Discussion
Climate change is considered one of the most critical threats to biodiversity, yet literature documenting climate change impact on tropical insects remains scarce (Taheri et al., 2021). Given that tropics contain enormous insect diversity, understanding climate vulnerability to tropical species is essential. Here, comparing the suitability distribution for 242 Bangladeshi butterfly species, I revealed that climate warming could severely impact many butterfly species. While the suitability might remain similar in some parts of the country (e.g., north-east), the future suitability could expand (e.g., towards north-east) and contract (e.g., from the central) in other parts. Although the direction of the shift in climatic suitability could be multi-dimensional, most species might shift towards the north. This finding is similar to VanDerWal et al. (2013), who analysed the distribution changes of 464 Australian bird species and found that both temperature and precipitation influence changes in species’ distributions, such that the resulting patterns of range shift are complex. In addition, many Bangladeshi butterflies could shift towards the north-east and south-east direction, mostly mountainous and forested habitats, whereas the remaining parts of the country mainly consist of grasslands (Chowdhury et al., 2021b). This finding supports that i) there could be a decrease in grassland species and an increase in forest species (Kwon et al., 2021) and that ii) elevational shifts are more pronounced than poleward shifts for tropical butterflies (Colwell et al., 2008; Chen et al., 2009).
Due to climate warming, the suitable habitat could primarily shift for all the Bangladeshi butterfly species (except for Spindasis elima ); however, there could be a striking difference between the threatened and non-threatened species. While the suitable habitat could increase for 58% of species and decrease for 42%, most of the threatened species could experience range contraction. The centroid shift for the threatened species (55km2) could be lower than the non-threatened species (77km2); therefore, threatened butterfly species could be more vulnerable to climate change if they cannot adapt to the changing environment. This finding is similar to Mattila et al. (2011), where authors analysed the distribution shift among the threatened and non-threatened Finnish butterflies. Although the range of Finnish butterflies shifted substantially, threatened species moved very little (Mattila et al., 2011). Although Finnish threatened butterflies did not show any consistent direction of the shift (Mattila et al., 2011), I obtained contrasting result that most threatened Bangladeshi butterflies could shift toward a south-south-west direction. Additionally, the mean elevation of Bangladeshi butterflies could increase by about 2.4 times in the future climatic condition, which could be more extreme for the threatened butterflies. To safeguard the existential status of Bangladeshi butterflies, especially threatened species, the conservation managers, planners, and policymakers should take this into account in their efforts.
Climate warming will lead many species to move towards higher elevations to track suitable climatic niches (Walther et al. 2002). Nevertheless, the ability to track climatic niche differ markedly depending on the migratory behaviour of the species (Warren et al., 2001; Hill et al., 2002). By analysing 46 British butterflies, Warren et al. (2001) showed that the distribution range expanded for half of the mobile and generalist butterfly species, whereas the other generalists and 89% of the habitat specialists declined in distribution size. In another study, Chowdhury et al. (2021a) showed that while expanding its geographic range, the migratory tawny coster butterfly maintains its native climatic niche. Here I found that the amount of suitable habitat for the migratory species could increase significantly and that migratory butterflies could experience a more pronounced shift in suitability in the future climatic conditions comparing to the non-migrants. This result supports the hypothesis that migratory species perform better in tracking climatic niches (Warren et al., 2001).
The existing biodiversity is facing an existential crisis due to several anthropogenic stressors; however, none is as pervasive as climate change (Halsch et al., 2021). The current climate change has already impacted 27 million km2 (18.3% of land), which took place in all biomes (Elsen et al., 2021). Climate change has negatively impacted the distribution of 47% of 873 terrestrial non-volant threatened mammal species and 23.4% of 1.272 threatened birds species (Pacifici et al., 2017). Here, I showed that the future climate could be unsuitable for 2-30% of Bangladeshi butterfly species depending on the climate change scenarios. If the climate changes at a similar rate – as it is now, many butterfly species will go extinct. Wallace’s understanding of tropical climate as “genial” dismisses the view that tropical climates can be vulnerable to climate change (Colwell et al., 2008). While safeguarding tropical insects, several steps need to be taken urgently.
Our current understanding of how climate change impacts tropical insects is limited: Detailed studies are required to assess climate change vulnerability (Basset & Lamarre, 2019; Janzen & Hallwachs, 2019; Montgomery et al., 2020; Wilson and Fox, 2021) — identifying species-level details on how climate change impacts physiology, life history and evolutionary traits; diversity, abundance, and distribution. Understanding the key drivers of species sensitivity and how it varies among different groups (e.g., threatened and non-threatened; migratory and non-migratory) is essential, where natural history collections can provide important historical information on climate change impact (Scheper et al., 2014; Kharouba et al., 2019; but see Ries et al., 2019). Further, developing decisive conservation strategies for tropical insects are needed, where spatial conservation prioritisation approaches can be beneficial and identifying factors limiting species’ capabilities to colonise areas beyond their historical limits is vital (Lewthwaite et al., 2018). Developing efficient conservation strategies requires long-term biodiversity data, yet such data is vastly unavailable from most tropics. To solve the biodiversity data gap, citizen science activities can be highly beneficial. A two-week citizen science project, for instance, could generate comparable spatial coverage for social wasp species as four decades of recording by expert amateurs (Sumner et al., 2019; Wilson and Fox, 2020). Assessing climate change’s impact on Bangladeshi butterflies was only possible because of utilising the citizen science data, given a vast proportion of the collated species distribution records were from Facebook. Scientists can engage the wider community to document insect distribution data from the tropics; however, developing standardised monitoring protocol with detailed guidelines before organising such surveys.
The climatic niche evolution rate of species is slower compared to the climate change rate (Quintero & Wiens, 2013), and failure to track climatic conditions may lead to range contractions and/or local extinctions (Moritz & Agudo, 2013; Wiens, 2016; He et al., 2018). Besides, climate change is not just a future threat but impacts many species in the current time (McCarty et al., 2001). Using the niche-overlapping analysis, I showed that about 54% of the climatic niche of Bangladeshi butterflies could shift in the future. Immediately detecting climate change response is vital to improve (or initiate) management strategies, counter species loss, and mitigate biodiversity impacts. Here, specific management strategies should include prioritising species-specific requirements. For example, butterflies mostly are tied to specific habitat types, and therefore, to survive, they will require coordinated movement of climatic conditions habitat types (including vegetation structure) to track conditions spatially (Peterson et al., 2004). Further, the adverse effects of climate warming on pollinators can exacerbate by homogenous and fragmented landscapes, limiting range shift opportunities and reducing micro-climatic buffering (Vasiliev and Greenwood, 2021), which emphasises the importance of considering landscape connectivity and habitat heterogeneity in conservation planning (Vasiliev and Greenwood, 2021). Similarly, increasing the number of floral resources could attract diverse insect groups, creating connectivity between forest patches could support range expansion, and protected habitat over elevational and vegetational gradients could help range-restricted species.
Protected area (PA) coverage in the tropics is relatively low. For example, only 4.6% of terrestrial land in Bangladesh is now protected, and most are too small to support viable biota (Chowdhury et al., 2021f). Increasing PA coverage could insulate species from human-induced climate change and meet the Post 2020 biodiversity framework targets, aiming to achieve 30% PA coverage by 2030 (Convention on Biological Diversity, 2020). When proposing new PAs, scientists need to include connectivity planning emphasising climate change impact, ensuring landscape permeability, creating the network of habitats, future habitat conditions, and prioritising intact habitats. In tropical mega-populated countries like Bangladesh, many people live inside or surrounding PAs; therefore, scientists need to engage local people in conservation.
This is the first study highlighting how climate change could impact the suitability distribution of South Asian butterflies; however, these results need to be interpreted cautiously. For example, the spatial distribution data of Bangladeshi butterflies are highly biased and are mostly distributed in the major cities (Chowdhury et al., 2021b, f), which might impact findings. However, I followed two approaches to handle sampling bias: i) spatial thinning process, where I filtered single occurrence record from each raster pixel (Aiello-Lammens et al., 2015), and ii) used the ‘checkerboard2’ evaluation method, which handles overinflation of model performance, at least that portion resulting from biased sampling (Muscarella et al., 2014; Chowdhury et al., 2021e).