The degradation of forests in tropical agricultural landscapes has reduced biodiversity and may negatively affect ecosystem services. However, the role of forest fragments in the provision of ecosystem services for agroecosystems, such as biological control, is still not well understood. This study analyses the relationship between tree diversity and arthropod communities in forest fragments and adjacent sugarcane fields around Jaboticabal (São Paulo, Brazil) to evaluate the benefits of preserving Atlantic Forest fragments. We tested the hypotheses: (1) arthropod diversity and the abundance of key functional groups are favoured by tree diversity, (2) diversity and abundance of these functional groups in sugarcane fields depend on adjacent forests, and (3) the insect community composition of forest fragments and sugarcane fields is similar due to spillover effects. Arthropods were sampled in ten forest fragments over two years (2020 and 2022) using traps (Malaise, yellow pan) and direct tree sampling. We identified arthropods to morphospecies (2020) and family level (2022) and associated families to functional groups. We found that the quality of forest fragments, particularly tree species richness, was positively related to forest arthropod richness and frequency (occupied samples) of functional groups involved in ecosystem services such as predators but also of phytophagous insects. There was also a strong positive correlation between insect family richness and functional groups of forest fragments and adjacent sugarcane fields, indicating potential spillover and spatial connectivity between habitats. Although insect community composition was significantly different between forest fragments and sugarcane fields, NMDS ordination revealed partial overlap of family communities, suggesting many shared taxa in both habitats. This study highlights the importance of forest fragments in preserving insect biodiversity and related ecosystem services in tropical agricultural landscapes and advocates for the conservation and restoration of these habitats. However, the benefits in terms of crop damage reduction and yields remain to be confirmed.

The study was conducted in 10 fragments of Atlantic Forest (Mata Atlântica) and adjacent sugarcane fields in the municipality of Jaboticabal, São Paulo state, Brazil. Five fragments were located on shallow slopes (plateau fragments) and five were remnants of riparian forest close to rivers and streams (gallery fragments). Plateau fragments typically exhibited a higher cover of shrub and liana species compared to river valley fragments. River valley fragments are protected as Permanent Preservation Areas (APPs), in particular to preserve the water quality of piped water catchment areas. Arthropods were sampled in April 2020 and in August 2022, using direct and indirect (trap) collection methods. Three fragments had to be replaced in 2022 due to access problems and forest fires. In each forest fragment, we collected data in a randomly selected 50 m x 20 m plot, 5 m from the edge. Arthropods were also sampled in sugarcane fields adjacent to the forest fragments.

In 2020, a preliminary study was conducted focusing on arthropod biodiversity in forest fragments. We defined woody plants larger than 5 cm in diameter at breast height as trees and identified them by species. Sampling was limited to 20 trees associated with five 1 m² subplots in each plot (tree closest to each corner). We used a simple habitat quality index to characterize forest fragments. The index uses tree cover as a positive criterion and vine and grass cover as negative criteria. Cover was estimated as the vertical projection of plant organs above the ground using a scale from 0 to 1 with 0.1 intervals. The quality index was calculated as the sum of the three parameters, with liana and grass cover adjusted as negative values ​​and tree cover as positive values.

In August 2022, a more detailed method was used to obtain more information about the characteristics of the plant and arthropod community. All trees in the 1000 m² plots were identified and insects were sampled from all trees during the dry season. On each tree, we carefully cut a randomly selected branch at a height of 2–5 m with long-reach pruning shears and manually captured all arthropods. The sampled arthropods were transferred to 70% diluted alcohol before identification. In 2020, insects and spiders were identified to the morphospecies level. In 2022, we identified the sampled insect individuals to the family level to obtain more information about taxonomic and functional groups, but we did not include spiders. We assigned all insects sampled in 2022 to the following functional groups, using knowledge of the life history feeding traits of most species in each family: parasitoids, predators, omnivores, and phytophagous arthropods. Families that could not be clearly assigned to a functional group were considered omnivores and ultimately removed from functional group analyses. For insects that switch diets between larval and adult stages, we used larval feeding behavior, since larvae generally have a higher demand for resources and are the main consumers in the life cycle.

In 2022, we additionally used traps to sample flying insects. A Malaise trap was placed in each forest fragment plot. In sugarcane fields, we installed two yellow traps (42 x 28 x 8 cm) 15 m apart and 5 m from the edge of the sugarcane field. Each trap was filled with 2 L of water and 10 drops of detergent before exposure for a period of 48 hours. Likewise, the insects were preserved in 70% alcohol, identified at the family level and classified into functional groups.