.2. Survival during hibernation is estimated in three winters and for five populations by means of recaptures in spring of beetles that had been released in autumn. The number of hibernated beetles in spring is estimated with the methods of Craig (Southwood 1978; the method is based on the number of individuals recaptured once, twice, etc.). By comparing this number with the number of marked beetles released in autumn, the surviaval chance during hibernation can be estimated ( Table V-5). This method gives a mean survival chance of 0.33, but there is some indication that there is a high variation from year to year (v.c.=0.7), and a smaller variation between populations in the same year (mean v.c.=0.4). On the average, survival rates in winter seem to be higher than during the active season: per month the mean survival in winter is 0.80 (per week 0.95, compare with Table V-1). Furthermore, we may have underestimated survival in winter, since the Craig methods suppose a survival chance Q=1 during the sample period.

3.5. Causes of death
3.5.1. No systematic study has been made of predation or food supply as causes of death, but the estimates of survival rates and some available observations do give some indication of the importance of each of these factors.
3.5.2. Predation by fishes or ducks was occasionally observed, but many times we saw fishes or ducks swimming nearby or even through a group of whirligig beetles without paying any attention to them. We preformed a small experiment with two ducklings in a shallow bath. In addition to whirligig beetles, many small watersnails and larvae and pupae of mosquitoes were present. At first the ducks hunted the beetles, but as soon as they detected the other prey under water they ignored the fast-swimming beetles and ate the other animals that were much easier to catch.
3.5.3. The beetles are probably not favoured as food, as they seem to have a bad taste (Ochs 1969). This may be the reason why they usually stay in groups during the day. Apart from the lower chance per individual to be caught if one is in a group (Cushing and Jones 1968), potential predators may more easily recognize a group of beetles as unfavourable than a single beetle. Fish in an aquarium at first try to eat beetles released into the aquarium, but usually the beetles are spit out again. After some time the fish ignore the beetles. In experiments in pools with fish without experience with whirligig beetles, the beetles were eaten, but possibly also spit out most of the time. We have the impression that beetles are more frequently attacked in such experiments than in pools where they have been present for a longer time. Another indication that in natural populations predation is not very important quantitatively, is that no consistent differences were found between the populations in ditches (with few or no fishes and/or ducks) and populations in pools (with both fishes and ducks).
3.5.4. We have no data about food. However, the frequent ovipositions during the reproduction period (Chapter IV Reproduction) can be an indication that in general food is not in short supply. In carabid beetles it has been shown that egg production decreases with a shortage of food (Baars and van Dijk 1984, van Dijk 1983). Death by starvation probably occurs infrequently.
3.5.5. Little is known about diseases and parasites of whirligig beetles. The population in a fishpool in our area was strongly infected by Laboulbeniales moulds (Ascomycetes). Nearly every beetle had mould fruit-bodies along its sides. But there is no indication that the survival chance of the beetles is influenced by this parasite (Scheloske 1969, Meijer 1975). In other populations beetles were rarely found with these parasites.
3.5.6. An indication of the proportion of beetles with maladjustments congenital or resulting from developmental failures may be the frequency of abnormalities of the elytra we recorded while marking the beetles. Of 15051 males from 13 different years and/or populations on the average 2.34 per cent (st.dev. 0.40) had abnormalities in shape or punctation of the elytra; among 10878 females 3.14 per cent (st.dev. 1.11) had abnormalities. The difference between males and females is significant (t-test: df=24, t=2.43, P<0.05), which may be one of the causes for the lower survival chance of females.

4. DISCUSSION

4.1. Several causes of mortality have been indicated. There seems to be no key factor for mortality such as a specific predator. The chance to survive may be lower outside a group than inside one. We have suggested that a single beetle may be less recognizable than a group as bad-tasting prey. The risk of death seems to be connected with the activities of the individual beetle itself and with generally unfavourable circumstances like cold weather over a longer period or a dusty water-surface where a beetle can get stuck by accident.
There is probably a decrease in the chance of survival from spring to autumn. This should not be caused by age (cf Table V-4). The risk of death could increase during the year if there should be an increase in predation by young ducks and fishes that have no experience with whirligig beetles and still have to learn that the beetles are unfavourable prey.
4.2. A number of factors may account for the lower survival chance of females than of males. For ovipositing a female has to be under water, where the eggs are laid in rows on waterplants, and are highly visible for predators (fish) and not able to escape as quickly as on the surface. Secondly, we found a significant higher frequency of physical maladjustments among females than males. The frequency of fatal maladjustments or of abnormalities that hinder the ability to swim, for example, is possible also higher for females.
4.3. The low variation in survival chance during the season, both between different populations and between different years indicates that survival chance is either determined by many and/or rather constant environmental factors, which do not vary much over space and time, or are mainly connected with features of the beetles themselves.
4.4. The variation in winter-mortality is much greater than that in mortality during the active season. Mortality during hibernation therefore generally contributes more to the variation in population size than mortality during the rest of the year. The variation from year to year is about twice that between populations in the same year. This may indicate that winter mortality depends more on general weather conditions (severe winters) than on local conditions.

return to Contents

REFERENCES

Baars MA, Dijk ThS van (1984) Population dynamics of two carabid beetles at a Dutch heathland II. Egg productionand survival in relation to density. J Anim Ecol 53:389-400
Bailey NTJ (1951) On estimating the size of mobile populations from recapture data. Biometrika 38:293-306
Begon M (1979) Investigation Animal Abundance capture-recapture for biologists. Edw.Arnold London
Bishop JA, Sheppard PM (1973) An evaluation of two capture-recap-ture models using the technique of computer simulation. In: The mathematical theory of the dynamics of biological populations (Ed. MS Barlett & RW Hiorns)
Cushing DH, Jones FRH (1968) Why do fish school? Nature 218:918-920
Dijk ThS van (1983) The influence of food and temperature on the amount of reproduction in carabid beetles. Rep 4th Symp Carab'81
Eijk RH van der (1983) Population dynamics of gyrinid beetles I. Flight activity of Gyrinus marinus Gyll. Oecologia (Berlin) 57:55-64
Eijk RH van der (1986) Population dynamics of gyrinid beetles II. Reproduction Oecologia (berlin) 69:31-40
Fisher RA, Ford EB (1947) The spread of a gene in natural conditions in a colony of the moth Panaxia dominula L. Heredity 1.2: 143-172
Jolly GM (1965) Explicit estimates from capture-recapture data with both death and immigration-stochastic model. Biometrika 51.1 & 2:225-247
Lack D (1943) The age of the blackbird. BritBirds 36:166-175
Manly BFJ, Parr MJ (1968) A new method of estimating population size, survivorship, and birth rate from capture-recapture data. Trans Soc Br Ent 18:81-89
Meyer J (1975) Carabid Migration Studied with Laboulbeniales (Ascomycetes) as Biological Tags. Oecologia (Berl) 19:99-103
Ochs G (1969) The ecology and ethology of whirligig beetles. Arch Hydrobiol Suppl 35,37:375-404
Sachs L (1982) Applied Statistics. A Handbook of Technics. Springer New York
Scheloske HW (1969) Beiträge zur Biologie, Ökologie und Systematik der Laboulbeniales (Ascomycetes) under besonderer Berücksichtigung der Parasit-Wirt-Verhältnisses. Parasitolog Schr Reihe 19:176 pp
Sheppard PM, MacDonald WW, Tonn RJ, Grab B (1969) The dynamics of an adult population of Aedes aegypti in relation to dengue haemorrhagic fever in Bankok. J Anim Ecol 38:661-702
Southwood TRE (1978) Ecological methods (2nd ed).Chapman and Hall, London

return to Contents

return to text (Methods)

APPENDIX  Comparison of three methods for estimating survival chance

I. When frequent recaptures of marked individuals are available several methods can be used to estimate the survival chance and/or population size during the period of sampling. In some methods survival chance is assumed to be constant (Lack 1943. Fisher and Ford 1947. Bailey 1951). The identical methods of Jolly (1965) and Seber (1965, 1973) only assume equal capture chances of all individuals throughout the sample period. For the method of Manly and Parr (1968) it is sufficient when individuals have an equal capture chance per sample. We tested the Jolly-Seber method, but it appeared not to be reliable, particularly for open populations. Probably because in our case the survival chance between two sampling events is small (Bishop and Sheppard 1973), so that too few data were available to estimate reliably the necessary Z- and R-values.
2. More accurate estimates of the survival rate and population size can be obtained with the Fisher & Ford method (explained by Sheppard et.al. 1969. Begon 1979. see also Southwood 1978). The strength of the Fisher and Ford estimation is that the data from a number of samples can be taken together, so that small sample effects are averaged (Begon 1979). But the method assumes a constant survival chance. Separate estimates must therefore be made for spring, summer and autumn, i.e. per generation. The Manly-Parr method is used in addition to the Fisher & Ford method because the Manly-Parr method is even not affected by age-dependent mortality. A third method is based on estimating the average period that a beetle has been present in the population. This mean time of presence corresponds with a mean survival chance per individual of 0.5 over that period. This is the most direct method of estimation and also the most detailed, because the survival chances are given per date of release. A disadvantage is that the method gives a minimum estimate since an individual will usually survive for some time after the last recapture.
3. The three methods for estimating survival chance (Fisher & Ford (FF), Manly & Parr (HP), and by means of the time spent in the population (TS)) differ in the way capture-recapture data are treated. The FF and TS methods both use the time individuals are available in the population for capture. FF by using the age of the marked individuals. TS by estimating the time individuals are present in the population. In other words. FF looks backwards to the history of a marked individual. whereas TS looks forward to what will happen to a marked beetle. The HP method uses the decrease in the number of marked beetles between the day of release and that of recapture.
4. The three methods are compared in the table below for the total season, and per generation in spring, summer and autumn. The methods give significantly different values. In general TS gives higher survival values than FF. and FF higher than HP. On the average the variation coefficients of the FF-estimates (per generation and per sex) are about 1.5 times those of the HP-estimates, and the latter are about twice those of the TS-estimates (0.1258, 0.0885, 0.0384 resp.). This may mean that FF estimates less accurately than HP, and HP less accurately than TS. Since TS give minimum estimates of the survival chance and TS in fact provides the highest values for survival chance, it is the most reliable of the three methods. The TS method will be used further in this study. Although the three methods give significantly different values, their estimates appear to be correlated, that is, the methods estimate relative differences in the survival of different populations about equally (product-moment correlation tests per generation: P<0.05 (3x), P<0.02 (2x), P<0.002 (4x)).

Table V-6.Comparison of three methods of estimating survival chance
Significance (2-tailed): z >1.96: P<0.05, z>2.33: P<0.02, z>3.10: P<0.002
SURVIVAL CHANCE PER WEEK WILCOXON TEST
(n, z-values)		CORRELATION TEST
(n, z-values)
METHODFFMP TSFF - MP FF - TSMP-TSFF - MPFF-TS MP-TS
TOTAL PERIOD0.86400.80750.8394534.38572.46526.00549.48588.76539.40
SPRING0.91450.81350.8666374.1639 3.07404.54382.12417.24422.38
SUMMER0.67840.74710.8526383.4043 5.41424.85435.43444.21435.43
AUTUMN0.67040.72510.759581.5415 1.7091.36113.52173.70113.52

REFERENCES

Bailey NTJ (1951) On estimating the size of mobile populations from recapture data. Biometrika 38:293-306
Begon H (1979) Investigation Animal Abundance capture-recapture for biologists. Edw.Arnold London.
Bishop JA, Sheppard PH (1973) An evaluation of two capture-recapture models using the technique of computer simulation. In: The mathematical theory of the dynamics of biological populations (Ed. HS Barlett& RW Hiorns)
Fisher RA, Ford EB (1947) The spread of a gene in natural conditions in a colony of the moth Panaxia dominula L. Heredity 1.2: 143-172
Jolly GH (1965) Explicit estimates from capture-recapture data with both death and immigration-stochastic model. Biometrika 51.1 & 2:225-247
Lack D (1943) The age of the blackbird. BritBirds 36:166-175
Hanly BFJ, Parr HJ (1968) A new method of estimating population size, survivorship, and birth rate from capture-recapture data. Trans Soc Br Ent 18:81-89
Seber GAF (1965) A note on the multiple-recapture census.. Biometrika 52. 249
Seber GAF (1973) The Estimation of Animal Abundance and Related Parameters. 506 pp Griffin. London
Sheppard PM. MacDonald WW. Tonn RJ. Grab B (1969) The dynamics of an adult population of Aedes aegypti in relation to dengue haemorrhagic fever in Bankok. J Anim Ecol 38:661-702
Southwood TRE (1978) Ecological methods (2nd ed).Chapman and Hall.London

return to text (Methods)

return to Contents