Within-population beta diversity
In general, nMDS based on Bray-Curtis and Euclidean distances revealed similar clustering patterns. The nMDS plots including pre- and post-disturbance data reflected a strong clustering by time (Figure 4A, B, E, F). However, when corrected for the temporal effect, there were also clear effects of temperature (Figure 4C, G) and range (Figure 4D, H). These observations were also supported by the mGLM fitted on post-disturbance time points. Community composition differed with time, between temperatures, between ranges (p = 0.002), and also the interaction between temperature and range was significant (p = 0.012, Table S2c).
Beta diversity within-populations (based on Bray-Curtis and Euclidean distances) increased from field to the climate room (pBray-Curtis = 0.024, pEuclidean = 0.004) but did not differ between ranges (p ≥ 0.525) and declined sharply following the disturbance treatment (p < 0.001, Figure 3K, P). After the disturbance, within-population beta diversity varied with time (p < 0.001) and followed an increasing trend (Figure 3L, M, Q, R) that was best fitted with a polynomial function (Table S3). Distances did not vary significantly between ranges (p ≥ 0.137). The main effect of temperature was highly significant for both distance measures, with substantially higher within-population beta diversity at 22 °C, compared to 15 °C (p < 0.001, Figure 3N, S). In addition, significant p-values were found for the interactions between range and time(pBray-Curtis = 0.015, pEuclidean < 0.001), temperature and time (p < 0.001), and range and temperature (p ≤ 0.001). Post-hoc comparisons indicated that the significant interaction between range and temperature was explained by higher beta diversity at 22 °C in native populations compared to non-natives °C(pBray-Curtis = 0.004, pEuclidean < 0.001), whereas natives and non-natives did not differ in terms of within-population beta diversity at 15 °C (Figure 3O, T, see Table S2d for all statistical output).