Higher avian biodiversity, increased shrub cover, and proximity to continuous forest may reduce pest insect crop loss in small-scale oil palm farming

One of the key ecosystem services offered by avian biodiversity within agricultural landscapes is natural predation. Nonetheless, the current use of biological control agents such as farmland birds in oil palm plantations is relatively limited. The present study aimed to assess the potential roles of avian biodiversity, particularly insectivores which provide natural predation against oil palm herbivorous insects. We also investigated the influence of local- and landscape-scale variables on foliage damage (crown or frond). Our data showed that crown damage decreased with increasing farmland bird richness (overall and insectivore), shrub cover, dried biomass, and elevation, but increased with epiphyte cover, oil palm height, and distance to continuous forest. Frond damage was negatively related to bird richness (overall, insectivore, and non-insectivore) and non-insectivore abundance, elevation, and shrub cover, while increased with insectivorous abundance, epiphyte cover, oil palm height, and distance to continuous forest. We also found that plantations (≥ 50 ha) were more susceptible to foliage damage from pest insects than smallholdings (<50 ha). There was no evidence that indicated the influence of forest patches on foliage damage. Our study highlights the economic value of conserving biodiversity, most notably, farmland birds and continuous forests with respect to biological control of defoliating pests and maintaining yield productivity in oil palm cultivation. Growers, particularly major plantation companies should make oil palm farming more biodiversity-friendly in order to increase the number of biological control agents such as birds.

Methods
Study design and foliage damage assessments

Foliage damage assessments were conducted at 30 predetermined sampling points established at each study site, totalling 180 sampling points across all study sites. All points were spaced at >500m apart to maximise spatial coverage. Foliage damage assessments were made based on the approach used by previous studies (e.g., Gilbert & Gregoire, 2003, Priwiratama et al., 2019; Ruslan et al., 2019). At each sampling point, eight oil palms, located within a 20 m radius from the centre of the sampling point were selected for the herbivorous pests (e.g., bagworms) damage inspection at two levels, namely crown and frond. Crown damage evaluations were conducted through visual inspections (using binoculars) to assess the extent of herbivorous pests affecting the entire foliage. The value representing crown damage at each sampling point was derived from the averaged percentage values of damage observed on eight selected palms. Damage observed from the frond base to lower crowns was categorized as ranging from 0% to 50%, damage extending to middle crowns was classified between 50% and 75%, and damage reaching upper crowns was classified between 75% and 100%. Frond damage caused by herbivorous pests was visually inspected on the 17th frond using a standard assessment method. The value representing frond damage for each sampling point was derived from the averaged percentage values observed on the eight selected palms (0% - no damage to 100% - severe damage). Evaluating oil palm health and productivity through the analysis of the 17th frond is a well-established method employed by agronomists (see, Rendana et al., 2015; Irvanto et al., 2022; Peng et al., 2022). In a nutshell, crown and frond evaluation indicate the coarse-scale (overall foliage damage) and fine-scale (detailed foliage damage) assessment, respectively.

Bird sampling

A 10-minute point count was conducted at each sampling point where foliage damage assessments were conducted. All observed or heard birds within 50 metres of the point's centre were recorded. All point counts were conducted in the morning between 0700 – 1030 hours, during good weather. Recorded birds through vocalisations and sightings were identified to species and bird data were sorted and grouped into two groups, namely insectivores and non-insectivores based on local bird guides (e.g., Wells 1999, 2010; Robson 2011; Billerman et al., 2020). We determined bird overall richness, insectivorous bird richness, non-insectivorous richness, bird overall abundance, insectivorous bird abundance and non-insectivorous bird abundance from the surveys.

Local- and landscape-scale variables 

We measured nine local- and landscape-scale variables at each sampling point. The values corresponding to the local-scale variables were obtained by calculating the average of four readings conducted within a 50 m x 50 m vegetation plot at each sampling point. These measured variables include, i) height of oil palm measured using Haga Altimeter, ii) grass cover - refers to coverage percentage (%) of herbaceous plants with narrow leaves estimated within 1m x 1m quadrat established on harvesting paths, iii) shrub cover - refers to coverage percentage (%) of perennial woody plants with stems and branches estimated within 1m x 1m quadrat established on harvesting paths, iv) canopy cover – refers to average percentage (%) of shade, v) epiphyte cover – refers to average percentage (%) of epiphyte persistence on oil palm trunk, vii) dried biomass – refers to weight (kg) of organic matter, debris, and litter cover (e.g., pruned fronds) on interplanting rows, and viii) elevation  - refers to altitude measured above sea level (m). Landscape-scale variables measured in this study include, i) distance to the continuous forest – refers to the shortest measured distance (km) to a forest fragment that exceeds 10,000 hectares in size and ii) distance to the forest patch – refers to the shortest measured distance (km) to a forest fragment that under 10,000 hectares in size (Hazwan et al., 2022; Oon et al., 2023). Landscape-scale variables were measured using the ruler tool in QGIS.