Win-win opportunities combining high yields with high multi-taxa biodiversity in tropical agroforestry

Agricultural expansion and intensification are the primary drivers of the current biodiversity crisis. Increased agricultural productivity often comes at the expense of biodiversity, creating urgent ecological-economic trade-offs, particularly in rapidly transforming tropical landscapes. The UN Decade on Ecosystem Restoration (2021–2030) emphasizes the need for solutions that prevent, halt, and reverse ecosystem degradation. Agroforestry offers promising restoration opportunities, but requires more system-specific knowledge for broader implementation of win-win solutions. Degraded lands, common in tropical regions due to subsistence agriculture and shifting cultivation, are particularly important targets for restoration. However, agroforestry can also lead to biodiversity loss if established by clearing forests. The land-use history of agroforests—whether established on open land or by forest thinning—significantly impacts biodiversity outcomes. Low-yielding agroforestry systems might necessitate more land, potentially increasing forest conversion and overall biodiversity loss. This study focuses on vanilla agroforestry in Madagascar, a tropical biodiversity hotspot with exceptionally high endemism rates, but facing significant challenges in biodiversity conservation and human development due to extreme poverty. Madagascar is the world's largest vanilla producer, with smallholders responsible for most of the production. The high market price of vanilla has boosted the socioeconomic benefits for Malagasy smallholders, leading to expansion of vanilla cultivation. Vanilla, a hemi-epiphytic orchid, is typically grown in agroforests, either by converting forest or utilizing fallow land. Converting fallow land into vanilla agroforests can partially restore biodiversity and ecosystem functions, unlike forest-derived agroforestry which contributes to forest degradation. Extensive fallow land, a byproduct of shifting cultivation for hill rice production, is widespread in northeastern Madagascar. This has resulted in significant old-growth forest loss, necessitating land-use solutions that balance conservation with agricultural production. This research quantifies the effect of vanilla cultivation on multiple taxa and uses yield data from 30 vanilla agroforests to identify yield-biodiversity trade-offs, assessing the biodiversity value of forest and fallow-derived agroforests compared to old-growth forest, forest fragments, and fallow land. The study distinguishes between overall species richness and endemic species richness to account for Madagascar's high endemism and the vulnerability of endemic species to land-use change.

The study's introduction provides a thorough review of the existing literature on the biodiversity crisis, particularly in tropical agricultural landscapes. It highlights the ecological-economic trade-offs inherent in increasing agricultural productivity and the need for solutions that balance both. The literature review emphasizes the role of agroforestry as a potential solution and the importance of land-use history in determining its success or failure. Existing research on agroforestry systems, including those involving cacao, is discussed, noting the variable results found in studies across different regions. The review stresses the need for more system-specific knowledge tailored to the unique characteristics of different regions and crops. Specific attention is given to Madagascar's unique biodiversity challenges and the potential for vanilla agroforestry to address them while also improving the livelihoods of smallholder farmers. The importance of understanding both overall species richness and endemic species richness in the context of Madagascar's high endemism is clearly established.

This study was conducted in northeastern Madagascar's SAVA region, a biodiversity hotspot characterized by tropical lowland rainforest and high deforestation rates. Ten villages were selected based on criteria including the presence of vanilla agroforests, forest fragments, and fallows, and the absence of coconut plantations and extensive water bodies. In each village, three vanilla agroforests (with varying canopy closure), one forest fragment, and two fallows (herbaceous and woody) were selected. Additionally, ten plots were studied in a protected old-growth forest (Marojejy National Park). Plots were established with a 25-meter radius, and data was collected on seven taxa: trees, herbaceous plants, birds, amphibians, reptiles, butterflies, and ants. Vanilla yield, vine length, vine age, and planting density were assessed on 36 vanilla plants per agroforest plot. Canopy closure was measured using hemispherical photography, while slope and elevation were determined using a digital surface model. Landscape forest cover within a 250-meter radius was also calculated. Understory vegetation cover was visually estimated. Soil samples were collected and analyzed for various chemical properties. The tree inventory involved identifying all trees with stems ≥ 28 cm DBH, while herbaceous plants were sampled in 32 m² subplots. Bird data were collected using point counts, amphibian and reptile data via time-standardized search walks, butterfly data with fruit traps and netting, and ant data with bait and pitfall traps. Species endemism was determined for each taxon. Statistical analyses included glmmTMB models and linear mixed-effects models to analyze biodiversity-yield relationships, accounting for land-use history and various environmental and management variables. Differences in species richness between land-use types were assessed using appropriate statistical tests.

The study revealed that higher vanilla yields were not generally associated with decreased species richness or endemic species richness for trees, herbaceous plants, birds, amphibians, or ants. However, negative relationships were found between vanilla yield and butterfly richness, and a variable relationship existed between yield and reptile richness, depending on land-use history. The overall and endemic diversity of all taxa combined was also unrelated to vanilla yields. Compared to old-growth forests, forest-derived vanilla agroforests showed significant losses in species richness for several taxa (birds, trees, amphibians), but butterflies showed a significant increase. Fallow-derived vanilla agroforests, however, exhibited significant gains in overall and endemic species richness for several taxa compared to fallows. Gamma diversity was generally highest in old-growth forests. Species composition differed significantly across land-use types for most taxa. Vanilla yields increased with planting density and vanilla vine length, but were unrelated to canopy closure, landscape forest cover, or other environmental factors. Trade-offs between yield-increasing variables and tree and reptile richness were observed, but positive relationships were found with endemic herbaceous plants. Other taxa were largely influenced by non-yield-related management and environmental variables such as canopy closure and landscape forest cover. Specifically, higher landscape forest cover and canopy closure positively influenced endemic herbaceous plants, trees, reptiles and ants. The study also found that vanilla age was unexpectedly unrelated to vanilla yield, while yield tended to increase with labor input for hand-pollination. The proportion of endemic species varied widely among taxa, with amphibians exhibiting the highest proportion (98%). Sampling coverage was satisfactory for most taxa, though some rarefaction curves indicated that increasing sampling effort could reveal further differences in biodiversity.

The findings challenge the common expectation of ecological-economic trade-offs in tropical agriculture. The study shows that increasing vanilla yields within the current management practices is not generally associated with biodiversity losses. It highlights the potential of vanilla agroforestry to restore biodiversity on degraded fallow lands, a prevalent land-use type in the study region. This supports the goals of the UN Decade on Ecosystem Restoration. The study found that while vanilla yield increased with higher planting density and vine length, this resulted in trade-offs for tree and reptile diversity. However, other factors largely determined biodiversity in vanilla agroforests, offering opportunities for farmers to increase yields without compromising biodiversity. The results corroborate findings from other agroforestry systems, such as cacao, where yield and biodiversity were often found to be unrelated. Madagascar's vanilla agroforestry system differs from others because shade vegetation is unrelated to yield, suggesting a high intensification potential without biodiversity trade-offs. The study emphasizes the significance of considering land-use history in agroforestry research and management, with fallow-derived systems offering considerable conservation opportunities. The high profitability and permanence of vanilla agroforests offer a potential way to break the cycle of shifting cultivation and forest degradation. The study also calls for policy support and economic incentives, such as contract farming or payments for ecosystem services, to promote the establishment of fallow-derived vanilla agroforestry.

This study demonstrates the potential for win-win solutions in tropical agriculture, combining high yields with high biodiversity. While strict protection is needed for remaining old-growth forests, fallow-derived vanilla agroforestry offers a valuable tool for biodiversity restoration within agricultural landscapes. Management strategies can be employed to simultaneously increase both yields and biodiversity. Vanilla agroforestry presents significant opportunities for ecologically and economically sustainable land management in Madagascar, aligning with global ecosystem restoration goals.

The study's reliance on self-reported data for land-use history, while confirmed by other observations, could introduce some uncertainty. The sampling design, while comprehensive for the study area, may not be fully generalizable to other regions with differing environmental or socioeconomic conditions. The focus on a single crop, vanilla, limits the generalizability of the findings to other agroforestry systems. Certain rarefaction curves did not reach an asymptote, suggesting further biodiversity differences could emerge with increased sampling effort.