Composite function and Biomod2 for evaluating the influence of climate change on the distribution of Aedes aegypti and Aedes albopictus in China

Abstract

Vector-borne diseases transmitted by Aedes, including dengue fever, Chikungunya fever, Zika virus, and yellow fever, represent major global public health threats. This study utilized the Biomod2 modeling framework, incorporating 19 bioclimatic variables, to simulate the current and future geographical distributions of Aedes aegypti and Aedes albopictus in China under climate change scenarios (SSP2–4.5 and SSP5–8.5). The results indicated that under future climate scenarios, highly suitable regions for both Aedes would decrease in area, while moderately suitable regions would expand. The co-presence probability analysis revealed that highly suitable regions for both species would concentrate in southern and southeastern China, with notable areas in Yunnan, Guangxi, Guangdong, and Hainan. From current to 2090s, the centroid would shift to northeast under SSP2–4.5 and SSP5–8.5. For Ae. aegypti, the most important variables were isothermality (bio3, 44.05 %), precipitation of the wettest quarter (bio16, 27.87 %), and mean temperature of the coldest quarter (bio11, 22.4 %). For Ae. albopictus, the mean temperature of the coldest quarter (bio11, 54.12 %), annual precipitation (bio12, 22.76 %), and precipitation of the coldest quarter (bio19, 13.47 %) were most significant. These findings highlight the potential impacts of climate change on the distribution dynamics of dengue vectors and provide a basis for developing targeted surveillance and control strategies to mitigate future transmission risks.

1. Introduction

Vector-borne arboviral illnesses—dengue, Chikungunya, Zika and yellow fever—now rank among the most rapidly expanding global health threats [1]. Their transmission dynamics are governed by a complex interplay of drivers [2], of which climate variables are arguably the most pervasive [3]. Ambient temperature modulates every stage of the mosquito life-cycle: egg eclosion, larval development, adult survival, gonotrophic cycle length, and the extrinsic incubation period of the virus itself [1,3]. Within thermal optima these processes accelerate, amplifying vectorial capacity; outside these bounds both vector and pathogen are rapidly constrained [4,5]. Atmospheric moisture governs egg desiccation risk and adult longevity, while rainfall patterns determine the availability and persistence of oviposition sites [6,7]. Wind speed and barometric pressure further shape dispersal trajectories and host-seeking behavior [8]. Climate change is already amplifying the frequency and intensity of extreme weather events—prolonged droughts, intense precipitation, and land-falling cyclones—thereby redistributing surface water and reshaping the spatiotemporal mosaic of mosquito habitat. The net effect is a non-linear, region-specific reconfiguration of transmission risk that traditional empirical models struggle to anticipate [9,10].

Recent studies have further underscored the profound impacts of climate change on ecological systems and disease transmission dynamics. For instance, Lang et al. [11] revealed an increasing phenological asynchronization between plants and animals under warming scenarios, which may disrupt trophic interactions and ecosystem stability. Chen et al. demonstrated that drought events—especially extreme droughts—are projected to significantly reduce vegetation productivity, particularly in dryland ecosystems, thereby altering habitat suitability for vector species [12]. Moreover, Chen et al. emphasized that while vegetation growth may increase under high CO₂ scenarios, ecosystem resilience—especially in equatorial regions—could decline, increasing vulnerability to disturbances such as drought and fire [13]. These findings collectively highlight the complex and nonlinear responses of ecosystems to climate change, which may indirectly influence vector-borne disease risks. In addition, Lawlor et al. [14] provided a comprehensive synthesis of global species redistribution patterns under climate change, noting that while many species are shifting their ranges poleward or upslope, the direction and magnitude of these shifts are highly variable. Such redistributions may alter the geographic overlap between vectors, pathogens, and human populations, thereby reshaping disease transmission landscapes. These insights reinforce the importance of integrating climate-ecosystem-disease interactions into predictive modeling frameworks for vector-borne diseases like dengue.

Dengue fever, an acute infectious disease caused by the dengue virus and transmitted to humans through the bites of Aedes aegypti or Aedes albopictus, has become an increasingly severe public health concern worldwide in recent years [15]. The clinical spectrum of dengue ranges from mild influenza-like illness to severe and potentially fatal manifestations such as dengue hemorrhagic fever and dengue shock syndrome [16,17]. The absence of specific antiviral treatments or widely effective vaccines makes prevention and vector control essential [18]. Furthermore, urbanization and associated environmental changes have compounded the challenges of outbreak prevention and response [19,20]. The rapid spread of the virus and its various strains has drawn significant global attention [21,22]. The distribution of dengue fever is vast, ranging from tropical and subtropical regions of Asia to parts of America, Africa, and Oceania. Statistics indicate that millions of people are infected with the dengue virus each year, imposing a considerable disease burden [23,24]. Given these emerging challenges, understanding how climate change influences the spatial distribution of Aedes aegypti and Aedes albopictus is critical for anticipating future dengue risk. In China, Ae. aegypti is primarily found in coastal areas south of 22° north latitude, such as Guangdong, Guangxi, Xishuangbanna, and Hainan. Aedes albopictus is widely distributed across China, ranging from Hainan to the northern regions of Shenyang and Dalian, west to Longxian and Baoji, and southwest to the Tibet. With the continuous exploitation and utilization of natural resources by humans, as well as the large-scale movement of people caused by tourism, the range of dengue fever has gradually expanded, and the frequency of its prevalence has been increasing. Given that there is no vaccine prevention and no specific treatment for the disease to date, it is particularly important to carry out research on the geographical distribution of Ae. aegypti and Ae. albopictus.

Recent studies have modeled the future distributions of Ae. aegypti and Ae. albopictus using CMIP6 projections and single-algorithm approaches such as MaxEnt [25,26]. Despite growing evidence that climate change is altering vector distributions, it remains unclear how simultaneous changes in multiple meteorological drivers will jointly reshape the spatial overlap of Ae. aegypti and Ae. albopictus across China. Moreover, single-model frameworks may underestimate prediction uncertainty and overlook non-linear interactions among predictors. To overcome these limitations, we employed Biomod2—an ensemble platform integrating ten commonly used species distribution models—and further introduced a Composite function to quantify the co-presence probability (CPP) of both vectors [[27], [28], [29]]. This combination allows us to (i) reduce model-specific biases, (ii) map the simultaneous risk zones that are most relevant for dengue transmission, and (iii) explicitly test the hypothesis that climate change will alter the spatial concordance of the two vector species.

2. Methods

2.1. Distribution data

Georeferenced occurrence of Ae. aegypti and Ae. albopictus were assembled through a systematic sweep of open-access repositories (GBIF, CNKI) and peer-reviewed literature indexed in Web of Science, PubMed and Medline [30]. ENMTOOL was used to proofread and screen the obtained distribution points, excluding the impact of overfitting simulation caused by high spatial correlation [31,32]. Finally, 10963 points of Ae. aegypti) and 8975 points of Ae. albopictus) were chosen (Fig. 1).

Fig. 1.

Distribution data of Ae. aegypti and Ae. albopictus in the world. Green points, occurrence data of Ae. aegypti. Blue points, occurrence data of Ae. albopictus.

2.2. Environmental variables

The 19 bioclimatic variables were selected based on their established fundamental ecological relevance to organism survival, development, reproduction, and population dynamics [28,33]. These variables represent annual trends, seasonality, and extreme environmental conditions that are known to fundamentally limit species' geographic distributions. A detailed list of all variables along with their general ecological justification was provided in Table S1. The periods were diveded into current (1970–2000) and future (2050s, 2070s) (2.5 arc-minutes) [33]. Multiple studies indicate that the BC-CSM2-MR model is a medium-resolution climate system model developed by the National Climate Center of China, and it performs well in East Asia, particularly in China [34,35]. Therefore, this study selects it as the climate model for future climate projections. To span the anticipated range of future anthropogenic forcing, we adopted the SSP2–4.5 and SSP5–8.5 scenarios as representative medium- and high-end pathways, respectively. SSP2–4.5 depicts a socio-economic trajectory consistent with nations that pursue sustainable development while gradually decarbonizing, thereby offering a plausible mid-range radiative forcing of approximately 4.5 W/m² by 2100. Conversely, SSP5–8.5 represents a fossil-fuel-intensive future characterized by rapid economic growth and delayed mitigation actions, culminating in a high-end forcing of ∼8.5 W/ m² and providing an upper-bound estimate of potential climate impacts [36].

2.3. Model construction

On the R (version 4.2.3) platform, package Biomod2 (version 4.2.5) was used to model the distribution of Ae. aegypti and Ae. aegypti. The modeling algorithms included Maximum Entropy Model (MaxEnt), Random Forest (RF), Artificial Neural Network (ANN), Generalized Linear Model (GLM), Enhanced Regression Tree Model (BRT), Classification and Regression Tree Model, Surface Range Envelope Model (SRE), and Flexible Discriminant Analysis (FDA) [37]. For MaxEnt, hyper-parameters were fine-tuned through ENMeval (RM = 0.5, feature class = LQ); all other techniques relied on Biomod2 defaults. Background points were generated at a 3:1 ratio to presences to balance prevalence. Seventy-five percent of occurrences were randomly retained for training in each iteration, and the entire procedure was repeated ten times per algorithm, yielding an ensemble of 80 calibrated outputs [28,37,38].

To quantify the relative contribution of each environmental predictor, we leveraged the internal ranking algorithms within Biomod2. Model reliability was then gauged by two complementary metrics: the area under the receiver-operating characteristic curve (AUC) and the true-skill statistic (TSS) [37]. AUC ranges from 0.5 (random discrimination) to 1.0 (perfect discrimination); values ≥0.80 denote strong predictive power, whereas those <0.70 indicate unsatisfactory performance. TSS spans 0–1, with scores >0.70 signifying high predictive accuracy and scores <0.50 suggesting low reliability. Using these thresholds, we selected the five top-performing single models—ranked jointly by AUC and TSS—and integrated them into an ensemble to enhance overall robustness. [27,29,37].

2.4. Calculation of co-presence probability

In order to further clarify the potential risk of dengue fever, we introduced a composite probability function to calculate the co-presence probability of the two Aedes.

$$\mathit{CPP}=\sum _{i=0}^n{P}_i\ast W_i\left(i=0,12,.\dots \mathrm{n}\right)$$

CPP represents the co-presence probability. P_i represents the presence probability of the i-th Aedes (0–1). W_i represents the weight of P_i, reflecting its local abundance and, consequently, its potential contribution to pathogen transmission.

Since worldwide absolute abundance data (e.g., CDC light-trap or larval surveys) are unavailable, we used the number of occurrence records compiled for Biomod2 as an empirical proxy for relative abundance. The underlying biological assumption is that, under comparable sampling effort or reporting probability, species recorded more frequently are likely to represent larger local populations and therefore exert a greater influence on arbovirus transmission. Weights were calculated as:

$${\mathrm{W}}_{\mathrm{I}}=\frac{{\mathrm{N}}_{\mathrm{i}}}{\sum _{\mathrm{k}=1}^2{\mathrm{N}}_{\mathrm{k}}}$$

With denoting the total number of occurrence records for species i. In the present dataset (Fig. 1), Ae. aegypti and Ae. albopictus are represented by 10,963 and 8975 records, respectively. This weighting scheme is internally consistent with the modeling data and avoids potential biases introduced by external datasets.

3. Results

3.1. Evaluation of model accuracy

By synthesizing the top five single-algorithm outputs (RF, GBM, MaxEnt, GLM, CTA) into an ensemble, predictive performance improved markedly. For Ae. aegypti, the combined model reached an AUC of 0.948 and TSS of 0.771, while for Ae. albopictus the corresponding metrics were 0.968 and 0.81, respectively. These gains confirm that the ensemble approach yields more reliable habitat projections than any constituent model alone (Fig. S1).

3.2. Geographical distribution under current climate situation

In China, the highly suitability region of Ae. aegypti was primarily located in the southern coastal regions and some warm and humid areas, including most parts of Guangdong, the southern part of Fujian, the majority of Taiwan, Hainan, the southern part of Guangxi, the southern part of Yunnan, and the southeastern part of Tibet (Fig. 2 A), with an area of 38.18 × 10⁴ km². Notably, Yunnan and Guangdong were identified with high suitability areas of 11.36 × 10⁴ km² and 10.90 × 10⁴ km², respectively. Under current climate condition, the highly suitability regions of Ae. Albopictus lies within the thermally mild, moisture-rich band stretching from the southeastern seaboard to the middle Yangtze basin (Fig. 2B). The largest contiguous swaths of high-suitability habitat occur in Guangxi (15.61 × 10⁴ km², Guangdong (15.61 × 10⁴ km²) and Hunan (11.71 × 10⁴ km²).

Fig. 2.

Suitability regions of Ae. aegypti, Ae. albopictus and their CPP in China under current climate situation.

To elucidate the transmission risks of dengue fever, this study employed a joint density function to assess the co-presence probability (CPP) of Ae. albopictus and Ae. aegypti. The results indicated that areas of highly suitability were primarily concentrated in the southern and southeastern regions of China (Fig. 2C), including Yunnan, Guangxi, Guangdong, Hainan, Fujian, Taiwan, and Hong Kong. Notably, Yunnan and Guangxi have significant high suitability regions, measuring 6.5 × 10⁴ km² and 7.97 × 10⁴ km², respectively. Moderately suitability regions were predominantly found in parts of Sichuan, Jiangxi, Guizhou and Zhejian (Fig. 2C).

3.3. Distribution pattern changes of suitability regions

Under the SSP2–4.5 scenario, the highly-suitability region of Ae. aegypti contracted from 38.18 × 10⁴ km² at current to 28.38 × 10⁴ km² in the 2050s and then marginally recovered to 29.57 × 10⁴ km² by the 2090s, whereas the moderate-suitability region expanded markedly from 56.17 × 10⁴ km² to 107.44 × 10⁴ km² before contracting slightly to 85.86 × 10⁴ km² (Fig. 3 A, 4 A). Under SSP5–8.5, the highly-suitability region declined more sharply to 18.23 × 10⁴ km² in the 2050s and subsequently rebounded to 31.00 × 10⁴ km²; the moderately-suitability region increased from 56.17 × 10⁴ km² to 65.88 × 10⁴ km² and then to 99.56 × 10⁴ km² (Fig. 3D, 4D).

Fig. 3.

Potential suitability region of Ae. aegypti, Ae. albopictus and their CPP in 2050s under climate change scenarios. (A) Ae. aegypti under SSP2-4.5; (B) Ae. albopictus under SSP2-4.5; (C) CPP under SSP2-4.5; (D) Ae. aegypti under SSP5-8.5; (E) Ae. albopictus under SSP5-8.5; (F) CPP under SSP5-8.5.

Fig. 4.

Potential suitability region of Ae. aegypti, Ae. albopictus and their CPP in 2090s under climate change scenarios. (A) Ae. aegypti under SSP2-4.5; (B) Ae. albopictus under SSP2-4.5; (C) CPP under SSP2-4.5; (D) Ae. aegypti under SSP5-8.5; (E) Ae. albopictus under SSP5-8.5; (F) CPP under SSP5-8.5.

For Ae. albopictus, SSP2–4.5 drove an increase in highly-suitability region from 121.77 × 10⁴ km² to 138.05 × 10⁴ km² (2050s) and to 140.05 × 10⁴ km² (2090s), while moderately-suitability region decreased from 143.39 × 10⁴ km² to 130.11 × 10⁴ km² and 131.22 × 10⁴ km², respectively (Fig. 3B, 4B). SSP5–8.5 produced a similar upward trend in highly-suitability region (121.77 → 138.05 → 140.05 × 10⁴ km²) and a comparable decline in moderately-suitability region (143.39 → 121.57 → 120.66 × 10⁴ km²) (Fig. 3E, 4E).

Regarding CPP, SSP2–4.5 led to a modest expansion of highly-suitability region from 55.59 × 10⁴ km² to 57.66 × 10⁴ km² (2050s), followed by a retreat to 53.43 × 10⁴ km² (2090s); moderately-suitability region rose steeply from 47.67 × 10⁴ km² to 98.26 × 10⁴ km² before easing to 87.88 × 10⁴ km² (Fig. 3C, 4C). Under SSP5–8.5, highly- suitability region declined from 55.59 × 10⁴ km² to 41.45 × 10⁴ km² (2050s) and recovered slightly to 44.57 × 10⁴ km² (2090s), whereas moderately-suitability region increased from 47.67 × 10⁴ km² to 56.26 × 10⁴ km² and then to 103.14 × 10⁴ km² (Fig. 3F, 4F).

3.4. Centroid migrations of suitability regions under climate change scenarios

We quantified range-shift trajectories by tracking the geographic centroids of suitability region under each scenario.

For Ae. aegypti: SSP2–4.5 drove an initial north-east displacement of 175.37 km (108.86°E, 24.44°N → 109.86°E, 25.88°N) by the 2050s, followed by a 89.47 km south-west retraction (109.21°E, 25.26°N) in the 2090s, yielding a net north-east shift of 89.74 km over the full period. Under SSP5–8.5, the centroid first drifted 27.26 km south-east to 108.90°E, 24.16°N, then rebounded 182.58 km north-east, culminating in a cumulative north-east migration of 161.4 km by the 2090s (Table S2).

For Ae. albopictus: SSP2–4.5 prompted a modest 10.74 km south-east move (110.94°E, 28.61°N → 111.01°E, 28.53°N) before a 15 .13 km north-east correction, producing a net 13.46 km north-east displacement. Under SSP5–8.5, the centroid moved 12.87 km south-east initially, followed by a 5.87 km north-east retreat, resulting in a 13.31 km south-east net shift (Table S2).

CPP: SSP2–4.5 generated two successive north-east vectors of 39.02 km and 11.22 km, producing an overall 47.9 km north-east translation. In SSP5–8.5, the centroid first shifted 19.23 km south-east, then 16.98 km north-east, for a cumulative 28.8 km north-east migration by the 2090s (Table S2).

3.5. Key variables

For Ae. aegypti, the top three variables with the higher percent contribution rate were the isothermality (bio3, 44.05 %), the precipitation of wettest quarter (bio16, 27.87 %), the mean temperature of coldest quarter (bio11, 22.4 %). From the perspective of permutation importance, the top two variables were the annual mean temperature (bio1, 33.37 %) and the precipitation of wettest quarter (bio16, 21.62 %) (Fig. S2).

For Ae. albopictus, mean temperature of coldest quarter (bio11) provided 54.12 % of the model contribution, followed by annual precipitation (bio12) and precipitation of coldest quarter (bio19,). Permutation importance ranked annual mean temperature (bio1, 44.77 %), annual precipitation (bio12, 18.09 %), isothermality (bio3, 13.46) and min temperature of coldest month (bio6, 13.01 %) (Fig. S2).

According to the response curve between environmental variables and species distribution probability, the suitable range of environmental variables for the distribution of Ae. aegypti can be determined. When the value of annual mean temperature was in the range of 9.95 °C–18.45 °C, with the increase in temperature, the predicted distribution probability of Ae. aegypti increased, and decreased rapidly when the temperature was higher than 18.45 °C, and the suitable range was 9.95 °C–32.21 °C. Response curves indicated that the probability of Ae. albopictus presence rose across an annual mean temperature window of 8.34 °C–13.06 °C, declined steeply beyond 27.41 °C, and yielded an overall thermal envelope of 8.34 °C–28.35 °C (Fig. 5 A). The vector required mean temperatures of the coldest quarter to exceed −3.21 °C (Fig. 5B).

Fig. 5.

Response curve between environmental variables and presence probability.

4. Discussion

To more reliably assess the synergistic impacts of climate change on the distribution of dengue vectors, this study employed the Biomod2 ensemble platform alongside a composite function to quantify species co-presence probability (CPP). This integrated approach offers several key advantages over single-model studies. First, the Biomod2 framework harmonizes ten distinct algorithmic perspectives, mitigating the intrinsic biases and uncertainties associated with any single model and thereby producing a more robust and consensus-driven projection of habitat suitability [39]. Second, and more critically, the composite function provides a novel and pragmatic metric for direct risk assessment. Rather than examining each species' distribution in isolation, the CPP calculates the spatially explicit probability of both Ae. aegypti and Ae. albopictus co-occurring, which is the scenario most conducive to dengue virus transmission and outbreak potential. This allows us to identify geographic hotspots where the presence of both vectors amplifies public health risk, a nuance that single-species models cannot capture. Ultimately, this methodology not only enhances prediction accuracy but also translates model outputs into a more actionable format for targeted surveillance and control planning. Nevertheless, the present study was based on a single-climate projection from the BCC-CSM2-MR model, which precluded quantification of the full uncertainty envelope inherent to multi-model ensembles. Integrating the broader CMIP6 multi-model ensemble in future investigations would therefore provide a more comprehensive understanding of how climate-projection uncertainty propagates into mosquito-distribution simulations.

Climate warming is driving species redistribution and reshaping the ecological niches of disease vectors by altering temperature and precipitation patterns [[40], [41], [42], [43], [44]]. Unlike most studies emphasizing the northward shift of suitable habitats, our Biomod2 ensemble simulation reveals a seemingly contradictory scenario: under SSP2–4.5 and SSP5–8.5 scenarios, the high-suitability area decreased by 26 % and 42 %, respectively, while the total suitable area increased by 48–120 × 10⁴ km². This “high-suitability area's contraction, total-suitability area's expansion” pattern aligns with the findings of Li et al. based on MaxEnt under SSP1–2.6, which indicated the expansion of Ae. albopictus to the west [45], yet differs from the uniform expansion assumption of Liu et al. through simple linear extrapolation, highlighting the advantage of multi-model integration in capturing non-linear responses [25]. The centroid shifted northeast by 47.9–161.4 km (2090s) was primarily controlled by the winter mean temperature (bio11) threshold (> − 3.21 °C), reflecting the mosquito's tracking of low-temperature limits rather than high-temperature upper limits. The ecological consequences include the emergence of new transition zones between the Yangtze and Yellow River basins, where winter conditions now meet the egg overwintering requirements, and increased summer precipitation extends the larval development season, thereby increasing the probability of virus spillover. Similar processes have been validated in Europe: Pasquali et al. noted that for every 100 km Ae. albopictus advances northward, the risk of Chikungunya outbreaks increases by 30 % [46].

Meteorological conditions can directly or indirectly affect the spread of infectious diseases, especially vector borne diseases [43]. The spatiotemporal distribution of vector organisms is easily influenced by climatic factors, and temperature, rainfall, humidity, and light can significantly alter the spread of diseases by affecting the reproduction, development, behavior, and population dynamics of vectors [1,3,47]. In addition, the reproduction and expansion of pathogens in vector organisms can also be influenced by climate factors, thereby altering the dynamics and activity range of the host population [2,48,49]. The distribution and growth of insects are closely related to their living environment. In nature, some species are distributed in limited areas, while others are distributed globally [50,51]. Some insects only appear during mild seasons, while others maintain population dominance over several seasons [52]. The species-environment relationship is an important aspect in the study of ecological needs and spatial distribution of a species. Laboratory studies demonstrate that Ae. aegypti cannot complete a full generation below 14 °C and exhibits markedly reduced fecundity at 15–16 °C [24,53], confirming thermal minima as primary distributional constraints. The species also requires high atmospheric moisture, thriving at 85–95 % relative humidity and showing a positive correlation between ambient humidity and occurrence [53]. Our response curve identifies a threshold of 118.31 mm of precipitation during the wettest quarter, above which the likelihood of Ae. aegypti presence increases markedly, corroborating earlier reports that elevated humidity facilitates colonization. Thus, the vector's range is delimited jointly by winter temperature minima and rainfall-mediated breeding opportunities.

Our study confirmed that bio11, bio12 and bio19 were the three most influential variables for Ae. albopictus. Experimental data indicate that the species cannot complete a generation below 15 °C, whereas 20–35 °C accelerates development [58]. Model output aligns with these thermal limits, assigning a lower threshold of −3.21 °C for bio11 and 8.34 °C–13.06 °C for bio1, coinciding with the subtropical and temperate monsoon zones where the vector is currently concentrated.

5. Conclusions

This study projects a northeastward shift in the climatic suitability for Ae. aegypti and Ae. albopictus in China under climate change, with highly suitable areas contracting but moderately suitable areas expanding. Key drivers include cold-quarter temperature and precipitation patterns. These findings align with global trends of poleward species migration but contrast with some single-model studies that projected broader expansion, highlighting the improved reliability of our ensemble approach. The results suggest potential new dengue risks in central and southwestern China, urging targeted surveillance and adaptive control strategies in transitioning regions. Future work should integrate socioeconomic and ecological factors to refine predictions. This study provides a critical foundation for climate-resilient public health planning. We note that human-mediated transport and trade strongly accelerate vector dispersal, yet their future trajectories remain difficult to parameterise within CMIP6 climate scenarios; integrating shared socio-economic pathways will be essential to reduce projection uncertainty.

Declaration of generative AI and AI-assisted technologies in the writing process

During the preparation of this work the author(s) used [Kimi+/https://www.kimi.com/] in order to improve readability and language. After using this tool, the authors reviewed and edited the content as needed and take full responsibility for the content of the published article.

Funding

This research received no external funding. The authors declare that no grants, equipment, or other financial support were provided for the work described in this paper.

CRediT authorship contribution statement

Jianjun Xu: Writing – review & editing, Writing – original draft, Visualization, Validation, Supervision, Software, Resources, Project administration, Methodology, Investigation, Formal analysis, Data curation, Conceptualization. Rulin Wang: Project administration, Methodology, Investigation, Formal analysis, Data curation. Zhenhan Mo: Software, Resources. Han Zhang: Visualization, Validation. Yujing Zhang: Validation, Supervision.

Declaration of competing interest

The authors declare that there are no conflicts of interest.

Appendix A. Supplementary data

Supplementary material 1

Supplementary material 2

Supplementary material 3

Supplementary material 4

Data availability statement

Distribution data of Aedes aegypti and Aedes albopictus were available at https://doi.org./10.6084/m9.figshare.28648607; Environmental variables were available at https://www.worldclim.org/.