Skip to main content
NCBI home page
Scientific Reports logoLink to Scientific Reports
. 2025 Jul 1;15:22080. doi: 10.1038/s41598-025-05987-x

Ensemble distribution model of Muga Silkworm (Antheraea assamensis) and its primary host plant Soalu (Litsea monopetala) in defined climate space

Kuladip Sarma 1,✉, Vivek Chetry 1,✉, Amal Bawri 2, Keshob Jyoti Borah 1, Bulbuli Khanikor 1, Tarali Kalita 3, Banasri Mech 1, Malabika Kakati Saikia 1, Manas Das 1
PMCID: PMC12214580  PMID: 40594578

Abstract

Muga silkworm is one of the most economically and culturally important insect found in northeastern region and has immense future potential for entrepreneurship development owing to its sericogenic nature. The distribution of Muga SilkWorm (MSW) in wild are extremely important determinant for rearing a disease resistant domesticated variety that can significantly improve the yield of the later in terms of silk production. This paper aims to explore the distribution of the wild MSW in northeastern part of India in connection with the distribution of its primary host plant viz., Litsea monopetala(LM) Soalu in wild. Species Distribution Models (SDMs) have been used to know the potential distribution of MSW in historical climate scenario and also the impact of future climate has been assessed using a two climate models (CMCC-ESM2 and HadGEM3) with two different scenarios (SSPs) for each model. The climate space was defined in two dimensional space using Principal Component Analysis (PCA) to simplify the models. Altogether two model algorithms were used to get the final ensemble model for both the MSW and the two primary host plants. The model performed for MSW was found to be good with average AUC value greater than 0.80. On the contrary, the model performed for Litsea monopetala was found to be excellent with average AUC value greater than 0.90. Similarly, the ensemble models performed using future climatic data under the CMCC-ESM2 and HadGEM3 models for MSW have shown AUC values within acceptable range (0.78-0.82), whereas, model of the host plant (LM) have shown AUC values in higher range i.e. within excellent model category (0.90-0.95). Thus, in this reduced climate dimension, the potential distribution of the MSW and LM have been compared among current and future climate scenarios of two different models and district wise distribution potential has been calculated. The overlapping potential habitat of the invasive plant MSW and its host plant LM in Northeast India, particularly Assam and Arunachal Pradesh, is projected to experience significant and critical changes under future climate conditions between 2040 and 2060. The present research identifies climate-driven habitat changes, which may impact host plant availability and, in turn, silkworm survival and silk production. The results also highlight the importance of climate-resilient sericulture policy, adaptive agriculture practices, and conservation strategies to maintain Muga silk production under changing environmental conditions.

Keywords: Algorithm, Assam, Bioclimatic variables, Litsea monopetala, Muga silkworm, PCA, Sericogenic, SDM

Subject terms: Climate-change ecology, Ecological modelling, Entomology

Introduction

Substantial evidence is available on the diversity and distribution of flora and fauna in a species rich environment to assess the impact of climate, specifically in multidimensional climate space1. Traditional exploration studies on flora and fauna usually focused on understanding diversity indices along the current location of species occurrences2. That in modern day science has very minimal implications as ecologists are now interested in assessing the future prospects of species under study. Valuing the economic and other socio-cultural aspects of the species, they are being on varying scale. For example, with the pivotal emphasis on sustainable natural resource use, the regional fauna and flora having economic importance have given the highest importance at various levels including conservation and sustainable extraction. Therefore, it is of utmost necessity to study the predictive distribution pattern of such species in a multidimensional climate space defined by bioclimatic variables3. Although this is not a new concept of understanding geographic pattern of species in current and future climate space, the focused species prioritization is being overlooked in studies. In contrast, SDMs are available for species in many cases, including most studies on invasive species3–5.

MSW is one of the most economically and culturally important insects found in the northeastern region and has immense future potential for entrepreneurship development due to its sericogenic nature6. MSW belongs to the genus Antheraea that is one of the largest genera of the Saturniidae family, most of which are known for silk production. Most of the species of genus Antheraea are polyphagous in nature and depend on variety of host plant species for their survival. The primary food plants,  Machilus gamblei (Som) and Litsea monopetela (Soalu), play a pivotal role in the life cycle of MSW7.  Machilus gamblei, classified under the Lauraceae family, stands out as a medium-sized tree characterized by grey bark, elliptic lanceolate leaves, and globose fruits. This species is not only distributed extensively throughout Assam but is also found in both natural and cultivated populations. Litsea monopetela, another Lauraceae member, is a middle-sized evergreen tree featuring greyish-brown bark, broadly elliptic leaves, and ovoid blackish fruits. Its wide distribution in the foothill region of Assam further emphasizes its significance in the habitat of MSW.

The distribution of MSW in the wild are extremely important determinant for rearing a disease-resistant domesticated variety that can significantly improve the yield of the later in terms of silk production. Despite its potential, Sericulture in Northeast India has a number of obstacles including the control of disease, the deficiency of contemporary infrastructure, and the effects of climate change8. Production is seriously threatened by silkworm diseases as pebrine, flacherie, and grasserie. Understanding these illnesses and creating resistance strains have not been the focus of research yet. Long-term threat also comes from climate change, which may modify the microclimates necessary for silkworm rearing.

Recently this species has been recognized by tagging Geographical Indications by Government of India owing to its specific geographic origin and reputation due to to such origin. This species is commercially exploited for producing golden silk and is endemic to Northeast India9. The making and wearing of muga silk cloth is an integral part of Assamese culture and also that contributes in the local economy of the state of Assam. MSW has its known distribution in the Brahmaputra valley in Assam, East, West and South Garo hills of Meghalaya, Mokokchung, Tuensang, Kohima and Wokha districts of Nagaland, Lohit and Dibang valleys, Changlang and Papum Pare districts of Arunachal Pradesh, Tamenglong district of Manipur and Cooch ehar district of West Bengal10–12.

The distribution information of MSW has been reported in very few published scientific articles. Although MSW is commonly believed to be a single race, its wild counterpart exhibits remarkable diversity in color polymorphism and voltinism. Wild populations are found in higher altitudinal regions of Meghalaya, Nagaland, and Arunachal Pradesh. This diversity has been documented in various studies along with the potential of the silk industry in the northeast region13–17. As all the reported distribution locations are based on direct evidence of the presence of the species, more area of suitable occurrences can be anticipated from Northeastern Region of India. However, no reports are available in terms of modelling approach for predicting suitable habitat of the species from elsewhere. The present study presents a novel comprehensive assessment of the potential distribution of the economically significant silkworm, MSW and its primary host plant, Litsea monopetala, under current and future climate scenarios using an ensemble modeling approach. Ensemble methods reduce bias and variance, leading to more reliable predictions compared to individual models, and enhance predictive accuracy, which helps mitigate the challenges of modeling rare species with limited occurrence data18. Additionally, individual models may overfit training data, whereas ensemble approaches smooth out extreme predictions, improving generalizability19. However, despite their advantages, the impact of model tuning on ensemble performance remains understudied. Many researchers rely on default settings, which may not be optimal and could affect model accuracy and reliability20.By integrating five advanced algorithms (Maxent, RF, GAM, CTA, and SVM), the study enhances predictive accuracy, crucial for forecasting Muga silk production, a valuable industry in Northeast India. Furthermore, the use of Coupled Model Intercomparison Project (CMIP6) climate projections provides future-oriented insights into habitat shifts, ensuring long-term sustainability of both the silkworm and its primary host plant. The present study identifies climate-induced vulnerabilities affecting host plant availability, which directly impacts Muga silkworm survival and silk yield, thereby offering essential information for conservation and plantation management strategies to sustain the silk industry in a changing climate. Therefore, this paper aims to explore the distribution of the wild MSW in northeastern part of India in connection with the distribution of one of the primary host plants viz., Litsea monopetala Sualu in wild.

Results

Ensemble models for MSW and host plant LM

The model performed for MSW was found to be good with average AUC value 0.83 (Fig. 1). The model evaluation statistics has shown that Random Forest (RF) has the highest AUC value (0.88) among all the five different model algorithms used in the ensemble model followed by MAXENT (0.87) and SVM (0.85) (Table 1). Both all the algorithms have shown higher correlation (r value > 0.7) except CTA vs. MAXENT (r=0.59). On the contrary, the model performed for Litsea monopetala was found to be excellent with average AUC value 0.94 (Fig. 2). For this ensemble model, the AUC value range has followed similar pattern as that of MSW being highest for RF (0.99) and lowest for CTA (0.90)(Table 1). However, in this model, all the algorithms have shown higher correlation (r value > 0.7). The PCs considered for both these ensemble models (for MSW and LM) have shown to be contributed in the respective models in varying percentages (Fig. 3.

Fig. 1.

Fig. 1

Open in a new tab

Ensemble model output of current distribution of MSW in NE India: (A) Habitat suitability map; (B) Principal Component (PC) contribution; (C) Model performance across algorithms. Cyan bars represent AUC values and red bars represent the percentage of models retained in the ensemble and (D) Correlation matrix among different algorithms.

Table 1.

Model evaluation statistics for MSW and its host plant (Litsea monopetala) in NE India.

Model algorithm AUC Sensitivity Specificity TSS Kappa
Antheraea assamensis
MAXENT 0.87 1.00 0.75 0.75 0.01
RF 0.88 0.92 0.83 0.75 0.75
GAM 0.78 1.00 0.56 0.56 0.03
CTA 0.78 0.44 0.62 0.06 0.06
SVM 0.85 0.92 0.79 0.71 0.71
Litsea monopetala
MAXENT 0.94 0.96 0.92 0.88 0.06
RF 0.99 1.00 0.97 0.97 0.97
GAM 0.94 0.97 0.91 0.88 0.36
CTA 0.90 0.94 0.86 0.80 0.80
SVM 0.93 0.96 0.90 0.86 0.86
Open in a new tab

Model performance was categorized as poor (0.5–0.6), fair (0.6–0.7), good (0.7–0.8), very good (0.8–0.9), and excellent (0.9–1) based on AUC values

AUC, area under the curve, TSS true skill statistic.

Fig. 2.

Fig. 2

Open in a new tab

Ensemble model output of current distribution of LM in NE India: (A) Habitat suitability map; (B) Principal Component (PC) contribution; (C) Model algorithm wise AUC values and (D) Correlation matrix among different algorithms.

Fig. 3.

Fig. 3

Open in a new tab

Percentage contributions of PCs in four SSPs of two corresponding models for both (A) MSW and (B) LM in Northeast India.

Predictions in future climate

The ensemble models performed for MSW using future climatic data of the two models; CMCC-ESM2 and HadGEM3, have shown AUC values within acceptable range (0.78-0.82), though the TSS values varies from 0.53-0.63.(Table 2). The other evaluation statistics did not show much differences. However, models of the host plant have shown AUC values in higher range i.e. within excellent model category (0.91-0.93) and the TSS values also lies in the higher range, varies from 0.82-0.86. PC1, PC2 and PC3 have been contributed in varying percentage in the two SSPs of both the GCMs (Fig. 3) . However, for the CMCC-ESM2 model, the maximum percentage area under high suitability threshold was calculated for the optimistic scenario with intermediate challenge (SSP2-4.5:21.21%) followed by pessimistic mitigation challenge scenario (SSP5-8.5:19.50%) and lowest in the current distribution model (17.77%) for MSW (Fig. 4). Whereas; for the HadGEM3 model, the maximum percentage area under high suitability threshold was calculated for SSP2-4.5 (20.30%) followed by SSP5-8.5 (19.33%).

Table 2.

Future climate model evaluation metrics in comparison to the current distribution model.

GCM SSPs AUC Sensitivity Specificity TSS Kappa
Antheraea assamensis
CMCC-ESM2 SSP2-4.5 0.78 0.73 0.80 0.53 0.32
SSP5-8.5 0.82 0.84 0.79 0.63 0.41
HadGEM3 SSP2-4.5 0.79 0.78 0.76 0.54 0.31
SSP5-8.5 0.81 0.84 0.73 0.57 0.56
Current model 0.83 0.85 0.71 0.56 0.31
Litsea monopetala*
CMCC-ESM2 SSP2-4.5 0.91 0.93 0.89 0.82 0.56
SSP5-8.5 0.93 0.94 0.92 0.86 0.60
HadGEM3 SSP2-4.5 0.92 0.95 0.89 0.84 0.58
SSP5-8.5 0.91 0.95 0.88 0.83 0.56
Current model 0.94 0.96 0.91 0.87 0.61
Open in a new tab

As there are only a few coordinates available for Persea bombyciana, for host plant, the model is run only for Litsea monopetala

AUC area under the curve, TSS true skill statistic, Kappa Cohen’s Kappa coefficient, GCM: Global Climate Model; SSP: Shared Socioeconomic Pathway.

*One of the primary host plant of MSW

Fig. 4.

Fig. 4

Open in a new tab

Future climate distribution of MSW in NE India for four different Shared Socioeconomic Pathways (SSPs): For CMCC-ESM2 model (A) ssp2-4.5 (B) ssp5-8.5; For HadGEM3 model (C) ssp2-4.5 and (D) ssp5-8.5.

Similarly,the PCs have been observed to contribute in varying percentages across all four SSPs for the two GCMs conducted for LM, the host plant. Here also, for the CMCC-ESM2 model,the maximum percentage area under high suitability threshold was calculated SSP5-8.5 (07.21%) followed by SSP2-4.5 (07.04%) and lowest in current distribution model (06.32%). (Fig. 5; Table 3). In contrast, for HadGEM3 the maximum percentage area under high suitability threshold was calculated for SSP2-4.5 (07.85%) followed by the pessimistic scenario SSP5-8.5 (07.06%) and lowest in current distribution model (06.32%).

Fig. 5.

Fig. 5

Open in a new tab

Future climate distribution of Litsea monopetala in NE India for four different Shared Socioeconomic Pathways (SSPs): For CMCC-ESM2 model (A) ssp2-4.5 (B) ssp5-8.5; For HadGEM3 model (C) ssp2-4.5 and (D) ssp5-8.5.

Table 3.

Estimated suitable habitat areas for the Muga Silkworm and its primary host plant under future climate projections and the current distribution model.

GCM SSPs Area (in percentage)
No suitability Suitability
Low Moderate High
Antheraea assamensis
CMCC-ESM2 SSP2-4.5 34.04 25.96 18.79 21.21
SSP5 8.5 35.67 30.48 14.35 19.50
HadGEM3 SSP2-4.5 29.80 27.40 22.50 20.30
SSP5 8.5 25.07 35.35 20.24 19.33
Current model 30.98 30.95 20.30 17.77
Litsea monopetala*
CMCC-ESM2 SSP2-4.5 79.51 8.14 5.31 7.04
SSP5 8.5 81.88 5.86 5.05 7.21
HadGEM3 SSP2-4.5 81.87 5.81 4.46 7.85
SSP5 8.5 81.69 5.39 5.86 7.06
Current model 81.42 6.96 5.30 6.32
Open in a new tab

One of the primary host plants of MSW. Suitability thresholds were determined based on ensemble species distribution modeling (SDM) using different Global Climate Model (GCM) and Shared Socioeconomic Pathways (SSPs)

In Assam and Arunachal Pradesh, most districts have accommodated less than 1% of the total highly suitable area for MSW, with 9 districts in Assam and 6 in Arunachal Pradesh falling into this category. (Fig. 4). In contrast, only a few districts in these states are having more than 5% high suitability. Under the SSP2-4.5 scenario, the number of districts with<1 % suitability generally decreases across most states, while those with 1–5 % suitability increase, particularly in Assam (10 districts) and Arunachal Pradesh (8 districts) for the model CMCC-ESM2 whereas; the SSP5-8.5 scenario shows a similar pattern, with a slight rise in districts having >5 % suitable area in Assam (6 districts) and Arunachal Pradesh (1 district). For model HadGEM3, both the scenarios indicate further shifts, with a reduction in districts with<1 % suitability in Assam and Arunachal Pradesh, and a concurrent increase in districts with 1–5 percent suitability. Notably, Arunachal Pradesh shows an increasing number of districts with >5 % suitability in the SSP5-8.5 scenario compared to other projections. Overall, the projected climate scenarios suggest an expansion of moderately suitable areas (1–5 %) at the expense of districts with very low suitability, with some regions experiencing an increase in highly suitable (>5 %) areas.

Study extent and methods

Study extent

The model extent has been bounded by the administrative boundary of seven northeastern states and Sikkim has been excluded as no preliminary occurrence data was recorded from the state. As MSW is endemic to northeast India specifically, care has been taken to incorporate distribution records from all the seven states considered for the model extent.

Methods

The data on species occurrences was obtained from published sources and field surveys. Nineteen bioclimatic variables with a spatial resolution of 30 arc seconds, sourced from the WorldClim database (www.worldclim.org) following the methodology of Hijmans et al., were utilised in the modelling process. Bioclimatic variables encompass factors such as long-term trends (e.g., average annual temperature and annual precipitation), seasonal variations (e.g., temperature and precipitation range over the year), and extreme conditions (such as the coldest and warmest month temperatures and three-month periods of wet and dry spells). These variables are derived from monthly temperature and rainfall data. For future climate prediction, the upgraded version of Intergovernmental Panel on Climate Change (IPCC) as per sixth assessment report (AR6), the Coupled Model Intercomparison Project (CMIP6), showcased five different Shared Socioeconomic Pathways (SSP) viz., SSP1–2.6, SSP2–4.5, SSP3-7.0, SSP4–6.0, and SSP5–8.5. This pathways are weighted based on two types of socioeconomic challenges viz., mitigation challenges and adaptation challenges21.The CMIP6 Shared Socioeconomic Pathways (SSPs) define various climate futures: SSP1–2.6 envisions a sustainable, low-emission world; SSP2–4.5 follows a moderate path with some climate action; SSP3–7.0 depicts a fragmented world with high emissions and regional conflicts; SSP4–6.0 describes abrupt inequalities with moderate-to-high emissions; and SSP5–8.5 represents a fossil-fuel-driven scenario with extreme warming22. The present study used future climate change scenarios from the CMIP6-based CMCC-ESM2 and HadGEM3 models. CMCC-ESM2 is based on the coupling between the climate coupled model CMCC-CM2, that accounts for interactive dynamics of atmosphere, ocean, sea-ice and land components, with the inclusion of the marine biogeochemistry to fully represent the global carbon cycles23. On the other hand, the HadGEM3 model includes a coupled atmosphere-ocean configuration, with or without a vertical extension in the atmosphere to include a well-resolved stratosphere, and an Earth-System configuration which includes dynamic vegetation, ocean biology and atmospheric chemistry24.These models have also been used by various researchers for distribution prediction of biota in Asia25–27.

The R package ’SSDM’ with the ’ensemble model’ function in version 4.0.2 was employed to generate and assess the models. The “SSDM” package in R is a comprehensive computational platform offering various methodological tools and parameterization options, including pseudo-absence selection, variable contribution assessment, model accuracy evaluation, consensus forecasting across models, species assembly design, and computation of weighted endemism28. We used the first three axes of the principal component analysis (PCA) of 19 bioclimatic variables downloaded from the Worldclim database to define a three-dimensional climate space. The number of components that explained minimum 80 percent cumulative proportion of variance were selected for the model. Principal component analysis was performed using the function rasterPCA built under the R package RStoolbox29. Subsequently, the model outputs were brought into ArcGIS 10.1 for the ultimate distribution mapping. The ensemble model used five algorithms: Maxent, Random Forest (RF), Generalized Additive Model (GAM), Classification Tree Analaysis (CTA) and Support Vector Machine (SVM) . A brief description of each algorithm is given below.

Maxent

The function in a species distribution model known as “MaxEnt” (Maximum Entropy) incorporates environmental data for known-presence locations as well as a large number of “background” locations [36]. The background locations are created at random from the raster data used to extract the environmental data. The result of this presence only model is a model object that may be used to forecast the appropriateness of other sites to anticipate the entire range of a species, for example.

Random Forest

The Random Forest system, an extension of the Classification and Regression Trees (CART) framework, is implemented in R using the randomForest package with two main arguments: a predictor variable data frame and a response vector. When the response variable is categorical, Random Forest performs classification; otherwise, it conducts regression.

Generalized additive model (GAM)

GAM is a flexible modeling technique, suited for capturing nonlinear, unspecified relationships between predictor variables and a response variable30. GAMs generalize the family of generalized linear models (GLMs), by replacing the linear functional form by a sum of smooth functions31.

Classification tree analysis (CTA)

CTA is a machine learning algorithm that used to identify the cutpoint on an ordered variable, or assignment rule for a categorical variables, that optimally discriminates between two or more classes (eg, outcome categories)32.

Support vector machine (SVM)

SVM is a classical machine learning algorithm, through mathematically complex. This machine learning algorithm has been extensively used in various fields because of its flexibility in solving range of classification problems33.

Model evaluation

Various metrics are employed to assess model performance, with the choice depending on study objectives. Some metrics are threshold-dependent, while others, like correlation coefficient and Area Under the Receiver Operator Curve (AUC), are threshold-independent. A high AUC indicates that areas predicted as suitable are indeed where the species is present, while lower values suggest otherwise. However, using AUC for evaluation has received criticism, especially regarding the spatial extent for selecting background points. Model validation involves splitting data into training and test sets, with discrimination based on AUC values. Kappa coefficient is another assessment, categorizing performance based on thresholds. In this study, four potential distribution areas were categorized from 0 to 1: No Suitability, Low Suitability (0.2–0.4), Moderate Suitability (0.4–0.6), and High Suitability (0.6–1) following IPCC guidelines.

Discussion

The impact of climatic variables on distribution and abundance of insects with economic importance have been least studied and the present study presents a detailed investigation report on Muga Silkworm (MSW) and its interactions with one of the primary host plants in a climate space defined by ninteen bioclimatic variables freely available in worldclim database. These understanding can help in formulating effective management and mitigation strategies for the species in its native range34. The identification and exploration of suitable habitats of MSW in wild is key for management of better variants of commercial MSW, the wild variety being known for resistance to various bacterial and other infections35. This might ultimately lead to the understanding of the distribution and specific habitat requirements of MSW for developing a disease resistance variety that can eventually increase the silk production. This study describes a detailed impact of environmental variables on distribution of both MSW and one of its primary host plant LM in the wild. The host plant specific study for phytophagous insects is thus, useful in understanding underlying impact of climate change36. As the host plant and herbivorous insects are greatly linked by their complex interactions, the later may be affected directly or indirectly with the subsequent changes in the population dynamics of the host plant due to climate change37.The present study also depicted a similar kind of dependence probability of the MSW on the host plant LM and almost 90 percent of the high suitable area of LM was found within the high potential range of MSW. In contrast, the highly suitable area of LM covered less than 30% of the MSW’s highly suitable habitat, meaning that 70% of the area where MSW is found must rely on other host plants, such as Machilus gamblei , Litsea cubeba, Litsea nitida, and Litsea salicifolia.38. Also due to its preference for elevations below 1000 meters, LM is mainly distributed in the Himalayan foothills. In Northeast India, it is largely confined to the plains and foothills of Assam, where such lower elevations are common, unlike the other, more mountainous states in the region39. However, most of the LM’s high suitable area was found in Assam and thus, the dominance of the species as a host plant of MSW in wild in eastern Assam in particular is very high.Assam’s climate further supports this distribution pattern, which features warm, humid summers and mild winters, with annual temperatures ranging from 3°C to 36°C. These conditions closely match the environmental preferences of LM, which thrives in these favorable temperature ranges across Assam and are conducive to the growth and proliferation of LM throughout the state40. Additionally, the longstanding practice of Muga silk cultivation in Assam has led to the intentional propagation and conservation of LM. Recognizing its vital role as a host plant for the Muga silkworm, local communities have historically promoted its growth to sustain silk production. This cultural emphasis has further contributed to the widespread presence of LM in the state41. Furthermore, the high suitable habitat of MSW is widespread as revealed by the current model, maximum in Assam followed by Arunachal Pradesh, Manipur, Meghalaya,Nagaland and Mizoram in terms of area covered by the distribution probability. Therefore, there is a huge scope of exploration of other secondary host plants of the species in the wild in other states of Northeast India.

Projecting into the future, under the moderate climate change scenario (SSP2-4.5), our model indicates that the highly suitable habitat for the MSW will become more concentrated and interconnected in northern Assam and along the southern border extending into Meghalaya. This suggests a potential shift in the species’ distribution, creating a more continuous and potentially robust population in these regions. However, this projected distribution does not fully align with the predicted habitat suitability of the host plant, LM. Specifically, the model forecasts a contraction of LM’s range with a corresponding expansion of its distribution towards the northern and southern extremities of the region. Conversely, under the more severe climate change scenario (SSP5-8.5), the model reveals a surprising similarity to the current distribution pattern, particularly regarding the areas of high suitability. This suggests that while moderate climate change might induce a spatial shift, more drastic changes might not significantly alter the existing high-suitability zones. Notably, a stark contrast emerges in Manipur, where the high suitability areas are predicted to completely disappear under the pessimistic SSP5-8.5 scenario, highlighting the region’s heightened vulnerability to severe climate change. The observed discrepancy between the high suitability areas for the Muga silkworm (MSW) and its primary host plant, Litsea monopetala (LM), suggests a potential role for alternate host plants, such as  Machilus gamblei. In regions where MSW exhibits high suitability but LM’s distribution is limited, the suitability of these alternate host plants becomes a crucial factor. These areas may offer viable habitats for MSW, provided that the alternate host plants can thrive under the prevailing environmental conditions, indicating the importance of considering host plant diversification for sustainable Muga silkworm cultivation. This dietary flexibility may offer some resilience against climate-induced habitat shifts, although proactive conservation strategies remain essential to manage potential range declines42,43.

The results indicate that Assam and Arunachal Pradesh show the highest prospects for MSW suitability under future climate scenarios. Both states exhibit a consistent increase in the number of districts with 1–5% and > 5% high suitability areas, particularly under SSP2-4.5, SSP5-8.5, and the high greenhouse gas emission scenarios (SSP2-4.5 and SSP5-8.5). Arunachal Pradesh, in particular, experiences a notable increase in districts with >5% high suitability under SSP5-8.5, suggesting that climate change might enhance habitat suitability in some regions of the state. This could be attributed to its diverse topography, which can create optimal environmental conditions such as moderate temperatures, high humidity, abundant host plants and other favourable microclimatic conditions which are critical for MSW development and silk production despite larger climatic change44. It is noteworthy here that the domesticated MSW in Assam is concentrated in districts such as Sivasagar, Lakhimpur, Jorhat, Golaghat, Dibrugarh, Tinsukia,  Charaideo, Dhemaji, Goalpara, Kokrajhar, Kamrup (R), and Darrang38,45 which is also evident from the present study. Meghalaya also shows moderate potential, with an increase in districts having 1–5% suitability across scenarios. However, the number of districts with >5% suitability remains relatively low, indicating that while some areas may become more favourable, they might not reach the highest suitability threshold. Nagaland and Manipur show limited changes, with most districts remaining in the<1% suitability category. Mizoram consistently exhibits low suitability, with no recorded increase in districts falling into higher suitability categories, suggesting that the state may not be a viable area for future expansion of MSW habitats. The findings suggest that climate change may lead to a shift in the geographical distribution of highly suitable areas rather than a uniform expansion. The increasing number of districts in the 1–5% suitability range implies that moderate suitability might become more widespread, which could enhance the overall potential for MSW cultivation, even if fewer areas reach the highest suitability threshold (> 5%). This could influence conservation and silk production strategies, emphasizing the need for targeted interventions in states like Assam and Arunachal Pradesh, where prospects appear most promising. In the fiscal year 2023-24, Assam’s total muga silk production reached 250 MT. Approximately 44,263 families are engaged in muga culture across 15,397 hectares of land. Thus, the result of the study could be very useful in developing a strategy for sustainable and profitable muga silk based industry particularly in Assam and Arunachal Pradesh owing to maximum highly suitable area inclusion. Additionally, the observed patterns highlight the importance of microclimatic variations in shaping species distribution under climate change, suggesting that localized environmental factors should be considered in future habitat suitability assessments.

The modelling approach to understand about distribution is a useful tool for least known wild fauna like wild MSW. MSW has been studied for its distribution and commercial rearing potential, mainly in the domesticated varieties. The wild varieties are least explored and the host plant specificity has not been studied. As the species has very high commercial as well as cultural value, it is important to have a healthy population for better production and survival. The industry potential as a whole for the silkworm has been studied and challenges have also been listed17. However, a better scientific approach to know its climate niche in both current and future climate is very important as many of the indigenous people are involved with the industry46,47. In this paper we have presented a interactive distribution understanding of MSW and its associated host plant in both current and future climate that will help scientific community as well as farmers to come up with better management practices for MSW in future course of time. Several management practices are crucial for cultivators based on the findings regarding potential impacts of climate change on Muga silkworms (MSW) and its host plant LM. To ensure the long-term sustainability of Muga sericulture, a diversified approach to host plant cultivation is essential48. Cultivators should actively promote mixed plantations of host plants that offer resilience against climate-induced declines in a single host species. Habitat connectivity and conservation efforts, such as preserving natural vegetation corridors and conserving wild host plant populations, will support Muga silkworm dispersal and genetic diversity49.

Despite its contributions, this study has certain limitations that should be acknowledged, which also present opportunities for future research. Refining distribution models by incorporating high-resolution environmental layers that account for microclimatic conditions and biotic factors–such as interspecies competition and predator-prey dynamics–could improve prediction accuracy. The 30-arc-second resolution bioclimatic data used may not fully capture microclimatic variations essential for species survival at finer scales. Additionally, key anthropogenic factors, including land-use changes, pesticide application, and deforestation, were not explicitly considered, despite their potential impact on MSW populations. Furthermore, species occurrence data and location records of primary host plants are limited, as the model relies on available records that may not accurately reflect the species’ actual distribution, particularly in remote areas. Conducting extensive field surveys could enhance data reliability and improve model accuracy.

Supplementary Information

Supplementary Information 1. (373.6KB, jpg)
Supplementary Information 2. (333.7KB, jpg)

Acknowledgements

We sincerely acknowledge the Department of Zoology, Gauhati University, Assam, India for providing administrative and logistic support to the authors. We extend our sincere gratitude to the authors, contributors, and farmers who provided the secondary information on Muga silkworms in Northeast India. We also deeply appreciate the reviewers’ valuable contributions to improving this manuscript.

Author contributions

Conceptualization: K.S. and V.C.; Data Collection: K.S., A.B. and V.C.; Methodology: K.S., V.C. and K.J.B.; Writing original draft: K.S., V.C. and K.J.B.; Writing review and editing: K.S., V.C., A.B., K.J.B., T.K., B.M., B.K., M.K.S. and M.D.

Data availibility

The datasets used and analysed during the current study are available from the corresponding author on reasonable request.

Footnotes

Publisher’s note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Contributor Information

Kuladip Sarma, Email: kldpsarma@gauhati.ac.in.

Vivek Chetry, Email: vivekchetry127@gmail.com.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-025-05987-x.

References

  • 1.Coelho, M. T. P. et al. The geography of climate and the global patterns of species diversity. Nature622, 537–544 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 2.Nautiyal, S., Bhaskar, K., Imran Khan, Y. D. Methodology for Biodiversity (Flora and Fauna) Study 13–37 (Springer, 2015). 10.1007/978-3-319-15464-02 .
  • 3.Sarma, K. et al. Impact of climate change on potential distribution and altitudinal shift of critically endangered amentotaxus assamica dk ferguson in Arunachal Pradesh Himalaya India. Theor. Appl. Climatol.155, 261–271 (2023). [Google Scholar]
  • 4.Norberg, A. et al. A comprehensive evaluation of predictive performance of 33 species distribution models at species and community levels. Ecol. Monogr.89(3), 01370 (2019). [Google Scholar]
  • 5.Rodríguez-Castañeda, G., Hof, A. R., Jansson, R., Harding, L. E. Predicting the fate of biodiversity using species’ distribution models: Enhancing model comparability and repeatability (2012) [DOI] [PMC free article] [PubMed]
  • 6.Bhattacharyya, M. & Goswami, C. Muga industry-a pride of Assam: An estimation of employment generation, Kamrup District. Space Cult.3, 42. 10.20896/saci.v3i2.149 (2015). [Google Scholar]
  • 7.Jigyasu, D. K., Patidar, O. P., Kumar, A. & Shabnam, A. A. Muga silk insect (Antheraea assamensis). In Commercial Insects 126–145 (CRC Press, 2023)
  • 8.Haloi, K., Kalita, M. K., Nath, R. & Devi, D. Characterization and pathogenicity assessment of gut-associated microbes of muga silkworm antheraea assamensis helfer (lepidoptera: Saturniidae). J. Invertebr. Pathol.138, 73–85 (2016). [DOI] [PubMed] [Google Scholar]
  • 9.Boro, P. & Borah, S. D. Biodiversity of sericigenous insects in north-eastern region of India: A review. J. Entomol. Zool. Stud.8(4), 269–275 (2020). [Google Scholar]
  • 10.Baruah, J. P. Muga silkworm, antheraea assamensis helfer (lepidoptera: Saturniidae)-an overview of distribution, life cycle, disease and control measure. Munis Entomol. Zool. J.16(1), 214–220 (2021). [Google Scholar]
  • 11.Tikader, A., Vijayan, K. & Saratchandra, B. Muga silkworm, antheraea assamensis (lepidoptera: Saturniidae): An overview of distribution, biology and breeding. Eur. J. Entomol.110(2), 293–300 (2013). [Google Scholar]
  • 12.Das, R. & Das, K. Effect of fungal and bacterial diseases in different instar muga silkworm, antheraea assamensis helfer (lapidoptera: Saturniidae) in different crop seasons. Munis Entomol. Zool.12(2), 578–82 (2017). [Google Scholar]
  • 13.Kakati, L. et al. Diversity and ecology of wild sericigenous insects in Nagaland, India. Trop. Ecol.50(1), 137–146 (2009). [Google Scholar]
  • 14.Luikham, R., Keisa, T. J., Bidyapati, L., Sinha, A. & Peigler, R. S. Biodiversity of sericigenous saturniidae of Manipur in India. Munis Entomol. Zool.12(2), 500–507 (2017). [Google Scholar]
  • 15.Kumar, R., Chutia, P., Ahmed, M., Rajkhowa, G. & Singh, N. Checklist of wild silk moths of North East India (lepidoptera: saturniidae, bombycidae). Munis Entomol. Zool.11(2), 508–514 (2016). [Google Scholar]
  • 16.Shangpliang, J. W. & Hajong, S. Diversity, species richness and evenness of wild silk moths collected from Khasi hills of Meghalaya, North East India. J. Entomol. Zool. Stud.3(1), 168–173 (2015). [Google Scholar]
  • 17.Devi, M. Muga Silk Industry in Assam Its Forward and Backward Linkages with the Regional Economy. PhD thesis, Utkal University (2008)
  • 18.Breiner, F. T., Guisan, A., Bergamini, A. & Nobis, M. P. Overcoming limitations of modelling rare species by using ensembles of small models. Methods Ecol. Evol.6(10), 1210–1218 (2015). [Google Scholar]
  • 19.Valavi, R., Guillera-Arroita, G., Lahoz-Monfort, J. J. & Elith, J. Predictive performance of presence-only species distribution models: A benchmark study with reproducible code. Ecol. Monogr.92(1), 01486 (2022). [Google Scholar]
  • 20.Hao, T., Elith, J., Guillera-Arroita, G. & Lahoz-Monfort, J. J. A review of evidence about use and performance of species distribution modelling ensembles like biomod. Divers. Distrib.25(5), 839–852 (2019). [Google Scholar]
  • 21.O’Neill, B. C. et al. The roads ahead: Narratives for shared socioeconomic pathways describing world futures in the 21st century. Glob. Environ. Change42, 169–180. 10.1016/j.gloenvcha.2015.01.004 (2017). [Google Scholar]
  • 22.Riahi, K. et al. The shared socioeconomic pathways and their energy, land use, and greenhouse gas emissions implications: An overview. Glob. Environ. Change42, 153–168 (2017). [Google Scholar]
  • 23.Lovato, T. et al. Cmip6 simulations with the cmcc earth system model (cmcc-esm2). J. Adv. Model. Earth Syst.14(3), 2021–002814 (2022). [Google Scholar]
  • 24.Andrews, M. B. et al. Historical simulations with hadgem3-gc3.1 for cmip6. J. Adv. Model. Earth Syst.12(6), 001995 (2020). [Google Scholar]
  • 25.Boral, D. & Moktan, S. Modelling current and future potential distribution of medicinal orchids in Darjeeling, Eastern Himalaya. Plant Eco.225, 1–14 (2024). [Google Scholar]
  • 26.Ali, F., Khan, N., Khan, A. M., Ali, K. & Abbas, F. Species distribution modelling of monotheca buxifolia (falc.) a. dc.: Present distribution and impacts of potential climate change. Heliyon9(2), e13417 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 27.Bhuyan, A. et al. Assessing current and future potential habitat of vatica lanceaefolia (roxb.) blume, a critically endangered tree species of Northeastern India. Theor. Appl. Climatol.156(2), 1–13 (2025). [Google Scholar]
  • 28.Schmitt, S., Pouteau, R., Justeau, D., De Boissieu, F. & Birnbaum, P. ssdm: An r package to predict distribution of species richness and composition based on stacked species distribution models. Methods Ecol. Evol.8(12), 1795–1803 (2017). [Google Scholar]
  • 29.Leutner, B., Horning, N. & Leutner, M. B. Package ‘rstoolbox’. R foundation for statistical computing, Version 0.1 (2017)
  • 30.De Bock, K. W., Coussement, K. & Poel, D. Ensemble classification based on generalized additive models. Comput. Stat. Data Anal.54(6), 1535–1546 (2010). [Google Scholar]
  • 31.Hastie, T. & Tibshirani, R. Exploring the nature of covariate effects in the proportional hazards model. Biometrics46, 1005–1016 (1990). [PubMed] [Google Scholar]
  • 32.Linden, A. & Yarnold, P. R. Using data mining techniques to characterize participation in observational studies. J. Eval. Clin. Pract.22(6), 839–847 (2016). [DOI] [PubMed] [Google Scholar]
  • 33.Vapnik, V. The Nature of Statistical Learning Theory (Springer, 2013)
  • 34.Müller, J. et al. Weather explains the decline and rise of insect biomass over 34 years. Nature628, 349–354 (2023). [DOI] [PubMed] [Google Scholar]
  • 35.Kalita, T. & Dutta, K. Characterisation of cocoon of different population of Antheraea assamensis (Lepidoptera: Saturniidae). Orient. Insects54(4), 574–590 (2020). [Google Scholar]
  • 36.Merrill, R. M. et al. Combined effects of climate and biotic interactions on the elevational range of a phytophagous insect. J. Anim. Ecol.77, 145–155 (2008). [DOI] [PubMed] [Google Scholar]
  • 37.Cornelissen, T. Climate change and its effects on terrestrial insects and herbivory patterns. Neotrop. Entomol.40, 155–163 (2011). [DOI] [PubMed] [Google Scholar]
  • 38.Nath, R., Nath, S. K. & Devi, D. Study and conservation of host food plants of Muga silkworm, antheraea assamensis (helfer) of Assam. Nat. Environ. Pollut. Technol.7(1), 83 (2008). [Google Scholar]
  • 39.Choudhury, S., Ghosh, A. C., Choudhury, M. & Leclercq, P. A. Essential oils of litsea monopetala (roxb.) pers a new report from India. J. Essent. Oil Res.9(6), 635–639 (1997). [Google Scholar]
  • 40.Devi, B., Chutia, M. & Bhattacharyya, N. Food plant diversity, distribution, and nutritional aspects of the endemic golden silk producing silkworm, antheraea assamensis-a review. Entomol. Exp. Appl.169(3), 237–248 (2021). [Google Scholar]
  • 41.Singh, S. et al. Conservation of muga silkworm, antheraea assamensis helfer in the natural habitats at different geographical location. Plant Archives 09725210 (2022)
  • 42.Betzholtz, P.-E., Pettersson, L. B., Ryrholm, N. & Franzén, M. With that diet, you will go far: Trait-based analysis reveals a link between rapid range expansion and a nitrogen-favoured diet. Proc. R. Soc. B280(1750), 20122305 (2013). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 43.Hill, G. M., Kawahara, A. Y., Daniels, J. C., Bateman, C. C. & Scheffers, B. R. Climate change effects on animal ecology: Butterflies and moths as a case study. Biol. Rev.96(5), 2113–2126 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 44.Baruah, J. P. Muga silkworm, antheraea assamensis helfer (Lepidoptera: Saturniidae): An overview of distribution, life cycle, disease and control measure. Munis Entomol. Zool. J16, 214–220 (2021). [Google Scholar]
  • 45.Sengupta, K. et al. Cytological Studies on Different Non-Mulberry Silkworm Species Found in Assam. Annual Report, Central Muga Eri Research Station, Titabor, Assam 25–39 (1975)
  • 46.Chutia, B. C. Wild silk moth diversity in north-eastern region of India: A potential source for novel silk. Souvenir-cum-Compendium 40 (2022)
  • 47.Kalita, T. & Dutta, K. Biodiversity of sericigenous insects in Assam and their role in employment generation. J. Entomol. Zool. Stud.2(5), 119–125 (2014). [Google Scholar]
  • 48.Sujatha, G. et al. A comprehensive review of the effect and mitigation of climate change on sericulture. Int. J. Environ. Clim. Change14(7), 776–88 (2024). [Google Scholar]
  • 49.Warson, J., Baguette, M., Stevens, V. M., Honnay, O. & De Kort, H. The impact of habitat loss on molecular signatures of coevolution between an iconic butterfly (alcon blue) and its host plant (marsh gentian). J. Hered.114(1), 22–34 (2023). [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplementary Information 1. (373.6KB, jpg)
Supplementary Information 2. (333.7KB, jpg)

Data Availability Statement

The datasets used and analysed during the current study are available from the corresponding author on reasonable request.

Articles from Scientific Reports are provided here courtesy of Nature Publishing Group

RESOURCES