The promotion of native vegetation as a habitat for natural enemies, which could increase their abundance and fitness, is especially useful in highly simplified settings such as Mediterranean greenhouse landscapes. Spiders as generalist predators may also be involved in intra-guild predation. However, the niche complementarity provided by spiders as a group means that increased spider diversity may facilitate complementary control actions. In this study, the interactions between spiders, the two major horticultural pests, Bemisia tabaci and Frankliniella occidentalis, and their naturally occurring predators and parasitoids were evaluated in a mix of 21 newly planted shrubs selected for habitat management in a highly disturbed horticultural system. The effects of all factors were evaluated using redundancy analysis (RDA) and the generalized additive model (GAM) to assess the statistical significance of abundance of spiders and pests. The GAM showed that the abundance of both pests had a significant effect on hunter spider’s abundance, whereas the abundance of B. tabaci, but not F. occidentalis, affected web-weavers’ abundance. Ordination analysis showed that spider abundance closely correlated with that of B. tabaci but not with that of F. occidentalis, suggesting that complementarity occurs, and thereby probability of biocontrol, with respect to the targeted pest B. tabaci, although the temporal patterns of the spiders differed from those of F. occidentalis. Conservation strategies involving the establishment of these native plants around greenhouses could be an effective way to reduce pest populations outdoors.

Agricultural landscapes are constantly changing as farmers adopt new production practices and respond to changing environmental conditions. Some of these changes alter landscape structure with impacts on natural pest control, pesticide use, and conservation of biodiversity. In rice agroecosystems the effect of landscape structure on natural enemies and pest suppression is often poorly understood. Here we investigate the effect of landscape composition and configuration on a key pest of rice, the brown planthopper (Nilaparvata lugens). Using N. lugens as sentinel prey coupled with predator exclusions, we investigated landscape effects on herbivore suppression and rice grain yield at multiple spatial scales in two regions of Bangladesh. Ladybird beetles and spiders were the most abundant natural enemies of N. lugens with landscape effects observed at all scales on ladybird beetles. Specifically, ladybird beetles were positively influenced by road edges, and fallow land, while spiders were strongly influenced only by rice phenology. Predator exclusion cages showed that N. lugens abundance significantly increased in caged plots, reducing rice gain yield. We also used an estimated biocontrol service index that showed a significant positive relationship with landscape diversity and a significant negative impact on pest density and yield loss. These results suggest that promoting fallow lands and fragmented patches between rice fields could lead to more sustainable insect pest management in rice agroecosystems, potentially reducing the practice of prophylactic insecticide use.

Viruses are a limiting factor in maize production areas around the world. The knowledge of the interactions between agroecosystems and the virus-vector-host system is limited, but a landscape-scale approach could help fill this gap. In this study, we show how the use of multiple spatial scales, i.e. 200, 500 and 1000 m, is a novel methodology for explaining the incidence of two closely related potyviruses: maize dwarf mosaic virus and sugarcane mosaic virus. To determine the factors involved in virus incidence, we recorded the proportion of surrounding crops and non-crop habitats at the landscape scale and, at the field scale, we recorded the planting date, the maize field area, the crop rotation and the weed diversity in the edges. In addition, we estimated the numbers of aphids with sticky yellow traps. Virus incidence in maize and in alternative grass hosts was determined by DAS-ELISA. Generalised linear mixed models were fitted using the multimodel inference method. The results showed that the most predictive model for the incidence of both potyviruses was at a scale of 200 m, but for the aphid abundance, it was at a scale of 500 m. Maize dwarf mosaic virus incidence was most affected by field management, and sugarcane mosaic virus by landscape variables. The planting date and the weed diversity in the edges were the field variables with the highest positive effects on both potyviruses. Moreover, both viruses were positively related to the abundance of aphids, and maize dwarf mosaic virus was only related to the cover of Johnson grass in the edges. Non-crop habitats had negative effects on potyvirus incidence at all spatial scales, showing that biodiversity in the landscape decreases the incidence of viruses. Here, we show that the early planting, the management of edges and the presence of non-crop habitats are key factors.