Based on the research objectives and fundamentals of the optimized MaxEnt model, a framework was constructed to simulate the current and future suitable distributions of kiwifruit species ( Figure 1 ). The framework is divided into four sections: (1) data collection: occurrence records of kiwifruit (Actinidia spp) species and environmental factors (bioclimatic, topographic, and soil factors) were obtained; (2) data preprocessing: occurrence records were converted to CSV files, samples that did not match environmental factors were filtered, environmental factors were all at 30” resolution, and input factors with the least amount of redundancy were filtered through correlation analyses; (3) parameter optimization and model setup: The ‘ENMevel’ package in R was utilized to select the optimal parameters from 72 candidate models, and subsequently, the MaxEnt model of kiwifruit species was established to predict the future period. And (4) Evaluation and analysis: Model accuracy was assessed using the AUC (Area Under the Curve) metric. Key environmental factors were identified through contribution ratios, response curves, and jackknife tests. Additionally, the SDMTool was employed to analyze changes in habitat suitability zones and their center of mass over time. A multi-factor correlation analysis was conducted to explore the relationship between suitability areas and kiwifruit yield trends.

Shaanxi Province is located in the transitional zone between the northern and southern climates in China, spanning various landforms such as plateaus, mountains, and plains. The annual temperature variation ranges from 31.57°C to 35.62°C, and the altitude gradient covers from 6 meters to 678.03 meters (Zhang and Li, 2022). This environmental heterogeneity poses strict requirements on the climatic adaptability of kiwifruit. Research has shown that extreme high - temperature stress can directly inhibit the physiological functions of plants (Raza et al., 2023), causing photosynthetic products to be preferentially allocated to vegetative growth and significantly reducing the carbohydrate and vitamin C content in fruits. Meanwhile, the imbalance in the precipitation pattern increases the risk of soil anoxia through extreme rainfall events (Furtak and Wolińska, 2023), threatening root respiration and leaf photosynthetic efficiency. In terms of hydrological conditions, during the fruit - swelling period, the soil water content needs to be maintained in the appropriate range of 6508.30 - 11332.82 mm. However, the kiwifruit’s intolerance to waterlogging (Li et al., 2021) requires that the planting area has efficient drainage capacity to avoid metabolic inhibition caused by root anoxia. Analysis of soil characteristics shows that kiwifruit in Shaanxi Province is concentrated in loam areas. Its pore structure can ensure the exchange efficiency of air and water and maintain moderate water - holding capacity. The soil organic carbon content and PH further screen suitable habitats by regulating nutrient availability. The distribution data of kiwifruit in Shaanxi Province (n = 259) were collected through the Global Biodiversity Information Facility (GBIF), the Chinese Virtual Herbarium (DHC), and field surveys. To ensure the temporal consistency between the species distribution data and the environmental data (from 1970 to 2020), orchard data recorded during this period and from orchards with trees over five years old were selected (n = 207). The ENMTools software was used to eliminate spatially redundant records at a resolution of 1 km. Finally, 99 valid distribution points (Field survey data is 3.03%) were obtained for model training ( Figure 2 ).

The study encompassed a total of 26 environmental factors, comprising 19 climatic factors, 4 soil factors within the lower soil layer, and 3 topographic factors ( Table 1 ). The climate data were obtained from the World Climate Database (http://www.worldclim.org/), and the baseline climate (1970-2020), future climate scenarios 2050 (2041-2060) and 2070 (2061-2080) from the BBC_CSM climate modelling system were used in this study. The future climate scenarios selected were SSP126, SSP245, SSP370, and SSP585, four emission models containing 19 bioclimatic factors important to species distribution and elevation data, with a spatial resolution of 30″(approximately 1km×1km). Slope and aspect data were calculated from elevation data. Slope and aspect data were calculated from elevation data. Soil data were obtained from the Food and Agriculture Organization of the United Nations (https://www.fao.org/) HWSDv1.0. The survival range of the kiwifruit root system is considered to be between 30-60cm, thus making it suitable for growth in slightly acidic soil with deep, fertile and loose layers and adequate drainage. It is imperative that the soil is rich in organic matter to ensure the full extension of the root system and the absorption of nutrients. The selection of the soil property data of the layer, including soil pH, soil coarse grain size, soil organic carbon content, and soil water content, was made with a spatial resolution of 30’ (approximately 1km×1km). The vector boundary maps involved in the study were obtained from the standard map service system (http://bzdt.ch.mnr.gov.cn/). Through the results of a single simulation test, the environmental factors that contributed less than 0.5% in the percentage contribution of environmental factors in the MaxEnt model output were excluded. The correlation between the factors was also analyzed using correlation, and the heat map of correlation coefficients ( Figure 3 ) was used to retain factors with correlation coefficients less than 0.80. Finally, 11 environmental factors were selected as the dominant environmental factors in this study.

The MaxEnt model is employed to predict the potential geographic distribution of the species, utilizing distribution data and environmental factor data to construct a relationship model. This model is then used to analyze the distribution suitability. A random selection of 25% of the distribution data was designated as the test set, with the remaining 75% of the distribution data being utilized as the training set for the construction of the kiwifruit habitat model with 10 operations. The contribution of environmental factors to species distribution was calculated, and response curves were plotted to analyze the environmental factors affecting species distribution and their thresholds. The area under the ROC curve, i.e. the AUC value was plotted for the simulation accuracy test of the model, and its magnitude gives a better indication of the accuracy of the simulated values of the model. An AUC of 0.5 to 0.6 indicates a poor simulation, 0.6 to 0.7 indicates a fair simulation, 0.7 to 0.8 indicates an accurate simulation, 0.8 to 1.0 indicates a very accurate simulation (Chicco and Jurman, 2023).

The ASCII data obtained from the model were converted to raster data using ArcGIS, and the presence probability P was reclassified. The minimum training presence logistic threshold (Mtp) in the MaxEnt model was utilized to ascertain the suitability of the distribution. The criteria for delineating the suitable distribution area are as follows: P<0.1 for non-suitable area, 0.1≤P<0.3 for low suitable area, 0.3≤P<0.6 for medium suitable area, and 0.6≤P for high suitable area.

The MaxEnt model performance was evaluated using the Area Under the Curve (AUC) metric ( Figure 5 ). The optimized model achieved an average AUC of 0.940 for training data and 0.918 for test data, reflecting a 7.55% improvement over the original model ( Table 2 ). These results confirm that the model provides high predictive accuracy for kiwifruit habitat distribution. Based on the suitability classification, kiwifruit habitats were reclassified into four classes ( Figure 5 ), with a suitability index of 30 per cent as the threshold, and areas with a suitability index greater than 30 per cent were classified as suitable habitat for kiwifruit. Suitable habitats for kiwifruit plants were mainly located at the foot of the foothills at low and middle elevations. The total area of suitable habitats was 83,000 km², of which 13,100 km² were highly suitable habitats and 32,600 km² were moderately suitable habitats. These habitats are mainly located in the northern foothills of the Qinling Mountains, the loess irrigation areas of northern Shaanxi, the Danjiang River basin of southern Shaanxi, and the Hanjiang River basin of southern Shaanxi, including mainly the cities of Weinan, Hanzhong, Xi’an and Baoji.

Considering the contribution of environmental factors, the variations in the probability of kiwifruit presence in response to environmental factors were analyzed, as well as the importance of each environmental factor in MaxEnt, suitable environments and key factors affecting kiwifruit growth. As illustrated in Figure 6 , graphs and margins of error are integrated for the response curves between each factor and the probability of kiwifruit presence. The probability of kiwifruit presence varies parabolically with the range of coldest season mean temperature, elevation, and annual temperature. The probability of occurrence of kiwifruit plants was found to be 0.58 at the coldest quarterly mean temperature of 2.3°C, representing the optimal winter temperature conditions for kiwifruit plants. At an altitude of 483 m, the probability of kiwifruit plant occurrence reached 0.69, representing the optimal altitude conditions. Conversely, the probability of kiwifruit occurrence was 0.64 when the annual temperature range was 32°C, signifying the optimal annual temperature difference. Conversely, an increase in soil moisture content was observed to result in a decline in the probability of kiwifruit plant occurrence.

Furthermore, the most significant factors affecting the spatial distribution of kiwifruit were BIO11 (32.81%), DEM (31.04%), BIO7 (9.18%), and r_swr (9.06%), as demonstrated in Table 3 . The cumulative contribution of the first four factors was more than 80%. The thresholds and means of the primary environmental factors were derived using a probability threshold of >0.5 for suitable kiwifruit habitat. As illustrated in Table 3 , values above 0.5 are considered optimal in terms of applicability probabilities. For instance, the mean temperature of the coldest quarter ranged from 0.69°C to 3.83°C, with a mean temperature of 2.26°C. The range of elevation was from 282.36 m to 678.03 m, with a mean elevation of 480.20 m. The distribution of kiwifruit sample sites was superimposed with the habitat suitability zones and contours of the environmental factors. This showed that the distribution points mainly occurred within the optimal thresholds for the environmental factors ( Figure 7 ).

The results of the importance of the environmental factors and the regularized training gain in MaxEnt are demonstrated in Figure 8 . The results of the folding knife test demonstrated that the elevation data exhibited the most significant gain in the model, and the MaxEnt model, which disregards the elevation drive, resulted in a substantial decrease in gain. When a single factor was considered, elevation, the coldest quarterly mean temperature, the annual temperature range, and the driest quarterly mean temperature exhibited higher values of normalized gain, strongly suggesting their significant contribution to model performance.

Kiwifruit is predominantly distributed near the foothill margins, one of the most sensitive areas to climate change. Therefore, the MaxEnt model was used to project potential suitable areas for kiwifruit in the 2050 and 2070 under scenarios SSP126, SSP245, SSP370 and SSP585. Figure 9 shows the spatial distribution of future kiwifruit suitability zones, categorized into high, medium and low suitability zones. By the 2070, significant fragmentation of kiwifruit suitable habitat is expected in all scenarios. Geographically, the distribution of kiwifruit in Shaanxi Province gradually expands to higher latitudes and altitudes as time and radiation intensity increase, and highly suitable areas recede to the south, focusing mainly on Chenggu and Hanbin in the Han River basin of southern Shaanxi. Under the SSP126 scenario, highly suitable areas for kiwifruit in Shaanxi Province will gradually disappear, and under the SSP370 scenario, the northern foothills of the Qinling Mountains, the Weibei Plateau Irrigation Area, and the Danjiang River Basin in southern Shaanxi gradually degenerate into low suitable areas. In the long term, the area of suitable habitat for kiwifruit is expected to rise and then fall under the influence of relatively low social development and emission trajectories. The area of low suitability areas will continue to rise, the area of medium suitability areas will rise and then fall, and the area of high suitability areas will continue to fall. The area of suitable habitat for kiwifruit is projected to continue to rise under relatively high development and emission trajectories. Under all scenarios, the area of suitable habitat disappears mainly in Shangzhou District in the Danjiang River Basin in southern Shaanxi Province, and expands mainly in the marginal parts of the current suitable habitat, mainly in Huangling and Xunyi.The area of suitable habitat under the SSP245 emission scenario would increase at a higher rate in the 2050, and the degraded area would also gradually increase by the 2070. Under the SSP370 and SSP585 emission scenarios, the area of extended relatively less point habitat is expected to decrease significantly, as shown in Figure 10 . This suggests that lower and higher SSP and emission pathways may have significant negative impacts on kiwifruit habitat.

As illustrated in Figure 11 , the temporal progression of the suitable area for kiwifruit cultivation is delineated according to specific SSP scenarios. It is evident that the area of highly suitable areas is in continuous decline, with the high emissions scenarios (SSP370, SSP585) resulting in a more than 50% reduction in the area of highly suitable areas by the 2050, from 13,100 km² to 4600 km² and 5200 km², respectively. The SSP126 emissions scenario and the SSP245 emissions scenario demonstrate a decline of 46.56% and 25.95%, respectively. By the 2070, the area of the high suitability region for the four emission scenarios (126, 245, 370, and 585) decreases to 0.27, 0.55, 0.42, and 0.5 million square kilometers, respectively. For the medium suitable area, the area first increases and then decreases, and by the 2050, the area of medium suitable area under the SSP245 emission scenario increases from 32,600 km² to 54,200 km², which is an increase of more than 66.26%. By the 2070, the area of the medium suitable area under the SSP370 and SSP585 emission scenarios decreases to less than the area of the contemporary medium suitable area, and the area of the medium suitable area under the other two emission scenarios remains higher than the area of the contemporary medium suitable area. Conversely, the area of low suitability is projected to increase under all emission scenarios. By the 2070, the area of low suitability zones for the four emission scenarios (126, 245, 370, and 585) is 606, 626, 659, and 63,200 square kilometers, respectively.

ArcGIS was utilized to calculate the location and directional movement of the center of mass of the kiwifruit suitability area for each scenario ( Figure 12 ). The current kiwifruit suitability center is located in Chang’an District at coordinates 108.786°E, 33.853°N. In the SSP126 scenario, the center of distribution of kiwifruit is projected to move in a southwesterly direction through Huyi until it reaches Ningshaan County in 2080 at coordinates 108.571°E, 33.810°N. In the SSP245 scenario, the center is projected to move in a southwesterly direction from 2040 to 2060, then in a northeasterly direction, and finally, from 2060 to 2080, it is projected to move in a southwesterly direction to the northern part of Chang’an District with coordinates 108.719°E, 34.177°N. In the SSP370 scenario, the center of mass is projected to move in a northwesterly direction from 2040 to 2060; however, it is predicted to move south-eastward to reach Huyi by 2060 to 2080, with coordinates 108.609°E, 33.921°N. In the SSP585 scenario, the center of mass is projected to move in a north-westerly direction from 2040 to 2060, and then move south-westerly to reach Huyi, with coordinates of 108.642°E, 33.915°N. In summary, based on the projected future climatic conditions, the center of distribution of kiwifruit is expected to shift slightly northwards and westwards with time.

As demonstrated in Figure 9 , the future primary producing countries continue to be classified as suitable zones. In conjunction with the feasibility of data collection, 31 main producing counties with average kiwifruit yield data of more than 100 tonnes (annual production) from 2018 to 2020 were collected, and the climate suitability zone hierarchy map and the main kiwifruit producing counties were superimposed and analyzed, so as to count the consistency of the main kiwifruit producing counties with the kiwifruit climate suitability zones simulated in the model, as shown in Figure 13 .The study revealed that the kiwifruit yields were situated within the medium and high suitability zones. The analysis indicated that six and 18 counties, respectively, exhibited kiwifruit production in the medium and high suitability zones. Furthermore, the study found that the spatial overlap between kiwifruit yield and climate suitability zones was 78.80%, suggesting a high degree of congruence between the distribution of kiwifruit-producing counties and the climate suitability zones within the prevailing climatic conditions.

The average yield in the main producing counties was subjected to qualitative analysis, with the area of high suitability zone, the area of medium suitability zone, the area of low suitability zone, the maximum temperature of the hottest month (BIO5), the range of annual temperature variation (BIO7), the mean temperature of the driest season (BIO9) and the mean temperature of the coldest season (BIO11) all taken into consideration. Without considering technological progress, kiwifruit yield in the main producing counties was correlated with the area of the three types of suitable zones, with the strongest correlation being with the area of the medium suitable zone. It is predicted that the kiwifruit yield in the medium-suitable zone will undergo an initial increase, followed by a subsequent decrease over the next 40 years. Concurrently, the high-suitable zone is expected to experience a continuous decline. This analysis suggests that the fluctuations in the areas of the high-suitable zone, the medium-suitable zone, and the low-suitable zone will result in a factor kiwifruit yield in Shaanxi Province between 2040 and 2060. The impact of climate change is projected to be the predominant factor influencing this variability, though it is not anticipated to result in a substantial reduction in yield. However, a decline in yield is predicted to occur in certain areas during the 2060–2080 period.

The study finds that kiwifruit distribution is mainly concentrated in southern Shaanxi and the Guanzhong region, with the most suitable habitats located in Weinan, Hanzhong, Xi’an, and Baoji. Shaanxi’s diverse geography creates a unique ecosystem, offering a range of ecological niches suitable for kiwifruit cultivation. Furthermore, the heterogeneous landscape of the Qinling Mountains has been shown to promote rapid species diversification. The foothills of the Qinling Mountains and their surrounding areas are biodiversity hotspots characterized by high species diversity, a large proportion of endemic species and unique plant communities (Huo et al., 2023). Projections indicate that kiwifruit’s suitable habitat will decline significantly under all four climate scenarios (SSP126, SSP245, SSP370, and SSP585) (Wang et al., 2024). The principal cause of this decline and fragmentation is believed to be climate warming.

In response to climate change, the center of kiwifruit cultivation is shifting geographically. The findings indicated that kiwifruit demonstrated a high degree of suitability in the Huyi District of Xi’an. The center of mass was located to the north of the current center of mass in different time periods and under different emission scenarios, except for the SSP126 emission. Consequently, the prediction indicates that the area of optimal conditions will coincide with the movement of the center of mass and expand in a northerly direction, influenced by future meteorological changes. The main manifestation of the kiwifruit suitable areas in Shaanxi Province responding to climate change is the movement towards higher latitudes and altitudes, which is consistent with the research results of other scholars (Wang et al., 2024; He et al., 2018). This finding also is consistent with the findings of related studies examining global warming that some species will migrate to higher dimensions and elevations (Zhang et al., 2018).

Kiwifruit sampling sites were largely within optimal environmental thresholds, confirming the high predictive accuracy of the model ( Figure 7 ). The presence of abundant rainfall, complex topography, and diverse microclimates in Shaanxi has been demonstrated to facilitate optimal growth conditions for kiwifruit plants, with optimal growth observed at low to medium altitudes within zones characterized by low but not excessively cold winters. The study finds that kiwifruit thrives in cold-season temperatures between 0.69°C and 3.83°C, aligning with previous findings by GAO (Gao et al., 2022). The coldest seasonal mean temperature has a direct impact on the overwintering growth of kiwifruit, and low temperatures can cause frostbite and frost damage to kiwifruit plants (Liu et al., 2023). Altitude is recognized as one of the most significant environmental factors within the ecosystem, exerting a direct influence on the microenvironment of kiwifruit. Research has demonstrated that kiwifruit exhibits optimal productivity within the range of 20–610 meters above sea level. However, studies have also shown a decline in mean fruit physical traits with increasing altitude (Zenginbal and Ozcan, 2018). At higher altitudes, lower temperatures are observed, along with higher atmospheric pressure and radiation intensity. Kiwifruit exhibits a high degree of intolerance to direct sunlight, with intense summer light potentially leading to leaf scorching, particularly in high-temperature environments, thereby increasing susceptibility to sunburn (Li et al., 2022). Consequently, kiwifruit cultivation is predominantly situated in lower altitudes.

The four soil factors – soil pH, soil coarse grain size, soil organic carbon content, and soil water content – accounted for a relatively high proportion of soil water content in the modelled contribution (Zheng et al., 2023). It has been established that kiwifruit has an increased demand for water during the expansion period, and insufficient water will result in a decrease in the quality of the fruit per weight. On the other hand, since kiwifruit combines waterlogging intolerance, excessive soil moisture tends to inhibit root respiration and jeopardize fruit tree health (Torres-Ruiz et al., 2016). Soil coarse-grainedness determines the pore structure of the soil, and larger pores allow rapid passage of air and water, but the soil’s water-holding capacity is poor, so kiwifruit is mostly planted in loamy soils (Gentile et al., 2021).

The study analyzes the impact of climate change on kiwifruit’s geographical indications (GIs). The results indicate that the area deemed highly suitable for kiwifruit cultivation is projected to experience a substantial decline under various climate change scenarios. The modelled highly suitable acreage would disappear under the SSP126 scenario, and these findings can be used to develop scientific recommendations for kiwifruit planning and climate-smart agricultural policies in different regions in response to climate change. For instance, agricultural investors can utilize these predictions to assess potential future high-quality kiwifruit production areas, target investments, construct factories and arrange industrial layouts. It is important to note that the majority of kiwifruit growers in China are smallholder farmers who lack the financial resources and technical capabilities to mitigate the adverse effects of climate change on kiwifruit cultivation. Consequently, this study may provide recommendations for kiwifruit climate-smart cultivation, with the aim of adjusting long-term planting plans and optimizing cultivation patterns.

In order to comprehend the sensitivity and adaptation of crop ecosystems to climate change, it is necessary to consider the interactions between ecological and socio-economic factors (Tang et al., 2021). This study identified key environmental factors affecting kiwifruit, but limited data restricted analysis of biological interactions and human activity impacts. In future research, a micro perspective should be adopted to understand the interaction between factors and the suitability of kiwifruit GI cultivation by focusing on agricultural, policy and economic benefits.