We investigated the food composition of the Southeast Asian Rhacophorid frog species Polypedates leucomystax, an invasive species introduced to Okinawajima in the Ryukyu Archipelago, during their breeding activities and the month immediately after that. We examined the stomach contents of 190 males and 54 females, uncovering 215 prey items (including at least 56 species) from 113 males and 21 females. The rate of empty stomachs was relatively high, 45.1% for all frogs (40.5% for males and 61.1% for females). Most prey were arboreal, with a minority being obligately terrestrial or aquatic. Insects constituted the majority of the diet (61.4%), followed by other arthropods, gastropods, and a vertebrate (an agamid lizard neonate). Orthopterans were the dominant prey group in terms of number (29.8%), frequency of occurrence (43.3%), and volume (51.6%). This species is considered an opportunistic feeder, a conclusion corroborated by comparisons with locally varied dietary data from populations of conspecific or closely related species in Southeast Asia. The average prey length was approximately one-third of the body length, but instances of the consumption of larger prey were observed. Similarly, males are able to swallow prey up to two-thirds of their gape width and females up to one-third. The data obtained in this study cover most of the active season of this species on the island, and thus provide a basis for elucidating the ecological impact of this species.
Introduction
The Rhacophorid treefrog, Polypedates leucomystax (Gravenhorst, 1829), is a medium-to-large arboreal species indigenous to Southeast Asia. It was unintentionally introduced to the Ryukyu Archipelago in southwestern Japan from the Philippines (Kuraishi et al., 2009). Since its first confirmation in Okinawajima in 1964 (Kuramoto, 1965), this species has demonstrated significant invasive capabilities (as a frequent stowaway) and rapid range expansion in this archipelago. As of 2023, this species has been recorded on 36 islands in the archipelago extending south of Tokunoshima, plus Kitadaitojima of the Daito Islands (Ota et al., 2004, 2008; Maenosono and Toda, 2007; Taba et al., 2013; Takao et al., 2013; Aoyagi et al., 2022; Okinawa Amami Nature Conservation Office, Ministry of the Environment, 2023).
The presence of this species poses a potential ecological risk to the highly endemic terrestrial faunas in this archipelago (Ota, 2002; Sengoku and Toda, 2019), prompting a few eradication attempts based primarily on the precautionary principle (Toda, 2019). The potential negative impacts of this species on native faunas in the archipelago include (1) competition for food resources or breeding habitats with native anurans (Ota, 2002), (2) disruption of native anuran breeding activities (Ota, 2002), (3) introduction of novel parasites to local faunas, including co-occurring amphibians (Hasegawa, 1993, 1994; Hasegawa and Ota, 2017; Hasegawa et al., 2018) and predators feeding on this frog (Kadota and Maenosono, 2007), and (4) predation on insects in high-density frog areas (Sengoku and Toda, 2019).
However, apart from parasite issues (e.g., Hasegawa et al., 2018), few attempts have been made to scientifically validate these concerns. For instance, there is only limited published data on the diet of the Japanese populations of this species, which is essential for evaluating the degree of feeding niche overlap between this species and indigenous animals. We investigate its diet in Okinawajima and report the results here. Our data will form the basis for assessing the severity of concerns mentioned above, and could aid in prioritizing measures against this well-established, resilient frog species and other invasive alien species in this region.
Materials and Methods
Study site and collection
The frogs investigated in this study were collected from seven localities in Ogimi Village, located in the northern part of Okinawajima. This area is a critical biodiversity conservation site and is registered on UNESCO's World Heritage List (UNESCO, 2021). Three of these localities were artificial ponds, two were water tanks, one was a roadside ditch, and one was a paddy field. The mating call of this species was heard from late April to late October in these localities. Co-occurring anurans that were considered to be potentially suffer from competitive pressure from this species were Hyla hallowellii (Hylidae), Fejervarya kawamurai (Dicroglossidae), Microhyla okinavensis (Microhylidae), Buergeria japonica, and Zhangixalus viridis (Rhacophoridae).
The frogs were collected between July and November in 2019, September and November in 2020, and May and June in 2021. Surveys commenced an hour after sunset. We captured the frogs by hand, kept them in plastic bags, and then euthanized them with a 10% water solution of Acetonechloroform Hemihydrate. The euthanasia was performed at the collection site because national law prohibits the transportation of live animals of this highly invasive species.
Laboratory processing
In the laboratory, we measured the snout-vent length (SVL) and gape width (GW) of the frogs. Afterward, we dissected the frogs to extract their stomach contents. The sexes and maturity stages were determined by the developmental degrees of the testes in males or the gonads in females. We examined the extracted stomach contents under a binocular microscope and identified each animal item to the lowest taxonomic category. Stomachs containing only well-digested amorphous material or non-animal remains (e.g., sand grains or plant fragments) were considered empty, aligning with the results of previous studies that used stomach flushing or forced regurgitation methods to obtain stomach contents from live frogs (e.g., Yap, 2015).
Diet analysis and prey dimension
For each prey taxon, we calculated the following three values: the numeric percentage (%N), the volumetric percentage (%V), and the occurrence percentage (%F). The maximum length and width of each food item were measured using a digital caliper. The volume of each food item was calculated using the formula for an ellipsoid:
where V is the volume, L is the length, and W is the width (Magnusson et al., 2003).
Depending on the prey type, we made certain modifications in calculating the volume of food items. For gastropods, we replaced length with the maximum shell height. For lizards, we did not include the long, thin tail in the body length.
We estimated the original body proportions of partial or fragmentary prey items from measurements of conspecific samples found in the prey items or museum specimens based on the assumption that the length of these body parts is linearly proportional to the maximum length and width of their body. For prey items that we could not infer the original body proportions from the remaining parts, we used the average volume of the same prey taxon in the food items (e.g., the volume of an unmeasurable “unidentified Orthoptera” item was the average volume of other orthopteran prey).
Statistical methods and calculations
The Chi-square test was used to investigate the intersexual differences in the vacuity index (the rate of empty stomachs). The Fisher's exact test was used to examine the difference in the males' monthly vacuity indexes. The normality of the variables was confirmed by the Shapiro–Wilk test. The statistical significance of several intersexual differences in specific values (prey item numbers and dimensions, total stomach content volume, and taxonomic variety of prey) was tested using Wilcoxon's rank sum tests.
Results
Frogs and body size
Our study involved 244 individual frogs, with 190 males and 54 females. The frogs were found mostly on shrubs or grasses, but some were also found on the ground. Males' SVL ranged from 41.3 to 65.9 mm (median: 49.2; x̄ ±SD=49.9±3.76), while females' SVL ranged from 56.9 to 83.1 mm (median: 73.0, x̄ ±SD= 72.8±5.3). All males were mature, and apart from one young female (SVL of 56.9 mm), all females were considered adults. The smallest mature female recorded an SVL of 64.1 mm. Due to the significant sexual size dimorphism of this species, males and females were analyzed separately in the following size-related comparisons.
Vacuity index and number of food items
Out of the 244 individuals, 134 (54.9%) had food items in their stomachs. The vacuity index for males was 40.5% (77/190), and for females, it was 61.1% (33/54). This intersexual difference was statistically significant (Chi-square test, χ2=7.20, P<0.01). Meanwhile, no statistically significant difference was observed in males' monthly vacuity index from May to November (Fisher's exact test, P=0.07). It was high in May (65%), August (50%), and October (38%) and low in June (28%), September (31%), and November (20%). The small sample size of females precluded a meaningful analysis of the monthly vacuity index variation.
A total of 215 prey items (186 items from 113 males and 29 items from 21 females) were found. The median number of prey among individuals with food in their stomachs was one (x̄ =1.65) for males and one (x̄ =1.38) for females, with no statistically significant intersexual difference (Wilcoxon's rank sum test, W=1288, P=0.47). The maximum number of prey found in a single stomach was seven for males and three for females.
Prey dimension
The median measurements of 139 prey items from male stomachs were as follows: length 9.7 mm (range: 0.4–66.0 mm), maximum width 3.1 mm (range: 0.3–9.3 mm), and volume 40.4 mm3 (range: 0.02–1248.6 mm3). The measurements for the 21 items from female stomachs were as follows: length 19.7 mm (range: 0.6–81.5 mm), maximum width 6.0 mm (range: 0.4–7.4 mm), and volume 413.3 mm3 (range: 0.05–826.2 mm3). These median values were significantly larger in females than in males (Wilcoxon's rank sum test; length: W=726, P<0.01; width: W=763, P<0.01; volume: W=636.5, P<0.01). The median total stomach content volume was 128.7 mm3 (range: 0.02–1248.6 mm3) in males (n=84) and 476.5 mm3 (range: 1.14–1316.5 mm3) in females (n=16), with a statistically significant intersexual difference (Wilcoxon's rank sum test, W=335, P<0.01).
The median and mean ratios of the length of the longest prey item to the frog's SVL were 24–33% in both sexes, reaching up to 140% in males and 126% in females. When compared to the frog's GW, the median and mean ratios were 76–101%, with the ratio reaching 446% in males and 369% in females. Regarding the width of the widest prey item found in each stomach, the median and mean ratios to the frog's SVL were approximately 8–9% in both sexes. These ratios reached 20% in males and 10% in females. When compared to the frog's GW, the median and mean ratios were 24–27% in both sexes, with the ratio peaking at 64% in males and 33% in females.
Prey taxa
We identified 56 species from these food items (Appendix), the bulk (85.6%N, 94.8%F, and 86.6%V) of which consisted of arthropods (predominantly insects: 61.4%N, 73.9%F, and 74.0%V), followed by gastropods (14.0%N, 20.2%F, and 12.0%V) and vertebrates (0.5%N, 0.8%F, and 1.4%V) (Table 1, Appendix). Orthopterans were the most important prey group of this frog. They represented the largest proportion of food items by number (29.8%N), followed by arachnids (20.0%N) and gastropods (14.0%N) (Table 1). Similarly, orthopterans were the most frequently consumed taxonomic group (43.3%F), followed by arachnids (28.4%F) and gastropods (20.2%F). In terms of volume, orthopterans accounted for more than half of all prey items (51.6%V), followed by gastropods (12.0%V) and Phasmatodea (7.0%V) (Table 1). The Eneopterid cricket Cardiodactylus guttulus was identified as the principal prey species, accounting for 17.2%N (37 individuals), 25.4%F, and 38.7%V of the diet (Appendix). This and several other prey species are arboreal and typically found in lower vegetation, such as bushes or grasses. However, we noted some ground-dwelling species, such as the field cricket Teleogryllus occipitalis and spiders Pardosa spp., and even some aquatic species, such as the diving beetle Copelatus oblitus and freshwater snail Austropeplea ollula.
The median, mean, and SD of the prey variety at the species level (i.e., that of Appendix) for all frogs with food in their stomachs were 1.0 and 1.49±0.81 taxa. These values were 1.0, 1.51±0.85 (maximum=6) for males and 1.0, 1.33±0.58 (maximum=3) for females, respectively, and there was no significant intersexual difference in the medians (Wilcoxon's rank sum test, W=1292.5, P=0.44). Similarly, the prey varieties at the order level (i.e., the taxonomic level shown in Table 1) were 1.0, 1.51±0.85 (range: 1–3) for males and 1.0, 1.33±0.58 (range: 1–3) for females, respectively, and there was no significant intersexual difference in the medians (Wilcoxon's rank sum test, W=1271.5, P=0.53).
The prey taxa consumed by males and females were similar. For instance, the most frequently foraged prey was the cricket Cardiodactylus guttulus in both sexes (male: 14.5%N, 23.0%F; female: 41.4%N, 57.1%F). Several other prey species (the stick insect Sipyloidea sipylus, beetle Phaeochrous emarginatus, and spider Neoscona scylla) were also common to both sexes. However, gastropod consumption by females was infrequent, unlike males (Table 1).
The majority of prey items for both males and females were less than 25 mm in length. Prey exceeding this length consisted exclusively of wholly or partially soft-bodied species, such as the stick insect Sipyloidea sipylus, a lepidopteran larva, Acridid or Tettigoniid orthopterans (Euconocephalus sp. and Psyrana ryukyuensis), the great house centipede
Table 1.
Summary of %N, %F, and %V for prey classified according to higher taxonomic categories. Boldface fonts and underlined text represent the top three and maximum values, respectively.
Thereuopoda clunifera, and the agamid lizard Diploderma polygonatum polygonatum.
The prey item with the largest relative width (20.3% of the frog's SVL and 63.9% of the frog's GW) was an unidentified land snail with a maximum shell width of 9.4 mm. This prey was consumed by a male with a SVL of 46.2 mm and a GW of 14.7 mm. In terms of volume, the largest items were a lepidopteran larva (1248.6 mm3) and Thereuopoda clunifera (1154.5 mm3). Both items had a total length of 46.0 and 45.0 mm, respectively, and were found in multi-folded states in the stomachs of males with SVLs of 53.4 and 49.2 mm, respectively.
Most prey animals were common species that were not regarded as either declining or threatened. The exception was the lizard Diploderma polygonatum polygonatum, an endangered species listed on national or prefectural red lists (Ota, 2014; Tanaka and Chigira, 2017). The reconstructed SVL was 30.6 mm, suggesting that it might have been a slightly grown neonate.
Discussion
Prey and its composition
The diet of this species primarily consists of arthropods, especially orthopteran insects, with the majority of them being arboreal species. On average, the prey's length is approximately 30% of the frog's body length in both sexes, although they may occasionally consume prey larger than their body length. Similarly, males capable of swallowing up to two-thirds of their gape width and females up to one-third.
This species has often been considered a non-selective feeder in Southeast Asian studies, especially when compared to sympatric ant-specialist feeders, such as several microhylid frogs (Berry, 1965; Yap, 2015; Le et al., 2018). Prey compositions vary locally according to the composition of local prey guilds, as Berry (1965) predicted. For instance, while orthopterans emerged as the most numerous and frequently consumed prey category in Okinawajima and Vietnam (as “Polypedates cf. leucomystax”: Le et al., 2018), they were overtaken by hymenopterans (mostly Formicid ants) in the Philippines, Malaysia, and Singapore (Berry, 1965; Almeria and Nuñeza, 2013; Yap, 2015; Gersava et al., 2020). As Katsube and Ota (2007) proposed, the prevalence of these prey types could reflect their high abundance in the primary habitats of this frog in Okinawajima (secondary forests, shrubs, and harvested areas). The dominance of the arboreal cricket, Cardiodactylus guttulus, could also be attributed to its abundance in Okinawajima's forest-edge lower vegetation. In addition, unlike males, females rarely consume gastropods, which might be due to differences in reproductive activity. Females may occupy places with fewer gastropods and temporarily visit wetlands for breeding.
Implications for the ecological impacts of P. leucomystax
Regarding the predation pressure from this frog on native animals, particular caution may be necessary for threatened arboreal species with body dimensions small enough to be preyed upon by P. leucomystax, typically those less than 25 mm in body length. Additionally, the species' predation on an arboreal lizard confirmed in this study may not be an isolated case. Hasegawa (1984) reported some endoparasites typically found in lizards in the rectum of an Okinawajima P. leucomystax, suggesting that the consumption of a gecko could potentially explain the frog's infection.
Concerning potential ecological conflicts with indigenous anurans (Ota, 2002; Sengoku and Toda, 2019), the lack of dietary data for these native species hinders the evaluation of the existence or extent of competition for food resources between P. leucomystax and these native species. It is expected that P. leucomystax and the co-occurring indigenous Rhacophorine species (Zhangixalus owstoni on Ishigakijima Island and Z. viridis on Okinawajima Island and adjacent islands) have similar feeding habits and prey components. This assumption is supported by the dietary data of the closely related Z. arboreus (see Toda and Kusano, 2013), which mirrors that of the P. leucomystax presented in this study.
It should also be noted that our nutritional data for Okinawan P. leucomystax represent only their dietary habits during their breeding activities and the month immediately after that. However, this period corresponds the most of expected active seasons for this tropical species on the island. Therefore, the data presented here provide a foundation for elucidating the ecological impacts from this introduced species upon indigenous faunas.
Acknowledgments
We express our gratitude to T. Sasaki (Fujukan) for granting us access to museum specimens under his care, and to M. Osada (Fujukan) for literature. We also thank to valuable comments and suggestions from two anonymous reviewers. This research was supported by the Environment Research and Technology Development Fund (4–1901) of the Environmental Restoration and Conservation Agency of Japan. This research also received partial support from MEXT KAKENHI Grant Number 22H05240.
© 2024 by The Herpetological Society of Japan