1. Introduction
Climate change/global warming has significantly impacted the distribution of various ecosystems, and the effects of future climate change will likely modify the habitat, scope, and distribution of myriads of species [
1,
2,
3]. According to the Fifth Assessment Report (AR5) of the Intergovernmental Panel on Climate Change (IPCC), the average global surface temperature is anticipated to rise 0.3–4.8 °C by the end of this century [
4,
5]. Global warming has triggered an array serious of environmental issues, such as changes in the spatial patterns of species, which threatens species diversity and sustainable development [
6]. A continually warming climate might lead to the extinction of geographically limited species and endemic species that are unable to adapt to unusual climatic conditions [
7]. Increases in drought, heat stress, and extreme weather events may produce an unprecedented series of pressures for species [
8,
9]. As endangered plants have very specific habitat requirements [
10], the determination of their current geographical distributions, population status, and identification of specific threats and extinction risks are key for their conservation. Therefore, an elucidation of the distribution dynamics of endangered plant species under climate change is beneficial toward the formulation of robust protection strategies.
The evolutionary trajectories, ecological habits, and terrestrial distribution of plant species are affected, restricted, and driven by climate change and human activities, which are also the key to the formation of ecosystem biodiversity [
11,
12,
13,
14]. The Last Glacial Maximum (LGM) refers to the most recent period in Earth’s history when the glaciers were at their thickest and sea levels at their lowest, roughly between 24,000–18,000 years ago [
15]. Due to rapid climatic deterioration, some species became extinct, while the distribution areas of most surviving species also dropped sharply [
16,
17,
18], as many surviving species migrated to glacier shelters [
17,
19]. With post-glacial warming (Holocene), surviving plant populations began to expand from these shelters to new suitable habitats.
Ecological niche models (ENMs) (also known as species distribution models), may be employed to predict the spatial distributions of target species, assess the potential responses of organisms to climate change, and determine species niches based on their environmental conditions [
20,
21]. The ENMs can forecast the potential terrestrial distribution of species in selected landscapes through pattern matching, or through effective statistical linking where species exist within certain environmental variables [
22]. Its predictive results are stable and reliable, and the estimation accuracy of the distribution of endangered species is high, even for small sample sizes [
23]. MaxEnt software analyzes climate data to estimate past and future species distribution for potentially suitable growth areas to determine sites for relatively stable species habitats, as well as their migration and dispersal pathways [
14]. MaxEnt has been successfully employed for nature reserve design, endangered species surveys, infectious diseases, exotic species risk assessments, and climate change impact studies on plant habitats, for example,
Semiliquidambar cathayensis [
14],
Bacillus anthracis [
22],
Ageratina adenophora [
24], and
Ziziphus spinosa [
25].
Cremastra appendiculata (D. Don) Makino (Orchidaceae) is a perennial herb with underground pseudobulbs that are utilized as medicine, with narrow oval and long-stalked single leaves that emanate from the tops of the pseudobulbs. According to traditional Chinese medicine,
C. appendiculata has the effect of clearing heat and detoxifying the body, moistening the lungs, and relieving cough, and activating blood circulation to relieve pain. It can be used externally to treat snake and insect bites and skin burns and taken internally to fight liver cancer and breast cancer.
Cremastra appendiculata (D. Don) Makino typically grows in forest or ravine wetlands at altitudes of from 500–2900 m [
26]. In China, it is distributed across the Gansu, Shaanxi, the Yangtze River Basin, as well as in Southwest and Southern China.
Cremastra appendiculata grows well in humus soil, is a cross-pollination plant, and due to the unique structure of its flowers, the fruiting rate is only 1.3%–2% under natural conditions, as it must be fertilized with the help of insects [
27]. It has been listed as a Grade-II state-protected plant by the central government. In addition to its excellent ornamental value,
C. appendiculata is one of the sources of the traditional Chinese medicine “Shancigu”. Its pseudobulbs contain chemical compounds such as bibenzyl, phenanthrenes, and alkaloids [
26], and studies have confirmed that the pseudobulbs have significant positive effects for the treatment of cancer. Thus, wild sources of
C. appendiculata have been subject to predatory harvesting, coupled with the restriction of its own reproductive mechanisms and the destruction of its shaded habitats. Its habitat area has gradually dwindled, which has resulted in a sharp drop in the number of scattered and wild distributions, leading it to the verge of extinction. Thus,
C. appendiculata has been added to the National Key Protected Wild Plants list, which means that it is in urgent need of protection. Therefore, accurate estimates of the distribution of potentially suitable habitats for
C. appendiculata in the context of climate change are of great significance for the conservation and sustainable use of its resources.
This study employed an optimized maximum entropy model to analyze the distribution patterns and changes in potentially suitable growth regions for C. appendiculata in the LGM, MH, modern times (1970–2000), under multiple shared socioeconomic pathways (SSP1-2.6, SSP2-4.5, SSP3-7.0, and SSP5-8.5) over different periods (2050s, 2070s, and 2090s). The objectives of this research work were to: (1) predict the potential geographic distribution of C. appendiculata under different climate scenarios, (2) understand the relevant environmental factors that affect the geographical distribution of C. appendiculata, (3) predict the relatively stable habitats of C. appendiculata under climate change, and (4) provide a theoretical basis for the conservation of germplasm resources and the delineation of suitable artificial cultivation areas for C. appendiculata.
3. Results
3.1. Model Parameter Optimization and Accuracy Analysis
In this study, seven environmental variables were screened to construct a predictive model. These included the mean diurnal range (max temp-min temp) (BIO02), temperature seasonality (BIO04), mean temperature of wettest quarter (BIO08), mean temperature of coldest quarter (BIO11), precipitation of driest month (BIO14), precipitation seasonality (BIO15), and precipitation of warmest quarter (BIO18). The contribution percentages of these variables for model construction were BIO02 (51.1%) > BIO04 (38.5%) > BIO08 (3.8%) > BIO11 (2.3%) > BIO14 (2.3%) > BIO15 (1.2%) > BIO18 (0.8%) (
Table 1;
Figures S1 and S2).
Based on 108 distribution points for
C. appendiculata and seven climatic variables, the MaxEnt model was used to predict the distribution of potentially suitable habitats for
C. appendiculata in China. According to the model optimization results, the FC was QP, the RM was 0.3, the model omission rate was 0.0303, and the delta AICc value was 0. The mean value of the training AUC (AUC
TRAIN) was 0.9589 ± 0.0023, the mean value of the test AUC (AUC
TEST) was 0.9539 ± 0.0070, and the absolute value of the difference between the training AUC and the test AUC (|AUC DIFF|) was 0.005, which indicated that the model had an excellent prediction (
Figure S3).
Based on the MTSPS threshold (0.1349), the spatial units for this study were divided as follows: 0–0.1349 unsuitable; 0.1349–0.4233 low suitability; 0.4233–0.7116 moderate suitability; 0.7116–1 high suitability.
3.2. Current Potentially Suitable Regions
According to the results predicted by the model, the total present potentially suitable growth area for
C. appendiculata was 213.9 × 10
4 km
2, which was mainly distributed across Sichuan, Guizhou, Chongqing, Anhui, Hubei, Taiwan, Zhejiang, Tibet, and Yunnan Provinces (
Figure 2). The high, moderate, and low potential suitable areas were 8.44 × 10
4 km
2, 57.71 × 10
4 km
2, and 147.76 × 10
4 km
2, respectively. The highly suitable regions were mainly distributed across southern Anhui and Shaanxi, western Hubei, central and northern Guizhou, central Taiwan, and a few areas in Sichuan.
3.3. Potential Past Suitable Regions
Our models predicted that the highly suitable growth regions for C. appendiculata decreased significantly during the LGM compared to modern times. The area of the highly suitable area was only 0.28 × 104 km2, accounting for 3.26% of the present highly suitable area, which included only central Chongqing, eastern Sichuan, and a very small portion of southern Anhui. In the MH, the highly suitable area for C. appendiculata was significantly increased compared with LGM, which was 8.38 × 104 km2, accounting for 99.30% of the modern highly suitable area. The northern part of Taiwan transitioned from the moderately suitable regions in the MH period to the modern highly suitable regions. Compared with the MH, the highly suitable regions in southern Anhui also increased, and the low suitable regions in Guangxi decreased slightly in modern times.
3.4. Potentially Suitable Areas in the Future
Overall, the potential
C. appendiculata habitat area is estimated to decrease to varying degrees over the next three eras. Except for the 2090s, the potential habitat area of
C. appendiculata was predicted to show a decreasing trend with intensifying climate severity (SSP2-4.5 → SSP5-8.5). (
Figure 3,
Figure 4 and
Figure 5,
Figures S4 and S5;
Table 2).
Under the SSP2-4.5 scenario, the total area of potential suitable habitats for C. appendiculata revealed a shrinking trend (67.98%, 58.59%, and 52.66% of corresponding present values). From the 2050s to the 2090s, southeastern Shandong and eastern Jiangsu transitioned from originally low suitable habitats to unsuitable habitats for C. appendiculata growth. The highly suitable areas showed an obvious trend of contraction, accounting for 24.99% (2050s), 15.36% (2070s), and 9.47% (2090s) of the corresponding contemporary values, respectively. The highly suitable areas in northern Guizhou, central Chongqing, and southern Shaanxi gradually shifted to moderate or low suitable areas.
Under the SSP5-8.5 scenario, the total potential habitat area of C. appendiculata exhibited a trend of initially shrinking and then expanding over time. The percentages of contemporary corresponding values were 59.91% (2050s), 45.88% (2070s), and 85.86% (2090s). From the 2050s to 2070s, the coastal areas of Shandong and Jiangsu, eastern Hubei, and northeastern Fujian gradually shifted from low suitable habitats to non-suitable habitats. From the 2070s to 2090s, the low suitable habitats gradually recovered, and the area of high suitable habitats increased from 0.67 × 104 km2 (2070′s) to 4.26 × 104 km2 (2090s), which were still primarily distributed across Shaanxi, Chongqing, Hubei, Anhui, and Taiwan Province.
3.5. Relatively Stable Habitat
Relatively stable suitable habitats refer to areas where species are relatively less affected by climate change, where the predicted results under different climatic scenarios are different (
Figure 6 and
Figure S1;
Table 3). With increasing climate severity (SSP1-2.6 → SSP3-7.0), the area of relatively stable suitable habitats for
C. appendiculata decreased (130.01 × 10
4 km
2 → 82.95 × 10
4 km
2), which accounted for 60.78%, 47.75%, and 38.78% of the total area of the present potential suitable habitats, respectively, eventually stabilizing at SSP5-8.5. Under the SSP5-8.5 scenario, the total area of relatively stable
C. appendiculata habitats was 84.40 × 10
4 km
2, which accounted for 39.46% of the total area of the present potential habitat. Furthermore, the predictions also indicated that central Sichuan, the entirety of Guizhou, southern Chongqing, western Hubei, northwestern Hunan, and southern Anhui were relatively stable suitable habitats for the growth of
C. appendiculata under any climate scenario.
3.6. Shifts in the Distribution Center of Suitable Habitats
From the results of the model simulations, the centroids of the potential habitats of
C. appendiculata showed a tendency to shift to the northwest under the SSP2-4.5 and SSP3-7.0 scenarios, while the centroids of the habitats initially moved to the northwest and then to the southeast under the SSP1-2.5 and SSP5-8.5 scenarios (
Figure 7). Under the SSP2-4.5 scenario, the centroids of the suitable
C. appendiculata habitats shifted from Yongding District, Zhangjiajie City, and Hunan Province to Lichuan City, Enshi Tujia, and Miao Autonomous Prefecture, Hubei Province (2050s), Fengdu County, Chongqing City (2070s), and Dianjiang County, Chongqing City (2090s) over time, with migration distances of 174.74 km, 66.51 km, and 61.49 km, respectively. Under the SSP5-8.5 climate scenario, the potential suitable habitats centroid shifted by 231.37 km (2050s), 192.64 km (2070s) to the northwest, and finally 337.55 km (2090s) to the southeast, from the present Zhangjiajie City to the 2090s in Xuanen County, Enshi Tujia, and Miao Autonomous Prefecture.
According to elevation changes, the elevation of the center of the suitable C. appendiculata habitats decreased from 484 m in the LGM to 285 m currently. Under the SSP1-2.6 scenario, the suitable C. appendiculata habitats shifted to higher elevations, while under the SSP2-4.5 and SSP3-7.0 climate scenarios, the elevation of suitable C. appendiculata habitats increased and then decreased. However, under the SSP5-8.5 scenario, the elevation of the center of suitable C. appendiculata habitats increased to 766 m in the 2050s, decreased to 330 m in the 2070s, and finally increased to 691m in the 2090s.
5. Conclusions
For this study, an optimized maximum entropy model was used to estimate the distribution of potential suitable habitats for C. appendiculata in China under different climate scenarios in the past (LGM and MH), modern times, and future eras. The results revealed that the suitable habitats area of C. appendiculata in the past was comparable to the total area of current potential suitable habitats. Among them, the highly suitable areas were primarily distributed across Shaanxi, Chongqing, Guizhou, Hubei, Anhui, and Taiwan Province. In the future, except for SSP5-8.5 2090s, the area of potentially suitable C. appendiculata regions will be reduced to varying degrees. Meanwhile, this study found that the relatively low impact areas of C. appendiculata were mainly distributed across Sichuan, Guizhou, Chongqing, Hubei, and Anhui Provinces. Furthermore, the central point of the potential suitable habitats of C. appendiculata moved to the northwest under the SSP2-4.5 and SSP3-7.0 scenarios, while the central point of the suitable habitats initially transitioned to the northwest and then moved to the southeast under the SSP1-2.5 and SSP5-8.5 scenarios. These research results will provide some theoretical references for the protection of wild resources and artificial cultivation of the endangered medicinal plant C. appendiculata.