Effects of habitat structure and land-use intensity on the genetic structure of the grasshopper species Chorthippus parallelus

Land-use intensity (LUI) is assumed to impact the genetic structure of organisms. While effects of landscape structure on the genetics of local populations have frequently been analysed, potential effects of variation in LUI on the genetic diversity of local populations have mostly been neglected. In this study, we used six polymorphic microsatellites to analyse the genetic effects of variation in land use in the highly abundant grasshopper Chorthippus parallelus. We sampled a total of 610 individuals at 22 heterogeneous grassland sites in the Hainich-Dün region of Central Germany. For each of these grassland sites we assessed habitat size, LUI (combined index of mowing, grazing and fertilization), and the proportion of grassland adjoining the sampling site and the landscape heterogeneity (the latter two factors within a 500 m buffer zone surrounding each focal site). We found only marginal genetic differentiation among all local populations and no correlation between geographical and genetic distance. Habitat size, LUI and landscape characteristics had only weak effects on most of the parameters of genetic diversity of C. parallelus; only expected heterozygosity and the grasshopper abundances were affected by interacting effects of LUI, habitat size and landscape characteristics. The lack of any strong relationships between LUI, abundance and the genetic structure might be due to large local populations of the species in the landscape, counteracting local differentiation and potential genetic drift effects.


Summary
Land-use intensity (LUI) is assumed to impact the genetic structure of organisms. While effects of landscape structure on the genetics of local populations have frequently been analysed, potential effects of variation in LUI on the genetic diversity of local populations have mostly been neglected. In this study, we used six polymorphic microsatellites to analyse the genetic effects of variation in land use in the highly abundant grasshopper Chorthippus parallelus. We sampled a total of 610 individuals at 22 heterogeneous grassland sites in the Hainich-Dün region of Central Germany. For each of these grassland sites we assessed habitat size, LUI (combined index of mowing, grazing and fertilization), and the proportion of grassland adjoining the sampling site and the landscape heterogeneity (the latter two factors within a 500 m buffer zone surrounding each focal site). We found only marginal genetic differentiation among all local populations and no correlation between geographical and genetic distance.  agriculture, including arable land and grasslands, but also contains one of the largest continuous broadleaved forests in Germany. The proportion of grasslands is comparatively small as major parts comprise intensively used arable land (figure 1).

Study sites and land-use intensity
For each of the 22 grassland sampling localities, we assessed habitat size and various land-use types: hay meadows (six sites), mown pastures (eight sites) and pastures (eight sites). Meadows were mown once (one site) twice (four sites) or three-times (one site) per year. Hay meadows and mown pastures were fertilized. Mown pastures and pastures were grazed either by sheep (seven sites) or cattle (nine sites). For our analyses, we used a quantitative, continuous index of LUI. This index combines the main human management measures of the grasslands, mowing, grazing and fertilization, into a single standardized index (see [13] for a detailed description of how the index LUI is calculated). For this study, we used averaged management measures over the period [2006][2007][2008]. For statistical analyses, we additionally carried out analysis using land-use categories, by dividing all sites into the three categories: low LUI (index less than 1.5), medium LUI (between 1.5 and 2.0) and high LUI (more than 2.0).
To analyse the effect of landscape structure, we assessed the environment in a 500 m buffer around a grassland plot on the basis of CORINE (2006) data (Pan-European project CORINE land cover-CLC, available from http://www.corine.dfd.dlr.de, accessed January 2014). From this, we calculated the percentage of grassland (CORINE types 231, 321) in the circle, the most suitable habitat of C. parallelus. Additionally, we calculated habitat diversity as Shannon diversity index based on the surrounding environment, split into the following categories: arable land (

Abundance of Chorthippus parallelus
We used standardized sweep-net samples (round sweep net with a 30 cm diameter) to assess the abundance of C. parallelus. Twenty double sweeps along each of three transects (total 60 double sweeps) were performed twice a year (June and August) during the years 2008-2012 by the same person. Arthropods collected from the three transects were pooled and preserved in 70% ethanol. Orthopterans were separated from other insects. Individuals of C. parallelus were counted. We used the summed number of individuals over the 5 years as measure of C. parallelus abundance to account for fluctuations in population densities among years. We neither performed further census calculations nor calculated any density estimates, as studies showed that extrapolations of sweep netting on square metres and meadow sizes will result in much more inaccurate values because of the unknown local species distribution [11,14].

Population genetic analyses
We tested for distortion of microsatellite data due to stutter bands, large allele dropout or null alleles using the program MICRO-CHECKER v. 2.0 [19]. Tests of Hardy-Weinberg equilibrium and linkage disequilibrium were conducted with the program ARLEQUIN v. 3.5 [20]. We calculated four parameters of genetic diversity for each population: mean number of alleles A, observed heterozygosity H o and expected heterozygosity H e using the same program, while FSTAT v. 2.9.3.2 [21] was used to calculate allelic richness AR, the mean number of alleles based on the lowest number of individuals (here 20 samples) with the rarefaction option. Further, we calculated locus-and locality-specific allele frequencies with this program. Analyses of molecular variance (AMOVAs) were performed to partition the genetic variance on three levels: genetic variance located among populations, among individuals within populations and within individuals. Respective fixation indices were calculated with the program ARLEQUIN. To test for potential correlations between genetic and geographical distance (isolation-by-distance), we correlated pairwise genetic distances [F ST /(1 − F ST )] with the natural logarithm (ln) of the geographical distance, using the ISOLATION BY DISTANCE WEB SERVICE v. 3.23 (http://ibdws.sdsu.edu/) [22] with 10 000 permutations to test for significance.

Overall statistical analysis
We used generalized linear mixed effects models (GLMM) to test for potential relationships between the following parameters: abundance of C. parallelus with size of grassland site, with LUI, with the percentage of surrounding grassland (within a 500 m radius) and with the Shannon diversity index (500 m radius). In a second analysis, we tested for potential relationships between all four parameters of genetic diversity (A, AR, H e , H o) and abundance of C. parallelus, land use, and all other landscape parameters. In both analyses, we included two-way interactions between land use and habitat size and landscape variables to test whether grassland size and edge-habitat-size effects depend on LUI. We used a stepwise model selection by AIC (backward and forward selection) with the function stepAIC in the package MASS in R v. 2.14.0 (http://www.r-project.org/) [23,24]. Abundance of C. parallelus, habitat size and percentage of surrounding grassland were ln-transformed prior to analysis to improve normality of residuals and homoscedasticity.

Genetic diversity and differentiation
We found no significant linkage disequilibrium and deviations from Hardy-Weinberg equilibrium and only marginal effects due to null alleles. Genetic diversity was homogeneously distributed over all

Effects of land use and landscape
Habitat size, LUI and surrounding landscapes differed among the studied sites. Habitat size ranged from 2.63 to 128 ha (with a mean of 21.4 ± 29.3 ha s.d.). LUI varied from very low (0.59) to very high (2.73) (with a mean of 1.67 ± 0.67 s.d.). The percentage of grassland in the adjoining environment varied from 0 to 88.0% (with a mean of 39.4 ± 32.2%). The heterogeneity of the surrounding environment measured as Shannon diversity index ranged from 0.00 to 1.12 (0.62 ± 0.29). Values for each site are given in table 1.
The sweep netting resulted in total numbers of C. parallelus individuals between 3 and 124 per site (mean 24 ± 2.97 s.d.). The abundance of the grasshopper species was significantly correlated with the size of patches, habitat heterogeneity of the surrounding environment and LUI ( figure 2 and table 3). The  abundance was negatively correlated with habitat size at low and medium LUI; a positive relationship was observed at high LUI. Landscape heterogeneity generally affected the abundance of C. parallelus negatively, but the relationship was stronger at medium and high LUI. Genetic diversity was not strongly affected by habitat size, LUI and the surrounding landscape, except for the expected heterozygosity. Here, a positive relationship between habitat size and H e could be observed at high LUI (figure 2). Further, H e was negatively affected by the abundance at low LUI, but positively at medium and high LUI (figure 2).

Genetics of a widespread and abundant invertebrate
The genetic analysis on the 22 local populations of the common and abundant meadow grasshopper revealed a low level of genetic differentiation across all populations, no isolation-by-distance, and an almost homogeneously distributed genetic diversity. Correlations between abundance, parameters of genetic diversity and biotic and abiotic characteristics of the grassland sites were significant yet only weak. These results suggest that the strong fragmentation of grassland habitats has no effect on the genetic diversity and differentiation of C. parallelus. Negative effects of fragmentation such as genetic impoverishment and strong population differentiation often observed in organisms living in similar fragmented habitats may be overcome by large population sizes and strong gene flow among local sub-populations.
Our data are in line with the high abundance of C. parallelus, but contradict the general assumption that this grasshopper species is mostly flightless and shows rather low dispersal ability [9,12]. The species' comparatively broad ecological tolerance allows the species to inhabit a variety of habitat types within a large geographical range and leads to its widespread occurrences and high abundances [12]. The observed low genetic differentiation and homogeneously distributed and high genetic diversity have frequently been observed for many generalist grasshopper species, while, in contrast, specialist species often show opposite genetic patterns indicating low genetic diversity and strong differentiation  [25][26][27][28]. Similar trends can be found for other invertebrates, such as butterflies: here the most specialized butterflies of Europe, representatives of the genus Maculinea, show a comparatively low genetic diversity and high differentiation among local populations, while other lycaenids such as the widespread Polyommatus icarus or Polyommatus coridon show a comparatively high genetic diversity and low genetic differentiation [4].

Land-use intensity, species abundance and genetic diversity
The abundance of C. parallelus was not significantly affected by LUI. This opposes other ecological studies documenting a reduction in Orthopteran abundance due to the impact of mowing [29,30]. Humbert et al. [29,30] and Gardiner & Hassall [31] suggested that lowered abundances might be the result of a combination of mortality caused directly by the physical damage during mowing as well as the high sward temperatures created by removal of the standing crop. In their study, the abundance of C. parallelus and Chorthippus albomarginatus showed a significant decline in the abundance in study plots which are mowed compared with unmanaged control swards (which we did not incorporate in our analysis) [31]. Generally, there is an impact continuum of grassland management on grasshopper populations, depending mainly on date of the season, and exact method and frequency of management [11], so that LUI will somewhat superpose the different local situations.
In our study, the abundance of C. parallelus was significantly affected by habitat size and the heterogeneity of the surrounding environment, yet the direction and significance of the respective relationship depended on LUI. Chortippus parallelus seems to benefit from increasing habitat size but suffers from higher environmental heterogeneity under high LUIs. Greater patch sizes and lower landscape diversity in the surrounding patches probably result in larger areas of suitable habitat and subsequent larger population sizes and increasing connectivity.
The strong variation in the abundances of C. parallelus across our study area might additionally be the result of differences in abiotic conditions such as microclimate (temperature, humidity), or different plant species compositions [11]. Previous studies showed that the fecundity of C. parallelus is positively affected by the temperature and moisture of the grasslands [32], and that diet can strongly affect its fitness [8,33]. Thus, differences among local population abundances might rather come from different biotic and abiotic factors such as soil condition (humidity), elevation and inclination-having a higher relevance than the LUIs. A final factor affecting the values of the abundance of the grasshopper might be due to sampling bias.
Taking into account that with 10 double sweeps about 8 m 2 are swept, and that only 25% of all individuals become sampled [11], our results from sweep-netting underline that the local population sizes must be very high, ranging from several thousands to some tens of thousands per study site (50 m × 50 m). The high and equally distributed genetic diversity found within populations might be the result of high population sizes ranging at a very high level-at which differences might play a rather negligible role for potential genetic effects [34].
Genetic parameters showed no significant correlation with any of the habitat characteristics (e.g. size of the grassland site, LUI, the percentage of grassland in adjoining landscape or heterogeneity of the surrounding environment). Only the expected heterozygosity was significantly affected by the abundance of C. parallelus and by habitat size; this correlation was strongly positive but only found in plots with high LUI. Furthermore, this goes in line with higher abundances of C. parallelus found in these plots. This relationship between population sizes and genetic diversity (here expressed by expected heterozygosity) can frequently be observed in wild populations [35]. However, expected heterozygosity is an estimate (and not measurements of in situ diversity found in individuals) based on allele frequencies and the Hardy-Weinberg equilibrium assumptions, which makes the significance of this correlation between LUI and genetic diversity arguable [36].
In conclusion, the lack of effects from different LUIs can be explained by two scenarios. First, the unspecific habitat requirements and high ecological tolerance of C. parallelus may lead to high abundances, even in landscapes with intensive land-use regimes. This allows the species to exist in large population sizes counteracting potential genetic drift effects. It needs to be added that the total area of grassland in the study area is still quite large, so possibly stronger fragmentation and decreased total habitat area may lead to a decrease in genetic diversity and higher differentiation. Second, even if habitats and populations are small (as observed in some sites for some years), high abundances can easily counteract potential drift effects by gene flow from adjoining populations, and thus prevent the loss of genetic diversity. These two explanations might play a pivotal role in buffering potential effects