Fire regime and spatial distributions of leaf litter- and ground-dwelling ants (Hymenoptera: Formicidae) across the tapia woodland of Madagascar

Abstract

The Central Highland of Madagascar has a native vegetation formation known as tapia woodland that is able to withstand regular fires. However, the ant fauna of this habitat remains poorly understood. This study compares the distribution of ant species in recently burned (<1 yr since fire) and unburned (>4 years since fire) tapia vegetation, which is dominated by the tapia tree Uapaca bojeri (Phyllanthaceae). Three quantitative inventory methods—mini-Winkler, monolith, and pitfall traps—were employed along a 200-m transect with 20 plots per site. In total, 155 ant species were collected, comprising 146 native species (95%) and 8 introduced species (5%). A statistical analysis revealed no significant differences in species richness between the burned and unburned plots for each method. Divergent patterns in species composition were observed between paired burned and unburned plots across 3 sites: Ambositra (56 vs 64), Ibity (23 vs 42), and Itremo (60 vs 59). Aggregating data from paired burned and unburned plots increased the species richness per locality. At Itremo, the combined species richness was 86, compared to 59 in unburned plots alone. Similarly, at Ibity, despite fire negatively impacting vegetation structure, the combined species richness was 51, versus 43 for unburned sites. Introduced ant species did not significantly differ between burned and unburned sites, with at least 4 species recorded at each tapia formation. The discovery of ground-nesting Camponotus andrianjaka, the first ant species in Madagascar found to have repletes, indicates an adaptation to arid environments and a possible strategy to escape fire.

Introduction

Tapia woodland is dominated by the tapia tree, Uapaca bojeri (Baill, 1958), and is an ecosystem found in the Central Highland of Madagascar surrounded by widespread grassland (Goodman and Ratsirarson 2000, Kull 2002, Malaisse et al. 2013a). Tapia woodland is a rare native vegetation form that has withstood anthropogenic pressure for decades. The formation comprises several clusters of different sizes, stretching from Arivonimamo at its northeastern edge to the Isalo Massif at its southwest delimitation. Tapia woodland features less biodiversity in terms of plants, invertebrates, and other organisms than the forest ecosystem (Bond et al. 2008, Świerczewski and Stroiński 2012, Razafimanantsoa et al. 2013, Ravelomanana et al. 2022). The tapia woodland ecosystem includes a few endemic plant species (Rabehevitra and Lowry 2009, Silander et al. 2024) as well as a few newly described endemic ant species (Rakotonirina and Fisher 2022, Rasoamanana and Fisher 2022). Botanical diversity within tapia woodlands surpasses that observed in the vast surrounding grassland. Despite the apparent disparity between these 2 distinct vegetative ecosystems, they exhibit a notable overlap of C₄ grass species (Bond et al. 2008, Solofondranohatra et al. 2018, Rakotomalala et al. 2021b). These grass species, prevalent in areas classified as natural grasslands, distinguish the natural grassland ecosystem from anthropogenically derived grasslands. Most of the natural vegetation in all ecoregions of Madagascar is impacted and degraded by wildfire.

Malagasy ants are diverse and well studied for invertebrates. Several publications related to Malagasy ant ecology and systematics have been published annually for a number of years (Rasoamanana and Fisher 2022, Ramamonjisoa et al. 2023, Ravelomanana et al. 2023, Rasoarimalala et al. 2024). Ants in general are known for their diversity, abundance, and omnipresence across diverse terrestrial ecosystem, and are increasingly being recognized as useful tools for land managers to monitor ecosystem conditions (Underwood and Fisher 2006). An estimated 794 valid ant species belonging to 58 genera are described from Madagascar (Fisher 2022, Antweb 2024). Most described species were collected in humid and dry forest. Multiple studies for monitoring programs or rapid inventories were undertaken across habitats ranging from humid forest to dry forest, but greater diversity was observed in humid forest (Fisher and Robertson 2002, Dejean et al. 2010, Rakotonirina 2010, Ravelomanana and Fisher 2013, Rakotomalala et al. 2021a). For example, ant species richness in the humid forest of Anjanaharibe (n = 180) (Fisher 1999) was greater than that found in the grassland ecosystem in Ankazomivady (n = 31) (Fisher and Robertson 2002).

The forest cover of Madagascar has steadily diminished in recent decades, and continues to be threatened by human activities. Less than 10% of the original extent of Madagascar’s forests remain. Tapia woodland is one of the rarest natural vegetation endemic ecosystems of Madagascar. It is both adapted to fire and has proven able to withstand human. Derived from sclerophyllous forest, tapia woodland includes 4 native woody species known to exhibit fire tolerance: U. bojeri (Bail, 1958), Sarcolaena oblongifolia (Gérard, 1915), Schizolaena microphylla (Perrier, 1925), and Asteropeia densiflora (Baker, 1882) (Kull 2002, Alvarado et al. 2013, Silander et al. 2024). No complete inventory has ever been conducted to characterize the diversity of tapia woodland ants. Frequent burning and uncontrolled fires fueled by the surrounding flammable grasslands pose a permanent threat to the tapia woodland ecosystem. While fire has shaped this vegetation for decades (Kull et al. 2005), fire occurrence was different depending on the locality and the management implemented by the park manager. This explains in part the great variation in floristic richness and composition between Itremo and Ibity tapia formations (Kull 2002, Alvarado et al. 2013, Malaisse et al. 2013b). There has been limited study of the interactions between ants and fire. However, little is known about the ant diversity of tapia woodlands and how these engineering invertebrates respond to fire compared to in the savannas and woodlands of Africa (Andersen and Yen 1985, Parr et al. 2007). In the last 2 decades, most of the 50 newly described ant species of Madagascar have been found in forest ecosystems (Rakotonirina and Fisher 2013, 2022, Rasoamanana and Fisher 2022), with relatively few species described from grassland and woodland.

The 2 main objectives of this study are to inventory ants in the 5 main tapia woodland formations in the Central Highland and to compare their diversity in burned and unburned plots. In addition, we evaluated 3 qualitative and quantitative collecting methods to determine the most efficient approach to conduct a rapid inventory program in tapia woodlands. This study elucidates the spatial distribution of ant species within the 5 prominent tapia woodland formations. Additionally, we present preliminary data describing how ant diversity correlates with fire regimes in 3 distinct tapia woodland localities, Ambositra, Ibity, and Itremo.

Materials and Methods

Study Area and Investigation Periods

Tapia trees grow on western slopes and are found in several clusters around the Central Highland of Madagascar (800 to 1,600 m). The 4 major localities include Imamo, the “Col des Tapia,” Itremo, and Isalo (Kull and Birkinshaw 2022). Tapia woodland covers an area of approximately 2,600 km² (DEF 1996). The climate is generally subhumid, with temperatures ranging from 20 to 30 °C and yearly rainfall ranging from 800 to 1,500 mm (Fig. 1). The geology and soils of the Central Highland are notably diverse. Tapia woodlands are associated with a variety of geological formations, including basement rock in the northern part around Arivonimamo, quartzite formations in the Ibity Ambositra-Itremo region, and sandstone formations in the southern part around Isalo (Moat and Smith 2007).

Fieldwork was conducted during the rainy season, which falls between January and April (2010 and 2012). Inventories were carried out in 5 localities: Ambositra, Arivonimamo, Ibity, Itremo, and Isalo. At each of the 5 localities, sites were chosen that were unburned for at least 4 years (Fig. 2). In addition, at 3 of the localities, Ambositra, Ibity, and Itremo, sites were inventoried that were burned less than 1 yr previously (Fig. 3).

The 5 sites differed in terms of protection status. Ibity, Itremo, and Isalo are protected areas managed by local NGOs, respectively, the Missouri Botanical Garden for the 2 first sites and Madagascar National Park. Ambositra and Arivonimamo are nonprotected areas. Nonprotected areas, managed by the local community, are known as COmmunity BAsed (COBA).

Sampling Design and Methods

Sampling at each of the 8 sites (3 burned, 5 unburned) involved 3 methods with no replicates (Mini-Winkler, pitfall, monolith). Each method deployed a transect 200 m long, with 20 subsamples located at 10 m intervals for a total of 24 transects for 8 sites (8 sites × 3 methods). Standardized collecting methods were employed in parallel for litter- and ground-dwelling ants. We focused on litter-dwelling ants, combining the use of pitfall traps and the mini-Winkler method according to the Ants of the Leaf Litter (ALL) protocol (Agosti and Alonso 2000). For ground-dwelling ants, the monolith was used. The standardized collecting methods were: (1) Mini-Winkler—This method of leaf litter sifting is effective for collecting litter-dwelling ants (Fisher 1998, Agosti and Alonso 2000). It involved establishing 20 plots of 1 m² that were delineated, extracted, and sampled at 10 m intervals along the transect line. Litter samples were sifted with 1 cm mesh grid sieves. A minimum of 1 liter of litter from each plot was used for ant extraction. Each litter sample was placed in a single mesh bag and left in mini-Winkler sacks for 48 h. (2) Pitfall trap—This is the most common method of sampling active ground species such as ants (Fisher and Razafimandimby 1997, Parr and Chown 2001). Twenty plots were established, each containing 1 trap consisting of a pitfall tube placed in a hole 20 cm deep by 11 cm in diameter. Container traps were filled with ethanol to which 3 drops of glycerin were added, and left in place for 48 h. (3) Monolith—This technique sampled subterranean ants (Fisher and Robertson 2002, Kolo 2006, Ravelomanana and Fisher 2013). It involves digging out a cube of soil 30 cm on a side in the 20 plots. The collected samples were preserved in ethanol and meticulously sorted at the Madagascar Biodiversity Center. Voucher specimens were deposited at the California Academy of Sciences, San Francisco, United States.

Specimens were identified using identification keys to the level of species (Blaimer 2012, Francisco and Fisher 2012, Yoshimura and Fisher 2012, Rakotonirina and Fisher 2013). For undescribed species, the online database AntWeb.org was used (Antweb 2024). Unnamed species were given number codes. Ant species not native to Madagascar were classified as “introduced.” Each species collected during the survey was photographed and specimen records and images are accessible on Antweb.

Data Analysis

Only records of ant workers were used for analysis, even though queens and males were often captured in traps (Fisher and Robertson 2002). The observed number of ant species occurring per method was compared with Chao-2 estimator (analysis was generated from the combination of all methods used). A nonparametric estimator of species richness, Chao-2, was selected because it is based on incidence data and is widely used and accepted (Magurran 2004). Species accumulation curves were calculated for each site and used to assess the efficiency of the various sampling methods (Longino and Colwell 1997). Each curve was plotted as a function of the number of subsample and species occurrences (observed number of species and Chao-2). Curves were drawn using Estimates 9 software (Colwell 2006). The shape of the species accumulation curves depended on the ordering of samples; 100 randomizations were done to avoid this problem and smooth the curve (Colwell and Coddington 1994). When the curve reached an asymptote, all species diversity in the area was assumed to have been captured. However, reaching an asymptote for hyperdiverse organisms such as arthropods is rare.

Variations in species richness of ants across sampling sites were tested using analysis of variance (ANOVA), followed by Tukey’s post hoc tests across sampling sites (site was used as the independent variable). The measure of dissimilarity used was the Bray–Curtis index (Legendre and Legendre 1998). A matrix representing the presence and absence of species in each of the 5 study sites was analyzed using ordination techniques. The same data were subsequently used to create a dendrogram based on a dissimilarity matrix, with the Bray–Curtis index serving as the distance measure. The software XLstat was used to draw the dendrogram and the statistical analysis.

Results

Richness and Efficacy of the Inventory Methods Across All Sites

For the 8 study sites using all 3 methods, the total species richness recorded was 155 (Table 1). Of these, 120 ant species were identified and described, with the remaining 34 identified to morphospecies. Notably, 95% of these species were endemic to Madagascar, with 8 introduced species. The mini-Winkler method captured the highest number of species (114 species, with 36 unique species), followed by the monolith method (97 species, with 19 unique species), and the pitfall trap (76 species, with 8 unique species). For all 3 methods combined, Isalo had the greatest number of species (n = 67), followed closely by Ambositra (n = 65). The site with the lowest species richness was Ibity, with 23 species collected. Within a site, the number of species collected by each sampling method differed significantly (1-way ANOVA, F = 9.207, df = 2, P < 0.001). The mini-Winkler method collected the highest number of species at 6 of the 8 sites, while the monolith method was most effective at 2 sites.

In terms of sampling effort, none of the species accumulation curves from each method reached an asymptote, indicating ongoing species discovery (Figs. 4 and 5). The nonparametric estimator for all main curve for Chao-2 (Fig. 4 or 5) also indicates unexplored diversity exists. At least 10 additional species were added for each site sampled. The exception was at Ibity, where all methods reached an asymptote, but this site had the lowest number of species captured (n = 23) (Fig. 5a). The combination of the mini-Winkler and pitfall methods, known as the ALL protocol, collected more species than other method combinations (Fig. 5). The greatest number of species was collected at Itremo (n = 59), followed by Ambositra (n = 50).

Species Richness and Composition in Unburned Sites

In the 5 unburned tapia woodland sites, 28 genera comprising 135 species were identified. Endemic species represented 95% of the total ants collected (n = 126), with 8 introduced species. Among the introduced species were Camponotus maculatus (Fabricius, 1782), Cardiocondyla emeryi (Forel, 1881), C. wroughtonii (Forel, 1890), Erromyrma latinodis (Mayr, 1872), Nylanderia bourbonica (Forel, 1886), Strumigenys maxillaris (Baroni Urbani, 2007), Technomyrmex albipes (Smith, 1861), and T. pallipes (Smith, 1876).

A comparison of the species richness collected by each method across the 5 unburned localities showed significant differences (1-way ANOVA, F = 5.166, df = 2, P < 0.01). On average, 27 ant species were collected per sampling. Species richness for each locality is presented in the Supplementary Tables. Higher species richness was observed in the protected areas, with 67 species at Isalo and 65 species at Ambositra, an unprotected area. The lowest species richness among the protected areas was at Ibity (n = 43). Despite Ibity’s protected status, uncontrolled fires burned much of the area annually, making it challenging to find plots that remained unburned for 4 years (Fig. 6). Analysis based on the Jaccard index highlighted differences in species composition between sites. Geographically close sites, like Itremo and Ambositra, had a similarity of up to 50%, while Arivonimamo, Ibity, and Isalo had similarities ranging from 35% to 40%. The proportion of unique ant species associated with each locality averaged 5% to 20%, with Isalo having 20% of its species (n = 28) unique to its locality.

This weak similarity can be associated with the species composition collected from each subfamily, the habitat structure (e.g., tree density, litter quantity after 4 years unburned), and ant biology (Fig. 6). Some genera were represented by more than 10 species, while others were limited to 2 species. In general, the ant fauna in unburned tapia woodland sites was dominated by Myrmicinae (48%, n = 65), followed by Formicinae (26%, n = 35), and Ponerinae (15%, n = 15). Other subfamilies were less abundant (<5%) (Fig. 6). The rarest subfamilies were Amblyoponinae (n = 2) and Dolichoderinae (n = 5). Some genera, such as Camponotus (n = 20), Hypoponera (n = 9), Nylanderia (n = 9), Pheidole (n = 11), and Strumigenys (n = 10), were relatively abundant in at least 3 localities. The main differences between the 5 tapia sites studied were soil type, climate, and woodland management.

Species Richness and Composition in Burned Sites

Across the 3 burned study sites, 98 species belonging to 24 genera were identified, averaging 47 ant species per site. The most species-rich site was Itremo (62 species; 63% of all species), while the site with the lowest richness was Ibity (23 species; 23% of all species). There was no statistical difference in the number of introduced species collected between burned and unburned sites (Student’s test, P > 0.05). Compared to the richness in unburned sites (n = 135 species), 20 unique ant species were recorded in the burned sites. Among these, Proceratium fhg-mala was the only species of Proceratinae collected in this study. Most of these unique ant species were collected by the mini-Winkler method (n = 12) and predominantly belonged to the subfamily Myrmicinae (n = 10). Other families were represented by 1–4 species. The proportion of subfamilies across the 3 sites remained broadly constant for the 5 most frequently represented subfamilies (Amblyoponinae, Dolichoderinae, Formicinae, Myrmicinae, and Ponerinae; Fig. 6). There was no statistical difference in species collected by each sampling technique (1-way ANOVA, F = 3.338, df = 2, P > 0.05). However, differences in species composition show complementarity in species collected, as observed through the shape of the species accumulation curves for different combinations of 2 and 3 methods (Fig. 5). Hierarchical cluster analysis showed that species similarity between the 8 sites, including burned and unburned plots, was low, ranging from 25% to 45% (Fig. 7). Each tapia woodland has its own ant fauna resulting from different geological and climatic conditions, which influence the dominant plant species and, consequently, the microclimate and ant distribution. The species similarity between Isalo, with sandstone formations, and Arivonimamo, with quartzite formations, is low. Additionally, these 2 localities are geographically distant, separated by approximately 530 km.

Discussion

Ant Diversity and Species Composition

Large-scale ant inventory programs have been conducted in the forest ecosystems of Madagascar over the past 2 decades (Fisher 1998, Fisher and Robertson 2002, Ravelomanana and Fisher 2013), but surveys of tapia woodland have been limited. Whether tapia woodland is a vestige of the island’s sclerophyllous forest or not is arguable. Perhaps because of its lower diversity, this ecosystem has been less of a priority for biological conservation programs (Koechlin et al. 1974, Kull 2002, Gautier et al. 2018). Furthermore, because tapia woodland features less biological diversity than forest ecosystems, it was labeled a degraded anthropogenic formation by Kull (2002). However, recent studies on floristic compositions, such as the dominance of C₄ grasses (Nanjarisoa et al. 2017, Lehmann et al. 2022) and the presence of a few endemic trees, like the tapia tree U. bojeri, demonstrate the unique plant diversity of this woodland and its resilience to human pressures such as burning. This study represents the first comprehensive inventory of leaf litter- and ground-dwelling ant diversity in tapia woodland. Of the 155 species recorded, 147 species were endemic to Madagascar (95%) (Table 1). This species richness represents 19% of the total ant fauna described from Madagascar (155/794 species) (Fisher 2022). Based on the ant diversity observed in the 8 localities with each sampling method, there were statistically significant differences in species richness. Typically, 53 ant species were collected per locality. The highest species richness was at Itremo (n = 59) and the lowest at Ibity (n = 43). The earlier study on floristic diversity conducted by Alvarado et al. (2013) reached a similar conclusion, highlighting the unequal diversity between these 2 sites. Despite this pattern, tapia woodlands offer a reservoir of fresh biological material, including potentially undiscovered ant species, and present significant opportunities for scientific exploration. Among the recently described species from this woodland are Camponotus andrianjaka and C. tapia, both of which belong to the Formicinae (Rasoamanana and Fisher 2022). According to the species distributions documented on Antweb (2024), these species are present only in tapia woodland. The wide grassland surrounding the tapia formation in the Central Highland is inhospitable to most endemic plants save for the tapia tree and some Sarcolaenaceae species known to be fire tolerant. Our findings underscore the distinctiveness of ant species diversity within the tapia woodland ecosystem relative to the surrounding vegetation, highlighting the urgent need for conservation efforts to safeguard the remaining biodiversity of this habitat.

Other factors affecting the composition of ant species include ant biology, soil type, climate, and fire regime. The 8 localities studied were geographically distant, and these abiotic factors differ depending on the site location. Our findings indicate that soil type and climate together explain the variation in species composition and similarity based on the Jaccard index (Fig. 7). These parameters are entirely different and the diversity of ant species identified is distinct for Isalo (n = 67) and Itremo (n = 59), The climate, based on average annual rainfall, is dry at Isalo (900 mm) and temperate at Itremo (1,500 mm). At Isalo, the soil is ferruginous and sandy compared to Itremo, which has quartzite and ferritic soil (Moat and Smith 2007). In Isalo, we observed a greater number of species living in the soil when using the monolith method compared to Itremo, where we found higher species diversity in the leaf litter (mini-Winkler). The temperature is higher at Isalo, and rainfall is lower, with the number of rainy days ranging from 80 to 90 (Moat and Smith 2007). The dry season ranges from nonexistent to over 8 months long. Within this generally arid environment, the monolith method played a pivotal role in unveiling intriguing aspects of ant behavior. The method led to the excavation of nests, revealing fascinating insights into the subterranean habitats of certain ant species. Notably, the endemic species C. andrianjaka (Rasoamanana and Fisher 2022), identified in nests located at Isalo and Itremo, exhibited an intriguing adaptation by establishing colonies as deep as 2 m underground. This discovery underscores the importance of employing specialized techniques to explore and comprehend the diverse ecological strategies adopted by ant communities in response to the challenges posed by extended dry periods and the imperative to escape wildfires. The workers have hypertrophied abdomens similar to replete ants that belong to Myrmecocystus (Formicinae). In the New World, some workers of Myrmecocystus serve as living pantries for the colony during lean periods. This is the first record of the presence of replete or honey pot ants in Madagascar.

Efficiency of Methods for Rapid Assessment of Ant Leaf Litter- and Ground-Dwelling Ants

Here, we tested 3 quantitative methods of sampling ants in woodland habitats. An evaluation of their efficiency will help the design and optimization of future studies in similar habitats. Sampling efforts were estimated by drawing species accumulation curves for each single and combined method (Figs. 4 and 5). Each method collected a complementary number of species; however, the mini-Winkler method collected the greatest number of species and unique species (n = 98/35). Monolith sampling was the second-most important method (n = 70/20; Fig. 5). The current study shows that each method requires specific conditions and does not provide a complete inventory on its own. Estimator Chao-2 estimates more species expected than species richness observed. It predicted even higher species diversity, despite the choice of a 200 m of transect per sampling method to increase the sampling efforts. The heterogeneity of the habitat sampled is great, increasing the number of microhabitats for ants to colonize. Our on-site observations reveal that the habitat heterogeneity of tapia woodland demands tailored methodologies for effective ant collection. Mini-Winkler and pitfall traps have primarily been used on a large scale in forest ecosystems (Fisher 1996) and, secondly, in open landscapes such as savanna grasslands in Madagascar (Fisher and Robertson 2002, Ravelomanana and Fisher 2013) and elsewhere (Andersen et al. 2008). According to these findings, the best combination methods were mini-Winkler combined with monolith (Figs. 4 and 5). The monolith is not widely used or documented for ground-dwelling ant fauna in Madagascar (Ravelomanana and Fisher 2013). Introducing the monolith method, employed for the first time in this ant survey for tapia woodland, produced a notable increase in the overall number of species collected, adding 10 to 25 species depending on the locality. This is the first published study to demonstrate the importance of using the monolith alongside the ALL protocol.

Fire Regime and Conservation Challenges

Before human settlements and the uncontrolled use of fire, the structural and floristic diversity of the tapia woodland was somewhat different (Goodman and Junger 2014). For example, giant herbivorous tortoises, now extinct, were undoubtedly present in the past and would have shaped past vegetation. The absence of tortoises can be explained by human activities such as hunting or wildfire. The giant reptiles could not survive the onslaught of frequent anthropogenic fires. For invertebrates like ants, we found that areas not burned for at least 4 years compared with plots recently burned (1 yr) had similar species richness: for example, Itremo had 59 species in unburned plots versus 60 species in burned plots. Regarding the efficiency of the 3 sampling methods, there is no statistical difference in the number of species collected. However, the species composition in paired burned and unburned sites differed. Additionally, the burned sites had 20 ant species absent from unburned sites. Fire can interact positively with ant composition based on the fire regime and biomass available, as observed in our data for Itremo and Ambositra (Fig. 5b and 5c). Wildfire had spared the tapia in these 2 sites for 4 years. One explanation for the different composition of species in burned sites is that some species are better adapted to infrequent fires, or not all of the plot is burned; this is the case for wildfires during the rainy season. A wildfire ignited by lightning preserves more leaf litter during the rainy season and covers relatively small patches of woodland. Blazes begun this way are considered to result in less destructive fires and tend to burn smaller areas than anthropogenic fires. Half of these 20 unique species belong to the subfamily Myrmicinae, which typically features scavenger workers that build subterranean nests, such as with Pheidole navoatrensis-complex (Salata and Fisher, 2020), and Carebara sampi (Azorsa and Fisher, 2018). Ground-nesting ants can thus escape the effects of fire more effectively than leaf litter ants such as Ravavy miafina (Fisher, 2009), and Anochetus madagascarensis (Forel, 1887), whose niches are directly impacted by fire.

This investigation provides insights into the biology of Malagasy ant species, specifically their ability to switch habitat preferences depending on the availability of litter resources and habitat disturbance (burned or not). This data should be augmented with detailed information on plant species, particularly endemic grasses, to enhance the quality of our data set and deepen understanding of the conservation measures required to protect these woodlands. Notably, inadequate management has contributed to the rarity of Malagasy C₄ grass species. These grassland formations have become increasingly vulnerable and unable to endure the intense pressure from recurrent human-ignited fires (Bond et al. 2023). Further ecological data are necessary to evaluate the suitability of these 20 ant species as bioindicator species for burned lands. To gain comprehensive insights, it is essential to replicate this ant study in regions with diverse fire regimes across the Central Highland of Madagascar. Such information will contribute to the effective management of extensive woodlands and grasslands, as recommended by Parr and Chown (2001) for the South African savanna.

Acknowledgments

We thank the Madagascar Biodiversity Center and Entomology Department at the University of Antananarivo for helping to realize this research. Special thanks to the ant team and field assistants: Michelle Eposito, Rodolphe Ndriamanisarison, Jean Jacque Rafanomezantsoa, Tahina Rajaonera, Balsama Rajemison, Jean Claude Rakotonirina, Chrislain Ranaivo, Clav Randrianandrasana, Nicole Rasoamanana, Hanitra Rasoazanamavo, and Ricca Raveloson. We thank Steve Goodman for his comments on the manuscript, which were done in the context of his 2034–2024 Madagascar Fulbright U.S. Scholar grant.

Author contributions

Andrianjaka Ravelomanana (Conceptualization [equal], Data curation [lead], Formal analysis [lead], Methodology [lead], Software [equal], Supervision [lead], Validation [equal], Writing—original draft [lead], Writing—review & editing [lead]), Lala Ravaomanarivo (Conceptualization [equal], Methodology [supporting], Project administration [supporting], Resources [equal], Visualization [equal], Writing—review & editing [equal]), Vonjison Rakotoarimanana (Conceptualization [equal], Formal analysis [equal], Writing—review & editing [supporting]), Herisolo Andrianiaina Razafindraleva (Visualization [equal], Writing—review & editing [equal]), and Brian Fisher (Conceptualization [equal], Data curation [equal], Formal analysis [equal], Funding acquisition [lead], Methodology [equal], Project administration [equal], Supervision [equal], Validation [equal], Writing—review & editing [equal])

Funding

The Norvig Family Foundation and Lakeside Foundation supported this research.

Conflicts of interest. None declared.

References