Crickets as indicators of ecological succession in tropical systems, New Caledonia

Crickets (Ensifera, Grylloidea) are not commonly used as ecological indicators in contrary to other Orthoptera (e.g., grasshoppers and katydids). However, they are sensitive to environmental changes and abundant in tropical regions. To evaluate whether crickets are relevant bioindicators of tropical ecosystems, we investigated cricket assemblages along a tropical ecological gradient. We collected crickets during both day and night in southern New Caledonia for three stages of ecological succession: open shrubland, preforest, and forest. Simultaneously, we measured several environmental variables, such as temperature and relative humidity, at each sampling site. Cricket species assemblages showed a clear response to ecological succession. The highest and lowest species richness and abundances of individuals were, respectively, found in forest and shrubland, with species specialized in each ecological stage revealing the conservation value of each of these stages. Similar results were found when considering only the part of cricket communities with the ability to acoustically communicate. This work is part of a larger research program about Neocaledonian crickets and contributes to support the use of acoustic approaches to monitor tropical environments. In conclusion, these findings highlight the potential value of crickets as an environmental indicator in tropical ecosystems. The results also contribute to the discussion of the intrinsic conservational value of shrublands in New Caledonia and similar ecotypes.

Two of the most common diversity indices are species richness and evenness (Ricotta, 2005). Such diversity indices are dependent upon species inventories, which are time-consuming (Lawton et al., 1998), expensive, and require highly skilled specialists (Levrel et al., 2010).
The United Nations has therefore strongly encouraged the scientific community to develop biodiversity indicators for biodiversity conservation (Strategic Plan for Biodiversity , United Nations, 2020, IPBES, 2019. Bioindicators are generally used to assess environmental change, to monitor specific disturbances, or to estimate taxonomic diversity (Gerlach et al., 2013;McGeoch et al., 2011). Such indicators serve several applications, including assessing the condition of the environment (Niemi & McDonald, 2004), following ecosystem restoration programs (Alignan et al., 2014), or prioritizing the management of habitats (Fartmann et al., 2012). The challenge in selecting taxa as bioindicators in complex habitats, such as tropical forests, is that they require sufficient knowledge about their distributions and natural fluctuations, with a fairly resolved taxonomic knowledge. This is frequently absent in tropical environments (Moritz et al., 2001).
Among invertebrates, Orthoptera (spanning greater than 28,000 valid species; Cigliano et al., 2020) are widely used to assess ecological or environmental changes (Gerlach et al., 2013). They are a key resource for upper food web species, such as lizards, spiders, and birds (Borges et al., 2013;Pretelli et al., 2014;Shea et al., 2009;Taylor et al., 2012) and play an important role in recycling organic matter (Prather et al., 2013). Related to ecological changes, Orthoptera species richness and composition have been negatively associated with mowing and grazing (Rada et al., 2014;Rambo & Faeth, 1999), and they are regularly used to assess the success of restoration programs (Alignan et al., 2014(Alignan et al., , 2018Hugel, 2012;Rácz et al., 2013). Among Orthoptera, grasshoppers and katydids are commonly used to assess environmental change and restoration in temperate regions where they are well-diversified and can be easily identified, especially in open habitats (Fartmann et al., 2012;Marini et al., 2009;Schirmel et al., 2011). Grasshoppers were also found to be markers for forest succession in Western Amazonia (Amédégnato, 1997;Amédégnato & Descamps, 1980). Crickets (Orthoptera, Grylloidea) are the other large group of Orthoptera, with approximately 6000 species (Cigliano et al., 2020).
Crickets are studied as models in ethology, neurobiology, physiology, and bioacoustics (Gerhardt & Huber, 2002;Kulkarni & Extavour, 2019) but receive less focus in ecology. Most ecological studies assessing restoration or habitat management using Orthoptera assemblages have not separated grasshoppers, katydids, and crickets. In particular, only a few have considered the cricket community alone (e.g., Hoffmann et al., 2002;Szinwelski et al., 2012). While underexplored, crickets as potential bioindicators in the endangered tropics hold significant potential because of their high level of diversification and endemism in the tropics (Cigliano et al., 2020), abundance, and local ecological specializations (Desutter-Grandcolas, 1995, and use of recognizable acoustic signals Diwakar & Balakrishnan, 2007) when using passive acoustics for monitoring.
The goal of the present study was to evaluate whether crickets, at the species and community level, might be qualified as relevant bioindicators of tropical ecosystems. The effect of vegetation and bioclimatic attributes along shrubland to forest gradient on cricket assemblages was assessed. Research was performed in New Caledonia, which was advantageous for two reasons. First, a detailed description of the cricket species living in the area of interest was needed Desutter-Grandcolas et al., 2016), and second, New Caledonia is representative of various tropical ecosystems. Crickets represent a well-diversified part of the original and disharmonic insect fauna of New Caledonia (Grandcolas et al., 2008), with 40 genera and more than 180 species recorded today in the archipelago. A total of 19 of the genera and more than 90% of the species are endemic to the territory . Guessestimate of the total richness of the cricket fauna, based on the putative pattern of species assemblage in New Caledonia , leads to the hypothesis that the actual number of crickets in New Caledonia should be much more important, especially because of the high microendemism that characterizes this fauna (Grandcolas et al., 2008;Nattier et al., 2011). Nevertheless, the cricket fauna living in our research area is well described Desutter-Grandcolas et al., Desutter-grandcolas et al., 2016). New Caledonia is additionally a biodiversity hotspot (Myers et al., 2000), with a high endemism rate (Grandcolas et al., 2008), and experiences multiple disturbances, including species invasion, and habitat loss (Ibanez et al., 2019;Jaffré, Bouchet, & Veillon, 1998;Jaffré, Rigault, & Dagostini, 1998;Pellens & Grandcolas, 2009). Therefore, furthering ecological knowledge in this context holds high value for local conservation plans by helping to evaluate forest restoration programs, which are of high importance due to mining activity in New Caledonia (Losfeld et al., 2014).
In this study, a large taxonomic inventory was conducted during both day and night to measure the overall abundance, species richness, composition, and structure of the cricket communities. Data were collected in three ecological stages defined by a gradient of microclimate and vegetation characteristics (forest, preforest, and shrubland) in the south of the New Caledonian archipelago's main island, on ultramafic soil. New Caledonian ecosystems are not naturally fire-prone ecosystems (McCoy et al., 1999;Stevenson et al., 2001), but fire is promoted by human populations since their arrival.
On these soil compartments, the succession is marked by the transition from open shrubland to arbustive shrubland to preforest to forest stages. Each stage shows clear-cut vegetation formation in terms of stem density, litter cover, and stratification (Jaffré, Rigault, & Dagostini, 1998;McCoy et al., 1999). The context of succession on ultramafic soils is similar to South Africa and Southwest Australia, with shrubland that resembles the South African fynbos or the Southwest Australian kwongankaroo (Pillon et al., 2021).
Analyzing these data, the following questions were addressed: (1) Are assemblages of crickets following the ecological succession? (2) Are any specific cricket species an indicator of the ecological stages? And (3) is acoustic monitoring a suitable method for the evaluation of the ecological stages? This research is part of a larger research program conducted in New Caledonia focusing on biodiversity monitoring using taxonomic and acoustic methods. The results will thus be interpreted and discussed within the larger findings of this program. In particular, a section of the discussion will be dedicated to Passive Acoustic Monitoring (PAM) (Sugai et al., 2019) for extracting information about the cricket community indicating their tropical habitat.

| Study sites
Cricket communities were studied in the southern region of Grande Terre, the main island of New Caledonia, an archipelago located in the Southwest Pacific Ocean. Sites were selected along the same geological ultramafic substrate at low altitude (231 ± 66 m; Table 1).
To study the ecological succession of cricket fauna, three vegetation stages-forest, preforest, and shrubland-were considered. Four sites were chosen within each stage for a total of twelve sites. Preforest can be as dense as the forest but is composed of different plant species, characterized by a greater canopy openness and lower diameter tree stems. As in the forest, the bare ground is covered. The tree layer of preforest includes an assemblage of Styphelia cymbulae, Hibbertia lucens, Alphitonia neaocaledonica, and Gymnostoma deplancheanum, while the herbaceous layer is dominated by Lepidosperma perteres.
Shrubland is characterized by a high bare ground expanse and an absence of a continuous tree layer. More specifically, shrublands encompass small patches of vegetation with a thin leaf litter, separated by bare ground. The maximum vegetation height is around 3 m. Shrublands were dominated by shrubs but also include herbaceous species such as Sannantha sp., Eugenia sp., Alphitonia neocaledonica, Hibbertia lucens, and H. pancheri.

| Cricket sampling
Crickets were collected between November 2013 and April 2014. In each of the twelve sites selected (see section 2.1), two squared parcels with 10 m sides were delimited. Sampling was performed under clear sky meteorological conditions from hours 0900 to 1700 by day, and 1900 to 0000 by night. On each parcel, crickets were collected using the same collection method of ten 30-minute sessions, with five total collections by day and five by night. Crickets were located using both sight and song cues in the field only based on human perceptions, which is the most appropriate sampling method for crickets (Touroult et al., 2021). Another common method to collect insects seems inefficient for crickets (Touroult et al., 2021) such as (1) light trap that rarely attracts crickets, (2) pitfall that only captures specimens foraging in the leaf litter, (3) fogging, a method previously tried in New Caledonia leading to collecting mostly juveniles that could not be identified (LDG, pers. obs), or (4) interception trap that captures only flying insects, very rarely crickets. For each individual, activity (e.g., singing, eating, mating, resting) and microhabitat (e.g., trunk, leaf litter, vegetation, rocks, height from ground) were noted before the capture. Collected specimens were kept for identification, as most revealed new species to science . Specimens are deposited in the Museum national d'Histoire naturelle de Paris with a reference collection in the Nouméa IRD center in New Caledonia. Species identifications were performed in the laboratory by comparison with the specimens in the deposited collections and using the classification derived from the extensive molecular phylogeny of crickets (Chintauan-Marquier et al., 2015). Based on previous taxonomic descriptions including acoustic production (i.e., Desutter-grandcolas et al., 2016), we were able to attribute to each species a presence or absence of singing abilities. Of the 1030 individuals collected, 54 were discarded from the analyses. These specimens were not identifiable at the species level because of their destruction due to unfavorable TA B L E 1 Location and GPS information of the 12 sites. The coordinate reference system used was WGS84  (Table S1) and the list of species full names (Table S3) are provided in the supplementary information.

| Environmental variables
During cricket sampling, environmental attributes were measured in each parcel, including the percentage of bare ground, the percentage of three vegetation layers (herbaceous, shrub, and tree), the number and diameter (dbh) of stems, the vegetation height, and the plant species richness. Canopy openness was extracted from photos taken with a 180° hemispherical (fisheye) lens in each parcel corner and in its center. From the photos, canopy openness was then calculated using the Gap Light Analyzer software (version 2.0, Simon Fraser University, Cary Institute of Ecosystem Studies) by estimating the percentage of light in the forest overstorey (Frazer et al., 1999).
Daily temperature (°C) and relative humidity (%) were recorded every 30 minutes for 15 days at a 50-cm height above the ground using waterproof thermo-hygrometer sensors (HOBO U23 Pro v2; HOBOware software). A table of the environmental variables is available in the Supplementary Information (Table S2).

| Statistical analysis
Two datasets were created for statistical analyses. The first included the environmental attributes averaged per site, while the second described the cricket assemblages, defined here as the number of individuals per species averaged per site. All statistical analyses were performed using R (software; R Core Team, 2019), with the type I error set at 5%.
To evaluate differences between ecological stages a Kruskal-Wallis test followed by the post hoc Conover test for multiple comparisons with the Bonferroni correction was applied (Conover & Iman, 1979). This test was run separately on species richness (number of species), overall abundance (total number of individual crickets), and environmental variables. This test was computed using the function posthoc.kruskal.conover.test of the "PMCMR" R package (Pohlert, 2014).
To visualize the differences in cricket assemblages along the ecological succession, a Nonmetric Multidimensional Scaling (NMDS) was performed on cricket assemblages across sampling sites (Legendre & Legendre, 2012;Minchin, 1987). The Bray-Curtis distances of cricket assemblage were calculated between sites. NMDS analysis with 1000 iterations and two dimensions was then performed on each of these distance matrices. To explain each of the four NMDS observations, the environmental variables were fitted into ordination, and their significance was assessed with a permutation test (n = 999 permutations). The percentage of variance in cricket assemblages that can be explained by the ecological stages was assessed with a permutation test (n = 999 permutations) applied to the same Bray-Curtis distance matrix. This analysis was similarly conducted on cricket assemblages and summarized in four ways: (1) cricket species abundances, (2) stridulatory cricket species abundances, (3) presence or absence of cricket species, and (4) presence or absence of stridulatory cricket species. The function metaMDS was used to perform the Bray-Curtis distance and the NMDS analyses, while the permutation test was performed using the function envfit for environmental data. Lastly, the permutation test applied on the distance matrix of cricket assemblages was performed using the function adonis. These three functions are from the "vegan" R package (Oksanen et al., 2019). The function s.class from the "ade4" R package was used for the graphical scatter diagram representation of the ecological stages in the NMDS space (Dray & Dufour, 2007).
The beta diversity (β) of cricket assemblages measured as the Sorensen dissimilarity was partitioned in the turnover (β SIM ) and the nestedness (β NES ) components, being, respectively, the Simpson dissimilarity and the nestedness-resultant fraction of Sorensen dissimilarity. Abundances were first reduced to a simple presence or absence of the cricket species at each site. Then, their overall beta diversity was partitioned using the function beta.multi from the "betapart" R package .
The same analysis was conducted on the presence or absence of the cricket species for each ecological stage.
To evaluate the most relevant species between paired ecological stages, a similarity percentage analysis (SIMPER) was performed on both the abundances per cricket species and the presence or absence of cricket species (Clarke, 1993;Warton et al., 2012). The Bray-Curtis dissimilarity was decomposed in species contribution.
This analysis was performed using the function simper to the "vegan" R package (Oksanen et al., 2019). The R code used to compile, manipulate, and analyze the data is available as Appendix S4 and https:// github.com/agasc/ Anso-et-al-2022-Biotr opica.

| Cricket species along the succession
A total of 976 specimens, 30 species, and 16 genera were collected, belonging to four families (Gryllidae, Phalangopsidae, Trigonidiidae, and Mogoplistidae) and six subfamilies ( Table 2). The most abundant species were Bullita fusca (Gorochov, 1986) and Agnotecous azurensis Desutter-Grandcolas, 2006 in both forest (40% and 31%, respectively) and preforest stages (35% and 25%, respectively). In shrublands, Koghiella flammea  was the most abundant species (88%). Species richness increased gradually along the ecological succession, from the lowest number of species in shrubland (7 species), followed by observed increases in preforest (15 species) and forest (20 species) (Figure 1). Each ecological stage was additionally found to harbor exclusive species. Forest demonstrated the highest ratio of unique species (60% including 12 species) followed by shrublands (43% including 3 species) and preforest (14% including 2 species). Forest and preforest shared eight species, TA B L E 2 List of cricket species from the taxonomic inventory in alphabetical order. The symbol a indicates the number of specimens collected Collected species additionally provided several unique characteristics. For instance, 70% of the cricket species collected were stridulatory species . It was also found that forests contained the lowest percentage of acoustic species when compared to total species richness (65%), followed by shrublands (71.4%) and preforest (78.6%) ( Table 2). Of all species collected, a total of eleven were collected only once during the entire taxonomic inventory. During this inventory, a larger number of species were collected at night (90%) than during the day (60%). Several climatic parameters differed in the succession. Mean and maximum temperature, along with daily variation, followed the ecological gradient, (i.e., the highest and lowest values for shrubland and forests, respectively). Mean and minimum relative humidity differed significantly between forest and the two other ecological stages, with lowest and highest values for shrublands and forests, respectively. Daily humidity variation followed the ecological gradient (i.e., highest in shrublands, followed by preforest and subsequently forest). The minimum temperature and the maximum humidity were not significantly different between the successional stages.

| Alpha diversity of cricket assemblages
Species richness of crickets was not significantly different between sites of differing ecological stages due to large standard deviation values. This was indicative of a high variation of intra-site species richness values. However, the overall abundance of crickets was significantly lower in shrubland than preforest and forest (Table 3).

| Beta diversity of cricket assemblages
Each ecological stage showed clearly separated cricket assemblages ( Figure 2) based on NMDS ordination (stress = 0.065, two dimensions). Thirteen environmental variables significantly con- Results from Section 3.3 that demonstrate a nonsignificant difference in species richness between sites of different ecological stages further support this idea.

| Relevant species by successional stage
SIMPER analysis demonstrated that specific cricket species were keys in distinguishing communities in the succession ( "mute species") and found in every forest site but only at two preforest sites. The two species A. clarus and C. meurieri were found to contribute the most to the difference between preforests and forests.

| DISCUSS ION
The results of this study highlight the central role of crickets in conservation biology and demonstrate that crickets could be relevant bioindicators of the restoration process and integrity of the environment in tropical systems.

F I G U R E 1
Cricket species assemblages along the ecological succession. * indicates stridulatory species.

| Cricket assemblages along ecological succession
Cricket assemblages clearly responded to the different stages along the environmental gradient, with several distinct species in each stage. Overall cricket abundances were highest in forested habitats (i.e., forest and preforest), suggesting that these habitats provide the best trade-off between food resources (Barberena-Arias & Aide, 2003;Williams et al., 2008), shelter from predators (Brouwers & Newton, 2009), and favorable moisture conditions. Oviposition sites may also be more important and various in forested habitats than in more open ones. This is especially important for litter-dwelling species, explaining higher densities in closed habitats (Huber, 1989).
These results are similar to those obtained for forest-dwelling grasshoppers in Western Amazonia by Amédégnato and Descamps (1980).
Lower cricket densities in shrublands might be explained by smaller optimal vegetation patches, but also warmer mean and high temperatures, lower relative humidity, and limited food resources (Shik & Kaspari, 2010). Moreover, in drier habitats, such as shrublands, cricket density may be further constrained by higher predation rates in combination with limited shelter (McCluney & Sabo, 2009, 2014. Accumulated species richness similarly showed a pattern related to ecological succession, with the highest number of species found in forested habitats (forest and preforest). This presumably reflected greater habitat complexity and microhabitat availability (Shik & Kaspari, 2010). However, based on the number of species found at each site, the differences were not significant between ecological stages. This could mean that a larger number of sites is needed to fully describe the cricket assemblages living in forests as compared to preforests or shrublands. Although preforest and forest share a large number of cricket species, the latter is characterized by a higher percentage of unique species. This pattern is expected for rich forest ecosystems (Rahman et al., 2012) and comparable with fruit-feeding butterflies, birds, leaf litter amphibians, and the crickets documented in Brazilian forests (Barlow et al., 2007;Szinwelski et al., 2012). While habitat complexity and heterogeneity might explain higher species richness in forests (Shik & Kaspari, 2010), they were surprisingly close to preforests in their vegetation attributes; however, they were clearly different in climatic conditions (i.e., temperature and relative humidity).
These observations highlight the importance of fine vegetation attributes or microhabitats (e.g., dead trunks, woody debris, and gaps around roots) along with microclimate condition (i.e., high moisture, low variation of temperature and humidity) in explaining community differences between these two advanced stages of forest regeneration, that are forest and preforest, in the southern part of New Caledonia. Among the ten species exclusively found in the forest, eight TA B L E 3 Mean values (± standard deviation) of cricket assemblages, vegetation attributes, and climatic parameters in the three ecological stages calculated from values averaged per sites. P represents the significance of the values after post hoc Kruskal-Wallis tests according to Conover for pairwise multiple comparisons of the ranked data between stages Fo-Sh***, Fo-Pr**, Pr-Sh** *p < .05; **p < .01; ***p < .001; Fo signifies Forest, Pr Preforest, and Sh Shrubland.
were found on vegetation strata, interestingly suggesting a close relationship to vegetation attributes (i.e., plant species or structure), and likely further interrelated with the older age of forest habitat.
Some cricket species were previously described as potential indicator of an ecological stage such as Pseudotrigonidium caledonica, previously described as inhabiting forest understory and singing on low vegetation with a unique and recognizable low-frequency song .
However, the statistical analyses presented in this paper did not reveal Pseudotrigonidium caledonica (Otte, 1987) to be a significant Scatter diagrams (coefficient for the inertia ellipse size = 1.5) were drawn according to the habitat type. Significant fitted environmental variables are shown with the following abbreviations: HER for herbaceous layer, SHR for shrub layer, TRE for tree layer, BAR for bare ground, VH for vegetation height, DN for number of stems, DBH for number and diameter of stems, CO for canopy openness, MT for mean temperature, HT for maximum temperature, DT for daily temperature variation, MH for mean humidity, LH for minimum humidity, and DH for daily humidity variation.

Forest Shrubland
Preforest TA B L E 4 Contribution percentage of the most influential species for each pair of habitats calculated based on species abundances, with cumulative percentages in parentheses. The most influential species have with a minimum cumulative percentage of 70% of the contributions indicator of the forest. This is due to the fact that this species was collected only once in our survey, which highlights an important bias of our analyses towards rare species.
Previously, Desutter-Grandcolas (1997) have described the cricket community in New Caledonia forested habitat for different soil types (volcano-sedimentary rocks) than those studied here (ultramafic soil). Their cricket community structure was described as close to our forest community (in terms of species richness and dominant species), with 23 species identified belonging to the same genera, but all different at the species level.
Taken together, the studies seem to confirm cricket genera specialization for habitats of a given vegetation type and plant structure  and the microendemism pattern of Orthoptera species in New Caledonia (Grandcolas et al., 2008).

| Implications for conservation
The findings presented here provide strong evidence that cricket species and their assemblages are sensitive indicators of ecological succession in New Caledonia, with potential for applications in conservation. There are several characteristics that make them highly suitable as bioindicators for terrestrial monitoring in New Caledonia. In combination with the bioindicator evaluation by Gerlach et al. (2013), New Caledonian crickets are found in high densities throughout the year, can be easily sampled by sight or hearing, are sensitive to ecological stresses, and have a thorough taxonomic understanding including various ranges of bioclimatic preferences (Desutter-Grandcolas, 1997;Desutter-grandcolas et al., 2016;Otte et al., 1987). Available data on assemblages of crickets in scientific literature tend to show that crickets could be good bioindicators in  (Gerlach et al., 2013).
All the ecological stages considered in this study are important for conservation as each harbors a distinctive assemblage of cricket species and their destruction would lead to irreversible loss of unique biodiversity in New Caledonia. This provides valuable insight for managers and decision-makers and further highlights the large biodiversity of New Caledonian forests, with numerous endemic genera and species.
Open shrublands are usually less studied for their fauna (Chazeau, 1993)

| Recommendations and perspectives for passive acoustic monitoring
The sampling design is highly important when studying crickets.
Previous research on the response of cricket communities to vegetation succession has shown contrary or inconclusive results; however, this could be due to a lack of taxonomic expertise, inadequate sampling methodology, and/or the type of habitat (nontropical areas or grasslands for example are usually poor in cricket species). For example, while Araújo et al. (2015) found no relationship between habitat degradation (from highly degraded to natural forested habitats) and cricket abundances, the passive trapping system (i.e., pitfall trap) used was not adequate to sample cricket communities (Gardiner et al., 2005), potentially leading to bias in the results. Because numerous cricket species live on vegetation, rocks, or tree trunks, with restricted displacements on leaf litter (Desutter-Grandcolas, 1992, 2002 (Desutter, 1990 (Gasc et al., 2018). Thereby, Gasc et al. (2018) concluded that only 6 days of passive acoustic recordings were necessary to describe the calling cricket community. Fourthly, calling cricket communities reliable bioindicators of the surrounding environment and their analyses can reveal major environmental perturbation.
In the present work, community analysis was performed only with stridulating crickets revealing the same succession-based specialization as compared to the analyses using all species. As shown by Gasc et al. (2018), the presence of the invasive ant Wasmannia auropunctata resulted in a significant reduction in cricket species and their acoustic activity.
In contrast with other methods, such as fogging or traditional taxonomic inventories, the ecoacoustic approach is advantageous as it is noninvasive, easy to implement, allow data collection over large temporal and spatial scales, and in remote places (Sueur & Farina, 2015). An acoustic approach for biodiversity monitoring is meaningful in species-rich ecosystems, such as tropical rainforests (Deichmann et al., 2018), where taxonomic impediments may be overcome by the recognizable taxonomic unit of songs (Riede, 1998(Riede, , 2018. Acoustic monitoring could not, however, be used to characterize the sampled communities by their specific components unless a preliminary taxonomic effort was conducted. When possible, taxonomic inventory to establish ground knowledge combined with passive acoustic monitoring to sample natural environments seems an effective strategy, and especially necessary in the tropics where biodiversity still needs description. The research effort in ecoacoustics conducted the last decade in New Caledonia within the same research program, including the present work, brought together sufficient taxonomic, ecoacoustic, and ecological knowledge to confirm the value of crickets as bioindicators and support the use of a passive acoustic approach to monitor ecosystems Gasc et al., 2013Gasc et al., , 2018.
Because crickets are globally distributed, particularly in endangered biodiversity hotspots and in the tropics, they can be a key tool and metric for conservation programs.

ACK N OWLED G M ENTS
We would like to thank Frédéric Rigault (IRD Noumea), Jean-Yves Meunier (IRD Nouméa), Léa Colmas, and Alexia Omont (Master students) who provided help with field work, habitat characterization (FR) and mapping our sampled localities. We are also grateful to Eric Vidal (head of IRD Nouméa) for his support, Alexandre Millon for helpful comments on our statistical analysis, and Garett Pignotti for providing thoughtful comments on the entire paper and English edits. We also thank Thomas Ibanez who provided helpful insight into community analysis. We thank the New Caledonian Direction

CO N FLI C T O F I NTE R E S T
The corresponding author confirms on behalf of all authors that there have been no involvements that might raise the question of bias in the work reported or in the conclusions, implications, or opinions stated.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data and code that support the findings of this study are openly available in the Dryad Digital Repository: https://doi.org/10.5061/ dryad.2280g b5w3