Introduction

Rubus idaeus is a prominent species in raspberry cultivation, which belongs to the Sect. Idaeobatus group of the Rubus genus and is grown extensively worldwide, particularly in the Americas and Europe1. Raspberries have boast appealing colors, distinct flavor, and notable health benefits1,2. They were consumed as fresh fruit, and seed and leaf hold abundant economic and medicinal value. Over 100 raspberry-derived products have been developed and utilized across pharmaceutical, cosmetics, agriculture and food industries3,4,5. Raspberry cultivation in China spanned over a century, initially introduced into Heilongjiang Province via Russia. Subsequently, numerous varieties from Europe and North America were introduced6. Up to now, raspberry has been cultivated with abroad varieties ‘Heritage’, ‘Fertodi Zamatos’ and ‘Tulameen’ which are the main varieties in Xinjiang, Yunnan, Henan, Heilongjiang, Liaoning and Hebei in China. However, prolonged domestication has led to a notable decline in both morphological and genetic diversity of R. idaeus7,8. Exploring and harnessing native wild resources for breeding purposes can significantly alleviate the time and effort required for introducing and domesticating foreign varieties, while also addressing the issue of diminished genetic diversity7,9. According to the Flora of China, R. idaeus exhibits a broad natural distribution in the northern regions of China. Among them, R. idaeus var. idaeus is naturally distributed in Jilin, Liaoning, Hebei, Shanxi, and Xinjiang. R. idaeus var. borealisinensis, a transitional form between R. idaeus and R. sachalinensis, is found in Inner Mongolia, Hebei, and Shanxi provinces, while R. idaeus var. glabratus is found only in Heilongjiang. The natural regeneration of R. idaeus include sexual and asexual reproduction methods, mainly asexual, which further limiting the genetic diversity of natural R. idaeus. The growth and development of R. idaeus was affected strongly by the environment. The seeds of R. idaeus have both physical and physiological dormancy10,11. The response of dormancy and germination to temperature is closely related to the ecological distribution of the species12. Thus, in view of clarifying resources for breeding, it’s necessary to conduct precise distribution surveys for wild R. idaeus resources, for securing long term sustainability and breeding initiatives for the species.

Global climate change exacerbates the frequency and severity of short-term droughts worldwide and potentially will alter plant distribution patterns and jeopardizing biodiversity13. Drought, recognized as one of the most constraining factors, poses a significant threat to both food security and ecological stability, and affects the sustainable advancement of socio-economic sectors14,15. Drought, as an environmental stressor, diminishes plant yields by compromising their eco-physiological performance, and affects growth and development16. While agronomic techniques can mitigate some adverse effects on cultivated plants, drought can precipitate a substantial reduction or even extinction of species in wild populations. R. idaeus harbors substantial developmental promise and boasts widespread distribution in China17. Nevertheless, pivotal questions linger: What is the present status of wild resources? How does climate change impact these wild populations? Are there unexplored areas conducive to R. idaeus that remain untapped? Consequently, predicting R. idaeus suitable habitats under climate change scenarios will furnish vital theoretical guidance for the judicious exploration and exploitation of R. idaeus germplasm resources in the future.

Species Distribution Models (SDMs) are pivotal tools for estimating the potential distribution of species, relying on the ecological niche concept18, Common models include the generalized linear model (GLM), bioclimate analysis and prediction system (BIOCLM), ecological niche factor analysis (ENFA), and maximum entropy (MaxEnt)19. Among these, the MaxEnt model stands out as the most favored and extensively utilized, owing to its robustness in predicting species’ spatial distribution, with minimal influence from sample size20,21,22. In this study, we conducted a comprehensive nationwide survey of R. idaeus in China. We gathered data on 19 climate factors and 3 topographic factors to predict the suitable areas for R. idaeus under future climate scenarios using the MaxEnt model. Results provide valuable data sources for germplasm resource utilization and breeding of R. idaeus, and a reliable theoretical framework for its widespread adoption, conservation and protection.

Materials and methods

Sample sites and survey areas of R. idaeus

Wild R. idaeus sampling areas were obtained through the Chinese Virtual Herbarium (https://www.cvh.ac.cn/), the Global Biodiversity Information Network (https://www.gbif.org/), the Flora of China and the floras of Xinjiang, Inner Mongolia, Shanxi, Hebei, Beijing, Henan, Jilin, Heilongjiang, etc. Then, the sample data were screened, and the uncertain, duplicate, misidentified and artificially cultivated samples were eliminated. A total of 352 sample sites were obtained. Then, ENMTools were used to deleted the duplicate data, and the only one specimen record datum for the 2.50′ × 2.50′ grid of the distribution area was selected20. Finally, a total of 90 sample sites of R. idaeus were determined for MaxEnt model construction (Fig. 1).Based on above sample data, 19 survey areas of R. idaeus were selected for further exploitation (Fig. 1). The survey areas were in Xinjiang, Inner Mongolia, Shaanxi, Shanxi, Henan, Hebei, Beijing, Jilin, and Heilongjiang 9 provinces (autonomous regions or municipalities).

Fig. 1
figure 1

Survey areas and sample sites of R. idaeus using Arcmap 10.8.0 software (URL: https:// www. arcgis. com/ index. html).

Full size image

Environmental variables

Nineteen bioclimatic variables and three terrain variables were used for MaxEnt model construction (Table S1). The 19 bioclimatic variables were obtained from WorldClim website (https://worldclim.org/). The future climate data uses Coupled Model Intercomparison Project Phase 6 (CMIP6) and the Global Climate Models (GCM) BCC-CSM2-MR. Four Shared Socio-economic Pathways (SSPs) (SSP1-2.6, SSP2-4.5, SSP3-7.0, SSP5-8.5) in 2050s (2041 ~ 2060) and 2090s (2081 ~ 2100) were selected. 3 terrain variables (DEM, Aspect, Slope) were obtained according to Li et al20. All the spatial resolution was unified to 2.50′. The global national administrative unit boundary data were downloaded from Resource and Environment Science and Data Center (https://www.resdc.cn/), Chinese academy of science.

MaxEnt model construction and accuracy evaluation

The latitude and longitude information of 90 sample sites and 22 environmental variables were imported into MaxEnt v3.4.4 software together for modeling operations. The parameters were as follows: 25% of sample sites were set as test data and the remaining 75% as training data; repeated by ten times; The model was set to operate 1000 times. The environmental variable checked ‘Create response curves’, ‘Do jackknife to measure variable importance’, and the rest of field parameters were left as default.

The Pearson correlation among 22 variables was analyzed using ENMTools, and the contribution rates of each environmental variable based on 90 distribution points were obtained by MaxEnt software (Figure S1). The variables with correlation |r|< 0.8 were kept. When correlation |r|> 0.8 between variables, the only one variable with higher contribution rates was further selected for model construction. Finally, 11 environmental variables were obtained for model construction (Table 1).

Table 1 Percent contribution and permutation importance of variables used in MaxEnt model.
Full size table

The accuracy of the prediction model was evaluated by the Receiver Operating Characteristic Curve (ROC) and the Area under roc curve value (AUC). The AUC values range between 0 and 1, and divided into five grades: invalid prediction (0.5 ≤ AUC < 0.6), poor prediction (0.6 ≤ AUC < 0.7), average prediction (0.7 ≤ AUC < 0.8), good prediction (0.8 ≤ AUC < 0.9), and excellent prediction (0.9 ≤ AUC ≤ 1).

Suitable habitat grade classification

The result of suitable habitat was reclassified by equal interval method using spatial analyst tools in ArcGIS v10.8. The habitat was divided into four levels23: highly suitable habitat (P > 0.60), moderately suitable habitat (0.30 < P < 0.60), poorly suitable habitat (0.10 < P < 0.30), and unsuitable habitat (P < 0.10). The P value is the habitats suitability of the species.

Suitable habitat spatial variation

The SDM Toolbox v2.6 in ArcGIS v10.8 software was used to carry out spatial variation analysis, and the result showed by four types of spatial variations: range expansion, no occupancy, no change, range contraction. Also, centroid and migration of suitable habitats in different periods were calculated using SDM tools.

Results

MaxEnt model accuracy and environmental variables contribution

Through the analysis of variable contribution percentages and Pearson correlation, we obtained 11 crucial environmental variables for constructing the MaxEnt model to predict suitable habitat distribution for R. idaeus (Figure S1A; Table S1). Based on MaxEnt model, ROC curve was used to predict the accuracy of the suitable habitat of R. idaeus (Figure S1B). It was showed that the mean AUC value was 0.993, close to 1, with a standard deviation of 0.002. This indicated the model was reliable with high stability and accuracy.

The results of MaxEnt model construction showed that precipitation of the coldest quarter (bio19, 22.9% contribution rate), precipitation of the warmest quarter (bio18, 19%), temperature seasonality (bio4, 15%), mean temperature of the coldest quarter (bio11, 15%), and slope (13.7%) were the important variables and had great influence on the suitable habitat distribution of R. idaeus (Table 1). Together, these variables accounted for an impressive cumulative contribution value of 85.6%. Particularly noteworthy were the permutation importance values of mean temperature of the coldest quarter (bio11) and precipitation of the coldest quarter (bio19), standing at 70.6% and 19.8%, respectively, which were significantly higher than other variables (Table 1). The jackknife test results showed that when considering a single environmental variable, the model’s training gain was notably higher for mean temperature of the coldest quarter (bio11) (Figure S1C), followed by precipitation of the driest quarter (bio17) and precipitation of the coldest quarter (bio19). When a single variable was ignored, the gain of the bio11 was significantly reduced, which indicated that bio11 has more efficient information than other variables. Above all, these findings underscored the critical role of mean temperature of the coldest quarter (bio11), precipitation of the coldest quarter (bio19), precipitation of the warmest quarter (bio18), and temperature seasonality (bio4) were found as key environmental factors influencing the distribution of R. idaeus.

Further, the analysis of above four environmental variables were carried out and obtained the crucial threshold ranges associated with highly suitable habitats, defined by a probability of occurrence exceeding 0.60. For precipitation variables, the threshold range for precipitation of the coldest quarter (bio19) was determined to be 9.75 mm. In the case of precipitation of the warmest quarter (bio18), the range extended from 293.06 to 386.22 mm, with a peak value observed at 328.00 mm (Fig. 2A,B). Regarding temperature variables, optimal suitability was observed within specific threshold ranges. For mean temperature of the coldest quarter (bio11), the range was found to be from − 11.52 to − 8.33°℃. Similarly, temperature seasonality (bio4) exhibited peak suitability within the range of 1032.15 to 1138.99°℃ (Fig. 2C,D).

Fig. 2
figure 2

Response curves showing the relationship the probability of presence of R. idaeus and variables. (A–D), Response curve of R. idaeus and bio4 (temperature seasonality), bio11 (mean temperature of coldest quarter), bio18 (precipitation of warmest quarter), bio19 (precipitation of coldest quarter), respectively.

Full size image

Suitable habitat of R. idaeus under current climate scenario

The prediction results of suitable habitat under current climate scenario showed that the total area conducive to R. idaeus in China encompassed 2373.39 × 103 km2 (Table 2), equivalent to 24.72% of China. Among them, the highly suitable habitat covers an area of 151.60 × 103 km2, accounting for 6.39% of the total suitable habitat. The higher distribution appears in regions such as Shanxi, northern Hebei, Beijing, and northeast Shaanxi, and the smaller in Inner Mongolia, Jilin and Liaoning (Fig. 3). The moderately suitable habitat covered an area of 785.26 × 103 km2, comprising 33.09% of the total suitable area. This moderately suitable habitat spans regions in the northwest (northern Xinjiang, northern Shaanxi, southeast Gansu, and southern Ningxia), north (Shanxi, central and southeast Inner Mongolia), east (Hebei, Shandong) and northeast China (Liaoning, Jilin, and Heilongjiang). The poorly suitability area predominantly surrounded the highly and moderately suitable habitats, covering an expanse of 1436.53 × 103 km2, representing 60.53% of the suitable habitats. Above all, the primary potential suitable habitats for R. idaeus in China were identified across 11 autonomous regions and municipalities, namely Xinjiang, Inner Mongolia, Gansu, Ningxia, Shaanxi, Shanxi, Hebei, Beijing, Liaoning, Jilin, Heilongjiang (Fig. 3).

Table 2 Areas of suitable habitats for R. idaeus in China under different climate scenarios.
Full size table
Fig. 3
figure 3

Suitable habitats for R. idaeus under current climate scenarios using Arcmap 10.8.0 software (URL: https:// www. arcgis. com/ index. html).

Full size image

Suitable habitat of R. idaeus under future climate scenarios

Under various climate scenarios, the distribution of suitable habitats for R. idaeus exhibited diverse trends from the present to 2050s (Table 2; Fig. 4A). Specifically, under the SSP1 scenario, the total suitable habitat area will be increased to 2401.37 × 103 km2, with an alteration of 1.17% compared to the current scenario. Among them, the area of the moderately suitable habitats will be increased by 16.70%, while the area of the highly and poorly suitable habitat will be decreased by 1.22% and 9.75% respectively. Under the SSP2, SSP3 and SSP5 scenarios, the total area of suitable habitats exhibited reductions of 3.54%, 9.12% and 5.29%, respectively. Notably, the area of highly suitable habitats was decreased across all scenarios, except for the SSP3, where it showed a modest increase of 5.10%.

Fig. 4
figure 4

Changes of the suitable habitats of R. idaeus under different climate scenarios in 2050s and 2090s using Arcmap 10.8.0 software (URL: https:// www. arcgis. com/ index. html). (A), 2050s, average for 2041–2060; (B), 2090s, average for 2081–2100.

Full size image

From the present to the 2090s, there was a consistent downward trend in the total area of suitable habitats and highly suitable habitats across all four different SSP scenarios (Table 2; Fig. 4B). Specifically, the total suitable habitats under SSP1, SSP2, SSP3, and SSP5 experienced reductions of 3.60%, 6.31%, 1.87%, and 10.33%, respectively. The diminishing suitable habitat areas projected for the 2050s and 2090s suggested that climate change had a significantly adverse impact on the distribution of R. idaeus.

Spatial pattern changes in the suitable habitat under the different climatic scenarios

The future dynamics of R. idaeus distribution were poised to undergo shifts. In comparison to the present scenario, by the 2050s, expansions of R. idaeus were predominantly observed in regions encompassing Xinjiang, Inner Mongolia, Gansu, Sichuan, Shaanxi, Henan, Shandong, Heilongjiang, Jilin, and Liaoning under the SSP1 scenario. Particularly the expansion along the southern boundaries of habitats, expands from western Sichuan to Penglai and to Qixia in eastern Shandong (Fig. 4A). Notably, R. idaeus demonstrated expansion into desert and arid regions in Xinjiang, central Inner Mongolia, Gansu, and Qinghai. Meanwhile, contractions were evident in northern Xinjiang, northeastern Inner Mongolia, eastern Qinghai, western Sichuan, and eastern Tibet. Under the SSP2 scenario, the overall expansion and contraction regions were like those of SSP1. With climate change in future, the expansion of desert arid zones in Xinjiang, central Inner Mongolia, Gansu, and Qinghai has notably diminished, and the suitable area in northern Inner Mongolia has further reduced. Under the SSP3 scenario, the expansion primarily occurred in Xinjiang, Henan, southern Shandong, and Heilongjiang, while significant contractions were observed in the northeast of Inner Mongolia, Xinjiang, Tibet, Sichuan, Gansu, and Qinghai. Under the SSP5 scenario, R. idaeus expansion primarily unfolded along the southern boundary of the suitable area, while contractions occurred notably in Inner Mongolia, Xinjiang, Tibet, Sichuan, and Qinghai. Especially at the northern edge of the suitable area in Inner Mongolia, the contraction was distributed in two bands, one extending from Oroqen Autonomous Banner in Inner Mongolia to Baotou city, and the other from Oroqen Autonomous Banner in the north to Naiman Banner and Kurun Banner in the south (Fig. 4A).

In the 2090s, Under the SSP1 scenario, R. idaeus expansion was notably pronounced in southern Xinjiang, central Inner Mongolia, southern Shaanxi, Henan, Shandong, and Heilongjiang (Fig. 4B). The contraction mainly occurred in northern Xinjiang, area of eastern Qinghai bordering Gansu, Sichuan, Tibet, and northeastern Inner Mongolia. Under the SSP2 scenario, the pattern resembled SSP1, with expansion regions remaining unchanged. However, expansion in central Inner Mongolia, southern Shaanxi, and Henan was relatively reduced, while expansion intensified in Heilongjiang. Contractions decelerated in northern Xinjiang compared to SSP1. Under the SSP3, R. idaeus expanded towards the east and south of its distribution boundary while shrinking in the west and north. Under the SSP5, the expansion area further diminished, and contractions intensified, especially in northeastern Inner Mongolia (Fig. 4B).

The centroid and migration patterns of suitable habitat under various climate scenarios were analyzed. And the result showed that, under the current climate scenario, the centroid was situated in Baotou City, Inner Mongolia (109°46′22.386", 40°39′3.845"), and it would migrate to different degrees under four different CO2 emission scenarios in the future (Fig. 5). Under the SSP1 scenario, the centroid remained in Baotou City. By the 2050s, it would shift 54.85 km eastward to 110°20′28.932′′,40°38′45.931′′, and by the 2090s, it would shift to the southwest to 110°16′1.524′′,40° 37′37.232′′, with the shortest shift being merely 7.40 km. Under the SSP2 scenario, the centroid was predicted to migrate northeastward by 69.27 km to Guyang County (110°27′21.690", 40°52′12.554′′) by the 2050s. Subsequently, by the 2090s, it would further migrate northeast to 110°33′31.331′′, 40°56′49.070′′, with a shift distance of 12.39 km. Under SSP3, the centroid’s migration path entailed a 74.39 km northeastward movement and arrived at Guyang County (110°26′31.654′′, 41°1′53.418′′). Then, followed by a 52.62 km southeastward migration to Wuchuan County (110°59′2.281′′, 40°58′11.730′′) by the 2090s. Under SSP5, the centroid initially migrated northeastward to Tumd Right Banner (110°32′48.703′′, 40°46′36.793′′), covering a migration distance of 75.67 km. By the 2090s, it continued its northeastward trajectory, relocated 15.90 km, and arrived at Wuchuan County (110°42′24.271′′, 40°49′0.235′′) (Fig. 5). Above all, the suitable habitat centroid of R. idaeus was always located in the Inner Mongolia, with the migration path showing an eastward trend.

Fig. 5
figure 5

The centroid migration of suitable habitats of R. idaeus.

Full size image

Discussion

MaxEnt model is widely used for predicting suitable habitats19,20,24, with its accuracy critically impacting prediction outcomes. A larger AUC value of the ROC curve indicates a more precise prediction result. In this study, the AUC value in the ROC curve reached 0.993 (Figure S1B), showing an exceptionally high precision. Furthermore, the results of 19 survey areas in R. idaeus distribution area also affirmed the accuracy of the model (Figure S2A). Among the survey areas, there were 4, 2 and 10 survey areas fall into low, medium and high suitability habitats, and no wild R. idaeus was found in the moderately and poorly suitable habitats Yueba town of Ankang city, Huanglongshan of Yan’an city in Shaanxi Province and Laojunshan of Henan province (Figure S2A). R. idaeus was presented in all regions classified as highly suitable habitats, except for areas in northeast Shaanxi. We speculated that the presence of the Yellow River, a grade one river separating Shaanxi and Shanxi, served as the primary barrier hindering the westward spread of the species across the province, thus explaining the absence of wild R. idaeus in Shaanxi. Previous studies have indicated that long-term geographical features such as tall mountains or rivers can influence regional migration patterns, potentially leading to genetic and geographical discontinuities25,26, further supporting our hypothesis. In addition, in Shaanxi Province, the cultivation of R. idaeus varieties such as ‘Heritage’ and ‘Tulameen’ has been reported in Yulin City, Taibai County, Baoji City of Guanzhong area and Chenggu County, Hanzhong City. In Henan province, Fengqiu of Xinxiang City is known as ‘the raspberry town in China’. Above reports proved the suitability of Shaanxi and Henan province for the growth of R. idaeus. The distribution of R. idaeus observed in the investigation closely aligned with the current predicted suitable areas by the MaxEnt model, providing further validation of the model’s accuracy and supporting its application for predicting future distributions.

Temperature and precipitation are critical environmental factors, and directly affect the growth and development process of plants. Here, we found four crucial environmental variables for R.idaeus distribution involved in temperature and precipitation: mean temperature of the coldest quarter (bio11), precipitation of the coldest quarter (bio19), precipitation of the warmest quarter (bio18), and temperature seasonality (bio4) (Fig. 2). Improper temperature can adversely affect plants by disrupting processes such as photosynthesis, respiration, and transpiration27,28,29. During winter, the growth of R. idaeus requires a certain chilling temperature to ensure normal growth and development, with a minimum temperature threshold not lower than − 30 ℃7. It was consistent with our result that the temperature is critical to the distribution of R. idaeus35,36,37. In our study, we found that R. idaeus growth reached optimization conditions when temperature seasonality (bio4) ranged from 1032.15 to 1138.99, and mean temperature of the coldest quarter (bio11) ranged from − 11.52 to − 8.33 ℃, (Fig. 2A,B). Growth was inhibited when temperature seasonality over 1580.03 or below 851.35, and mean temperature of the coldest quarter (bio11) over 0.33 ℃ or below − 17.67 ℃. In addition, the response of dormancy and germination to temperature was closely related to the ecological distribution of the species, with a certain amount of humidity and temperature11,12 is required to stop the dormancy. In other research studies, we found similar results on the prediction of suitable areas of species with dormancy properties under climate change, such as for Toxicodendron vernicifluum, Crataegus pinnatifida, Rosa canina et al30,31,32. These studies also provided a reliable theoretical basis for our results.

Water is essential for plant survival, yet an excess or deficiency can hinder growth and development. Waterlogging leads to anaerobic or anoxic conditions for plants28,33,34. Drought stress has an impact on insufficient water absorption, initially causing wilt of leaves and, if prolonged, stunted growth, yellowing, leaf drop, and even death of the plant28,33. In winter, the aboveground part of R. idaeus wither, and excessive rainfall can cause excessive humidity around the underground roots, leading to root rot and even death, while a certain amount of precipitation is still needed to maintain the basic water required for survival. In the subtropical areas of southern China, the precipitation in winter is higher than that is in the north, as well as the temperature. R. idaeus are intolerant of waterlogging, and its growth requires a certain amount of cold storage to break dormancy35. In southern China, R. idaeus cannot overwinter normally, which is one of the reasons for the natural distribution absence of R. idaeus in southern China. It was clear that precipitation in winter plays an important role in the growth and development of R. idaeus. Herein, bio19 was the precipitation of coldest quarter and the contribution of bio19 was 22.1970 in our MaxEnt model (Table S1). Our study suggests that R. idaeus thrived optimally with a precipitation of the coldest quarter (bio19) of about 9.75 mm and is suitable for growth with a precipitation of the warmest quarter (bio18) ranging from 293.06 to 386.22 mm (Fig. 2C,D). This result was also similar to the overall winter climate conditions in northern China, which has colder climate and less rain36,37. Different plant species exhibit varying growth habits and adaptations to environmental variables19. For example, the potential distribution of Emeia pseudosauteri were influenced by temperature seasonality, altitude and distance to rivers38, and for Thrips tabaci Lindeman, the distribution were influenced by annual mean temperature, annual precipitation and precipitation seasonality39. Based on MaxEnt model and Marxan model, precipitation and temperature were more crucial for Agastache rugosa distribution compared to topographic and soil factors40. In field cultivation, it is crucial to satisfy the needs of irrigation, which is related to the survival ability of plants under unforeseen drought and waterlogging in the view of climatic changes. Under drought stress, it must be paid attention to the labor and groundwater costs of increased artificial irrigation that is used to keep a balance between planting and water resources conservation7,41. Some raspberry species, such as R. idaeus var. strigosus, exhibit drought resistance, however, selection for drought resistance plants has not been intentionally utilized in breeding for new raspberry cultivar42. In recent years, raspberry planting has developed widely in northwest China, where drought poses a significant limitation43. It is thus important that the research on drought-resistant genotypes of R. idaeus and their utilization need to be further exploited.

In this study, we found that the optimal habitats for wild R. idaeus primarily situated in the north of China, like Xinjiang, Inner Mongolia, Gansu, Ningxia, Shaanxi, Shanxi, Hebei, Beijing, Liaoning, Jilin, Heilongjiang (Fig. 3). At present, the main production areas of raspberry are Liaoning, Jilin, Heilongjiang, Henan, Hebei, Xinjiang, Yunnan. Meanwhile there are a small number of cultivation in Gansu, Ningxia, Qinghai, Inner Mongolia, Shaanxi, Shanxi, Shandong, Beijing and Tianjin44. Our results and these reports of production areas further confirmed that the moderately and poorly suitable habitat of R. idaeus were reliable, and a moderate scale of R. idaeus could be planned for planting. Moreover, Lu et al45 analyzed the climatic factors of the origin of different cultivated varieties and the climatic conditions of different regions in China, and obtained the suitable areas of cultivars45. Comparing our findings with Lu’s study, it’s evident that cultivated varieties tend to occupy more southerly habitats compared to wild R. idaeus, with wild populations extending even to the northern boundaries inaccessible to cultivated varieties (Fig. 5). The utilization of wild resources with hybrization with cultivars, and cultivation range of raspberry may be further expanded, as well as the resistance of raspberries to drought may be further improved.

According to our findings, in 2050s and 2090s, the primary habitats of R. idaeus remain unchanged. The centroid of distribution shifted towards the east and the variation of total suitable habitats areas was minor. The distribution area of R. idaeus decreased with the climate change, and the variation was dominated by the poorly suitable habitat (Fig. 4; Table 2). Interestingly, in the 2050s, under the low-emission scenario, there was a projected expansion of suitable habitats compared to current scenarios, suggesting a positive regulatory effect of temperature changes induced by elevated carbon dioxide concentrations within a certain threshold. We speculated that the appropriate increase in CO2 concentration may promote the development process of plants by increasing biomass, improving photosynthetic efficiency and water use efficiency and further promote stress resistance of plants46,47,48. Our results showed highly conducive habitats for R. idaeus, and these highly and moderately suitable areas were predicted to be found rather steady under future climate scenarios as well. It is conducive to establish long-term raspberries production bases to ensure sustainability and long-term conservation and breeding of the species.

This study predicted the suitable habitats of R. idaeus under both current and future climate scenarios. By combining the summarization of distribution of wild and cultivated R. idaeus through literature and field investigation, we propose that the cultivation scale can be appropriately expanded in the moderately and poorly suitable habitats in the future. However, these results are indicative because this study only considered large-scale climate and topographical factors but can be considered as a baseline for further exploitation. Indeed, in actual cultivation, soil, human activities, and economic conditions need to be fully considered to formulate long term management and cultivation plans for the species.