Electronic Supplementary Material s44

Electronic Supplementary material

Distribution patterns of European lacustrine gastropods – a result of environmental factors and deglaciation history

Elisavet Georgopoulou1*, Thomas A. Neubauer1, Mathias Harzhauser1, Andreas Kroh1, Oleg Mandic1

1Geological-Paleontological Department, Natural History Museum Vienna, Burgring 7, 1010 Vienna, Austria

*Corresponding author: Elisavet Georgopoulou, e-mail: , tel: +43 1 52177-254

Appendix S1

To test for the influence of habitat availability and endemic species we included in the current Appendix further analyses carried out in two different datasets. The results of these were subsequently compared to the original dataset of the manuscript (OD). The first set (Set 1) is almost identical to the OD, but surface area was replaced by lake perimeter. Surface area and lake perimeter were never included in the same model because of high collinearity (Table A1). Lake perimeter was reconstructed like surface area in Google™ Earth 7 and was used as a surrogate for habitat availability (see Dehling et al., 2010 and references therein). In the second set (Set 2) the predictor variables are the same as the OD, but here single lake endemics (SLE) were removed in order to evaluate the effect of endemism on the differences between the lake groups. For both sets we implemented stepwise multiple regressions (Tables A2, A3) and evaluated the spatial autocorrelation of the regressions’ residuals using Moran`s I index as for the OD. For each lake group, the resulting regression models of the two sets were compared to the model of the OD using ANOVA. Additionally, the beta diversity analysis was repeated for Set 2 in order to test how endemic species affect species composition (Table A4)

Regarding both Sets low or no spatial autocorrelation was observed in the models’ residuals (Fig. A1, A2). The comparison of Set1 with the OD revealed that none of models differs significantly from the original one except for the model for LG4 (F = 10.841, P = 0.001). In all models, the contribution of perimeter was similar to that of surface area and was prominent only in LG4 (Table A2). This provides indication that habitat availably is an insufficient predictor of lacustrine gastropod richness suggesting that lake faunas in a pan-European scale and within the formerly glaciated areas are not in equilibrium (see also Dehling et al., 2010). However, faunal evolution seems to be stronger linked to habitat availability in lakes not directly affected by glaciations (LG4). This is likely affected by the higher longevity of some of these lakes (e.g. Balkan lakes). To disentangle, however, the faunal patterns within LG4 additional information are required.

The comparison of the OD with Set 2 did not yield significant differences and thus we conclude that the presence of the SLE does not alter the patterns of species richness. Accordingly, differences in species composition were not strongly influenced by the exclusion of the SLE but the pairwise beta diversity was lower (compare Table 3 and Table A4). The balance between species loss and species turnover between the LGs remained the same indicating that the patterns of beta diversity of European lacustrine gastropods are independent of SLE.

REFERENCES

Baselga, A. (2012) The relationship between species replacement, dissimilarity derived from nestedness, and nestedness. Global Ecology and Biogeography, 21, 1223–1232.

Dehling, D.M., Hof, C., Brändle, M. & Brandl, R. (2010) Habitat availability does not explain the species richness patterns of European lentic and lotic freshwater animals. Journal of Biogeography, 37, 1919–1926.

Table A1 Multicollinearity test results for the complete dataset and the four lake groups (LGs) 1–4 for Set 1. Variation inflation factor greater than 10 indicates increased multicollinearity.

Lat / Long / Alt / Area / Perim / Temp / Prec / Isol
Complete dataset / 2.95 / 1.45 / 1.59 / 36.31 / 36.84 / 2.86 / 1.41 / 1.14
LG1 / 3.06 / 3.23 / 1.41 / 40.93 / 41.26 / 3.02 / 2.57 / 1.24
LG2 / 7.05 / 2.39 / 1.41 / 35.81 / 36.15 / 7.28 / 4.27 / 1.14
LG3 / 2.17 / 2.82 / 5.35 / 103.14 / 98.20 / 4.68 / 1.67 / 1.47
LG4 / 2.72 / 1.51 / 2.50 / 36.87 / 36.64 / 4.16 / 1.12 / 1.32

Lat, latitude (decimal degrees), Long, longitude (decimal degrees), Alt, altitude (metres), Perim, lake perimeter (km), Temp, mean annual temperature (°C), Prec, annual precipitation (mm) and Isol, Isolation (km).

Table A2 Predictor variables affecting gastropod species richness of European lakes as proposed by stepwise multiple regressions (MR) for Set 1. Shown are the adjusted R2 and P-values (P) of the MRs (R2adj), Akaike’s information criterion (AIC), the difference between initial and final Akaike’s information criterion (ΔAIC), the estimate of the multiple regression (Est. MR), significance of the variables in the multiple regression (P MR) and the independent contribution for each variable as provided by hierarchical partitioning results (Ind.).

R2adj / P / Lowest AIC / ΔAIC / Variables / Est. MR / P MR / Ind.
Complete dataset / 0.2679 / <0.0001 / -1920.02 / 1.99 / Lat / -0.4053 / 0.1363 / 0.031
Long / 0.0896 / 0.0677 / 0.014
Perim / 0.1628 / <0.0001 / 0.095
Temp / 0.5819 / <0.0001 / 0.060
Prec / -0.7413 / <0.0001 / 0.068
Isol / -0.0960 / 0.0027 / 0.006
LG1 / 0.3248 / <0.0001 / -530.8 / 5.37 / Lat / -5.0253 / <0.0001 / 0.095
Alt / -0.1800 / <0.0001 / 0.049
Perim / 0.0412 / 0.144 / 0.007
Prec / -0.9181 / <0.0001 / 0.187
LG2 / 0.5448 / <0.0001 / -811.3 / 0 / Lat / -6.5865 / <0.0001 / 0.155
Long / 0.2889 / <0.0001 / 0.058
Alt / -0.1060 / 0.0116 / 0.014
Perim / 0.1868 / <0.0001 / 0.090
Temp / 0.4459 / 0.1263 / 0.065
Prec / -0.6920 / 0.0017 / 0.143
Isol / -0.1487 / 0.0081 / 0.029
LG3 / 0.5852 / <0.0001 / -161.58 / 4.7 / Lat / 9.8998 / 0.0183 / 0.043
Perim / 0.0751 / 0.1362 / 0.132
Temp / 1.2999 / <0.0001 / 0.328
Prec / -0.9282 / 0.0958 / 0.111
LG4 / 0.3637 / <0.0001 / -752.81 / 5.92 / Lat / 2.7703 / <0.0001 / 0.065
Perim / 0.2837 / <0.0001 / 0.303

Table A3 Predictor variables affecting gastropod species richness of European lakes as proposed by stepwise multiple regressions (MR) for Set 2. Shown are the adjusted R2 and P-values (P) of the MRs (R2adj), Akaike’s information criterion (AIC), the difference between initial and final Akaike’s information criterion (ΔAIC), the estimate of the multiple regression (Est. MR), significance of the variables in the multiple regression (P MR) and the independent contribution for each variable as provided by hierarchical partitioning results (Ind.).

R2adj / P / Lowest AIC / ΔAIC / Variables / Est. MR / P MR / Ind.
Complete dataset / 0.2789 / <0.0001 / -1954.63 / 3.15 / Long / 0.0725 / 0.129 / 0.014
Area / 0.0994 / <0.0001 / 0.105
Temp / 0.6171 / <0.0001 / 0.083
Prec / -0.7854 / <0.0001 / 0.076
Isol / -0.1072 / 0.0004 / 0.006
LG1 / 0.3236 / <0.0001 / -530.46 / 5.35 / Lat / -4.9786 / <0.0001 / 0.095
Alt / -0.1790 / <0.0001 / 0.049
Area / 0.0245 / 0.141 / 0.006
Prec / -0.9223 / <0.0001 / 0.187
LG2 / 0.5505 / <0.0001 / -817.33 / 0.22 / Lat / -8.3529 / <0.0001 / 0.231
Long / 0.3035 / <0.0001 / 0.052
Alt / -0.1302 / 0.0004 / 0.017
Area / 0.1146 / <0.0001 / 0.093
Prec / -0.4331 / 0.0109 / 0.134
Isol / -1656 / 0.0030 / 0.032
LG3 / 0.5765 / <0.0001 / -161.65 / 4.67 / Lat / 10.7108 / 0.0112 / 0.05
Temp / 1.4717 / <0.0001 / 0.410
Prec / -1.0605 / 0.0572 / 0.139
LG4 / 0.3856 / -785.18 / 7.16 / Lat / 3.4185 / <0.0001 / 0.085
Area / 0.1503 / <0.0001 / 0.297
Temp / 0.2855 / 0.153 / 0.010

Lat, latitude (decimal degrees), Long, longitude (decimal degrees), Alt, altitude (metres), Area, lake surface area (km2), Temp, mean annual temperature (°C), Prec, annual precipitation (mm) and Isol, Isolation (km).

Table A4 Pairwise beta diversity of the lake groups (LGs) 1–4 computed with the Jaccard dissimilarity measure (Baselga, 2012) when single lake endemics are excluded. βjac is the overall beta diversity, βjtu and βjne are the dissimilarity components referring to species replacement and species loss, respectively.

Groups / βjac / βjtu / βjne
LG1–LG2 / 0.41 / 0 / 0.41
LG1–LG3 / 0.44 / 0.31 / 0.13
LG1–LG4 / 0.62 / 0.23 / 0.39
LG2–LG3 / 0.39 / 0.22 / 0.17
LG2–LG4 / 0.46 / 0.33 / 0.13
LG3–LG4 / 0.50 / 0.18 / 0.32

Figure A1 Moran’s I correlograms for the residual of multiple regression models for the complete dataset and lake groups (LGs) 1–4 for Set 1. Equal number of pairs in default distance classes is selected. Geographic distances are measured in kilometres.

Figure A2 Moran’s I correlograms for the residual of multiple regression models for the complete dataset and lake groups (LGs) 1–4 for Set 2. Equal number of pairs in default distance classes is selected. Geographic distances are measured in kilometres.