Condition and spatial distribution
The condition of individual bush-crickets
influence their spatial distribution
Dag Ø. Hjermann
Department of Biology, Division of Zoology, PO Box 1050 Blindern, N-0316 Oslo, Norway
Abstract
Males of many species of grasshoppers and crickets (Orthoptera) form aggregations when calling for females. However, to our knowledge it has not been reported that the males' spatial position within aggregation correlates with mating success, such as in many species of lekking birds and mammals. Males of the wart-biter Decticus verrucivorus, a bush-cricket (Tettiginiidae), are attracted by other males' song and so have an innate tendency to aggregate. They also provide no parental care. Their mating system therefore resembles a lek (but not a "classical" lek, since males hold floating territories). In this experiment, I "created" animals of good and poor condition by rearing approximately half of the animals under suboptimal conditions. The animals were released in equal density in eight experimental patches of three sizes (20, 40 and 80 m2), and subsequently tracked by recording their positions four times per day during three weeks. Males that were raised under suboptimal conditions moved less, sang less and had lower song amplitude than the other animals, confirming that they were of poorer condition. The movement patterns revealed that in medium-sized patches, poor-condition animals of both sexes stayed more in the edge zone (< 1 m from habitat edge) than expected by chance, while good-condition animals used the edge zone and interior equally much. The same tendency was found in the only small patch with good-condition animals during most of the experiment. During the peak hours of male singing, however, poor-conditioned females (but not poor-conditioned males) moved towards the center of the patch, where predominantly good-condition males were advertising. In the large patches, however, animals used the interior and the periphery equally much regardless of condition. The results indicate that in the small and medium patches, poor-condition males are repelled by males of better condition because they latter sing louder or more frequently. The spatial segretation between good- and poor-condition females was surprising; the most plausible explanation appears to be that females compete for access to attractive males. Wart-biter males may only mate once a day under ideal conditions, and more infrequently under Norwegian climatic conditions. The lack of differences in the edge-interior use in the large patches may be caused by larger habitat area (less need or higher cost of defending the interior) or by larger population size (the optimal behavioral strategy depends on population size). The interaction between condition and patch size is interesting in the light of the excessive habitat fragmentation of many habitats, including the wart-biters' natural habitat.
Introduction
In many species of grasshoppers and crickets (Orthoptera), the males form aggregations when calling for females (e.g., Campbell and Clark 1971; Schatral 1984). In some species, males seem to clump mainly because the preferred microhabitat, usually tall plants, is clumped, i.e., there is no need to invoke an innate tendency to clump to explain the observed distribution (Arak and Eiriksson 1992). In these species, clumping may reduce the ability of males to attract females (Arak et al. 1990), so when the area of suitable microhabitat is limited, males may be forced to settle more densely than optimal. In contrast, males of the wart-biter Decticus verrucivorus are attracted to other singing males (postive phonotaxis; Schatral et al. 1985; Keuper et al. 1986). This, together with the impression that wart-biter males congregate in smaller areas than the preferred vegetation, lead Weidemann et al. (1990) to propose that the wart-biter mating system is a resource-based lek, i.e., that positive phonotaxis leads to a denser aggregation than one would expect from the distribution of microhabitat alone. The male does not provide parental care (Wedell and Arak 1989; Wedell 1993), so it indeed appears likely that wart-biters actually display a lek in the relative broad sense ("an aggregated male display that females attend primarily for the purpose of mating"; Höglund and Alatalo 1996:6).
In many species of lekking birds and mammals, "dominant" males tend to stay in the middle of the male aggregation (Tab. 3.1 in Höglund & Alatalo 1995). However, this has to my knowledge never been reported in Orthoptera. In the current study, I studied the spatial movement patterns of wart-biters that were manipulated so they were of either good or poor condition. The animals were individually marked and their movements were followed during an extended period of time in experimental patches of different sizes. I specifically tested whether "dominant" males tended to stay in the middle of the male aggregation, and whether the spatial orginazations of the animals was affected by habitat or population size.
Methods
Study species and experimental design
The wart-biter (Decticus verrucivorus L.) is a relatively large bush-cricket (length 3-4 cm); it has wings, but most individuals are not capable of flying more than a couple of meters (Ander 1947). In Norway, adult males are largely found in July and August. Adult males emerge before females, probably to avoid sperm competition by mating with virgin females (Wedell 1992). The diet of the wart-biter is a combination of insects (especially acridid grasshoppers) and plant food. Both instars and adults are extremely heat-loving, and the habitat (in Scandinavia) is low meadow or grassland vegetation, preferably including microhabitats with tall vegetation (male singing perches) and vegetation-free spots (where females prefer lay their eggs; Ingrisch and Boekholt 1982; Cherrill and Brown 1990; Cherrill et al. 1991). Adults sing mostly between 1000 and 1200, on hot days also around 1500. Males move frequently around during these periods, singing for a few minutes from each perch. Females approach the male when they want to mate; mating is very short (1-2 minutes).
The experiment was performed during the summer of 1997 at Oslo University’s experimental station at Evenstad, SE Norway. Eight habitat patches were made by mowing parts of a continuous area of seminatural pasture. Because the available area was shaped as a narrow rectangle, all habitat islands were 4-6 m wide, while their lengths varied from 4 to 16 m. Two islands were relatively large (ca. 80 m2), two were medium (ca. 40 m2), and four were small (ca. 20 m2) (Fig. 1). Since wart-biters prefer a habitat mosaic of open ground and vegetation (Cherrill and Brown 1990), I created a number of vegetation-free spots within each patch. The patches were protected from bird predation using nets suspended 1.52 m above the ground (the wart-biters could move freely through it). To the north and south, adjacent patches were separated by 20-25 metres of short-cut lawn. To the western side there was 80-100 cm of earth, then a 60-cm fence of transparent plastic seaparating the
study area from a corn field. On the eastern side, there was ca. 2 m short-cut lawn between the patch and a 60 cm high fence of solid metal. There were no wart-biters on the study site in advance. The animals were caught in semi-natural meadows and roadsides in five locations in SE Norway. Some were caught as 2.3. instars and reared in the laboratory until they were adult, while others were caught as adults shortly before the experiment. In the laboratory, the insects were kept individually in plastic boxes (measuring 16 x 8 x 5.5 cm) with a bit dry vegetation (to climb on to prior to moulting) and fed with cabbage leaves and ground dried catfood. Partly because there was not enough space for the final moult, the laboratory conditions were clearly suboptimal, as evidenced by small physical deformities and lower song performance and movement activity in males (see Results). I therefore refer to laboratory-reared and adult-caugh animals as the groups "poor condition" and "good condition", repectively. The animals were marked by gluing one end of a red, numbered 80 x 4 mm plastic tape to the top of the pronotum, cooled down and released in the experimental habitat patches at about 5p.m. to avoid escape reactions because of handling. On June 24 at 0500, adults were released in equal densities in each of the eight patches (4, 8 and 16 in the small, medium and large ones, respectively) in approximately even sex ratios (male:female number was 5:3 in one medium patch and 9:7 in the large ones) and with approximately even proportions of good- and poor-condition animals. However, 23 of the animals migrated between patches during the study, especially from the small patches; migration rate was especially high in good-condition males. The populations in the small patches were therefore biased towards poor-conditioned females at the end of the study. Animals were usually tracked four times a day on almost every day from day 4 to day 23 of the experiment and on two days thereafter. In each tracking session, I measured the position of each individual relative to a 25 x 25 cm grid which was marked in the patches using small poles. During tracking, the animals were usually not disturbed so much that they moved. I also noted activities such as singing, mating and egg-laying. For each male, I assessed the song volume on a scale from 0 (not audible) to 3 (very loud). Temperatures were recorded (usually 20 - 30 times/day) using an ordinary ethanol thermometer lying sun-exposed on ground level leaves in order to mimic the body temperatures of basking wart-biters.
Movement activity
Movement activity or speed was analysed by linear regression, based on the distance between two subsequent recordings on the same day (r) and the time interval (t) between the two recordings. The transformation loge(r2) was used as the response variable, while the predictor variables were ln(t) together with covariates (these transformations are suitable to describe random walks and similar movement patterns; Johnson et al. 199x, Hjermann 2000). I used the following covariates: the identity number of the animal (ind) and the mean temperature during the time interval between recordings (temp, calculated from temperature measurements smoothened out using 2nd degree polynomial regression), and temp2. I tested whether any of the covariates could be excluded from the model using Mallow's Cp. On the individual level I tested the effect of patch size, sex, catch location (the geographic origin of the animals), and condition (lab-reared vs. wild-caught) on movement activity, using each individual’s estimate of the indparameter as an indicator of movement activity.
Use of the edge zone vs. the patch interior
As a condensed measure of choice of microhabitat within each patch, I calculated the use of edge zones vs. the patch interior for each individual. The edge zone of the patch was defined as the part of the patch lying within 1 m of the patch edge. Because the patches were relatively small, I assumed the entire patch to be equally available to the animals. I defined the selection index SI following Manly (1974, eq. 3): SIedge=ratioedge/(ratioedge + ratiointerior), where ratio = (proportion of use)/(proportion of area). SIedgewas calculated for each patch*animal combination with at least five observations. SIedgecan vary from 0 to 1. Animals outside the patch were not included when SIedgewas calculated.
A rough measure of the range area of each animal was calculated from the median as well as the 10th and 90th percentiles of the x- and y-coordinates (xmed, x10, x90, etc.). The range area was calculated as the area of the irregularly diamond-shaped area between the points (x10, ymed),(xmed, y90), (x90, ymed),(xmed, y10). I did this for all animal-patch combinations with at least eight observations.
Results
Movement and song activity
The animals moved quite much on sunny days, but some of them could spend several days within a limited area (Fig. 2). The optimal model (as measured by Mallow's Cp) to explain movement length between subsequent observations on the same day (r) included all the variables considered: the time between observations (t), the average temperature during that time interval (temp), temp2, and the identity of the individual (ind):
log(r2) = 1.047 log(t) + 0.72 temp - 0.0109 temp2 + ind + e
All factors were significant with P < 0.001. The random component e had an estimated variance of 3.82 and appeared normally distributed. In the case of a random walk, the coefficient of log(t) is theoretically expected to equal one (Johnson 1992); thus, the value of 1.047 (SE: 0.147) indicates that the wart-biters' movement paths were close to a random walk on this scale. The temp and temp2 coefficients indicate that the movement length increased with the temperature (measured in the sun) for temperatures up to 28-30° C and flattened out thereafter. Lastly, the significance of the ind parameter (P < 0.0001) indicates that movement activity differed between individuals. Movement activity, as indicated by each individual's estimate of ind, did not depend on patch, patch size, catch location or sex (ANOVA, P0.40). There was an interaction between sex and condition (F1,50=4.76, P=0.034): males caught as adults (males in good condition) were significantly more active than males caught as instars and reared in the lab (males in poor condition; t21=5.68, P < 0.001). The same tendency prevailed for females but was not statistically significant (t29=2.10, P = 0.16).
As expected, singing activity was highest in from ca. 10 a.m. to about noon and declined thereafter (Fig. 3). In addition to display higher movement activity, males in good condition also tended to sing 3-4 times more often than males in poor condition (song frequency per individual; t21=5.89, P0.0001; Fig. 3). When they sang, good-condition males tended to sing much less loud than poor-condition males (means 1.08 vs. 2.79 on a scale from 0 to 3; t21=4.00, P0.001). Egg-laying was observed only 22 times (mostly just outside the patches); it was most frequently observed in the afternoon and evening (Fig. 3). There was no difference in egg-laying frequency with regard to condition (Fisher's exact two-sided test, P = 0.64).
Use of edge zone vs. patch interior
Use of the edge zone as measured by SIedge did not differ among the sexes, but differed among patches for animals reared in the lab. In the medium-sized patches, poor-
condition animals stayed significantly more in the edge zone than in the interior (selection indices > 0.5, Fig. 4, Tab. 1). In the small and large patches, the selection indices indicate that poor-condition animals on average used edges and interiors equally much. This effect of patch size was not found among good-condition males, and the interaction between patch size and condition was significant (Tab. 1). However, because most good-condition animals left the small patches early in the experiment, the estimates of good-condition animals in these patches are based on only four animals, of which two were were male immigrants who stayed for a short time. In the only small patch were the two groups coexisted for an extended period, the two good-condition animals (one male, one female) stayed more in the patch interior (SIedge = 0.32 and 0.37) than the three poor-condition animals (SIedge = 0.45, 0.62 and 0.69). Thus, this patch shows the same tendency as the medium patches: poor-condition animals of both sexes used the edge more than good-condition animals.