Introduction

The oil palm tree produces the highest oil yield among all vegetable crops globally1,2, making the industry crucial for meeting the growing demand for vegetable oil across domestic use and industrial manufacturing sectors3. This crop thrives in tropical climates and is predominantly cultivated in rainforests and semi-deciduous forest zones4. Notably, the largest exporters of palm oil are concentrated in Asia, particularly within the Southeast Asian region5. Malaysia rank among the top global exporters, with a total of 26,660, 448 tonnes of palm oil products exported in 20246. Despite their extensive experience and dominance in the industry, Malaysia continue to face significant challenges, including losses due to pest infestations, crop diseases, and ongoing controversies surrounding environmental and social impacts7,8.

However, Pests remain among the greatest challenges of the oil palm industry, causing significant yield losses and economic damage across major producing countries. Defoliator pests, particularly the Bagworm (Metisa plana) (Lepidoptera: Psychidae) have caused significant foliage losses, ranging from 3.88% in India and up to 50% in Malaysia9. Plant borers, particularly the Asiatic rhinoceros beetle (Oryctes rhinoceros) (Coleoptera: Scarabeidae), is also a major pest in oil palm plantation and have been reported to cause crop loss of up to 65.5% in moderately infested regions8. To reduce the damage caused by these insect pests, the Malaysian Palm Oil Board (MPOB) employed the use of entomopathogenic microbes, particularly Bacillus thuringiensis in infested oil palm plantations at Perak and Johor from 2005 to 2015 and saved approximately RM 18,814,766 from a potential 40% yield loss10. Although many types of chemical pesticides have been used to control pest populations8, they still persevere in the plantation as they develop immunity11. Additionally, pests of oil palm plantations were diverse as they targeted all stages of the oil palm tree development. Furthermore, most chemical pesticides are non-specific and may kill beneficial insects or even cause secondary poisoning among predators12.

A more sustainable approach of controlling pest populations is using biological control which were highlighted in the Integrated Pest Management such as using entomopathogenic microbes, rat pathogens, pheromone traps and predator birds (insectivore and carnivore)8,13. In the case of rat control, utilising Tyto alba javanica was more cost-effective than using rodenticides14. Although the idea of insectivorous birds controlling insect populations in oil palm plantations was popular, there are still no reports that provide evidence of their direct contribution in controlling insect pests in oil palm agriculture. There were multiple methods employed by researchers to evaluate the effectiveness of insectivorous birds in consuming insects. Koh, through the utilisation of enclosure methods, found that insectivorous birds can reduce damage to oil palm tree15. Similarly, Razak et al. found a positive correlation of bird richness with oil palm yield16, indicating the diverse bird community potentially reduces pest populations, thus increasing the productivity of the plantation. Additionally, increasing attraction of natural predators via volatile compounds in insect frass by injecting mesoporous silica nanoparticles in plants may also contribute in recruiting biological control agent to prey on oil palm pest17. However, some studies argued the effectiveness of insectivorous birds to control oil palm pests as arthropods were found to have a greater predatory rate on artificial caterpillar18,19.

Many studies relied on indirect observations such as improved oil palm yield and healthier oil palm trees in areas with high bird richness15,16,20,21. Insectivorous birds act as natural predators of herbivorous insects that damage oil palm foliage. Studies have shown that increased bird richness correlates with decreased foliage damage, highlighting the importance of avian biodiversity in pest control15,21. Nonetheless, these studies often lack direct evidence linking specific bird species to pest consumption15,22. Traditionally, bird dietary studies involve destructive and invasive methods such as dissection or chemically induced regurgitation23,24. However, these methods are ethically contentious and require skilled professionals to identify degraded insects and handle hazardous chemicals. Another method for dietary studies is by direct observation. However, this method proves to be a challenge as maintaining vision on feeding birds as well as identifying the prey species are difficult25. Thus, ecologists turned their attention to modern methods such as molecular techniques that can produce similar or greater results with reduced time and effort on the field compared to traditional sampling methods. Consequently, identifying which insectivorous bird species prey on agricultural pests is vital for oil palm planters to effectively promote their presence in plantations. A recent metabarcoding study by Arazmi et al. provided evidence of agricultural insect pest consumption by aerial insectivorous birds in areas adjacent to oil palm and paddy fields plantation, highlighting their critical ecological role in natural pest management26. While limited to swiftlets and swallows, this study demonstrates the potential of metabarcoding for quantifying the contribution of insectivorous birds to insect pest control.

Recent advancements in molecular techniques, particularly DNA metabarcoding, have provided powerful tools for identifying and monitoring insect pests in agricultural ecosystems27,28,29. Metabarcoding involves amplifying a hypervariable region of a conserved gene, which is unique between species and serves as a ‘barcode’. These hypervariable regions are placed between conserved regions where primers can bind during PCR amplification. The Cytochrome c oxidase 1 gene became the target region of amplification for many primers that are used in insect studies. This approach allows ecologists to easily identify insects present in ecosystems including elusive and nocturnal species which are usually difficult to identify visually30. Despite the many benefits of using the molecular approach to analyse species diversity, it is not without limitations. Metabarcoding cannot be used for abundance estimation, other than it may have incomplete barcode reference and rely on destructive sampling method30,31. Methods of overcoming these limitations include using relative read abundance (RRA) as a proxy for abundance estimation in prey diet, assigning to higher taxonomic ranks for low percentage identification sequences and soaking samples in lysis buffers instead of crushing them to preserve the sample26,32. Additionally, the metabarcoding method faces many challenges including high contamination risk during workflow, errors during PCR amplification and biases during biostatistical analyses31. Thus, precautionary steps to reduce risk of contamination needs to be taken while handling the sample as well as using suitable pipelines for the bioinformatics. Indeed, molecular methods offer a higher taxonomic resolution for insect identification, especially in dietary studies.

Due to the lack of evidence on the role of birds as biological control of pest insects in oil palm plantation, our study embarked on investigation on the insect composition in the diet of targeted insectivorous birds using the metabarcoding approach. Oriental magpie robin (Copsychus saularis) (Passeriformes: Muscicapidae), a common insectivorous bird found in oil palm plantation was chosen in this study, which incorporates three different oil palm plantation landscapes; monoculture, oil palm with forest patch, and oil palm next to forest reserve. C. saularis have high potential of providing ecological services to the OPP as it was reported to consume insect pests15,22,33. The selection of different oil palm landscapes is driven by the hypothesis that habitat heterogeneity and connectivity to natural forests significantly influence insect prey availability, thus elucidating the insect species composition in the bird diets ultimately, their ecosystem service of pest control, other than non-pest insects. In addition, we also compared the result with a sympatric forest-associated species, White rumped shama (Copsychus malabaricus) (Passeriformes: Muscicapidae) in plantations with forest patch, to distinguish the extent of which they utilize or spill over into these modified environments to forage. This helps understand the functional spillover of both sympatric species in a heterogenous oil palm landscape in terms of their dietary intake, thus elucidating the value of these modified landscapes for forest-dependent species. Therefore, this study provides evidence-based recommendations for sustainable oil palm management practices that promote biodiversity and natural pest suppression.

Material and method

Ethics

This research was conducted in accordance with ethical standards set by the Universiti Kebangsaan Malaysia Animal Ethics Committee (Ref: FST/2023/FARAH SHAFAWATI/22-MARCH/1328-MARCH-2023-OCT.-2025). All methodologies pertaining to the bird handling and faecal collection procedures from the birds were reviewed and approved by the committee. Necessary permits were secured from relevant authorities, including Felda Global Ventures Plantation Malaysia (FGVPM), Department of Wildlife and National Parks (Ref: JPHLTN.600-6/1/4 JLD2 (119)) and Forestry Department of Peninsular Malaysia (Ref: JH/100 Jld. 35(14)).

Sampling locality

This study was carried out in oil palm plantations located in four different states (Pahang, Terengganu, Johor, and Kelantan) in Peninsular Malaysia (Fig. 1). The site selection is based on three criteria where the oil palm plantation landscape exhibits: (1) monoculture plantation (MCP), (2) forest patch (FPP) and (3) adjacent large forest (NFP). The sampling was conducted in each state, consisting of these landscapes between May 2023 to August 2024.

Fig. 1
figure 1

The map of study sites of each state in Peninsular Malaysia, (a) Kelantan, Aring; (b) Terengganu, Setiu; (c) Kelantan, Aring; (d) Johor, Tenggaroh. The birds represent the target species of this study, (e) White rumped shama (Copsychus malabaricus) and (f) Oriental magpie robin (Copsychus saularis). NFP: Next to Forest Plantation; MCP: Monoculture Plantation; FPP: Forest Patch Plantation.

Full size image

Faecal collection

We targeted two insectivorous bird species particularly the oriental magpie robins (Copsychus saularis) and white rumped shama (Copsychus malabaricus) (Fig. 1). The birds were captured using mist nets and were stored in modified containers for faecal collection before release. We used containers similar to Borrelli et al. but with smaller dimensions (20 cm×25 cm×30 cm)34. To ensure reduced contamination of our sample, we used disposable plastic sheets to collect bird faeces and sprayed the container with 70% alcohol after processing each bird. The faecal samples were stored in a 2 ml microcentrifuge tube and preserved with absolute alcohol. The tubes were then immediately stored in -20 °C freezer until DNA extraction. The final sample used for Next Generation Sequencing analysis was only done on samples from three states due to their high-quality DNA as mentioned in Table 1.

Table 1 Summary of faecal samples collected from two bird species across three landscapes in Peninsular Malaysia. The codes represented the landscapes which the samples were collected in and subsequently used during NGS metabarcoding.
Full size table

Sample pre-processing

The chemicals used to preserve the faecal sample prior to DNA extraction were removed for further processing. As a precaution, the sample was centrifuged to spin down all the remaining preservatives which were then removed via pipetting. Additionally, the sample was incubated for 5 min at room temperature with opened tube lids to allow the remaining preservatives to evaporate. The resulting faecal matter was then weighed and samples with < 50 mg were pooled with another sample of the same species from the same landscape.

DNA extraction

The genomic DNA was extracted using QIAamp® Fast DNA Stool Mini Kit following the manufacturers protocol with modifications. DNA concentration was determined with a NanoDrop™ One Microvolume UV-Vis Spectrophotometer. Most of the extracted DNA has low concentrations, hence we performed PCR reaction twice using the same condition.

DNA QC and PCR

For amplicon PCR, primer pairs LCO149035 and HCO1777 were used36 which targeted the 286-bp region of cytochrome c oxidase subunit I (COI). The amplifications were performed in triplicates of 25 µl mixture containing 12.5 µl of PerfeCTa qPCR ToughMix, 5.5 µl ddH2O, 1 µl of 10mM forward and reverse primer and 5 µl of DNA template (~ 10 ng). The thermocycling conditions were as follows: 2 min 30 s at 94 °C, 30 s at 94 °C, 30 s at 44 °C, 45 s at 65 °C and a final extension of 10 min at 65 °C. The second PCR was carried out using 1 µl of DNA template from the first PCR product. Amplification of targeted DNA barcode was then reconfirmed by visualising the DNA on gel via electrophoresis and proceeding to library preparation.

Library Preparation and second PCR

Two-step library preparation was used, consisting of two rounds of PCR. The first PCR was amplified using primers with overhang adapters (Forward overhang: 5’TCGTCGGCAGCGTCAGATGTGTATAAGAGACAG’3 and Reverse overhang: 5’GTCTCGTGGGCTCGGAGATGTGTATAAGAGACAG’). The second PCR involves attaching dual indices, to the overhang adapters using Illumina Nextera XT Index Kit v2 following manufacturers protocols. The quality of libraries was measured using Agilent Bioanalyzer 2100 System by Agilent DNA 1000 Kit and fluorometric quantification by Helixyte Green™ Quantifying Reagent.

Next generation sequencing (NGS)

The libraries were then normalised and pooled according to the protocol recommended by Illumina and proceeded to sequencing using the 300 PE MiSeq platform. All sequencing was performed by Apical Scientific Sdn. Bhd.

Sequence processing and bioinformatic analysis

The raw sequences were obtained in FastQ format with forward and reverse reads. To ensure the sequences were suitable for subsequent analysis, a quality assessment of the raw reads was carried out using fastqc. Following the quality assessment, the primers and adaptors were removed using cutadapt 3.5. Next, Paired-end reads were merged using Divisive Amplicon Denoising Algorithm 2 (DADA2) V1.1837. Next, removal of chimeric sequences was performed using DADA2 to create an amplicon sequence variants (ASVs) matrix. ASVs is a sequence variant that represents individual sequences at a single nucleotide resolution. An ASV table of read counts per sample was generated. The taxonomy of ASVs was identified by blasting against the NCBI database. All sequences shorter than 150 or longer than 600 bp were removed from further analysis. Visualisation using bar plots and Venn diagram was generated from the ASV table using R v3.6.1. Subsequently, the rarefaction curve of each sample was generated to confirm that ASV diversity plateaued, indicating adequate sequencing depth had been obtained.

Statistical analysis

All statistical analysis was carried out using R package v3.6.1. Alpha diversity was assessed to evaluate the insect diversity in the sample using Chao 1, Shannon, Simpson and Observed. Next, a β-diversity analysis with bray-curtis similarity was used to see how the insect communities were different between landscape types. To visualise the similarity of insects between different landscapes, a Principal Coordinate Analysis (PCoA) was generated using Bray-Curtis as the distance measure. Analysis of variance (ANOVA) was then used to test whether there was a significant difference of diet between landscapes.

Result

Sequencing

The sequencing produced 3,014,901 raw sequences from nine faecal samples of C. saularis and two faecal samples from C. malabaricus across the three landscapes (Tables 1 and 2). Following filtering, denoising and removing chimeras, 2,532,507 sequences remained. A total of 270 unique ASVs were identified from the sequences using COI gene analysis. Taxonomic identification revealed three classes, 13 order, 25 family, 29 subfamily and 35 genera of organisms in the bird diet. However, some ASVs cannot be identified to the species level due to low percentage identification with the reference database. Only 0.24% of the cleaned reads were identified to the species level while 99.27% ASVs were identified to the genus level. The alpha diversity reveals that ASVs were the highest in natural habitats (NFP) but were reduced in disturbed habitats (FPP) and suffered further in plantations (MCP) (Table 2). Surprisingly, the mean diversity for C. saularis, which is a non-forest bird, was higher than C. malabaricus, a forest-associated bird in FPP. MCP2 sample also harbours the greatest number of raw reads and alpha diversity from other samples. All alpha diversity measures indicate no significant difference of ASVs between landscapes (Fig. 2).

Table 2 Total number of reads from metagenomic sequencing which were filtered to remove low quality reads. The mean (SD) represent the mean number of reads for each oil palm landscape types (MCP, FPP and NFP). Four measures of alpha diversity (Shannon, Simpson, Chao 1 and Observed) are shown to compare genetic diversity within insect communities in different landscapes.
Full size table
Fig. 2
figure 2

Alpha diversity measures in bird diets using different indexes (i.e. Chao-1, Observed, Shannon and Simpson). Significant difference between landscapes is indicated by p < 0.05.

Full size image

The rarefaction analysis revealed that the observed number of reads have plateaued, indicating that adequate ASVs have been successfully detected for all samples (Fig. 3). PCoA analysis indicates a distinct clustering of MCP and NFP (Fig. 4). However, FPP and FPPX have no clustering. The ASVs detected are visualised in the Venn diagram with MCP harbouring the greatest number of unique ASVs (105) followed by NFP (78), FPP (50) and FPPX (27) (Fig. 5).

Fig. 3
figure 3

Rarefaction curve which were constructed by using the number of observed insect taxa (y-axis) as a function of the number of sequencing reads (x-axis) for each individual faecal sample. The plateauing of the curves indicates that the sequencing depth was sufficient to capture the majority of the insect community diversity present in each sample.

Full size image
Fig. 4
figure 4

Principal coordinate analysis (PCoA) based on Bray-Curtis visualising the diet clustering between landscapes. The figure visualizes the relationships between different insect communities sampled from various landscapes. Axis.1 explains 20.36% of the variation, and Axis.2 explains 17.66% of the variation in community composition. Points represent individual samples, grouped by landscape type: FPP (Forest Patch Plantation), FPPX (Forest Patch Plantation X), MCP (Monoculture Plantation), and NFP (Next to Forest Plantation). Samples with similar community compositions are clustered closer together.

Full size image
Fig. 5
figure 5

Venn diagrams showing the distribution of shared and unique insect taxa across different landscapes. The diagrams illustrate the overlap in insect Amplicon Sequence Variant (ASVs) found in the faecal samples. (a) A three-way diagram comparing the three main landscapes: Next to Forest Plantation (NFP), Monoculture Plantation (MCP), and Forest Patch Plantation (FPP). The numbers represent the count of unique and shared taxa among these landscapes. (b) A two-way diagram showing the overlap in taxa between two species of birds in the same plantation types: FPP (C. saularis) and Forest Patch Plantation X (C. malabaricus).

Full size image

The bird’s diet in MCP is dominated by a single insect order in each sample which were Orthoptera, Blattodea and Orthoptera in MCP 1, MCP2 and MCP3 respectively (Fig. 6). In FPP, the birds reveal a mixed diet consisting of two dominant insects in each sample which were Lepidoptera and Hymenoptera in FPP1, Orthoptera and Blattodea in FPP2 and Orthoptera and Blattodea in FPP3. Birds in NFP on the other hand have shown a diverse diet consisting of at least three insect orders for each individual. NFP1 exhibits the greatest diet diversity consisting of Orthoptera, Stylommatophora and Blattodea. Meanwhile, NFP2 harbours Stylommatophora, Orthoptera, and Blattodea. NFP3 harbours Orthoptera, Hymenoptera, Blattodea and Lepidoptera. Surprisingly, FPPX1 and FPPX2 indicate only three insect orders consumed by forest associated birds which were Blattodea, Orthoptera and Stylommatophora.

Fig. 6
figure 6

Relative read abundance of insect order between samples. The figure illustrates the variation in dietary composition among samples from different landscapes.

Full size image

Although there is a substantial number of genera identified in this study, only some of them are greatly represented in the relative read abundance bar graph, while others have low reads (Fig. 7). When comparing the C. saularis diet composition by landscape, birds in habitats with greater complexity exhibit the greatest diversity with nine insect genera (NFP) while habitats with moderate diversity exhibits eight insect genera (FPP) and lowest diversity exhibiting only six (MCP) insect genera (Fig. 7). In contrast, C. malabaricus (FPPX) only harbors three insect genera in their diets.

Fig. 7
figure 7

Relative read abundance of insects in different landscapes and species. The figure reveals distinct differences in dietary composition between landscapes (MCP, FPP and NFP) and species (FPP and FPPX).

Full size image

Based on the heat map, the birds generally prefer Blattella sp. and Gryllus sp. over other insects as they are highly represented in the bird’s diet (Fig. 8). Notable insect genera were the Episymploce sp., Mythimna sp. and Dolichopoda sp. which were prevalent in only one sample each suggesting a high abundance of those insects in the landscape. Occasionally, the birds consume other invertebrates such as Pseudoxya sp., Parmarion sp., Agrotis sp., Carebara sp., Anopheles sp., and Megachiles sp. Interestingly, the diet of C. malabaricus and C. saularis are more similar when they are in the same landscape type, suggesting a diet overlap as shown by the Hierarchical Clustering. Additionally, Hierarchical clustering further revealed that the dietary composition of birds in NFP closely resembles those from FPP, whereas samples from MCP exhibit higher similarity to FPP diets.

Fig. 8
figure 8

Heatmap exhibiting the relative abundance of taxa between samples. Similarities between samples are exhibited using Hierarchical Clustering with UPGMA and Bray-Curtis distance matrix.

Full size image

Among the pest insects detected in bird diets, only Mythimna sp. were largely consumed in FPP1 (Table 3). Meanwhile, Pseudoxya sp. are represented in three samples such as FPP3, MCP1 and NFP1. Odontotermes sp. and Oxya sp. were not preferred by C. saularis as they have very low relative read abundance. In contrast, predation on beneficial insects is limited with Carebara sp. representing the greatest relative read abundance followed by Megachile sp. and Yaginumaella sp.

Table 3 Consumption of pests and beneficial invertebrates in oil palm landscape. The table lists the relative abundance of specific insect genera categorized as either pests or beneficial species. The percentages represent the proportion of each taxon found within the total insect community of each faecal sample. The samples are grouped by their landscape of origin: monoculture plantation (MCP), forest patch plantation (C. saularis) (FPP), forest patch plantation X (C. malabaricus) (FPPX), and next to forest plantation (NFP).
Full size table

Discussion

This study demonstrates the diet of insectivorous birds which are abundant in oil palm plantations, namely C. saularis and C. malabaricus while highlighting their potential to become biological control agents for insect pests. Our findings provide evidence from metabarcoding data on the diets of these birds at a higher taxonomic resolution (i.e. genus-level) than previous studies22,38. Our analysis of bird diets across three landscape types suggests that these birds have surprisingly varied diets, with a total of 35 genus detected across all samples.

We found that the diet composition of C. saularis in NFP has the greatest diversity, while FPP harbours moderate diversity and MCP exhibits the lowest diversity. This phenomenon is likely due to degraded habitats such as MCP harbouring lower habitat complexity which induced survival pressure on insect species. This causes a limited diversity but promotes abundance of resilient insects in the plantation39. According to Lucey & Hill, ants exhibited greater resilience to land-use changes compared to butterflies due to their tolerance39. The study found that air and soil temperature in oil palm plantations were 2 °C hotter than in natural landscapes thus exerting greater impact on butterflies than ants. High abundance of certain insects likely caused behavioural changes among predators such as C. saularis to target them in the poor diversity landscape. One of the behavioural changes includes their foraging strategy. These birds primarily forage on the ground, feeding on ground-dwelling insects such as grasshoppers, ants and termites. This is shown in the Hierarchical Clustering where the diets of birds in NFP resembles FPP, whereas samples from MCP are more similar to FPP diets (Fig. 8). However, in a different ecosystem these birds have mixed hunting strategies such as sallying and ground foraging38. Similar to Ashitha T & Seedikkoya K, the diet composition of C. saularis were dominated by ground dwelling prey including termites, ants and black soldier fly larvae38. Due to C. saularis being a generalist insectivore, they consume their prey opportunistically rather than selectively. Hence, the prevalence of certain insects in the bird’s diet composition likely reflects their abundance within the plantation ecosystem.

We found that C. saularis primarily consumes palatable insects such as grasshoppers, moths, ants and termites. These insects are likely favoured due to their characteristics of having soft texture bodies making them easier to capture and subsequently consumed. In this study, Blattela sp. and Gryllus sp. are particularly abundant in the bird’s diet, indicating the bird’s choice of food across landscape (Fig. 8). Additionally, these insects contain high nutritional content, particularly proteins which are important especially during the breeding season where the energy demand of chick-rearing birds is heightened40,41. The preference for soft-bodied insects also suggests that C. saularis optimises its diet for efficiency. Although Oryctes rhinoceros contains high nutritional content similar to grasshoppers and moths40, there was no evidence of C. saularis feeding on these pests albeit being abundant in the plantation. This is likely due to the characteristics of O. rhinoceros such as being large and having hard carapaces, making them hardly palatable and inefficient to be processed before being consumed by C. saularis. Additionally, their larvae are primarily buried in the ground8, greatly limiting potential predators. Notably, there were reports of another insect-eating bird, the Lesser coucal (Centropus bengalensis) exhibiting a ‘digging’ behaviour at 1–2 cm into the ground to feed on Setothosea asigna while Acridotheres javanicus has been recorded to consume O. rhinoceros larvae33,42. This suggests that these birds can potentially control O. rhinoceros population in oil palm plantation. These findings highlight the diverse adaptations between bird species to feed on insects that requires specialised strategies or unique physiological traits.

Surprisingly, MCP2 harbours the greatest diversity in all samples. This indicates that this particular bird consumes multiple insect types even in the degraded habitat (MCP). Although MCP exhibits the lowest habitat complexity, there are still some features within the plantation that attract high insect abundance such as high epiphyte persistence and high understorey vegetation43,44. These habitat features are predominantly in valley regions of the plantation where the area is not regularly maintained due to hazardous terrains. Apart from harbouring greater understorey vegetation and epiphyte persistence, this area also harbours flowing water that acts as a small riparian zone. This zone greatly enhanced the local biodiversity by providing microhabitats suitable for aquatic and semi-aquatic species45,46,47. Additionally, we also detected some waterbird species in this area such as White-breasted waterhen (Amaurornis phoenicurus) (Gruiformes: Rallidae), Purple heron (Ardea purpurea) (Pelecaniformes: Ardeidae) and Chinese pond heron (Ardeola bacchus) (Pelecaniformes: Ardeidae) highlighting the significance of this area to wildlife species dependent on water availability. Given the high abundance of insects in oil palm valley regions, C. saularis greatly benefitted due to increased prey availability and abundance in these microhabitats as shown in this study. On another note, increasing understorey vegetation in the valley regions within the plantation so that it may act as forest corridors may be an efficient solution to enhance prey diversity of C. saularis and potentially facilitates spillover of forest-associated birds to further support ecosystem function in oil palm plantations20,48.

Interestingly, C. malabaricus which is a forest-associated bird harbours the lowest diet diversity even when compared to C. saularis in FPP. Forest-associated insectivorous birds are thought to consume greater insect diversity compared to farmland insectivorous birds due to them inhabiting landscapes with high habitat complexity39,49. However, the low diversity observed in C. malabaricus diets is likely influenced by the availability of insects within the landscape similar to C. saularis in plantations. Although FPP harbours moderate habitat complexity, the area is mainly secondary growth forests that are dominated by pioneer insect species. These pioneer insects likely exhibit some degree of dominance in the area39, causing C. malabaricus to primarily feed on them due to their abundance. Interestingly, this study found that C. malabaricus primarily consumes ground-dwelling insects such as Orthoptera and Blattodea. This may be attributed to its hunting strategy in the forest. Although there were limited studies on the feeding behaviour of this species, its hunting strategy may be similar to its oil palm counterpart which is C. saularis. Additionally, findings from Mohd-Azlan et al. indicates that HCV areas in Borneo are dominated by Coleoptera (1029), Hymenoptera (567) and Blattodea (219) while this study found Blattodea, Orthoptera and Parmarion to be the main diets of C. malabaricus50. Hence, C. malabaricus may be selective in their prey selection and primarily consume ground-dwelling prey while opportunistically consuming other insects. On another note, the presence of insects from the modified environment in the bird diet, suggests their functional spillover into the oil palm landscape. This also indicates that the heterogeneous landscape benefitted the forest-dependent species by increasing and diversifying the food sources48,51,52.

The bird’s diet is stable across landscapes indicating a relatively similar prey overall. This can be explained due to the limited diversity but abundant prey in oil palm plantation as mentioned earlier. Insects that can survive in oil palm plantations mainly consist of resilient insects that can withstand habitat changes39. Notably, Mythimna sp. were detected only in landscapes near natural habitats enhancing the notion that C. saularis are feeding opportunistically. These insects became the main food source for birds inhabiting oil palm plantations. Still, there are some changes in insect composition especially in the NFP landscape as shown by PCoA analysis indicating distinct clustering between NFP sites. In this habitat, forest insects may ‘spillover’ into the plantation and be consumed by C. saularis who feeds opportunistically. However, this spillover effect is not significant enough to influence the ANOVA results, indicating a limited spillover effect of forest insects into the oil palm plantation. Parallel to our findings, Lucey & Hill also found little evidence of spillover effect of insects into the oil palm plantation using butterflies and ants in their study39.

Although C. saularis prefers neutral insects, they were also found to consume both pest and beneficial insects (Fig. 8; Table 3). This indicates that C. saularis have high potential to act as biological control agent in the oil palm plantation. However, a previous report has highlighted how insectivorous birds such as Orthotomus sp. may provide a net negative effect on the ecosystem due to predation of beneficial insects22. Nevertheless, this study proves that C. saularis predation on pests are greater than beneficial insects, hence highlighting its potential to be used in the plantation. Earlier research has also found similar results in the study of insectivorous birds in soybean fields where their data suggest that birds consume a significantly higher proportion of herbivory arthropods compared to beneficial insects27. In contrast, C. malabaricus was found to be neutral as they consume neither pests nor beneficial insects of oil palm. C. malabaricus primarily consumes neutral insects that do not affect the oil palm. However, changes in insect composition in the area may affect the diet of C. malabaricus.

Interestingly, we have observed some pest beetles such as Red palm weevil (Rhyncophorus ferruginous) (Coleoptera: Curculionide) in our study area, but none have been detected using NGS. This indicates that C. saularis may not be as efficient predators of the pest as the NGS did not pick up their DNA in the bird faeces. Although Desmier De Chenon & Susanto have reported C. saularis to consume coleopterans, they are likely ground-dwelling beetles22. This suggests that C. saularis can only efficiently control ground-dwelling pests such as orthopteran species which are economically important8. Additionally, primer bias may be a recurring issue in this study as Yu et al. has previously reported that LCO1490 and HCO1777 has a low detection rate on hymenopterans53. However, Paula reported a 92–97% detection rate of COI Folmer primers on insect species54. On the other hand, we did not detect any Mythimna sp. in FPP with visual observations, but the NGS successfully captured the DNA from bird faeces. This study proves that NGS is excellent in identifying insect species from the insect predator bird’s diet. Some species that are highly elusive and difficult to capture using conventional sampling methods may be detected in the predator diet27. Hence, NGS can be an additional method to complement data from conventional sampling to further enhance taxa detection.

Conclusion

This study provides vital evidence regarding the diets of C. saularis and C. malabaricus in oil palm landscapes, highlighting the ecological interactions between these species and their prey availability. Despite plantations being a degraded habitat, the presence of understory vegetation and persistence of epiphytes became a boon to the birds as they attract insects for feeding and shelter. Our study confirms the positive functional spillover of C. saularis from the forest edge into the oil palm landscapes as biological control of insect pests, whereas C. malabaricus shows minimal functional roles in biological control of insect pests. Despite this, this study highlights the value of the oil palm landscape in supporting the forest-dependent species via providing additional food sources. We also found that C. saularis poses minimal ecosystem disservices, given its tendency to consume pest species more frequently than beneficial insects. However, we acknowledge that this study is not conducted in plantations during a pest outbreak. Hence, future studies incorporating long term monitoring supplemented by molecular analysis could further refine this dietary assessment. Additionally, future dietary studies involving NGS can focus on other insectivorous bird species in oil palm plantation such as Yellow vented bulbul (Pycnonotus goiavier), White throated kingfisher (Halcyon smyrnensis) and Common tailorbird (Orthotomus ruficeps) as they are similarly abundant in the landscape and were also reported to consume insects.