Climate change poses a significant threat to the future of global food production with implications for agricultural productivity and food system sustainability (Lobell et al., 2011; Dang and Dabalen, 2017; Ray et al., 2019). In addition to climate change, there is widespread global concern about biodiversity loss due to agricultural expansion and intensification (Newbold, 2018; Beckmann et al., 2019). Besides its inherent, intrinsic value, biodiversity underpins vital ecosystem services that support food production (Dainese et al., 2019), such as the biological control of crop pests and diseases by natural enemies (Bommarco et al., 2018). As with food production, biodiversity and biodiversity-based ecosystem services are fundamentally dependent on climate and, hence, vulnerable to climate change (Pecl et al., 2017; Nunez et al., 2019).

Climate smart agriculture (CSA) has been developed as a climate adaptation framework for multiple aspects of agriculture, from field-level practices to food supply chains (Venkatramanan and Shah, 2019). CSA implementation has targeted developing regions such as the sub-Saharan Africa (SSA) (Neufeldt et al., 2013), which are most vulnerable to climate change and have limited capacity to cope, and invest in adaptive institutions and technologies (Fankhauser and McDermott, 2014; Porter et al., 2015; van Wijk et al., 2020). The primary goal of CSA is to achieve triple wins by contributing to three objectives; increasing productivity, increasing the resilience of food production, and mitigating greenhouse gas emissions (Lipper et al., 2018). However, there are gaps in evaluating the success of CSA approaches as studies have predominantly focussed on productivity outcomes (Dinesh et al., 2015; Rosenstock et al., 2019; van Wijk et al., 2020). In a limited number of cases, investigations of the resilience benefits have also been linked to productivity gains, where surplus food minimized the effects of climate impacts and increased the resilience of farmers (Dinesh et al., 2015; van Wijk et al., 2020). With such a narrow agricultural productivity and food security focus, the possible impact of CSA practices on other components that affect agricultural system resilience and environmental sustainability, such as biodiversity and ecosystem service provisioning (Thornton et al., 2018), is a crucial knowledge gap. Evaluation of these broader components would help develop a better understanding of the system-level effectiveness of, and trade-offs associated with the CSA framework. This is critical to enable the identification of agricultural strategies that improve food system multifunctionality necessary for building resilience to climate change (Thornton et al., 2018; van Wijk et al., 2020).

Here, we evaluate the effects of local-scale farm management, including CSA practices, on invertebrate diversity and agricultural yields, focusing on pest prevalence and suppression in the East Usambara Mountains of Tanzania. Specifically, we examine relationships among (i) invertebrate richness, abundance and volume of crop pests and natural enemies, (ii) percentage crop damage and, (iii) annual income from crops (USD ha⁻¹) under varying combinations of CSA and traditional farm management practices. Landcover diversity and configuration around the fields (Duarte et al., 2018; Zhang et al., 2020), and management practices in the neighboring fields affect the invertebrate diversity, pests, and natural enemies through spatial flow and interactions (Redlich et al., 2018). The understanding of the effects of the landscape context and the associated processes on local agrobiodiversity and related ecosystem services is mainly based on studies in large-scale and commercial farming systems in temperate ecosystems (Rusch et al., 2016; Woodcock et al., 2016; Aguilera et al., 2021). Only a few studies have examined the ecosystem services in biodiverse smallholder farming landscapes in SSA (Mkenda et al., 2019). Hence, we also account for the effects of landscape context and the influence of management in the neighboring fields on invertebrate richness, crop damage and income.

Our study provides new insights into the implications of CSA farm management for biodiversity and ecosystem services at local and landscape scales in SSA, which can inform future agricultural transformation strategies in the region and beyond.

Our study is the first to evaluate the ecological implications of CSA practices deployed in different combinations at field-scale. It helped identify potential synergies and trade-offs among the targeted objectives of CSA practices, annual agricultural yields and income, and outcomes for pest and natural enemy diversities and associated ecosystem service of biocontrol (Table 3). We showed that fields with mulches and Fanya juu trenches had the greatest synergy with the targeted objectives in terms of higher annual agricultural income and lower pest pressures. Surprisingly, the fields with maximum CSA practices did not have significantly different yields or invertebrate diversities, indicating that certain practices (mulches and Fanya Juu trenching) had a significant effect, but lots of practices combined ceased to produce any effect. This may simply be due to different practices having contrasting effects on biodiversity and yields – some positive and others negative, resulting in a no-net effect. Interventions such as greater spacing between crops and spice plantations had lower invertebrate diversities (pests as well as natural enemies), but higher annual returns. This indicates a possible trade-off between biodiversity and agricultural production. Further work in this area is needed to consider trade-offs in agricultural production over time, examining short term losses in the context of long-term goals.

We found that fields with trenches and mulching, primarily adopted to increase soil quality and moisture conservation in the fields, generally also had improved pest suppression. This was indicated by lower abundance and volume of pests, and higher crop yields and annual agricultural earnings. These fields also had a lower volume of natural enemies, but no impact on natural enemy richness or abundance. The role of mulches in reducing pest densities and diversities have been observed in a number of other studies (Altieri et al., 1985; Hooks et al., 1998; Brown and Tworkoski, 2004; Massaccesi et al., 2020) and is generally attributed to the greater density and diversity of ground-dwelling predators (Carabidae, Staphylinidae, and Arachnids) in the fields with live or compost mulching (Altieri et al., 1985; Schmidt et al., 2004). Past studies have also indicated the positive effects of mulching on predator populations and diversity (Schmidt et al., 2004). Natural enemy abundance and richness in the mulched fields in our study was not significantly different from other field management clusters. However, we observed that natural enemy volumes were significantly lower as most natural enemy individuals caught in the mulching fields were of smaller body sizes (mainly Parasitoids). This may be due to lower habitat diversity in and around the mulching fields, compared to the other fields which had agroforestry, spice plantations and intercropping and supported functionally diverse natural enemy communities comprising greater body-size distribution and diversity.

We did not find any other study evaluating the effect of Fanya juu trenches on invertebrates and our study is potentially the first to suggest that Fanya juu trenches and associated live mulching with grass strips offer significant benefits in terms of pest suppression in addition to the targeted objective of reduced run-off and soil erosion. Live mulches are generally associated with reduced pest infestations and crop damage (Altieri et al., 1985; Johnson and Gurr, 2016; Rehman et al., 2019), as also observed in our study. Our observations are comparable to the effects of push-pull technology where Napier grass has often been used as a “pull” crop, which attracts pests away from the crop and thereby reduces crop damage (Khan et al., 2011). Although it was not planted with the targeted objective of reducing crop pests, planting and live mulching with Napier grass demonstrates that there may be considerable synergies between CSA and conservation and pest management agricultural practices like push-pull. Thus, our study shows that the fields with mulching and trenching had the best outcomes in terms of pest suppression and productivity, and no significant reduction in non-pest or natural enemy diversity.

Among the other field clusters in this study, CS-Intercrop fields, where the maximum number of CSA technologies were deployed, were not significantly different from the monocrop traditional fields in terms of pest or natural enemy richness and abundance, crop damage or yields. Pest volumes, however, were the highest in CS-Intercrop fields. This may be because CSA fields containing improved crop varieties exhibited enhanced plant growth and higher foliar biomass, which attracted a larger functional or body-size diversity of insect herbivores (Real-Santillán et al., 2019). Furthermore, wider crop row spacing, and spice intercrops had lower richness of pest communities, higher crop damage, but greater yields than all other fields. This suggests that despite higher crop damage in food crops (maize and beans), the annual yields in Spice-Intercrop fields were not affected as spice plantations (clove and cardamom) and intercrops (cassava and bananas) mediated the net effect of pest damage.

Our study also showed that farm characteristics such as farm size, number of times it is cultivated, management practices in the neighboring fields and landscape context influenced the pests and natural enemies in the focal field as observed in other studies which are mainly from commercial farming landscapes (Rodriguez-Saona and Thaler, 2005; Madeira et al., 2016; Schulz-Kesting et al., 2021). Hence, this is one of the few studies from biodiverse smallholder farming landscapes in SSA that demonstrates such process are important in a wide range of farming systems. Particularly, farm size was associated with a reduction in abundance and volume of natural enemies. Increasing farm sizes is associated with fragmentation of natural landcover and landscape homogenisation, which may lead to a reduction in biocontrol as natural enemy communities with lower densities and a smaller range of body sizes are known to be less effective in pest control (Letourneau and Goldstein, 2001; Staton et al., 2021). Furthermore, neighbors practicing intercropping was associated with reduced pest abundance and an increase in natural enemy richness in local variety maize monocrops, mainly dominated by Coccinellids. Pest abundance was strongly related with an increase in crop damage and reduction in yields, while natural enemy richness was associated with a reduction in crop damage. This highlights the importance of local heterogeneity created by the difference in management practices between fields, which leads to spatial ecological interactions and flow of ecosystem services between neighboring fields where the characteristics of one field influences pest pressure and biocontrol in the other field (Landis et al., 2000; Sarthou et al., 2014).

Landscape heterogeneity, as indicated by forest edge length, had positive association with pest richness and negative relationship with crop damage. The effect of landscape configuration is typically context-dependent and has often been overlooked in the landscape-pest and biocontrol literature (Karp et al., 2018). An increase in landscape heterogeneity may lead to an increase in pest diversity resulting in aggravated pest pressures and crop damage (Martin et al., 2013; Zhang et al., 2020). In contrast, an increase in pest richness, diversity and evenness can also reduce pest-driven crop damage as some studies indicate (Crowder et al., 2010; Lundgren and Fausti, 2015). Diverse and more evenly distributed pest communities have higher inter-species competition, which reduces the dominance and impact of single and most impactful pest species. In our study, fields with species poor pest communities (2–3 species) and with greater mean abundances (mean 132 ± 74 individuals) were dominated by Diptera, Orthoptera and Lepidoptera groups, and had 4% higher crop damage and 28% lower annual earnings than fields with species-rich pest communities. This supports the possibility that lower pest richness with greater density and dominance of few and more specialized crop herbivores like locusts, fall armyworm, stalk borers and African cotton bollworm may lead to greater crop damage and yield losses. Our findings thereby suggest that diversity decreases herbivory potentially diluting the impact of dominant pests on crop production (Civitello et al., 2015). Therefore, scaling up of CSA must take in to account broader landscape-scale perspective and focus on maintaining heterogeneity in agricultural landscapes through mosaics of monocrop and intercrop fields intermixed with natural and semi-natural landscapes.

Use of improved crop varieties is widely promoted as a climate adaptation (Thierfelder et al., 2016). Our study showed that use of improved crop varieties for drought resistance and shorter growing season in a neighboring field was associated with increased pest richness and reduced yields in the focal fields with local varieties, mainly of maize. This could mean that a shift to improved varieties among one group of farmers may lead to increased vulnerability of another group, thereby highlighting a form of maladaptation. Further studies are needed to explore the spatial flows and interactions of ecosystem services and disservices due to the adoption of new technologies.

We also found a strong effect of crop seasonality as months coinciding with maize plantation (January) and harvest (September), which are themselves timed with respect to expected rains, had the highest levels of pest and natural enemy diversity, abundance, and volume. The effect of sampling season has also been observed across the globe and shows that it is important to take in to account the temporal variability when evaluating ecological responses to agricultural interventions (Habel et al., 2018; Fanelli, 2020; Scheiner and Martin, 2020). Further, one of the sampling occasions, November 2019, coincided with flash floods in the study area during which the invertebrate communities significantly declined in richness, abundance, and volumes. This sudden decline was more likely due to the destructive effect of flash floods on invertebrate habitats as observed in other studies (Hickey and Salas, 1995; Moi et al., 2020). It also reveals the fragility of any farming system, whether they are CSA or not, in the face of increasing climate extremes.

Adoption of CSA may be necessary in the future to reduce agriculture's vulnerability to climate change and ensure food security for an estimated 9 billion people by 2050 (Taylor, 2018). CSA, presently, is focused primarily on technical fixes to increase agricultural productivity and adaptation (Dinesh et al., 2015). However, food production under climate change will also depend largely on sustainably increasing farmer's resilience to climate risks by developing adaptation strategies and managing trade-offs (risks and conflicts) between land use change, biodiversity, and food production (Chandra et al., 2018; Taylor, 2018). Our study highlights the need for agriculture industry to evaluate interactions, synergies, and trade-offs between the different outcomes relating agriculture change from a multi-dimensional perspective. Understanding of risks and challenges farmers face, adaptation strategies they use, and which strategies provide the best outcome in terms of productivity, food security net agricultural income, and environmental stability will be crucial for building more resilient and sustainable food systems of future (MacFall et al., 2015; Aggarwal et al., 2018).

The present study is an evaluation of management practices at local scales. Even though we sampled for multiple months in a year there may be variations and multi-year effects we could not account for. The CSA practices in the study areas were introduced only 3 years before the present study was implemented. Although the three-year time frame may be enough to effect the changes in some invertebrate communities - pests especially, there may still be time lags in responses of indirectly affected trophic groups (e.g., natural enemy communities) causing delayed biodiversity changes (Essl et al., 2015). Overlooking time lags may result in an underestimate of biodiversity impacts and associated trade-offs with food production. Hence, longer-term studies are needed to fully understand the ecological implications and biodiversity impacts of agricultural interventions. Our study, nevertheless, points toward possible trends.

Our study identifies mulching and Fanya juu trenches as CSA practices with potential for maximizing synergies among food production, biodiversity conservation and pest suppression via biocontrol. Furthermore, it also shows that that increasing inter-row spacing of Maize to 70–100 cm and integrating food crops like maize and beans with cash crops, such as spice plantations, improves grain yield and annual income. Broadly, our study highlights the importance of local management and landscape characteristics in maintaining biodiversity. Our study has also uncovered the real risk of negative impacts of CSA intervention and maladaptation on farmer livelihoods. In this study, increased pest richness and abundance, lower yields and income were evident in local-variety fields neighboring those where improved varieties are used, suggesting the use of improved varieties by farmers is creating increased vulnerability among neighboring farmers. The need for further research on assessing biodiversity impacts, and for holistic and long-term evaluation and monitoring of CSA across field and landscape scales is therefore very clear. CSA evaluations must look beyond productivity alone because the trade-offs and synergies among biodiversity, agricultural production and food security, livelihoods and health often interact and feedback on each other. Finally, more local-scale case and long-term studies and syntheses are needed to understand the combined and interactive effects of agriculture intervention and global change processes across different geographies and food production models.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

HT conceived the idea, designed, and implemented this study with support from SSai, WK, SSal, SMau, and HS. All authors have contributed ideas that inform the structure and narrative of the paper, provided editorial support, agreed to the published version of the manuscript, and contributed to the article and approved the submitted version.

This work was funded by the Biotechnology and Biological Sciences Research Council through UK Research and Innovation, as part of the Global Challenges Research Fund: AFRICAP Programme, Grant Number BB/P027784/1. Research in Tanzania was conducted under COSTECH permit number: 2019-305-NA-2016-101. RD was supported by the University of Leeds MSc in Biodiversity and Conservation programme. Fieldwork in Tanzania was supported by the Economic and Social Research Foundation. The funders had no role in the design of the study, in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.