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

CHAPTER II SUMMARY

1. INTRODUCTION
1.1. The central purpose of this study is to test different views concerning the biological significance of dispersal activity between populations. There are important controversies both about the causes of dispersal and about its role in population dynamics.
1.2. One traditional approach starts from the supposition that emigration should only occur due to adverse conditions in the present habitat. Risks connected with staying in the population should have become at least as high as the risks during dispersal before emigration is expected to occur. Otherwise, dispersal would be accompanied by increased mortality, which would lead to reduction or elimination of the dispersal behaviour by natural selection.
1.3. An alternative approach was introduced by Den Boer. He supposes that exchange of individuals between populations will decrease the chance of extinction of a species by 'spreading the risk'. In his view it is not necessary to assume bad conditions to understand the occurrence of emigration. It seems more probable that animals - at least insects such as carabid beetles - emigrate when the conditions for emigration are favourable.
1.4. The first view approach focuses on the mortality risk to the emigrating and not-emigrating individuals, Den Boer's approach on the possible advantages of the exchange of individuals between populations for the species as a whole. Supporters of the first approach object that advantages of dispersal, which only occur at the population level, are not individual characteristics of the sort which are the subject of natural selection.
1.5. Population biology is an empirical science, or should be, and a case study could throw light on this problem. I therefore looked for a species in a situation that would give the opportunity to investigate the problem satisfactorily.
The waterbeetle Gyrinus marinus Gyll. (whirligig beetle) appeared to be a favourable species. The adult beetles live in groups on the water surface near the banks of pools and ditches. The beetles can easily be captured and recaptured, and can be observed by eye. They disperse both by swimming and by flight. Egg production, size of offspring and the chance of survival can be estimated by simple methods. I concentrated upon an area of about 15 pools with Gyrinus-populations (Fig III-3). Some pools are connected by ditches, others are isolated and whirligig beetles could only reach these by flight.
1.6. From 1974 to 1983 field data were collected with the help of experiments and by capture-recapture studies of marked beetles. Next, computer simulation models were developed to investigate the consequences of the field data for the population dynamics of Gyrinus marinus in particular, and for other species in general, by changing the values for reproduction, survival, dispersal, etc.

2. FIELD DATA

2.1. Methods
2.1.1. Beetles were captured, marked with pin-pricks on the elytra and released again at the place where they were caught. The pattern of the pricks over the rows of dots on the elytra giving an individual number (Fig VI-1). By the captures/recaptures we got information about the exchange of beetles between different populations, the size of offspring, the chance of survival and the number of beetles (population size).
2.1.2. The technique of marking with pin-pricks gives too few data to permit following exchange by swimming. Therefore, in 1978 the beetles were marked with paint-spots, so that they could be observed without the necessity of recapture. The marks could be recorded while the beetles were on the water surface. Beetles of the same population were marked with the same color, and each day all over the study area we could note the numbers of beetles with different colors. After about 40 - 50 days there were too few marked beetles left and they were too mixed together to give data on exchange. A new set of beetles was then marked and released, and in this way in four periods throughout the year new beetles are released. The experiments were carried out in the upper part of the study area with pools that are interconnected by ditches (Br-complex).
2.1.3. Egg production can be studied by putting females for 24 hours in a tube or petri-dish with moist paper. Eggs are laid on the paper, and if kept moist they will develop and hatch after about one week.

2.2. Reproduction
2.2.1. In April, when hibernation ends, females start laying eggs. Egg laying is continued until mid-August. The eggs are laid in rows under water on water plants. The larvae live on the bottom of the pools and they pupate in a cocoon outside the water on water plants or in the soil.
2.2.2. A female oviposites at least once a week and is probably fertilized after each oviposition. At the end of June the first tenerals emerge. These beetles start reproduction about 10 days after emerging. The recruitment of this summer generation emerges from mid-September onwards. Because of the low survival chances there occur three more or less separate generations in a year: a reproducing spring generation (i.e. the hibernated autumn generation), a reproducing summer generation, and a hibernating autumn generation (Fig III-2).
2.2.3. In Table IV-1) the egg production of the spring and summer generations are compared. Egg production in summer is lower than in spring because fewer females oviposit and because each female lays fewer eggs per oviposition in summer.
2.3.4. Development from egg to teneral depends on temperature. Therefore, eggs laid from April to mid-May develop slower than eggs laid from mid-May onwards (about 10 weeks and 6 weeks, respectively).
2.3.5. On the average 1 or 2 tenerals emerge per oviposition, in summer somewhat fewer than in spring (Table IV-5). The mean number of tenerals per female/per month in spring is about twice that in summer (8 as compared with 3 or 4). This difference is mainly due to the smaller number of weeks for oviposition in summer than in spring.
2.3.6. The variation in recruitment is greater than we found for egg production. This difference in variation between egg production and recruitment indicates that the development from egg to teneral is more important for the dynamics of Gyrinus-populations than egg production, so that factors such as the reactions of females concerning the places to lay eggs should have only a limited effect on recruitment size.
A key-factor analysis gives as a result that variation in recruitment is most influenced by variation in the viability of the eggs, and by variation during larval and pupal development. Variation in egg production seems not to be important. Thus, conditions below the water surface (eggs and larvae) are of more importance for the rate of reproduction than the conditions at the water surface (adult beetles).

2.3 Survival of adults
2.3.1. The survival chances of adult beetles per week and per month in isolated pools are shown in Table V-1.
The survival rate decreases from spring to autumn and males have a better chance of survival than females; in spring a male on average has a life expectancy of 50 days, a female of 42 days. In autumn this expectancy is decreased to 20 and 19 days, respectively. Fortunately, variation coefficients were very low (less than 0.1). Thus, the chance of survival of different populations was about equal.
2.3.2. The survival chance of freshly emerged beetles (tenerals) is lower than that of beetles older than about three weeks, but we also found a decreasing survival chance with age.
2.3.3. At the end of October the beetles start hibernating below the water surface. On the average 33 per cent of the beetles survived the winter and appeared again in April. This corresponds with a mean chance of survival per week of 0.95. Variation in survival during hibernation is higher between different years (v.c = 0.7) than between populations (v.c. = 0.4) >2.4. dispersal by flight
2.4.1. Exchange of individuals by flight between populations was recorded within the study area by the recapture of marked beetles in pools that did not have water connections with the pool in which the marked beetles were released.
2.4.2. Such exchange is observed when emigration is followed by immigration within the study area. But some proportion of the emigrated beetles will have left the study area or will have died. Thus immigration data give only minimum estimates of emigration.
The intensity of this exchange is very low: in most cases fewer than 1 per cent of the marked beetles were recaptured after flight activities (Table II-1). Males were more frequently recaptured as flying immigrants than females.
Flight occurs from April to October, thus also during the reproduction period.
2.4.3. To get a better impression of the emigration rate from an isolated population we clip the wings of some of the beetles. In the weekly recaptures the decrease in numbers of both clip-winged and full-winged beetles was followed. No differences in rate of decrease were found between clip-winged and full-winged beetles (Fig VI-3).
But this experiment was carried out in 1977 when the weather was rather cold. However, when we compare the decrease rate of the clip-winged beetles in 1977 with that of full-winged populations in 1976 (a year with many warm days) no differences can be found (Fig VI-4). Thus, emigration by flight has no measurable influence on the decrease in numbers of a population. Given the estimated decrease rates of population size we can estimate that at least 5 per cent of the beetles have to emigrate before we would be able to detect this in the decrease of population size.
2.4.4. It is obvious that flight activities depend on weather conditions, and some experiments confirm this relationship. Fig I-4 (or Table VI-2) shows that no flight occurs below 18^oC.
The beetles have no power to fly against the wind: only at low wind velocities can they choose their own direction of flight (Fig I-5, Table VI-4). They only exceptionally fly at all if there is wind, because their efforts to take off usually fail then.
Daily observations of beetles at an artificial, small, shallow pool only recorded flight on sunny days without wind, without rain and with temperatures above 19^oC.
This dependence on weather conditions may be the cause of the differences in exchange by flight we found in 1974, 1976 and 1977. There appeared to be a relationship between the number of days with favourable weather for flight in a year and the exchange of individuals we recorded (Table II-2).
2.4.5. The lower exchange by females may be due to the many favourable days that occur during reproduction, when the flight ability of females is lower than after the reproduction period, as we found in laboratory experiments.
In these experiments we compared the flight of males and females. Fig VI-6 shows that in spring when all females reproduce the flight activity of males is higher than that of females, but after reproduction was finished more females than males flew away. During the summer generation this phenomenon recurs. The different flight activity of females during and after reproduction can be explained by the relations we found between the flight activities of females and the number of eggs per female, the proportion of reproducing females and body weight (Table II-3). Experiments also show that recently emerged beetles (tenerals) show less flight activity than older beetles (Table VI-6).
2.4.6. Differences in individual flight ability were tested in repeated experiments with the same individuals. Beetles that showed flight activity in the first test showed in later tests a greater willingness to fly than beetles that had not flown the first time. Thus, we may assume that there are differences in flight ability between individual beetles (Table II-4).

2.5. dispersal by swimming
2.5.1. At the Br-complex (cf. Fig III-3) there is extensive exchange by swimming between (sub)populations. On the average after 3 - 4 weeks most of the populations consist for more than 40 per cent of immigrants (Table VII-11).
The pools in the centre of the complex get more immigrants than the pools situated at the outer edges.
2.5.2. The emigration rate from a population can be estimated when both the chance of survival in the population and of the emigrants, and the number of these emigrants that have immigrated elsewhere, are known.
There is some decrease of the mean emigration rate per week from 0.36 in the first period (spring) to 0.20 in the fourth period (autumn), see E-values in Table VII-4. The centre populations on the average have a higher emigration rate than the border populations.
2.5.3. Survival during dispersal can be estimated by comparing the number of beetles that has emigrated and the number that is found as immigrant elsewhere. On the average we found a survival rate of about 72 per cent (see Q_e-values in Table VII-4), without significant differences between pools or periods. The loss to a population by emigration is about equal to the loss by mortality.
2.5.4. The relationship between emigration from and immigration into a population can be expressed as the dispersal-ratio Db = immigrants/ emigrants. In most cases Db<1, so, generally a population lost more beetles by emigration than it got by immigration. But, in eight out of the 24 cases Db>1 so that these populations is directly benefitted by dispersal (Db-values in Table VII-4).
2.5.5. With the paint-marked beetles no difference could be made between males and females, but with the pinprick-marked beetles we could compare the proportion of males and of females recaptured after immigration by swimming. Males were recaptured more frequently than females, so that we may assume that males emigrate more frequently than females (Table VII-1).
2.5.6. As might have been be expected, a reverse relationship is found between the distance between two pools and the exchange between these pools (Table II-5). However, there are exceptions. Apparently factors other than distance can also be important. We found, that the number of alternative routes, the route along which a pool can be reached, and the distribution of aggregations of beetles over the area can influence the degree of immigration into a population.
2.5.7. If emigration were a reaction to adverse conditions we could expect a relationship between emigration and population size. But if emigration occurs by accident, or because the conditions for emigration are favourable, no relationship between emigration and population size should be expected.
In our case, however, we found a reverse relationship; emigration from larger populations is smaller than from smaller populations. Moreover, immigration is probably greater into larger populations (Table II-6). Thus, there seems to be a tendency to stay in and to go to larger populations, possibly due to the attraction that groups of beetles exert on individuals that pass by.
2.5.8. We further looked at the influence of weather conditions on exchange between populations. The numbers of beetles that exchange during various weather conditions are compared in Table VII-13. We found that exchange between pools is significantly lower if the minimum temperature at night is below 8^oC. There may be more exchange in rainy than in dry weather. These results may be explained by the fact that the majority of swimming-activities occur at night and that the orientation of the beetles is disturbed by rain, so that a beetle may loose its way and arrive somewhere else.

3. COMPUTER SIMULATIONS

3.1. Key-factor analysis
3.1.1. The field data were used to feed computer simulation-models by which we could evaluate the results of the field study in regard to the consequences for Gyrinus-populations both of the small rate of dispersal by flight and of the high dispersal rate by swimming. Different models were constructed.
3.1.2. A first model uses the mean values and variation coefficients, as these were found for the chances of survival per developmental stage, for dispersal activities, and for the number of offspring per female. Stochastically throughout the year, we simulate the changes in numbers of individuals per stage, starting from the numbers of eggs laid, via the numbers of larvae and pupae, the numbers of tenerals, to the numbers of beetles that survive hibernation.
This was repeated 100 times, each time starting with 1000 males and 1000 females in April. The numbers were computed per week. The mean course of these numbers is shown in Fig VIII-2. After hibernation the number of beetles is generally somewhat higher than at the start of the year before. However, in this computation the real situation was highly simplified.
3.1.3. The losses of individuals in each stage were treated according to the method of key-factor analysis. In this way we could test which stages may be especially responsible for the variation in population size over the years. The average results of this exercise are given in Table VIII-2 and Table VIII-3.
The higher the k-value the greater the decrease in numbers during that stage. A stage with a high k-value, but a low standard deviation, nevertheless has little influence on the variation in the number of beetles in spring over the years.
We therefore calculated the product of k*st.dev. to find the stages that may contribute most to the variation in population size. Most important for population size seems to be the survival from hibernation, followed by the development from egg to teneral in both generations.
3.1.4. The fact that hibernation may be the key-factor (which is corroborated by a high correlation between survival during hibernation and population size after hibernation) introduces an permanent factor of instability in the population dynamics of these beetles. It is obvious that the beetles can not "protect" themselves against bad winter conditions once the winter has started.
There are two possibilities to minimize the effects of the highly variable chance of survival during hibernation in spring by behaviour of the beetles: (1) density-dependent reaction in egg production, (2) spreading the risk over as many habitats as possible by dispersal activities.
The field study does not give any indication that density-dependent reactions would play an important part. The role of dispersal is traced by simulation models, of which the results are given below.

3.2. Simulation models with dispersal
3.2.1. Dispersal occurs both by flight and by swimming. The influence of the dispersal activities as found in the field on population size, and on the chance of population survival, must be investigated by complicated models that simulate not only the process of dispersal as well as possible, but also the processes of reproduction and survival.
3.2.2. It would take too much space to explain the model fully here, so I restrict myself to some general remarks.
Survival, reproduction and dispersal are simulated as separate processes. The model considers individuals, not groups of individuals. Each time a value is needed it is randomly chosen from the log-normal distribution that was fitted to the mean value and variation coefficient of the field data.
There are 10 or 20 "habitats" at the start with 500 individuals or with a randomly chosen number. The beetles can exchange between the habitats.
A diagram of the model is given in Fig VIII-3.
We made different versions of the model, depending on the problem we wanted to investigate.
3.2.3. First, we analysed the consequences of dispersal activities. In this model a second species (species B) is introduced, which is identical to the first one (species A), except that it lacks the ability to exchange between the habitats. In this way the situation with dispersal activities can be compared with the same situation without dispersal.
The simulation was run with variation coefficients 0.3, 0.5 and 0.7. In this way the influence of variability (instability) could be estimated. The results after 20 "years" are given by the mean net reproductive rate R = (P₂₀/P₀)0.05 (P_x= population size in year x), as well as by the number of populations (N) that survived.
Dispersal can be low (5 %), or high (50 %). The survival chance per individual during dispersal (i.e. the chance to immigrate into another habitat) is 0.5 (Table II-7).
3.2.4. If the variability is low species A has a lower R and a lower N than species B: dispersal activities are disadvantageous in that case. But with increasing variability the situation for species B becomes worse: both R and N decrease.
Even a low dispersal intensity decreases the chances of extinction. A low dispersal rate appears even to be better in this regard than a high one.
The fluctuations in population size of the A-species are importantly reduced by its dispersal activities (Fig VIII-5). On the short term the advantages of staying seem greater than those of dispersal, but on the long term the not-dispersing species risks extinction. Although the dispersing A-species lives in smaller populations it will survive the B-species (Table II-7).
3.2.5. To test the genetic consequences of dispersal we built in the possibility of natural selection on dispersal activity. For Gyrinus no clear evidence is found for such selection, but we imagined there are two genotypes M and B: M can fly, and B cannot fly (for example macropterous and brachypterous, respectively). In our model only the homozygotic macropterous genotype (MM) is actually capable of flight.
Fig VIII-6 gives an example of the results of such simulations. Without dispersal the genotypes are distributed according to a Mendelean distribution, but when dispersal occurs the BB-type makes up an increasing part of the population. The MM-type, however, is not selected away. The course of numbers of the MM-type is much more stable than that of the MB- and the BB-types (Fig VIII-7).
3.2.6. These simulations seem to confirm the field observations, which suggest that dispersal occurs when the circumstances are favourable for dispersal activities. However, because of dependence on both variability, and on survival chances in the population as well as during dispersal, the intensity of dispersal is not neutral. The best "strategy" seems to be to show a low but non-zero dispersal intensity.