Morphological divergence of domatia in ant-free populations of the widespread Neotropical myrmecophyte Miconia tococa (Melastomataceae)

Abstract

Co-evolving organisms experience multiple selection pressures that may lead to trait mismatches among different populations and sites. In defensive ant–plant mutualisms, host plants (myrmecophytes) produce specialized shelters (domatia) to harbour specialized ant-partners in exchange for protection against enemies. Although populations of myrmecophytes without ants occur in some locations, there are no records of changes in domatia morphology—at the population level—due to the absence of symbiotic ants. We conducted broad-scale samplings of Miconia tococa (Melastomataceae) populations across the Brazilian Cerrado and a 2-year transplant experiment to test whether domatia morphology changes when symbiotic ants are naturally absent. Domatia were 33.9% smaller in ant-free populations than in ant-inhabited populations. Transplants revealed that host plants from ant-inhabited sites still developed larger domatia than those from ant-free sites, even in the absence of ant-partners. These findings point to a change of M. tococa traits associated with biotic defences where symbiotic ants are absent. What may have begun as a plastic adjustment to ant-free environments appears to have been transformed into fixed (genetic) interpopulation differences over time, indicating a potential local destabilization of the mutualism or a mechanism to stabilize the interaction at the landscape scale.

Introduction

In mutualisms, traits co-evolve in response to selection for mutual access to rewards and services (Anderson 2015). For example, in defensive ant–plant mutualisms, ant-plants (myrmecophytes) provide specialized shelter structures (domatia) for colonies of some ant species in exchange for defence against enemies and competitors (Heil and McKey 2003, Michelangeli 2005). This type of mutualism is generally obligate, i.e. neither plants nor ants can survive without each other (Moraes and Vasconcelos 2009), but different plant populations may associate with different ant species (Fonseca and Ganade 1996, Fiala et al. 1999, Brouat et al. 2001, Shenoy and Borges 2010). However, there are locations where myrmecophytes occur without being associated with any symbiotic ant, thus living without a biotic defence service (Janzen 1973, Putz and Holbrook 1988, Moraes and Vasconcelos 2009). The causes of such ant-free locations potentially include geographical barriers allied to the low dispersal capacity of symbiotic ants, local partner extinction by environmental changes, or reduced pressure from plant enemies (Kiers et al. 2010, Türke et al. 2010, Bartimachi et al. 2015, Plowman et al. 2017). Nevertheless, as the stability and persistence of specialized mutualisms and their associated traits require that partners coexist over time and space, a lingering uncertainty is whether isolation (e.g. plants living in the absence of symbiotic ants) could lead to a permanent change in plant mutualistic traits (Janzen 1973, Rickson 1977, Gutiérrez-Valencia et al. 2017).

Some myrmecophytes can modify their traits in response to the presence and quality of the partner. At the individual level, ant-free plants can reduce or cease the production of nutrient-rich rewards (Janzen 1973, Putz and Holbrook 1988, Heil et al. 1997), as they would represent a drain of plant resources without any benefits from partner services (Rickson 1977, Heil et al. 1997). Similarly, domatia plasticity has also been observed in some species. For instance, in response to high herbivory pressure, Cordia nodosa (Boraginaceae) trees can increase domatium volume, favouring larger ant colonies and potentially enhancing anti-herbivore defence (Frederickson et al. 2013); conversely, they can also reduce domatia volume when ant-protection is inefficient (Edwards et al. 2006). Finally, Hirtella myrmecophila (Chrysobalanaceae) can even prematurely abort domatia development to prevent the establishment of opportunistic ants (Izzo and Vasconcelos 2002). These examples indicate that some myrmecophytes may adjust the size and availability of domatia due to interaction quality and herbivory pressure. However, population-level studies of domatia morphology in response to the variation of mutualist protection are surprisingly rare (Shenoy and Borges 2010), and none has assessed the occurrence of permanent alterations in isolated populations. Thus, our understanding of the evolutionary direction of myrmecophytic traits in ant-free plant populations remains limited.

Here we investigated whether geographical isolation of partners leads to trait divergence in an ant–plant mutualism. We hypothesize that domatia morphology may be different in populations where symbiotic ants are naturally absent. We tested this by conducting a large-scale field sampling in ant-free and ant-inhabited populations of Miconia tococa (Desr.) Michelang. (Melastomataceae; formerly known as Tococa guianensis Aubl.; Michelangeli et al. 2019) across a large section of its distributional range (Fig. 1). We measured and compared domatia volume, which is the key morphological host trait directly associated with this mutualism. Over a period of 2 years, we also cultivated newly formed and uncolonized seedlings experimentally transplanted from different origins to an ant-free locality, testing whether the alteration of domatia morphology may be adaptive or plastic. We discuss the implications of these findings for the stability of the mutualism, particularly considering pervasive environmental changes found in the Cerrado biome of Central Brazil.

Methods

Study area and system

We surveyed M. tococa populations in 19 remnants of the Cerrado (the Brazilian savanna) in the states of Mato Grosso (MT), Goiás (GO), and Minas Gerais (MG), separated from each other by at least 40 km (Fig. 1). We found M. tococa in 15 locations, inhabiting areas adjacent to small streams or headwaters within the understorey of gallery forests (Supporting Information, Fig. S1). The surveyed region is characterized by a highly seasonal, tropical climate showing marked dry and rainy seasons (Alvares et al. 2013). Mean annual temperature is 20–25°C and accumulated annual precipitation is 1500–1700 mm (Table S1).

The genus Miconia (Melastomataceae) is widely distributed in the Neotropics (Michelangeli 2005). In the Cerrado, M. tococa is usually associated with symbiotic ant species from the genera Allomerus (Myrmicinae), Crematogaster (Myrmicinae), and Azteca (Formicidae; Fonseca and Ganade 1996). Their domatia develop early in the plant’s life, from the production of the first leaves (Leroy et al. 2010), and are formed by two pouches—adjacent, independent, ovoid chambers at the leaf base opening in the abaxial surface (Supporting Information, Fig. S1; Michelangeli 2005). In MT and GO states, plants occur associated mostly with the symbiotic ant Allomerus octoarticulatus (Fig. S1), whereas surveys carried out over the last two decades have shown that plant populations from MG state are ant-free or inhabited only by opportunistic ants, which do not offer protective services and do not influence levels of herbivory (Rosumek et al. 2009). In the M. tococa system, opportunistic species belong to the genera Brachymyrmex, Camponotus, Cephalotes, Crematogaster, Linepithema, Pheidole, Solenopsis, Tapinoma, and Wasmannia (Moraes and Vasconcelos 2009, Bartimachi et al. 2015).

Sampling and experimental design

As domatia are foliar structures in M. tococa, larger leaves presumably show larger domatia (Fonseca 1993). Thus, we collected 3–6 leaves per plant from 6–15 individuals in each occurrence location and measured both leaf and domatia dimensions. In total, we sampled 833 leaves from 151 plants across the 15 populations in January 2019 (Supporting Information, Table S1). To standardize leaf age, we collected only fully expanded, mature leaves in the apical portion of branches, between the last three nodes, from plants ranging between 1 and 2 m in height. We collected leaves showing low herbivory damage (<10%) to allow an accurate measurement of leaf and domatium dimensions. Ants found on leaves or within domatia were identified by specialists and comparisons with the reference collection of the Social Insect Ecology Laboratory (LEIS) at the Federal University of Uberlândia (UFU). We identified 13 ant species, three symbiotic and 10 classified as opportunistic (Table S2).

Leaf area was estimated with the equation: area = ×0.325 + 0.732 × LW, where L and W are maximum leaf length and width, respectively, with a strong relationship (R² = 0.98, N = 111, P < .001; Moraes and Vasconcelos 2009). Since M. tococa domatia are formed by two similar pouches (Supporting Information, Fig. S1), we estimated the volume of both structures separately and then summed the values to obtain the entire domatium volume. For this we used the adapted equation for a cone: $volume=1/3(πabc)$⁠, where a and b are the half-values of length and width, respectively, and c is height (Edwards et al. 2006, Frederickson et al. 2013).

To determine whether morphological differences in domatia are adaptive or a result of phenotypic plasticity, we used a dataset of plants from a transplant experiment conducted without the presence of symbiotic ants. This approach is frequently used to monitor the growth of plants from different populations in a common setting (Scheepens et al. 2010, Bradley St. Clair et al. 2013, Villemereuil et al. 2016). We collected seedlings from Cachoeira da Fumaça, MT, and Aragarças, GO (localities with natural occurrence of symbiotic ants), and Uberlândia, MG (natural absence). To control for potential ontogenetic effects, we only used newly formed seedlings (2–4 leaves and 5 cm in height) not yet colonized by A. octoarticulatus and present in environments with similar light, humidity, and temperature conditions. Seedlings were maintained in a glasshouse under the same conditions for 4 months until they reached 15–30 cm height. Thereafter, 60 seedlings (20 from each population) were planted in 20 points (one seedling from each origin) regularly distributed along a 400-m transect through a gallery forest remnant in the Gloria Experimental Farm, owned by UFU in Uberlândia-MG (Supporting Information, Fig. S1). After 2 years (from July 2011 to June 2013), we collected and measured the leaves and domatia sizes from living plants using the above-mentioned protocols.

Statistical analyses

We conducted Spearman’s rank correlations between domatia volume and leaf area for populations with and without symbiotic ants, including those in the transplant experiment. Since there was a positive relationship between domatia volume and leaf area (Supporting Information, Fig. S2), we incorporated the latter in further analysis as an offset term to correct for any ontogenetic factors that may influence domatia size in addition to the predictor variables of interest (see below). To test if domatia volumes differ among populations with and without symbiotic ants, we fitted a GLMM (generalized linear mixed model) adjusting a Gamma distribution with log-link function. Individual plant identity nested within population identity was treated as a random effect and the logarithm of leaf area as an offset term. We then fitted another GLMM to test if domatia volumes differ among the three populations from different origins in our transplant experiment. Leaf area was considered an offset, and individual plants the random effect. Post-hoc analyses were conducted via Tukey adjusted contrasts using the package emmeans (Lenth et al. 2022).

The GLMMs were built using the R-package glmmTMB v.1.0.2.1 (Brooks et al. 2017). We checked model fit by simulating residuals 1000 times and accessing their QQ plot and the plot of residual versus predicted values using the R-package DHARMa v.0.4.1 (Hartig 2020). We assessed the significance of models using type II Wald chi-square tests though the R-package car v.3.0.10 (Fox and Weisberg 2019). We used the ggpredict function available in the R-package ggeffects v.1.1.0 (Lüdecke 2018) to obtain the back-transformed model-adjusted (predicted) values. Analyses were conducted in R software v.4.2.2 (R Core Team 2023).

Results

We found that domatia volume was significantly different among populations (χ² = 33.79, d.f. = 1, P < .001; Fig. 2A). Domatia without symbiotic ants were on average 33.9% smaller than those with ants present, regardless of leaf area (marginal mean ± SE: 294.1 ± 0.1 vs. 444.8 ± 0.1 mm³, respectively). This pattern of relatively larger domatia was consistent across all populations of ant-inhabited plants when compared to ant-free plants (Supporting Information, Fig. S3). We also found significant effects in the transplant experiment when comparing plants from different origins growing in the same area (χ² = 23.07, d.f. = 2, P < .001; Fig. 2B). Plants from the Uberlândia population, where symbiotic ants are naturally absent, had domatia 38.2% and 36.7% smaller (182.2 ± 18.3 mm³) than plants from both Aragarças and Cachoeira (294.7 ± 18.9 and 287.7 ± 19.1 mm³, respectively), where symbiotic ants occur. Plants from ant-inhabited origins were not significantly different. These findings support the general pattern of relatively larger domatia in plants from sites with natural occurrence of ant-partners, regardless of the conditions in which the plant originally grows or leaf size.

Discussion

Intraspecific variability in ant–plant mutualistic traits have been seldom reported partly because of a lack of large-scale sampling of populations (Shenoy and Borges 2010). Here, we report data from an extensive plant collection, which covered a linear transect of ~1000 km of the geographical distribution of a widespread Neotropical myrmecophyte. The comparison between 15 populations of M. tococa in the Cerrado biome revealed that the absence of symbiotic ants can lead to changes in domatia morphology. Additionally, our experimental data indicate that divergence among populations has a genetic basis, highlighting potential outcomes of partner isolation for ant–plant mutualisms.

We found a marked difference in domatia volume between plants from ant-inhabited and ant-free populations. Previous studies have shown that plants from some myrmecophyte species may invest more in spacious domatia as a plastic response to attract ants, since larger structures allow plants to harbour more ants and thereby improve patrolling activity to reduce herbivory (Edwards et al. 2006, Frederickson et al. 2013, Chanam et al. 2014). Following this assessment, we presume that M. tococa plants can optimize the protection services by modulating the availability of nesting space for ants. Thus, plants from populations without ant-partners may have smaller domatia and limit energy expenditure as benefits from a biotic defence are not guaranteed, while they allocate available resources to other forms of defence, such as leaf density of trichomes (Bartimachi et al. 2015). However, it is important to note that domatia in M. tococa are small expansions/inflations of photosynthetic leaf tissue near to the petiole and, consequently, the requirement for additional cell and tissue production could be offset by photosynthesis of domatia. If the costs of producing domatia are low, the explanation for the observed variation in this mutualist trait cannot be reduced to a simple cost–benefit balance. Thus, further investigations are needed to elucidate whether the reduction in relative size of domatia in ant-free populations of M. tococa was also driven by other biotic and abiotic factors.

Furthermore, results from the transplant experiment were congruent with the observed patterns of geographical variation, as seedlings from populations associated with symbiotic ants, but growing without an ant-partner, still produced larger domatia than plants from an ant-free population. Contrary to observations in previous studies of morphological variation of domatia (Izzo and Vasconcelos 2002, Edwards et al. 2006, Frederickson et al. 2013), our experimental findings indicated that interpopulation divergence in M. tococa cannot be regarded as a simple plastic response. The formation of small domatia seems to be a fixed morphological adaptation of plants from populations that lost the mutualist interaction with an effective defensive ant species. Although we have experimentally detected such a response for a generation, this is presumably an ongoing process (Rickson 1977). We recognize that potential environmental effects during very early development and phenotypic plasticity in domatia size cannot be completely ruled out due to the use of seedlings instead of seeds in the experiment. However, we selected only newly formed and uncolonized seedlings, and we believe that this minimizes potential environmental effects due to their brief exposure in their original habitats. Nonetheless, these findings support our hypothesis that the presence or absence of symbiotic ants plays an important role in determining structural traits in myrmecophytes related to defensive mutualisms.

Some myrmecophytes may also adjust the size and availability of domatia in response to herbivory pressure (Edwards et al. 2006, Frederickson et al. 2013). Although we have not measured local leaf damage occurring over our large-scale assessment, our findings are not related to a presumed and unrecorded difference in herbivory among sites. In a previous ant-exclusion experiment with some M. tococa populations used here, low herbivory rates were recorded in two distant locations, Cachoeira da Fumaça, MT (ant presence/large domatia), and Uberlândia, MG (ant absence/small domatia; Bartimachi et al. 2015). Moreover, our transplant experiment revealed that even under controlled settings and low herbivory, plants from sites with ant-partners present produced larger domatia than those from ant-free sites. These results do not suggest that herbivory is unimportant but indicate that this selective pressure is not the primary factor explaining the observed differences in domatia sizes among ant-inhabited and ant-free populations. Hence, regardless of local herbivory rates, plants produce larger domatia in sites where M. tococa occurs with symbiotic ants.

Populations of myrmecophytes from the genus Cecropia also occur naturally along some Caribbean islands without symbiotic ants (Janzen 1973, Rickson 1977). In this system, the reduction or loss of food rewards by plants living on distant islands was recognized as evidence of a breakdown in mutualism (Gutiérrez-Valencia et al. 2017). Likewise, the apparent discontinuity in the distribution of M. tococa (Fig. 1)—potentially associated with the low dispersal capability of symbiotic ants (Türke et al. 2010)—may be the best explanation for the occurrence of ant-free populations reported in our study. Furthermore, the reduction in domatia size may indicate a potential dissolution of defensive mutualism in M. tococa populations isolated from symbiotic partners. This finding is important for evaluating mutualistic interactions, given the pervasive environmental changes found over the studied area in Central Brazil. Currently, the Cerrado landscape is dominated by a fragmented natural environment, with disconnected remnants surrounded by extensive pastures and crops (Carvalho et al. 2009). Such a situation may reinforce the isolation of dispersed populations of M. tococa, further pushing the studied defensive mutualism towards future collapse, and serves as a warming for what could happen to other specialist mutualistic interactions following anthropogenic changes, with unpredictable indirect impacts on ecosystem structure and function (Kamaru et al. 2024).

Our results suggest that selection may generate divergence in mutualistic traits between ant-free and ant-inhabited populations of myrmecophytes. What were potentially plastic adjustments to ant-free environments with low herbivory may have become interpopulational fixed (probably genetic) differences over time. The implications of these findings for the stability of defensive ant–plant mutualisms are scale-dependent. Locally, they could signal a dissolution of mutualism interactions in ant-free areas, as host plants become less suitable to symbiotic ants. Regionally, they may indicate a mechanism for the maintenance of this interaction in changing environments. Spatiotemporal variations in abiotic and biotic conditions result in locations where plants can persist even in the absence of ant-partners, while still retaining the ability to associate by not completely ceasing production of mutualistic traits. Our results highlight the importance of partner presence for the maintenance of mutualistic traits. However, further studies are necessary to test the extent of these patterns in other ant–plant mutualistic systems.

Acknowledgments

We thank Marcello Barbosa and Ana Silva for their hospitality, as well as Lenimar Barbosa and Jaqueline Silva for their help during field and laboratory work, and Rogério Gonçalves for map preparation. We also thank the editors and anonymous reviewers for valuable comments and suggestions that helped to improve the paper.

Funding

This research was supported by the Graduate Program in Ecology, Conservation and Biodiversity at the Federal University of Uberlândia, and by research grants from the Brazilian Council for Research and Scientific Development (CNPq grants 479135/2010-0 and 441142/2020-6 to H.L.V.) and the Research Foundation of Minas Gerais state (FAPEMIG APQ 03372-21). A.B. and F.T. were granted postdoctoral fellowships by the Brazilian Coordination for Improvement of Higher Education Personnel (CAPES) (PNPD; projects 88882.314746 and 1659767). J.C.F.C. was supported by the National Council for Scientific and Technological Development (CNPq; project 152014/2022-5).

Data Availability

Data will be shared upon reasonable request to the corresponding author.

References