Abundance of Odonata in different microhabitats at an oxbow lake in the Peruvian Amazon

The relationship between Odonata and vegetation in Amazonia has been studied primarily in streams. In this study, I examined the abundance of adult Odonata in two vegetation types (shrubs and herbs) surrounding an oxbow lake in the Peruvian Amazon. Daytime visual samplings of Odonata were carried out in time blocks along transects in each habitat. Thirteen taxa were identified. Five species were similarly abundant in both habitats, three used mainly herbs, and one mainly shrubs, with no variation among time blocks. The results suggest that most Anisoptera and Zygoptera are adapted to unshaded areas of the lake. Some Odonata also were observed during sunless days with light rainfall, suggesting they are adapted to rainy conditions in tropical climate.

Vegetation provides multiple uses for Odonata by providing microclimates, oviposition substrates, perching structures, and protection from predators or unfavorable weather (Buchwald 1992). Thus, vegetation can influence Odonata species richness (Ferreira-Perruqueti and De Marco Jr. 2002), endemism, and species turnover (Bota-Sierra et al. 2021). In Amazonia, numerous studies have assessed the relationship of Odonata to habitat integrity and highlighted the importance of vegetation (e.g., Carvalho et al. 2018). However, most of this research was conducted in streams (see the supplementary material in Gómez-Tolosa et al. 2021).
The Madre de Dios region in southwestern Amazonia in Peru harbours ∼200 Odonata species (Hoffmamn 2009), which amounts to one third of the known Peruvian odonate fauna (591 species, J. Hoffmann, pers. comm.). In Manu National Park (Madre de Dios, Peru), 136 species were recorded (Louton et al. 1996). Here, I evaluated habitat use by Odonata throughout the day in an oxbow lake, which is a water body formed by cut-off river channels.
The study was carried out at Manu National Park, specifically at Cocha Cashu (11°53'18.470"S, 71°24'28.307"W), a permanent oxbow lake in the Amazonian lowland connected to the Manu River by a drainage channel (Groenendijk et al. 2019). It has a 24-ha surface and a maximum depth of 2 m (Groenendijk et al. 2019), and is in the phytoplanktondominated state (Terborgh and Davenport 2013). Its margins are populated by herbs, shrubs, and isolated trees (Groenendijk et al. 2019). VOL. 52(3) 2022: 236 -240 ACTA AMAZONICA Odonata were surveyed over 18 non-rainy days during the beginning of the rainy season, from October 23 to November 16, 2017. I classified vegetation formations as shrubby or herbaceous ( Figure 1) and established 21 5-m transects (eight in shrubs and 13 in herbs) along the southern margin of the lake (900 m of shoreline). Transects were separated by at least 20 m to minimize spatial autocorrelation (sensu Oppel 2005). Odonata adults were observed along the transects with 10×42 binoculars within three time blocks: morning (0800 -1100 h), midday (1100 -1400 h), and afternoon (1400 -1600 h). Each transect was surveyed only once during a period of 15 min in each time block. Odonata were grouped according to morphology and coloration pattern. I collected dead individuals and deposited them in the entomology collection of Museo de Historia Natural of Universidad Nacional Mayor de San Marcos (MUSM) ( Table 1). Taxa were identified by comparing photographs and collected material with reference material from the MUSM collection. Odonata activity was also punctually recorded on rainy days.
In order to relate Odonata occurrence and abundance to a vegetation type, I used a multinomial species classification method (CLAM). It classifies species as generalists or specialists using their estimated relative abundances in two habitats (Chazdon et al. 2011). I considered a specialization value of 2/3 ("supermajority rule"; Chazdon et al. 2011), and the following adapted categories: mainly present in one vegetation type, present in both types, or too rare for classification. Fisher's exact tests were used to assess whether Table 1. Abundance of adult Odonata species observed along transects in shrubs and herbs along the shores of Cocha Cashu (Manu National Park, Peru) during October and November 2017. The P-value corresponds to Fisher's exact tests comparing the abundance in herbs and shrubs in the three time blocks (value in bold indicates significant difference).  Thirteen taxa were identified (Table 1, Figures 2 and 3). One was identified only to family level (Coenagrionidae), as it was not collected nor photographed. It had a green thorax with black longitudinal markings, and black abdomen with a blue tip (likely an Ischnura). Micrathyria and Nephepeltia are similar in external appearance, differing in the antenodal vein (Garrison et al. 2006), therefore they were grouped into a single taxon. Acanthagrion individuals were not identified to species level, as several species may coexist in the same area (Tennessen 2004). These three taxa were not included in the statistical analyses.
Temperature, relative humidity and shade determine the presence of Odonata in an area (May 1979). Due to their constant exposure to the sun, herbs provide a drier and warmer microhabitat than shrubs. Half of the recorded species were observed in both vegetation types and three mainly in herbs in all time blocks, indicating that species in the lake are adapted to unshaded areas.
Zygoptera are usually associated with dense canopy cover and shade (Alves-Martins et al. 2019), but only one species, a Protoneurinae, was recorded primarily in shrubs. Members of that subfamily are expected to prefer shaded environments due to their vulnerability to desiccation and overheating (Paulson 2006). However, other protoneurine, E. humeralis, was recorded equally on herbs and shrubs. Notably, both Telebasis species differed in their vegetation preference, as has been recorded previously for different closely related species of the same genus (May 1977). Lestes species are known to adapt their body temperature through behavior (Lambret and Stoquert 2011), which explains why Lestes sp. was found in both vegetation types.
As in my results, the Anisoptera B. furcata and E. peruviana were found to prefer herbaceous areas of a lake in Brazil (De Marco et al. 2005). Perithermis lais was reported to avoid open areas (Calvão et al. 2003) and their similar abundance in herbs and shrubs in my survey could be owed to the presence of the substrate where they roost (branches; Garrison et al. 2006) in herbaceous areas. Similarly, O. pachystigma was reported  All taxa, except Erythemis and Telebasis sp., were also observed being active during overcast days with light rain performing activities such as patrolling, ovipositing and fighting. In the Neotropics, some genera do not require sunlight for their daily activity (Paulson 2006). This was the first record of Brachymesia, Epipleoneura, Lestes, Protoneura, and Telebasis having little dependence on sunlight for their activity in the Neotropics The long rainy season in the study area may explain this adaptation to sub-optimal weather conditions. Future studies should expand on these aspects of odonate activity patterns.
My results highlight the importance of vegetation heterogeneity for Odonata in Amazonian wetlands. Since only one species was found to be more associated with shrubs, it is possible that most Odonata in this area have adapted to living in unshaded habitat, unlike along streams, where Zygoptera species are usually associated with canopy cover (Carvalho et al. 2018).

ACKNOWLEDGMENTS
I thank San Diego Zoo Global Peru for providing the course "Técnicas de Campo y Ecología Tropical en la Estación Biológica de Cocha Cashu", during which this study was conducted. I also thank Roxana P. Arauco for her advice during fieldwork, Dominc A. Evangelista for his insightful comments and review of the language, and two anonymous reviewers. This study was carried out under Resolución Jefatural del Parque Nacional del Manu #12-2020-SERNANP-JEF issued by Servicio Nacional de Áreas Naturales Protegidas por el Estado-Perú.