Experimental evidence for density-determined wing dimorphism in two bush-crickets ( Ensifera : Tettigoniidae )

Macroptery is common in many species of Orthoptera, but the causes are still discussed. Besides the assumption that macroptery is genetically determined, there is evidence that wing dimorphism is induced by environmental factors, particularly population density. However, most of the research is on pest species. In contrast, knowledge of wing dimorphism in species that occur at low population densites is still poor. Our study aims to test how density actually affects macroptery. As model organisms we chose two bush-cricket species of the genus Metrioptera (Ensifera: Tettigoniidae): While long-winged M. roeselii (Hagenbach, 1822) occur regularly, macropterous M. brachyptera (Linnaeus, 1761) are rare and are never observed outside their mating habitat. Nymphs of populations from the range core of both species (340 individuals each) were reared in groups of three and six individuals per 500 cm3 box, and individually. Our analyses revealed that development of macropters was mainly affected by the initial rearing densities. Compared with those reared individually the number of macropters was significantly higher among individuals reared at medium and high densities. The percentage of macropterous individuals was about twice as high in M. brachyptera as in M. roeselii, and the development of macropters significantly differed between the two species. These findings lead to the conclusion that macropterism is mainly influenced by density stress in both bush-crickets. Genetically determined wing dimorphism is unlikely, otherwise the observed high numbers of long-winged individuals of M. brachyptera, which are very rare under natural conditions, would never have developed in the laboratory. Macropterous M. brachyptera may rarely be found in the field, but we argue that this is due to low natural densities and, accordingly, to rare exposure to density stress. 599 * Corresponding author: e-mail: fartmann@uni-muenster.de ropterous M. brachyptera when reared at different densities. MATERIAL AND METHODS

The bush-cricket genus Metrioptera (Ensifera: Tettigoniidae) represents an ideal model system for studying wing dimorphism as it consists of several species that exhibit macroptery (e.g.Marshall & Haes, 1988;Fartmann, 1997;Thomas et al., 2001).For our experiments we used two related bush-cricket species (M.roeselii and M. brachyptera) that vary in their habitat requirements and their propensity to produce long-winged individuals in nature: While macropterous individuals of the habitat specialist M. brachyptera are rare and are never observed outside the mating habitat (Schouten et al., 2007), longwinged individuals of the habitat generalist M. roeselii occur regularly (e.g.Thomas et al., 2001;Simmons & Thomas, 2004;Poniatowski & Fartmann, 2008b).
Here, we test if and how density affects wing dimorphism in two related bush-cricket species.In addition, we want to explain why long-winged M. roeselii occur regularly in nature, whereas observations of macropterous M. brachyptera remain quite rare.If macropterism is induced by genetic factors, as suggested by Simmons & Thomas (2004), one would expect very few or no mac-ropterous M. brachyptera when reared at different densities.

Experimental design
Nymphs of Metrioptera roeselii and M. brachyptera (each n individauls = 340) were collected in the field between 9 and 22 May, 2008.Collection of nymphs took place in central Germany (Diemeltal, 51°28´N, 9°08´E; Medebacher Bucht, 51°10´N, 8°40´E) from populations at the core of the species range (distance of at least 100 km to the range margin) of both species (M.roeselii: n sites = 4; M. brachyptera: n sites = 3).Following Simmons & Thomas (2004) only first-and second-instar nymphs were collected, because macroptery may be induced during early nymphal stages (Köhler, 2002).These nymphs were reared to the adult stage under a 14L : 10D photoperiod at 25°C (SD = 2°C) (Higaki & Ando, 2003) in transparent plastic boxes (500 cm³) covered with dark nylon gauze.Humidity was kept at 45% (SD = 8%) because at higher humidities a high percentage of the nymphs fail to develop (Ingrisch, 1978).Nymphs of both species were reared in groups of three (nboxes = 40) or six (n boxes = 30) individuals per box.Moreover, forty nymphs of both species were kept individually as a reference (control).Box position was randomised.Preliminary studies showed that for successful rearing a mixed diet of plants and insects is necessary (Poniatowski & Fartmann, unpubl. data).Therefore, nymphs were fed ad libitum with fresh grass, blossoms and seeds as well as cricket food (JBL TerraCrick) to provide vitamins, proteins and minerals.Water was supplied by spraying the boxes every day and offering pieces of fresh cucumber (replaced every second day) (cf.Helfert & Sänger, 1975).
At maturity, fore-wing length of each individual was measured using a calliper gauge (0.1 mm accuracy).As a reference, short-winged (brachypterous) individuals of both species were also measured in the field (M.roeselii: n individuals = 79, n sites = 4; M. brachyptera: n individuals = 119, n sites = 6).A bush-cricket was classified as long-winged (macropterous) whenever wing length clearly exceeded the usual measure (Figs 1 and 2), i.e. if they reached the end of the abdomen or overlapped the hind knees (Ramme, 1951).

Statistical analyses
As our data did not fit the assumption of a t-test (i.e., a normal distribution; Kolmogorov-Smirnov test), a Mann-Whitney U-test was used to compare two independent samples.Differences between more than two continuous variables were analysed using Kruskal-Wallis H-test (incl.Mann-Whitney U-test with Bonferroni correction).Differences in frequencies of macropterous individuals and survival rate were compared for the three rearing densities using ²-test.Where observed values were 0, 1 was added to each of the three classes (Leyer & Wesche, 2007) to allow ²-test.Frequencies of survival rate per box were correlated with initial density per box using Spearman's rank correlations (test of significance: two-tailed).All statistical tests were done using SPSS 11.5 statistical package.
To evaluate which parameters affect the development of macropters a binomial generalized linear mixed-effects model (GLMM: lmer, Bates et al., 2008) using R-2.9.0 (R-development-core-team 2009) was used (Table 1).The significance of the predictor variables and interactions were assessed using likelihood ratio tests (Type III test).Nonsignificant predictors were excluded from the final model.
There were also significant differences between the rearing densities concerning the percentage of macropters

601
Nymphs were collected from different sites (see also material and methods).(Table 2).The GLMM showed that the development of macropters can be explained by medium and high initial rearing densities (Table 3).However, the percentage of macropters did not increase with increase in rearing density (Table 2).The percentage of macropterous individuals was about twice as high in M. brachyptera as in M. roeselii (Table 2), and the probability of macroptery differed significantly between the two species (Table 3).The median of wing length was highest at medium densities, although there was no significant difference between medium (3 individuals/500 cm³) and high (6 individuals/500 cm³) densities (Figs 3 and 4).Individually-reared M. brachyptera had significantly longer wings than field-grown individuals (Fig. 4), while those of M. roeselii did not differ (Fig. 3).

DISCUSSION
Our laboratory findings show that the macroptery of both bush-crickets (M.roeselii and M. brachyptera) is determined by density stress.Sänger & Helfert (1975) obtained similar results for M. roeselii: 21% of all individuals became long-winged when reared at high densities, irrespective of the climatic regime (temperature and humidity).Other laboratory studies on bush-crickets also support the hypothesis of macroptery being densityinduced (Ando & Hartley, 1982;Sänger, 1984;Higaki & Ando, 2003).
Since temperature and humidity were the same at all rearing densities, these environmental parameters can be excluded as a driver of macropterism, otherwise there should have been high percentages of macropters among the individually-reared bush-crickets.Taking this into account, pheromones as main triggers of macropterism can also be excluded since all boxes were covered with nylon gauze allowing gas exchange between the boxes with different rearing densities.In contrast, tactile and possibly visual stimuli seem to be important.An excess of these stimuli due to crowding triggers the release of endo-602   3. Statistics of GLMM [Pseudo R² (Nagelkerke's) = 0.29; n individuals = 400]: Relationship between macroptery (response variable) and "density" as well as "species" (predictor variables)."Sex" was not a significant predictor.For more information see Table 1.
crines that induce the development of long wings (Zera & Denno, 1997;Zera, 2004).However, not all individuals reared under medium and high densities become macropters.There are two likely reasons for that: (i) There is a strong degree of individual variation in the affinity to become macropterous (genetic disposition) or (ii) the heterogeneity within the boxes leads to an uneven encounter rate among the individuals, which means that there is variation in the stimulation individuals received (cf.Harrison, 1980).However, phenotypic plasticity does not only differ intra-specifically but also inter-specifically.In M. brachyptera far more individuals developed into the long-winged form than in M. roeselii.M. brachyptera to be more sensitive to density stress.Hence, even some individually-reared individuals became longwinged and wings in general were longer than those of field-grown individuals.Probably this was a consequence of increased tactile stimuli due to disturbance during daily feeding of the bush-crickets and cleaning of the boxes.It might be argued that the good food quality promoted body and wing growth.However, this is unlikely as wing length did not differ between individually-reared M. roeselii and individuals in the field.
Besides population density habitat disturbance may have an influence on the percentage of macropters.Zera & Denno (1997) view macropterism amongst others as an evolutionary adaptation to disturbance: Whereas in disturbed habitats significantly more macropterous, highly mobile, individuals occur, in more persistent, undisturbed habitats short-winged individuals with a high fecundity dominate (e.g.Denno et al., 1996Denno et al., , 2001)).These observations made on planthoppers seem to be transferable, at first glance, to the two bush-crickets studied.M. roeselii is a species that is often long-winged (Simmons & Thomas, 2004;Poniatowski & Fartmann, 2008b) and colonises disturbed habitats like grassland and roadside verges (Marshall & Haes, 1988;Kleukers et al., 1997).M. brachyptera is a typical species of habitats with little or no land use (e.g.Molinia caerulea-dominated vegetation and abandoned calcareous grassland) (Poniatowski & Fartmann, 2008a;pers. observ.).Macropters are quite rare in the field (Marshall & Haes, 1988;Kleukers et al., 1997).Accordingly, we were surprised that the percentage of macropters in our experiment was higher in M. brachyptera than in M. roeselii.A possible explanation might be the different densities of the species in nature.According to Ingrisch & Köhler (1998), population densities of M. brachyptera vary between 0.1 und 3.8 adults/10 m² (n studies = 5), which is very low compared to M. roeselii (0.7-11 adults/10 m²; n studies = 5).Consequently, M. brachyptera is seldom exposed to density stress in its natural habitats.
In contrast to our results, Simmons & Thomas (2004) did not find any effect of density on the development of macropters.They regarded genetically induced macroptery as more likely than density-induced wing dimorphism.According to Simmons & Thomas (2004), in species with expanding ranges like M. roeselii, the development of macropters should occur more frequently at the edges of its range than in established populations at the core of the range, because of differences in the selective advantage of dispersal.Consequently, species with static range margins like M. brachyptera are considered lowdispersal specialists (Simmons & Thomas, 2004).Taking 603 Fig. 4. Metrioptera brachyptera: Wing length of short-winged individuals from the field and individuals reared at different densities in the laboratory (initial density: 1, 3 or 6 per 500 cm³ box).Kruskal-Wallis H-test for females: ² = 49.064,d.f.= 3, P < 0.001, and for males: ² = 89.747,d.f.= 3, P < 0.001 (For further explanations see Fig. 3).
this into account, one would expect no or very low numbers of long-winged individuals of M. brachyptera in our experiment.However, we observed the exact opposite: under medium and high densities more than 60% of all individuals became macropterous.The percentage of macropterous M. brachyptera was even significantly higher than in M. roeselii.A density stress-induced macroptery is thus more likely than genetically determined wing dimorphism in bush-crickets.Maybe there is a combination of both, as is the case in many insect species (Zera & Denno, 1997), with a dominance of density effects.