4 DISCUSSION

4.1 The potential geographical distribution of species

Habitat is a vital place for survival, reproduction and population development of species. Its quality can directly influence the distribution, quantity and survival rate of species (Hall et al., 1997; Zhang et al., 2019). We utilized the MaxEnt model to predict the potential geographical distribution of 25 rare, endangered and/or national protected plant species in Northwest Yunnan. The results showed that the potential distribution area (i.e., Region where HSI ≥ 0.4) of each species was between 826.33 km2 and 44,963.53 km2, which indicated that these species had obvious differences in their adaptability to environmental factors such as the topography and climate in Northwest Yunnan. The potential distribution area of these target species were mainly concentrated in the mid-western, mid-eastern and northern parts of the study area, which was in good agreement with the prediction results of the potential distribution area which studied by Zhuang et al. (2018a). The prediction results of this study will provide possible new areas for species distribution and field investigation. In addition, the potential distribution area of species overlapped greatly in the mid-western and northern parts of Northwest Yunnan. We roughly divided them into two parts. One part was close to the Gaoligong Mountain and Baima Snow Mountain NNRs, which were almost consistent with the hotspot distribution of key higher plant species in Northwest Yunnan (Ye et al., 2020a). Similarly, the other part was located in Shangri-La County, which might be related to the fact that Shangri-La is situated at the core area of the Three Parallel Rivers World Natural Heritage Site (Yang et al., 2013), with abundant landscape types, vegetation types, and ecosystem types. Consequently, the results reflected that these regions mentioned above will play a crucial role in biodiversity conservation in the future.

4.2 A model used to predict the distribution of species

The model prediction method can help to compensate for the difficulty in field investigation. The vertical peaks and horizontal valleys in Northwest Yunnan greatly limit the accessibility of field surveys. Hence, by combining the model prediction with field investigation, the field survey can be carried out in the order of habitat suitability, with priority given to the field investigation in the optimal suitability distribution area and the potential distribution area that has not been studied. In addition, the completeness, accuracy and reliability of species geographical distribution data are important and key links of division of key biodiversity areas. Meanwhile, the accuracy of the data and the precision of the model prediction are also mutually promoted (Zhang et al., 2016; Wu et al., 2016; Zhuang et al., 2018a). In this study, we tried our best to collect a lot of species distribution data through multiple approaches, but there are still some species with relatively less distribution data. For example, Coptis teeta is also distributed in Myanmar and other countries. Therefore, more species distribution data must be collected to construct a more accurate species distribution model. In addition, the distribution of species is not only determined by topography, climate and edaphic factors, but also influenced by social and economic structure, land use type, human disturbance and other social factors. In some cases, due to the influence of the local microenvironment, the areas predicted to be of lower suitability levels are actually the distribution areas of species. Therefore, the results obtained in this study are only advisory. A more precise and accurate potential distribution prediction needs to be supported by more comprehensive social and environmental factors and more precise and more reliable species distribution information.

4.3 Environmental explanations for the potential distribution of species

From an ecological point of view, environmental factors can affect the spatial distribution of species, as well as their habitat suitability (Zhang et al., 2019). In this study, we found that the cumulative contribution rate of Bio12 (25.92%), Bio19 (15.86%) and Pop (17.95%) to the MaxEnt model prediction results reached 59.73%. Furthermore, the contribution rate of Bio12 was the highest, which was consistent with the results obtained by Zhuang et al. (2018). In addition, the R 2 values of the GWR model were all above 0.80, indicated that the model has a reliable goodness of fit for explaining the potential distribution of species. Moreover, the water model (R 2 = 0.88, AIC = 7703.82) showed a higher explanation rate compared to other single models, followed by the temperature model (R 2 = 0.88, AIC = 7939.56) and topographical model (R 2 = 0.88, AIC = 7900.40), which were second only to the water model in their interpretation effect on the potential distribution of species. Accordingly, this result suggested that climate (temperature and water factor) and habitat heterogeneity (topographical factor) could play an important role in the prediction of potential distribution areas of species, which was coincide with the research results of Ștefănescu et al. (2017). However, compared with all single models, the comprehensive model (R 2 = 0.92, AIC = 6785.26) that combines all environmental factors had a better goodness of fit, which indicated that the potential distribution of species was the result of the combined effects of various environmental factors (Wang et al., 2018). The main thing worth mentioning is that, rare and endangered species are usually restricted to specialized edaphic or topographic or other environmental conditions which occupy a quite fraction of their climatically suitable range. Therefore, on the basis of this study, it is necessary to study the effects of other relevant environmental factors on the potential distribution of species.

4.4 Suggestions for the protection of rare and endangered plant species

In recent decades, due to the disturbance of human activities and the impacts of the external natural environment, especially climate change, the population size and distribution area of some rare, endangered and/or threatened species had been declining (Yu et al., 2014). Therefore, understanding the habitat suitability of species and its influencing factors is the basis of protecting rare and endangered plant species (Zhang et al., 2019). Habitat suitability plays an important role in the survival and development of species. Hence, habitat suitability assessment is the first step of effective conservation and scientific management of species, and can provide scientific basis for relevant departments to formulate valid conservation strategies. It has become a resultful method to protect rare and endangered species by scientifically predicting the potential distribution areas and habitat suitability levels of species and planning wild nature reserve in the best suitable areas (Xiao et al., 2011; Xu et al., 2015). The results of this study revealed that the habitat suitability of species near Gaoligong Mountain and Baima Snow Mountain NNRs and Shangri-La Country was relatively high, and these areas were also the core regions for the distribution of rare, endangered and/or national protected species in Northwest Yunnan. For this reason, it is suggested that to strengthen the conservation of these areas. In addition, Shangri-La Country had more habitat suitability distribution areas for species, but there were only three PNRs with a small area, namely, Napahai, Bitahai and Haba Snow Mountain PNRs (Ye et al., 2020a). It is recommended that appropriate expansion of the nature reserve should be carried out, with measures combining in-situ and ex-situ conservation to strengthen the protection of species in a state of isolation and/or fragmentation. In addition, the species distribution model is an estimate of the potential distribution of species, and its essence is prediction research. Therefore, the results of the model simulation cannot be used as the only basis for formulating strategies of species conservation, and corresponding field surveys should be carried out according to the simulation prediction results, so as to formulate more scientific and reasonable strategies for species conservation (Renner & Warton, 2013; Wen et al., 2019). Correspondingly, by simulating the potential distribution area of species, it can also provide some basic and scientific evidence for species reintroduction.