Climate change is rapidly altering biodiversity at all levels, from genes to ecosystems (Scheffers et al., 2016; IPCC, 2021), with significant consequences for essential ecological services like pollination. It is expected that by 2050 climate change will lead to an average temperature increase of up to 4˚C (IPCC, 2014). Overall, climate change has been associated with modifications at different levels of biological organization, affecting the maintenance of ecological processes and the provision of ecosystem services (Houghton et al., 2001; Malhi and Wright, 2004; Malhi et al., 2008; Scheffers et al., 2016; Ferreira et al., 2021), such as pollination (e.g., Botkin et al., 2007; Loarie et al., 2008; Giannini et al., 2017, 2020; Silva et al., 2019, 2022; Sales et al., 2021; Millard et al., 2023). In response to climate change, some species move to new more suitable habitats, as a way of adjusting their climatic envelope (e.g., Urban, 2015; Chen et al., 2011; Pecl et al., 2017). In this way, climate change acts as a strong environmental driver capable of reorganizing biological assemblages, affecting many species interactions, leading to incompatible geographic distributions between interacting species (Schweiger et al., 2008; Valiente-Banuet et al., 2015; Giannini et al., 2017, 2020; Sales et al., 2021). Then, climate change can cause spatiotemporal mismatches, changes in ecological networks with the formation of new interactions, and shifts in dispersal capacity (Memmott et al., 2007; Schleuning et al., 2020). In addition, incompatibilities in plant-animal interactions are also driven by changes in phenology, such as flowering time or animal emergence and migration (Fitter and Fitter, 2002; Kharouba and Vellend, 2015; Cohen et al., 2018; Kharouba et al., 2018).

Bees are widely reported to play a key role in ecosystem functioning, contributing to the pollination of most wild plants and crops (Klein et al., 2007; Ollerton et al., 2011; Oliveira et al., 2024). However, bees have experienced globally declines in abundance and richness due to land use change, biological invasions, pesticides, and agricultural expansion (Potts et al., 2010; Dicks et al., 2016; Zattara and Aizen, 2021; Albacete et al., 2023). In addition to these stressors, climate change has also been widely reported as one of the main causes of pollinator declines across the globe (IPBES, 2016; Potts et al., 2010, 2016; Giannini et al., 2017; Lima and Marchioro, 2021; Sales et al., 2021), as observed for honeybees and bumblebees (Becher et al., 2013; Kerr et al., 2015; Soroye et al., 2020). Warmer temperatures, changes in precipitation regimes, and higher frequency of extreme weather events affect the behavior and survival of bee species (Goulson et al., 2015; Soroye et al., 2020; IPCC, 2021). In general, rising temperatures can negatively impact plant-pollinator interactions through reduced visitation rates (e.g., Rader et al., 2013; Maluf et al., 2022). For example, watermelon crops under extreme climate change scenarios incur a 14.5% decline in pollination services provided by managed bees (Rader et al., 2013). Then, plant responses to climate change are also influenced by how animals respond to these changes (Schleuning et al., 2020). Thus, disruptions in plant-animal interactions intensified by the effects of climate change can impact biodiversity and the status quo of species interactions.

As a major threat to the maintenance of natural and managed populations globally (Pacifici et al., 2017; Potts et al., 2016; Rafferty, 2017), climate change is a major potential driver of reduced agricultural yields (IPBES, 2016; BPBES/REBIPP, 2019; Giannini et al., 2017, 2020; Sales et al., 2021; Millard et al., 2023). The reduction in pollinator populations driven by climate change affects the provision of pollination services for many agricultural crops that are pollinator-dependent and economically important, such as tomato, guava, coffee and avocado (Costanza et al., 2014; IPBES, 2016; BPBES/REBIPP, 2019; Giannini et al., 2017, 2020; Sales et al., 2021), all of which could reduce the productivity of agricultural crops and negatively affect the global economy (IPBES, 2016). Since pollinators play an important role in the production of many fruits and seeds that are consumed by humans, making a large and strong contribution to global agricultural production (Giannini et al., 2015; Dicks et al., 2016; Novais et al., 2018), compromising the pollination process in pollinator-dependent crops can result in low economic yields (Garibaldi et al., 2013), which directly affects human food security (IPBES, 2016; Oliveira et al., 2023).

The total area cultivated with pollinator-dependent crops has expanded in recent decades (Aizen et al., 2019; Oliveira et al., 2023), and, consequently, the need for pollinators and pollination has also increased. Thus, it is urgent to understand how to mitigate the effects of climate change on crop pollination to ensure biodiversity persistence and safeguard food sovereignty. In Brazil, for example, 60% of all crops depend on pollinators (Giannini et al., 2015). Furthermore, Brazil ranks 5^th in its human population, is the 3^rd largest agricultural producer and exporter, and is arguably Earth’s most megadiverse country (e.g., Lopes et al., 2021; Porto et al., 2020; Oliveira et al., 2023a). However, food insecurity and hunger are still a major problem in Brazil, which returned to the hunger map in 2021, with over 60 million people facing moderate to severe food insecurity in 2020 (FAO, 2021). Understanding crop-pollinator interactions occurring in Brazil thus appears as an excellent opportunity to examine the effects of climate change on selected cases of self-incompatible or highly pollinator-dependent crops and discuss the implications for crop pollination and food security in light of the future scenarios.

In this study, we seek to understand how climate change simultaneously affects the range and spatial distribution of three economically important food plants native to Brazil and their respective main pollinators. Here we focus on Passiflora edulis (Passionfruit), Bertholletia excelsa (Brazil-nut), and Eugenia uniflora (Pitanga or Brazilian cherry) as model species to investigate the impacts of climate change on plant-pollinator interactions. These species are key in agroforestry systems, which restore ecosystems by enhancing soil productivity and supporting biodiversity while mitigating climate change (Urruth et al., 2022; Lima et al., 2023). In addition, these three food plant species are also excellent models because they are self-incompatible (i.e., rely on cross-pollination to set fruits and seeds), and essentially dependent on pollinators; in other words, yields are reduced by at least 90% in the absence of pollinators (Klein et al., 2007). We tested the hypothesis that the current spatial distribution of self-incompatible food plant species and their main pollinating bees will be highly modified under different scenarios of projected climate change, resulting in a mismatch in their new regions of occurrence, thereby threatening crop pollination services. Therefore, we expected a reconfiguration of the areas of occurrence leading to a spatial mismatch in plant-pollinator interactions.

Considering food plants, model performance experienced robust support from the occurrence data given that mean values for AUC and TSS were high at 0.84 and 0.62, respectively ( Table 1 ). The model performance for pollinating bees was also robust with 0.82 and 0.81 mean values for AUC and TSS, respectively ( Table 2 ). The ecological niche models generated showed that the variables BIO12 (annual precipitation) and elevation had the highest contribution to the current, SSP-3.70, and SSP-5.85 scenarios for all food plant and pollinator species compared to any other predictor. In addition, elevation and BIO19 (coldest quarter precipitation) explained more of the variation in species distribution under the SSP-2.45 scenario.

We observed that climate change was associated with contractions of suitable habitats for all plant and pollinator species under any climate scenario examined here. Comparing the current and any future scenarios, climate change was associated with a reduction in suitable habitat areas for all three food plant species. Specifically, suitable habitats may be reduced by up to 52.5% for Passiflora edulis, 70% for Bertholletia excelsa, and 54.5% for Eugenia uniflora ( Figure 1 ). We documented that reductions in suitable habitat areas were greatest under the SSP-5.85 scenario for P. edulis and B. excelsa ( Figure 1 ), and the SSP-3.70 scenario for E. uniflora ( Figure 1 ). Considering the pollinators, climate change also marked reduced suitable habitat areas for all three pollinator species. Overall, habitats could be reduced by up to 39.6% for Xylocopa frontalis, 51.4% for Eulaema mocsaryi, and 44.1% for Apis mellifera ( Figure 2 ). The highest reductions were under the SSP-2.45 scenario for X. frontalis and E. mocsaryi and the SSP-3.70 scenario for A. mellifera ( Figure 2 ).

Overall, suitable habitats for P. edulis will likely be across the northern and southern regions of Brazil within the Atlantic forest. For B. excelsa, relictual suitable habitats contracted from the seasonally-dry peripheral to the core portions of the species distribution within the Amazon. For E. uniflora, we observed that suitable habitats also moved towards the central-southern portions of the species range, with species distributions within the Atlantic forest. Regarding pollinators, suitable habitats for X. frontalis moved to the northeastern and southeastern range limits, with marked losses in suitable habitats central-western Brazil, whereas suitable habitats for E. mocsaryi shifted to northwestern Brazil. For honeybees, remaining habitat patches shifted to southeastern and southern Brazil, following marked loss of suitable habitats in the semi-arid northeast.

Overlaying species distribution maps, we observed that climate change is clearly likely to induce marked niche mismatches between food plants and their pollinators. Specifically, suitable habitats for the co-occurrence of plant-pollinator interactions could be reduced by up to 51.5% for P. edulis and X. frontalis, 76% for B. excelsa and E. mocsaryi, and 54% for E. uniflora and A. mellifera ( Figure 3 ). The niche mismatches between the food plants and their pollinating bees will impact the plant-pollinator interactions and, consequently, plant reproduction. These reductions in co-occurrence were most exacerbated under the SSP-2.45 scenario for P. edulis and X. frontalis, and for E. uniflora and A. mellifera. The worst predicted co-occurrence forecast for Brazil-nut trees and their orchid bee pollinators was under the SSP-5.85 scenario ( Figure 3 ). Our findings highlight not only how climate change can impact ecological networks by disrupting plant-pollinator interactions, but also their potential to pose threats to food security by reducing pollination and, consequently, the productivity of pollinator-dependent food plants that essentially rely on pollinators to set fruits and seeds.

Our results clearly indicate that climate change will likely reduce suitable habitats and alter the spatial distribution for pollinator-dependent food plants native to Brazil and their primary bee pollinators. Our findings are evidenced by the pattern of contractions in occurrence areas, which were most severe under some climate change scenarios, with divergent responses among the species examined here. Furthermore, our findings show that climate change can act as a potential driver of decoupling between each crop and its main pollinating bee, leading to co-occurrence mismatches. Reductions in suitable habitat areas were documented for all food plants and their pollinators under all climate change scenarios, indicating that climate change can potentially affect the occurrence and co-occurrence of plants and their pollinators, which corroborates our hypotheses regardless of the scenario we used. Therefore, shifts in habitat suitability induced by climate change can result in severe spatial mismatches, potentially exacerbating the risks of pollination failure.

Our results reinforce now widely observed and predicted patterns of climate-induced shifts species ranges in terms of suitability, some of which involving potential impacts on pollination (e.g., Giannini et al., 2013, 2017; Settele et al., 2016; Elias et al., 2017; Millard et al., 2023). The effects of climate change have been linked to significant reductions in pollination services (Settele et al., 2016; Sales et al., 2021; Millard et al., 2023). This poses intractable challenges to food production and food security as biotic pollination is a major ecosystem service ensuring crop yields in a large number of food plants (Klein et al., 2007; Giannini et al., 2015; Oliveira et al., 2024). From this perspective, our results are consistent with previously documented effects of climate change on agricultural crops and pollinators in Brazil. Specifically, projected climate change could reduce suitable areas for the occurrence of bees, which pollinate some crops, including tomato, persimmon, and passionfruit (e.g., Giannini et al., 2013, 2017; Elias et al., 2017). In addition, the interaction between agricultural land use and climate change is associated with population declines in insect pollinators, and these impacts are more severe in tropical regions (Millard et al., 2023). Watermelon production is also expected to decline due to climate change when crops are pollinated by managed honeybees, but the opposite was shown for native bee pollinators, reinforcing the importance of the native biota in mitigating climate change (Rader et al., 2013). Some species could therefore respond positively to climate change. In contrast to this study, it has been suggested that Brazil-nut trees could expand their distribution area by up to 6%, depending on the climate change scenario (Sales et al., 2020). However, when interactions with pollinators were examined, declines in pollinator diversity may still lead to a spatial mismatch and threaten the pollination of Brazil-nut tree stands (Sales et al., 2020). Regardless of the climate scenario, the distribution of wild Brazil-nut tree populations across the Brazilian, Peruvian and the Bolivian Amazon is likely to be constrained, not least because of mounting threats from widespread forest degradation through logging and wildfires (C.A. Peres, unpubl. data).

Climate change could be a strong environmental driver altering ecosystem functioning (Malhi and Wright, 2004), often leading to an ecological imbalance that favors generalist species over specialists (e.g., Burkle et al., 2013; Silva et al., 2019). The latter, in turn, have narrower ecological niches and become more susceptible to suitable habitat contraction or undergo local extinction (Blois et al., 2013; Silva et al., 2019). We observed a marked decline in suitable habitats. For Passiflora edulis and Eugenia uniflora, we observed a range contraction in the Atlantic forest region, whereas Bertholletia excelsa succumbs to substantial habitat loss across the Amazon. The Atlantic forest is one of the most susceptible biodiversity hotspots to climate change (Béllard et al., 2014), which is intensified by deforestation in this biome (Scarano and Ceotto, 2015). The Atlantic forest is expected to face pervasive biotic reorganization followed by a savannization process (Joly et al., 2014; Scarano and Ceotto, 2015), particularly in the Brazilian northeast where aridity is a major issue, which could explain the patterns we find for P. edulis and E. uniflora, but also for their pollinators (Xylocopa frontalis and Apis mellifera). Additionally, the risk of pollinator shortages for native crops has been documented in some counties in northern Brazil, although this is mainly concentrated in the northeast and the southeast, with the Atlantic forest and the Caatinga dry forest at higher risk of pollinator shortages (Oliveira et al., 2024). Considering Brazil-nut trees, reduced precipitation and increasing frequency of drought are expected in the eastern and southeast Amazon, the regions most affected by deforestation (Malhi et al., 2008). This is aggravated given that Brazil-nut tree populations have been widely affected by deforestation, habitat fragmentation, wildfires, and overexploitation of their seed crops (Peres et al., 2003; Brandão et al., 2021). Therefore, deforestation is a major factor boosting the effects of climate change on pollinators, which is reinforced by climate projections in the eastern Amazon indicating that 95% of all bee species will face declines in their area of occurrence, affecting crop pollination (Giannini et al., 2020).

Our projections indicate that climate change is a potential driver of niche mismatch between food plants and their pollinators. Brazil is one of the countries at highest risk of crop yield losses due to the impacts of climate change on the abundance of insect pollinators (Millard et al., 2023). The contraction in areas of biotic overlap may therefore represent a severe threat to Brazilian agriculture, given that the tropics will likely experience the highest risks of pollinator loss, which threatens the production of key pollinator-dependent crops, such as cocoa, mango, watermelon, passionfruit, tomato, and coffee (Giannini et al., 2013, 2017; Elias et al., 2017; Millard et al., 2023). As such, our worst-case scenario highlights the severe impacts on ecosystem functioning and food production, with downstream socioeconomic implications. Such negative effects of climate change on the pollination of food plants affect not only biodiversity but also subsistence and cash-crop yields, the national economy, and the livelihoods of local communities. The decline in local food production tends to increase the value of national trade, which may benefit the national economy, but reinforces inequality and rural poverty by eroding per capita purchase power (Cornelsen et al., 2015; Murphy et al., 2022). These effects will be even more intense in Brazil, which is a leading producer and exporter of tropical fruit crops (Oliveira et al., 2023) and one of the countries most vulnerable to climate change (Millard et al., 2023). A decline in pollination interactions induced by climate change heightens food insecurity (Millard et al., 2023). Furthermore, the reduction in crop diversity driven by the expansion of monoculture lands is a key driver of pollinator declines, which in turn could threaten the resilience of ecosystems and agricultural yields (e.g., Aizen et al., 2019; Oliveira et al., 2023). Overall, the combined impacts of land-use and climate change reduce long-term agricultural sustainability by increasing the likelihood of disrupted plant-pollinator interactions, which can ultimately lead to food shortages. Accordingly, the loss of pollinators will increase trade risks for major food importers such as China, western Europe and the USA, as the lack of pollinators will disrupt agricultural yields of pollinator-dependent cash crops, such as coffee and cocoa (Millard et al., 2023).

Therefore, decision-makers need to focus on developing public policies that prioritize climate change mitigation and encourage sustainable agricultural practices to protect key mutualistic interactions that benefit human societies, such as pollination. Resilient ecosystems not only benefit plants and their pollinators but are also vital in ensuring agricultural productivity and food security (IPBES, 2016). Thus, we need to rethink our agricultural practices, implementing biodiversity conservation as a crucial pillar to mitigate the ecological challenges imposed by climate change, while also ensuring resilience and food sovereignty. To achieve this, policy makers should urgently include the remaining areas where key mutualists are expected to co-occur into action plans for conservation, while maintaining species abundance and richness, and building more sustainable agricultural production pathways (Altieri and Nicholls, 2008, 2017; Garibaldi et al., 2016; Nicholls and Altieri, 2018; Oliveira et al., 2023). Furthermore, sustainable agriculture practices should be adaptable to a variety of production methods and conditions, and consider the entire agricultural system in their design (Muhie, 2022 and references therein). These practices can include climate-smart agriculture (a relatively new approach that ensures improved agricultural practices while lowering greenhouse gas emissions to assist underprivileged people in increasing agricultural productivity and income), organic farming (which promotes resource sustainability, environmental protection, animal welfare, food quality and safety, social justice, and market-driven payments for internalized costs), sustainable intensification (which increases crop yields without threatening the environment or converting natural areas into farmland), regenerative agriculture (a food and agricultural system focused on rehabilitation and preservation), and integrated nutrient management (a technique for safely disposing of agricultural wastes by incorporating inorganic and organic fertilizers in a balanced and integrated manner to preserve soil fertility and give plants the right amount of nutrients). These practices can achieve the goals of life on land and zero-hunger of the Sustainable Development Goals, especially in major emergent agricultural economies such as Brazil.