A further five specimens were too damaged to be identified and we

A further five specimens were too damaged to be identified and were excluded. All species were classified into three habitat-preference categories: sand-dwelling, open ground-dwelling and forest-dwelling, Paclitaxel based on information from Hansen (1964), Koch (1989–1992), Lindroth (1961) and Palm (1948–1972). A few species did not fit into any of the three categories and were classified as ‘indifferent’. The categories sand-dwelling and forest-dwelling selleck products included species specialized for living, or mainly living, in the respective habitats, whereas open ground-dwelling species also included

generalists and species occurring in other habitats. The species in each category are hereafter referred to as ‘sand species’, ‘open ground species’ and ‘forest species’. Red-listed species were defined after Gärdenfors (2010). Data analysis For each site, the beetle data collected were pooled. All species data were Selleck Staurosporine included in the analysis, despite some differences in sampling intensity. To handle these differences, sampling intensity, calculated as the

number of trap days per site, was included in all regression models and in the ordinations as a covariable. The SAR was tested using two models: the commonly used log–log power function, S = c A Z (Arrhenius 1921; Tjørve 2003), and a curved model called the quadratic power function, S = 10(b0+b1 logA+b2 (logA)2) (Chiarucci et al. 2006), where S = species number, A = area, z = the slope (z value) and c and b x are constants. The models were chosen to fit our empirical data and according to Dengler (2009) both models generally perform well. The species numbers were log10(n + 1) transformed since they included zero-values.

The area variables were log10-transformed in accordance with the models. Two measures representing the size of the sand pit (total area and area of bare ground) were tested parallel to see their relative ability in predicting species number. The z values were calculated without sampling intensity as a covariable. Linear regressions were performed to analyze the effects of the measured environmental variables on the numbers and proportions of all beetle species and carabid species, respectively. The variables were tested both individually and in multiple regressions by stepwise regression (combining both forward selection and backwards elimination) to identify Urease significant variables (p < 0.05). For the multiple regressions, the covariable sampling intensity was added afterwards when the significant subset of variables had been identified. The adjusted R 2 values were used throughout, so that the number of explanatory variables included would not influence the goodness of fit. For carabids, the data from the study site Nyboda were not included in the regressions that included the proportion of species, as the low total number of species (two) gave a misleading value (and an outlier) for the proportion of sand species (100%).

Comments are closed.