Increased dispersal rates and distances in density-stressed bush crickets

The limitation of dispersal due to habitat fragmentation is considered as one key factor for an increasing risk of extinction especially in sedentary species with low dispersal ability'. A few studies on grasshoppers and crickets indicate that dispersal ability even of sedentary species may have been underestimated in the past. It is argued that certain environmental conditions can trigger and increase dispersal in such species. In a first set of experiments emigration rates of the bush cricket Metrioptera brachyptera (L., 1761) were recor­ ded in ’’microcosms” (cage-populations) with various „population“ densities. The proportion of emi­ grants was correlated with increasing densities. In a second set of experiments individuals stressed by high densities as well as individuals that were not submitted to this stress were released on a soccer field and dispersal patterns of both groups were recorded. Stressed individuals dispersed significantly faster and further than unstressed ones. The results indicate that dispersal is induced by certain environmental condi­ tions suggesting episodic colonization patterns which occur even in sedentary species. Kurz Hügeligen Beißschrecke Metrioptera brachyptera (L., 1761) wur­ den in einem ersten Schritt die Emigrationsraten bei verschiedenen Populationsdichten in “Mikrokosmen” (Käfigpopulationen) untersucht. Der Anteil der Individuen, die aus den Käfigen auswanderten, war positiv mit der Individuendichte in den Käfigen korreliert. In einem zweiten Schritt wurden sowohl Indivduen, die unter Dichtestress gehalten wurden, als auch Tiere, die keinen hohen Individuendichten ausgesetzt waren, auf einem Fußballplatz freigelassen und ihr Ausbreitungsmuster untersucht. “Gestresste” Individuen di­ spergierten signifikant schneller und weiter als “ungestresste”. Die Ergebnisse weisen darauf hin, dass Ausbreitungsverhalten induzierbar ist und auch bei standorttreuen Arten auftreten kann. Es ist bei diesen jedoch keine regelmäßige Erscheinung, sondern besitzt Ausnahmecharakter.


Introduction
The limitation of dispersal due to habitat fragmentation is considered one of the key factors for the increasing risk of extinction of populations and metapopulations (SOULÉ 1986, GILPIN & K A N SK ! 1991, P r i m a c k 1993). Inter-patch dispersal, i.e. the movement of individuals between discrete patches of suitable habitat, is essential for colonization of vacant patches and is therefore of basic importance for metapopulation models (e.g. H ANSKI 1985, H A n s k i & TH O M AS 1994). One question arising from this concept is how to assess the contribution of single local-populations towards the persistence of the whole metapopulation and whether large or small local-populations provide more individuals for interpatch dispersal. Therefore, many studies focus on dispersal behaviour and both emigration-and immigration patterns of metapopulations (TH O M AS et al. 1992, HANSKI et al. 1994, KlNDVALL 1996, Recent studies reveal underestimations of both patch occupancy and dispersal distances even in sedentary species (H IL L et al. 1996, LE W IS et al. 1997, K lN D V A LL 1999. Regarding this, possible differences in motivation of individuals to disperse which are emphasized by H AN SSO N (1991), may generate a bias in such estimations. Dispersal motivation may be affected by population density but usually emigration-and immigration rates are assumed to be independent of patch sizes and population densities (e.g. H ANSKI & TH O M AS 1994). Recently, these basic parameters of metapopulation modelling have been discussed as being variable and dependent on population density, habitat quality and spatial features of the landscape (KUUSSAARI et al. 1996, BAGUETTE et al. 1998, KlN D VALL et al. 1998, KlN D VALL 1999. The triggers of emigration and dispersal, respectively, seem to be of crucial importance for mobility patterns (BA K E R 1984). High population densities are reported to play an important role in inducing dispersal of grasshoppers and crickets (A lK M A N & H EW ITT 1972, SÄNG ER 1984, L A U ß MANN 1998. Nevertheless, JO E R N & GAINES (1990) assume a genetically fixed dispersal beha viour that possibly is connected with invariable emigration rates. Density-dependent dispersal indicates that individuals, parts of populations or even entire species can switch from "sedentary" to "mobile" in a "facultative migration" (SOUTHWOOD 1962) as a reaction to environmental adversities (JOHNSON 1969). The above mentioned aspects led to the following hypothesis of the present study: Even very sedentary species are able to cover great distances and cross so-called dispersal barriers under certain environmental conditions. These patterns, however, occur only occasionaly at times when individuals of these species are forced to emigrate. Both emigration rates and dispersal ability are assumed to be positively correlated with popu lation density. Therefore, density-stressed populations are thought to produce a higher proportion of emigrants that cover great distances. The aim of the study is to show that emigration rates can increase with population density and to reveal a shift from sedentary to dispersive in individuals of the bush cricket Metrioptera brachyptera exposed to different density stress.

Methods
M icrocosm -experim ents to m easure em igration rates at different densities Eight acrylic glass cages (160 x 60 x 40 cm) housed eight "modehpopulations" of M. brachyptera in "microcosms" (c£ FRASER and KEDDY 1997). To provide comparable "habitat conditions" between cages, the cages had an identical setup: each had a three to five cm litter layer, two flower pots (20 cm in diameter) and was covered with gauze ( Fig. 1). The flower pots w7ere planted with Molinia caerulea which wT as the predominant species in the habitat where MAdgb/m?-individuals had been collected (wet heathland and bog mar gins). A 20 cm long rectangular plastic nozzle (20 x 6 cm) was fixed in a 45° angle to a rectangular hole in the front side of each cage acting as an exit similar to a fish trap (see fig. 1). A transparent plastic bag was attached to the end of the nozzle. I IV uf iCV6'r F ig. 1: Model-habitat o f M etrioptera brachyptera used in the density-dependent em igration experiments. In each experimental replicate eight acrylic glass cages housed eight «m odel-populations» of M . brachyptera. To provide comparable «habitat conditions» between cages, the cages had an identical setup: each cage had a three to five cm litter layer, two flower pots and was covered w ith gauze. The flowrer pots were planted with M olinia caerulea w hich was the predom inant species in the habitat the M etrioptera-individuals had been col lected (see methods).
In the cages, different numbers of individuals were kept at an approximately even sex ratio ranging from five to 40 individuals per cage (5, 10, 15, 20 etc.). A total number of 180 individuals for each replicate was collected on the day prior to the experiment and transported in separate cigarette boxes. For each replicate new7 individuals were caught in the field. During the seven days of each experimental replicate all individuals found to have emigrated into the plastic bags were counted twice a day (11:30 h and 17:00 h) by closing the bags below the nozzle exit and removing it with the enclosed individuals. These wT ere then released to a holding cage and a new7 bag w7as then fixed to the nozzle. After the 11:30 h counting, each "model-population" was provided with six flowers of ILeontodon autumnalis or Sonchus arvensis as a food supply in addition to Molinia caerulea. The cages were exposed on the roof of the institute building and after one replicate, each cage was moved to the former position of another one. Three of the replicates were carried out in 1997 followed by two in 1998 for both males and females: Aug. 5 -Aug. 11, 1997 (juvenile/adult) Aug. 15 -Aug. 21, 1997 (juvenile/adult) Aug. 29 -Sep. 4, 1997 (adult) Aug. 5 -Aug. 11, 1998 (adult) Aug. 21 -Aug. 27,1998 (adult) The same experimental design was used in a sixth replicate (Aug. 14 -Aug. 20,1998) to measure density-dependent emigration exclusively for females. All replicates were car ried out at comparable weather conditions favourable enough to allow behavioural pat terns like mating and egg-laying.
R elease-experim ents w ith individuals exposed to different densities To investigate whether population density has an impact on dispersal behaviour, the mobility patterns of individuals which experienced no density-stress were compared to "stressed" individuals (taken from the high-density-cages of the microcosm-experiments). This was done by releasing the two groups simultaneously on the lawn of a soccer field (100 x 62 m). This location was chosen to increase resight probability. The 130 "stres sed" individuals were released at one intersection of the centre circle with the centre line whereas the 224 "unstressed" individuals were released at the other intersection of the centre circle and the centre line of the soccer field. The two groups of individuals were tagged differently either with a silver permanent marker ("unstressed") or with a white one. The experiment lasted for 16 days and eight control walks were conducted either at 10:00 h, 16:00 h or 20:00 pm depending on weather conditions. No control walks were carried out during rainfall. Each resight session lasted 90 minutes and was performed on transects in a minimum intertransect distance of three meters. The transects extended to a maximum of 50 meters beyond the margin of the soccer field. Each day, the distances between release and resight point were measured for both "stres sed" and "unstressed" individuals. Thus, the number of recorded distances of one day corresponds to the number of resighted individuals of that particular day. The number of recorded distances of different days can, however, stem from the same individuals.

Results
In the five replicates carried out with males and females, a significant positive correlati on (Pearson, R=0.96, p 0.001, n=40) between the density of stocked individuals and the proportion of emigrants was found (Fig. 2a). The difference in the proportion of emi grants was significant between the cage stocked with five individuals and that stocked with 10 bush-crickets (p 0.05), and it was highly significant between the former cage and all cages with 15 individuals or more (p 0.001, ANOVA, post-hoc-test Bonferroni). Dif ferences between cages with more than five individuals were not significant. Thus, the density-dependent increase of emigration rates seems to be between the cage holding five individuals and the one holding 10 individuals. In order to detect sex-related diffe rences in emigration rates, the proportion of emigrants in both sexes was analyzed seperately (Fig. 2b). In males as well as in females the proportion of emigrants was positively correlated with density (males: Pearson, R=0.693, p 0.001, n=40; females: R=0.4913, p 0.001, n=40), but between the sexes no significant difference in the pro- portion of emigrants could be detected (/-test, 1=1.558, p=0.123, n=40). In the "early" mixed-sex-replicates (1. and 4.: Aug. 5 -Aug. 11) lower overall emigration rates than in the three later ones were found (3. -5.: Aug. 15 -Sep. 4) (Fig. 2c). The emigration rates of the sixth replicate (exclusively with females) differed consi derably from the replicates with both males and females (see fig. 2a). Here, even the lowdensity cages had a high proportion of emigrants. Furthermore, the emigration rates tend to decrease with an increasing density of stocked individuals. During the eight resight sessions of the release-experiment on the soccer field, 62 distances of at least 42 unstressed individuals and 72 distances of at least 40 stressed individuals were recorded (Resighting proportion: 28% vs. 55%; unstressed vs. stressed). Although the released specimens were not marked individually, the minimum number of resighted specimens could be obtained from specific characteristics (sex, colour, injuries) and by the number of simultaneously resighted individuals of the same day. The stressed individuals covered significantly higher distances in significantly less time than the unstressed ones (Mann-Whitney-U-test, U=1231.0, Z=-4.493, p 0.001, n=132) (Fig. 3). Dispersal distances increased rapidly until the third day after the release but started to slow down after the third day. On the 11th day after the release the maximum distance of 75 m from the release point was observed in a stressed individual far beyond the margin of the soccer field. A linear increase of distances with time was found in unstressed individuals. Therefore, the maximum distance from the release point of an unstressed individual (57 m) could be only recorded on the 16th day after the release. days after release

Triggers of em igration
The results of the cage-experiments suggest a positive correlation between density and emigration rates, at least in the model-populations of M. brachjptera examined. The main increase of emigration rates between five and 10 individuals per cage indicates that the resource supply (respectively the "habitat quality") of the cages may be sufficient for a density of about five individuals. The non-linear regression suggests that a certain thres hold (i.e. a "carrying capacity" of the cages) has to be exceeded before density-depen dent emigration is induced. Density-dependent emigration in grasshoppers and crickets is reported by SÄNGER (1984), R e m m e r t (1992) and K lN D V A L L et al. (1998). Since the later replicates show a higher overall proportion of emigrants, "older" indivi duals seem to be more sensitive to density-stress. Similarly, males of Metrioptera bicolor move longer distances per day later in the season (K lN D V A L L , pers. com.). It is known that the intensity' of male stridulation e. g. the number of stridulating males act as a trigger of emigration (M c H u g h 1972, REMMERT 1992). A r a k et al. (1990) suggest that mating success of male Pettigonia viridissima will be maximized when singing males space out as far as possible. In butterflies, an increase of male harassment on mated females or an increase of territorial fights at high population densities leads to emigration (e. g. Although carried out only once, the high emigration rates of the low-density cage in the female replicate seem to contradict a positive correlation between emigration and densi ty. But it may, however, point at an increase of emigration rates at very low densities as well as at high densities in the mixed-sex-replicates. The low female density with absent males may be interpreted (by the female) as a sign of bad habitat quality (e.g. bad egglaying conditions) and thus induce emigration. This is supported by the high emigration rates at very low densities as was shown for butterflies (KUUSSAARI et al. 1996), an effect that was possibly caused by a lack of conspecific attraction. Although the mixed-sex-replicates reveal no different emigration rates of males and females, the female replicate suggests -at least under certain circumstances (e.g. male absence) -sexual differences in the response to high densities. However, no differences were detected under the more or less equal sex ratio given in the mixed-sex-replicates. Nevertheless, sexual differences in the emigration behaviour may occur, if, for instance, the operational sex ratio shifts towards a dominance of one sex (e.g. due to the mating status of females). These aspects of different sexual movement patterns need further investigation.
In addition to ''running" further and faster, as it may be supposed because of the release-experiments, density-stress provides another way to considerably increase disper sal distances in rather immobile grashoppers and crickets: getting long wings. Increasing population densities ("crowriing") lead to the production of certain pheromones in in stars of Locustinae-grasshoppers. These pheromones are known to induce a shift from a sedentary to a migratory phase with long wings (NO LTE 1977, D A LE & TO B E 1990, LO H E R 1990 wrhich then emigrates in the well-known swarms (FARRO W 1990). Long winged macropterous morphs are described from several european crickets and grass hoppers which are usually apterous or mesopterous and therefore rather immobile (Tab. 1). The most impressive example may be the nowadays rare species Volysarcus denticauda. This usually apterous cricket is reported to have covered great distances as a long-winged (!) morph in the 1940's probably due to "crowding" (EBN ER 1950/51, E N G E L 1951. BRU N ZEL (1999) observed long-winged morphs also in M. brachyptera performing flights of more than 20 meters.

Impact of density-induced dispersal-patterns on the conservation of populati ons in fragm ented landscapes
The release experiments of this study suggest that individuals which experienced high densities dispersed faster and further than individuals which were not exposed to densi ty-stress. But the significant differences found are linked to faster spreading of stressed indviduals rather than to higher distances covered. The resighted unstressed individuals, although fewer, seem to have covered similar distances but much later than the stressed ones. It took unstressed individuals longer to cover similar distances as stressed ones, resulting in a longer exposure to predators before they reach a new7 patch. Regarding this, the surprisingly lower resighting-proportion of unstressed individuals (28% : 55%; unstressed : stressed) may in fact be caused by predators: birds were observed preying on released Mririopifera-individuals. A constant rate of loss due to predation would inevitably lead to a lower resighting-proportion of slow7er-spreading individuals. This also would explain the higher proportion of lower dispersal distances which are found in the unstres sed individuals: most of them may get lost before being able to cover high distances. The hypothesis o f a faster spreading o f stressed individuals is supported by RlEGERT et al. (1954). They note that density-stressed grasshoppers disperse faster than supposedly unstressed individuals (low density). However, they did not test individuals stocked at low' densities. AIRMAN & HEWITT (1972) observed decreasing dispersal speed in the course o f a release experiment and linked that to the decreasing density due to the spreading.
Increasing density in M. brachyptera seem to cause a higher proportion o f em igrants covering greater distances in shorter time. High population densities in sedentary, im m obile grasshoppers can produce m acropterous m orphs w'hich are k n o w n to perform dispersal flights (UVAROV 1977, FARROW 1990. Howyever, RITCHIE et al. (1977) do not consider m acroptery as im portant for dispersal due to the "oogenesis-flight-syndrom e" .
The results o f this study support the hypothesis that dispersal is positively correlated w ith population density and therefore is condition-dependent, such that gene flow' and colonization m ay be episodic. T he higher proportion o f stressed em igrants dispersing faster than unstressed individuals could lead to an increase o f colonization probability and thus to an expansion o f a species (Fig. 4) basic parameter of metapopulation-modelling, the positive effect of a density-induced increase of emigration rates and dispersal distances on the probability" of colonizing vacant patches should be taken into account even in sedentary species. As results of KEAN & B a r l o w (2000) suggest, the density-dependence of modelling-parameters has a considerable impact on metapopulation persistence.