Vertically transmitted bacterivorous nematodes are consistent nest inhabitants in the Azteca-Cecropia ant-plant mutualism

Abstract

Nematodes play key roles in natural and agricultural ecosystems. They contribute to organic matter transformation and the stability of soil food webs. Beyond their free-living forms, many nematode lineages have evolved in close associations with insects, ranging from mutualistic and commensal to parasitic interactions. Recent studies have revealed that nematodes are common in tropical ant–plant mutualisms and are particularly relevant in ant-made organic matter piles, or “patches”, within ant nests. To investigate nematode community dynamics during ant colony growth and their consistency across closely related ant species, we analysed patches from 65 ant colonies of the Azteca-Cecropia ant-plant mutualism using 18S rRNA metabarcoding combined with morphology-based quantification methods. Bacterivorous nematodes from the order Rhabditida were present in all samples, regardless of the ant or plant species and the colony developmental stage. Members of Tylenchida and Dorylaimida were also detected, though sporadically. Our results support the previously proposed vertical transmission of bacterivorous nematodes from mother to daughter colonies as well as horizontal transmission among patches within the same ant colony. Moreover, nematode community composition remained stable throughout colony development but varied between ant species. These findings demonstrate that nematodes constitute a persistent and functionally important component of this ant–plant mutualism.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-025-34231-9.

Introduction

Nematodes are the most abundant and ubiquitous metazoans on Earth¹. They are distributed across all trophic levels, and thus, they are usually classified based on their feeding strategy (i.e., herbivores, bacterivores, fungivores and omnivores)². Due to their high adaptability and their notable phenotypic plasticity, many species are involved in complex ecological networks that often include symbiotic associations with bacteria, plants and other animals^3,4.

It is estimated that nematodes are associated with 40,000-500,000 species of arthropods worldwide, many of which display a social behaviour, such as bees, termites, and ants⁵. Entomophilic non-parasitic nematodes are often characterized as bacteria-feeding opportunists². These bacterivorous nematodes have repeatedly evolved the ability to use insects as bio-vehicles to access bacteria-rich environments^6–8. This dispersal strategy, known as phoresy, is characterized by the development of a highly resilient ‘dauerlarva’, a resistant, non-feeding third-stage juvenile (L3)⁶. Dauerlarvae exhibit a behaviour (standing on the tail and waving) as a “sit-and-wait” host-tracking and dispersal strategy⁹ facilitated by a specialized cuticle-attachment mechanism^6,10.

While the association is clearly beneficial for the nematodes, the effects recorded so far on the fitness and development of their insect partners range from mutually beneficial to detrimental^11–13. For example, in dung beetles (Coleoptera, Scarabaeoidea), the presence of Diplogastrellus (Rhabditida, Diplogastridae) nematodes, which are transgenerationally inherited and sexually transmitted within the brood balls, promotes the growth of beetle offspring¹¹. On the contrary, in burying beetles, the presence of phoretic nematodes in the bodies of mating females leads to a significant reduction in brood size, juvenile survival, and juvenile mass¹². Furthermore, recent research on fungus-growing termites (Blattodea, Isoptera) indicates a commensal relationship with nematodes^7,14. Their presence in the termite nest appears to be harmless to the termites. However, nematodes are rare in chambers with brood or are even suppressed by the termites’ grooming activity¹⁴.

Ants have been associated with nematodes for at least 20–30 million years¹⁵. As in other insects, most ant-associated nematodes are non-parasitic, saprobiontic bacterivores, primarily belonging to the families Rhabditidae, Diplogastridae and Panagrolaimidae of the order Rhabditida⁸. Ant-nematode interactions are particularly frequent in tropical ant-plant mutualisms worldwide^8,16,17. In these mutualistic relationships, ants protect their host plant from herbivores, pathogens, and overgrowing vegetation in exchange for nesting space within hollow plant structures (domatia) and access to nutritional resources, such as extrafloral nectar or food bodies^18–21. Within the domatia, nematodes coexist with fungi and bacteria in organic matter piles built by ants and known as ‘patches’^{8,16,17,22–25}. Several studies have proposed that ant-associated nematodes may be transmitted vertically from mother to daughter colonies, as they are frequently detected in the infrabuccal pockets and attached to the bodies of alate queens^16,17,26.

The dynamics and composition of the patch communities with respect to fungi and bacteria have been extensively studied^22,23,27,28, but no research has been conducted on the occurrence and composition of the nematode communities. To address this gap, we investigated the diversity of nematodes inhabiting ant-made patches in one of the most prominent and abundant ant-plant mutualisms in the tropical America, the Azteca-Cecropia complex, by comparing: (i) ant colonies at different developmental stages , and (ii) different species of the respective ant and plant partners. By combining amplicon sequencing of the partial 18S rRNA gene with morphological identification methods, we analysed patches from 65 colonies of three Azteca species (Formicidae, Dolichoderinae) inhabiting Cecropia spp. (Urticaceae) from which fungal and bacterial communities had already been examined^22,23. We hypothesize that (i) nematodes are consistently present in patches during early colony foundation, supporting vertical transmission from mother to daughter colonies; (ii) nematodes are present in all patches as the ant colony expands within the Cecropia, supporting horizontal transmission among patches within the same ant colony; and (iii) nematode community composition is influenced by the ant species rather than the plant species. Our study of the nematode community composition in an ant–plant mutualism provides an important foundation for understanding the ecological roles of nematodes within this complex ecosystem, as well as their potential effect on the fitness and development of both the ant colony and the host plant.

Materials and methods

Patch sample collection

Samples were collected in wet lowland forests and pastures in the conservation zone ACOSA (Área de Coservación Osa) in Puntarenas, Costa Rica in close vicinity to the Tropical Field Station La Gamba (08°42’03”N, 83°12’06”W, 70 m a.s.l.). All plant and ant samples collected in this study were obtained with permission from the “Sistema Nacional de Áreas de Conservación” of Costa Rica (permits SINAC INV‑ACOSA‑001–16,” and “SINAC INV‑ACOSA‑013–18). For this investigation, we sampled 55 Cecropia trees (C. peltata, C. obtusifolia and C. insignis) colonized by 68 Azteca colonies. The Cecropia species were determined based on the leaf characteristics when the leaves were properly developed²⁹. Plants that were young and still lacked the distinctive characteristics required for precise identification were labelled as “Cecropia sp.“. These identifications were made by VEM and MN based on their extensive Cecropia spp. collections from the same area in 2013. Voucher specimens are deposited at the Herbarium of the University of Vienna (WU) under accession number 1971/5268.

After transversally opening colonized Cecropia stems, we collected ant-made patch samples from the nesting space of Azteca spp. colonies using a dental probe. Subsequently, the ant species (A. alfari, A. constructor and A. xanthochroa) were identified through the morphology of the ant queen following J. Longino’s Azteca species description³⁰ and through the characteristics of the ant nest²⁸.

Patch samples were classified in three categories depending on the developmental stage of the ant colony (Fig. 1). In the initial and young developmental stages, ant colonies are spatially limited to a single plant internode and usually maintain only one patch. These patches were analysed separately. In established ant colonies, in which ants maintain multiple patches throughout the nesting space, patches of a single ant colony were either combined in the field or separately sampled and sequenced by tree sections as described in Nepel et al. (2023)²² and then, bioinformatically pooled.

Fig. 1.

Graphic illustration of the Azteca-Cecropia mutualism and the associated nematodes inhabiting ant-made patches. (A) Representation of the ant colony development and the patch formation, including the patch categories used in this study: initial patch (IP); young patch (YP); established patches (EP). (B) Stereomicroscope image showing numerous nematodes on the patch surface. (C) Scanning electron microscope (SEM) image of nematode eggs within the patch material. (D–E) Microscopic images showing morphological features of a female Sclerorhabditis nematode.

In the area of sampling, the abundance of the different Azteca and Cecropia species was notably uneven: only one Cecropia specimen was identified as C. insignis; from A. xanthochroa, only founding queens with initial patches were found in the study area; and all young ant colonies were identified as A. alfari colonies. Consequently, we were unable to obtain an equal number of samples for each ant and plant species, and for each ant colony developmental stage (Supplementary Information S1).

Morphological identification and quantification of nematodes

Nematodes were morphologically identified and quantified from homogenized patch samples of four ant colonies (two established A. alfari and two established A. constructor colonies). First, alive nematodes were extracted from the patch matrix by centrifugation at 1800 g for 4 min using a highly dense non-ionic solution (Nycodenz 1.16 g/mL 1x PBS). Subsequently, the supernatant, where most of the nematodes were detected, was transferred to a new 2 mL tube. The nematodes were pelleted by centrifugation (18000 g for 3 min) and washed with 0.2 μm filtered H₂O three times, each time followed by centrifugation. The pellet was stored in RNA-later. Finally, nematode suspensions were subsampled and specimens were manually classified, whenever possible, to their respective species based on species-specific morphological features³¹, and, manually quantified using a microscope (Zeiss Axioplan 100 x Plan-Neofluar).

Metabarcoding

For metabarcoding, patch samples from 68 ant colonies stored in RNAlater solution (1:4) were washed twice with a phosphate buffer by centrifugation at 14 000 rpm for 1 min. DNA extraction was performed following an adapted phenol-chloroform based extraction protocol³², using three rounds of mechanical lysis via bead beating (30 s at 6.5 m s-1)³³. Amplicon libraries targeting the partial 18S rRNA gene were generated by a three-step PCR approach for highly multiplexed amplicon sequencing composed by: (a) a semi-nested PCR protocol using metazoan selective primers (NemF and 18Sr2b) first, followed by amplification with NF1 and 18Sr2b primers^34,35; (b) a barcoding PCR step³⁶. Detailed information about the primer sequences, PCR protocol and programs used in this study is provided in Supplementary Information S2. Library preparation and MiSeq Illumina sequencing (JMF-2010-2) was performed by the Joint Microbiome Facility (JMF, University of Vienna, Austria) using a 2 × 300 bp cycles paired-end mode and the MiSeq v3 Reagent kit (Illumina).

Sequence processing and analysis

Amplicon sequence data were processed as described in Pjevac et al. (2021)³⁶ following the recent recommendations for metabarcoding studies of nematodes communities³⁷. First, amplicon sequence variants (ASVs) were inferred using the DADA2 R package version 1.2.0³⁸ with R v4.1.1³⁹ by applying trimming at 220/230 nt with allowed expected errors of 2/4. Second, ASVs were taxonomically classified using RDP classifier⁴⁰ and the SILVA v.138 SSU database⁴¹. Third, the taxonomic classification of the 50 most abundant ASVs was revised by BLASTN using the nucleotide collection (nt) database (updated on the 11th of October 2023). In the count and taxonomy tables, we performed a filtering process where we discarded: (i) singleton ASVs, (ii) ASVs not classified as phylum Nematoda, and then, (iii) samples with less than 2000 reads. As a result, 65 samples out of 68 were kept for downstream analyses (Supplementary Information S1).

Downstream analyses were performed in R v4.1.2 and RStudio 2021.09.1⁴². When several patch samples were sequenced from the same established ant colony, their sequence data were merged by adding up read counts using the R packages ampvis2 v2.7.11⁴³. The nematode alpha and beta-diversity metrics in individual ant colonies were analysed using the R packages ampvis2 v2.7.11 and vegan v2.6-4⁴⁴. For both diversity metrics, we first rarefied the read counts using the minimum read count per sample that was higher than 2000 reads. Alpha diversity was calculated using Shannon index and the differences between groups were statistically tested with p value of 0.05 by using: (i) Wilcoxon test when comparing two groups; and (ii) Kruskal-Wallis and post-hoc pairwise Wilcoxon test when comparing three groups. The beta diversity was visualized by PCoA using Bray-Curtis distances and categorical variables were statistically tested using PERMANOVA test (p < 0.05)⁴⁵. Since the sample size design was notably unbalanced in most beta diversity comparisons, additional PERMDISP tests⁴⁶ were performed to evaluate the heterogeneity of dispersions. Relative read abundances of ASVs belonging to frequent nematode groups were statistically compared between A. alfari and A. constructor colonies using Kruskal-Wallis and post-hoc Wilcoxon test (p < 0.05).

For analysing the nematode diversity and community composition among different tree sections of established ant colonies, we used the separately sequenced patch samples per colony. The alpha and beta diversity indexes were calculated and statistically tested as described above.

Results

Morphological identification reveals three Rhabditida species in the patches of established ant colonies

By microscopy and morphological identification methods, we identified three nematode species in the patch samples of established Azteca spp. colonies inhabiting Cecropia spp. plants. Due to suboptimal fixation, only the structure of the lip region and stoma could be used for species identification. The nematode specimens were almost exclusively female. As far as it could be seen, they were similar in body size and shape, amphidelphic gonads with a median vulva and conoid elongated tail. Sclerorhabditis is characterized by (in lateral view) a dorsal and a ventral thorn shaped sclerotization directed toward the mouth opening (Fig. 2A). The cylindrical stomatal tube is relatively short and wide. In contrast, Diploscapter has a long, narrow mouth tube and dorsally and ventrally an outward-facing, sclerotized hook-like appendage of the lips (Fig. 2B). Specimens of the third species could be distinguished by a cup-shaped stoma, in which – due to the poor state of preservation – more special structures could not be differentiated (Fig. 2C). Other features also remained unclear, especially the structure of the pharynx, and thus, it was not possible to determine on a morphological basis whether it belongs to the family Diplogastridae (Rhabditida).

Fig. 2.

Comparison of the anterior ends of the three species of nematodes detected in the patches. (A) Sclerorhabditis cf. neotropicalis, (B) Diploscapter sp., (C) two sketches of a species of Rhabditida which could not be assigned to any genus, having a cup-shaped stoma.

While Sclerorhabditis specimens from the patches are likely belonging to the Sclerorhabditis cf. neotropicalis species²⁵, slight morphological differences in the male posterior end of these individuals were detected (Supplementary Information S3).

Metabarcoding analysis detects Rhabditida, Tylenchida and Dorylaimida orders in patches across different ant colony developmental stages

Metabarcoding analysis of the partial 18S rRNA gene from 65 Azteca spp. ant colonies (Supplementary Information S1) resulted in 1.61 × 10⁶ total reads and 200 ASVs. For downstream analysis, we used 108 ASVs (90% of total reads) which were assigned to the phylum Nematoda. The non-Nematoda ASVs were mostly assigned to the phylum Arthropoda (29 ASVs and 9% of total reads) or to other eukaryotic groups (61 ASVs and 1% of total reads).

The most prominent taxonomic order was Rhabditida with a mean relative read abundance of 93 ± 21% and 93 ± 15% in A. alfari and A. constructor colonies, respectively (Fig. 3A). ASVs from this order were found in every ant colony and mostly assigned to the genera Sclerorhabditis, Diploscapter and unclassified Rhabditida. The 11 most abundant unclassified Rhabditida ASVs were closely related (> 97.5% similarity, Supplementary Information S4) and are therefore likely to belong to the unidentifiable Rhabditida species detected by morphological analysis.

Fig. 3.

Comparison of nematode communities inhabiting ant-made patches across different ant colony development stages. (A) Relative read abundances (%) of abundant genera (> 0.5%) per ant colony, grouped by Cecropia plant species (green), Azteca ant species (grey) and ant colony developmental stages (brown). Low abundant taxa (< 0.5%) are merged as “Rare”. (B) Alpha diversity (Shannon Index) and beta diversity (PCoA showing Bray-Curtis dissimilarity) metrics of each ant species. Statistical significance (p < 0.05) is calculated by Kruskal-Wallis or Wilcoxon test and by Permanova and Permdisp tests, respectively.

The metabarcoding data showed additional ASVs assigned to the order Tylenchida (genera Aphelenchoides and Anguina) were detected in: (i) 12 out of 36 A. alfari colonies with an average 7 ± 21% mean rel. read abundance, and, (ii) in 13 out of 23 A. constructor colonies with an average 2 ± 9% mean rel. read abundance (Fig. 3A). This order exhibited a sporadic dominance in certain initial and young patches, but was generally rare in established patches (0.5% mean relative abundance). Moreover, ASVs assigned to the taxonomic order Dorylaimida (mainly genus Mesodorylaimus) were detected only in A. constructor established colonies (6 out of 19) mostly nesting in C. obstusifolia trees (Fig. 3A). Dorylaimida accounted for 18 ± 18% rel. read abundance in these colonies.

Nematode diversity and community composition remain stable across ant colonies of different ages

The alpha diversity of the nematode communities (Fig. 3B) was stable across the different developmental stages of the ant colonies, both in A. alfari and A. constructor colonies (p = 0.18, x² = 3.40; and p = 0.19, x² = 1.68, respectively). Similarly, the nematode community composition (Fig. 3B) did not significantly correlate with the ant colony developmental stages in A. constructor patches (p = 0.36, F = 1.05), while it was slightly correlated in A. alfari colonies (p = 0.04, F = 1.86). Furthermore, no significant differences were detected in the alpha and beta diversity analyses when comparing tree sections of established patches (p = 0.19, x² = 3.29; and p = 0.51, F = 0.91, respectively; Supplementary Information S5).

The nematode community composition is ant species specific

In order to evaluate whether the diversity or community composition of nematodes inhabiting the patches exhibits a significant correlation with the ant or plant species, we conducted alpha and beta diversity analyses across colonies. The Shannon Index showed that the nematode communities in A. alfari colonies exhibited a significantly higher diversity than those in A. constructor colonies in both, the initial (p = 0.037, x² = 4.36) and the established (p = 0.005, x² = 8.06) patches (Fig. 4A). Furthermore, the Bray-Curtis dissimilarity and the Permanova test revealed statistically significant differences in the nematode communities between A. alfari and A. constructor in both, the initial (p = 0.006, F = 4.65) and the established (p = 0.001, F = 12.71) patches (Fig. 4A).

Fig. 4.

Comparison of nematode communities inhabiting ant-made patches between the different ant and plant species. (A) Alpha diversity (Shannon Index) and beta diversity (PCoA showing Bray-Curtis dissimilarity) metrics of initial and established patches between ant species. (B) Alpha diversity (Shannon Index) and beta diversity (PCoA showing Bray-Curtis dissimilarity) metrics of established patches between plant species. Statistical significance (p < 0.05) is calculated by Wilcoxon test and by Permanova and Permdisp tests, respectively.

In contrast, the nematode alpha and beta diversity in established patches of each ant species did not vary between plant species (C. peltata and C. obstusifolia) (Fig. 4B). Given the unbalanced sample size between the tested groups (Supplementary Information S1), an additional PERMDISP test was performed in beta diversity analyses to ensure sufficient statistical robustness (Fig. 4).

Rhabditida species are shared between ant species but differ in relative abundance

To further evaluate the effect of the ant species on the nematode communities, we compared the relative abundances of the three most abundant nematode species (> 10% mean rel. abundance) in the established patches of each ant species based on metabarcoding analysis (Fig. 5A) and morphological identification methods (Fig. 5B).

Fig. 5.

Relative abundance (%) of the most abundant nematode groups from the order Rhabditida in established patch samples of A. alfari (ALF) and A. constructor (CON) colonies based on: (A) 18S rRNA sequence reads from amplicon sequencing; and (B) nematode specimen counts following morphological identification methods. Statistical significance (p < 0.05) is calculated by Wilcoxon test.

Both relative read abundance and morphological nematode counts indicated that the unclassified Rhabditida group was significantly less abundant in A. constructor colonies than in A. alfari colonies (p < 0.001) (Fig. 5). Instead, Diploscapter was significantly more abundant in the established patches of A. constructor colonies compared to A. alfari colonies (p < 0.001) (Fig. 5).

Discussion

The nematode inhabitants of the Azteca-Cecropia patches

Three distinct orders of nematodes (Rhabditida, Tylenchida and Dorylaimida) exhibiting diverse putative feeding strategies were identified through metabarcoding analysis of ant-made patches in the Azteca-Cecropia mutualism (Fig. 3A). As observed in other ant-plant mutualisms and social insect nests^8,16,17, bacteria-grazing Rhabditida nematodes were by far the most abundant and frequent group across the 65 Azteca colonies investigated. Large masses of these nematodes were observed in-situ, actively moving on the patch surfaces of established colonies (Supplementary Information S6).

In addition to the dominant Rhabditida, nematodes from Tylenchida (e.g., Aphelenchoides) and Dorylaimida (e.g., Mesodorylaimus) appeared sporadically in the patches. Members of the genus Aphelenchoides species are known to feed on either fungi or plant tissue⁴⁷, which likely explains their high relative abundance in initial and young patches, where plant material is the dominant substrate and fast-growing fungi are present²³. Aphelenchoides species were also isolated from fungus-growing termite nests¹⁴ and the oviducts of carpenter bees⁴⁸. Nematodes of the genus Mesodorylaimus are typically omnivorous⁴⁹ and, in some cases, predators of other nematodes⁵⁰. This feeding behaviour is consistent with their presence in patches of established A. constructor colonies, which often have a high biomass of bacterivorous rhabditids. Overall, these findings indicate that the ant-made patches offer previously unrecognized niche opportunities that facilitate the development of nematodes from multiple trophic levels.

Transmission and development of the patch nematode communities

The homogeneity of the nematode diversity and community composition between the initial and established patches shown in this study provides key information about the transmission and spatio-temporal dynamics of the nematode patch communities in relation to the ant colony development (Fig. 3B). Previous studies have shown that alate Azteca queens carry Rhabditida nematodes either externally attached to their bodies¹⁶ or internally within their infrabuccal pockets, where they store patch particles collected from their mother colonies²⁶. After entering a hollow Cecropia stem, the founding queen creates a pile of parenchyma, which she likely inoculates with the nematodes, bacteria, and fungi she carries in her infrabuccal pocket²⁶. As the colony grows, the worker ants expand the nest by colonizing additional internodes of the tree and building new patch structures²³.

While we did not observe any alate queens leaving their mother colony for the nuptial flight, we managed to catch a few young queens that recently entered the Cecropia stem with rhabditid nematodes in their freshly made initial patch. The combination of evidence from previous studies that alate queens carry phoretic nematodes^16,17,26, and our finding that Rhabditida nematodes are present in all patches from the earliest stages of colony foundation strongly suggests vertical transmission of these nematodes through the queen from mother to daughter colonies. Furthermore, the high similarity in nematode community composition between freshly made patches in recently grown plant internodes and the patches in older domatia (Supplementary Information S5) supports horizontal transmission of Rhabditida nematodes across internodes as the ant colony expands within the Cecropia tree.

Unlike rhabditids, tylenchids and dorylaimids were not widespread among the different developmental stages of the ant colonies (Fig. 3A). The Mesodorylaimus ASVs (Dorylaimidae, Dorylaimida) were found to be dominant only in established patches from A. constructor colonies inhabiting Cecropia obtusifolia trees inhabited by large ant colonies with numerous individuals and thus, Mesodorylaimus might be rather transmitted via patrolling of ant workers or patch visitors like mites or flies. Like in human-made compost piles^51,52, Mesodorylaimus appeared only after the fast-growing bacterivorous nematodes had reached dominance within the patches during the earlier phase of patch maturation. In contrast, the Aphelenchoides ASVs (Aphelenchoididae, Tylenchida) were relatively abundant in some initial and young patches of all three Azteca species but not in established ones. This suggests that they may have been introduced by the queen from the surrounding environment during the plant colonization. Their absence in established patches suggests that Aphelenchoides feeds on plant material, since the substrate of initial and young patches is predominantly plant-derived parenchyma, whereas in older patches it is ant-derived exoskeletons of dead nestmates.

Influence of the Azteca ant and Cecropia plant species on the nematode communities

The bacteria-grazing Rhabditida nematodes detected in the patches belong to the genera Diploscapter and Sclerorhabditis (both 0.5–1 mm long) and to an unclassified Rhabditida species characterised by a notably small body size. All three Rhabditida species were present together in 29 out of the 31 established colonies investigated. However, it appears that the patch of each ant species favours the development of different Rhabditida nematodes (Figs. 4A and 5), even though these ant species coexist in the same geographical area and environment. The differences in the relative abundance of Diploscapter and the unclassified Rhabditida group in A. alfari and A. constructor, characterized by differing nematode body sizes, may be due to the unique patch habitat that each ant species creates²² and which could influence whether the habitat is suitable for a particular nematode species⁵³. Nematodes need water films for movement and feeding^54,55. Furthermore, in soil, the distribution of nematode size has been shown to correlate with the distribution of aggregate and pore size, as larger nematodes require larger inter-aggregate spaces than smaller nematodes^53,54. In the domatia, A. constructor forms relatively big, three dimensional and very moist patches, whereas A. alfari patches are thin, dry and sandy²³. The low moisture level and physical structure in A. alfari patches may favour smaller nematodes as has been observed in soils under drought stress⁵⁶. Overall, the ant species is a significant driver of the nematode diversity within these patches, as it has recently been shown for the fungal and bacterial communities in the patches of the Azteca-Cecropia association^22,23,28.

Ant-plant-nematode associations: parasitism, commensalism or mutualism?

The first record of rhabditid nematodes associated with ants was a 20–30 million years old Dominican amber dauerlarvae of a diplogastrid nematode next to a fossil Azteca ant⁵⁷. More recent fossil evidence indicates that nematodes and ants have coexisted for at least 40 million years¹⁵ and during which a multitude of symbiotic relationships have emerged.

Associations between nematodes and ants were traditionally classified as commensal or antagonistic, with the latter proposed to have evolved from commensal interactions which form the basis of most nematode–insect relationships⁵⁸. The occurrence of nematode dauerlarvae in the postpharyngeal glands of ants has been documented since the 19th century⁵⁹ and is now reported from numerous species of ants⁸. In their adult stages, these nematodes typically feed on free-living bacteria inhabiting the ant nest. When external conditions become unfavourable (e.g., low humidity or food scarcity), dauerlarvae enter the glands and use ants as vehicles for dispersal^6,57. Since minor damage to postpharyngeal glands of alates had been documented, rhabditid nematodes were initially considered facultative parasites of ants⁵⁷.

Although the interaction between nematodes and ant-plant mutualisms also includes phoresy, recent studies have proposed that ant-plant-nematode associations are rather commensal or even mutualistic^8,16,17. Our results indicate that the association with Azteca ants is clearly advantageous for the nematodes. Ant-made patches provide a stable and protected microenvironment rich in bacteria which ensures a continuous nutrient supply. Moreover, alate queens facilitate nematode dispersal during colony founding as dauerlarvae attach to their bodies and are transported to new nesting sites^16,26. These benefits are likely shared across other ant–plant mutualisms^8,17, suggesting that such associations represent an effective and recurrent ecological strategy for the nematodes.

In contrast, the benefits for the ants, and for the Azteca–Cecropia complex, are less evident. Ants are known to maintain high nest hygiene to prevent infections⁶⁰. Therefore, large nematode populations would likely not be tolerated if they posed a health risk for the ant colony. Although context-dependent fitness effects cannot be fully excluded, as observed in other systems⁶¹, we found no evidence of detrimental effects in the ant colonies which remained healthy, or, the host trees that showed no signs of stress. Instead, the consistent occurrence of abundant nematodes in all colonies, even in domatia with the greatest activity next to brood and queens, rather supports the hypothesis that nematodes play a functional role within this complex ecosystem.

Previous studies have suggested that nematodes may serve as a nutritional supplement for ant larvae during periods of food scarcity and that bacteria could be cultivated to nourish the nematodes, which would in turn provide food for the ants^8,16,17. However, attempts to verify this in the Azteca–Cecropia association have been unsuccessful, as neither predation nor active feeding on nematodes has been observed⁸. Beyond potential trophic interactions, we propose that nematodes may contribute to the stability and functioning of microbial communities within the patches of the Azteca–Cecropia mutualism and other ant–plant associations.

As shown in forest soils, agricultural systems, and composts^49,62–65, bacterivorous nematodes could provide a wide range of beneficial ecosystem services within patches. First, they may facilitate organic matter decomposition and nutrient recycling. By feeding on bacterial biomass and excreting ammonia, these nematodes likely contribute to nitrogen mineralization in the patches⁶⁶. Because certain bacterial taxa within the patches fix atmospheric nitrogen at high rates²⁸, nematode grazing could further enhance nitrogen cycling. Second, bacterivorous nematodes could contribute to nest hygiene by reducing bacterial loads and potentially suppressing pathogenic strains. Third, their constant movement through the patch surface likely promotes aeration and mixing of organic material (Supplementary Information S6). Overall, these observations suggest that the phoretic association between Azteca ants and bacterivorous nematodes is not merely commensal but possibly mutualistic, with nematodes acting as micro-scale ecosystem engineers that help maintain the stability of the patch ecosystem.

Conclusions

The large-scale 18S rRNA gene metabarcoding approach used in this study combined with morphological identification methods provided valuable semi-quantitative insights into the taxonomic diversity and community composition of nematodes inhabiting the ant-made patches of the Azteca–Cecropia mutualism. While bacterivorous nematodes from the order Rhabditida like Sclerorhabditis and Diploscapter were present in all samples, members of Tylenchida and Dorylaimida were only sporadically detected. Our findings support the previously hypothesized vertical transmission of nematodes from mother colonies to daughter colonies, and also suggest horizontal transmission between individual patches within the same ant colony. Moreover, each Azteca ant species appears to create and maintain a distinct microenvironment within its nest, favouring the development of specific Rhabditida nematodes in the patches.

Based on these results, we conclude that bacterivorous nematodes represent a permanent and functionally relevant component of this complex ant–plant ecosystem. Further research is needed to determine whether these nematodes contribute to the stability and functionality of the patches by, for example, selectively grazing on particular bacterial taxa, maintaining a balanced bacterial growth, or, contributing to organic matter degradation and nutrient recycling.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

V.E.M., M.N. and D.W. designed the research. V.B.S., V.E.M. and M.N. performed the research. V.E.M. and M.N. identified the plant and ant species. V.B.S., B.H. and W.S. contributed with analytical tools. V.B.S., B.H. and W.S. processed and analysed the data; V.B.S analysed and interpreted the results. V.E.M., W.S. and D.W. provided significant intellectual contribution. V.B.S wrote the original draft of the manuscript. All authors contributed to the final version of the manuscript and approved it for publication.

Funding

This research was funded in whole, or in part, by the Austrian Science Fund (FWF) [https://doi.org/10.55776/P31990 to VEM]. For open access purposes, the author has applied a CC BY public copyright license to any author accepted manuscript version arising from this submission.

Data availability

All sequence data (raw sequence reads and metadata) are accessible on NCBI under the BioProject accession number PRJNA777006. The 18S rRNA amplicon sequencing dataset and the R script generated in this study are available in the Figshare public repository at https://doi.org/10.6084/m9.figshare.c.7499025.

Declarations

Competing interests

The authors declare no competing interests.

Ethics declaration

This research was conducted with authorization from the Costa Rican authorities who provided the following research permits “R‑046–2015‑OT‑CONAGE‑ BIO,” “SINAC INV‑ACOSA‑001–16,” and “SINAC INV‑ACOSA‑013–18.”. These permits allowed the collection of Cecropia plants colonized by Azteca ants. We would like to emphasize that Cecropia species are pioneer, fast-growing plants which are widely distributed throughout Costa Rica. They are particularly common in the area of sampling along roadsides, in domestic gardens, and in abandoned agricultural fields. Some samples were even collected from trees that had been felled by local residents due to interference with roadside power lines. Consequently, the sampled plants are not endangered, and their collection did not pose a conservation concern. In accordance to these research permits, we followed the national laws of Costa Rica “Ley de Conservación de la Vida Silvestre N° 7317,” “Ley de Biodiversidad N° 7788,” and “Ley Orgánica del Ambiente N° 7554” and applied the General Protocol “MINAE‑SINAC‑P‑001” for the use of the Protected Wildlife Areas of the National System of Conservation Areas. Benefits from this research accrue from sharing our data on public databases.