Discussion
As documented in Guisan et al. (2014) and Sexton et al. (2017), recent applications of niche models (based on ENMs and PCA-env) analyze niche shifts have provided new insights to the roles of abiotic climate and geographical conditions in shaping range limits. Moreover, intraspecific variation along allopatric genetic clades in the potential for realized niche shifts has important implications not only for forecasting biological invasion, but for studying the process of origin and evolution across clades within species level.
Our results agree with previous progressive uplifts of Tibet (Mulch et al. 2006). Molecular datings of E. D, W. a and W. b are almost identical to the fourth uplift of high mountain ranges and aridification of Central Asia (ca. 20 to 10 Ma), and the divergence times of clades E. A, E. B and E. C are coincident with the final extension of the last uplift (ca. 10 Ma to present) predicted by previous studies (Favre et al. 2015). However, future researches with more detailed complementary analyses are needed to understand a link among clades molecular dating, gene flow routes and past geological and climatic changes across these regions.
Our results show PCA-env based approach well supports E. A, E. B and partial of E.C clades with significant divergence. However, ENMs based method just supports divergence of E. B in bi-directions, while the divergence of E. A exhibits one-sided significance, the opposite comparisons not deviated from null expectation. While inconsistent conclusions drawn in divergence of E. C based two methods. In literatures, inconsistent conclusions regarding niche conservatism or divergence have been frequently reported based on different approaches. For example, Guo et al. (2013) applied the ordination and ENMs for the globally introduced Phragmites australis and found inconsistent results. Our results based on two approaches draw almost the same conclusion in one-sided tests, but PCA-env based approach is sensitive to niche divergence, while ENMs is more inclined to niche conservatism in bi-directions. We analyze reasons leading to these differences further and find the native-range model, such as ENMs, always under-predicts the species’ invaded range in exactly the types of climates in which there was evidence of niche expansion as shown in previous studies. Strubbe et al. (2013) showed that the predictive performance of native range ENMs increased with increasing niche overlap and decreased with increasing niche change. Similarly, Tingley et al. (2014) found that a native-range ENMs under-predicted the extent of the species’ Australian invasion.
In addition, we found E. A, E. B and E. C clades within S. boulengeri have an obviously expansion of climatic niche (Figure 3h and Supplementary material Appendix Figure A5). Intriguingly, they shift beyond the realized niche with the new conditions but still overlap the fundamental niche, which provides positive proof for the niche conservatism hypothesis (Wiens 2004, Kozak and Wiens 2006) and niche divergence hypothesis (Graham et al. 2004, Evans et al. 2009) are not contradictory. We found a gradient of realized niche change in the invaded ranges across clades within S. boulengeri : niche stasis on E. D (96.3%) and W. a (96.6%), niche unfilling in W. b (75.2%), and niche expansion (vs. E. D and W. a separately) in E. A (mean = 99.3%), E. B (mean = 97.1%) and E. C (mean = 90%). Our results seem to be inconsistent with conclusion of previous studies in Petitpierre et al. (2012), in which realized niche shifts between the native and invaded ranges were largely due to niche unfilling. Our results are also different from the results of cane toad in Tingley et al. (2014): the shift in the realized niche of the cane toad Rhinella marina was solely due to niche expansion. In our results, niche expansion into novel environments is more popular than niche unfilling, suggesting that our niche divergence due to niche expansion in the shifted range and thus represents true niche changes. Why do E. D and W. a fail to fill its fundamental niche in its native range? One possibility is that the presence of closely related species (S. glandulatus and/or S. mammatus ) may prevent S. boulengeri from colonizing suitable environments south of its present range. Indeed, previous study found there were low rates of interspecific hybridization (Chen et al. 2009). We cannot exclude dispersal limitation in the native range as a possible contributing factor, such as Jinsha River and Yalong River (Li et al. 2009), which also can enforce stable parapatric range boundaries. Future studies will be able to test this hypothesis using laboratory or field experiments.
Numerous examples of rapid adaptation in non-native niche suggest that rapid evolution may be common during invasions in species level (Alström et al. 2015, Bartels et al. 2012, Sherratt et al. 2017). The degree to which species adapt to novel environments is important to a range of topics in ecology and evolution (Wiens et al. 2008), but is of special concern for the study of intraspecific niche evolution (Tingley et al. 2016). In our study, niche divergence caused by niche expansion indeed accompanied key morphological innovations of preadaption in novel climates vs. niche unfilling and stability. Our finding of significant phylogenetic signal in SVL (Table 5) and Elevation (Table 4 and Figure 2c), indicates that these acquired data are not random and our results are robust. Furthermore, our findings of significant phylogenetic signal in these traits are consistent with previous studies (Freckleton et al. 2002, Blomberg et al. 2003, Oufiero et al. 2011). We found that Elevation (AIC = 21.3; P = 0.002), Isothermality (AIC = 24.47; P = 0.007), Mean diurnal range (AIC = 29.33; P = 0.037) and Max temperature of warmest month (AIC = 29.31; P = 0.037) are significantly negative predictors of SVL under phylogenetic models, which suggest S. boulengeri toads from warmer and more arid environments tend to be larger, which is in concert with true records in our field work.
Several factors may underlie the observed pattern of SVL variation inS. boulengeri clades. One possibility pertains to the expected relationship between fasting endurance and SVL (Mautz 1982). The second possibility is ecological release in novel shifted areas may allow for larger SVL (Losos and Queiroz 1997, Yoder et al. 2010). The third is likely that maintenance of preferred body temperature influences the evolution of SVL (Oufiero et al. 2011). Our results highlight reduced competitors (ecological release) in newly shifted niche may be the most likely reason for enlarged SVL in E. A, E. B and E. C clades.
Moreover, we found species tolerance of newly shifted niche tends to have morphological attributes important for locomotor performance. These traits may be a key preadaptation in toads that helps overcoming the challenge of insufficient precipitation or high temperature in novel habitats, which are in accord with the evolutionary shifts mechanistic model highlighted by prior studies (Tingley et al. 2014, Phillips et al. 2010, Kolbe et al. 2010). In our study, LAHL (mean value: 27.19 mm in clade E. A, 26.27 mm in clade E. B and 23.95mm in E. C, 21.77–23.57 mm in the remained clades, Supplementary material Appendix Table A1), HLL and FL have the same trend. Interestingly, these character values, without size-correction, have a high phylogenetic signal, best fitted BM model. However, once size-corrected, these character values will have a completely different scenario — with a low phylogenetic signal but trait correlations still exist, there seems to be a trade-off strategy by locomotor performance combined enlarged SVL for speed and endurance in thermal reaction norms (Angilletta et al. 2003). Collectively, we found phenotypic plasticity (i.e LAHL, HLL and FL) and evolution changes (i.e. SVL) may together contribute to niche expansion towards adapting novel niche. Indeed, because the proximate mechanisms that underlie variation between body length and locomotor performance can be complex, quantifying the fitness consequences of the resulting trade-offs will be challenging, novel analytical tools and optimization models are needed in further studies.