Weak edge effects on trees in Bornean rainforest remnants bordering oil palm

Many tropical forests are dominated by edge habitat, with consequences for forest structure, carbon stocks, and biodiversity. However, effects highly variable context-dependent, poorly quantified in oil palm landscapes. We studied 10 lowland rainforest remnants bordering mature plantations on Borneo, surveying 0.2 ha plots along transects running perpendicular to the (ten 1.6 km transects, 5-6 per transect; 57 total). examined how proximity affected plot-level structure (canopy cover, number size of stems ≥10 cm diameter), aboveground microclimate (air temperature light intensity), tree community composition richness. The largest trees were significantly smaller (up 21% reduced diameter) near edges, was up 30% lower (model-fitted average = 64.7 Mg ha−1 at 50 m from edge, versus 92.3 1600 m), strongest within 300m edges. these significant relatively small context existing variation, distance-from-edge explaining <13% total variability maximum or carbon. In addition, there generally no any other component diversity, only a weak effect microclimate. conclude that limited this system may reflect low structural contrast between palm, invasion pioneer plantations, which diminished influence heterogeneous remnants. Abstract Malay is available online material. Sebahagian besar kawasan hutan tropika dipelopori oleh habitat pinggir, yang turut mempengaruhi struktur kanopi hutan, stok karbon dan kepelbagaian biologi. Walau bagaimanapun, kesan pinggir seperti ini sentiasa bergolak mengikut keadaan semasa, kajian seumpamanya dikawasan bersempadan dengan ladang kelapa sawit masih kurang. Di tanah pamah kami telah memilih di sawit. Tinjauan dilakukan melalui penubuhan plot-plot sepanjang transek bersebelahan ( i.e., transek; plot bagi setiap jumlah plot). Kami mengkaji bagaimana diperingkat akan (penutup kanopi, bilangan saiz batang ⩾ atas tanah, iklim mikro (suhu udara keamatan cahaya), komposisi komuniti pokok kekayaan spesis. Jumlah terbesar jauh lebih kecil (sehingga diameter semakin berkurangan) berhampiran peringkat juga adalah sehingga rendah (purata model-dipasang ha-1 pada dari berbanding ketara dalam lingkungan 300 daripada jarak pinggir. agak konteks variasi sedia ada, dimana terdapat kebolehubahan maksimum atau karbon. Secara amnya tiada terhadap mana-mana komponen lain, seperti, kepelbagaian, hanya sangat kecil. berpendapat bahawa terhad mungkin boleh mengambarkan perbezaan fungsi antara matang, penyesaran perintis terhad. Keadaan hujan pelbagai. Agricultural expansion has caused widespread loss rainforests, support an outstanding diversity species provide valuable ecosystem services (Lewis et al., 2015). This predicted continue coming decades food resources growing human population (Laurance 2014). Deforestation agriculture causes reduction area increased fragmentation remaining forest, biodiversity key functions such as storage (Haddad Forest creates where abiotic biotic changes (“edge effects”) can drive ecological 2002; Pfeifer 2017). Approximately 20% 100m (Brinck 2017); it therefore critical we understand impacts edges have communities, both quantify environmental agricultural devise management strategies limit detrimental effects. typically hotter, drier, brighter windier than interior vulnerability desiccation fire 2002). Abiotic lead effects, elevated recruitment mortality, causing rapid turnover 2006). manifests declines slow-growing, late successional higher wood densities, whilst fast-growing pioneers better suited disturbed conditions become more dominant (Tabarelli 2012). Thus, long-term compositional shifts occur, resulting taxonomically functionally distinct communities (Santos 2008), stand-level (e.g., stem number, size, canopy density) (Broadbent 2008) richness (Oliveira 2004), but not always (Ibáñez Increased mortality large high density cause biomass (de Paula 2011), magnitude variable, be mediated adjacent matrix (i.e., land cover) type (Melito 2018). Not do vary magnitude, also often unclear far they permeate into forest. studies report penetration distances less 500 m, albeit considerable variation exact 2008). Estimates remotely-sensed data, however, indicate losses >10% penetrate least 1.5 among regions (Chaplin-Kramer Hence, commonly reported (Ries 2004); part due study designs context-dependent nature Landscape configuration exert strong depth, define assess their significance (Harper 2005). Given becoming increasingly common (Fischer 2021), patches important refuges wildlife contribute stocks landscapes (Fleiss 2020), examine across range human-modified To date, been little research border (Elaeis guineensis) plantations. surprising given world's most consumed vegetable (Meijaard estimated cover over 20 M globally (Descals 2021) around half all new occupying converted Oil perennial crop reach 13m height industrial forming closed stable understory (Luskin & Potts, 2011). It buffer severity neighboring (Fitzherbert lack found Fleiss al. (2020) data Borneo suggest (AGC) remnants, 22.5% 114 (Ordway Asner, 2020). Similarly, Nunes (2021) had growth during 2015-16 El Niño event. evidence accumulation 448 (Qie 2017), suggesting toward taxa. Qie (2017) included types study, inhabited areas, clear cut logging, regenerating logging roads. if plant consensus scale operate. industry expected consider when quantifying expansion. Edge account 19% production greenhouse gas (GHG) emissions, GHG footprint (Lam 2019); will benefit robust estimates AGC Roundtable Sustainable Palm (RSPO) Zero-Deforestation Commitments requires “High Conservation Values” biodiversity) Carbon Stocks” conserved (Rosoman 2017; RSPO, levels could enhanced understanding landscapes, potentially enabling conservation practices outcomes. conducted field surveys Sabah, Malaysian Borneo. Forests region contain some highest (Asner 2018) hotspots (Myers 2000). Sabah lost approximately 40% its since 1973 (Gaveau 2014) fragmented plantation now accounting >20% Sabah's (MPOB, 2019). increasing change, because persist test hypotheses fewer plots, temperatures, levels. hypothesis taxonomic functional trees, taxa, (<500 a.s.l.) mixed dipterocarp State (Figure S1; Table S1), June October 2019. All sites reserves protected disturbance timber hunting), containing areas except Site 6 1 Whilst detailed histories available, likely selectively logged intensities past commercial site 1984, formally declared although many experienced encroachment felling hunting) (Stride Edges 19–49 years old (mean 36 years), bordered expanses Neighboring palms 12.6 tall (SD 3.75 m) 7.5 apart, varied vegetation typical (Luke placed single transect each site, >800 avoid multiple (Porensky Young, 2013). Transects >4 apart spatial autocorrelation. defined point natural became taller 5 (UN FAO, 2012); usually characterized hard boundary our sites, sometimes separated minor tracks. comprised six circular (25 radius, ha) spaced concentrate sampling effort close S1c), greatest expected. three contained five either remnant features made final inaccessible. total, sampled (total 11.2 ha). Within plot, used nested survey design maximize efficiency S1d) followed standard protocols (Marthews 2014; Phillips 2016) inventory live (including palms) above breast (dbh). Trees identified genus level, known (23.1% stems), consultation botanists Danum Valley Field Centre herbarium. Tree eye, same person (AJ). AJ's validated against “tangent method”, estimate (Larjavaara Muller-Landau, 2013), closely correlated giving us confidence estimates. For calculated individual sampled, mean dbh (cm) (m), characterize structure. measured using spherical crown densiometer, taking four readings (N, E, S, W) center then converting measure (proportion 0–1). daily air (°C) daytime intensity (lum/ft2) Hobo® loggers (see Appendix S1 details). assigned (g/cm3) values finest level Global Wood Density Database (Chave 2009; Zanne 2009), published allometric equations Goodman 2013) produce (Mg C ha−1) As composition, community-weighted (CWM) (g/cm3), mixed-resolution (12.9% 82.4% 4.6% family level) abundance weightings. trait linked strategy, lower-wood having faster rates, preference earlier status (Slik, genus-level analyses (number genera) diversity. worked challenges reliably identifying analyzing give reliable results species-level identifications (Slik 2003), greater sample sizes (Imai Genus-level disturbance-driven floristic 2006; Michalski 2007; Slik Furthermore, patterns forests, making detectable resolutions (Ganivet 2020; Imai Stride 2018), taxonomically-conserved trait, 72.5% explained Our information. R version 4.0.2 (R Core Team, mixed-effects models on: (stems plot), (cm), 0–1), ha−1), (°C), (nine total; 1). Data analyzed lme4 package (Bates Maximum ln-(natural log) transformed improve model fits, based diagnostic plots. models, ln-transformed distance fixed order linearize edge-response curves, following Ibanez (2017), appropriate concentrated closer edge. fitted error distribution families link (Table 1) ensure assumptions met “site” random intercept site-level response variables. modelling procedure CWM (two models; evaluate similarity relation proximity, computed Bray–Curtis dissimilarity (Magurran, relative abundances genera. performed ordination non-metric multidimensional scaling (NMDS) 1000 iterations, vegan (Oksanen permutational multivariate analysis variance (PERMANOVA) differences classes (treated categorically analysis), 999 permutations calculate significance. determine whether taxa even absence pooled 3 classes: “edge” (50 + 100 plots), “intermediate” (200 400 “core” (800 plots). randomization approach generate (abundance-based) genera amongst categories (Appendix chi-square goodness-of-fit actual differed what would expect chance approach). If gain (as form), see unique class, classes, alone. confirm resolution repeated higher-resolution subsets sensitivity S1). surveyed 2403 representing 138 S3) 54 families. heterogeneity example, 11.3 255.9 81.9 ± 44.4), 8 31 4.8). There according index scores 0.8 0.11, 0–1 scale). components height, declining Model-predicted showed 26.2% decrease 32.9 44.6 m; Figure 1a; p < .01), similar (21.3%) (65.9 compared 83.7 cm; 1b; .01). very resulted 29.9% decline (from vs. 1c; .05). variables weakened distance, change occurring 13% dbh, 10% 5% R2M). significant, proportion stems, density; 1; S2). marginally S2f). increase 25.54°C 25.71°C (p .05; 1d), 0.5% variance. (R2C 0.95), different days. 2b). clustering NMDS 2a), PERMANOVA supported conclusion (R2 0.07, .98), indicating communities. S2e), composition. (edge, intermediate, core); difference observed abundance-based (χ2(6) 7.47, .3). Sensitivity qualitative results, minority did indicative possible biologically driven entirely outlier rare (<1 individual/ha) findings 1c). Ordway Asner reporting primarily those 70 dbh; 2013; 1b). occurred independent shifts, elsewhere (Silva probably make disproportionately contribution Large S5) worst They susceptible wind damage (Gora Esquivel-Muelbert, tend (Magnago Indeed, sensitive drought-induced (Phillips 2010) sheltered speeds (Jackson 2021). reduce created after ceased, subsequent contributed patterns, removal rates accessible. Nonetheless, argue still edge-related effect, creation maintenance finding (21% 26% height) (via edge-facilitated felling), and/or inhibited (Nunes Over 1), supporting 300–500 2021; Small dominate (Taubert without further experience severe degradation stand associated stocks. Effects wider consequences, importance numerous processes (Pinho Contextualizing essential meaningful .05, .01 .01, respectively; baseline Distance-from-edge 13%, 10%, AGC, including accounted additional 7%, 40%, 22% respectively R2 values). drivers inter-site factors selective logging) topography, outweigh Liu Slik, contrary hypotheses, systems absent. density), (<0.2°C temperature). overall appear here think unlikely subset conclusions Other shown determinant fragments, mediating (Driscoll Hatfield Kupfer 2006), so, extent) factor study. developed canopies complexity open habitats pastureland 2002) annual crops sugarcane maintained plantation-forest Structural directly mediates strength gradients control (Arroyo-Rodríguez Mesquita 1999), thus maturation provided influence. 2.8°C hotter once 2011); hence buffered regrowth “sealing” (Didham Lawton, 1999) occur short (Ewers Banks-Leite, act barrier dispersal disturbance-adapted minimizing shifts. Study small, degraded one another (Benchimol Peres, 2015; Laurance Magnago Santos source seed rain facilitate spread (Jesus 2012), propagule pressure (Nascimento prevent establishment monoculture isolation landscape Scriven 2015), 100–1000 (Corlett, 2009). system. suggests resilience note considerably system, local soil type, climate, patch geometry palm. age formation years, AGC. ongoing residual shortly creation. biased older 46–49 ago S1); therefore, about temporal dynamics limited. increases impact particularly younger established 19 ago. Mortality periodic replanting) worsened droughts, frequent climate (Cai 2018; worsen, decays Although present cohorts threshold. seedlings saplings independently adult post-edge altered interactions early life stages (Krishnadas 2019; Luskin (2018) Bornean signaling potential extinction debt. reductions future. Further needed time lags, pose threat integrity Studies should seek management, replanting. nevertheless AGC; implications future sustainable Adoption sustainability criteria, retention boost persistence must considered; reason, recommendations prioritize >200 (Lucey recognizing edges). Most fall well target (Scriven unpublished data), indicated compromise ability irregularly shaped maintain high-carbon continues grow, taken developing Overall, like (2018), vulnerable possibly maintaining longer term. suspect current (extensive mature) themselves (isolated heterogeneous). young lags. ubiquitous, absent scale, inform JAA, CJM, SS, NP, JKH involved project conceptualization administration. assisted supervision, SS funding acquisition. JKH, AJ designed methodology. JAA collected data. JT administration Malaysia. led writing original draft, authors contributing reviewing editing paper. funded Unilever. thank Biodiversity Council, Management Committee, Forestry Department, Wildlife Parks, South East Asia Rainforest Research Partnership (SEARRP) permissions carry out (SaBC Access License ref. no. JKM/MBS.1000-2/2 JLD.8 (91)). Wilmar Int. Ltd access fieldwork, staff Herbarium assistance identification, Hannah Grice collection. Chris Thomas, Colin Beale, Susannah Emily Waddell advice analyses, reviewers, Julia Chatterton, James Joyce, Henry King Unilever, helpful comments manuscript. corresponding author confirms behalf conflict interest. openly Dryad Digital Repository: doi: 10.5061/dryad.gqnk98sqj (Anderson 2022). Please note: publisher responsible content functionality information supplied authors. Any queries (other missing content) directed article.