Contents lists available at ScienceDirect

Ecological Indicators

journal homepage: www.elsevier.com/locate/ecolind

Original Articles

Resilience to fire and climate seasonality drive the temporal dynamics ofant-plant interactions in a fire-prone ecosystem

Fernanda V. Costaa,⁎, Nico Blüthgenb, Arleu B. Viana-Juniora, Tadeu J. Guerrac, Laura Di Spiritoa,Frederico S. Nevesa

a Insect Ecology Lab, Biological Sciences Institute, Federal University of Minas Gerais, Belo Horizonte, Brazilb Ecological Networks Group, Department of Biology, Technische Universität Darmstadt, GermanycGraduate School in Botany, Federal University of Minas Gerais, Belo Horizonte, Brazil

A R T I C L E I N F O

Keywords:Campo rupestreFlammable ecosystemsFire managementInteraction networksThermal nichesThermal tolerance

A B S T R A C T

Animal-plant interactions have a major influence on ecosystem structure and functioning. Understanding towhat extent the temporal dynamics of interactions is determined by climate and disturbances is thus relevant topredict ecological and evolutionary outcomes in a changing world. Here, we assessed whether the temporaldynamics of ant-plant interactions in a mountainous fire-prone ecosystem is driven by seasonal variation inabiotic conditions, and to what extent fire disturbance alters this dynamic. We also examined the thermal re-sponses of foliage-dwelling ants in order to predict the effects of seasonal oscillations of temperature on antactivity. To do so, we monitored ant-plant associations in 35 sampling plots for one year before an unmanagedfire has occurred. Then, 26 burned and nine unburned plots were monitored for another year after fire. We foundthat warmer and wetter conditions let to increases in the diversity and frequency of ant-plant interactions,mainly via upturns in plant resource availability and ant foraging activity. Beyond the positive effects of tem-perature on interaction networks, however, ant species exhibited a low heterogeneity and a huge overlap inthermal niches. Moreover, fire has led to short-term negative effects on the diversity and frequency of inter-actions in ant-plant networks. In spite of it, these network metrics in burned plots took up to half a year to returnto similar levels from unburned plots, highlighting the resilience of ant-plant interactions after fire disturbances.This study shows that wide thermal niches of ant species and fire resilience likely beget ant-plant networksreliability over seasons. The high overlap and broad thermal niches of ant species interacting with plant re-sources suggest that ant diversity plays a minor role in the tolerance against climatic changes in this fire-adaptedcommunity. These findings open a new pathway to explore thermal responses of species and their ecologicalinteractions in broader gradients of environmental conditions and ecosystem disturbances. We advocate thatlong-term studies comprising assisted burnings are desirable to forecast the impact of fire regimes and how theirsynergy with climate would affect the functioning of fire-prone ecosystems. Lastly, this study adds evidence thatstudied interaction networks can be useful to monitor the impacts of environmental changes such as anthro-pogenic disturbances, being representative of many even more complex species interactions in ecosystems.

1. Introduction

Animal-plant interactions have a major influence on ecosystemstructure and functioning (Valiente-Banuet et al., 2015). Ants act asmutualist and antagonist of a great diversity of plants (Bronstein, 1998;Rico-Gray and Oliveira, 2007), mediating many ecosystem functionsand processes whose temporal dynamics depend on environmental fil-ters (Del Toro et al., 2015; Dell et al., 2014). Climatic conditions andanthropogenic disturbances figure among the key drivers of ant speciesoccurrence and their interactions with plants (Gibb et al., 2015;

Paolucci et al., 2016). Therefore, to predict the ecological and evolu-tionary outcomes of ant-plant interactions in a changing world(Parmesan, 2006; Siepielski et al., 2017), we need to understand towhat extent their temporal dynamics depend directly on climate vari-ables and associated disturbances such as fire (Lehmann et al., 2014).

The dynamics of ant-plant interactions might result from oscilla-tions in plant resources availability and ant foraging activity uponthem. Indeed, the effect of climatic seasonality on plant phenology andfood availability for ants is well-known (e.g., Belchior et al., 2016).Besides, it is remarkable that ant foraging activity and ant richness are

https://doi.org/10.1016/j.ecolind.2018.05.001Received 2 October 2017; Received in revised form 24 April 2018; Accepted 2 May 2018

⁎ Corresponding author at: Graduate School in Ecology of Tropical Biomes, Federal University of Ouro Preto, Ouro Preto, Brazil.E-mail address: costa.fvc@gmail.com (F.V. Costa).

Ecological Indicators 93 (2018) 247–255

Available online 21 May 20181470-160X/ © 2018 Elsevier Ltd. All rights reserved.

T

http://www.sciencedirect.com/science/journal/1470160Xhttps://www.elsevier.com/locate/ecolindhttps://doi.org/10.1016/j.ecolind.2018.05.001https://doi.org/10.1016/j.ecolind.2018.05.001mailto:costa.fvc@gmail.comhttps://doi.org/10.1016/j.ecolind.2018.05.001http://crossmark.crossref.org/dialog/?doi=10.1016/j.ecolind.2018.05.001&domain=pdf

positively influenced by temperature, humidity, and rainfall (Gibbet al., 2015; Kaspari, 1993). Temperature is described as the mainabiotic predictor of activity and distribution of thermally constrainedgroups such as ants, in which species display specific tolerances totemperature oscillation (Diamond et al., 2012; Peters et al., 2016).Ecological communities may differ in their responses to temperaturefluctuations according to species thermal niches (Arnan et al., 2015;Kühsel and Blüthgen, 2015). For instance, some communities includespecies with narrow and non-overlapping thermal breadths, i.e., lowcomplementarity in thermal niches. As result, strong temperaturefluctuations may act as environmental filters reducing species diversity.In contrast, other communities encompass species with broader andoverlapping thermal breadths, thus species diversity in these commu-nities are less affected by temperature oscillations (Arnan et al., 2015;Kühsel and Blüthgen, 2015). This divergence is most likely to occur inhabitats that differ in temperature oscillation, e.g., highly seasonalenvironments which might comprise communities with high hetero-geneity in their thermal breadths, in response to the experienced se-lection for distinct optimal conditions (Arnan et al., 2015; Kaspari et al.,2015). Hence, it is expected that marked seasonality, i.e., the temporalperiodicity in climatic conditions (Tonkin et al., 2017), regulates plantresource availability and patterns of ant foraging with extensive effectson the dynamics of ant-plant interactions.

Seasonality as its related disturbances such as fire are most promi-nent in fire-prone ecosystems (Lehmann et al., 2014), wherein thetemporal activity of species (Andersen et al., 2014) and vegetationphenology (Alvarado et al., 2017) are shaped by burning regimes. Infact, the biota in these environments presents adaptations that allowthem to survive fire events, as a result of their association over evolu-tionary time (Bond and Keeley, 2005; Whelan, 1995). Moreover, recentevidence shows that structure and composition of ant communitiesfrom flammable environments such as grasslands and savannas are ableto recover after burning events (reviewed by Vasconcelos, Maravalhasand Cornelissen 2016). Likewise, plants from fire-prone vegetationusually exhibit a rapid recovery after burning (Maurin et al., 2014), andsome species even re-sprout and bloom only in response to fire(Figueira et al., 2016). Alternatively, fire promotes negative impacts onants by simplifying vegetation structure (Kimuyu et al., 2014), de-creasing the availability of nesting sites, and causing direct mortality ofcolonies, features that together negatively impact ant-plant interactions(Fagundes et al., 2015).

Despite the considerable data on the effects of climate and fire onplants (e.g., Veldman et al., 2015) and animals (e.g., Gibb et al., 2015),little is known about how these environmental filters influence animal-plant interactions. To date, no empirical study has addressed to whatextent both drivers may affect the temporal variation of ant-plant in-teractions. Here, we assessed whether the dynamics of ant-plant inter-actions in a seasonal fire-prone ecosystem is determined by climate andwhether this dynamic is altered by an unmanaged fire event. To do so,we studied ant-plant interactions through ecological networks ap-proach, which has been broadly used to evaluate ecosystem structureand functioning (Okuyama and Holland, 2008). We chose the diversityand frequency of interactions once these network metrics represent asimple and meaningful way to describe quantitative patterns and thecomplexity of species-rich interacting communities (Blüthgen, 2010;Vázquez et al., 2005), i.e., how many species are actually in associationand how frequently they truly interact. Thus, we focused on camporupestre, a tropical megadiverse mountaintop grassland ecosystem(Silveira et al., 2016), wherein plant-related rewards comprisingflowers, fruits and extrafloral nectaries are used as food by ants (Costaet al., 2016). This fire-prone environment has been subjected to re-current anthropogenic fires (Alvarado et al., 2017; Figueira et al., 2016)and climatic filters that regulate species distribution (Fernandes et al.,2016) and vegetation phenology (Rocha et al., 2016). Current findingshave shown a short-term negative effect of fire on campo rupestre ve-getation structure (Le Stradic et al., 2016) and ant assemblages (Anjos

et al., 2017; Neves et al., 2016). Likewise, recent studies suggest thatclimatic conditions in this ecosystem change linearly along the eleva-tional gradient (from 800 to 1400m a.s.l.) constraining ant occurrence(Fernandes et al., 2016), an abiotic-limiting trend that is consistent inseveral mountainous ecosystems (Bishop et al., 2016; Peters et al.,2016).

As follows, we have three expectations. First, we expected a positiveand linear effect of temperature, rainfall, and humidity on the diversityand frequency of ant-plant interactions. This expectation is based onexisting evidence that warmer and wetter conditions linearly increaseresource availability and ant activity (see Rocha et al., 2016; Fernandeset al., 2016), when the amplitude of climatic conditions is seasonally-wide, but not extreme as observed for very high mountains (e.g., Dunnet al., 2007). Second, that temperature oscillation over seasons wouldlead to high heterogeneity in ant species thermal responses, as seasonalhabitats might comprise communities with a set of species with highvariability in their optimal conditions (Arnan et al., 2015; Kaspari et al.,2015). Finally, we predict that fire would lead to short-term negativeeffects on interactions’ diversity and frequency, as plant and ant com-munities from flammable ecosystems are supposed to be resilient to firedisturbances (Andersen et al., 2014; Figueira et al., 2016). Here, weadopted the traditional view of resilience as the time required for anecological system to return to a steady-state following a perturbation(Gunderson, 2000), assuming that unburned plots should reflect astable state for ant-plant interactions.

2. Material and methods

2.1. Study area

This study was carried out in seven campo rupestre sites at Morro daPedreira Environmental Protection Area, a buffer zone of Serra do CipóNational Park, in the southern region of the Espinhaço Mountain Range,southeastern Brazil (19°17’49” S, 43°35”28” W, Fig. S1). At higher al-titudes (upper to 900m asl.), the region is featured by campo rupestre, arocky montane savanna composed by a species-rich vegetation, highlevels of plant endemism, a large number of threatened plant species(Silveira et al., 2016), and high ant richness (Costa et al., 2015). Fireregimes in this region are moisture-dependent, which means that itsoccurrence is often related to the dry season length and rainfall dis-tribution along the season. Additionally, there is a strong spa-tial–temporal variation in fire frequency in the region, wherein in thelast 30 y, 51% of all affected areas got burned between one and fourtimes, 22% between five and nine times, and 2% were burned ten timesor more (Alvarado et al., 2017). Recent data indicate an interval of twoyears between two fire events, reflecting the occurrence of large fires inshorter time intervals (Figueira et al., 2016). On the landscape level, thewidespread rocky outcrops, associated with rivers and riparian forests,create gaps in the fuel layer that prevent fire spread (Figueira et al.,2016). Generally, most fires are mainly anthropogenic and superficial,consuming fine fuels of herbaceous layer (Figueira et al., 2016). Theclimatic regime is characterized as tropical altitudinal (Cwb) accordingto Köppen’s classification (Alvares et al., 2013), comprising markedlydry and cold winters and warm and wet summers. Mean Annual tem-perature is ca. 22 °C, wherein daily minimum and maximum values are33 °C and 28 °C for the warmest month (February), and 13 °C and 7 °Cfor the coldest month (July). Mean annual rainfall is ca. 1,500mm,ranging from 75 to 340mm during rainy season (October to April, >60mm per month), while throughout dry season it ranges from 7 to32mm (May to September,< 40mm per month) (Alvarado et al.,2017).

2.2. Sampling design

In each site, we selected the larger rocky outcrop that also wascloser to the weather station to install the sampling transect. On the

F.V. Costa et al. Ecological Indicators 93 (2018) 247–255

248

rocky outcrops happen the higher vegetation structure wherein organicmatter accumulates, while surrounding them, a matrix comprisedmostly by grasslands occurs (i.e. forbs and graminoids) (Silveira et al.,2016). Additionally, the seven campo rupestre sites are similar in termsof altitude (from 1100 to 1200m a.s.l.), climatic conditions and vege-tation structure. The minimum distance between sites is 1.44 km andthe maximum 10 km (see Fig. S1). At each site, we delimited onetransect of 200m in length and one meter in width, which was dividedinto 20 plots (10×1m). Then, in order to randomly sample a broaderand spatially independently area, we drew five plots in each transect, ina way they were at least 30m away apart from one another, totaling 35plots. In each plot, we recorded ant-plant interactions on all herbs,rosettes, subshrubs, shrubs, and trees that were fully accessible to us,those with 50–200 cm in height. We monitored the marked plantsquarterly, at the peak and at the end of the rainy and dry seasons (re-spectively, January, May, July and October), from January 2014 toOctober 2015, plus one sampling during the peak for the rainy season(February) of 2016, totaling nine sampling periods.

2.3. Climatic variables

Close to each transect there is a climatic station from the local LongTerm Ecological Research Project (PELD in Brazilian acronym), which isequipped with a precise data logger (Onset HOBO® U30). The sensorsare settled at 2m above ground and record several climatic variables,uninterruptedly, in every five minutes. Here, we choose three variablesthat very often are pointed as good predictors of ant activity and di-versity: air temperature (°C), air humidity (%) and rainfall (mm)(Kaspari et al., 2015; Kaspari and Valone, 2002). To assure a refineddata, we used mean values of temperature and humidity recorded at thetime when each plot was monitored (i.e., records from the beginning tothe ending time of each plot). For rainfall measurements, we used theaccumulated of the sampling month, since we avoided to sample inrainy periods. These three abiotic variables were included in-dependently in our climatic models, as they are weakly correlated (seeTable S1).

2.4. Fire disturbance

According to records provided by Morro da Pedreira EnvironmentalProtection Area’s staff, the study sites have not been burned since 2009.After approximately one year monitoring the study sites, in September2014, fire outbreaks started in this region and persisted for two months,affecting a wide geographical area (∼15.000 ha) (ICMBio, 2014). Thisextensive burning event was unmanaged, most likely from anthro-pogenic origin (see more details in Figueira et al., 2016) and superficial(i.e., especially over fine fuel biomass with small thickness). At the endof burning events, 26 of the 35 plots were burned and nine plots re-mained unburned. For most of the plots, we could conclude one year ofsampling before the fire has started (see Table S2). Thus, we continuedto monitor the studied sites for one more year post-fire in order to assesswhether this disturbance would affect the temporal dynamics of inter-actions.

2.5. Tracking of ant-plant interactions

We observed each plant for ca. 3 min between 08:00–12:00 and14:00–17:00, avoiding to sample in periods when ant activity is verylimited, i.e., rainy periods and very hot hours (adapted from Belchioret al., 2016; Blüthgen et al., 2004). To obtain representative activities ofants on plants, we recorded every type of interaction taking place ineach sampled plant: ants feeding upon extrafloral nectary (EFN inter-action), floral nectar or pollen (flower), glands and fruit secretions(fruits), and honeydew droplets from trophobiont hemipterans asso-ciated with sampled plants (honeydew). We computed these interac-tions involving plant-derived resources when ants were seen feeding

quietly upon them with mouthparts in contact to the source, for periodsup to several minutes (Rico-Gray, 1993). When we observed an ant onan individual plant that did not provide any food source, when an antleft a plant without making contact with resources of any type, or whenan ant was using a plant as nesting site we defined the interaction as“visit” (see Costa et al., 2016 for a similar approach). We also recordedthe number of worker ants in each plant to estimate the recruitmentrate of each species, which is related to foraging activity and numericaldominance.

We collected vouchers of plants and ants for taxonomic identifica-tion. To identify the ants we used the key by Baccaro et al. (2015) andconsulted a specialist from Federal University of Paraná (UFPR). Wedeposited ant vouchers in the entomological collection Padre JesusSantiago Moure at UFPR. We identified the plants with the support ofmany botanists from Federal University of Minas Gerais (UFMG) anddeposited vouchers in the herbarium of the Botanical Department(UFMG).

2.6. Data analysis

2.6.1. Climatic effectsTo test whether the temporal dynamics of ant-plant interactions is

affected by climatic conditions we used generalized linear mixed effectsmodels (GLMMs, lmer function for data with normal distribution andglmer for non-normal data, both with lme4 package in R). We fitted fixedand random effects that account for our repeated measures design (i.e.,longitudinal data, sensu Crawley, 2013), wherein each plot was sampledquarterly for two years. We include plot identity nested within sites as arandom effect on the intercept for all models (1|sites/plots) to copewith our repeated temporal sampling (nine measures per plot) (Bateset al., 2014). As fixed effects, we used the climatic variables describedabove, as they are weakly correlated (see Table S1): dependent vari-able∼mean temperature+mean humidity+ accumulated rainfall.All abiotic variables were scaled in order to standardize for differencesin their units and amplitudes.

As dependent variables, we used two network metrics: interactions’frequency per plot and interactions’ Shannon diversity per plot (H2 –see Bersier et al., 2002; Blüthgen et al., 2008). In our local networks(i.e., plot level), consistent with other studies on ant-plant interactions(e.g., Ivens et al., 2016), each interaction frequency is computed basedon the occurrence of an ant species in an individual plant, not thenumber of workers recruited per plant. Therefore, total interactions’frequency reflects the number of independent plant individuals onwhich an ant species has been recorded in association, e.g. if a plot hasa frequency equal to 2, it means that ant A and plant B were observed inassociation in two independently plants occurring in the plot. Hence,we included all records from all interaction types that occurred betweenants and plants within each plot. From those, we built weighted ma-trices with plant species as rows and ant species as columns and filledcells with the number of events observed between one plant species iand one ant species j. Each matrix was standardized and used to com-pute the frequency and diversity of interactions. In total, we had 315matrices/networks that correspond to each plot (n= 5 per site), withineach site (n=7 sites), in each sampling period (n=9 for each plot).Recruitment data was used in additional analyses (Table S7 and S9), inorder to unveil the role of basal biological mechanisms (i.e., antsforaging activity) on the response of whole ant-plant community (i.e.,network level).

To test whether temperature oscillation predicts the heterogeneityin ants’ thermal responses, we assessed the thermal niche of each antspecies. We computed a thermal niche model that is based on occur-rence-weighted temperature conditions of each species’ activity (seeKühsel and Blüthgen 2015), i.e., the occurrence is computed when aspecies is recorded in a distinct plot-temperature. This weighted ap-proach considers the relative temperature preferences (rates) as well asthe reliability (number of observations per temperature) to characterize

F.V. Costa et al. Ecological Indicators 93 (2018) 247–255

249

a species’ niche. Thus, we used the precise temperature during theperiod in which each plot was monitored and defined the occurrence-weighted mean temperature for each ant species across all plots as itsthermal optimum. As a proxy for a species’ thermal niche breadth, wecalculated the occurrence-weighted standard deviation of mean tem-perature. To test if species activity responds significantly to tempera-ture, we performed a null model of an expected thermal niche for eachspecies if it can occur on every plot with the same likelihood in allsurveys irrespective of the temperature. The comparison between theobserved thermal optimum and the optimum obtained from null modelsthus shows whether a species’ activity deviates significantly from theaverage sampling condition. We performed 1000 randomizations tocalculate how often the expected thermal niche is higher or smallerthan the observed temperature for each species (α=5%) (see Chistéet al., 2016 for a related null model approach along land-use gradientsinstead of temperature).

It is worth pointing out that thermal niches as defined by activitydata in the field have the advantage that they mirror realistic beha-vioral responses, not only physiological constraints such as laboratorymeasurements. On the other hand, several environmental factors andbiotic interactions may influence species’ activities, and while the nullmodel controls for the microclimatic conditions during the surveys andfor the overall abundance of each ant species, such context-dependencycannot be disentangled from their thermal niches. Some studies haveshown that laboratory measurements across ant species do correspondto activities in the field (Arnan and Blüthgen, 2015) or to microhabitatpreferences (e.g. Kaspari et al., 2015), but such independent climaticmeasurements across the whole ant community were beyond the scopeof our study.

2.6.2. Fire effectsTo test if the fire has affected the temporal dynamics of interactions

we fitted mixed models where fire (two levels variable with burned andunburned plots) and the interaction with sampling period (nine levelsvariable corresponding to sampling periods) were fixed effects: de-pendent variable ∼fire× sampling period. Sites and plots weregrouped as random effects following a structure were the intercept varyamong sites and the plots are nested within sites (1|sites/plots) (Bateset al., 2014). We built one model for network interactions’ frequencyand one model for network interactions’ diversity (H2).

The residuals of all GLMMs models were evaluated, as well as the

suitability of error distribution chosen. The complete models weresimplified until minimum suitable models by backward selection basedon P-value. All statistical analyses were performed in R (R DevelopmentTeam, 2017).

3. Results

Over two years, we monitored a total of 1,114 individual plants,representing 107 species. From these, a total of 1,905 interaction eventswere recorded in 873 plants from 99 species and 33 families (Table S3),which provided food or were used as nesting and foraging substrate by43 ant species from 16 genera and 6 subfamilies (3858 workers) (TableS4). Ants foraged on plants to consume nectar from extrafloral nectaries(EFNs) (23% of the records), nectar or pollen from flowers (7%), hon-eydew from trophobiont hemipterans (6%), and pulp or fruit secretions(2%). The remaining 62% represents visit. The most representativeplant families that provided ants with resources were Malpighiaceae(e.g., EFNs in Tetrapterys microphylla with 7% of all records), Fabaceae(e.g., EFNs in Chamaecrista papillata with 6%), Velloziaceae (e.g.,shelter in Barbacenia flava with 6% and Vellozia nivea with 5%), Myr-sinaceae (e.g., EFNs in Myrsine monticola with 5%), and Asteraceae(e.g., secretory structures in Symphyopappus reticulatus with 5%). Themost common ant subfamily was Formicinae (50% of records) andMyrmicinae (39%), and the most common species were Cephalotes pu-sillus (35% of records), Camponotus crassus (12%), Camponotus rufipes(10%), Brachymyrmex cordemoyi (9%), and Camponotus trapeziceps(6%). C. pusillus, C. rufipes, and C. crassus were the only species thatoccurred in all sampling periods and had the highest interaction fre-quencies in all resources.

3.1. Response of ant-plant interactions to climate

All climatic variables presented marked seasonality (Fig. 1). Inter-actions diversity ranged from 0.69 to 2.83 (1.27 ± 0.1259, mean ±SD) and increased with temperature (22 ± 3.55 °C) and humidity(67 ± 12.6%), whereas interactions frequency ranged from 2.0 to 27.0(5.95 ± 0.81), increasing with temperature and rainfall(40 ± 30.33mm) (Fig. 2, Table S5).

Across seasons, ant species thermal optimum ranged from 14 °C(Nesomyrmex sp1) to 28 °C (Cephalotes depressus). Despite this greatvariation, the overall thermal responses of most species correspond to

Fig. 1. Monthly values of rainfall, temperature, and humidity in the seven campo rupestre sites monitored for two years.

F.V. Costa et al. Ecological Indicators 93 (2018) 247–255

250

the mean thermal optimum and mean niche breadth of the wholecommunity (i.e., 22.0 °C ± 2.3 °C; mean ± SD; dashed lines in Fig. 3),indicating a high overlap in thermal responses. Only nine species (21%of species) had a thermal optimum and niche breadth significantlydeviating from the null model (Fig. 3). Amongst these, three speciesshowed a significant preference for warmer conditions (right side ofvertical dashed line, Fig. 3), and two species showed preferences forcolder conditions (left side of vertical dashed line, Fig. 3), i.e., respec-tively, being more active in higher and lower temperatures than theaverage sampled all over the seasons.

3.2. Fire effects on the temporal dynamics of interaction networks

Fire impacts on the temporal dynamics of ant-plant interactionswere evident only one month post-fire (Oct 2014; Table S6), a periodduring which burned plots had lower interactions diversity (five timesless) and total interactions frequency (seven times less) than unburnedplots (Fig. 4). Thus, fire effects on species interactions were highlytransient, since differences between burned and unburned plots were nolonger observed four months post-fire.

4. Discussion

We showed for the first time the direct influence of climate and firedisturbance on the temporal dynamics of ant-plant interactions in amountaintop fire-prone seasonal vegetation. As expected, ant-plant

interactions were highly affected by climatic seasonality and positivelyresponded to increases in temperature, humidity, and rainfall. Foliage-dwelling ant species, though, exhibited low heterogeneity in theirthermal niches under temperature oscillation over seasons, which evi-dence a high tolerance of ants to temperature fluctuation. Moreover, wefound that an immediate consequence of fire was the significant re-duction in the frequency and diversity of ant-plant interactions, mostlikely due to decreasing in plant-derived resources availability, as wellas the reduction in ant foraging activity upon them. Nevertheless, thevalues of diversity and frequency of ant-plant interactions in burnt plotsreached similar values of unburnt plots after less than half a year post-disturbance, highlighting the resilience of ant-plant interaction net-works to fire.

We observed that diversity and frequency of interactions are sea-sonally regulated by increases in temperature, humidity, and rainfall, ina sense that interactions peaks match rainy seasons peaks (Fig. 4). Itseems that at local scale, warmer and wetter conditions positively in-fluence ants activity (e.g., Kaspari and Valone, 2002) and ant-plantinteractions (e.g., Belchior et al., 2016), whereas at global scales theireffects diverge from positive (e.g., Dunn et al., 2009) to negative (e.g.,Leal and Peixoto, 2016). This conflict may reflect how climatic condi-tions are assessed (e.g., measurements at sampling moment vs. in-ferences) and analyzed (e.g., correlations vs. direct effects). Here, webring a fine-scale picture that represents a direct effect of climate oninteractions dynamics. The available literature has shown that thetemporal variation of ants feeding upon EFNs (Belchior et al., 2016) and

Fig. 2. Relationship between climate and ant-plant interactions monitored for two years in campo rupestre. Numbers in x-axis are minimum, medium and maximumvalues of climatic variables. Points in temperature and humidity illustrations are mean values for each site in each sampling period. Points in rainfall illustration aremean values for each sampling period, considering all sites together. Vertical and horizontal gray lines are the standard errors of dependent and predictor variables,correspondingly. Curves were fitted with parameters from GLMM models.

F.V. Costa et al. Ecological Indicators 93 (2018) 247–255

251

the structure of ant-EFNs networks (Lange et al., 2013) are determinedby nectar availability, which depends on seasonality. In fact, we ob-served that ant recruitment on plants is highly correlated to the abun-dance of plants providing resources (see Table S7), being the latter alsoreliant on climatic influence (see Table S8). Moreover, the frequency ofant interactions with plants bearing EFNs, which are the main plant-derived source in campo rupestre (see Costa et al., 2016), is higher at theonset and peak of rainy seasons, when plants are flushing new leavesand the nectar availability is greater (Alves-Silva and Del-Claro, 2014).Thus, it appears that plant resources availability and ants feeding uponthem, trigger interactions response to climate.

The studied ant community exhibited broad and high overlappedthermal responses, i.e., low thermal complementarity. Ants from sea-sonal environments are supposed to have high complementarity inthermal responses, because temperature variability provides availablethermal niches for species with different thermal optima (Arnan et al.,2015). Contrary to our prediction, our results indicate that seasonaldaily-temperature fluctuations in campo rupestre (up to 20 °C degrees)do not constrain ant occurrence over seasons. The high overlap ofthermal niches, linked with the evidence that ant recruitment on plantsstrengthens the positive response of interactions to climate (Table S9),indicates that temperature here is triggering ant foraging activity,

rather than limiting species occurrence. Low thermal complementaritycan be assumed as less tolerance to climate changes, but only if speciesexhibit narrow niches (Arnan et al., 2015; Kühsel and Blüthgen, 2015).Here, the low heterogeneity of ant thermal responses is likely due to thebroad niches that overlay, suggesting a high tolerance of ant commu-nity to temperature seasonality. This thermal tolerance of campo ru-pestre ants most likely increases the capability of ant-plant interactionsto recover from climate fluctuations.

The effects of fire on ant-plant interactions were negative buttransitory, in line with our predictions. Interactions diversity and fre-quency were lower in burned plots, but they quickly recovered up tofour months after fire. The earliest impacts of burning in fire-proneecosystems may include simplification of vegetation structure(Maravalhas and Vasconcelos, 2014) and limitation of cavities to beused as nest (Fagundes et al., 2015). However, many plants in camporupestre are resistant to fire, as vegetation structure promptly recoversafter burning (Figueira et al., 2016; Le Stradic et al., 2016). In fact, wefound that the abundance of plants providing resource for ants is highlycorrelated with ant-plant interactions dynamics (Table S7); and fire didnegatively affect plant abundance in a similar pattern as observed forinteraction networks (Table S8). Therefore, fire is clearly affecting re-source availability, which prompts ants’ activity and consequently

Fig. 3. Thermal characterization of ant community showing the mean temperature of plots where each species occurred (weighted by occurrences in distincttemperatures) with weighted standard deviations (corresponding to niche breadth). Asterisks show that an ant species’ occurrence significantly differs from the nullmodel. Species on the right side are more heat-tolerant, while species in the left side prefer cooler temperatures than the average sampled during the study (dashedvertical line). The information inside parentheses corresponds to the numbers of plots in which each species occurred (from 315 possibilities in total).

F.V. Costa et al. Ecological Indicators 93 (2018) 247–255

252

promotes changes in interactions dynamics.Short-term impacts of fire are well documented for ground-dwelling

ants from campo rupestre (Anjos et al., 2017; Neves et al., 2016), Cer-rado (Maravalhas and Vasconcelos, 2014), African savanna (Parr et al.,2004) and Australian savanna (Andersen et al., 2006). The main directeffect of a single fire event on ants is the increasing of colony mortality,mainly for species that nest in twigs and small branches (Kimuyu et al.,2014). Actually, campo rupestre vegetation is composed mostly of herbsand small shrubs (Giulietti and Pirani, 1997) that may not supportshelter structure for most ant species. As a result, we believe that themajority of ants that forage on vegetation nests in natural cavities in the

ground and cavities of fire-adapted plants (e.g. the huge part of ob-served nests were on the rosettes of Velloziaceae species), which actoffering protection against superficial burning. Additionally, structuraland demographic responses of flora may vary extremely from few daysto more than three years, depending on plant life-form and adaptivestrategies (Figueira et al., 2016). Together, these evidence support ourfindings that ant-plant interactions in fire-prone ecosystems are re-silient to fire, as previously documented for ant (Andersen et al., 2014;Parr et al., 2004) and plant (Figueira et al., 2016; Le Stradic et al.,2016) communities.

To our knowledge, this is the first empirical study that provide

Fig. 4. Fire effects on the temporal dynamics of interactions diversity (A) and interactions frequency (B), both represented by the mean (points) and standard error(vertical limits) measured per plots in each sampling period. Asterisks indicate periods when burned and unburned plots significantly differed from one another.

F.V. Costa et al. Ecological Indicators 93 (2018) 247–255

253

evidence of how climatic conditions drive the temporal dynamics ofant-plant interaction networks and to what extent an unmanaged firealters this dynamic. In summary, fire has altered the seasonal dynamicsof interactions, leading to negative short-term impacts on species in-teractions, which revealed to be resilient to this disturbance. Moreover,our results point out that climate seasonality predicts the temporalvariation of ecological interactions mainly due to changes in resourceavailability and consumer activity upon them. Despite the high oscil-lation of heat over seasons, ants exhibited high tolerance (i.e., broadthermal niches) to temperature fluctuation, which supported the re-liability of interactions over seasons. These findings open a newpathway to explore the thermal responses of species and ecologicalinteractions across broader gradients of environmental conditions andecosystem disturbances. Long-term experimental studies comprisingassisted burnings are desirable to forecast the consequences of the in-creases in human-induced fires and how its synergy with climate wouldaffect ecosystem functioning in fire-prone ecosystems. Finally, thisstudy adds evidence that studied networks can be useful to monitorenvironmental changes, as the impacts of anthropogenic or naturaldisturbances, being representative of many even more complex speciesinteractions in a planet where these events are increasingly frequent.

Authors’ contributions

FVC, TJG and FSN conceived the ideas and designed methodology;FVC and LDB collected the data; FVC, NB and ABV analyzed the data;FVC, NB, FSN and TJG wrote the manuscript. All authors contributed tothe drafts and gave final approval for publication.

Acknowledgments

We are grateful to many colleagues who helped us in many ways.We thank Marco A. R. Mello and Affonso H. N. Souza for their com-ments in an earlier version of this manuscript and the latter for the mapproduction. We thank Affonso H. N. Souza, Alice C. Leite, Fábio T.Pacelle and Humberto Brandt for field support; Rodrigo Feitosa for antspecies identification; Fernando A. O. Silveira and João R. Stehmann forplant species identification; and Melanie Chisté for the help withthermal niche analysis. We also are grateful to two anonymous re-viewers for their helpful comments, Chico Mendes Institute forBiodiversity (ICMBio), Serra do Cipó National Park, and VelloziaReserve for their logistic support. FV Costa received a PhD scholarshipfrom the German Academic Exchange Service (DAAD) and BrazilianCoordination for the Improvement of Higher Education Personnel(CAPES). Our study was funded by the Conselho Nacional deDesenvolvimento Científico e Tecnológico (CNPq), Fundação de Amparo àPesquisa do Estado de Minas Gerais (FAPEMIG), PELD-Campos Rupestresda Serra do Cipó, and Alexander von Humboldt Foundation.

Appendix A. Supplementary data

Supplementary data associated with this article can be found, in theonline version, at http://dx.doi.org/10.1016/j.ecolind.2018.05.001.

References

Alvarado, S.T., Fornazari, T., Cóstola, A., Morellato, L.P.C., Silva, T.S.F., 2017. Drivers offire occurrence in a mountainous Brazilian cerrado savanna: tracking long-term fireregimes using remote sensing. Ecol. Indic. 78, 270–281. http://dx.doi.org/10.1016/j.ecolind.2017.02.037.

Alvares, C.A., Stape, J.L.J.L., Sentelhas, P.C., De Moraes Gonçalves, J.L., Sparovek, G., deMoraes Gonçalves, J.L., Sparovek, G., 2013. Köppen’s climate classification map forBrazil. Meteorol. Zeitschrift 22, 711–728. http://dx.doi.org/10.1127/0941-2948/2013/0507.

Alves-Silva, E., Del-Claro, K., 2014. Fire triggers the activity of extrafloral nectaries, butants fail to protect the plant against herbivores in a neotropical savanna. Arthropod.Plant. Interact. 8, 233–240. http://dx.doi.org/10.1007/s11829-014-9301-8.

Andersen, A.N., Hertog, T., Woinarski, J.C.Z., 2006. Long-term fire exclusion and ant

community structure in an Australian tropical savanna: congruence with vegetationsuccession. J. Biogeogr. 33, 823–832. http://dx.doi.org/10.1111/j.1365-2699.2006.01463.x.

Andersen, A.N., Ribbons, R.R., Pettit, M., Parr, C.L., 2014. Burning for biodiversity: highlyresilient ant communities respond only to strongly contrasting fire regimes inAustralia’s seasonal tropics. J. Appl. Ecol. 51, 1406–1413. http://dx.doi.org/10.1111/1365-2664.12307.

Anjos, D., Campos, R., Campos, R., Ribeiro, S.P., 2017. Monitoring effect of fire on antassemblages in Brazilian rupestrian grasslands: contrasting effects on ground andarboreal fauna. Insects 8, 64. http://dx.doi.org/10.3390/insects8030064.

Arnan, X., Blüthgen, N., 2015. Using ecophysiological traits to predict climatic and ac-tivity niches: lethal temperature and water loss in Mediterranean ants. Glob. Ecol.Biogeogr. 24, 1454–1464. http://dx.doi.org/10.1111/geb.12363.

Arnan, X., Blüthgen, N., Molowny-Horas, R., Retana, J., 2015. Thermal characterizationof European ant communities along thermal gradients and its implications for com-munity resilience to temperature variability. Front. Ecol. Evol. 3. http://dx.doi.org/10.3389/fevo.2015.00138.

Baccaro, F.B., Feitosa, R.M., Fernandez, F., Fernandes, I.O., Izzo, T.J., Souza, J.L.P., Solar,R., 2015. Guia para os gêneros de formigas do Brasil. Editora INPA, Manaus. http://dx.doi.org/10.5281/zenodo.32912.

Bates, D., Mächler, M., Bolker, B., Walker, S., 2014. Fitting linear mixed-effects modelsusing lme4. Drug Inf. J. 35, 1215–1225.

Belchior, C., Sendoya, S.F., Del-Claro, K., 2016. Temporal variation in the abundance andrichness of foliage-dwelling ants mediated by extrafloral nectar. PLoS One 11,e0158283. http://dx.doi.org/10.1371/journal.pone.0158283.

Bersier, L.F., Banašek-Richter, C., Cattin, M., 2002. Quantitative descriptors of food-webmatrices. Ecology 83, 2394–2407.

Bishop, T.R., Robertson, M.P., Van Rensburg, B.J., Parr, C.L., 2016. Coping with the cold:minimum temperatures and thermal tolerances dominate the ecology of mountainants. Ecol. Entomol. 42, 105–114 10.1111/een.12364.

Blüthgen, N., 2010. Why network analysis is often disconnected from community ecology:a critique and an ecologist’s guide. Basic Appl. Ecol. 11, 185–195. http://dx.doi.org/10.1016/j.baae.2010.01.001.

Blüthgen, N., Fründ, J., Vazquez, D.P., Menzel, F., Vázquez, D.P., Menzel, F., Vazquez,D.P., Menzel, F., 2008. What do interaction network metrics tell us about speciali-zation and biological traits? Ecology 89, 3387–3399.

Blüthgen, N., Stork, N.E., Fiedler, K., Blu, N., Stork, N.E., Fiedler, K., Blüthgen, N., Stork,N.E., Fiedler, K., 2004. Bottom-up control and co-occurrence in complex commu-nities: honeydew and nectar determine a rainforest ant mosaic. Oikos 106, 344–358.

Bond, W.J., Keeley, J.E., 2005. Fire as a global “herbivore”: the ecology and evolution offlammable ecosystems. Trends Ecol. Evol. 20, 387–394. http://dx.doi.org/10.1016/j.tree.2005.04.025.

Bronstein, J.J.L.J., 1998. The contribution of ant-plant protection studies to our under-standing of mutualism. Biotropica 30, 150–161.

Chisté, M., Mody, K., Gossner, M., Simons, M., Köhler, G., Weisser, W., Blüthgen, N.,2016. Losers, winners and opportunists: how grassland land-use intensity affectsorthopteran communities. Ecosphere 7, e01545. http://dx.doi.org/10.13140/RG.2.1.3702.8566.

Costa, F.V., Mello, M.A.R., Bronstein, J.L., Guerra, T.J., Muylaert, R.L., Leite, A.C., Neves,F.S., 2016. Few ant species play a central role linking different plant resources in anetwork in rupestrian grasslands. PLoS One 11, e0167161. http://dx.doi.org/10.1371/journal.pone.0167161.

Costa, F.V., Mello, R., Lana, T.C.T.C., Neves, F.S.F.D.S., 2015. Ant fauna in megadiversemountains: a checklist for the rocky grasslands. Sociobiology 62, 228–245. http://dx.doi.org/10.13102/sociobiology.v62i2.228-245.

Crawley, M.J., 2013. In: The R Book, Second ed. Wiley, United Kingdom. http://dx.doi.org/10.1007/s13398-014-0173-7.2.

Del Toro, I., Ribbons, R.R., Ellison, A.M., 2015. Ant-mediated ecosystem functions on awarmer planet: effects on soil movement, decomposition and nutrient cycling. J.Anim. Ecol. 84, 1233–1241. http://dx.doi.org/10.1111/1365-2656.12367.

Dell, A.I., Pawar, S., Savage, V.M., 2014. Temperature dependence of trophic interactionsare driven by asymmetry of species responses and foraging strategy. J. Anim. Ecol.83, 70–84. http://dx.doi.org/10.1111/1365-2656.12081.

Diamond, S.E., Sorger, D.M., Hulcr, J., Pelini, S.L., Toro, I. Del, Hirsch, C., Oberg, E.,Dunn, R.R., 2012. Who likes it hot? a global analysis of the climatic, ecological, andevolutionary determinants of warming tolerance in ants. Glob. Chang. Biol. 18,448–456. http://dx.doi.org/10.1111/j.1365-2486.2011.02542.x.

Dunn, R.R., Parker, C.R., Sanders, N.J., 2007. Temporal patterns of diversity: assessingthe biotic and abiotic controls on ant assemblages. Biol. J. Linnean Soc. 91, 191–201.http://dx.doi.org/10.1111/j.1095-8312.2007.00783.x.

Dunn, R.R., Agosti, D., Andersen, A.N., Arnan, X., Bruhl, C.A., Cerdá, X., Ellison, A.M.,Fisher, B.L., Fitzpatrick, M.C., Gibb, H., Gotelli, N.J., Gove, A.D., Guenard, B., Janda,M., Kaspari, M., Laurent, E.J., Lessard, J.-P., Longino, J.T., Majer, J.D., Menke, S.B.,McGlynn, T.P., Parr, C.L., Philpott, S.M., Pfeiffer, M., Retana, J., Suarez, A.V.,Vasconcelos, H.L., Weiser, M.D., Sanders, N.J., 2009. Climatic drivers of hemisphericasymmetry in global patterns of ant species richness. Ecol. Lett. 12, 324–333. http://dx.doi.org/10.1111/j.1461-0248.2009.01291.x.

Fagundes, R., Anjos, D.V., Carvalho, R., Del-Claro, K., 2015. Availability of food andnesting-sites as regulatory mechanisms for the recovery of ant diversity after firedisturbance. Sociobiology 62, 1–9. http://dx.doi.org/10.13102/sociobiology.v62i1.1-9.

Fernandes, G.W., Almeida, H.A., Nunes, C.A., Xavier, J.H.A., Cobb, N.S., Carneiro,M.A.A., Cornelissen, T., Neves, F.S., Ribeiro, S.P., Nunes, Y.R.F., Pires, A.C.V., Beirão,M.V., 2016. Cerrado to Rupestrian Grasslands: Patterns of Species Distribution andthe Forces Shaping Them Along an Altitudinal Gradient. In: Fernandes, G.W. (Ed.),Ecology and Conservation of Mountaintop Grasslands in Brazil. Springer

F.V. Costa et al. Ecological Indicators 93 (2018) 247–255

254

http://dx.doi.org/10.1016/j.ecolind.2018.05.001http://dx.doi.org/10.1016/j.ecolind.2017.02.037http://dx.doi.org/10.1016/j.ecolind.2017.02.037http://dx.doi.org/10.1127/0941-2948/2013/0507http://dx.doi.org/10.1127/0941-2948/2013/0507http://dx.doi.org/10.1007/s11829-014-9301-8http://dx.doi.org/10.1111/j.1365-2699.2006.01463.xhttp://dx.doi.org/10.1111/j.1365-2699.2006.01463.xhttp://dx.doi.org/10.1111/1365-2664.12307http://dx.doi.org/10.1111/1365-2664.12307http://dx.doi.org/10.3390/insects8030064http://dx.doi.org/10.1111/geb.12363http://dx.doi.org/10.3389/fevo.2015.00138http://dx.doi.org/10.3389/fevo.2015.00138http://dx.doi.org/10.5281/zenodo.32912http://dx.doi.org/10.5281/zenodo.32912http://refhub.elsevier.com/S1470-160X(18)30336-4/h0050http://refhub.elsevier.com/S1470-160X(18)30336-4/h0050http://dx.doi.org/10.1371/journal.pone.0158283http://refhub.elsevier.com/S1470-160X(18)30336-4/h0060http://refhub.elsevier.com/S1470-160X(18)30336-4/h0060http://refhub.elsevier.com/S1470-160X(18)30336-4/h0065http://refhub.elsevier.com/S1470-160X(18)30336-4/h0065http://refhub.elsevier.com/S1470-160X(18)30336-4/h0065http://dx.doi.org/10.1016/j.baae.2010.01.001http://dx.doi.org/10.1016/j.baae.2010.01.001http://refhub.elsevier.com/S1470-160X(18)30336-4/h0075http://refhub.elsevier.com/S1470-160X(18)30336-4/h0075http://refhub.elsevier.com/S1470-160X(18)30336-4/h0075http://refhub.elsevier.com/S1470-160X(18)30336-4/h0080http://refhub.elsevier.com/S1470-160X(18)30336-4/h0080http://refhub.elsevier.com/S1470-160X(18)30336-4/h0080http://dx.doi.org/10.1016/j.tree.2005.04.025http://dx.doi.org/10.1016/j.tree.2005.04.025http://refhub.elsevier.com/S1470-160X(18)30336-4/h0090http://refhub.elsevier.com/S1470-160X(18)30336-4/h0090http://dx.doi.org/10.13140/RG.2.1.3702.8566http://dx.doi.org/10.13140/RG.2.1.3702.8566http://dx.doi.org/10.1371/journal.pone.0167161http://dx.doi.org/10.1371/journal.pone.0167161http://dx.doi.org/10.13102/sociobiology.v62i2.228-245http://dx.doi.org/10.13102/sociobiology.v62i2.228-245http://dx.doi.org/10.1007/s13398-014-0173-7.2http://dx.doi.org/10.1007/s13398-014-0173-7.2http://dx.doi.org/10.1111/1365-2656.12367http://dx.doi.org/10.1111/1365-2656.12081http://dx.doi.org/10.1111/j.1365-2486.2011.02542.xhttp://dx.doi.org/10.1111/j.1095-8312.2007.00783.xhttp://dx.doi.org/10.1111/j.1461-0248.2009.01291.xhttp://dx.doi.org/10.1111/j.1461-0248.2009.01291.xhttp://dx.doi.org/10.13102/sociobiology.v62i1.1-9http://dx.doi.org/10.13102/sociobiology.v62i1.1-9

International Publishing, Stanford, USA, pp. 345–377. http://dx.doi.org/10.1007/978-3-319-29808-5_15.

Figueira, J.E.C., Ribeiro, K.T., Ribeiro, M.C., Jacobi, C.M., França, H., Neves, A.C. de O.,Conceição, A.A., Mourão, F.A., Souza, J.M., Miranda, C.A. de K., 2016. Fire inRupestrian Grasslands: Plant Response and Management. In: Fernandes, G.W. (Ed.),Ecology and Conservation of Mountaintop Grasslands in Brazil. SpringerInternational Publishing, Stanford, USA, pp. 415–448.

Gibb, H., Sanders, N.J., Dunn, R.R., Watson, S., Photakis, M., Abril, S., Andersen, A.N.,Angulo, E., Armbrecht, I., Arnan, X., Baccaro, F.B., Bishop, T.R., Boulay, R.,Castracani, C., Del Toro, I., Delsinne, T., Diaz, M., Donoso, D.A., Enriquez, M.L.,Fayle, T.M., Feener, D.H., Fitzpatrick, M.C., Gomez, C., Grasso, D.A., Groc, S.,Heterick, B., Hoffmann, B.D., Lach, L., Lattke, J., Leponce, M., Lessard, J.-P., Longino,J., Lucky, A., Majer, J., Menke, S.B., Mezger, D., Mori, A., Munyai, T.C., Paknia, O.,Pearce-Duvet, J., Pfeiffer, M., Philpott, S.M., de Souza, J.L.P., Tista, M., Vasconcelos,H.L., Vonshak, M., Parr, C.L., 2015. Climate mediates the effects of disturbance on antassemblage structure. Proc. R. Soc. B Biol. Sci. 282http://dx.doi.org/10.1098/rspb.2015.0418. 20150418 20150418.

Giulietti, A., Pirani, J., 1997. Espinhaço Range Region, Eastern Brazil. In: Davis, S.,Heywood, V., Herrera-Macbryde, O., Villa-lobos, J., Hamilton, A. (Eds.), Centres ofPlant Diversity: A Guide and Strategy for Their Conservation, pp. 397–404.

Gunderson, L.H., 2000. Ecological resilience–in theory and application. Annu. Rev. Ecol.Syst. Annu. Rev. Ecol. Syst 31, 425–439.

ICMBio, 2014. Incêndio destrói quase 15 mil hectares na Serra do Cipó, in: BoletimInformativo – APA Morro Da Pedreira E Parque Nacional Serra Do Cipó. p. 15.

Ivens, A.B.F., von Beeren, C., Blüthgen, N., Kronauer, D.J.C., 2016. Studying the complexcommunities of ants and their symbionts using ecological network analysis. Annu.Rev. Entomol. 61, 353–371. http://dx.doi.org/10.1146/annurev-ento-010715-023719.

Kaspari, M., 1993. Body size and microclimate use in Neotropical granivorous ants.Oecologia 96, 500–507. http://dx.doi.org/10.1007/BF00320507.

Kaspari, M., Clay, N.A., Lucas, J., Yanoviak, S.P., Kay, A., 2015. Thermal adaptationgenerates a diversity of thermal limits in a rainforest ant community. Glob. Chang.Biol. 21, 1092–1102. http://dx.doi.org/10.1111/gcb.12750.

Kaspari, M., Valone, T.J., 2002. On ectotherm abundance in a seasonal environment –studies of a desert ant assemblage. Ecology 83, 2991–2996. http://dx.doi.org/10.1890/0012-9658(2002) 083[2991:OEAIAS]2.0.CO;2.

Kimuyu, D.M., Sensenig, R.L., Riginos, C., Veblen, K.E., Young, T.P., 2014. Native anddomestic browsers and grazers reduce fuels, fire temperatures, and acacia ant mor-tality in an African savanna. Ecol. Appl. 24, 741–749. http://dx.doi.org/10.1890/13-1135.1.

Kühsel, S., Blüthgen, N., 2015. High diversity stabilizes the thermal resilience of polli-nator communities in intensively managed grasslands. Nat. Commun. 6, 7989. http://dx.doi.org/10.1038/ncomms8989.

Lange, D., Dáttilo, W., Del-Claro, K., 2013. Influence of extrafloral nectary phenology onant-plant mutualistic networks in a neotropical savanna. Ecol. Entomol. 38, 463–469.http://dx.doi.org/10.1111/een.12036.

Leal, L.C., Peixoto, P.E.C., 2016. Decreasing water availability across the globe improvesthe effectiveness of protective ant-plant mutualisms: a meta-analysis. Biol. Rev.http://dx.doi.org/10.1111/brv.12307.

Le Stradic, S., Hernandez, P., Fernandes, G.W., Buisson, E., 2016. Regeneration after firein campo rupestre: Short- and long-term vegetation dynamics. Flora – Morphol.Distrib. Funct. Ecol. Plants 1–10. http://dx.doi.org/10.1016/j.flora.2016.12.001.

Lehmann, C.E.R., Anderson, T.M., Sankaran, M., Higgins, S.I., Archibald, S., Hoffmann,W.A., Hanan, N.P., Williams, R.J., Fensham, R.J., Felfili, J., Hutley, L.B., Ratnam, J.,Jose, J.S., Montes, R., Franklin, D., Russell-Smith, J., Ryan, C.M., Durigan, G.,Hiernaux, P., Haidar, R., Bowman, D.M.J.S., Bond, W.J., San Jose, J., Montes, R.,Franklin, D., Russell-Smith, J., Ryan, C.M., Durigan, G., Hiernaux, P., Haidar, R.,Bowman, D.M.J.S., Bond, W.J., 2014. Savanna vegetation-fire-climate relationshipsdiffer among continents. Science (80-.) 343, 548–553. http://dx.doi.org/10.1126/science.1247355.

Maravalhas, J., Vasconcelos, H.L., 2014. Revisiting the pyrodiversity-biodiversity hy-pothesis: long-term fire regimes and the structure of ant communities in a Neotropicalsavanna hotspot. J. Appl. Ecol. 51, 1661–1668. http://dx.doi.org/10.1111/1365-2664.12338.

Maurin, O., Davies, T.J., Burrows, J.E., Daru, B.H., Yessoufou, K., Muasya, A.M., van derBank, M., Bond, W.J., 2014. Savanna fire and the origins of the “underground forests”of Africa. New Phytol. 204, 201–214. http://dx.doi.org/10.1111/nph.12936.

Neves, F.D.S., Lana, T.C., Anjos, M.C., Reis, A.C., Fernandes, G.W., 2016. Ants in burned

and unburned areas in campos rupestres ecosystem. Sociobiology 63, 628. http://dx.doi.org/10.13102/sociobiology.v63i1.779.

Okuyama, T., Holland, J.N., 2008. Network structural properties mediate the stability ofmutualistic communities. Ecol. Lett. 11, 208–216. http://dx.doi.org/10.1111/j.1461-0248.2007.01137.x.

Paolucci, L.N., Maia, M.L.B., Solar, R.R.C., Campos, R.I., Schoereder, J.H., Andersen, A.N.,2016. Fire in the Amazon: impact of experimental fuel addition on responses of antsand their interactions with myrmecochorous seeds. Oecologia 182, 335–346. http://dx.doi.org/10.1007/s00442-016-3638-x.

Parmesan, C., 2006. Ecological and Evolutionary Responses to Recent Climate Change.Annu. Rev. Ecol. Evol. Syst. 37, 637–669. http://dx.doi.org/10.1146/annurev.ecolsys.37.091305.110100.

Parr, C.L., Robertson, H.G., Biggs, H.C., Chown, S.L., 2004. Response of African savannaants to long-term fire regimes. J. Appl. Ecol. 41, 630–642. http://dx.doi.org/10.1111/j.0021-8901.2004.00920.x.

Peters, M.K., Hemp, A., Appelhans, T., Behler, C., Classen, A., Detsch, F., Ensslin, A.,Ferger, S.W., Frederiksen, S.B., Gebert, F., Haas, M., Helbig-Bonitz, M., Hemp, C.,Kindeketa, W.J., Mwangomo, E., Ngereza, C., Otte, I., Röder, J., Rutten, G.,Schellenberger Costa, D., Tardanico, J., Zancolli, G., Deckert, J., Eardley, C.D., Peters,R.S., Rödel, M.-O., Schleuning, M., Ssymank, A., Kakengi, V., Zhang, J., Böhning-Gaese, K., Brandl, R., Kalko, E.K.V., Kleyer, M., Nauss, T., Tschapka, M., Fischer, M.,Steffan-Dewenter, I., 2016. Predictors of elevational biodiversity gradients changefrom single taxa to the multi-taxa community level. Nat. Commun. 7, 13736. http://dx.doi.org/10.1038/ncomms13736.

R Development Team, 2017. R: A language and environment for statistical computing.doi:3-900051-07-0.

Rico-Gray, V., 1993. Use of Plant-Derived Food Resources by Ants in the Dry TropicalLowlands of Coastal Veracruz. Mexico. Biotropica 23, 301. http://dx.doi.org/10.2307/2388788.

Rico-Gray, V., Oliveira, P., 2007. The Ecology and Evolution of Ant-Plant Interactions.The University of Chicago Press, London, UK.

Rocha, N.M.W.B., Carstensen, D.W., Fernandes, G.W., Stradic, S. Le, Buisson, E.,Morellato, L.P.C., 2016. Phenology Patterns Across a Rupestrian GrasslandAltitudinal Gradient. In: Fernandes, G.W. (Ed.), Ecology and Conservation ofMountaintop Grasslands in Brazil. Springer International Publishing, Stanford, USA.

Siepielski, A.M., Morrissey, M.B., Buoro, M., Carlson, S.M., Caruso, C.M., Clegg, S.M.,Coulson, T., DiBattista, J., Gotanda, K.M., Francis, C.D., Hereford, J., Kingsolver, J.G.,Augustine, K.E., Kruuk, L.E.B., Martin, R.A., Sheldon, B.C., Sletvold, N., Svensson,E.I., Wade, M.J., MacColl, A.D.C., 2017. Precipitation drives global variation innatural selection. Science 355, 959–962. http://dx.doi.org/10.1126/science.aag2773.

Silveira, F.A.O., Negreiros, D., Barbosa, N.P.U., Buisson, E., Carmo, F.F., Carstensen,D.W., Conceição, A.A., Cornelissen, T.G., Echternacht, L., Fernandes, G.W., Garcia,Q.S., Guerra, T.J., Jacobi, C.M., Lemos-Filho, J.P., Le Stradic, S., Morellato, L.P.C.,Neves, F.S., Oliveira, R.S., Schaefer, C.E., Viana, P.L., Lambers, H., 2016. Ecology andevolution of plant diversity in the endangered campo rupestre: a neglected con-servation priority. Plant Soil 403, 129–152. http://dx.doi.org/10.1007/s11104-015-2637-8.

Tonkin, J.D., Bogan, M.T., Bonada, N., Rios-Touma, B., Lytle, D.A., 2017. Seasonality andpredictability shape temporal species diversity. Ecology 98, 1201–1216. http://dx.doi.org/10.1002/ecy.1761.

Vasconcelos, H.L., Maravalhas, J.B., Cornelissen, T., 2016. Effects of fire disturbance onant abundance and diversity: a global meta-analysis. Biodivers. Conserv 10.1007/s10531-016-1234-3.

Vázquez, D.P., Morris, W.F., Jordano, P., 2005. Interaction frequency as a surrogate forthe total effect of animal mutualists on plants. Ecol. Lett. 8, 1088–1094. http://dx.doi.org/10.1111/j.1461-0248.2005.00810.x.

Valiente-Banuet, A., Aizen, M.A., Alcántara, J.M., Arroyo, J., Cocucci, A., Galetti, M.,García, M.B., García, D., Gómez, J.M., Jordano, P., Medel, R., Navarro, L., Obeso,J.R., Oviedo, R., Ramírez, N., Rey, P.J., Traveset, A., Verdú, M., Zamora, R., 2015.Beyond species loss: the extinction of ecological interactions in a changing world.Funct. Ecol. 29, 299–307. http://dx.doi.org/10.1111/1365-2435.12356.

Veldman, J.W., Buisson, E., Durigan, G., Fernandes, G.W., Le Stradic, S., Mahy, G.,Negreiros, D., Overbeck, G.E., Veldman, R.G., Zaloumis, N.P., Putz, F.E., Bond, W.J.,2015. Toward an old-growth concept for grasslands, savannas, and woodlands. Front.Ecol. Environ. 13, 154–162. http://dx.doi.org/10.1890/140270.

Whelan, R.J., 1995. The Ecology of Fire. Cambridge University Press, Cambridge, UK.

F.V. Costa et al. Ecological Indicators 93 (2018) 247–255

255

http://dx.doi.org/10.1007/978-3-319-29808-5_15http://dx.doi.org/10.1007/978-3-319-29808-5_15http://refhub.elsevier.com/S1470-160X(18)30336-4/h0155http://refhub.elsevier.com/S1470-160X(18)30336-4/h0155http://refhub.elsevier.com/S1470-160X(18)30336-4/h0155http://refhub.elsevier.com/S1470-160X(18)30336-4/h0155http://refhub.elsevier.com/S1470-160X(18)30336-4/h0155http://dx.doi.org/10.1098/rspb.2015.0418http://dx.doi.org/10.1098/rspb.2015.0418http://refhub.elsevier.com/S1470-160X(18)30336-4/h0165http://refhub.elsevier.com/S1470-160X(18)30336-4/h0165http://refhub.elsevier.com/S1470-160X(18)30336-4/h0165http://refhub.elsevier.com/S1470-160X(18)30336-4/h0170http://refhub.elsevier.com/S1470-160X(18)30336-4/h0170http://dx.doi.org/10.1146/annurev-ento-010715-023719http://dx.doi.org/10.1146/annurev-ento-010715-023719http://dx.doi.org/10.1007/BF00320507http://dx.doi.org/10.1111/gcb.12750http://dx.doi.org/10.1890/0012-9658(2002) 083[2991:OEAIAS]2.0.CO;2http://dx.doi.org/10.1890/0012-9658(2002) 083[2991:OEAIAS]2.0.CO;2http://dx.doi.org/10.1890/13-1135.1http://dx.doi.org/10.1890/13-1135.1http://dx.doi.org/10.1038/ncomms8989http://dx.doi.org/10.1038/ncomms8989http://dx.doi.org/10.1111/een.12036http://dx.doi.org/10.1111/brv.12307http://dx.doi.org/10.1016/j.flora.2016.12.001http://dx.doi.org/10.1126/science.1247355http://dx.doi.org/10.1126/science.1247355http://dx.doi.org/10.1111/1365-2664.12338http://dx.doi.org/10.1111/1365-2664.12338http://dx.doi.org/10.1111/nph.12936http://dx.doi.org/10.13102/sociobiology.v63i1.779http://dx.doi.org/10.13102/sociobiology.v63i1.779http://dx.doi.org/10.1111/j.1461-0248.2007.01137.xhttp://dx.doi.org/10.1111/j.1461-0248.2007.01137.xhttp://dx.doi.org/10.1007/s00442-016-3638-xhttp://dx.doi.org/10.1007/s00442-016-3638-xhttp://dx.doi.org/10.1146/annurev.ecolsys.37.091305.110100http://dx.doi.org/10.1146/annurev.ecolsys.37.091305.110100http://dx.doi.org/10.1111/j.0021-8901.2004.00920.xhttp://dx.doi.org/10.1111/j.0021-8901.2004.00920.xhttp://dx.doi.org/10.1038/ncomms13736http://dx.doi.org/10.1038/ncomms13736http://dx.doi.org/10.2307/2388788http://dx.doi.org/10.2307/2388788http://refhub.elsevier.com/S1470-160X(18)30336-4/h0285http://refhub.elsevier.com/S1470-160X(18)30336-4/h0285http://refhub.elsevier.com/S1470-160X(18)30336-4/h0290http://refhub.elsevier.com/S1470-160X(18)30336-4/h0290http://refhub.elsevier.com/S1470-160X(18)30336-4/h0290http://refhub.elsevier.com/S1470-160X(18)30336-4/h0290http://dx.doi.org/10.1126/science.aag2773http://dx.doi.org/10.1126/science.aag2773http://dx.doi.org/10.1007/s11104-015-2637-8http://dx.doi.org/10.1007/s11104-015-2637-8http://dx.doi.org/10.1002/ecy.1761http://dx.doi.org/10.1002/ecy.1761http://refhub.elsevier.com/S1470-160X(18)30336-4/h0310http://refhub.elsevier.com/S1470-160X(18)30336-4/h0310http://refhub.elsevier.com/S1470-160X(18)30336-4/h0310http://dx.doi.org/10.1111/j.1461-0248.2005.00810.xhttp://dx.doi.org/10.1111/j.1461-0248.2005.00810.xhttp://dx.doi.org/10.1111/1365-2435.12356http://dx.doi.org/10.1890/140270http://refhub.elsevier.com/S1470-160X(18)30336-4/h0325

Resilience to fire and climate seasonality drive the temporal dynamics of ant-plant interactions in a fire-prone ecosystemIntroductionMaterial and methodsStudy areaSampling designClimatic variablesFire disturbanceTracking of ant-plant interactionsData analysisClimatic effectsFire effects

ResultsResponse of ant-plant interactions to climateFire effects on the temporal dynamics of interaction networks

DiscussionAuthors’ contributionsAcknowledgmentsSupplementary dataReferences