Natural enemy and pollinator insects provide important ecosystem services for agriculture; however, crops are intensively managed and the occurrence of these insects within the agricultural matrix depends on the spillover from adjacent habitats (Haan et al., 2020). These beneficial insects often rely on non-crop habitats such as forests, meadows and grasslands in the surrounding landscape to find a permanent and diverse supply of food items, refuge from crop management, nesting sites and overwintering habitats (Tscharntke et al., 2012; Bartual et al., 2019). Beneficial insects can also rely on landscape diversity, foraging in multiple crop and non-crop habitat types, to fulfill resource requirements and complete their life cycle (Tscharntke et al., 2012; Sirami et al., 2019). Landscape metrics related to landscape configuration such as habitat connectivity and edge density also influence the spillover of beneficial insects and how far they can disperse into an agricultural matrix (Haan et al., 2020).

Cross-habitat spillover between crop and non-crop habitats is not unidirectional and the flux of organisms can occur from natural habitats to crops and vice versa (Sackett et al., 2009; Tscharntke et al., 2012). Multiple factors can determine the magnitude and direction of spillover in agricultural landscapes including crop management, and high contrast in habitat structure and primary productivity between crops and natural habitats (Frost et al., 2015). Natural habitats are unmanaged, provide a constant supply of resources for biodiversity and often act as a source of beneficial insects and ecosystem services to adjacent agroecosystems (Bianchi et al., 2006; Tscharntke et al., 2012). In contrast, the pulses of resource availability in crops (e.g., pest outbreaks and mass flowering) can make the agricultural matrix a temporary source of beneficial insects to adjacent natural habitats (Rand et al., 2006; Frost et al., 2015; Madeira et al., 2016). Such cross-habitat movements are crucial for persistence of beneficial insects in agricultural landscapes where many species rely on different resources that are spatially separated in different crop and non-crop habitats (Tscharntke et al., 2012).

Recent studies conducted in Brazilian coffee-producing landscapes provide evidence that Atlantic Forest is a crucial habitat for insect conservation, and that high forest cover and proximity enhance the occurrence of beneficial insects into adjacent crops, resulting in high coffee fruit set and low levels of pest infestations (Saturni et al., 2016; Hipólito et al., 2018; Aristizábal and Metzger, 2019; Medeiros et al., 2019a, 2019b, González-Chaves et al., 2020). These studies estimated how landscape structure affects the occurrence of beneficial insects and provision of ecosystem services within coffee plantations; however, further investigation is required to better understand how landscape features regulate the spillover of beneficial insects between forest remnants and adjacent coffee plantations.

Medeiros et al. (2019a) collected bees, flower flies and predatory wasps in 16 coffee-producing landscapes to investigate how landscape structure affects alpha and beta diversity of beneficial insects. In each of the 16 landscapes, beneficial insects were collected in forest remnants and adjacent coffee plantations and all individuals collected in both habitats were aggregated to calculate alpha and beta diversity at landscape level. For predatory wasps the authors found that total species richness and abundance increased, and species loss decreased with increasing landscape forest cover. Here, we used the same dataset as Medeiros et al. (2019a) to assess the effects of landscape composition and configuration on the spillover of wasps between forest remnants and adjacent coffee plantations. We used incidence and abundance-based dissimilarity indices to estimate the inter-patch movement of wasp species and individuals and we assumed that high dissimilarity means less flow (and thus less spillover) between forest and coffee habitats.

We hypothesized that the dissimilarity of paired wasp communities between forest and coffee habitats decreases with increasing forest cover, landscape diversity and edge density in the surrounding landscape, whereas pesticide usage does not affect dissimilarity patterns. We expected that forest cover would increase wasp spillover between habitats by increasing the availability of unmanaged nesting sites, whereas high diversity of habitat types and edge environments would provide alternative and complementary feeding habitats. We also expected that the spillover of wasps between forest and coffee habitats is mainly regulated by landscape features that minimize the negative impact of pesticide usage in crop matrix. We provide important information on how landscape structure regulates the spillover of beneficial insects in forest remnants and adjacent crops. Such information is crucial for landscape planning and decision-making focused on promoting the synergy between biodiversity conservation and the provision of ecosystem services in coffee-producing landscapes.

We collected a total of 9847 wasps (Vespidae: Polistinae and Eumeninae) classified into 75 species and 23 genera (Fig. 2 and supplementary material S1). Wasp species richness was higher in coffee plantations (68 species) than in forest remnants (43 species). Wasps were also more abundant in coffee plantations (6376 individuals) than in forest remnants (3471 individuals). The species Agelaia multipicta (Haliday, 1836) and Agelaia vicina (de Saussure, 1854) were dominant in both habitats and together encompassed 66% of the individuals collected in forest remnants and 40% in coffee plantations. Landscape explanatory variables and pesticide usage varied considerably along the 16 landscapes.

Fig. 2.

Number of individuals per species in each of the 16 pairs of forest and coffee habitats along the gradient of forest cover (dark gray bars on the top) in southeast Brazil. The letters C and F represent the paired coffee and forest habitats in each of the 16 sites/landscapes.

(0.5MB).

All four dissimilarity indices (Sorensen, Simpson, Bray-Curtis and its balanced variation component) decreased with increasing forest cover at the finer scale (250 m - Fig. 3) suggesting spillover of wasps between forest and coffee habitats increased with increasing forest cover in the surrounding landscape. The explanatory variables landscape diversity, edge density and pesticide usage did not affect wasp dissimilarities between forest remnants and coffee plantations. The nested and abundance gradient components did not respond to any of the explanatory variables (Supplementary material S2).

Fig. 3.

Linear models representing the influence of landscape composition on spillover of wasps between forest remnants and adjacent coffee plantations as represented by the influence of forest cover on dissimilarity indices. Incidence-based indices are represented by Sorensen and Simpson dissimilarity indices, respectively, for A and B. Abundance-based indices are represented by Bray-Curtis index and its balanced variation component, respectively for C and D.

(0.18MB).

Wasp dissimilarity decreased with increasing forest cover in the surrounding landscape, suggesting that forest amount increases the spillover of wasps between forest remnants and adjacent coffee plantations. In contrast, wasp dissimilarity did not respond to landscape diversity or edge density.

Beneficial insects encompass many functional groups that perceive landscape composition and configuration in different ways. Some species rely on landscape complementation (Dunning et al., 1992) and benefit from landscape diversity and edge environments, while other species rely on habitat amount (Fahrig, 2013) and strongly dependent on a single habitat or few habitat types. Our findings suggest that forest cover (habitat amount) in the surrounding landscape is the main factor regulating wasp spillover between forest and coffee habitats at a local scale. All four dissimilarity indices decreased with increasing forest cover in the landscape resulting in high diversity and equability between forest and coffee habitats. This means that forested landscapes support more diverse wasp communities because of the high habitat amount that improves the accessibility of local communities to the regional species pool by increasing spillover opportunities and availability of nesting habitats.

For wasps, forest remnants are important sources of undisturbed nesting sites (Souza et al., 2014; Oliveira et al., 2017) and food resources (Sobek et al., 2009) that are crucial for wasp persistence in periods of scarcity in farmland. Nesting sites are particularly important for wasps, which are central foragers and move repeatedly among nesting and feeding habitats (Charnov, 1976; Richter, 2000). In our study system, we assumed that wasps use forest remnants as nesting habitats and coffee plantations as feeding habitats. We argue that coffee plantations cannot provide nesting habitats for wasps mainly due to nest removal that is a common practice among farmers to avoid sting accidents during crop management and harvesting operations. At the habitat level, coffee plantations supported higher species richness and abundance of wasps than in adjacent forest remnants. Such a pattern can be associated with high primary productivity of crops leading to a higher density of consumers in the crop matrix than in adjacent natural habitats (Rand et al., 2006; Frost et al., 2015). However, we highlight that wasp alpha diversity in both forest and coffee habitats increased with increasing forest cover revealing that crop dominated landscapes supported only subsets of species found in forested landscapes (Medeiros et al., 2019a).

The lack of effect of pesticide usage on wasp dissimilarity indices can be also associated with the rapid recolonization of crops promoted by the spillover from forest remnants (nesting habitats; Souza et al., 2010, 2014) that act as permanent sources of wasps to adjacent coffee plantations (Medeiros et al., 2019b). Moreover, landscape diversity and edge density did not affect wasp spillover suggesting that nesting sites are not available in other crop and non-crop habitats nor in edge environments.

The four dissimilarity indices (Sorensen, Bray-Curtis, Simpson, and Balanced variation) responded to forest cover at the smaller scale. This points out that most wasp species can spillover between habitats within landscapes at a radius of 250 m that is in accordance with the very few studies available from the Neotropical region suggesting that some wasp species have a dispersal range of up to 300 m from nesting sites (Santos et al., 2000; da Cruz et al., 2006, Ribeiro-Filho et al., 2008). However, we highlight that the alpha and beta diversity of wasps at the landscape level responded to forest cover up to 1000 m (Medeiros et al., 2019a). Therefore, the effects of forest cover on the landscape wasp species pool seems to occur at larger spatial scales.

Forest cover reduced species loss and increased diversity of wasps and provision of biological control services in the same 16 landscapes used in the present study (Medeiros et al., 2019a, 2019b). Forest cover is also important for other beneficial insects and their ecosystem services in coffee producing landscapes. In our study region, Saturni et al. (2016) found an increment of 28% in coffee fruit set that was associated with the presence of bees, which were positively correlated with forest amount in the surrounding landscape. Similar results were found for predatory ants, which decreased presence, infestation, and damage of a major coffee pest in forested landscapes, but not in structurally simple landscapes (Aristizábal and Metzger, 2019). The positive effects of forest remnants on beneficial insects and the provision of ecosystem services in adjacent coffee plantations were also observed in Central America (Roubik, 2002; Ricketts, 2004), southeast Asia (Pries et al., 2007; Krishnan et al., 2012) and Africa (Milligan et al., 2016; Berecha et al., 2015). Therefore, forest conservation and restoration initiatives should be incorporated in landscape planning and management to improve the spillover of beneficial insects and their ecosystem services in farmland creating a win-win scenario for biodiversity and coffee farmers.

The study was designed by H.R.M, C.A.R and M.C.R.; Data collections were performed by H.R.M.; Data analyses were performed by F.M. All authors actively contributed to writing and revising the manuscript.

The authors declare that there are no competing personal and financial interests.