https://orcid.org/0000-0001-8345-7986
Figures
Abstract
Habitat loss and degradation, mainly driven by agricultural expansion and intensification, alter ecological processes, ecosystem services and human well-being at a global scale. Pollinator populations in degraded agricultural areas tend to collapse, which in turn can reduce the effective pollination of agricultural crops and food production, particularly in leading food producing countries, such as Brazil. We sought to understand how the vulnerability to pollinator failure and the economic value of pollination (EVP) are associated with farmland size, farmer income, habitat loss, and the size of the human footprint. We also examine socioeconomic predictors related to income inequality levels across Brazilian municipalities and phytogeographic domains. We show that 58% of all leading crops in Brazil are dependent to some degree on animal pollination, and agricultural production in 96.8% of all municipalities is at least partly vulnerable to pollinator failure. Soybean, coffee, cotton, tomato and cocoa accounted for 84% of the total economic valuation of cropland pollination in Brazil, with soybean alone representing more than half of this value. Including soybean, vulnerability and EVP were positively correlated with the Human Development Index (HDI), Gross Domestic Product (GDP), habitat loss, and farmland size. On the other hand, these key indicators were negatively associated with the human footprint and the proportion of low-income farmers. Excluding soybean, however, vulnerability increased across low-income municipalities, particularly across the Atlantic Forest, Cerrado, and Caatinga, revealing strong biome contrasts. Overall, EVP was consistently reduced in areas under heavier human footprint, suggesting that anthropogenic pressures reduce pollination benefits. Our results also show that soybean cultivation masks underlying social and regional disparities in agricultural vulnerability. Thus, the evaluation of pollination risks can be masked by commodity production. Enhancing ecological processes and promoting diversified, pollinator-friendly agriculture may strengthen food production, particularly for tropical low-income farmers in vulnerable regions, while contributing to ecosystem service provision and human well-being.
Citation: Oliveira W, Porto RG, Cruz-Neto O, Tabarelli M, Viana BF, Peres CA, et al. (2025) Yield vulnerability of low-income smallholders to pollinator declines in Brazil is biome-dependent. PLoS One 20(11): e0337328. https://doi.org/10.1371/journal.pone.0337328
Editor: Nickson Erick Otieno, National Museums of Kenya Ornithology Section, KENYA
Received: August 28, 2025; Accepted: November 6, 2025; Published: November 25, 2025
Copyright: © 2025 Oliveira et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: All relevant data are within the paper and its Supporting Information files.
Funding: The author(s) received no specific funding for this work.
Competing interests: The authors have declared that no competing interests exist.
1. Introduction
Agricultural expansion is one of the main drivers of land degradation [1,2], leading to the loss of natural habitats and biodiversity, consequently disrupting ecosystem dynamics and decreasing their capacity to maintain ecological functions and services [3,4], mainly in tropical regions (e.g., [1]). In tropical developing countries, biodiversity loss and degraded ecosystem functions can be inextricably intertwined with social inequality [5]. This is particularly concerning in Brazil, as more than half of the population (116.8 million people) are under food insecurity or hunger (e.g., [4,6,7]), even though the country is one of the leading food producers [6,8]. In Brazil, as in many other tropical countries, most of the food production for domestic consumption comes from smallholders and family farming [9]. In total, 76.8% of all Brazilian farms belong to family units [10], indicating that decreases in agricultural production, due to pollinator loss, are expected to affect the entire supply chain, from family farmers to large producers, particularly the poor and the rural population [11]. A wide variety of crops cultivated in the tropics depend on pollinators [12,13]. In Latin America, smallholder farmers are responsible for producing more than half of the food consumed by humans [14,15]. Yet, because of the relentless agricultural expansion in the tropics, smallholders are now more vulnerable to declines in animal pollinators, which can aggravate food insecurity [16]. Brazilian agriculture, spanning from smallholders to large-holders, is heavily reliant on biotic pollinators [13,17].
Animal pollination is crucial both as an ecosystem function and as an ecosystem service for people, providing essential inputs to crop yields worldwide [18]. Whether they are cultivated or wild, approximately 90% of all flowering plant species [19] are pollinated by a wide diversity of pollinators, including bees, other insects (e.g., moths, butterflies, wasps, and beetles), bats, birds, and other vertebrates (e.g., [13,19–21]). Natural pollinators ensure or enhance the fruit set, yield, fruit quantity, and quality of over 75% of all global food crops [12,18,20]. Consequently, pollination service provides essential micro- and macronutrients to human health, especially lipids, and vitamins [22–24]. In this context, pollination emerges as a nature contribution to people, as it generates positive benefits for humans through nature [25,26]. Therefore, maintaining the delivery of ecosystem goods has been recognized as imperative for the Sustainable Development Goals [27].
Nevertheless, despite the critical importance of pollination to human well-being, this vital ecosystem service has been severely impacted by human activities on natural ecosystems. Cropland expansion is one of the main drivers of pollinator decline and is still a common strategy to increase crop production, especially in developing countries [28]. Notwithstanding, in both the developed and developing worlds, the area allocated to pollinator-dependent crops has increased faster than that of non-dependent crops [4,29,30]. However, such a trend is largely associated with growing areas of large-scale mechanized monoculture (such as soybean) and, consequently, greater isolation from natural habitats [4,30]. The demand for animal pollination services is rising while pollinator abundance and diversity have been declining [21,29]. In Brazil, for example, a scenario of a pollination crisis could reduce the production of food derived from pollinator-dependent crops by 13.5% – 41.6%, resulting in monetary losses of US$4.86–14.56 billion per year [11]. Losses in pollination services can induce simultaneous degradation of several other ecosystem services [31], and, therefore, cause interlinked poverty spirals and major risks for humankind [32].
As an essential asset in agricultural systems, decreases in pollination services can reduce cropland productivity [30] and economic gains [33,34]. Annually, the economic global input of pollinators to agriculture, inflation-adjusted for 2020, conservatively ranges from US$195 billion to US$387 billion [34], but this value is likely underestimated as economic valuations fail to represent the complex sets of pollinator benefits, including the value of locally grown crops for trade and subsistence, especially in tropical countries [34]. Studies addressing the economic contribution of pollinators overlook the interdependence between biotic pollination and family farming, smallholders, and low-income farmers (but see [16,35–37]. Poverty research does not consider the impacts of pollinator losses and research on the impact of biodiversity loss on poverty rarely reflects pollinators as an essential part of biodiversity [16].
More detailed research on the relationship between pollinator loss and various agricultural systems, considering the spectrum of agricultural income, can provide stronger policy arguments to advocate for pollinator protection [21]. Here, we aimed to explore the complex interactions between environmental, agricultural, and socioeconomic factors, while developing a nationally representative assessment of pollination contribution to agriculture, accounting for rural and socioeconomic diversity. Specifically, we evaluate how the economic value of pollination (EVP) and the vulnerability to pollinator failure correlate with socioeconomic predictors [Gross Domestic Product (GDP), Human Development Index (HDI) and Gini coefficient] and agricultural profile (farm size and farmers income), and how these variables relate to the degree of degradation of administrative jurisdictions. We tested the hypotheses that socioeconomic inequality indicators and agricultural profile influence municipal vulnerability to pollination service loss, specifically, municipal counties with smaller farm sizes, with higher proportion of low-income farmers, with lower HDI and higher Gini coefficient are more vulnerable to losses of pollination services. Additionally, we evaluated whether municipalities exposed to higher EVP tend to have higher GDP and HDI.
2. Materials and methods
2.1. Agricultural production
We selected the 60 most important agricultural crops in Brazil in terms of total production volume in 2019 [10]. This recent annual harvest (2019) is representative of the country’s typical average agricultural production and the relative contribution of different crops to the overall nutrient supply. These crops comprise almost the entire Brazilian agricultural production. The cultivar list includes Brazilian staple foods (such as rice, beans, and oranges), crops used in livestock feed that result in other staple foods (e.g., eggs, meat and dairy products), and important export commodities (e.g., soybean, sugar cane, cotton, maize, and wheat). We obtained the production value of all crops grown in each of 5,499 Brazilian municipalities in 2019 on the basis of data from the Brazilian Institute of Geography and Statistics [10].
2.2. Socioeconomic indicators, farm size, and landholder income
To account for the municipality-scale level of development, we used the human development index (HDI), a widely used indicator that encompasses three key dimensions of human development: living standard, knowledge, and health. To assess the economic aspect, we obtained the GDP of each municipality, and to measure the municipal inequality, we obtained the Gini coefficient of all Brazilian municipalities. All indicators are available at IBGE [10].
To determine a municipal index of farmers’ income, we apply an opportunity cost methodology to classify types of family farmers [38]. Family farming income is categorized into types A, B, C, and D, where type A has the highest income level, while D represents low-income farmers. We then established the municipal index of low-income farmers from the proportion of farmers of type D for each county, available from IBGE [10]. To determine an index of farmland size by municipality, an average of farm size was calculated from the number and total area of rural properties across all Brazilian municipalities, available from IBGE [10].
2.3. Financial efficiency of crops
We calculated the financial efficiency of all crops by yield in relation to the total production area, from municipal production data available at IBGE [39]. We then related the financial efficiency of all crops to the dependence of each crop on biotic pollinators.
2.4. Land degradation
We accessed data on the cumulative human pressure on the environment as of 2009 at a spatial resolution of ~1 km, for the entire Brazilian territory. This measure of human footprint is based on eight variables, including built-up environments, human population density, electric power infrastructure, croplands, pasture lands, roads, railways, and navigable waterways. The data set was produced by Venter et al. [40] and is available in the Mollweide projection (https://doi.org/10.7927/H46T0JQ4). For data on primary habitat loss, we used Brazil’s annual land use and land cover maps over 36 years (1985–2020) at a 30-m resolution, available from the MapBiomas project (https://mapbiomas.org).
2.5. Dependence on pollinators and Economic Value of Pollination (EVP)
Considering that most food crops rely on biotic pollination to a varying extent, we applied the currently most widely accepted standard of relative dependence on pollinators in world agriculture [18,41]. Based on the degree to which crop production and quality in fruits and seeds are associated with animal pollination, crop types examined here were assigned to one of five categories of pollinator dependence (D), adapted from Klein et al. [18] and Gallai et al. [33]: (1) essential (D = 0.95); production is reduced by 90% or more when pollination services are lacking; (2) great (D = 0.65): reduction of 40% to 90%; (3) modest (D = 0.25): reduction of 10% to 40%; (4) little (D = 0.05): reduction of 0% to 10%; (5) no differences in yields (D = 0) under conditions with and without animal-mediated pollination.
Our database consisted of the production value of leading crops and their respective pollination dependence. We thus calculated, using a bioeconomic approach, the economic contribution of pollination (i.e., the economic value of pollination; EVP) across all municipalities, as well as their overall vulnerability to pollinator declines [33,42]. We adapted the equation proposed by Gallai et al. [33], which estimates the total production value of insect pollination (IPEV), with the variables crop i, where i ∈ [1; I], in each world region x, where x ∈ [1; X], quantity produced (Qix) and the dependence ratio of crop i on insect pollinators (Di) and the price of crop i per unit produced in region x (Pix):
Instead of using the price of each crop and quantity produced (Pi × Qi), we substituted the production value (PVix) available in the IBGE database, which represents the production price of crop (i) paid to producers at each municipality (x), which we consider, in this case, a more accurate approach:
2.6. Vulnerability to pollinator declines
Based on Gallai et al. [33], vulnerability is defined as the potential production loss attributable to a lack of pollinators, which is calculated as the ratio of EVP to the economic production value of an area (x) and can be stated as:
2.7. Statistical analysis
In this study, municipal counties were adopted as the unit of analysis, totaling 5,499 records. Indicators of crop yield vulnerability to pollinator decline and the economic value of pollination (EVP) across Brazilian municipalities were examined via generalized linear mixed models (GLMMs). Vulnerability to loss of biotic pollination was modeled using a zero-inflated beta regression with logit link (glmmTMB package), while EVP was modeled using a Tweedie regression with a log link [43]. Standardized predictors (z-scores) are as follows: farm size (log-transformed), Gini coefficient, Human Development Index (HDI), GDP (log-transformed), the human footprint index, habitat loss (log-transformed), proportion of low-income farmers, and the interaction between low-income farmers and biome. Biome-level differences were evaluated by extracting estimated marginal means with Tukey-adjusted pairwise comparisons for vulnerability (emmeans and multcomp packages) and by comparing total EVP values across 2000 bootstrap resamplings, from which pairwise differences and p-values were obtained. Multicollinearity among predictors was tested using variance inflation factors (VIF, car package), with all values < 3, indicating no major collinearity problems. Subsequently, in order to evaluate fit-quality of selected models, we also assessed model residuals using the DHARMa package, which applies a simulation-based approach to generate scaled (quantile) residuals for fitted generalized mixed models [44]. This diagnostic framework allows for the detection of typical model misspecifications (e.g., deviations from residual uniformity, overdispersion, and residual autocorrelation). To account for potential spatial autocorrelation, Moran’s Eigenvector Maps (MEMs) were incorporated into the models [45]. The MEMs are spatial predictors that maximize spatial autocorrelation measured by Moran’s coefficient. Spatial autocorrelation of residuals was further evaluated using Moran’s I index. After adding MEMs, the initially high Moran’s I values slightly dropped, suggesting that spatial structure was partially controlled (S1 Table). All statistical analyses were performed using R v. 3.5.3 [46].
3. Results
3.1. Economic value of pollination and vulnerability at the national level
Our results indicate that 35 (58%) of the leading agricultural crops across Brazil are dependent on animal pollination to some degree (Table 1). The spatial patterns of agricultural economic value and the economic value of pollination (Fig 1 and 2) reflected the widespread contribution of pollinators across the country. Agricultural production value, including soybean monoculture (Fig 1A), was spatially concentrated in the Central-West and Southern regions, reflecting the dominance of soybean cultivation in Brazil. When soybean croplands were excluded (Fig 1B), the spatial pattern changed, revealing an even distribution of production values across the Northeast, Southeast, and Northern portions of Brazil. A similar pattern was observed for the economic value of pollination. The inclusion of soybean (Fig 2A) emphasized regions under intensive grain production, particularly the Central-West. By excluding soybean (Fig 2B), there was a strong dependence on diversified agriculture in the Northeast, Southeast, and parts of the Northern region; i.e., soybean thus masked regional vulnerabilities and the economic importance of pollinator-dependent crops at the national scale. Overall, agricultural production in 96.8% of all municipalities was at least partly vulnerable to the loss of pollinators (Fig 3). Considering the agricultural vulnerability to pollinator loss, we observed that, including soybean (Fig 3A), the highest vulnerability was again concentrated in the Central-West and Southern Brazil. Excluding soybean (Fig 3B), however, vulnerability was more evident along the Atlantic coast and in the Northeast.
(A) Considering 60 leading crops; and (B) Considering 59 leading crops (excluding soybean). Source: IBGE 2021.
(A) Considering 60 leading crops; (B) Considering 59 leading crops (excluding soybean). Source: IBGE 2021.
(A) Considering 60 leading crops; (B) Considering 59 leading crops (excluding soybean). Source: IBGE 2021.
The modestly dependent soybean yield (35% of all agricultural production) had the highest EVP, which represented up to US$7.8 billion, followed by coffee (US$1.1 billion, modestly dependent), cotton (~US$1 billion, modestly dependent), tomato (US$0.9 billion, greatly dependent) and cocoa (~US$0.6 billion, essentially dependent) (Table 1). The total economic value of pollination was ~ US$13.8 billion, more than half of which was accounted for by soybean cultivation. This translates into a contrasting economic value of pollination when soybean was either included or excluded from the analysis. With soybean (Table 2), EVP was positively associated with Human Development Index (HDI; Table 2, Fig 4A), Gross Domestic Product (GDP; Table 2, Fig 4B), Habitat loss (Table 2, Fig 4C), and farmland size (Table 2, Fig 4D). In contrast, EVP was negatively related to the human footprint (Table 2, Fig 4E) and proportion of low-income farmers (Table 2, Fig 4F). By excluding soybean (Table 2), patterns shifted substantially. Specifically, HDI and farmland size became negatively related to EVP (Table 2, Fig 5A and B), while GDP (Table 2, Fig 5C) and habitat loss (Table 2, Fig 5D) remained positively associated with EVP. The human footprint index and low-income farmers continued to exhibit a negative association with EVP (Table 2, Fig 5E and F).
Red shading indicates 95% confidence intervals.
Red shading indicates 95% confidence intervals.
The average vulnerability of Brazilian municipalities was as high as 0.138, indicating that ~14% of the overall agricultural production may be reduced by the loss of pollinators. When soybean was included, vulnerability was positively associated with income inequality (Gini coefficient; Table 3, Fig 6A), HDI (Table 3, Fig 6B), GDP (Table 3, Fig 6C), and rate of habitat loss (Table 3, Fig 6D), while the human footprint (Table 3, Fig 6E) and proportion of low-income farmers (Table 3, Fig 6F) were negatively related to vulnerability. When soybean was excluded, HDI, the human footprint, proportion of low-income farmers, and farm size were negatively associated with vulnerability (Table 3, Figs 7A, D, E and F) while GDP and habitat loss remained positively related (Table 3, Figs 7B and C).
Red shading indicates 95% confidence intervals.
Red shading indicates 95% confidence intervals.
3.2. Economic value of pollination and vulnerability at the biome scale
Contrasting trends in EVP and vulnerability across biomes emerged, regardless of whether or not soybean cultivation was included. Including soybean, most biomes exhibited negative correlations between poverty and EVP, especially the Amazon, the Atlantic Forest, and the Pantanal wetlands, where the incidence of low-income farmers increased and EVP sharply decreased (S1 Fig). On the other hand, the Caatinga and the Pampa exhibited positive correlations, suggesting that the relative importance of pollinators rises for low-income counties in these biomes. For example, in the Caatinga, greater vulnerability under socioeconomic constraints resulted from the predominance of pollinator-dependent crops (S1 Fig). When soybean was included in the production system, the Cerrado’s response was almost flat, indicating that the income structure had less impact on EVP (S1 Fig). When soybean was excluded, a more heterogeneous pattern emerged across biomes. Although the slopes were even steeper, the Amazon and the Atlantic Forest still showed noticeable declines in EVP as poverty increased (S2 Fig). Previously exhibiting minimal fluctuation, the Cerrado now had a clear decreasing trend, consistent with the trends observed in other biomes (S2 Fig). In contrast to the model including soybean, the Pantanal showed a significantly flatter relationship, whereas the Caatinga and Pampa maintained positive relationships (S2 Fig). These changes resulted from the fact that soybean is a partially pollinator-dependent crop and it is highly concentrated in particular regions.
The interaction between low-income farmers and biomes indicated that when soybean was included, most biomes exhibited gentle slopes, with only the Pampa and Caatinga showing stronger increases in vulnerability as the proportion of low-income farmers increased (S1 Fig). In contrast, when soybean was excluded, the biome-specific responses became much clearer. The Amazon and Pantanal exhibited decreasing vulnerability with higher prevalence of low-income farmers (S2 Fig). In contrast, the Atlantic Forest, Cerrado, Caatinga, and Pampa presented a pronounced positive relationship, with vulnerability sharply increasing as the proportion of low-income farmers increased (S2 Fig).
In general, the economic value of pollination in Brazil was concentrated in the Atlantic Forest and Cerrado. Nevertheless, when soybean was excluded, the EVP declined to approximately US$6.0 billion (Table 4). EVP varied significantly across biomes, with more striking differences when soybean was excluded from the analysis. Specifically, when soybean was included, the highest EVP was concentrated in the Atlantic Forest and the Cerrado, which did not differ significantly from each other and together represented the largest fraction of the overall value across all biomes (~US$11.2 billion, 81.1% of the total EVP; Table 4, Fig 8A). The Amazon and Caatinga supported intermediate EVP values, but the Amazon exhibited a significantly higher EVP than that observed in the Caatinga (Fig 8A). Meanwhile, the Pampa grasslands and the Pantanal wetlands supported the lowest scores (Fig 8A). In contrast, when soybean was excluded from the estimates, the Atlantic Forest clearly dominated, showing significantly higher EVP than all other biomes (~US$2.8 billion, equivalent to approximately 47% of the total EVP; Table 4, Fig 9A), followed by the Cerrado with considerably lower values (Fig 9A). Under this scenario, the Caatinga and Amazon supported intermediate values, but EVP was significantly higher in the Caatinga compared to the Amazon (Fig 9A). The Pampa and Pantanal presented only marginal values and did not differ from each other, although the Pantanal partially overlapped with some of the intermediate biomes (Fig 9A). We can therefore observe that soybean cultivation substantially increased the EVP in the Cerrado, bringing it to the same level as the Atlantic Forest. Conversely, since the EVP in the Caatinga did not increase with soybean, its relative importance increased, surpassing the Amazon when soybean was excluded, although their absolute values remained similar.
Bars indicate mean ± SE. Different letters above bars represent significant differences among biomes.
Bars indicate mean ± SE. Different letters above bars represent significant differences among biomes.
Vulnerability to pollinator loss across all Brazilian biomes was relatively low and homogeneous when soybean was included in the analysis. The Caatinga presented the highest vulnerability values, and significantly higher than the Cerrado, which presented the lowest scores (Fig 8B). In contrast, the Amazon, Atlantic Forest, Pampa, and Pantanal occupied intermediate positions lacking significant differences (Fig 8B). The vulnerability pattern shifted and the underlying biome disparities were more pronounced after soybean was removed. The Amazon exhibited an increase in vulnerability under this scenario, displaying intermediate values, while the Caatinga retained the highest vulnerability, far above all other biomes (Fig 9B). The Atlantic Forest and Pantanal also exhibited intermediate vulnerability, although the latter showed considerable variation and partially overlapped with the Caatinga and Cerrado. The Cerrado and Pampa were consistently the least vulnerable, did not differ from each other, but differed significantly from the Caatinga. Therefore, it was possible to detect that soybean reduced the vulnerability contrast between biomes, but its exclusion revealed a clearer gradient.
4. Discussion
Our results indicate that cropland production in the world’s leading tropical agricultural country depends on biotic pollination to varying degrees. Most of the ~ 5,500 Brazilian municipal counties (96.8%) are, to some extent, vulnerable to pollination failure, which is related to factors such as farmland size, farmer income, level of habitat loss, income inequality, and other socioeconomic indicators. Around 58% of all Brazil’s leading farmland crops rely on animal pollination, with key cultivars such as soybean, coffee, cotton, tomato, and cocoa accounting for 84% of the overall agricultural economic value linked to pollination. Although representing 35% of total production, soybean cultivation accounts for more than half of the total economic value of Brazilian agricultural production. Soybean croplands strongly shape the spatial and socioeconomic patterns of the economic value of pollination and crop vulnerability in Brazil. Soybean homogenizes the responses at the biome-level, while its exclusion amounted to positive associations in the Atlantic Forest, Cerrado, and Caatinga. The relationship between vulnerability and low-income farmers also varies across Brazilian biomes. Thus, patterns of vulnerability to pollinator loss are determined by pollinator dependence and crop diversification across biomes. In light of these challenges, we discuss how enhancing agricultural practices and ecological processes may significantly improve food security, especially for low-income farmers in vulnerable areas.
Our findings reinforce the notion that losses in ecosystem services, measured here by the expected yield losses from biotic pollinator failure, exert more pronounced consequences in an already vulnerable group of rural producers [47]. For instance, smallholder farmers in Madagascar frequently face food production risks, such as pest damage and crop losses, which result in direct negative impacts on income and livelihoods, highlighting the urgent need to reduce future vulnerability to high-risk exposure [48]. In Brazil, subsistence consumption of edible fruits by low-income communities in the Caatinga dry forest is negatively impacted by chronic anthropogenic disturbances and increased aridity [49]. In this context, it is important to understand how losses in ecosystem services will affect the food systems and agricultural yields of smallholders, often represented by the low-income rural poor who subsidise their livelihoods with forest resources. Our results provide empirical evidence for this concern: when soybean was removed from the analysis, municipalities containing higher percentages of low-income farmers showed greater vulnerability to pollinator losses, especially in biomes like the Atlantic Forest, Cerrado, and the Caatinga. Conversely, the dominance of soybean in central-western Brazil obscured such a socioeconomic vulnerability, resulting in weak correlations at the biome level. This contrast reveals that soybean production strongly homogenizes the national pattern, masking the elevated risks faced by smallholders in other regions. These findings (1) call attention to the role major crops can play in relation to vulnerability assessments, and (2) align with broader evidence that smallholders are disproportionately exposed to losses of ecosystem services, as their limited adaptive capacity, small economies of scale, reliance on diversified pollinator-dependent crops, and socioeconomic fragility increase their vulnerability to global change [47,48,50]. In general, low-income smallholders rely on agriculture for subsistence, so that reduced agricultural productivity can have severe impacts on food security and family income [51]. Our findings demonstrate that the interaction between rural poverty and pollination dependence amplifies agricultural vulnerability, and that this interaction, rather than uniform, varies markedly across Brazil’s phytogeographic biomes. This indicates that soybean masks socioecological inequalities and that smallholder vulnerability carries a strong structural component.
The spatial and socioeconomic heterogeneity observed in our results is largely shaped by the central role of soybean in the Brazilian national agricultural economy. Soybean tend to homogenize patterns, shifting the EVP and vulnerability to regions characterized by high mechanized production, which explains the positive association between EVP and variables such as HDI, GDP, habitat loss, and farm landholding size. Higher-income farmers typically embrace modern conventional agriculture, which is centered on intense use of synthetic fertilizers and pesticides, as well as mechanized monoculture production [52,53]. Conventional agriculture is widely known to be one of the main drivers of habitat degradation, which in turn reduces native biodiversity (e.g., [1]). As a consequence, floral diversity and resource availability for pollinators are reportedly decreased (e.g., [6,54]). This contrasts with smallholder farmers, who generally maintain organic or diversified agricultural practices, such as the use of agroforestry systems [55,56]. These systems are widely recognized for supporting higher agricultural diversity, conserving natural habitats, enhancing biodiversity, and promoting pollinator assemblages, thus contributing to the delivery of many ecological services, including pollination [56]. Smallholders are typically low-income farmers and their production is based on family farming, mainly food crops for subsistence [55–62]. Therefore, differences in agricultural practices adopted by farmers of different income levels are reflected in the delivery of ecosystem services, such as pollination. In this sense, the socioeconomic context influences the capacity for sustainable management and the conservation of suitable habitats available for pollinators. Thus, socioeconomic inequality translates into ecological inequality, with wealthier farmers driving habitat degradation (e.g., [47]), while a disproportionate number of poorer farmers depending more directly on ecosystem services but being impacted by habitat loss and degradation of ecosystem functioning [5]. Because they lack institutional support and are less resilient, individuals who are in vulnerable situations are disproportionately impacted [63]. In order to develop successful strategies that balance agricultural production with biodiversity protection, it is essential to incorporate socioeconomic and ecological factors into decision-making.
The Atlantic Forest and Cerrado domains had the highest EVP values, while the Caatinga and Amazonia grouped the most vulnerable municipalities. The contrasts between biomes allow us to interpret certain trends. Despite the additional economic importance, the combination of severe habitat fragmentation, family farming, and crop diversity results in high regional exposure to pollinator declines. Tropical forests are critically threatened by agricultural frontier expansion, mainly dominated by mechanized monocultures such as soybeans, sugar cane, and palm oil [1,4]. In Brazil, soybean cropland expansion alone accounts for a significant fraction of deforestation in the Atlantic Forest, Cerrado, Pampa, and Amazonian biomes [13,64–66]. However, 70% of all soybean plantations in Brazil are in the Cerrado and/or the Amazon [65]. In the Amazon, small-scale farmers depend on agricultural products as their primary source of income, but they continue to face major challenges such as access to markets and land degradation, especially in agricultural frontier areas [67]. The Amazon shows intermediate EVP values and high internal heterogeneity: in agricultural frontier areas, the expansion of soybeans and other monocultures increases habitat loss and puts pressure on family systems, while in rural areas that still retain much forest cover, dependence on pollination can be high locally. Therefore, excluding soybeans, vulnerability in the Amazon tends to increase and reveal hidden inequalities in pollination benefits. The Caatinga domain hosts a population of c. 29 million people who often subsist on natural resources, with family farming being prevalent in local communities [68]. However, the Caatinga has been greatly degraded by chronic disturbances and increased aridity, which could threaten the food systems in the Caatinga dry forest, including the production of edible fruits [49]. The pattern of low EVP in the Caatinga can be explained by the prevalence of low-income smallholders who largely cultivate subsistence crops. On the other hand, the Atlantic Forest had the highest EVP but concentrated municipalities containing smaller and more vulnerable farms. Family farming of food crops in this domain feeds a large fraction of Brazil’s population [69], which can explain the higher EVP values. However, the Atlantic Forest has been historically threatened by habitat loss and fragmentation, which results in biodiversity loss and lower flows of ecosystem services in small farms adjacent to natural habitat patches [70–72]. Therefore, we highlight that smallholders and low-income farmers are more vulnerable not only to socioeconomic factors but also to biotic pollination loss, which can severely reduce yield production and threaten food security.
Finally, we seek to strengthen an emerging discussion on enhancing agricultural yields for smallholders and low-income farmers in Brazil. Mitigation strategies that could combine the maintenance of natural forests, native biodiversity, and agricultural production within the same landscape could reduce the vulnerability of small farmers to pollinator losses. Food security depends on stable agricultural production, and ecological intensification can emerge as a nature-based solution to safeguard or restore pollination services and promote food security by reducing the land use footprint by sparing adjacent patches of natural habitats [54,73]. Our results show that soybean expansion homogenizes nation-wide patterns by masking regional vulnerabilities to pollinator loss, highlighting the need for policies that are not only commodity-oriented but also context-dependent for low-income farmers, i.e., considering the realities of smallholder farmers in accordance with the biome/region they live in. However, amid the growing pressures of global environmental change, it is important to highlight that habitat loss is not the only driver of negative impacts on the delivery of pollination services, but also that the interactive effects of climate change must be integrated into approaches. For example, seeking to understand how rising temperatures and changing precipitation regimes alter the survival, phenology, and spatial distribution of plants and pollinators, which consequently leads to frequent mismatches in plant-pollinator interactions and negatively affects the effectiveness of pollination services in agricultural landscapes (e.g., [74]). Furthermore, in many cases, climate change can exacerbate the impacts of natural habitat fragmentation and degradation. Integrating these drivers into land-use planning is particularly urgent in Brazil, where regional contrasts in pollination vulnerability are profound and where poor farmers disproportionately rely on ecosystem services for their livelihoods. Therefore, future researchers and decision-makers should consider climate change projections as a key factor in developing effective strategies for land-use planning and biodiversity conservation. This will enable policies that promote sufficient levels of retention of pollinator habitats. This can effectively support smallholder agricultural production, creating more ecologically sustainable and resilient agricultural systems in the face of global environmental change, capable of sustaining key ecosystem services, such as pollination. This can ensure that ecosystem services are maintained not only for export-oriented commodity agriculture but also for diverse local food systems. It is important to emphasize that the vulnerability of smallholder farmers is structural, stemming from historically persistent socioeconomic inequalities, limited adaptive capacity, and a strong reliance on pollinator-dependent crops. These structural factors must therefore be assessed for the resilience of agricultural production. Our findings reinforce the notion that ecological strategies alone are insufficient to reduce agricultural vulnerability if underlying social disparities cannot be simultaneously reduced. Therefore, by considering these factors, agricultural landscapes can become more socioecologically sustainable, supporting key ecosystem services such as pollination.
Supporting information
S1 Table. Model selection results for the Economic Value of Pollination (EVP) and Vulnerability to Pollinator Decline models based on the inclusion of Moran’s Eigenvector Maps (MEMs).
https://doi.org/10.1371/journal.pone.0337328.s001
(PDF)
S1 Fig. Relationship between the proportion of low-income farmers and (A) the economic value of pollination (EVP) and (B) the vulnerability associated with pollinator dependence across six Brazilian biomes, with soybean included.
Colored lines represent fitted trends for each biome.
https://doi.org/10.1371/journal.pone.0337328.s002
(TIFF)
S2 Fig. Relationship between the proportion of low-income farmers and (A) the economic value of pollination (EVP) and (B) the vulnerability associated with pollinator dependence across six Brazilian biomes, with soybean excluded.
Colored lines represent fitted trends for each biome.
https://doi.org/10.1371/journal.pone.0337328.s003
(TIFF)
References
- 1. Laurance WF, Sayer J, Cassman KG. Agricultural expansion and its impacts on tropical nature. Trends Ecol Evol. 2014;29(2):107–16. pmid:24388286
- 2. Schmitz C, van Meijl H, Kyle P. Land-use change trajectories up to 2050: Insights from a global agro-economic model comparison. Agric Econ. 2014;45:69–84.
- 3. Tilman D, Clark M, Williams DR, Kimmel K, Polasky S, Packer C. Future threats to biodiversity and pathways to their prevention. Nature. 2017;546(7656):73–81. pmid:28569796
- 4. Oliveira W, Galetto L, Tabarelli M, Peres CA, Lopes AV. Paradoxically striving for food security in the leading food-producing tropical country, Brazil. One Earth. 2023;6(5):455–8.
- 5. Mikkelson GM, Gonzalez A, Peterson GD. Economic inequality predicts biodiversity loss. PLoS One. 2007;2(5):e444. pmid:17505535
- 6. Oliveira W, Cruz-Neto O, Tabarelli M, Galetto L, Peres CA, Lopes AV. Markedly declining reproductive functional diversity of food plants in the world’s largest tropical country despite rapid cropland expansion. Agriculture, Ecosystems & Environment. 2023;357:108673.
- 7.
PENSSAN/VIGISAN. II inquérito nacional sobre insegurança alimentar no contexto da pandemia da COVID-19 no Brasil. São Paulo: Rede PENSSAN. 2022.
- 8. Schneider M, Biedzicki de Marques AA, Peres CA. Brazil’s Next Deforestation Frontiers. Tropical Conservation Science. 2021;14.
- 9. Graeub BE, Chappell MJ, Wittman H, Ledermann S, Kerr RB, Gemmill-Herren B. The State of Family Farms in the World. World Development. 2016;87:1–15.
- 10.
IBGE. Produção Agrícola Municipal. 2021. https://www.ibge.gov.br/estatisticas/economicas/agricultura-e-pecuaria/9117-producao-agricola-municipal-culturas-temporarias-e-permanentes.html?=&t=resultados
- 11. Novais SMA, Nunes CA, Santos NB, D Amico AR, Fernandes GW, Quesada M, et al. Effects of a Possible Pollinator Crisis on Food Crop Production in Brazil. PLoS One. 2016;11(11):e0167292. pmid:27902787
- 12.
Relatório temático sobre Polinização, Polinizadores e Produção de Alimentos no Brasil. São Carlos-SP: Editora Cubo. 2019.
- 13. Oliveira W, Colares LF, Porto RG, Viana BF, Tabarelli M, Lopes AV. Food plants in Brazil: origin, economic value of pollination and pollinator shortage risk. Sci Total Environ. 2024;912:169147. pmid:38065486
- 14. Fanzo J. From big to small: the significance of smallholder farms in the global food system. Lancet Planet Health. 2017;1(1):e15–6. pmid:29851589
- 15. Ricciardi V, Ramankutty N, Mehrabi Z, Jarvis L, Chookolingo B. How much of the world’s food do smallholders produce?. Global Food Security. 2018;17:64–72.
- 16. Cely-Santos M, Lu F. Intersections between rural livelihood security and animal pollination in Anolaima, Colombia. Geoforum. 2019;104:13–24.
- 17. Giannini TC, Cordeiro GD, Freitas BM, Saraiva AM, Imperatriz-Fonseca VL. The Dependence of Crops for Pollinators and the Economic Value of Pollination in Brazil. J Econ Entomol. 2015;108(3):849–57. pmid:26470203
- 18. Klein A-M, Vaissière BE, Cane JH, Steffan-Dewenter I, Cunningham SA, Kremen C, et al. Importance of pollinators in changing landscapes for world crops. Proc Biol Sci. 2007;274(1608):303–13. pmid:17164193
- 19. Ollerton J, Winfree R, Tarrant S. How many flowering plants are pollinated by animals?. Oikos. 2011;120(3):321–6.
- 20.
Potts SG, Imperatriz-Fonseca VL, Ngo HT. Summary for policymakers of the assessment report of the Intergovernmental Science-Policy Platform on Biodiversity and Ecosystem Services on pollinators, pollination and food production. Secretariat of the Intergovernmental Science-Policy Platform on Biodiversity and Ecosystem Services. Bonn, Germany: Secretariat of the Intergovernmental Science-Policy Platform on Biodiversity and Ecosystem Services. 2016.
- 21. Lopes AV, Porto RG, Cruz-Neto O. Neglected diversity of crop pollinators: lessons from the world’s largest tropical country. Perspect Ecol Conserv. 2021;19:500–4.
- 22. Eilers EJ, Kremen C, Smith Greenleaf S, Garber AK, Klein A-M. Contribution of pollinator-mediated crops to nutrients in the human food supply. PLoS One. 2011;6(6):e21363. pmid:21731717
- 23. Oliveira W, Silva JLS, Porto RG, Cruz-Neto O, Tabarelli M, Viana BF, et al. Plant and Pollination Blindness: Risky Business for Human Food Security. BioScience. 2019.
- 24. Porto RG, Cruz-Neto O, Tabarelli M, Viana BF, Peres CA, Lopes AV. Pollinator-dependent crops in Brazil yield nearly half of nutrients for humans and livestock feed. Global Food Security. 2021;31:100587.
- 25. Díaz S, Demissew S, Carabias J, et al. The IPBES Conceptual Framework — connecting nature and people. Curr Opin Env Sust. 2015;14:1–16.
- 26. Díaz S, Pascual U, Stenseke M, Martín-López B, Watson RT, Molnár Z, et al. Assessing nature’s contributions to people. Science. 2018;359(6373):270–2. pmid:29348221
- 27.
United Nations. Transforming Our World: The 2030 Agenda for Sustainable Development. New York: United Nations. 2015.
- 28. Potts SG, Biesmeijer JC, Kremen C, Neumann P, Schweiger O, Kunin WE. Global pollinator declines: trends, impacts and drivers. Trends Ecol Evol. 2010;25(6):345–53. pmid:20188434
- 29. Aizen MA, Garibaldi LA, Cunningham SA, Klein AM. How much does agriculture depend on pollinators? Lessons from long-term trends in crop production. Ann Bot. 2009;103(9):1579–88. pmid:19339297
- 30. Silva FDS, Carvalheiro LG, Aguirre-Gutiérrez J, Lucotte M, Guidoni-Martins K, Mertens F. Virtual pollination trade uncovers global dependence on biodiversity of developing countries. Sci Adv. 2021;7(11):eabe6636. pmid:33692110
- 31. Wratten SD, Gillespie M, Decourtye A, Mader E, Desneux N. Pollinator habitat enhancement: benefits to other ecosystem services. Agric Ecosyst Environ. 2012;159:112–22.
- 32. Christmann S. Do we realize the full impact of pollinator loss on other ecosystem services and the challenges for any restoration in terrestrial areas?. Restoration Ecology. 2019;27(4):720–5.
- 33. Gallai N, Salles J-M, Settele J, Vaissière BE. Economic valuation of the vulnerability of world agriculture confronted with pollinator decline. Ecol Econ. 2009;68:810–21.
- 34. Porto RG, de Almeida RF, Cruz-Neto O, Tabarelli M, Viana BF, Peres CA, et al. Pollination ecosystem services: A comprehensive review of economic values, research funding and policy actions. Food Sec. 2020;12(6):1425–42.
- 35. Steward PR, Shackelford G, Carvalheiro LG, Benton TG, Garibaldi LA, Sait SM. Pollination and biological control research: are we neglecting two billion smallholders?. Agriculture & Food Security. 2014;3:5.
- 36. Garibaldi LA, Carvalheiro LG, Vaissière BE, Gemmill-Herren B, Hipólito J, Freitas BM, et al. Mutually beneficial pollinator diversity and crop yield outcomes in small and large farms. Science. 2016;351(6271):388–91. pmid:26798016
- 37. Tibesigwa B, Siikamäki J, Lokina R, Alvsilver J. Naturally available wild pollination services have economic value for nature dependent smallholder crop farms in Tanzania. Sci Rep. 2019;9(1):3434. pmid:30837555
- 38.
INCRA/FAO. Novo retrato da agricultura familiar - o Brasil redescoberto. 2000.
- 39.
IBGE. Censo Agropecuario 2017. Rio de Janeiro: IBGE. 2019.
- 40. Venter O, Sanderson EW, Magrach A, Allan JR, Beher J, Jones KR, et al. Last of the Wild Project, Version 3 (LWP-3): 2009 Human Footprint, 2018 Release. Palisades, NY: NASA Socioeconomic Data and Applications Center (SEDAC). 2018.
- 41.
Roubik DW. Pollination of cultivated plants: a compendium for practitioners. Rome: Food and Agriculture Organization of the United Nations. 2018.
- 42. Barfield AS, Bergstrom JC, Ferreira S, Covich AP, Delaplane KS. An Economic Valuation of Biotic Pollination Services in Georgia. J Econ Entomol. 2015;108(2):388–98. pmid:26470149
- 43. Brooks ME, Kristensen K, van Benthem KJ, Magnusson A, Berg CW, Nielsen A, et al. glmmTMB balances speed and flexibility among packages for zero-inflated generalized linear mixed modeling. The R Journal. 2017;9:378–400.
- 44. Hartig F, 2021. DHARMa: Residual Diagnostics for Hierarchical (Multi-Level/ Mixed) Regression Models (R package).
- 45. Dray S, Legendre P, Peres-Neto PR. Spatial modelling: a comprehensive framework for principal coordinate analysis of neighbour matrices (PCNM). Ecological Modelling. 2006;196(3–4):483–93.
- 46.
R Core Team. R: A Language and Environment for Statistical Computing. Vienna, Austria: R Foundation for Statistical Computing. 2019.
- 47.
Kumar P, Yashiro M. The marginal poor and their dependence on ecosystem services: Evidence from South Asia and Sub-Saharan Africa. Marginality: Addressing the nexus of poverty, exclusion and ecology. London: Springer. 2014.
- 48. Harvey CA, Rakotobe ZL, Rao NS, Dave R, Razafimahatratra H, Rabarijohn RH, et al. Extreme vulnerability of smallholder farmers to agricultural risks and climate change in Madagascar. Philos Trans R Soc Lond B Biol Sci. 2014;369(1639):20130089. pmid:24535397
- 49. Oliveira W, Cruz-Neto O, Silva JLS, et al. Chronic anthropogenic disturbances and aridity negatively affect specialized reproductive traits and strategies of edible fruit plant assemblages in a Caatinga dry forest. For Ecol Manag. 2022;514:120214.
- 50. Morton JF. The impact of climate change on smallholder and subsistence agriculture. Proc Natl Acad Sci U S A. 2007;104(50):19680–5. pmid:18077400
- 51.
Hertel TW, Rosch SD. Climate change, agriculture and poverty. 5468. Washington, DC: World Bank. 2010.
- 52. Pretty J. Intensification for redesigned and sustainable agricultural systems. Science. 2018;362(6417):eaav0294. pmid:30467142
- 53. Le Capion A, Oury FX, Heumez E, Rolland B. Conventional versus organic farming systems: dissecting comparisons to improve cereal organic breeding strategies. Org Agr. 2020;10:63–74.
- 54. Aizen MA, Aguiar S, Biesmeijer JC, Garibaldi LA, Inouye DW, Jung C, et al. Global agricultural productivity is threatened by increasing pollinator dependence without a parallel increase in crop diversification. Glob Chang Biol. 2019;25(10):3516–27. pmid:31293015
- 55. Altieri MA, Nicholls CI. An Agroecological Basis for Designing Diversified Cropping Systems in the Tropics. Journal of Crop Improvement. 2004;11(1–2):81–103.
- 56. Altieri MA, Funes-Monzote FR, Petersen P. Agroecologically efficient agricultural systems for smallholder farmers: contributions to food sovereignty. Agron Sustain Dev. 2011;32(1):1–13.
- 57. Altieri MA, Nicholls CI. Scaling up Agroecological Approaches for Food Sovereignty in Latin America. Development. 2008;51(4):472–80.
- 58. Altieri MA, Nicholls CI. Agroecology: challenges and opportunities for farming in the Anthropocene. IJANR. 2020;47(3):204–15.
- 59. Tomich TP, Brodt S, Ferris H. Agroecology: A review from a global change perspective. Annu Rev Environ Resour. 2011;36:193–222.
- 60. Burchfield EK, Nelson KS, Spangler K. The impact of agricultural landscape diversification on U.S. crop production. Agric Ecosyst Environ. 2019;285:106615.
- 61. Tamburini G, Bommarco R, Wanger TC, Kremen C, van der Heijden MGA, Liebman M, et al. Agricultural diversification promotes multiple ecosystem services without compromising yield. Sci Adv. 2020;6(45):eaba1715. pmid:33148637
- 62. McGreevy SR, Rupprecht CD, Niles D. Sustainable agrifood systems for a post-growth world. Nat Sustain. 2022;:1–7.
- 63.
Birkmann J, Liwenga E, Pandey R. et al. 2022. Poverty, Livelihoods and Sustainable Development. In: Climate Change 2022: Impacts, Adaptation and Vulnerability. Contribution of Working Group II to the Sixth Assessment Report of the Intergovernmental Panel on Climate Change. Cambridge University Press, Cambridge, UK and New York, NY, USA. 1171–1274, https://10.1017/9781009325844.010
- 64. Gibbs HK, Rausch L, Munger J, Schelly I, Morton DC, Noojipady P, et al. Environment and development. Brazil’s Soy Moratorium. Science. 2015;347(6220):377–8. pmid:25613879
- 65. Lima M, Silva Junior CAD, Rausch L, Gibbs HK, Johann JA. Demystifying sustainable soy in Brazil. Land Use Policy. 2019;82:349–52.
- 66. Soterroni AC, Ramos FM, Mosnier A, Fargione J, Andrade PR, Baumgarten L, et al. Expanding the Soy Moratorium to Brazil’s Cerrado. Sci Adv. 2019;5(7):eaav7336. pmid:31328157
- 67. Garrett RD, Niles MT, Gil JDB, Gaudin A, Chaplin-Kramer R, Assmann A, et al. Social and ecological analysis of commercial integrated crop livestock systems: Current knowledge and remaining uncertainty. Agricultural Systems. 2017;155:136–46.
- 68.
Silva JMC, Barbosa LCF, Leal IR, Tabarelli M. The Caatinga: understanding the challenges. Caatinga - The Largest Tropical Dry Forest Region in South America. Cham: Springer. 2017. 3–19.
- 69. Nolasco C, Soler L, Freitas M, Lahsen M, Ometto J. Scenarios of Vegetable Demand vs. Production in Brazil: The Links between Nutritional Security and Small Farming. Land. 2017;6(3):49.
- 70. Girão LC, Lopes AV, Tabarelli M, Bruna EM. Changes in tree reproductive traits reduce functional diversity in a fragmented Atlantic forest landscape. PLoS One. 2007;2(9):e908. pmid:17878943
- 71. Lopes AV, Girão LC, Santos BA, Peres CA, Tabarelli M. Long-term erosion of tree reproductive trait diversity in edge-dominated Atlantic forest fragments. Biol Conserv. 2009;142:1154–65.
- 72. Tabarelli M, Lopes AV, Peres CA. Edge‐effects Drive Tropical Forest Fragments Towards an Early‐Successional System. Biotropica. 2008;40(6):657–61.
- 73. Bommarco R, Kleijn D, Potts SG. Ecological intensification: harnessing ecosystem services for food security. Trends Ecol Evol. 2013;28(4):230–8. pmid:23153724
- 74. Oliveira W, Cruz-Neto O, Silva JLS, Tabarelli M, Peres CA, Lopes AV. Climate change will lead to local extinctions and mismatched range contractions disrupting bee-dependent crop pollination. Front Bee Sci. 2025;3.