Introduction

Protected areas (PAs) play a key role in the efforts to preserve both local and global biodiversity, particularly when it comes to threatened species, habitats and ecosystems (Mittermeier et al. 2005; Ferreira et al. 2014; Azevedo-Santos et al. 2017; Peralta et al. 2019). The changing land use and the expansion of PAs have amplified the interaction between protected and unprotected areas worldwide (Watson et al. 2014) and have become central to the conservation of biodiversity and ecosystem services (Blanco et al. 2020). Therefore, it is increasingly vital to know the importance of protected areas and their surroundings for the conservation of local assemblages and for regional diversity (Blanco et al. 2020).

In Brazil, PAs cover more than 1.590 billion square kilometers, i.e., about 18.69% of the territory (MMA 2021). They are unevenly distributed across biomes, and still poorly studied with regard to their biodiversity (Oliveira et al. 2017; Fonseca and Venticinque 2018; Leal et al. 2020). The lack of data and the limited number of areas in certain regions create uncertainties about the effectiveness of PAs and the ways in which they interact with surrounding areas to preserve the country's biodiversity (Fonseca and Venticinque 2018).

Among the Brazilian forest biomes, the Atlantic Forest is considered a biodiversity hotspot (Mittermeier et al. 2011). This biome still suffers from consecutive impacts caused by land use changes linked to the cultivation of coffee, sugar cane, soy, corn and pasture areas, and also a large human population, forming the largest urban and industrial centres in the country, has caused a drastic reduction in native vegetation (Mello et al. 2020). The result is a heavily fragmented and anthropized landscape with only around 12% original vegetation (S.O.S Mata Atlântica 2018; MapBiomas 2021); most of which is located inside PAs. Another important issue to consider is that the creation of PAs in Brazil has been strongly focused on terrestrial biota and threats to them rather than freshwater systems and species (Fagundes et al. 2016; Samways 2017; Frederico et al. 2018; Reid et al. 2018; Basset and Lamarre 2019; Leal et al. 2020).

Invertebrates are often overlooked when it comes to conservation (Mammola et al. 2020), despite the fact that they constitute the vast majority of animal species, and that they are among the groups that suffer most from anthropogenic disturbances (Cardoso et al. 2020; Sundar et al. 2020). As the knowledge about the importance of these organisms to the ecosystems and the human population is still limited, they are frequently excluded or overlooked from conservation programs (Cardoso et al. 2020; Samways et al. 2020; Mammola et al. 2020). There are some rare exceptions, such as the State Wildlife Refuge Dragonflies of Serra de São José, explicitly created to protect dragonfly species in the Atlantic Forest (Bede et al. 2015), but there are very few protected areas designed to preserve aquatic insects in biodiversity hotpots around the world (Sundar et al. 2020). Current studies have shown that the PAs are not very efficient in protecting the biodiversity of freshwater (Fagundes et al. 2016; Frederico et al. 2018; Leal et al. 2020). However, few studies consider invertebrates (Fagundes et al. 2016; Frederico et al. 2018; Leal et al. 2020). This causes concern, as we still do not know how important PAs and surrounding areas are to the local and regional invertebrate diversity, and how they interact with each other.

Among freshwater aquatic invertebrates, dragonflies (Odonata) have been widely used as indicators of environmental health (Oliveira-Junior and Juen 2019; Voster et al. 2020; Ribeiro et al. 2021a, b). They are dependent on both aquatic and terrestrial ecosystems, and changes in these ecosystems are likely to affect the occurrence and reproduction of adults and the development of larvae. These modifications in turn affect the capacity for thermoregulation, choice of partners, availability of oviposition sites, selecting different groups of species in modified sites (De Marco et al. 2015; Oliveira Junior et al. 2015; Valente-Neto et al. 2016; Rodrigues et al. 2018). In the neotropical region the odonates are divided into two suborders, Zygoptera and Anisoptera (De Marco et al. 2015). In general, the Zygopterans are, considered to be more sensitive than anisopterans to landscape changes (from forest to open-areas) and aquatic habitat modification (as lotic for lentic and physicochemical water changes) due to their lower dispersal capacity and greater dependence on the environment for thermoregulation and suitable habitats for oviposition and development of larvae (De Marco et al. 2015; Carvalho et al. 2018; Rodrigues et al. 2018).

Human activities causing loss of forested habitats increase the local extinction risk dragonflies considered forest specialists and facilitate the colonization of generalist species (Rodrigues et al. 2016, 2018; Carvalho et al. 2018; Oliveira-Junior and Juen 2019), which lead to drastic changes in the local diversity. In the Atlantic Forest, the majority of areas surrounding PAs are anthropized landscapes where the natural forest cover is reduced, and the PAs are likely to be the only refuges available to some forest specialist species, particularly zygopterans (Ribeiro et al. 2021a, b).

Among the metrics used to assess biodiversity changes of Odonata, the richness, abundance and composition have been widely used to assess physical changes in channels and loss of surrounding vegetation in aquatic ecosystems. (Oliveira-Junior and Juen 2019; Ribeiro et al. 2021a, b). Another tool to detect diversity patterns are beta diversity analysis, also used to measure the impacts caused on the diversity between environments (Quintana et al. 2017; Altermatt and Holyoak 2012). Recently, Legendre and De Cáceres (2013) developed a single-number estimate of beta diversity through the calculation of the total variance of the species composition. This method allows the assessment of two other measures: species contributions to beta diversity (SCDB) and local contributions to beta diversity (LCBD) (Legendre and De Cáceres 2013). And can be used to understand drivers of beta diversity (e.g., Landeiro et al. 2018) which may indicate site’s conservation value (Legendre and De Cáceres 2013).

Based on this general expectation, we evaluated how the communities changed comparing a protected area and its surroundings on the Odonata assemblages in the Atlantic Forest. The structure (composition and richness) of the Odonata assemblages in streams within and around the PA was assessed. In addition, we assessed the contribution of each site to beta diversity (LCBD) and species contribution to beta diversity (SCDB) in streams within and outside the PA. We also checked if any of the species present could be regarded as bioindicators of pristine forest areas.

As a first prediction, we expected to find a higher richness of zygopterans at the sites within the PA, and a higher richness of anisopterans in the surrounding areas due to the PA maintaining more forested areas and greater physical integrity of the channels when compared to the surrounding streams. (De Marco et al. 2015; Rodrigues et al. 2016, 2018; Oliveira-Junior and Juen 2019; Ribeiro et al. 2021a, b). As the second prediction we expected a difference in the composition of the species assemblages, with the protected areas maintaining a greater diversity of Zygoptera and the surrounding areas a greater diversity of Anisoptera. As the third prediction, we expected that the PA maintain important sites (LCBD) for the regional beta diversity as well as maintaining species that contribute to the increase of regional beta diversity (SCDB); the PA housing the majority of species regarded as forest specialists, i.e. species dependent on forested areas for their survival (Carvalho et al. 2018). As the fourth prediction we expected that the PA may harbour a selected group of species that might be considered bioindicators of preserved areas (Rodrigues et al. 2019).

Materials and methods

Study area

The study was carried out in and around the Private Reserve of Natural Heritage (RPPN) Veracel Station, located in the southern part of the state of Bahia, between the municipalities of Santa Cruz Cabrália and Porto Seguro (16°23ʹ27ʺ S and 39°10ʹ28ʺ W). The RPPN Veracel Station has an area of 6069 ha. It is considered the largest private reserve in North-Eastern Brazil, and the second largest in the Atlantic Forest Biome. The area maintains primary vegetation with high biological diversity (RPPN 2016). The region’s climate is Af according to the Koppen classification, characterized by average annual temperature of 18 °C or higher, with very little annual temperature variation. And with significant precipitation, average rainfall of at least 60 mm each month, and high humidity.

All the streams, both within and outside the PA, were considered small and narrow (estimated from first to second order). The streams inside the PAs had the width varying between 0.78 cm the 5.01 cm (mean 2.76 cm and sd 1.35 cm) and depth varying between 0.08 cm the 0.35 cm (mean 0.28 cm and sd 0.12 cm). And the surrounding streams with width varying between 0.74 cm the 5.40 cm (mean 2.65 cm and sd 1.28 cm) of and depth varying between 0.12 cm the 0.63 cm (mean 0.19 cm and sd 0.15 cm). The streams sampled inside the RPPN were surrounded by a more closed and well-preserved canopy, formed by primary vegetation and without evidence of visual pollution, erosion or anthropogenic physical changes along the stream channel. The areas surrounding the reserve were composed of pasture and agriculture matrices with settlement areas and large rural properties. The streams generally did not have uniform riparian vegetation, and a larger proportion of the sites were more open to direct sunlight. At some sampling sites, the stream channel was interrupted by constructed dams for watering animals and/or other use within the property. We also observed a small amount of domestic waste and plastic bags.

Sampling sites

Sampling was carried out at 40 sites, 22 within the Veracel Station Conservation Unit and 18 sites surrounding the RPPN. Each site was sampled twice (September 2018 and February 2019), i.e., during both the rainiest and the driest season, in order to obtain a more comprehensive representation of the annual odonate assemblages (Fig. 1).

Fig. 1
figure 1

Map of the RPPN Veracel Station and its surroundings, located in the municipality of Porto Seguro, BA. Black dots show the sampling sites inside the RPPN, and grey dots show the sampling sites outside the RPPN

Full size image

Specimen collection, curation and identification

All adult Odonata specimens found were captured through active netting. For each point, a 100-m stretch along the water-line was sampled on both banks for 90 min with two collectors. The sampling was conducted on rain-free days at the time of peak odonate activity, from 9 am to 4 pm (Calvão et al. 2018). The collected specimens were placed in entomological envelopes and plastic boxes and later placed in freezers to be sacrificed. The adults were taken to the Laboratory of Aquatic Organisms (LOA) at the State University of Santa Cruz for further identification to the lowest possible taxonomic level by means of identification keys (Garrison et al. 2006; Lencione 2005, 2006, 2017). All specimens collected are deposited in the Zoological collection of Aquatic Organism Laboratory—LOA in the Santa Cruz State University, Ilhéus, Bahia, Brazil.

Statistical analyses

Species richness was calculated both for the total odonate community and separately for the suborders Zygoptera and Anisoptera. The data was analysed using a generalized linear model (GLM), using a log-linear model, with Poisson distribution (Gotelli and Ellison 2011). For this analysis, the richness (species numbers) was used as a response variable, and the two treatments (within and surrounding the RPPN) as predictor variables. A principal coordinates analysis (PCOA) of the abundance data was conducted to summarize the species composition inside and around the protected area, respectively, using the Bray–Curtis dissimilarity index (Legendre and Legendre 1998). This grouping was analysed through a permutational analysis of variance (PERMANOVA) (Anderson 2001), that was performed with 9999 replications (Anderson and Walsh 2013), to test for significant differences in species composition.

To evaluate the relation of beta diversity with the sampling sites in and around the APs, we calculated the total beta diversity (BDtotal) and we consider two measures: The local contribution to beta diversity (LCBD), to assess each site's contribution to beta diversity, and the contribution of the individual species to beta diversity (SCBD), to assess the contribution of each species to beta diversity (Legendre and De Cáceres 2013). We first applied Hellinger transformation to the community composition and then estimated BDtotal as the unbiased total sum of squares of the species composition data (Legendre and De Cáceres 2013). The relative contribution of each sampling unit to the beta diversity (LCBD) was estimated by the total variation in the species composition data (Legendre and De Cáceres 2013). To determine which taxa contributed the most to beta diversity patterns in streams within and around the RPPN, we calculated the SCBD. The species with SCBD values higher than the mean of all taxa are the species that make high contributions to beta diversity patterns (Legendre and De Cáceres 2013). The association of the local assemblies with beta diversity may indicate sites that contribute disproportionately to the regional species pool relative to species richness, and this may be particularly useful for identifying keystone sites (Legendre and De Cáceres 2013; Ruhí et al. 2017, Valente-Neto et al. 2020).

The indicator species analysis (IndVal) was calculated as proposed by De Cáceres et al. (2010). This analysis classify each specie showing the probability of them belong to a specific group of sites, in our case within the PA and its surroundings. The IndVal also enables us to evaluate two components for species considered to be indicators. Component A, also called specificity, is the probability that an indicator species belongs to one of the selected groups. Component B, also called fidelity, determines the probability for each species to be found either inside the PA or in the surroundings (De Cáceres et al. 2010). The analyses were performed using the statistical program R (R Core Team 2020) and the package indicspecies (De Caceres and Legendre 2009).

Results

We collected 722 specimens of 47 species belonging to 31 genera and eight families. Six of the species were only encountered within the PA, 29 species occurred only in the surrounding areas, and 14 species were found in both areas (Appendix 1). The species Aceratobasis cornicauda, Leptagrion macrurum, Leptagrion acutum, Forcepsioneura sancta, Heliocharis amazona and Phyllogomphoides sp. were found only at sampling sites within the protected area. The most abundant Zygoptera species were Heteragrion aurantiacum, Hetaerina rosea, Argia modesta with 202, 79 and 71 specimens, respectively. The most abundant species of Anisoptera were Erythrodiplax fusca and Erythrodiplax paraguayensis with 30 and 12 individuals, respectively (Appendix 1). Most species were represented by fewer than 10 specimens.

The total Odonata richness showed a significant difference in species number between the sites collected inside and outside the RPPN (AIC = 168.8, df = 36, p = 0.003), with an average of 4.57 species inside and 7.63 species outside the RPPN; the model explaining 21% of the variation in total richness in relation to areas, inside and outside the RPPN. The suborder Anisoptera also showed significant differences (AIC = 137.7, df = 36, p = p < 0.0001) between the sampled areas, with an average of one species inside and four species outside the RPPN. The model explained 31% of the variation in Anisoptera richness in relation to areas, inside and outside the RPPN. There was no significant difference in the richness of zygopterans between areas (p = 0.91), with an average of 3.5 species for the sites within the RPPN and 4.1 species in the sampling sites outside it (Fig. 2).

Fig. 2
figure 2

Average total richness of Odonata (in grey) (t =  − 3.53; p = 0.001), of Anisoptera (in black) (t =  − 3.75 and p = 0.001) and of Zygoptera (in red) (t =  − 1.25, p = 0.91), in streams within and outside the protected area (error bars indicate standard deviation)

Full size image

With regard to species composition, the assemblages found within and outside the PA differed significantly from each other (PERMANOVA test, p = 0.001 and R = 0.5018) (Fig. 3). The assemblages within the protected area were more similar to each other, suggesting less variation in their species composition. The compositions of the assemblages outside the RPPN were much more varied, indicating a greater variation in species composition between the sites outside the RPPN (Fig. 3). The species composition of the three sites outside the protected area (F01, FO3 and F07) was very similar to that of the sites within the protected area (Fig. 3).

Fig. 3
figure 3

Graph of the PCOa analysis showing the similarity in composition of the adult Odonata assemblages between the areas inside and outside the RPPN. Black dots denote areas within the protected area, and in grey dots denote the surrounding areas. Site codes shown in Appendix 2

Full size image

In relation to LCBD, eight sites had a significant contribution to the local beta diversity (p < 0.05; LCBD ≥ 0.04). They were all situated outside the protected area (Appendix 2). When we estimated the SCDB, 24 species showed higher contribution values to beta diversity regionally. Of these, six species were associated with streams located within the RPPN (H. amazona, L. macrurum, L. acutum, H. aurantiacum, H. longipes, E. cannacrioides). The first three were species found exclusively within the preserved area (Appendix 1).

Of the forty-nine species collected, seven qualified as indicators of either protected or non-protected areas. Three species were considered to be indicators of streams within the PA and four were indicative of the areas surrounding the PA (Table 1). The values of the specificity (A) showed a variation from 0.71 to 1.0, and the selected species displayed a high specificity to the sampling sites (inside and outside the PA, respectively). Heteragrion aurantiacum, Hetaerina longipes and Helicharis amazona were almost exclusively collected at the sampling sites within the PA and Ischnura capreolus, Telebasis corallina, Erythrodiplax fusca and Erythrodiplax paraguayensis at the sites outside the PA. Regarding the fidelity (B), the values ranged from 0.26 to 0.84, emphasizing that Heteragrion aurantiacum (B = 0.84) was present at most of the sites within the PA. All other species had a lower fidelity, meaning a low number of records within the sites sampled inside the RPPN as well as in the surrounding areas (Table 1).

Table 1 Result of IndVal test showing the indicator species for sites inside and outside the protected area, respectively
Full size table

Discussion

Our results highlight the importance of protected areas in the preservation of Odonata species that are forest specialists sensitive to change in natural environments, i.e., species dependent on pristine aquatic environments (Carvalho et al. 2018; Calvão et al. 2018; Rodrigues et al. 2018). Two examples from our study were the phytothelmatous species Leptagrion macrurum and L. acutum, which were found exclusively within the PA. L. acutum is categorized as critically endangered in the Red List of endangered species in Brazil (IUCN 2014, ICMBio 2018, Ribeiro et al. 2021b). Undisturbed sites inside the PA are therefore critical to the conservation of such threatened species.

According to our first prediction a higher species richness of the suborder Anisoptera was found in the surrounding area, compared to within the protected area. This relationship has been observed in other studies on Odonata in areas experiencing urbanization impacts (Monteiro Junior et al. 2013; Rodrigues et al. 2019), as well as in areas of intensive livestock grazing and agriculture (Carvalho et al. 2018; Calvão et al. 2018). It can be attributed to different degrees of change in the aquatic environments and their surrounding habitats, e.g., loss of riparian vegetation, siltation reducing the flow in channels from lotic to lentic, and others. These changes facilitate colonization of Anisoptera species that are usually more generalist and/or adapted to lentic and altered environments (Juen and De Marco 2012; Monteiro-Junior et al. 2013; Calvão et al. 2018; Rodrigues et al. 2018; Carvalho et al. 2018; Rodrigues et al. 2019), although no lentic habitats were present in our study area.

Contrary to our first prediction, the species diversity of adult zygopterans was similar among the sites sampled within the RPPN. Changes in the surrounding areas lead to the loss of riparian vegetation and changes in the flow of channels from lotic to lentic due antropic impacts. These changes may attract species adapted to these more lentic habitats. For example, Acanthagrion cuyabae, Acanthagrion gracille, Telagrion longum, Telebasis corallina, Neoneura sylvatica, Nehalennia minuta, Lestes forficula and Lestes tricolor were only collected outside the PA in this study. All of them are commonly found in lentic environments or open areas (Monteiro Junior et al. 2013; Rodrigues et al. 2016; Vilela and Ferreira 2016).

Corroborating our second prediction, the assemblages found within and outside the PA were different, with regard to species composition, the assemblages within the PA did not vary as much as the assemblages in the surrounding areas, indicating that protected areas maintain a more stable community than unprotected areas. Unprotected areas are subject to disturbances of varying magnitude (as flow modification, siltation, loss of water quality, change in physicochemical variables and flooding), enabling species with different eco-physiological requirements to colonize thiscertain sites (De Marco et al. 2015; Rodrigues et al. 2016, 2018). Previous studies on Odonata have already shown that species composition is altered as a result of changes in natural environments, possibly due to the introduction of one or more generalist species that may out-compete the specialist species. This pattern seems to be more evident in Zygoptera (Juen and De Marco 2012; Monteiro-Junior et al. 2013; Rodrigues et al. 2016, 2018; Carvalho et al. 2018; Rodrigues et al. 2019), and our study points in the same direction.

Corroborating in part our third prediction, our results show that sites within the RPPN have low LCDB values when compared to sites surrounding the RPPN. This is closely linked to the greater species richness found in the areas surrounding the RPPN. However, the SCDB analysis did, demonstrate that some of the species with a very high SCDB value, e.g., E. cannacrioides, H. amazona, L. acutum, L. macrurum, occurred almost exclusively in streams within the RPPN. The species L. acutum, L. macrurum are phytotelmatous, and tightly linked to forested areas (Carvalho et al. 2018; Rodrigues et al. 2019; Furieri et al. 2020; Ribeiro et al. 2021a, b). The species with high SCDB values which were only collected in streams surrounding the RPPN, e.g., Diastatops obscura, E. fusca, E. paraguayensis, E. umbrata, I. capreolus and T. corallina, are either open area specialists or habitat generalists (Rodrigues et al. 2018, 2019; Carvalho et al. 2018). Another relevant finding is that even though the difference in Zygoptera diversity between the protected area and its surroundings was not statistically significant, five of the six species contributing most to SCDB and registered within the RPPN belong to the suborder Zygoptera, emphasizing that PAs often harbour some exclusive species, especially from this suborder.

In accordance with to our fourth prediction, among the species tested as bioindicators, three species (H. aurantiacum, H. longipes and H. amazona) showed a preference for forested areas. The adults of these species favour more shaded places with a greater environmental integrity between the surroundings and the stream channel (Garisson et al. 2010, Ferreira-Puriquetti and De Marco 2002; Borges et al. 2019). Some studies have shown that their larvae are found in un-altered permanent streams (Garrison et al. 2010). Among the species suggested as indicators of the areas surrounding the PA, T. corallina and I. capreolus are generally associated with more open and lentic environments (Rodrigues et al. 2016, 2019).

To conclude, our results show that the composition of the odonate species assemblages may provide a means to assess the importance of protected areas to Odonata communities. Streams located within the RPPN harboured species with a high SCDB value and were good candidates to be keystone communities, contributing to the regional species pool. Our study highlights the importance of PAs to the maintenance of the regional species pool, to forest specialist species and to threatened species. Management strategies are needed to maintain the disproportionally great contribution of sites located within protected areas to the regional species pool. We particularly call attention to the conservation of the protected areas to provide the ecological conditions required by forest specialists, such as big trees, high humid conditions, stable ambient temperature, greater integrity of habitats and availability of sites for reproduction and development of larvae. In fact, the conservation of these sites inside protected areas play a vital role not only in the preservation of forest specialist dragonflies, but also many other taxonomic groups with similar habitat requirements (Fernández et al. 2004, 2012; Gray et al. 2016). These sites also provide other ecological services, such as carbon storage, maintenance of soil and water quality and decomposition among others (DeFries et al. 2010; Palomo et al. 2014; Cumming 2016; Maestre-Andrés et al. 2016). The conservation strategies for dragonflies in the Atlantic Forest could therefore be linked with comprehensive strategies for forest conservation and restoration. Assisting in the conservation of the Atlantic Forest’s biodiversity.