The Asian weaver ant, Oecophylla smaragdina (Fabricius, 1775), is widespread from western Asia to northern Australia (Wetterer 2017) and maintains arboreal territories through aggressive behaviour. Worker ants patrol tree branches and attack other animals because they are generalist predators, preying on insects and other arthropods and also attack humans who disturb their nests (Crozier et al. 2010). The Asian weaver ant has several ecological and economic functions regarding its ecological interactions: enhancing potential nitrogen (Pinkalski et al. 2015, Vidkjær et al. 2015, Vidkjær et al. 2016), shaping plant-pollinator interactions (Rodríguez-Gironés et al. 2013) and acting as biocontrol agent (Peng et al. 1999, Peng and Christian 2005, Peng and Christian 2013, William et al. 2015, Anato et al. 2017). Weaver ants are commonly used as biocontrol agents in several agricultural systems with potential to control major pests such as true bugs (Coreidae and Miridae), beetles (Chrysomelidae), aphids (Aphididae), caterpillars (Lepidoptera), leaf miners (Coleoptera), leafrollers (Lepidoptera), fruit flies (Drosophilidae), leafhoppers (Cicadellidae) and shoot borers (Lepidoptera) (Van Mele 2007).

The majority of the species in the butterfly family Lycaenidae are associated with ants, either mutualistically or parasitically, with interactions either being facultative or obligate (Pierce et al. 2002). Parasitic interactions usually involve myrmecophagy, with butterfly larvae consuming the ant brood. In the family Lycaenidae, 74 species show this behaviour during some stage of their life cycle (Pierce 1995). These parasitic-myrmecophagy interactions can be found between Africa and northern Australia (Fiedler 1991, Eastwood and Fraser 1999, Pierce et al. 2002, Terblanche and Van Hamburg 2003, Crozier et al. 2010, Table 1).

Liphyra brassolis (Westwood, 1864), is obligately myrmecophilous, with larvae being dependent on weaver ants (Oecophylla smaragdina) and eating the ant brood (Dodd 1902, Eastwood and Fraser 1999, Pierce et al. 2002). As L. brassolis feeds on the larvae of weaver ants, it might have the potential to affect ant populations. Hence, it is important to understand its occurrence in widespread crops, such as oil palm plantation, where O. smaragdina has the potential to control pests, such as leaf-eating caterpillars.

Sampling of weaver ant nests was conducted in an oil palm plantation located in the Province of Riau in Sumatra, Indonesia (0°56'35.3"N, 101°11'56.4"E) in August 2021. A localised outbreak of Clania tertia Templeton, 1847 (Lepidoptera, Pyschidae) bagworm was ongoing when the sampling was conducted. Twenty active weaver ant nests were chosen randomly on twenty individual palms. Only nests in which ants were present (confirmed using binoculars), constructed from live green palm leaflets, were sampled. The nest was collected, wrapped in a plastic bag and all connecting leaflets and fronds were cut using machete and pruning shears. Nest dimensions (length, width and depth) were measured. On the ground, nests were thoroughly dissected and checked for the presence of L. brassolis larva. Dissected nests were removed from bags and replaced at the base of the palm to allow ants to return.

During this period, as part of another study, we were conducting mass-rearing of weaver ants with the aim of propagation and release as biocontrol agent against leaf-eating caterpillar pests. All the colonies were fed on a 30% sugar solution and live insects or fish as a protein source. Observations of L. brassolis were conducted in this mass-rearing activity in 25 ant colonies that were being propagated in plastic bottles (Offenberg 2014) for integrated pest management (IPM).

Liphyra brassolis larvae, found in ant nests, were identified morphologically (Crozier et al. 2010, Eastwood et al. 2010, Dupont et al. 2016). The larva of this species is distinguished by its carapace-like structure with a hard cuticle both dorsally and laterally (Fig. 1), which is used to defend against ants attacks (Dupont et al. 2016). All speciments of L. brassolis larvae and O. smaragdina were housed in Smart Research Institute, Riau, Indonesia.

a

b

c

d

Figure 1.

Morphology of Liphyra brassolis that were collected from weaver ant nests during field observations in an oil palm plantation in Riau Province, Indonesia.

a: dorsal side of larva  
b: ventral side of larva  
c: pupal stage  
d: pupal exuvia  

A generalised linear model (GLM) with binomial errors was used to investigate the relationship between the weaver ant nest size and L. brassolis occurrence. The occurrence of L. brasollis larvae was used as a binary response variable with one nest sample per observation. Nest size was calculated as an ellipsoid volume using the three measured nest dimensions. Correction for overdispersion was conducted using the quasibinomial family since the residual deviance was larger than the degree of freedom (Crawley 2015). All statistical analyses were performed using the R statistical program version 4.0.2 (R Core Team 2020) in the R Studio version 1.1.423 (RStudio Team 2020) environment. Graphs were created using “ggplot2” and “tidyverse” R packages (Wickham et al. 2019). The generalised linear model was performed using the "lme4" package (Bates et al. 2015). The dataset of observation is deposited in GBIF, Global Biodiversity Information Facility (https://doi.org/10.15468/upzkak).

Liphyra brassolis larvae were present inside both weaver ant nests from the field (5 of 20 nests sampled) and mass-rearing observations (2 of 25 nests). In all seven nests with L. brassolis larvae present, only a single larvae was found (Suppl. material 1). A similar rate of occurrence of L. brassolis was reported during the ant harvesting season in Khao Chong, Thailand (this study: 16%, Eastwood et al. 2010: 21%). We found no evidence for a relationship between ant nest volume and L. brassolis larva presence (GLM, binomial errors, d.f. = 19, t = 0.78, p = 0.44; Fig. 2). This contrasts with a previous anecdotal observation mentioning that L. brassolis was found in larger weaver ant nests (Itterbeeck 2014).

Figure 2.  

Relationship between Liphyra brassolis larvae occurrence and weaver ant nest volume. Fitted line comes from a GLM with binomial errors.

We observed the larvae in the mass-rearing facility actively following the weaver ants during migration from the original disturbed leaf nest into the plastic bottle used to house the colony (Fig. 3). The presence of L. brassolis larvae in mass-rearing facilities should be avoided since consumption will reduce the number of ant brood. Mass-rearing operators should be trained to notice the L. brassolis larvae, so that larvae can be removed before the ant colony becomes established in the new nest. Education is needed here, since often local people misidentify L. brassolis larvae as a large ant or queen ant (Eastwood et al. 2010) . This is important because the weaver ant mass-rearing is commonly used to produce ant brood for use as bird feed (Prayoga 2015), a protein source for human consumption (Sribandit et al. 2008) or as an augmentation strategy for integrated pest management (IPM) programmes (Peng et al. 2014).

Figure 3.  

Presence of Liphyra brassolis (Lepidoptera: Lycaenidae), an obligate-parasitic myrmecophily larva, inside a weaver ant (Oecophylla smaragdina) colony in the mass-rearing facility.

Although adult L. brassolis have been reported in West Java, Indonesia (Peggie 2012), this is the first report of this species in an oil palm plantation to our knowledge, despite multiple other butterfly surveys in this habitat in Sumatra (Panjaitan et al. 2020, Panjaitan et al. 2021). In Thailand, the larva of L. brassolis is commonly reported in weaver ant nests on orchard plantations, being first documented during entomophagy by local people (Eastwood et al. 2010). Furthermore, the presence of L. brassolis larva in weaver ant nests has not yet been reported during brood harvesting activity by local people in Indonesia (Césard 2004).

Implementation of biological control against leaf-eating caterpillars in an oil plantation is needed to support sustainable palm oil practices. However, only a small number of studies have investigated the ecological function of O. smaragdina as a biological control agent in oil palm plantation (Pierre and Idris 2013). With specific interaction with O. smaragdina, the butterfly L. brassolis can be a bioindicator for weaver ant colony in oil palm plantations. Liphyra brassolis larvae are only ever found associated with O. smaragdina. However, other Lycaenid larvae have been observed in the nests of Oecophylla spp, such as Liphyra grandis and Zesius chrysomallus (Pierce et al. 2002).

Our finding indicates that further investigations of weaver ant symbionts in oil palm plantations could be useful, in particular, because this ant species has potency for controlling leaf-eating caterpillar pests. It would be worth conducting larger-scale observations over greater numbers of colonies in order to measure any potential impacts of L. brassolis larvae on colony fitness, both in the field and in mass-rearing facilities. Furthermore, this observation represents an additional data point for lepidopteran biodiversity in Sumatra, especially in oil palm landscapes.

ADA and KMY were supported by a Malaysian Ministry of Higher Education Fundamental Research Grant (FRGS/1/2014/STWN10/UMS/02/1). TMF was supported by a Czech Science Foundation Standard Grant (19-14620S). We would also like to thank to PT. Smart Tbk for supporting ADA to pursue his Masters study in the Universiti Malaysia Sabah and all SMARTRI staff for their support.

Malaysian Ministry of Higher Education Fundamental Research Grant (FRGS/1/2014/STWN10/UMS/02/1); Czech Science Foundation Standard Grant (19-14620S)

Institute for Tropical Biology and Conservation, Universiti Malaysia Sabah; Smart Research Institute, PT Smart Tbk, Indonesia

The data in the paper were collected as part of an ongoing leaf-eating caterpillar pest and its natural enemy monitoring in the plantations, following standard industry operating procedures.

ADA, KMY and TMF contributed to the conception and design of the study. ADA collected the data. ADA, KMY and TMF wrote the first draft of the manuscript. All authors contributed to manuscript revision and have read and approved the submitted version.

The authors declare no conflict of interest.