Tropical forests are the most important reservoirs of terrestrial biodiversity around the world. Despite there being many studies in these ecosystems,most of them focus on tropical wet forests (TWF), with less attention to the tropical dry forests (Mooney et al. 1995). Tropical dry forests (abbreviated herein as TDF) are seasonal ecosystems in which precipitation occurs only during a portion of the year and the decrease in rainfall causes changes in the patterns of greenish/senescence of the forest, resulting in a reduction of productivity and availability of resources for animals and, consequently, in the biological activity in the forest during the dry season (Martínez-Hernández et al. 2019). TDF are the tropical community with the greatest extension in Mexico, stretching along the Pacific coast from Sonora south to Chiapas in the Mexico-Guatemala border, with some intrusions in Central Mexico, northern Veracruz and the Yucatan Peninsula (Búrquez and Martnez-Yrízar 2010, Trejo 2010). The relevance of the TDF has been recognised by many authors as very rich ecosystems with an abundance of endemic animal and plant species, which are, however, at great risk due to deforestation and human activities (e.g. cattle ranching and agriculture, particularly in Mesoamerica, (Banda 2016, Miles et al. 2006, Sánchez-Azofeifa et al. 2005). Therefore, documenting the biodiversity of the TDF is an urgent task for the proper management and conservation of this ecosystem.

Seasonal fluctuations of richness and abundance of insects inhabiting tropical forests has been documented elsewhere (e.g. Kannagi et al. 2016, Kishimoto-Yamada and Itioka 2015, Wolda 1988). In Mexico, it has been monitored in some detail at localities such as the Chamela-Cuixmala Biosphere Reserve in Jalisco State (García-Aldrete and Ayala-Barajas 2004), but such studies are scarce for other sites. With the aim of filling this gap, in 1997, a group of entomologists from the Institute of Biology of the National Autonomous University of Mexico (UNAM) started a long-term project to document the richness and distributional patterns of selected groups of insects associated with the TDF in Mexico. Most of the localities studied are distributed along the Pacific slope from Sonora to Oaxaca, with a few studies in south-central Mexico (e.g. González-Soriano et al. 2008, González-Soriano et al. 2009, Noguera et al. 2002, Noguera et al. 2007, Noguera et al. 2009, Noguera-Martínez et al. 2012, Noguera et al. 2017Zaragoza-Caballero et al. 2003).

Odonata was included in that project as a “test group”, with the aim of comparing if their local assemblages vary in a similar way as those exhibited by other insect groups that are more directly associated with plants (e.g. Cerambycidae, Lampyridae, Syrphidae and Apidae). Dragonflies are predaceous insects that do not depend directly on plants from a trophic point of view, but forests can provide certain requirements for the adults, such as optimal microclimates for effective thermoregulation, conditions for an optimal foraging and a provision for nocturnal roosting or daytime shelter from both inclement weather and predators (Corbet 1999, Corbet et al. 2006, Paulson 2006). Vegetation can also be used as mating areas and feeding perches (Buchwald 1992).

To date, only a few studies documenting the patterns of diversity and abundance of Odonata have been done in the Mexican TDF: one at the Chamela Biosphere Reserve in Jalisco (González-Soriano et al. 2004) and another two at the Sierra de Huautla in Morelos and the Sierra de San Javier in Sonora (González-Soriano et al. 2009, González-Soriano et al. 2008). Additionally, Novelo-Gutiérrez & Gómez-Anaya (Novelo-Gutiérrez and Gómez-Anaya 2008) included a couple of sites with TDF (Pinolapa and Aguililla) in their study on the altitudinal distribution of Odonata assemblages in the Sierra de Coalcoman in Michoacan.

We are presenting here the results of a study on the patterns of richness, diversity and abundance of an Odonata assemblage in Santiago Dominguillo, Oaxaca, Mexico, which was originally carried out between November 1997 and October 1998.

The study area was the vicinity of the town of Santiago Dominguillo (referred to herein as Dominguillo), in the north-eastern part of the State of Oaxaca (17.6484, -96.91117, 760 m a.s.l.), south of the Tehuacan-Cuicatlan Biosphere Reserve. The area belongs to what is known as the Floristic Province of the Tehuacan-Cuicatlan Valley, which is considered part of the xerophytic Mexican region (Rzedowski 1991).

The climate is semi-warm type according to the Köppen climate classification modified by García (García 1988). Average annual precipitation is 521.5 mm and air temperature is 25.2°C (Jaramillo-Luque and González-Medrano 1983). The dominant vegetation in the area is TDF and Lysiloma divaricatum (Jacq.), Bursera aptera Ramírez, B. morelensis Ramírez, B. schlechtendalii Engl, Cyrtocarpa procera Kunth, Pachycereus weberi (J.M. Coult.) Backeb, Escontria chiotilla (F.A.C.Weber ex K.Schum.) Rose and Ceiba parvifolia Rose (Jaramillo-Luque and González-Medrano 1983, Ochoa 2001) are the dominant trees. Gallery forest is also present along rivers, with trees as tall as those found in contiguous TDF. Some of the flat areas in the region have been converted into pasture lands and mango plantations.

The area belongs to the hydrological region No. 28, which corresponds to the Papaloapan basin, a river draining into the Gulf of Mexico and the site corresponds to the Rio Grande sub-basin, which enters the Tehuacan-Cuicatlan Reserve from the south.

Sampling methods and regimes. Most samplings were done around the Rio Las Vueltas (also known as Rio de las Vueltas) and other minor tributaries around the Cuicatlan-Dominguillo vicinity, including the towns of Santiago Dominguillo (N 17.6484, W -96.91117) and San Pedrito Chicozapote (N 17.77218, W -96.9445), with an occasional collection also in Rio Grande, Presa Derivadora Matambo (N 17.6484, W -96.9117). Fieldwork was carried out monthly from November 1997 to October 1998. Samplings were done for five days every month, from 09:00 to 15:00 h (10:00 to 16:00 h during daylight savings time).

Diversity analysis. We analysed the diversity of the Odonata assemblage of Santiago Dominguillo and its temporal pattern. The assemblage structure was analysed through the abundance (number of individuals), richness (number of species observed, ⁰D), Shannon diversity (exponential of Shannon Index, ¹D) and Simpson diversity (inverse of Simpson Index, ²D) (Chao et al. 2014Jost 2006). The unit of measurement for ⁰D and ¹D is the number of effective species or Hill numbers: ¹D indicates the effective number of species equally abundant within an assemblage or community, while ²D is the effective number of the most abundant or dominant species equally abundant. These metrics are mere modifications of the indices used in previous works to analyse diversity (e.g. González-Soriano et al. 2008) and we used them in order to obtain direct measures of diversity that can be interpreted from a biological perspective (number of effective species instead of bits or nats) (Cultid-Medina and Escobar 2019).

The maximum expected richness value of diversity was estimated by non-parametric abundance, based Chao 1-bias corrected estimator (Chao 2005), as well as the expected diversity for ¹D and ²D using the estimators proposed by Magurran (Magurran 1988) and Chao et al. (Chao et al. 2013). All these calculations were done using the Spade R package (Chao et al. 2015). We also used the iNext package (Hsieh et al. 2016) to obtain the cumulative species curve for the whole Odonata assemblage from the Dominguillo Region, through the interpolation-extrapolation method proposed by Chao et al. (Chao et al. 2014).

To evaluate temporal diversity patterns for the Dominguillo Odonata assemblage, we performed the same diversity metrics described above (abundance, ⁰D, ¹D and ²D), based on the monthly information occurrence of collected odonates. Additionally, we evaluated the monthly phylogenetic divergence using the taxonomic distinctness (Δ*) and average taxonomic distinctness (Δ+) indices based on the abundance and incidence of the species, respectively (Clarke and Warwick 1998, Clarke and Warwick 1999, Clarke and Warwick 2001Warwick and Clarke 1995), which were calculated using the taxondive and taxa2dist functions of the Vegan R package (Oksanen et al. 2015). These measurements indicate that the higher value of Δ* and/or Δ+, the greater phylogenetic distance amongst individuals (Δ*) and species (Δ+) within an assemblage.

The taxonomic distinct is a measure to evaluate the phylogenetic divergence within the communities or assemblages according to the topology of their taxonomic hierarchy and it analyses the pattern of the phylogenetic relationships amongst taxa obtained from a sample or a complete assemblage, i.e. how closely related the species are or how evenly distributed are their evolutionary paths through the taxonomic hierarchy (Clarke and Warwick 1998, Clarke and Warwick 2001). Here, we included the Odonata taxonomic hierarchy: order, superfamily, subfamily, tribe, genus and species. The supraspecific taxonomic arrangement was based on Carle et al. 2015, Davies and Tobin 1985, Dijkstra et al. 2014.

To evaluate if the Dominguillo odonate diversity is affected by or related to abiotic factors, Pearson’s correlation analyses were done between the monthly species diversity (abundance, ⁰D, ¹D and ²D), the monthly phylogenetic diversity (Δ*, Δ+) and the average rainfall and temperature recorded in Dominguillo during the sampling time. All the analyses were done using Past software (Hammer et al. 2001). We also generated a heatmap with gplots v. 3.0.4 package of R (Warnes et al. 2012) in order to display the presence and abundance that each species exhibited monthly.

Monthly diversity and phylogenetic divergence analyses allowed us to evaluate how the species diversity and the taxonomic assemblage structure were related to monthly changes of temperature and humidity. In other words, those analyses allowed us to evaluate how the abiotic factors can be associated with the temporal structure of the Odonata community of the TDF.

For the phenological analyses, we considered the period from June-July to September as the rainy season and October to May-June as the dry season. This categorisation was based in the occurrence of individual events of rainfall higher than 15 mm, since events with lower rainfall are intercepted by the canopy (Cervantes 1988). Rainfall values correspond to the climatological norms of the Mexican National Meteorological Service and were obtained from the nearest climatological station of Dominguillo, Oaxaca (CONAGUA (Comisión Nacional del Agua) 2020).

All the material collected was deposited at the CNIN (Colección Nacional de Insectos del Instituto de Biología), UNAM, Mexico City.

List of Odonata species registered from Santiago Dominguillo, Oaxaca, Mexico, including phenology data and number of individuals collected (in parenthesis). Additional information in Suppl. materials 1, 2 and photographs of some species in Fig. 1.

Lestidae

Archilestes grandis (Rambur, 1842). Jun (1). (Fig. 1)

Platystictidae

Palaemnema domina Calvert, 1903. Jun (13), Jul (2), Aug (6).

Calopterygidae

Hetaerina americana (Fabricius, 1798). Nov (41), Jan (23), Mar (30), Apr (11), May (23), Jun (6), Jul (7), Aug (3), Sep (24), Oct (4). (Fig. 1).

Hetaerina occisa Hagen in Selys, 1853. May (1).

Hetaerina cruentata (Rambur, 1842). Jan (2), Mar (11), Apr (7), May (1), Jun (2), Jul (6), Aug (7), Sep (7), Oct (5).

Coenagrionidae

Apanisagrion lais (Brauer in Selys, 1876). Jul (1).

Acanthagrion quadratum Selys, 1876. Jan (1), Apr (4), May (9), Jun (6), Jul (2), Sep (2).

Argia anceps Garrison, 1996. Nov (8), Jan (9), Mar (14), Apr (7), May (17), Jun (4), Jul (7), Aug (8), Sep (19), Oct (6).

Argia extranea (Hagen, 1861). Nov (2), Jan (2), Mar (10), Apr (2), May (6), Jun (4), Jul (4), Aug (3), Sep (21), Oct (4).

Argia funcki (Selys, 1854). May (1), Jun (1), Jul (2), Aug (1).

Argia harknessi Calvert, 1899. Jan (1), Mar (5), Apr (1), May (1), Jun (3), Jul (2), Aug (2), Sep (4).

Argia immunda (Hagen, 1861). Nov (31), Jan (6), Feb (3), Mar (19), Apr (7), May (7), Jun (2), Jul (4), Aug (2), Sep (2), Oct (4).

Argia oculata Hagen in Selys, 1865. Jan (3), Mar (1), Apr (8), May (7), Jun (1), Jul (5), Aug (2), Sep (14), Oct (3). (Fig. 1).

Argia oenea Hagen in Selys, 1865. Nov (21), Jan (6), Feb (3), Mar (14), Apr (25), May (28), Jun (6), Jul (4), Aug (4), Sep (7), Oct (6).

Argia pallens Calvert, 1902. Nov (6), Jan (8), Mar (8), May (2), Jun (2), Jul (1), Aug (1), Oct (4).

Argia pipila Calvert, 1907. Aug (1).

Argia pulla Hagen in Selys, 1865. Nov (42), Jan (21), Feb (5), Mar (45), Apr (54), May (85), Jun (19), Jul (9), Aug (6), Sep (20), Oct (5). (Fig. 1).

Argia tezpi Calvert, 1902. Nov (6), Jan (8), Feb (7), Mar (29), Apr (15), May (28), Jun (9), Jul (1), Aug (2), Sep (5), Oct (2).

Enallagma novaehispaniae Calvert, 1907. Nov (2), Jan (1), Mar (4), Jun (5).

Enallagma praevarum (Hagen, 1861). Nov (6), Jan (11), Feb (2), Mar (5).

Enallagma semicirculare Selys, 1876. Mar (1).

Ischnura denticollis (Burmeister, 1839). Jan (1).

Leptobasis vacillans Hagen in Selys, 1877. Apr (1)

Telebasis salva Hagen in Selys, 1877. Nov (4), Jan (4), Feb (1), Mar (6), Apr (3), May (3), Jun (4), Aug (7).

Aeshnidae

Anax walsinghami McLachlan, 1883. Nov (1).

Gomphidae

Erpetogomphus elaps Selys, 1858. Nov (4), Jun (4), Jul (8), Aug (6), Sep (8), Oct (7).

Phyllogomphoides danieli González & Novelo, 1990. Jun (3), Jul (9), Aug (2).

Phyllogompoides suasus (Selys, 1859). Aug (3), Oct (2).

Progomphus clendoni Calvert, 1905. Jun (1), Jul (4).

Libellulidae

Brechmorhoga mendax (Hagen, 1861). Nov (1).

Brechmorhoga praecox (Hagen,1861). Nov (2), Mar (2), May (5), Jun (8), Jul (9), Aug (7), Sep (4), Oct (1).

Dythemis nigrescens Calvert, 1899. Mar (1), Jun (4), Jul (4), Aug (2).

Dythemis sterilis Hagen, 1861. Nov (7), Jan (5), Mar (8), Apr (2), May (6), Jun (15), Jul (7), Aug (1), Sep (2), Oct (1). (Fig.1e)

Erythemis plebeja (Burmeister, 1839). Jun (1).

Erythemis vesiculosa (Fabricius, 1773). May (2), Jun (1).

Erythrodiplax funerea (Hagen, 1861). Apr (1), May (2), Jun (2).

Erythrodiplax fusca (Rambur, 1842). Nov (12), Jan (4), Feb (3), Mar (8), Apr (4), May (5), Jun (8), Jul (2), Aug (6).

Erythrodiplax umbrata (Linnaeus, 1758). Apr (1). (Fig. 1)

Libellula croceipennis Selys, 1868. Nov (1), Apr (1), Jun (4), Jul (7), Aug (5), Sep (1), Oct (3).

Macrothemis hemichlora (Burmeister, 1839). Nov (1), Apr (2), May (5), Jun (7), Aug (3), Sep (1).

Macrothemis inacuta Calvert, 1898. Jun (6), Jul (3).

Macrothemis pseudimitans Calvert, 1898. Nov (4), Jan (5), Feb (1), Mar (6), May (1), Jun (9), Jul (6), Aug (3), Oct (3). (Fig. 1)

Miathyria marcella (Selys in Sagra, 1857). Jun (3).

Orthemis discolor (Burmeister, 1839). Jan (1), Mar (2), May (2), Jul (1), Sep (4), Oct (2).

Orthemis ferruginea (Fabricius, 1775). Nov (5), Jan (2), Apr (1), Jun (1), Oct (3).

Paltothemis lineatipes Karsch, 1890. Nov (1), May (1), Jul (3), Aug (1), Oct (1).

Pantala flavescens (Fabricius, 1798). Nov (1), Aug (1), Oct (1).

Perithemis mooma Kirby, 1889. Mar (1), Jul (3), Aug (2).

Pseudoleon superbus (Hagen, 1861). Nov (1), Mar (1), May (1), Jun (2), Jul (2), Aug (3), Oct (1). (Fig. 1)

Tramea onusta Hagen, 1861. Nov (2).

Analysis

Dominguillo is located in the north-eastern part of the State of Oaxaca, in what is known as the Cañada Region. It is part of the Tehuacan-Cuicatlan Biosphere Reserve, an interesting xerophytic area with a large proportion of endemic animal and plant species, but with poor knowledge on its odonate fauna.

Compared to both Huautla and Chamela, the low species richness found in Dominguillo could be the result of sampling around a homogeneous aquatic habitat and in a more restricted area. The reduced number of species of important groups (e.g. Aeshnidae with only one species) also denotes this. In Dominguillo, the Rio de las Vueltas is an exposed rocky stream with a permanent water flow and few zones of lentic backwaters. In contrast, the presence of a dam in the vicinity of sampling sites in Huautla and the occurrence of remnant pools downstream from the dam during the dry season, facilitate the presence of more species associated with lentic habitats, which, in turn, contributes to the increase in the richness of the site (González-Soriano et al. 2008). In relation to Chamela, although samplings were made at intermittent periods, they were carried out during the course of several years and included a wide array of habitats, such as coastal lagoons, swamps and temporal pools, all of which are very favourable for the species of the Coenagrionidae and Libellulidae families (González-Soriano et al. 2004). In addition, Chamela is a well-preserved forest within the Chamela-Cuitzmala Biosphere Preserve. On the other hand, Dominguillo is located in an ecotone between TDF and xerophytic vegetation, which is a much drier area compared to the TDF located in Huautla and Chamela.

A positive relationship between richness, diversity and precipitation has been found in other insect groups at this and other sites (e.g. Noguera et al. 2017). For instance, in several groups of Coleoptera (e.g. Cantharidae, Cerambycidae, Lampyridae, Lycidae and Phengodidae) and Hymenoptera (e.g. Encyrtidae at some sites), the highest number of species coincides with the beginning or the middle of the rainy season (Noguera-Martínez et al. 2012, Rodríguez-Velez et al. 2009, Rodríguez-Velez et al. 2011). In contrast, in San Javier, Sonora and Huautla, Morelos, more species of Encyrtidae were recorded during the dry season (Rodríguez et al. 2010, Rodríguez-Vélez and Wooley 2005). In San Javier, there is a visible peak of odonate adults during the wet season, but richness values in Huautla fluctuate throughout the year (González-Soriano et al. 2008).

In this study, the increase in both diversity and phylogenetic divergence of the odonate assemblage is apparently related to an increase in rainfall. Our results indicate that an increase in rainfall not only influences the distribution of abundance amongst species, but also the identity and/or the type of genera and/or families (i.e. the phylogenetic divergence) that constitute those temporal assemblages. When precipitation increases during the rainy season, the taxonomic distances amongst species also increase. During the wet season, high rainfall probably increases the availability of niches and resources (biotic and abiotic), which brings forth not only a higher number of taxa, but also allows the co-existence of taxa with more diverse ecological requirements. Taxonomic distinctness has been proven to be a highly useful index to evaluate the taxonomic structure of communities and assemblages in other odonates studies (Campbell and Novelo-Gutiérrez 2007).

A close examination of the Dominguillo assemblage (Fig. 4) reveals that the monthly species composition of the community varies throughout the year. Therefore, the increase in phylogenetic divergence of the assemblage observed during the rainy season is due to the recruitment of taxa that were not present early during the dry season. For example, Archilestes grandis Rambur, 1842, Palaemnema domina Calvert, 1903, Phyllogomphoides danieli González & Novelo, 1990, P. suasus Selys, 1859 and Progomphus clendoni Calvert, 1905 are species that were observed only during the wet season. This sub-group alone results in the addition of four genera, three families and four subfamilies that were not recorded during the dry season. Additionally, Apanisagrion lais Brauer in Selys, 1876 (Coenagrionidae) and Miathyria marcella Selys in Sagra, 1857 (Libellulidae) are two species belonging to unique genera within the study area that were recorded only during the wet season. All species of the “late spring/summer” group contributed significantly to an increase in the observed phylogenetic diversity during the wet season.

At a local scale, studies in Mexico reveal that the variation in the seasonal composition of the assemblage observed in our study seems to occur also at other localities with TDF (González-Soriano et al. 2009, González-Soriano et al. 2008). For example, several species of Coenagrionidae (especially in the Argia genus) and Calopterygidae (Hetaerina) are active as adults throughout most of the year in Dominguillo, Huautla and San Javier. On the other hand, species of the Gomphidae family (except perhaps for Erpetogomphus, which appears to have a more extended flying period) and the Platystictidae family (at some sites) occurred as adults only during the wet season (see also Campbell and Novelo-Gutiérrez 2007 for another ecosystem). Kishimoto and Itioka (Kishimoto-Yamada and Itioka 2015) mentioned that more long-term studies are necessary in order to understand the causes underlying seasonality in odonates. Here, we add that more inter-site comparative studies will shed some light on understanding the causes for these variations at this and other Mexican tropical dry forests. Unfortunately, climate change predictions for the entire Tehuacán-Cuicatlán Preserve suggest increased aridity, higher temperature and lower rainfall leading to reduced river flow and increased salinity and mineralisation of the drylands streams, leading to a loss of aquatic macroinvertebrates' biodiversity (López-López et al. 2019). According to our results, we suggest that those changes in temperature and precipitation regimes will not only affect odonate abundance and species richness, but also taxonomic structures of the TDF odonate assemblages.

The odonate assemblage from Santiago Dominguillo, Oaxaca consists of six families, 26 genera and 50 species. However, our data analysis suggests higher species richness in the region and, therefore, it would be interesting to extend the field work for a longer time in order to test this prediction. Dominguillo odonates did not show a significant relationship between abundance and richness with temporal variation in precipitation, which is opposite to that shown by other TDF insect assemblages. Nonetheless, both the monthly species diversity and the monthly phylogenetic diversity did show a significant relationship with temporal variation in precipitation. No significant relationship was found with temperature. In particular, the seasonal abundance pattern was different from those found in other TDF insects and more efforts should be made to explore their causes and associated factors. The high monthly phylogenetic diversity, registered during the rainy season, indicates a high variation in the temporal taxonomic composition within the odonate assemblage — a pattern that had not been recorded before in this order — and it might be related to ecological factors, such as competition amongst species.

Odonate TDF assemblages have been poorly explored around the world, in spite of the threats to their habitats caused by diverse environmental changes and human activities (Campbell et al. 2010, Sánchez-Bayo and Wyckhuys 2019). Climate change is a factor affecting dryland streams in the semi-arid Tehuacan-Cuicatlan Biosphere Reserve, leading to a loss of aquatic macroinvertebrates diversity (López-López et al. 2019). Here, we suggest that both species richness and taxonomic structures of odonate assemblages will be affected by those changes. Therefore, a better understanding of their spatial and temporal patterns would be helpful to evaluate some of the most relevant threatening factors and to design and plan relevant conservation strategies.

To Alonso Ramirez (North Carolina State University) and Cheryl Harleston, who kindly reviewed the text and made invaluable suggestions to improve it. To Ethan Bright, Ben Price and an anonymous reviewer, who made invaluable suggestions to greatly improve the manuscript. Special thanks to Biol. Ubaldo Melo Samper (SIBA-UNIBIO, UNAM) for his great help with the database arrangement and managing and MSc Enrique Ramírez García for providing some photographs. We wish to dedicate this work to our beloved friend Martin Leonel Zurita (R.I.P.), who also helped in the organization of the references.