Discussion

In this study we identified the overall effectiveness of biocontrol techniques in controlling insect pests of crops in sub-Saharan Africa, and identified patterns across biocontrol interventions, pest taxa, crop types and experimental design. Using a meta-analytical approach, we found that biocontrol interventions effectively reduced pest abundance and crop damage by over 50%, while increasing crop yield by more than 60%. The size of the yield increases highlights the great challenge posed by insect pests to smallholder crop production, which is in line with recent evidence estimating high crop losses to pests, especially in the absence of any control intervention (Oerke 2006; Savary et al. 2019). The substantial yield increase that biocontrol can provide could have an enormous impact on sub-Saharan food security if these practices are scaled up to regional level. Crucially, we showed comparable performance of biocontrol and synthetic pesticides on pest abundance, crop damage and crop yield, and a significant reduction in the loss of natural enemies, particularly following botanical pesticides application.

Biocontrol effectiveness across biocontrol intervention techniques

Pest abundance and crop damage were negatively affected by biocontrol across all interventions. Push-pull and botanical pesticides had the greatest effect on crop yield, increasing production by 92% and 80% respectively. This may be due to the highly effective companion crops utilised in push-pull technologies, which release bioactive chemicals that repel pests and attract natural enemies, while also suppressingStriga , a parasitic weed which causes up to 100% yield losses across SSA (Khan et al. 2014). The large yield increase observed in our synthesis may be due to a combination of the pest repellent and weed suppression abilities of push-pull implementation. Our findings indicate the potential of botanical pesticides to be an effective method of pest control in SSA. However, two thirds of the studies included here were carried out on research farms, which may be under more controlled settings compared to more realistic field conditions, potentially inflating the observed effect size.
Our review captured a small number of studies on classical biocontrol interventions, including augmentation, despite successful examples such as the control of the Cassava mealybug (Phenacoccus manihoti ) by the Encyrtid wasp (Anagyrus lopezi ) (Norgaard 1988). Conceivably these interventions may be hampered by the high costs involved in their research and production, such as insect rearing facilities (Neuenschwander 2004), and the growing concerns on the environmental risks of releasing exotic species (Van Lenteren et al. 2006). Therefore, they may only be implemented for highly widespread and devastating pests such as the Cassava mealybug or the Tomato leaf miner (Tuta absoluta ).

Biocontrol effectiveness across crop type and pest taxon

Cereals were the most studied crops in our meta-analysis, conceivably because they play a central role in the region’s food security, accounting for about 50% of total crop area and caloric intake (Robinson et al. 2015). Nonetheless, other crop types such as fruits, pulses and fibre should be included in future research in this area. Our study provides strong evidence of the effectiveness of biocontrol across all taxa, particularly against lepidopteran crop pests. The potential of biocontrol to reduce cereal crop damage by 60% is encouraging given the devastating damage caused, particularly on maize, by caterpillars including fall armyworm (Spodoptera frugiperda ), Diamondback moth (Plutella xylostella ), Crambid cereal stemborer (Chilo partellus ) and Maize stemborer (Busseola fusca ).

Biocontrol effect on natural enemies and non-target pests

Understanding the effect of biocontrol on natural enemy populations is crucial as they are both an indication of pest control potential and a measure of the impact of the pest control method on non-target species. Our results showed no overall change in NEA following biocontrol application when compared to untreated fields. Although, we found a significant decline in natural enemy abundance following botanical pesticides application. The most likely explanation for this is that the interventions have reduced prey availability for natural enemies, making them move to other more profitable foraging locations, but the direct negative impact of some interventions, such as some broad-spectrum botanical pesticides, cannot be excluded (Ndakidemi et al. 2016). The existing evidence for the effect of botanical pesticides on non-target species is conflicting, with some research showing that plant extracts such as neem, garlic and eucalyptus may cause mortality and have sub-lethal effects on beneficial insects (Simmonds et al.2002; Maia & Moore 2011), while other studies found no detrimental effect of pepper and garlic extract on natural enemies populations (Fening et al. 2013; Amoabeng et al. 2020). More research is needed to draw robust inferences on the repercussion of botanical pesticides on beneficial/non-target species before considering large-scale adoption.
Evidence is more consistent on the positive response of natural enemy populations to biocontrol interventions such as push-pull and field margins (Koji et al. 2007; Midega et al. 2008), which is in line with evidence from the global north on the benefits of habitat enhancement on natural enemy density and diversity (Blaauw & Isaacs 2012; Holland et al. 2017). However, we found that only 14% of the studies measured NE abundance following biocontrol application in sub-Saharan Africa. Natural enemy abundance should be measured more consistently in future studies to further elucidate direct and indirect effects of biocontrol on non-target species.
Furthermore, the most common outcomes measures reported in the studies focussed on the abundance of pests and/or natural enemies, while we did not find studies measuring their species diversity or functional group diversity. However, it has been shown that biocontrol is strengthened by increased natural enemy richness (Griffin et al. 2013; Katanoet al. 2015) and this is consistent across temperate and tropical regions (Letourneau et al. 2009). Ecosystem functioning can be stabilised by functional redundancy, by enabling functional groups to compensate for individual species fluctuations and increase the resilience of ecosystem against species loss (Rosenfeld 2002; Hooperet al. 2005). This is particularly relevant to understand the long-term impact of biocontrol on natural enemy communities and their pest suppression ability and should be explored in future research.

Biocontrol effectiveness compared to synthetic pesticides

When compared to synthetic pesticides, biocontrol interventions had a similar impact on pest abundance and crop damage, which is a critical finding for farmers who cannot access or afford chemicals. Crucially, natural enemy abundance was significantly reduced after synthetic pesticides application even over the short time scales of the studies examined. In the long term there could be greater reductions in pest and crop damage following biocontrol as a result of more abundant and diverse communities of natural enemies. In terms of a reduction in the negative environmental impacts associated with chemical pesticides, the benefits provided by more resilient natural enemy populations could be one of several indirect positive effects of opting out of conventional pesticide use. It is worth noting that most comparisons with synthetic pesticides were measured against botanical pesticides, therefore inferences for other biocontrol methods should be made with caution. Future research should aim to determine the effectiveness of biocontrol approaches, such as push-pull, when compared to synthetic pesticides to fill this knowledge gap.
A possible limitation of this study is the potential selection bias towards significant results, causing an overrepresentation in the published literature, a criticism that could be levelled against all meta-analyses. The two tests we used to assess publication bias yielded conflicting results, hence it is hard to know with certainty the scale of publication bias towards results where an effect was found. However, we show that crop losses to pests are significantly higher in untreated fields, supporting the idea that any crop protection intervention has the potential to improve yields substantially. The size of the yield gains shown in the current meta-analysis suggest there is a big opportunity to raise yields with biocontrol interventions.

Landscape composition and biocontrol

Our study set out to answer the question, “does the surrounding landscape composition affect the effectiveness of biocontrol interventions?”. However, we found a paucity of studies investigating either the effect of landscape composition on biocontrol effectiveness, or the relationship between landscape composition and natural enemy abundance. The research we found indicated a significant decrease of natural enemy density and predation/parasitism activity with isolation from natural habitat (e.g., Henri et al. 2015; Soti et al.2019). This is in line with recent research showing a similar effect of landscape complexity on pollinators and natural enemies in sub-Saharan regions (Ratto et al. 2021; Jordon et al. 2022) and a larger body of research particularly in the global north (Bianchiet al. 2006; Chaplin-Kramen et al. 2011; Shackelfordet al. 2013).
Furthermore, the sparse evidence we found focusing on the effect of landscape composition on biocontrol effectiveness showed inconsistent results. Midega et al. (2014) found that semi-natural habitat acted as a source of lepidopteran pests to the maize crop fields in Kenya, while Kebede et al . (2019) demonstrated that landscape simplification overrode the effect of intercropping practices and was the main driver of pest infestation levels. A key avenue for future research would involve large scale studies to identify clear patterns in the relationship between landscape complexity and natural enemy activity and the ecosystem service delivered to sub-Saharan agricultural systems. Furthermore, recent evidence from SSA showed that natural enemy diversity in crop fields is dependent on the land management of neighbouring fields (Tripathi et al. 2022). This highlights the need for further multi-scale studies to identify potential variation in biocontrol effectiveness across different land management contexts.
In conclusion, our findings provide the first quantitative synthesis of biocontrol effectiveness in SSA, indicating that biocontrol interventions have the potential to substantially reduce crop damage, increase crop yield while maintaining natural enemy populations within sub-Saharan agricultural systems. Our results further suggest that biocontrol has comparable performances to synthetic pesticides with reduced adverse impact on beneficial insects and ecosystems. Our results suggest that biocontrol represents an effective intervention for farmers who do not have access to pesticides, while it can maintain crop yields without associated negative pesticide effects.