Tree islands enhance biodiversity and functioning in oil palm landscapes

The rapid expansion of oil palm has accelerated the loss of tropical lowland rainforests, reducing biodiversity and altering ecosystem functions in affected landscapes. While oil palm supports economic development, it causes strong declines in species diversity and ecosystem services relative to forests. There is an urgent need for restoration strategies that can enhance biodiversity and ecosystem functioning without undermining agricultural productivity. One proposed approach is establishing native tree islands (through planting or natural regeneration) and species-rich agroforestry within plantations. However, robust empirical evidence is needed to assess ecological outcomes and potential productivity trade-offs. This study tests whether tree islands increase multidiversity and multifunctionality relative to conventional oil palm, how island area and planted tree diversity influence outcomes (grounded in island biogeography expectations that larger islands harbor more diversity and functions), and whether oil palm yields are compromised at local versus landscape scales.

The study builds on work showing oil palm expansion’s ecological impacts and the need for restoration in tropical agroecosystems. Prior research suggests that species-rich agroforestry and applied nucleation (tree islets) can facilitate biodiversity recovery, with theory (island biogeography) predicting larger habitat patches support more species and functions. Plant diversity can influence higher trophic levels and ecosystem functioning via complementarity, and structural complexity mediates many biodiversity–function relationships. Evidence from oil palm systems highlights trade-offs between multifunctionality and profit, but also indicates potential benefits of diversification, riparian buffers, and optimized management. The study situates tree islands as complementary to forest protection and certification schemes, noting the need to evaluate ecological outcomes together with productivity at relevant spatial scales.

Design: A large-scale ecosystem restoration experiment was conducted within a 140-ha industrial oil palm plantation in Sumatra, Indonesia. Fifty-two tree islands were established 3–5 years prior to assessment, varying orthogonally in area (25, 100, 400, 1,600 m²) and planted native tree species richness (0, 1, 2, 3, 6 species; zero denotes natural regeneration only). Conventionally managed oil palm monoculture plots served as controls. Some oil palms within island footprints were felled to create space for planted trees; adjacent palms remained. Measurements: Biodiversity was measured using 10 indicators spanning bacteria, fungi, plants, and animals. Ecosystem functioning was quantified with 19 indicators across domains: productivity (oil palm yield, aboveground biomass), resistance to invasion (native seed rain, resistance to invasive plants), pollination (pollinator abundance/activity, pollination rate), soil quality (soil P, decompaction, 1/soil C:N), predation/herbivory (vertebrate predators, arthropod predators, soil predators, soil herbivores), carbon/nutrient cycling (decomposers, litter decomposition, litter input), and water/climate regulation (evapotranspiration, water infiltration, microclimate buffering). Twelve vegetation structure indicators characterized structural complexity and tree vs oil palm dominance. Composite metrics: Multidiversity and multifunctionality were calculated as the number of indicators exceeding thresholds expressed as percentages of the maximum observed across islands and controls. Analyses considered species richness and abundance-weighted diversity (Shannon, Simpson). Yield: Oil palm yield was assessed per area within islands (local scale) and per island including adjacent palms (landscape-relevant compensation), with attention to thinning effects. Analyses: Effects of treatment (tree islands vs controls), island area, and planted tree diversity on indicators and composite metrics were tested using ANOVA, reporting F- and P-values. Structural equation models (SEMs) evaluated direct and indirect pathways: island area via tree dominance; planted tree diversity via structural complexity. Rarefaction/constant sampling area approaches were used to rule out passive sampling as the driver of area–biodiversity relationships.

• Tree islands increased biodiversity and ecosystem functioning relative to conventional oil palm monocultures. Ecosystem functioning overall: tree island effect F=6.2, P=0.016.
• Biodiversity responses varied by indicator: tree island × indicator interactions significant (species richness F=2.5, P=0.007; Shannon F=3.6, P=0.0002; Simpson F=3.0, P=0.001).
• Specific changes: +4.7 tree species and +2.5 bird species (species richness) in islands vs monocultures; decreased diversity of the most abundant seed species (−1.2 seed species based on Simpson diversity).
• Strongest functional increases in islands: water infiltration (+174% saturated hydraulic conductivity), litter input (+151% leaf litter biomass), activity of insectivorous bats and birds (+556%), and soil fertility (+14% 1/soil C:N).
• Multidiversity and multifunctionality were higher in islands across thresholds. At 50% threshold: biodiversity indicators reaching ≥50% of max were 4 in islands vs 2.5 in controls; ecosystem functioning indicators 6 vs 3.
• Larger island area increased restoration benefits: ecosystem functioning F=12.9, P<0.0001; biodiversity effects significant with indicator-dependent interactions (species richness island area×indicator F=5.1, P<0.0001; Shannon F=2.8, P=0.003; Simpson F=1.8, P=0.06).
• SEMs indicated island area influenced outcomes indirectly via increased tree dominance; larger islands had greater canopy tree dominance and thicker litter, enhancing multidiversity. Large islands particularly boosted functions related to predation/herbivory (predatory arthropods, soil herbivores) and carbon/nutrient cycling (decomposers).
• Planted tree diversity effects depended on indicator when considering abundances (planted diversity×indicator: Shannon F=2.3, P=0.014; Simpson F=2.8, P=0.004). Higher planted diversity increased structural complexity, benefiting some taxa (bats, herbs; abundance-weighted diversity), while more open structures favored others (seed rain, understorey arthropods). Through structural complexity, tree diversity had a negative indirect effect on multifunctionality (method-dependent).
• Yield: Within-island per-area oil palm yield was on average 24% lower than in conventional plots due to reduced palm density. However, per-island yield (including adjacent palms) showed no significant reduction; yield gains of surrounding palms compensated local losses, attributable to thinning effects. Over time, within-island yields may further decrease due to tree competition (especially at higher planted diversity), but effects remain negligible at plantation scale given small island area (2.8 ha <5% of 140 ha).

The experiment demonstrates that establishing native tree islands in oil palm plantations enhances multidiversity and multifunctionality without compromising landscape-level oil palm yield. Larger islands improve outcomes mainly by shifting canopy dominance from oil palms to trees, increasing habitat quantity/quality and structural attributes (e.g., litter layer), consistent with island biogeography and reduced edge effects. Planted tree diversity shapes habitat structural complexity, leading to taxon-specific and function-specific responses: structurally complex habitats favor some biodiversity components and functions (biomass, litter input, microclimate buffering), while more open habitats benefit others (seed rain, pollinators, arthropod predators). Consequently, a mosaic of island sizes and structural complexities may maximize landscape-scale gamma diversity and multifunctionality. The lack of yield penalty at the landscape-relevant scale, despite local reductions, supports the integration of tree islands as a viable restoration strategy in industrial plantations. These findings align with and extend previous evidence on diversification benefits and underscore the complementary role of restoration within agricultural matrices alongside strict forest protection and riparian buffers.

Planting native tree islands in oil palm-dominated landscapes yields robust, multidimensional ecological benefits—higher biodiversity, improved ecosystem functions, and increased multidiversity/multifunctionality—while maintaining overall oil palm productivity at the plantation scale. Larger islands provide disproportionate gains via increased tree dominance and habitat quality, and varying planted diversity modulates structural complexity with taxon-specific outcomes. Tree islands should complement, not replace, the protection of intact forests. Scaling up will benefit from integrating such set-asides into sustainability certification and management practices, designing mosaics of island sizes and structures, and expanding research across multiple landscapes and time horizons to optimize biodiversity–function–productivity synergies.

• Spatial scope: results derive from a single industrial plantation landscape in Sumatra, potentially limiting generalizability across regions and management contexts.
• Temporal scope: assessments occurred 3–5 years post-establishment; longer-term dynamics (e.g., tree maturation, competition effects on yields and functions) remain uncertain.
• Method dependence: multifunctionality outcomes varied with calculation method; planted diversity effects on multifunctionality were sensitive to analytical approach.
• Yield trade-offs: within-island local yield reductions (−24%) and potential further declines with tree growth may pose challenges for smallholders with limited land, though co-products and ecosystem service gains may offset losses.
• Indicator-specific responses: contrasting taxon responses to structural complexity suggest that single-structure designs may not optimize all biodiversity components simultaneously.