Introduction

Mutualisms between plants and insects represent a major topic in ecology and evolution [1]. An important benefit that insects provide to plants is seed dispersal, with myrmecochory (i.e., seed dispersal by ants) standing out as the sole true seed dispersal syndrome among invertebrates, observed in over 11,000 plant species [2–4]. Myrmecochorous seeds are characterized by an oily appendage, called elaiosome, which ants use as a food source [5,6]. While seeds provide food to ants, the latter can increase the reproductive success of plants by manipulating and moving their seeds to places with better development odds [4,7]. Most studies focusing on the advantages of myrmecochory to plants focus on two mechanisms: seed removal distance and deposition site [4]. On the other hand, the effects of ants’ physical and chemical manipulation of seeds have been poorly investigated ([but see 8,9]. Because not all seed disperser ants benefit plants [10,11], comparing ant species is fundamental to understanding the holy scenario on the possible effects of myrmecochory on seed germination, plant development, and consequently, plant population dynamics.

When ants manipulate seeds, they remove the elaiosome or pulp to feed on [8,12]. During this process, ants can also promote seed disinfection by actively removing microorganisms or deploying antimicrobial substances produced mainly by their exocrine glands [9,13]. Ant cleaning strategy prevents the growth of pathogenic fungi on non-myrmecochorous seeds [9,12] and changes the fungi communities surrounding ant nests [14], but ant effect on fungi communities of myrmecochorous seeds has never been tested.

Ant species employ distinct cleaning strategies to safeguard themselves and their colonies against pathogenic microorganisms. [15–17]. Also, the frequency of use and chemical composition of antimicrobial substances secreted by ants glands are highly variable among species [16]. Therefore, each ant species can have different outcomes in their disinfection efficiency against pathogens. It is important to note that ant hygienic strategies also inhibit a variety of microorganisms with no evident harm to ants, such as phytopathogenic fungi. [13,14,18]. Therefore, distinct hygienic strategies among ant species might play divergent roles in the fungi communities present on myrmecochorous seed.

Atta sexdens (Linnaeus, 1758) and Acromyrmex subterraneus (Forel, 1893) are two sympatric and genetically related species of leaf-cutting ants with distinguishing hygienic strategies [19]. The genus Atta immune strategy is based on the production of chemical compounds by exocrine glands, mainly the metapleural and mandibular glands [19]. On the other hand, ants of the genus Acromyrmex have a symbiotic association with Pseudonocardia. In this case, the bacteria grow under the ant tegument and produce antibiotics, which act as the primary line of defence against ant-pathogenic infections [19–21]. Finally, it is known that the secretion produced by the Pseudonocardia bacteria has a smaller spectrum of action against microorganisms when compared to metapleural gland compounds [19].

Those two ant species disperse seeds of the myrmecochorous plant Mabea fistulifera Mart.- Euphorbiaceae [22]. Mabea fistulifera is a pioneer tree species widely distributed in Brazil and presents a two-phase seed dispersal system (Diplochory). When fruits ripen and open, seeds are ejected onto the ground (up to 8 m, a process known as ballochory), where ants collect and transport them to their nest [7]. Both A. sexdens and Ac. subterraneus are known to retrieve M. fistulifera seeds from the ground and transport them to their nest. In the nest, they consume the elaiosome, and afterwards, they discard the seeds on the soil surface near the nest entrance [7,8,22]. Although those two species have similar dispersal behaviours, they present contrasting outcomes in seed germination [8,19,23]. The manipulation of M. fistulifera seeds by A. sexdens improves germination rates, while the manipulation of seeds by Ac. subterraneus could cause no change or even a decrease in the germination of those seeds [8,23]. Hence, variations in hygienic strategies between A. sexdens and Ac. subterraneus ants could directly impact microbial communities, including fungi, on myrmecochorous seeds, and potentially affect germination.

Based on this, our main objective was to experimentally compare the fungi richness and composition on seeds manipulated by Atta sexdens, Acromyrmex subterraneus and unmanipulated seeds. We tested two non-excluding hypotheses: i) fungi richness is higher in seeds manipulated by Ac. subterraneus (species that negatively affect seed germination), followed by unmanipulated seeds and seeds manipulated by A. sexdens (ant species that increase seed germination). ii) seeds manipulated by A. sexdens, Ac. subterraneus and unmanipulated seeds present dissimilar fungi compositions. By doing so, we provide insight into the role of third-party microbial interactions with ant species in myrmecochory.

Material and methods

Results

We found 24 fungal morphotypes from six families in the three seed treatments (S1 Table). Among those, we could identify 19 morphotypes to the Genera level, as five morphotypes did not have spores for precise identification at any level (i.e., sterile mycelium). In total, we found 19, 17 and 19 fungal morphotypes on seeds manipulated by Ac. Subterraneus, A. sexdens and control, respectively. Despite the similar numbers on the total fungi richness, the average number of fungal morphotypes per seed manipulated by Ac. Subterraneus presented around 25% higher species richness (mean ± SE, 2.04 ± 0.12, Chi = 9.1, N = 320, p = 0.02,) than the seeds manipulated by A. sexdens (1.58 ± 0.08) and control (1.36 ± 0.06), which did not differ from each other (Chi = 2.2034, N = 320, p = 0.13, Fig 1). The non-metric multidimensional scaling analysis (NMDS) showed three distinct fungi community compositions on M. fistulifera seeds (manipulated by Ac. subterraneus, manipulated by A. sexdens, and control, Fig 2). The ANOSIM similarity analysis supported those differences among treatment groups (R² = 0.27, F_2,13 = 2.408, p = 0.003).

Download:

• PNG
  larger image
• TIFF
  original image

Fig 1. Average richness of fungal morphotypes per seed from treatments.

i) seeds previously manipulated by Acromyrmex subterraneus, ii) seeds previously manipulated by Atta sexdens, iii) seeds not manipulated by ants (Control). Treatments followed by the same letter do not differ statistically among themselves. Points represent the average, bars the standard errors, and shapes the data distribution. Ac. subterraneus show a high number of fungal morphotypes than A. sexdens and control (Chi = 9.1, N = 320, p = 0.02).

https://doi.org/10.1371/journal.pone.0293377.g001

Download:

• PNG
  larger image
• TIFF
  original image

Fig 2. Result of non-metric multidimensional scaling analysis (NMDS).

Three groups of fungal morphotypes was formed, one for each treatment i) seeds previously manipulated by Acromyrmex subterraneus, ii) seeds previously manipulated by Atta sexdens, iii) seeds not manipulated by ants (Control). NMDS was based on the distance from the matrix obtained from the Bray-Curtis index.

https://doi.org/10.1371/journal.pone.0293377.g002

According to the Species Indicator Value, some fungal morphotypes were more closely associated with a specific seed treatment group. Penicillium sp.1 and Trichoderma sp.1 were the two most frequent morphotypes on seeds manipulated by Ac. subterraneus (IndVal = 0.9, p = 0.005; IndVal = 0.82, p = 0.031, respectively), while the Sterile mycelium sp.2 was mostly associated to the control group (IndVal = 0.92, p = 0.003, S1 Table). Finally, Fusarium sp.1 and Trichothecium roseum were mainly associated with the two groups of seeds manipulated by ants, regardless of their identity (IndVal = 0.85, p = 0.020; IndVal = 0.79, p = 0.039, respectively). No specific group was associated with A. sexdens following IndVal analysis (S1 Table).

Discussion

Here, we compared seeds manipulated by different ant species and found that Acromyrmex subterraneus, a species known to negatively affect the germination of those seeds [8], exhibited the highest richness of fungi compared to Atta sexdens and unmanipulated seeds. We also demonstrate that A. sexdens ants were unable to decrease fungal richness when compared to unmanipulated Mabea fistulifera seeds. Finally, we showed that fungi composition on these myrmecochorous seeds depends on the ant species which manipulate them. Despite the extensive research on the advantages that ants provide to plants regarding seed removal distance and deposition site [4,37–39], there is limited focus on how ant identity could affect fungi presence on myrmecochorous seeds. Here, our results support the hypothesis that the manipulation of myrmecochorous seeds by different ant species leads to variations in the fungal communities found on seeds. Furthermore, because those two ant species present contrasting effects on seed germination [8,19,23], we suggest that differences in ants’ hygienic strategies contribute to the observed outcomes in seed germination.

We do not support the hypothesis that ants act as seed cleaners by reducing fungal loads on seeds, as suggested for non-myrmecochorous seeds [9,12]. Ants protect themselves and their colonies against infection by releasing antimicrobial substances from external glands or through mutualistic microorganisms [15,19,40]. Mutualist plants may also benefit from this cleaning behaviour by experiencing decreased fungal growth on their seeds [9,12] and reduced pathogenic fungal loads near ant nests where seeds are typically deposited [14,41]. Seeds manipulated by A. sexdens showed a similar number of fungal morphotypes compared to unmanipulated seeds, while seeds manipulated by Ac. subterraneus had even higher fungal loads. Therefore, these leaf-cutting ants seem not to benefit myrmecochorous seeds by reducing fungal loads on them.

Our results suggest that Ac. subterraneus is acting as a seed contaminator instead of a seed cleaner as their manipulated seeds presented more fungal morphotypes than unmanipulated ones. Acromyrmex subterraneus is a fungus-growing ant, as such its nests have appropriate conditions for developing their symbiotic fungus [42,43]. However, the same conditions might favour other fungi found growing in the ants’ fungus garden, as demonstrated for other fungus-growing ant species[44,45]. Furthermore, ants of the genus Acromyrmex have their main defences against pathogens based on the antibiotic produced by mutualistic bacteria (Pseudonocardia sp.), which has a narrow spectrum of action when compared to glandular secretions commonly used by other ant genera [19–21]. The presence of other fungi inside the ant nest plus the lower spectrum of Ac. subterraneus hygienic strategies are likely to explain the increase of fungi infection on seeds manipulated by them compared to the control group. Moreover, Ac. subterraneus also changes fungi community on seeds, and Penicillium sp.1 and Trichoderma sp.1 seem to be favoured by this manipulation (S1 Table). Here, we suggest that the increase in fungi infection and the change in fungi composition make this ant species an inefficient seed cleaner, which might lead to the decrease in the germination observed on M. fistulifera seeds manipulated by Ac. subterraneus reported by our previous study [7].

Although A. sexdens do not actively clean seeds against fungi, their contact during seed dispersal can change fungi composition on the seeds, which could have contributed to the increased germination of M. fistulifera after being manipulated by A. sexdens seeds [23]. However, it needs to be taken with care as other factors, such as elaiosome removal and seed scarification promoted by ants during the manipulation, could also play a role in seed germination rates [8,46]. Previous studies that have suggested a reduction in fungal infection on non-myrmecochorous seeds through ant manipulation typically assessed only the presence of fungi after germination, without consideration for their specific identity [9,12]. Consequently, the reported fungal infections in prior experiments did not differentiate between fungal attacks that successfully acted as pathogens and those that developed within non-viable seeds. In short, this study indicates that the fungi community present on seeds and the changes in their composition caused by ants are important aspects to consider when measuring myrmecochorous seed fate.

In literature, ant species are classified as high-quality dispersers when they rapidly find and retrieve the seeds to their nests without damaging them [11,47]. For the myrmecochory dispersal syndrome, Giladi [4] described an ant species as an ideal seed disperser if it retrieves seeds: i) rapidly enough to reduce predation risk, ii) far enough to reduce parental competition, iii) to a suitable site for plant establishment, and iv) without damaging them. Building upon our earlier [7] and the current results, we propose that seed manipulation behaviour could be considered an additional factor influencing seed fate because variations in seed manipulation could have consequential effects on seed viability.

Although our results show changes in fungi communities on seeds after being manipulated by different ant species, we have constraints on identifying morphotypes to species. We recognize limitations in our identification of microbial species. Therefore, we suggest further investigation using modern molecular techniques (e.g., next-generation sequencing) to precisely identify fungi, their functional groups and pathogenicity [14,18]. Furthermore, the fine identification of fungi species might be valuable once some fungi lineages could also positively affect seed germination and plant development [48]. Those new methods might give us more detailed information on the perspective of microbial third parties on myrmecochory and their associated ant species.

Our study demonstrates that ants do not decrease fungal loads but fungal composition on myrmecochorous seeds. We further our knowledge about seed dispersal by ants showing that fungi present on seeds depend on the ant species that manipulated them, at least for leaf-cutting ants. In contrast to most studies on myrmecochory, we focus on the differences in ant species considering the cleaning strategies rather than their ability to remove seeds and deposition sites. Our results support the inclusion of one more factor in the myrmecochorous seed fate equation to determine the quality of a seed dispenser, the ant manipulation strategies [9,18,49]. Thus, we bring new insight into the comprehension of the co-evolutionary paths of the myrmecochory mutualisms, third-part interactions with fungi and their ultimate outcomes on seed fate.

Supporting information

Acknowledgments

We are grateful to the Departamento de Engenharia Florestal (UFV) for allowing access the area to collect seeds, to Terezinha M. C. Della Lucia’s and Karina Dias Amaral for allowed us to use their ant colonies, and to Meiriele da Silva for help with fungi identification.