Sexual harassment in heterogeneous landscapes can mediate population regulation in a grasshopper

(1)

Sexual harassment in heterogeneous landscapes can mediate population regulation in a grasshopper

Silke Bauer, Jo ¨rg Samietz, and Uta Berger

Institute of Ecology, Friedrich-Schiller-University Jena, Dornburger Str. 159, D-07743 Jena, Germany

Population regulation has been related to differences in the quality among habitats, which mediate differences in vital rates such that in poor habitats reproductive rates are lower than those in high-quality habitats. The spatial distribution of animals in such habitats depends on their preferences and the degree to which individuals have a free choice of a particular habitat. The identified mechanisms that lead to a particular spatial distribution and eventually to regulation mainly include foraging-related interference, for example, ideal free distribution, or simple selection of available high-quality habitats, that is, site-dependent habitat selection. However, in insect species these mechanisms might not be applicable, but density-dependent habitat selection still occurs. We therefore suggest a mechanism that refers to the nearly universal observation that matings also bear fitness costs.

Although these costs have been investigated on the individual level in many insect species, their consequences for population dynamics have not yet been addressed. In the grasshopper speciesStenobothrus lineatus, females in a nonreceptive mating status escape sexually approaching males by undirected jumps. By including such avoidance behavior in a spatially-explicit simulation model, we investigated its potential to result in progressive use of low-quality habitats at increasing population densities. In particular, we show that (1) such behavior changes habitat selection, (2) altered habitat selection results in population regulation, and (3) the degree of habitat heterogeneity influences regulation such that (4) heterogeneous habitats show fine- tuned regulation and homogeneous habitats tend to support large fluctuations.Key words:habitat quality, habitat use, individual interactions, mating behavior, spatial simulation model,Stenobothrus lineatus. [Behav Ecol]

P

opulation regulation refers to decreasing per capita performance at increasing population densities (see Cappuccino and Price, 1995). In classical models of population ecology, regulation is caused by density dependence of demographic parameters. In reality, the change in demographic parameters is not driven by density per se (Hengeveld and Walter, 1999; Walter and Hengeveld, 2000) but by individual responses, for example, in behavior and performance, to changes in local population size (see Both, 1998;

Dhondt et al., 1992; Uchmanski, 2000).

The role of differences among habitats in the relationship between density and vital rates and the habitat selection behavior of individuals has been emphasized in the discussion of population regulation (Fretwell, 1970; McPeek et al., 2001;

Morris, 2003; Pulliam, 1988; Pulliam and Danielson, 1991;

Rodenhouse et al., 1997). Several theoretical approaches have attempted to characterize the spatial distribution of individuals among habitat patches of different quality, and to identify the mechanisms that produce this pattern. The most prominent among these are the descendants of the ideal distributions. For example, a high number of competitors in a food-rich habitat may increase interference and thereby lead to choice of suboptimal but less crowded habitat (Brown, 1969; Fretwell, 1970; Sutherland, 1996).

When habitat choice is not caused by interference but by simple selection of highest-quality habitat patches available,

this is referred to as site-dependent habitat selection (originally preemptive habitat selection; McPeek et al., 2001;

Pulliam and Danielson, 1991; Rodenhouse et al., 1997). Thus, increasing population sizes result in a filling-up of high-quality sites and progressive use of low-quality sites. Consequently, population-level demographic parameters such as reproductive rates or survival rates decline, which in turn stops population growth or even leads to a shrinking population size. This mechanism has been determined to regulate populations of territorial animals, namely, mostly birds and mammals (see Kruger and Lindstrom, 2001; Rodenhouse et al., 1997, 1999; Saitoh et al., 1999; Wolff, 1997).

In insects, heterogeneity in quality of habitats and its consequences for individual performance have also been described. For example, host-plant density and quality affect growth of adult and pupal herbivorous insects (Doak, 2000;

Erelli et al., 1998) or survival rates of their larvae (Floater and Zalucki, 2000). However, in many insect species individuals neither interfere during foraging nor preemptively occupy habitat patches but nevertheless show a density-dependent habitat selection (Samietz, 1998). Thus, what is the mechanism behind their habitat selection? In this article, we propose a mechanism that is related to the nearly universal observation that matings also bear fitness costs to females and females might be selected to avoid additional matings (Johnstone and Keller, 2000). Male-induced harassment behavior has been reported from many insect species (Baguette et al., 1996; Johnstone and Keller, 2000; Magurran and Seghers, 1997; McLain and Pratt, 1999; Weigensberg and Fairbairn, 1994), and the fitness consequences for the affected individual are described as well.

However, a direct empirical investigation of the population dynamical consequences of such behavior remains difficult because this requires simultaneous observations of individual behavior and its population-level consequences. Therefore, this Address correspondence to S. Bauer, who is now at the Netherlands

Institute for Ecology (NIOO-KNAW), Centre for Limnology, Rijks- straatweg 6, 3631 AC Nieuwersluis, The Netherlands. E-mail:

s.bauer@nioo.knaw.nl. J.S. is now at Agroscope FAW Wa¨denswil, Swiss Federal Research Station for Horticulture, Postfach 185, CH-8820 Wa¨denswil, Switzerland. U.B. is now at ZMT Center for Tropical Marine Ecology, Fahrenheitstr. 1, D-28359 Bremen, Germany.

Received 8 July 2003; revised 15 July 2004; accepted 24 July 2004.

Advance Access publication 8 September 2004

(2)

question has rarely been addressed previously. We aim to fill this gap by using the grasshopperStenobothrus lineatus(Panzer 1796, Caelifera, Acrididae) as a model system because the heterogeneity of this grasshopper’s habitat has been described, the fitness consequences of reproducing in habitats of different quality are known, and several empirical findings suggest that density-dependent habitat selection might be caused by harassing interactions during mating (Samietz, 1998). The habitat ofS. lineatuscan be divided into microhabitats differing in vegetation structure, which determines the amount of solar radiation penetrating the vegetation (Samietz, 1998). In turn, the radiation energy reaching female grasshoppers affects their body temperature and consequently determines the rates of temperature-dependent processes, for example, maturation of eggs requires a temperature sum of 450-hours above a developmental threshold of 23.560.2C (Samietz and Ko¨hler, 1998). Females of S. lineatus show clear preferences for the microhabitat that allows the highest reception of radiation energy (Samietz, 1998). Although we would expect a spatial distribution of female grasshoppers that is directly connected to the spatial distribution of such ‘‘good’’ microhabitats, empirical results depart from this expectation—particularly under high population densities (Samietz, 1998). Instead, the only factor that significantly explained the spatial distribution of females was male density. When male density was high, more females were found in poor microhabitats (Samietz, 1998). Further- more, in the course of the season, the density of males tended to decrease earlier than did the density of females because males have a similar life expectancy but reach adulthood earlier. Thus, when the abundance of females peaks, the abundance of males has already declined. Consequently, females reacted to this male thinning with a return to good microhabitats. Moreover, lifetime mobility of females differed significantly between years and was significantly determined by population density.

Although males’ lifetime mobility remained on the same level, females’ mobility significantly increased with population, that is, male, density (Samietz, 1998).

A potential reason for these results lies in the occasionally observed female rejection behavior during male-mating attempts: Females that are not receptive to mating reject courting males by jerking with the hind legs. As this is usually ineffective to prevent further male copulatory attempts, females escape by horizontal jumps, which contrast with their walking locomotion and are only performed when they are disturbed (Richards and Waloff, 1954; Samietz, 1998). Females became first receptive when reaching maturity and later on, after ovipositing approximately five egg-pods (Samietz, 1998).

In the present study, we address whether this avoidance behavior indeed provides a link between individual habitat selection and population regulation with an individual-based, spatially-explicit simulation model. In particular, we hypoth- esized that (1) this avoidance behavior determines habitat selection in females, (2) at high population densities a higher proportion of the population uses low-quality microhabitats, and (3) the avoidance behavior balances habitat use such that the vital population rates respond to prevalent population densities; namely, population grows at low densities and declines at high densities. Furthermore, we expected the mating behavior to intertwine with habitat characteristics, and therefore, we investigated the role that habitat composition plays in population regulation.

THE MODEL

General model structure

The model tracks the life cycle and movement of each individual grasshopper in a spatially-explicit habitat matrix.

The simulated habitat was subdivided into a rectangular grid in which each cell is of 131-m size and coded a particular microhabitat quality. According to the empirical measure- ments of vegetation structure and microclimate (Samietz, 1998), three quality classes were distinguished: (1) poor, (2) medium, and (3) good. Good microhabitats have sparse vegetation and adequate sunlight, whereas in poor microhabitats the vegetation is too dense to allow sunlight to reach the ground. A zero quality was assigned to nonhabitat space in the spatial arrangements such as noncrossable shrubs, hedges, and forest (Figure 1).

The model is discrete in time and covers one adult period, for example, June–September, of S. lineatus. Starting with adult molt and ending with the death of the last individual, a simulation run covers approximately 120 days. One time- step corresponds to 0.5 h and one simulated day consists of 32 time-steps corresponding to a photo- and activity period of 16 h in summer.

Given the empirical findings that no egg laid in medium and poor patches will produce a surviving offspring (Sam Samietz ietz, 1998), the model individuals were distributed randomly over the good-quality cells at the beginning of a simulation.

Males reached adult stage before females (protandry), at a mean age of 16 days (females 21 days, SD ¼ 5 for both sexes); the overall sex ratio was 1:1 (Samietz, 1996). In each time-step, all animals move, reproduce, and interact according to the rules defined below.

The model used sex- and age-dependent mortality rates that had been empirically estimated in the field population by mark-resight studies (Samietz, 1998).

Movement and interaction between sexes

Individual movement was determined by sex-dependent movement probabilities, pm, that include the influence of habitat quality and mating behavior. According to the pronounced habitat preferences, female pm increased from Figure 1

Schematic distribution maps of good (white), medium (light grey), and poor (dark grey) microhabitats on two empirical study plots of S. lineatus(Leutratal near Jena/Germany). Black cells stand for noncrossable shrubs, hedges, or forest. The proportions of the microhabitats are 32% good, 29% medium, and 39% poor quality in plot 1 and 24% good, 43% medium, and 33% poor quality in plot 2.

(3)

0.1 in good-quality cells and 0.45 in medium-quality cells to 0.55 in poor-quality cells (Bauer, 1997, Samietz, 1998). Males moved about to search for females, hence more frequently and rather independently of habitat quality. Theirpm¼0.6 in all microhabitats. Within the scale of a grid cell, the direction of movement was independent of microhabitat quality.

However, zero-quality cells were assumed to be noncrossable, and movements leading out of the habitat or into a zero- quality cell were corrected such that the direction was restricted to the neighboring cells of nonzero quality.

Interactions took place when males and females met, that is, stayed in the same grid cell. Males always attempted copulation but the receptivity status of the female determined the outcome of this attempt. If not receptive, that is, the female had not completed the required five ovipositions, and faced with a male, the female performed an escape jump (pm¼1). Otherwise, it moved according to the microhabitat- dependentpm. The assumptions behind this encounter model reflect well the biology of this species. Females are usually far less mobile than are males but may jump between 0.9 and 1.2 m when pursued by males attempting copulation. Such jumps effectively stop pursuit as the female is by then beyond the reach of males. However, within this distance the very mobile male easily detects a female either visually, acoustically, or chemically in the time interval of 30 min.

To compare this harassment interaction with a more general negative interaction, we additionally investigated the outcome of the model when assuming interference interaction. Herein, individuals always interacted negatively, that is, performed escape jumps upon encounter, irrespective of sex, age, or receptivity status.

Reproduction

The temperature-dependent egg-maturation process determines oviposition frequency. Consequently, the higher the body temperature has been over a certain time, the more frequently a female can oviposit. Female body temperatures are determined by ambient temperatures and by heating-up when exposed to sun radiation. Microhabitat quality codes vegetation structure and density and thereby determines how much sunlight may penetrate the vegetation (see General Model Structure). Thus, we assumed in the model that microhabitat quality indirectly modifies body temperatures.

Body temperatures in each microhabitat were simulated by applying an empirical thermal model on the basis of hourly weather data of the year 1993 (meteorological station Jena, Germany; Samietz and Ko¨hler 1998). Whenever a mature female in the model accumulated 450-h, six eggs (i.e., one egg-pod) were assigned to the current grid cell.

To calculate the population size of the next generation N(t þ 1), empirically estimated egg and larval survival rates were applied to the oviposited eggs (Samietz, 1998). However, there are potentially two negative feedbacks caused by habitat heterogeneity: females oviposit less under high population densities, and egg survival is lower in nongood microhabitats.

Referring to the latter, we analyzed the impact of three different survival functions: (a) eggs survive only in good habitats with a rate of 0.17 (empirically estimated survival rates) hypothetical egg survival functions; (b) eggs survive in all microhabitats with the same rate, 0.056; and (c) eggs survive with a rate decreasing from good (0.1) or medium (0.05) microhabitats to poor (0.02) microhabitats.

Model scenarios

We analyzed two scenarios with our model. In the first, the model mimicked the conditions on two empirical study plots.

The sizes of plots 1 and 2 were 4950 and 1125 m², respectively;

the proportions of good, medium, and poor microhabitats were 0.32, 0.29, and 0.39 in plot 1 and 0.24, 0.43, and 0.33 in plot 2, respectively (Figure 1).

In the second scenario, we emphasized the role of habitat composition on individual interactions and population regulation. Therefore, we created hypothetical habitat maps of the size of study plot 1 (1125 m²) by varying the proportions of good, medium, and poor sites between 0.0 and 1.0, and characterized the composition of each map with Simpson’s evenness index (McGarigal and Marks, 1994):

E¼1P h²

11=H; ð1Þ

with H the total number of microhabitats and h the proportion of each microhabitat. The evenness may range between zero, when the whole habitat consists of only one microhabitat type, and one, when all microhabitat qualities equally contribute to the habitat composition. Thus, high evenness values result from similarly composed heterogeneous habitats; however, low evenness values may result from significantly differing habitats (e.g., habitats that contain virtually only good microhabitats have the same evenness value as habitats with only medium or poor microhabitats). As this potentially affects population dynamics, we systematically changed habitat composition by starting from different homogeneous habitats and progressively adding other microhabitat qualities while keeping the size of the habitat constant.

In particular, we approached a heterogeneous habitat composition via four pathways: In the first and second, we started from a good homogeneous habitat and progressively added medium and poor microhabitats. In the third and fourth, the initial habitat contained only medium and poor microhabitats, respectively, and good microhabitats were added.

For each habitat map, population densities were varied from 0.5–8 animals/m²and therefore covered the empirically observed densities of 1–5 animals/m²(Ko¨hler, 1999; Ko¨hler et al., 1999; Samietz, 1998).

Quantifying the consequences of habitat composition and individual interactions on population dynamics

We fitted N(t) versus N(tþ 1) data from our model to the density-dependent function of Maynard Smith and Slatkin (1973, MS model):

Ntþ1¼ kNt

1þaNt^b

; ð2Þ

where k is fertility,a andb determine the shape of density- dependent function, namely, its inflection point, and the compensation type.

A measure for the dynamic complexity of this function is its slope,c, at the equilibrium population size (for derivation ofc, see Doebeli and Koella, 1995; Johst et al., 1999):

c¼1þb 1 k1

ð3Þ

Ifjcj 1, the equilibrium is stable, and ifjcj.1, the system exhibits the period-doubling way to chaos. Thus,cindicates the potential of a population to show simple equilibrium dynamics or complex (i.e., oscillatory or chaotic) dynamics.

We chose this density-dependence function because it is highly flexible, less prone to overestimate density dependence and could be fitted to a wide range of empirical data (for comparison of different density-dependent functions, see Bellows, 1981).

(4)

RESULTS

Scenario 1: mimicking field conditions

With increasing population size, the relative duration of stay in different microhabitats changed (Figure 2). At low population densities, females seldom left good cells and almost completely avoided medium and poor cells, reflecting their low leaving tendency in good sites. In contrast, high population densities forced females to abandon good cells and to spend more time in medium and poor cells. Females’ spatial distribution at high densities approached a random distribution such that the proportion of their lifetime spent in microhabitats of a particular quality equaled this quality’s proportion of the whole habitat. In study plot 1, for example, a female spent approximately 24% of its lifetime under high densities in cells of good quality and the area of good-quality cells was 24% of the total area. Thus, at the high end of population densities, we found the same pattern (ideal free distribution of females) as expected from feeding interference.

Similarly, the number of egg-pods a female laid during her lifetime declined with increasing population density (Figure 3).

The average number of three egg-pods per female at low densities decreased to an average value of two at high population density. Thus, increased density led on average to lowered demographic rates. As a consequence, population sizes at low densities increased or remained the same in the following year, whereas at higher densities the population in the following year was considerably lower (Figure 4, intersec- tion of population sizeN(t)/N(tþ1) data with equality line).

Fitting the MS model to population sizes revealed only slight differences between the two plots (Figure 4). Fertilities were k¼1.26 in plot 1 andk¼1.49 in plot 2 and the corresponding b values 0.72 and 0.76. Therefore, both plots are similar in their values of dynamic complexity (plot 1,c¼0.84; plot 2,c¼ 0.76), and we thus expect for both plots similar population Figure 2

Female microhabitat use in the model given as the proportion of lifetime spent in microhabitats of a particular quality (for study plot 2). At low population densities, females stayed most of their lifetime in the highly preferred good microhabitats, whereas higher densities forced females to an augmented use of medium and poor microhabitats.

Figure 3

Per capita performance, namely, the average number of egg-pods per female, deteriorated with increasing population density (model results for study plot 2). The mean amount of egg-pods per female declined from three egg-pods at low population densities to two egg- pods at high densities (regressiony¼2.08þ0.85e^0.51x,R²¼.78).

Figure 4

Population sizesN(tþ1) were predicted from the number of egg- pods oviposited in each microhabitat type, where they survived with a microhabitat-dependent rate. Black dots indicate a population in which individuals interact via sexual harassment, and grey dots give population predictions when individuals interact via general interference. Under low population densities, the harassing population grows until intersecting the line of equality,N(tþ1)¼N(t). The population size at this point is considered the equilibrium population size; growing beyond this population size causes the population to decline. In contrast, the ‘‘interference’’ population intersects the equality line either not at all or at very low population sizes, indicating a continuously declining population trend. ModelN(t)/N(tþ1)-data of both study plots were fitted to the density-dependent function of Maynard Smith and Slatkin (1973; MS model).

(5)

dynamics with low fluctuations (undercompensation). Accord- ing to plot size, the inflection points differed remarkably about two orders of magnitude, suggesting that the negative effects of density dependence started to affect population growth at smaller population sizes on plot 1 compared with plot 2.

In comparison to the harassment interaction, interference led to remarkably differing outcomes. The number of egg- pods oviposited in good microhabitats was lower when individuals interfered regardless of receptivity status. Conse- quently, populations, in which individuals always interfered upon encounter had lower reproductive rates, and thus, their population size in the following generation was always lower

than in populations in which only the nonreceptive females performed escape jumps (Figure 4).

Scenario 2: hypothetical habitat maps

Fecundity in the hypothetical habitats differed considerably (Figure 5). In habitats with predominantly good microhabitats, fecundity k reached the highest values (k’ 4). In contrast, in habitats that contained mostly medium or poor microhabitats, k values were lower. When changing the habitat composition toward higher evenness, namely, the habitat contains approximately equal proportions of each Figure 5

Values for the dynamic complexity,c, and fecundity,k, originate from fits ofN(t)/N(tþ1) model values to the density-dependent function of Maynard Smith and Slatkin (1973, .Equation 2) and are shown in relation to the evenness of the habitat they originated from (Equation 1). Low evenness values (E’0.0) result from homogeneous habitats consisting mainly of one microhabitat quality, whereas heterogeneous habitats (E¼ 1.0) contain equal proportions of each microhabitat quality. We systematically changed habitat composition starting from four homogeneous habitats, that is, containing only one microhabitat quality, and added the other microhabitats, whereas the size of the habitat remained the same.

Thus, the proportions of the starting microhabitat qualities were decreased in exchange of the microhabitat quality indicated after the arrow.

To calculateN(tþ1) from the number of oviposited eggs in the model, we applied three egg-survival functions: (a) only eggs in good habitats survived with a rate ofsgood¼0.17 andsmedium¼spoor¼0; (b) eggs survived in all habitats with the same rate 0.056; (c) eggs survived in all habitats but survival gradually decreased from 0.1 in good or 0.05 in medium habitats to 0.02 in poor habitats. Accordingly, fecundity was high when the habitat was mainly composed of good microhabitats and the eggs survived mainly there (survival function a). As heterogeneity increases, habitats and thus fecundity become similar. The common pattern among all results was the finding that homogeneous habitats (E¼0) may tend to complex population dynamics (c.1) and thus high abundance fluctuations. Accordingly, when habitats depart from homogeneity, population dynamics becomes stable and we expect stable population sizes.

(6)

microhabitat type, k reacted accordingly. When adding medium or poor microhabitats,kslightly decreased, whereas when adding good microhabitats to predominantly medium or poor habitats k increased with a high rate. At high heterogeneity (E ’ 1), habitats resembled one another in their composition and consequently in fecundity.

Different egg-survival functions also changed the differences in k between the habitats such that these differences almost disappeared when eggs survived with the same rate in all microhabitats, high differences appeared when eggs survived only in good microhabitats (Figure 5a) and an intermediate difference when eggs survived with a decreasing rate from good to poor microhabitats (Figure 5c). However, the final value for katE¼ 1 differed considerably between egg-survival functions: egg survival 1 resulted in the highest values (k¼2.7), which decreased via egg survival 3 (k¼2.0) to egg survival 2 (k¼1.0).

The dynamic complexity,c, which indicates the potential of a population to show a particular type of dynamics (Equation 4), showed a similar pattern over all egg-survival functions:

when the evenness was low (E’0), the populations tended to higher c values and thus potentially to high abundance fluctuations. However, this pattern disappeared at evenness E’0.3 and did not occur at all when the habitat composition started with good microhabitats. Here we found only values of c slightly smaller than one, indicating a stable-equilibrium population dynamics. Thus, habitat composition determined the reproductive potential of a population (k) and, in addition, changed the strength of the negative feedback. This feedback is weak in homogeneous habitats and thus potentially leads to complex dynamics with higher abundance fluctuations, whereas a heterogeneous habitat provides a strong feedback and the populations are consequently regulated in a stable population dynamics.

Comparing our suggested regulating mechanism to the more general interference interaction revealed the following differences (data not shown): First, the k values in all egg- survival functions were generally lower than in the harassment interactions. Furthermore, when starting with predominantly good habitats, the decrease in k exceeded that of the harassing populations and accordingly, when starting with predominantly medium or poor habitats, k increased to a lesser extent. Thus, populations with interference interactions had lower demographic rates than harassing populations. However, the dynamical complexity patterns were similar to those in the harassing populations.

DISCUSSION

We proposed a mechanism for density-dependent habitat selection in insects or organisms that neither interfere during foraging nor select their habitat in a preemptive way. This mechanism relates to the nearly universal observation that matings also bear fitness costs, and therefore, females tend to avoid unnecessary rematings. A common way to avoid these costly rematings is to simply escape from approaching males.

However, by doing so, females may be forced into low-quality habitats and face another cost, a decreased reproductive rate.

In the present study, we tested the capacity of an avoidance behavior in the grasshopper species S. lineatus to result in density-dependent habitat use, which, in turn, affects population growth. In accordance with our initial hypothesis, we found density-dependent habitat use. Under low population densities, females spent most of their lifetime in good microhabitats, which have an open vegetation and thus allow a high transmission of solar radiation (Samietz, 1998; Samietz and Ko¨hler, 1998). In contrast, increasing population densities progressively forced females into medium and poor

microhabitats. As a consequence of this altered habitat use, per capita fecundity, namely, egg-production, decreased such that population growth declined.

The proposed mechanism that leads to differences in microhabitat use inS. lineatusis as follows: at low population densities, a female stays much of its life in good microhabitats, which promotes fast egg-maturation. Receptivity is rapidly regained, female escape jumps are rare, and the probability of entering a medium or poor microhabitat is small. In contrast, under high population densities, a female frequently meets a male when not receptive. Consequently, females often escape by undirected jumps and hence more likely end up in a microhabitat of poor quality.

By using an individual-based, spatially-explicit model, we tested whether this verbal model indeed leads to population regulation. Our suggested mechanism implicitly assumes that (1) females do not return to good-quality microhabitats for oviposition although this should be expected for fitness maximization, and (2) females avoid additional matings even if such an avoidance behavior potentially leads them into low- quality microhabitats. Although both assumptions seem to contradict the expectations (fitness maximization), several facts sustain their plausibility: First, grasshopper females may be simply physiologically unable to postpone oviposition until they reach a good-quality habitat (Kriegbaum, 1997; for a moth example, see Jallow and Zalucki, 1998). Furthermore, although this grasshopper perceives the quality of the current microhabitat and leaves, for instance, a low-quality habitat with a higher probability, it can not predict when it will find a more suitable site.

An explanation for the second assumption—why females do indeed avoid additional matings—might be the finding that during mating, a number of male insects transfer proteins that prevent the female from remating (for grasshoppers, see Hartmann and Loher, 1996, 1999). Moreover, a number of studies report that rematings are costly for females and that females avoid male harassment by simply escaping (McLain and Pratt, 1999; Pilastro et al., 2003; Pizzari and Birkhead, 2001).

Our suggested mechanism might be interpreted as a special form of a direct interference-interaction among individuals.

Although the underlying biological assumptions differ sub- stantially—interference assumes competition for food or other resources, which does not apply to the species considered here—the outcomes might be alike. Therefore, we compared the individual-interaction rules of our model with a model that engaged all individuals in interference interactions; that is, they always interacted negatively and thus escaped from each other at encounter regardless of sex, age, or receptivity status.

The outcomes showed that females oviposited more frequently in the harassment model, especially under low population sizes, but the differences in egg-numbers almost disappeared at higher population sizes. However, the models’ most important difference was that under harassment, females oviposited more egg-pods in good microhabitats even under high population sizes: the proportion off egg-pods oviposited in good microhabitats declines from approximately 0.72 at low population density to 0.50 at high population density, whereas this proportion remained approximately the same (0.44) in the interference encounter. Accordingly, the eggs oviposited in medium microhabitats increased from 0.22–0.46 under harassment, whereas this proportion remained at 0.48 under interference. Consequently, populations with harassment interactions more likely increase under low population sizes compared with a pure interference interaction. Thus, although the harassing interaction suggests a disadvantage for the fitness of the females at first glance, this might still be the best solution under the given circumstances.

(7)

These findings support what has been learned from feeding-interference models, for example, behavior-based interference models, namely, that the mechanism behind interference should be explicitly identified for accurately predicting the consequences of individual interactions on higher levels (e.g., populations; Norris and Johnstone, 1998;

Stillman et al., 1997; VanderMeer and Ens, 1997).

In our model, regulation depended on habitat heterogeneity. Even when the additional effect of egg-survival has been switched off by using the same survival rates in each microhabitat (Figure 5b), we found a tendency of complex dynamics in homogeneous habitats. As a general hypothesis, we here propose that a more heterogeneous habitat tightens population regulation (Rodenhouse et al., 1997) such that the population is always kept within particular boundaries and thus exhibits typically a stable-equilibrium dynamics. In contrast, we expect to find weak regulation in homogeneous habitats that potentially leads to high fluctuations in population numbers (Jaggi and Joshi, 2001; Rodenhouse et al., 1999). In support of this general conclusion, some studies have already indicated that heterogeneous habitats tend to stabilize population dynamics such that abundances vary only modestly and extreme population sizes, namely, outbreaks or extinctions, are rarely reached (Floater, 2001;

Floater and Zalucki, 2000; Johst et al., 1999; Kindvall, 1996;

Shachak and Brand, 1991; Sutcliffe et al., 1997). However, our results have also shown that in homogeneous habitats populations may not display complex dynamics in the density-dependence function (not necessarily in actual population dynamics) when the habitat consists mainly of high-quality patches and thus ensures high fecundity.

In populations of S. lineatus, high densities led to an augmented use of low-quality microhabitats and deteriorating developmental rates and, finally, to a reduced population growth. The important point is that regulation is not an intrinsic property of a population but emerges only in conjunction with habitat heterogeneity. This contrasts with the view of Ylioja et al. (1999), who define factors that act

‘‘exogenously,’’ that is, independently of population density, or ‘‘endogenously’’ to the regulatory process. Following this distinction, vegetation structure and microclimate fall into the exogenous category, but as our results have shown their effect might not only change the state of an individual but also change the response of individuals during interactions and finally result in change of population regulation. For example, a female that stays in good microhabitats gains from favorable conditions for egg-maturation, which in turn affects how she responds to a male. Hence, as habitat heterogeneity and individual behavior entangle, such a distinction has no explanatory value, and we therefore suggest that in future studies both should be considered in combination (Johst et al., 1999; Petrovskii and Blackshaw, 2003; Rodenhouse et al., 1999).

We thank Bart A. Nolet, Steffen Hahn, Ju¨ rgen Groeneveld, Klaus Reinhardt, Anne E. Houde, and two anonymous referees for their most valuable comments on earlier drafts of this manuscript. Gu¨ nter Ko¨hler is greatly acknowledged for his support and fruitful discussions. Karin Johst greatly helped with the analyses of the MS function. J.S. was supported by grants of the Deutsche Forschungsge- meinschaft (DFG grants Ko1494/1–1, Ko1494/1–2). This is publication 3385 of the Netherlands Institute of Ecology.

REFERENCES

Baguette M, Convie I, Neve G, 1996. Male density affects female spatial behaviour in the butterflyProclossiana eunomia. Acta Oecol 17:225–232.

Bauer S, 1997. Populationsdichteregulation durch Verhaltenseffekte?

Ein Simulationsmodell am Beispiel des Heidegrashuepfers Sten- obothrus lineatus (MSc thesis) Jena, Germany: Friedrich-Schiller- University.

Bellows TS, 1981. The descriptive properties of some models for density dependence. J Anim Ecol 50:139–156.

Both C, 1998. Density dependence in clutch size: habitat heterogeneity or individual adjustment? J Anim Ecol 67:659–666.

Brown JL, 1969. The buffer effect and productivity in tit populations.

Am Nat 103:347–354.

Cappuccino N, Price PW, 1995. Population dynamics: new approaches and synthesis. New York: Academic Press.

Dhondt AA, Kempenaers B, Adriansen F, 1992. Density-dependent clutch size caused by habitat heterogeneity. J Anim Ecol 61:

643–648.

Doak P, 2000. Habitat patchiness and the distribution, abundance, and population dynamics of an insect herbivore. Ecology 81:

1842–1857.

Doebeli M, Koella JC, 1995. Evolution of simple population-dynamics.

Proc R Soc Lond B 260:119–125.

Erelli MC, Ayres MP, Eaton GK, 1998. Altitudinal patterns in host suitability for forest insects. Oecologia 117:133–142.

Floater GJ, 2001. Habitat complexity, spatial interference, and

‘‘minimum risk distribution’’: a framework for population stability.

Ecol Monogr 71:447–468.

Floater GJ, Zalucki MP, 2000. Habitat structure and egg distributions in the processionary caterpillar Ochrogaster lunifer: lessons for conservation and pest management. J Appl Ecol 37:87–99.

Fretwell SDLHL, 1970. On territorial behaviour and other factors influencing habitat distribution in birds. Acta Biotheoretica 19:

16–36.

Hartmann R, Loher W, 1996. Control mechanisms of the behavior

‘‘secondary defense’’ in the grasshopper Gomphocerus rufus L. (Gomphocerinae Othoptera). J Comp Physiol A 178:329–336.

Hartmann R, Loher W, 1999. Post-mating effects in the grasshopper, Gomphocerus rufusL., mediated by the spermatheca. J Comp Physiol A 184:325–332.

Hengeveld R, Walter GH, 1999. The two coexisting ecological paradigms. Acta Biotheoretica 47:141–170.

Jaggi S, Joshi A, 2001. Incorporating spatial variation in density enhances the stability of simple population dynamics models.

J Theor Biol. 209:249–255.

Jallow MFA, Zalucki MP, 1998. Effects of egg load on the host-selection behaviour ofHelicoverpa armigera(Hubner) (Lepidoptera: Noctui- dae). Austr J Zool 46:291–299.

Johnstone RA, Keller L, 2000. How males can gain by harming their mates: sexual conflict, seminal toxins, and the cost of mating. Am Nat 156:368–377.

Johst K, Doebeli M, Brandl R, 1999. Evolution of complex dynamics in spatially structured populations. Proc R Soc Lond B 266:1147–

1154.

Kindvall O, 1996. Habitat heterogeneity and survival in a bush cricket metapopulation. Ecology 77:207–214.

Ko¨hler G, 1999. Aussterbeprozesse bei Heuschrecken. Bochum, Germany: Laurenti Verlag.

Ko¨hler G, Perner J, Schumacher J, 1999. Grasshopper population dynamics and meteorological parameters: lessons from a case study.

Ecography 22:205–212.

Kriegbaum H, 1997. Grasshopper reproductive strategies measured in the field: a tradeoff between age at maturity and egg production per day. Naturwissenschaften 84:157–159.

Kruger O, Lindstrom J, 2001. Habitat heterogeneity affects population growth in goshawkAccipiter gentilis. J Anim Ecol 70:173–181.

Magurran AE, Seghers BH, 1997. A cost of sexual harassment in the guppy,Poecilia reticulata. Proc R Soc Lond B 258:89–92.

Maynard Smith J, Slatkin M, 1973. The stability of predator- prey-systems. Ecology 5:167–180.

McGarigal K, Marks, BJ, 1994. FragStats: spatial pattern analysis program for quantifying landscape structure, 2.0 version. Corvallis, Oregon: Oregon State University.

McLain DK, Pratt AE, 1999. The cost of sexual coercion and heterospecific sexual harassment on the fecundity of a host-specific, seed-eating insect (Neacoryohus bicrucis). Behav Ecol Sociobiol 46:

164–170.

(8)

McPeek MA, Rodenhouse NL, Holmes RT, Sherry TW, 2001. A general model of site-dependent population regulation: population-level regulation without individual-level interactions. Oikos 94:417–424.

Morris DW, 2003. Towards an ecological synthesis: a case for habitat selection. Oecologia 136:1–13.

Norris K, Johnstone I, 1998. Interference competition and the functional response of oystercatchers searching for cockles by touch. Anim Behav 56:639–650.

Petrovskii SV, Blackshaw RP, 2003. Behaviourally structured populations persist longer under harsh environmental conditions. Ecol Lett 6:455–462.

Pilastro A, Benetton S, Bisazza A, 2003. Female aggregation and male competition reduce costs of sexual harassment in the mosquitofish Gambusia holbrooki. Anim Behav 65:1161–1167.

Pizzari T, Birkhead TR, 2001. For whom does the hen cackle?: the function of postoviposition cackling. Anim Behav 61:601–607.

Pulliam HR, 1988. Sources, sinks, and population regulation. Am Nat 132:652–661.

Pulliam HR, Danielson BJ, 1991. Sources, sinks, and habitat selection:

a landscape perspective on population dynamics. Am Nat 137:

550–566.

Richards OW, Waloff N, 1954. Studies on the biology and population dynamics of British grasshoppers. Anti-Locust Bull 17:1–182.

Rodenhouse NL, Sherry TW, Holmes RT, 1997. Site-dependent regulation of population size: a new synthesis. Ecology 78:2025–2042.

Rodenhouse NL, Sherry TW, Holmes RT, 1999. Multiple mechanisms of population regulation: contribution of site dependence, crowd- ing, and age structure. In: Proc. 22 Int. Ornithol. Congr., Durban.

(Adams NJ, Slotow RH, eds). Johannesburg: BirdLife South Africa;

2939-2952.

Saitoh T, Bjornstad ON, Stenseth NC, 1999. Density dependence in voles and mice: a comparative study. Ecology 80:638–650.

Samietz J, 1996. Zur Microhabitatnutzung einer Heuschreckenart (Insecta:Caelifera) in Halbtrockenrasen:Stenobothrus lineatus(PAN- ZER). Verh. Gfo. 26:569–573.

Samietz J, 1998. Populationsgefa¨hrdungsgradanalyse an einer Heus- chreckenart: Methoden, empirische Grundlagen und Modellbil- dung beiStenobothrus lineatus(Panzer). Go¨ttingen: Cuvillier Verlag.

Samietz J, Ko¨hler G, 1998. The impact of behavioural thermoregu- lation on reproductive rates in a grasshopper. In: Population and community ecology for insect management and conservation (Baumga¨rtner J, Brandmayr P, Manly BFJ, eds). Rotterdam:

Balkema; 63–73.

Shachak M, Brand S, 1991. Relations among spatiotemporal heterogeneity, population abundance, and variability in a desert.

In: Ecological heterogeneity (Kolasa J, Pickett STA, eds). New York:

Springer Verlag; 202–223.

Stillman RA, GossCustard JD, Caldow RWG, 1997. Modelling interference from basic foraging behaviour. J Anim Ecol 66:692–703.

Sutcliffe OL, Thomas CD, Yates TJ, GreatorexDavies JN, 1997.

Correlated extinctions, colonizations and population fluctuations in a highly connected ringlet butterfly metapopulation. Oecologia 109:235–241.

Sutherland WJ, 1996. From individual behaviour to population ecology. Oxford: Oxford University Press.

Uchmanski J, 2000. Individual variability and population regulation:

an individual-based model. Oikos 90:539–548.

VanderMeer J, Ens BJ, 1997. Models of interference and their consequences for the spatial distribution of ideal and free predators. J Anim Ecol 66:846–858.

Walter GH, Hengeveld R, 2000. The structure of the two ecological paradigms. Acta Biotheoretica 48:15–46.

Weigensberg I, Fairbairn DJ, 1994. Conflicts of interest between the sexes: a study of mating interactions in a semi-aquatic bug. Anim Behav 48:893–901.

Wolff JO, 1997. Population regulation in mammals: an evolutionary perspective. J Anim Ecol 66:1–12.

Ylioja T, Roininen H, Ayres MP, Price PW, 1999. Host-driven population dynamics in an herbivorous insect. Proc Natl Acad Sci USA 96:10735–10740.