Trade-offs between multifunctionality and profit in tropical smallholder landscapes

The study addresses how recent land-use transitions driven by smallholder farmers in tropical regions, especially shifts from traditional low-input systems to monocultures (rubber and oil palm), affect biodiversity, ecosystem functioning, and profitability. While smallholders manage a large share of agricultural land, the economic-ecological trade-offs of their land-use choices are not well quantified. The research seeks to determine the shape and magnitude of relationships between profit and ecological outcomes (biodiversity and ecosystem functions), including whole-ecosystem multidiversity and multifunctionality, and to identify landscape compositions that could mitigate trade-offs under varying profit expectations. The context is Jambi Province, Sumatra, Indonesia, a biodiversity hotspot undergoing rapid conversion from rainforest to plantations, with implications for achieving SDGs that balance livelihoods with ecosystem conservation.

The paper situates its contribution in literature on agricultural expansion and tropical biodiversity loss, noting the predominance of studies on large-scale estates while smallholder-driven transitions remain understudied. Prior work in the same region demonstrated that land-use choices follow profitability at the expense of ecological functions. Evidence of potential win-wins exists in some agroforestry systems (e.g., cocoa), but outcomes vary by crop and management intensity. The authors highlight gaps in linking economic functions directly to ecological outcomes across multiple taxa and functions, and in integrating whole-ecosystem measures (multidiversity and multifunctionality) across land uses and transitional stages.

Study region and design: Fieldwork in lowland Jambi, Sumatra, focused on two landscapes (Harapan and Bukit Duabelas) with four land-use systems: primary degraded lowland rainforest, jungle rubber (agroforestry), rubber monoculture, and oil palm monoculture. In 2012, 32 permanent 50 m × 50 m plots were established (4 land uses × 2 landscapes × 4 replicates). Monoculture rubber plots were 7–16 years old; oil palm 8–15 years.

Biodiversity sampling: A total of 26,894 species/OTUs across 14 taxa covering above- and belowground biota and trophic levels were surveyed. Methods included: tree inventories (DBH ≥10 cm), understory vascular plants, canopy ants and parasitoid wasps via canopy fogging (dry and rainy seasons), birds via point counts and automated recordings, bats via mist nets/harp traps, butterflies via sweep netting, litter macroinvertebrates via litter sieving and morphospecies identification, soil protists (testate amoebae), soil microarthropods (Oribatida, Mesostigmata), soil fungi (ITS metabarcoding), bacteria and archaea (16S rRNA metabarcoding). Standardization and replication procedures were applied per plot and taxon.

Ecosystem functions: 36 indicators across 10 functions were measured, including net primary production (litterfall, fine roots, latex and fruit harvests, stem increment), plant biomass C stocks (above-/belowground, fine roots, latex, fruit), soil organic carbon stocks to 2 m, soil fertility metrics (net N mineralization, extractable P, exchangeable cations), soil respiration and greenhouse gas fluxes (CO2, CH4, N2O), nutrient leaching (biweekly-monthly via suction lysimeters to 1.5 m; TDN, NH4+, NO3−, DOC, base cations, P, S with modeled drainage), decomposition (litterbags), plant transpiration (sap flux methods scaled to stand transpiration), and microclimatic conditions (air/soil temperature, humidity, soil moisture). Measurements largely spanned 2013–2016 depending on indicator.

Yields and profits: Rubber and oil palm yields were continuously monitored in the same plots from July 2014 to June 2016. Profit data came from two rounds (2012, 2015) of household surveys with 701 randomly selected smallholder households across 45 villages in five regencies, collecting detailed plot-level input-output data, prices, and management practices. Annual profits per hectare were calculated as revenues minus production costs (including priced family labor; 2015 exchange rate 1 USD = 13,389.413 IDR). Forest profits were set to zero (<1% marketed products).

Stakeholder interviews: 150 qualitative interviews (2012–2016) with households and actors at district, provincial, and national levels (government, NGOs, corporations) explored land-use history, spatial planning, drivers of transformation, and conflicts.

Statistical modeling: Yield-profit relationships for jungle rubber, rubber, and oil palm were modeled via linear regression using household survey data, revealing strong positive linear relationships with heteroscedastic residuals. Profits for ecological plots were predicted from measured yields using these relationships, including heteroscedastic variance. Because profits as predictors contained measurement error, Simulation-Extrapolation (SIMEX) was applied to correct bias in regression coefficients when relating profit to biodiversity (species richness; negative binomial GAMs with penalized splines) and to ecosystem function indicators (GAMs). SIMEX used λ in [0.1, 3], B=200 bootstraps per λ, quadratic extrapolation to λ=0.

Multidiversity and multifunctionality: For each plot, multidiversity (14 taxa) and multifunctionality (36 indicators, 10 functions; undesirable indicators inverted and indicators weighted within functions) were calculated as the proportion of taxa/functions exceeding thresholds from 1% to 99% of the mean of the top five values (to reduce outlier influence). Profitability relationships were assessed with simple linear regression across the full threshold range.

Landscape optimization: Conceptual 32-plot landscapes were generated in silico using a binary genetic algorithm (population size 500, 100 generations) to maximize biodiversity (per taxon), ecosystem function performance (weighted indicator sums), multidiversity, or multifunctionality, subject to minimum average profit constraints (0, 200, 400, 600, 800, 1000 USD ha−1 yr−1). Plot-level biodiversity/function and profit data were used with replacement to fill slots. The algorithm identified Pareto-optimal landscape compositions under each profit expectation.

• Biodiversity-profit trade-offs: Across most of the 14 taxonomic groups (total 26,894 species/OTUs), species richness declined non-linearly with increasing profit, with the steepest losses when transitioning from forest and jungle rubber to monocultures. Rubber and especially oil palm monocultures achieved higher profits (up to ~1000 USD ha−1 yr−1) but harbored lower biodiversity. When focusing on species also present in rainforest (47% of all species), negative relationships with profit were pervasive across all groups, indicating strong losses of rainforest species in monocultures.
• Ecosystem function-profit trade-offs: Among 36 indicators across 10 functions, many showed undesirable trade-offs with profit: soil and aboveground carbon stocks, soil respiration, and decomposition declined, while nutrient leaching and greenhouse gas fluxes increased with higher profits. Soil fertility indicators improved with profit largely due to inputs applied in oil palm on acidic Acrisols. Some indicators showed hump- or U-shaped responses (e.g., plant transpiration, microclimate), reflecting complex system-specific dynamics.
• Whole-ecosystem indices: Multidiversity and multifunctionality both declined with increasing profits across the entire threshold range (1–99%), with strongest negative slopes around 30–70% thresholds. Increasing profitability consistently reduced overall ecosystem diversity and functioning, even under lenient performance expectations.
• Landscape optimization: Maintaining high multidiversity required high proportions of rainforest in optimized landscapes. As profit expectations rose, rainforest share and multidiversity decreased in parallel; medium to high-profit (≥400–800 USD ha−1 yr−1) optimized landscapes were dominated by oil palm. For individual functions, optimal compositions varied: rubber-dominated landscapes favored high soil respiration and low nutrient leaching, whereas oil palm entered only at high profit constraints (>800 USD ha−1 yr−1), often with function losses. Some functions (e.g., limiting high soil greenhouse gas fluxes) could be sustained at relatively high profits, while others (e.g., decomposition, organic carbon storage) fell below 50% of their maximum potential even under optimal allocation at similar profit targets. Multifunctionality declined linearly from low- to high-profit landscapes.
• Social context: Oil palm expansion increased profits and is associated with improved household incomes, food security, and consumption for adopters in Jambi; at the national scale it may have reduced rural poverty. However, interviews revealed rising social inequality and land tenure conflicts linked to expansion.
• Land-use change context: From 1990 to 2013 in Jambi, rainforest cover fell from 49.5% to 34.5%, while rubber and oil palm increased from 26.4% to 32.5%. By 2017, ~99% of rubber and ~61% of oil palm area was smallholder-managed.

The findings demonstrate robust, often non-linear economic-ecological trade-offs in smallholder-dominated tropical landscapes: profit gains from transitioning to rubber and oil palm monocultures come at substantial costs to biodiversity, rainforest species persistence, and multiple ecosystem functions. Whole-ecosystem multidiversity and multifunctionality decline consistently with profit increases, challenging the sustainability of current development trajectories. Optimized landscape simulations suggest that biodiversity losses can be mitigated most effectively by retaining significant rainforest fractions and deriving profits primarily from oil palm to meet higher profit targets; however, maintaining high multifunctionality lacks a single optimal composition, as functions respond divergently to land-use mixes. These insights underscore the need for landscape planning that explicitly accounts for trade-offs and for policy instruments that realign private incentives with conservation and ecosystem functioning goals. While oil palm adoption improves several welfare dimensions for adopters, social disparities and land conflicts underline the importance of equitable governance and tenure security. The general patterns likely extend to other Indonesian and tropical lowland regions, though context-specific variations should be expected.

This interdisciplinary study quantifies the shape and magnitude of trade-offs between profitability and ecological outcomes across an unprecedented breadth of taxa and ecosystem functions in a tropical smallholder landscape. It integrates continuous yield monitoring, extensive biodiversity and ecosystem function measurements, profit modeling with measurement error correction, and landscape optimization. Key contributions include demonstrating pervasive biodiversity-profit and function-profit trade-offs, consistent declines in multidiversity and multifunctionality with rising profits, and identifying landscape compositions that can partially mitigate trade-offs under profit constraints. Policy implications center on redesigning economic incentives through combinations of regulatory measures, payments for environmental services, and certification schemes, supported by effective law enforcement and spatial planning that internalizes multifunctionality-profit trade-offs. Future research should evaluate the effectiveness and equity of incentive instruments, test transferability across regions and crops, monitor long-term ecological and socioeconomic outcomes, and refine optimization frameworks to include spatial configuration, connectivity, and social constraints.

• Geographic specificity: Results are specific to Jambi Province; extrapolation to other regions should consider ecological, socioeconomic, and policy differences.
• Land-cover mapping: The 2013 land-cover map was validated (overall accuracy 82.6%, Kappa 0.79), but the 1990 map’s accuracy was assumed comparable, introducing uncertainty in long-term change estimates.
• Profit estimation: Plot-level profits for ecological plots were predicted from yield–profit relationships (2015 prices) using SIMEX to address measurement error; residual biases and temporal price variability remain possible.
• Temporal coverage: Ecological and yield measurements covered limited periods (mostly 2013–2016); interannual variability and longer-term dynamics may affect outcomes.
• Taxa and functions scope: Although broad (14 taxa, 36 indicators), unmeasured taxa/functions and landscape spatial configuration effects (beyond composition) were not explicitly modeled.
• Smallholder focus: Large estate management practices may produce different relationships; this study emphasizes smallholder-managed systems.