Ecol Res (2017) 32: 693–702 DOI 10.1007/s11284-017-1490-z

O R I GI N A L A R T IC L E

Masaru Sakai • Shin-ichi Suda • Taichi Okeda Izumi Washitani

•

Identifying priority habitats and monitoring species for conservation and restoration of lentic Odonata habitats: assemblage nestedness on Amami-Oshima Island, Japan Received: 9 June 2017 / Accepted: 11 July 2017 / Published online: 4 August 2017 Ó The Ecological Society of Japan 2017

Abstract We investigated Odonata faunal and habitat characteristics (forest cover, emergent, submerged, floating-leaved and floating plant covers, pond area, NO3 , chemical oxygen demand, and presence/absence of a nonnative fish) in 10 ponds on Amami-Oshima Island. In total, 26 species of six odonate families were found, and we detected significant nestedness of species composition among the ponds (22 species in the most species-rich pond, and 8 species in the most species-poor pond). Species found only in the most species-rich ponds were: Anax nigrofasciatus nigrofasciatus, Acisoma panorpoides panorpoides, Agriocnemis famina oryzae, Rhyothemis severini, Anasiaeschna martini, Hemicordulia okinawaensis, Lyriothemis elegantissima, and Hydrobasileus croceus (hereafter referred to as the rare species). These rare species are generally known to preferentially inhabit ponds with lush emergent plants and/or to prefer cooler habitats shaded by forest cover, such as Anax nigrofasciatus nigrofasciatus, Anac. martini, He. okinawaensis, and L. elegantissima. In contrast, the common species also found in species-poor ponds were: Ischnura senegalensis, Pantala flavescens, Anax parthenope julis, Ictinogomphus pertinax, and Tramea verginia, which are known to prefer an open water surface as spawning habitat. These differences in habitat preference between the rare and common species may be among the major reasons for the significant positive effects of percent forest cover and emergent plants on Odonata species richness. These results suggest that nestedness helped identify precise habitat characteristics and rare species that should be considered for conserM. Sakai (&) Æ I. Washitani Faculty of Science and Engineering, Chuo University, 1-13-27 Kasuga, Bunkyo-ku, Tokyo 112-8551, Japan E-mail: [email protected] Tel.: +81-3-3817-7297 S. Suda Æ T. Okeda Research and Development Initiative, Chuo University, 1-13-27 Kasuga, Bunkyo-ku, Tokyo 112-8551, Japan

vation and restoration of lentic habitats on AmamiOshima Island. Keywords Biodiversity Æ Dragonfly Æ National park Æ Subtropical region Æ Wetland

Introduction Amami-Oshima Island (712 km2; 28.33°N, 129.37°E), a continental island that has been isolated from the Eurasian Continent since 1.7 Ma at the latest, is endowed with a moist subtropical climate (Mizuta 2016). About 85% of the land is covered with forests dominated by Castanopsis sieboldii and Pinus luchuensis (Kato 2000). The island harbors diverse fauna and flora characterized by a high proportion of rare and endemic species (Sugimura et al. 2003). The Castanopsis forests of the island form a particularly distinct lucidophyllous landscape and are important habitats sustaining the subtropical biodiversity. Because of its unique ecosystems, core parts of the forested area on Amami-Oshima Island were incorporated into the 34th National Park of Japan in 2017 with the aim of conserving biodiversity and an ‘‘outstanding landscape,’’ the traditional management targets of Japanese national parks. Until several decades ago, traditional small-scaled rice paddies with reservoir ponds were tilled along with river channels on Amami-Oshima Island. Agricultural lentic environments function as alternative wetland habitats for lentic flora and fauna and are therefore important components of a traditional sustainable ecosystem called ‘‘SATOYAMA’’ (Washitani 2001). However, the lentic environments on the Island dramatically disappeared when they were converted to sugarcane fields due to the governmental policy of reducing rice acreage in the 1970s. Consequently, the rice paddy area was reduced from 5024 ha in 1954 to only 88 ha in 1990 (Hagiwara 1992). The massive loss of

694

lentic habitat has resulted in a critical decrease in lentic biodiversity on Amami-Oshima Island (Karube and Kitano 2016), including species with northern or southern limits of their distributions (Kawai and Tanida 2005). In such a situation, we must develop conservation priorities based on a scientific evaluation of important habitats and species. In total, 197 species and 17 subspecies of odonates have been recorded in Japan, which is 1.7-fold greater than those in all of Europe (Sugimura et al. 1999). All species use both the aquatic and terrestrial environments throughout their life histories; 65% of the species utilize lentic environments as their habitats during the larval stage, and these formerly predominated in the natural wetlands and their surrogate agricultural still water, such as reservoir ponds and rice paddies (Sugimura et al. 1999). However, modern agricultural intensification has caused the loss and deterioration of such lentic environments, and it has been suggested that this is one of the major threats to lentic biodiversity (Washitani 2001; Naito et al. 2012). Fifty-six Odonata species are currently listed as endangered in the Japan Red Data Book. Odonata species have been broadly used as ecological quality indicators for land–water ecotones, aquatic habitat heterogeneity, and hydrological connectivity (e.g., Clark and Samways 1996; Chovanec and Waringer 2001; D’Amico et al. 2004). A nested animal assemblage structure, in which species found in species-poor sites represent subsets of a richer site (Patterson and Atmar 2000), is helpful for prioritizing valuable habitats and species to be monitored when considering animal conservation. Among assemblages featured from species presence–absence data, nestedness helps reveal specific habitat characteristics and species that should be conservation and restoration priorities (Fig. 1). Such nestedness has been reported for birds (Hansson 1998; Fischer and Lindenmayer 2005), insects (Schouten et al. 2007), vascular plants (Hansson 1998), and odonates (Kadoya et al. 2008a). In this study, lentic Odonata assemblages in irrigation and biotope ponds on Amami-Oshima Island were investigated, and we examined their nestedness to prioritize the habitat characteristics and species to monitor for purposes of the conservation and restoration of the lentic environment.

respectively. All study ponds were located in former rice paddy areas but the landscape surrounding the ponds no longer included rice paddies. Ponds P1 and P2 were located in a forested basin in Yamato village (210 m above sea level), whereas Ponds P3–P10 were located in agricultural areas of Kasari-cho, Amami city (6–45 m above sea level), which were cultivated mainly for sugarcane (Fig. 2). Ponds P1 and P2 were constructed as biotope ponds about 20 years ago, whereas Ponds P3–P10 were assumed to be irrigation ponds for previous rice paddies. Ponds P1 and P2 were close each other (within 200 m), and Ponds P3–P10 can fall within the range of a 5 kmdiameter circle (Fig. 2). Because most of Odonata species can migrate and colonize among habitats approximately within 1 km (Moriyama et al. 1990), Odonata metacommunity might be structured in the habitat groups of Ponds P1 and P2, and Ponds P3–P10. Odonate censuses Censuses of adult odonates, including dragonflies and damselflies, were conducted in 2015 and 2016 along the representative and accessible shorelines of each study pond. The censuses were performed on June 21–25 and September 7–10, 2015 and May 24–27, 2016 to cover the seasons of all phenological groups. However, because a long, heavy storm unpredictably hit Amami-Oshima Island during the census in June 2015 (total 249.5 mm at the Kasari weather station), an additional census was conducted on July 29–30, 2015 to complete all censuses for June. Each census (observe and catch) for each pond was completed in approximately 30 min in the morning and afternoon under sunny or slightly overcast weather conditions when odonates were active (e.g., flying and/or showing predatory and reproductive behaviors). The censuses were always done by the same investigators (M.S., S.S. and T.O.), and all species appeared during the censuses were recorded. Hence, the censuses enabled us to examine all potential Odonata species with the views from seasonal, diurnal, and meteorological variability, similar to Kadoya et al. (2008a; b). An odonate presence/absence matrix for further statistical analysis was made by pooling all the census data. All investigations were carried out after obtaining permission from local jurisdictional governments.

Materials and methods Study site

Habitat characteristics

This study was conducted in 10 ponds on Amami-Oshima Island, Kagoshima, Japan (Table 1; Figs. 2, 3). Amami-Oshima Island is endowed with a moist subtropical climate and is surrounded by coral reefs. Mean annual air temperature and precipitation, measured at the Kasari AmeDAS automated weather station (Amami city) from 2006 to 2015, were 21.8 °C and 2213 mm,

The percent cover of emergent, floating-leaved, floating, and submerged plants in each pond at 5% increments were also recorded because aquatic plant cover has positive and negative effects on the Odonata community, such that moderate aquatic plant cover serves as spawning habitat for adults and/or refugia for larvae from predation (Sugimura et al. 1999), whereas dense

695

cover of floating plants increases mortality due to oxygen depletion (Coetzee et al. 2014). Presence/absence data of non-native omnivorous fish (Cichlidae: Tilapia zillii and Oreochromis mossambicus) were also recorded for each pond. When the fish species were recorded at least once, it was considered to inhabit that pond. These two species are listed as non-native according to the list of the species that can threaten native ecosystems in Japan (MOE 2015). We assumed that the fish species can impact Odonata community through predation of larvae and alteration of the habitat by foraging surrounding aquatic plants because the fishes are large-sized omnivore. The NO3 and chemical oxygen demand (COD) of the pond water were measured using simplified analytical instruments (Pack Test, Kyoritsu Chemical Check Lab., Corp., Tokyo, Japan) during each census period. Three-time-measured aquatic plant cover and water quality data were averaged prior to the statistical analysis. An Amami-Oshima Island vegetation map was obtained from the website of the Biodiversity Center of

Japan, Nature Conservation Bureau, Ministry of the Environment (http://www.biodic.go.jp/trialSystem/top_ en.html). Percent forest cover within a 500-m radius of each study pond was estimated based on the map using ArcGIS 10.2.2 to evaluate the number of resting and refuge habitats for nepionic and adult odonates (Sugimura et al. 1999). The areas of the study ponds (m2) were also calculated based on the map.

Statistical analysis We tested the nestedness of the odonate presence/absence matrix using the binary matrix nestedness temperature calculator (BINMATNEST) (Rodrı´ guezGirone´s and Santamarı´ a 2006; 2010). As defined by the program, a matrix with perfect nesting has a temperature of 0°, whereas the temperature rises as disorder increases (no order remains at 100°). When significant nestedness was detected, the particular species and

(a)

(b)

Fig. 1 A conceptual illustration depicting that the nested Odonata assemblage enables us to identify priority habitats and monitor species for conservation and restoration. (a) Species D1, D2, and D3 inhabit Pond P1, which has good habitats; species D2 and D3 inhabit Pond P2, which has ordinary habitats; and only species D3 inhabits Pond P3, which has poor habitats. Specification of the habitat characteristics of Pond P1 that differs from the other ponds provides information on priority habitat characteristics for

conservation and restoration. Species D1 (i.e., rare species) can be a monitoring species as an indicator of good habitat quality. (b) Good habitat will be considered for conservation efforts, and the identified specific characteristics of a good habitat provide clues about how to restore poor habitats. In this context, the identified monitoring species will be appropriate for evaluating the restoration process

696 Table 1 Environmental characteristics (area, forest cover around pond, NO3 -N and chemical oxygen demand [COD] concentrations in pond water, and presence of non-native fish) of each study pond Pond

Area (m2) Forest cover (%) NO3 -N (ppm) COD (ppm) Non-native fish presence

P1

P2

P3

P4

P5

P6

P7

P8

P9

P10

782 73.2 0.10 7.33 0

1960 85.3 0.13 5.67 0

4070 46.5 0.10 7.33 1

415 5.1 0.13 9.67 0

61.1 7.3 0.13 16.70 0

437 48.8 0.10 5.00 1

875 34.0 0.10 12.30 0

925 1.1 0.27 5.00 1

172 2.9 0.20 9.33 1

9900 15.7 0.17 8.33 1

1 and 0 indicate presence and absence, respectively

Fig. 2 Map showing the location of the study ponds on Amami-Oshima Island

habitat characteristics that should receive conservational importance were assessed based on the simple rule given in Fig. 1. Generalized linear mixed models (GLMMs) were used to evaluate the environmental factors affecting the richness of the Odonata species in the ponds. We first checked the correlations between each pair of environmental variables using Pearson’s correlation test and excluded highly correlated variables (r < 0.6) from the explanatory variables of the model to avoid multicollinearity. The explanatory variable candidates were as follows: pond area, forest cover, emergent plant cover, floating-leaved plant cover, floating plant cover, submerged plant cover, concentrations of NO3 and COD,

and non-native fish presence (Tables 1, 2). Identifiers in the study ponds were included as a random intercept in each model to consider the expected spatial variability in species occurrence. A Poisson error structure was used for the response variables with a log link function. After determining the explanatory variables, we constructed models including all combinations of the explanatory variables and the models that yielded the lowest Akaike’s information criterion (AIC); those with DAIC <2 were selected for descriptive purposes (Burnham and Anderson 2002). The GLMMs were performed using the glmmML package (Brostro¨m 2017), and model selections were performed using the MuMIn package (Barton´ 2016) in R 3.1.2 (R Development Core Team 2015).

697

Fig. 3 Photographs of each study pond. (a)  (j) indicate Ponds P1  P10, respectively

698 Table 2 The coverage (%) of aquatic plants and dominant species for floating, floating-leaved, emergent, and submerged plants inside the study ponds Pond

Floating plant

Floating-leaved plant

Emergent plant

Submerged plant

%

Dominant species

%

Dominant species

%

Dominant species

%

Dominant species

Potamogeton distinctus

53.8

Typha domingensis, Eleocharis dulcis, Leersia sp. Typha domingensis, Eleocharis dulcis, Leersia sp. Panicum repens Leersia sp. Leersia sp.

0.2

Limnophila sessiliflora, Blyxa echinosperma

P1

6.5

Utricularia exoleta

11.7

P2

1.4

Utricularia exoleta

0.0

85.3

P3 P4 P5 P6 P7 P8 P9 P10

0.0 0.0 20.7 0.0 1.0 0.0 73.3 0.0

0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0

5.3 0.3 65.0 0.0 0.0 2.7 6.7 0.0

Utricularia aurea Eichhornia crassipes Eichhornia crassipes

Leersia sp. Arundo donax

0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0

Table 3 Binary data matrix of the Odonata assemblage closest to perfect nestedness, permutated by the binary matrix nestedness temperature calculator (BINMATNEST) Family

Coenagrionidae Libellulidae Aeschnidae Gomphidae Libellulidae Libellulidae Libellulidae Coenagrionidae Aeschnidae Libellulidae Libellulidae Macromiidae Libellulidae Libellulidae Aeschnidae Coenagrionidae Libellulidae Libellulidae Aeschnidae Libellulidae Coenagrionidae Libellulidae Aeschnidae Corduliidae Libellulidae Libellulidae

Species

Ischnura senegalensis Pantala flavescens Anax parthenope julis Ictinogomphus pertinax Tramea virginia Orthetrum sabina sabina Trithemis aurora Ceriagrion auranticum ryukyuanum Anax guttatus Rhyothemis variegata imperatrix Crocothemis servilia servilia Epophthalmia elegans Zyxomma petiolatum Orthetrum melania Anax panybeus Paracercion melanotum Orthetrum albistylum speciosum Brachydiplax chalybea flavobittata Anax nigrofasciatus nigrofasciatus Acisoma panorpoides panorpoides Agriocnemis femina oryzae Rhyothemis severini Anasiaeschna martini Hemicordulia okinawaensis Lyriothemis elegantissima Hydrobasileus croceus

PRDB

JRDB

* *

*

* ** *** * **

*

Pond P1

P2

P3

P4

P5

P6

P7

P8

P9

P10

1 1 1 1 1 1 1 1 1 1 1 0 0 1 1 0 1 1 1 1 1 1 1 1 0 1

1 1 1 1 1 0 1 1 1 1 1 0 1 1 1 0 1 1 1 0 1 1 1 1 1 0

1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 1 0 0 0 0 0 0 0 0 0 0

1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 0 0 0 0 0 0 0 0 0 0 0

1 1 1 1 1 1 0 1 1 1 1 0 0 0 0 1 0 1 0 1 0 0 0 0 0 0

1 1 1 1 1 1 1 1 1 0 0 1 1 0 1 1 0 0 0 0 0 0 0 0 0 0

1 1 1 1 1 1 1 0 1 0 1 1 1 1 0 1 0 0 0 0 0 0 0 0 0 0

1 1 1 1 1 1 1 0 1 1 1 1 1 0 0 1 0 0 0 0 0 0 0 0 0 0

1 1 1 1 1 1 1 1 0 1 0 0 0 1 0 0 1 0 0 0 0 0 0 0 0 0

1 1 1 1 1 1 1 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0

Bold 1 indicates the presence of Odonata species, whereas 0 indicates its absence. PRDB, prefectural red data book (Kagoshima Prefecture) and *, **, and *** in the column indicate ‘‘important from its distribution characteristic’’, ‘‘near threatened’’ and ‘‘vulnerable’’, respectively. JRDB, Japan red data book and * in the column indicates ‘‘endangered’’

Results The areas of the study pond varied widely, from 61 (Pond P5) to 9,900 (Pond P10) m2 (Table 1). Ponds P1 and P2 were predominantly surrounded by forest (73 and 85%, respectively), but the other ponds were surrounded primarily by arable land with less forest cover

(1–49%). NO3 -N was evenly low, but COD ranged from 5.0 to 12.3 ppm among the study ponds. Invasion by non-native fish was recorded for half the study ponds (Table 1). Ponds P1, P2, and P5 had lush emergent plant cover (54–85%), and Pond P9 was characterized by dense floating plant cover (73%). Submerged plant cover was generally low at all ponds but sparsely distributed at Pond P1 (0.2%) (Table 2).

699 Table 4 Akaike’s information criterion (AIC) ranking of the models that explain the number of Odonata species in ponds. +, positive effect on the number of Odonata species; –, negative effect; blank, no significant effect Rank

Explanatory variable Forest

1 2 3 4 5 6 7 8

+ + + + + + +

Emergent plant

Area

Floating-leaved plant

COD

– + + + +

– –

+ +

Floating plant

AIC

DAIC

54.6 54.8 55.8 55.9 56.4 56.4 56.5 56.6

0.00 0.16 1.20 1.31 1.79 1.80 1.81 1.92

COD stands for chemical oxygen demand, and the column of DAIC indicates differences between AIC values of the best model and selected model

The odonate presence/absence matrix showed that the faunal structure in our study was significantly nested (Table 3; T = 26.16°, P < 0.00001). The greatest number of Odonata species was found at Pond P1 (22 species) and decreased following the numerical order of the ponds, and the smallest was in Pond P10 (eight species). The rare species found at only one or two ponds were as follows: Anax nigrofasciatus nigrofasciatus, Acisoma panorpoides panorpoides, Agriocnemis famina oryzae, Rhyothemis severini, Anasiaeschna martini, Hemicordulia okinawaensis, Lyriothemis elegantissima, and Hydrobasileus croceus; the common species observed at all ponds were as follows: Ischnura senegalensis, Pantala flavescens, Anax parthenope julis, Ictinogomphus pertinax, and Tramea verginia; (Fig. 4). Assemblages in the most species-rich ponds lacked Epophthalmia elegans and Paracercion melanotum though the species were observed at greater than or equal to half the study ponds (Table 3). After the correlation check, we selected pond area, forest cover, emergent plant cover, floating-leaved plant cover, floating plant cover, and COD as explanatory variables for the full model because the pairs of floatingleaved plant cover and submerged plant cover (r = 1.00), floating plant cover and NO3 (r = 0.63), and presence/absence of non-native fish and emergent plant cover (r = 0.61) were highly correlated. The GLMMs ranked with low AICs indicated that forest cover and emergent plant cover consistently positively affected the number of Odonata species, and the models explained by either of the variables were selected as good models (Table 4). Several selected models showed a negative effect of pond area and positive effects of floating-leaved plant cover and COD on the number of Odonata species (Table 4).

Discussion Similar to those of previous studies, our results confirmed that the lentic Odonata assemblage in representative pond habitats on Amami-Oshima Island was

significantly nested (Kadoya et al. 2004, 2008b). The factors structuring nestedness have been reported for selective colonization (Kadmon 1995), selective extinction (Cutler 1991), habitat nestedness (Honnay et al. 1999), interspecific variation in sensitivity, and tolerance to environmental conditions (Worthen et al. 1996). These aspects are thought to be basically linked to the so-called ‘‘habitat preferences’’ of a target taxon and the habitat qualities of target factors. Odonate biology, including the habitat preference of each species, is well known and described in illustrated guide books (e.g., Corbet 1999; Sugimura et al. 1999). Based on such descriptions, we inferred the important factors for the observed nestedness. Anax nigrofasciatus, Ac. panorpoides, Ag. famina oryzae, R. severini, Anac. martini, He. okinawaensis, L. elegantissima, and Hy. croceus were identified as the rarest species in the present study and prefer lush emergent cover and/or submerged plants, which act as larval habitats according to Sugimura et al. (1999). A description in the same guide states that Anax nigrofasciatus, Anac. martini, He. okinawaensis, and L. elegantissima additionally prefer forest cover around ponds as shade and for roosts. The common species that appeared in all ponds, such as Is. senegalensis, Pan. flavescens, Anax parthenope julis, Ic. pertinax, and T. verginia, generally inhabit the open water surface, even artificial fountains, reservoir dams, and swimming pools with no aquatic plants (Sugimura et al. 1999). The most common species, Anax parthenope julis, is taxonomically closely related to the rare species Anax nigrofasciatus nigrofasciatus but more adapted to open water and less shade (Sugimura et al. 1999). The difference in habitat preference between the rare and common species corresponded well to the results of the GLMMs, indicating the positive effects of forest and emergent plant cover on the number of Odonata species. Presence of the common species that prefer the open water surface at the species-rich ponds indicating that the ponds (e.g., Ponds P1 and P2) had sufficient open water surface area, which effectively attracted the common species. In contrast, Pond P9, with a dense cover of invasive floating Eichhornia crassipes, which is known to

700

Fig. 4 Photographs of specimens of members of the Odonata species observed in this study; all specimens are male, except the Orthetrum sabina sabina, Rhyothemis severini, and Anaciaeschna martini specimens, which are females

701

flourish rapidly and impair macroinvertebrate community structure in lentic habitats due to detritus loading, light limitation and oxygen depletion (Coetzee et al. 2014), had few Odonata species. These results suggest that some heterogeneity in the habitat characteristics of the pond surface is an important factor determining odonate diversity; this does not contradict the consensus regarding the relationship between biodiversity and habitat heterogeneity (e.g., Benton et al. 2003; Kadoya and Washitani 2011). Thus, the number of Odonata species would not increase continuously with increased forest and aquatic plants cover, as indicated by the GLMM results. Odonata species richness can be maximized in areas with surrounding forest and emergent plants that provided appropriate shade combined with appropriate heterogeneity of water surface coverage. For example, absence of E. elegans which prefers deep pond with large open surface, and P. melanotum which prefers open water surface with sparse floating plants (Sugimura et al. 1999) at the most species-rich ponds implies that reducing aquatic plant cover would further enhance the diversity of habitats and species at the ponds. In relation to this, foraging pressure of the alien omnivorous fishes might reduce aquatic plants (e.g., Lahser Jr. 1967) because the emergent plant cover negatively correlated with the presence of the fishes. However, introduction of the fishes to ponds possibly induce not only plant reduction but also other impacts on native ecosystems such as over-predation of native insects. Therefore, introducing such fishes as a plant controller must be strictly avoided. A negative effect of pond area and positive effects of floating-leaved plants and COD were detected after selecting the models. Several studies have reported that larger ponds harbor more species than smaller ones (Oertli et al. 2002; Kadoya et al. 2004), which contradicts our result. This contradiction was caused by the fact that the most species-poor pond (Pond P10), which was constructed with cement and rubber and had no aquatic plants, was much larger than the other ponds. This interpretation is supported by the analysis excluding Pond P10, which resulted in no significant effect on pond area. Meanwhile, the positive effect of floatingleaved plants may include poor predictions because of the lack of the variability in these environmental factors (present only at Pond P1). Although floating-leaved plants could provide habitats for larval and adult odonates, the positive effect of COD remains questionable. Hassall et al. (2011) indicated that species richness of lentic organisms and COD show relationship of a downward convex in the range from 25 to 8710 ppm, but the causal relation is still unknown. Based on the nestedness of the Odonata assemblages, the presence of rare species can provide insights determining whether a particular pond functions as a good habitat. Additionally, forested ponds with lush emergent plants could be considered as priority habitats for lentic conservation on Amami-Oshima Island. Then, it should

be noted that structure of habitat network can also be an important factor determining priority habitats for conservation and restoration. Because the most-species rich ponds were close each other, a metacommunity structure in the ponds might also contribute its high species diversity. The lucidophyllous forest was originally established on Amami-Oshima Island because of the moist and warm climate, and traditional rice agriculture has provided lentic environments for animals. Therefore, the results presented here probably reflect factors that maintained the past biodiversity in historical landscapes formed by nature and humans. The rapid disappearance of the lentic environments on this island underscores the importance of conserving these unique communities. Humans should manage the vegetation inside and outside the lentic habitats to achieve successful future conservation and restoration. Our results emphasize how identifying priority habitats and species for monitoring provides important data regarding lentic biodiversity conservation in the newest national park of Japan. Acknowledgments We thank the anonymous reviewers for providing invaluable comments for the improvement of this manuscript. A portion of this study was supported by the Environmental Research Fund (4-1409) of the Ministry of the Environment, Japan, and the JSPS KAKENHI Grant Number 26292181. The authors declare no conflict of interest.

References Barton´ K (2016) MuMIn: multi-model inference. https://cran.r-project.org/web/packages/MuMIn/index.html. Accessed 21 Jul 2017 Benton TG, Vickery JA, Wilson JD (2003) Farmland biodiversity: is habitat heterogeneity the key? Trends Ecol Evol 18:182–188 Brostro¨m G (2017) glmmML: generalized linear models with clustering. https://cran.r-project.org/web/packages/glmmML/ index.html. Accessed 21 Jul 2017 Burnham KP, Anderson DR (2002) Model selection and multimodel inference: a practical information-theoretic approach, 2nd edn. Springer, New York Chovanec A, Waringer J (2001) Ecological integrity of riverfloodplain systems-assessment by dragonfly surveys (Insecta: Odonata). Regul River Res Manage 17:493–507 Clark TE, Samways MJ (1996) Dragonflies (Odonata) as indicators of biotope quality in the Kruger National Park, South Africa. J Appl Ecol 33:1001–1012 Coetzee JA, Jones RW, Hill MP (2014) Water hyacinth, Eichhornia crassipes (Pontederiaceae) reduces benthic macroinvertebrate diversity in a protected subtropical lake in South Africa. Biodivers Conserv 23:1319–1330 Corbet PS (1999) Dragonflies, behavior and ecology of Odonata. Cornell University Press, Ithaca Cutler A (1991) Nested faunas and extinction in fragmented habitats. Conserv Biol 5:496–505 D’Amico FS, Darblade S, Avignon S, Blanc-Manel S, Ormerod SJ (2004) Odonates as indicators of shallow lake restoration by liming: comparing adult and larval responses. Restor Ecol 12:439–446 Development Core Team R (2015) R: a language and environment for statistical computing. R Foundation for Statistical Computing, Vienna

702 Fischer J, Lindenmayer B (2005) Perfectly nested or significantly nested—an important difference for conservation management. Oikos 109:485–494 Hagiwara S (1992) Changes in agricultural activities in Amami Islands—focus on changes in crop types. Farm Manage Res 30:35–45 (in Japanese) Hansson L (1998) Nestedness as a conservation tool: plants and birds of oak-hazel woodland in Sweden. Ecol Lett 1:142–145 Hassall C, Hollinshead J, Hull A (2011) Environmental correlates of plant and invertebrate species richness in ponds. Biodivers Conserv 20:3189–3222 Honnay O, Hermy M, Choppin P (1999) Nested plant communities in deciduous forest fragments: species relaxation or nested habitats? Oikos 84:119–129 Kadmon R (1995) Nested species subsets and geographic isolation: a case study. Ecology 76:458–465 Kadoya T, Washitani I (2011) The satoyama index: a biodiversity indicator for agricultural landscapes. Agric Ecosyst Environ 140:20–26 Kadoya T, Suda S, Washitani I (2004) Dragonfly species richness on man-made ponds: effects of pond size and pond age on newly established assemblages. Ecol Res 19:461–467 Kadoya T, Suda S, Nishihiro J, Washitani I (2008a) Procedure for predicting the trajectory of species recovery based on the nested species pool information: dragonflies in a wetland restoration site as a case study. Restor Ecol 16:397–406 Kadoya T, Suda S, Tsubaki Y, Washitani I (2008b) The sensitivity of dragonflies to landscape structure differs between life-history groups. Landscape Ecol 23:149–158 Karube H, Kitano T (2016) Lentic aquatic insects of Amami Islands falling into crisis. In: Mizuta T (ed) Natural history studies in Amami—biodiversity in subtropical islands. Tokai University Press, Kanagawa, pp 250–270 (in Japanese) Kato M (2000) Anthophilous insect community and plant-pollinator interactions on Amami Islands in the Ryukyu Archipelago, Japan. Contr Biol Lab Kyoto Univ 29:157–252 Kawai T, Tanida K (2005) Aquatic insects of Japan: manual with keys and illustrations. Tokai University Press, Kanagawa (in Japanese) Lahser CW Jr (1967) Tilapia mossambica as a fish for aquatic weed control. Prog Fish Cult 29:48–50

Ministry of the Environment, Japan (MOE) (2015) List of nonnative species that threaten ecosystems in Japan. https://www.env.go.jp/nature/intro/1outline/list/list.pdf. Accessed 8 Jun 2017 (in Japanese) Mizuta T (2016) Natural history studies in Amami—biodiversity in subtropical islands. Tokai University Press, Kanagawa (in Japanese) Moriyama H, Iijima H, Harada N (1990) A favorable distribution pattern of aquatic biotopes in terms of the dispersability of dragonflies. People Environ 15:2–14 (in Japanese) Naito R, Sakai M, Morimoto Y (2012) Negative effects of deep roadside ditches on Pelophylax porosa brevipoda dispersal and migration in comparison with Hyla japonica in a rice paddy area in Japan. Zool Sci 29:599–603 Oertli B, Joye DA, Castella E, Jude R, Cambin S, Lachavanne JB (2002) Does size matter? The relationship between pond area and biodiversity. Biol Conserv 104:59–70 Patterson BD, Atmar W (2000) Analyzing species composition in fragments. Bonn Zool Monogr 46:9–24 Rodrı´ guez-Girone´s MA, Santamarı´ a L (2006) A new algorithm to calculate the nestedness temperature of presence-absence matrices. J Biogeogr 33:924–935 Rodrı´ guez-Girone´s MA, Santamarı´ a L (2010) How foraging behaviour and resource partitioning can drive the evolution of flowers and the structure of pollination networks. Open Ecol J 3:1–11 Schouten MA, Verweij PA, Barendregt A, Kleukers RJM, de Ruiter PC (2007) Nested assemblages of Orthoptera species in the Netherlands: the importance of habitat features and lifehistory traits. J Biogeogr 34:1938–1946 Sugimura M, Ishida S, Kojima K, Ishida K, Aoki T (1999) Dragonflies of the Japanese Archipelago in color. Hokkaido University Press, Hokkaido Sugimura K, Yamada F, Miyamoto A (2003) Population trend, habitat change and conservation of the unique wildlife species on Amami Island, Japan. Glob Environ Res 7:79–89 Washitani I (2001) Traditional sustainable ecosystem ‘SATOYAMA’ and biodiversity crisis in Japan: conservation ecological perspective. Glob Environ Res 5:119–133 Worthen WB, Carswell ML, Kelly KA (1996) Nested subset structure of larval mycophagous fly assemblages: nestedness in a non-inland system. Oecologia 107:257–264