Study species. Our laboratory studies included four anuran taxa, all of which are terrestrial and breed in wide array of freshwater habitats including ponds, marshes and swamps. Two species belong to the “true toads” (family Bufonidae). Our target species was the eastern-Japanese common toad (Bufo japonicus formosus; mean total length [=TL] of tadpoles = 30 mm21), and the parapatric bufonid was the western-Japanese common toad (Bufo japonicus japonicus; mean tadpole TL = 35 mm21). The other two taxa are members of the family Ranidae, both of which are broadly sympatric with B. j. formosus: the Japanese brown frog (Rana japonica; mean tadpole TL = 38 mm21) and the montane brown frog (Rana ornativentris; mean tadpole TL = 43 mm21). Larval body sizes in the two groups used in the experiment spanned a similar range (Table 1).
Tadpoles of all four species were derived from eggs collected in natural waterbodies from two sites (B. j. formosus and R. japonica - Tochigi prefecture, B. j. japonicus and R. ornativentris - Okayama prefecture). Tadpoles of B. j. formosus, R. japonica, and R. ornativentris were found in the same waterbodies (Haramura, personal observation), confirming that competition is likely to occur in nature. Tadpoles of all four species were maintained in groups in 120 L plastic containers (66 x 86 x 34 cm). Tadpoles were fed algal pellets (Hikari Algae Wafers, Kyorin) ad libitum, and water was changed weekly. The tadpoles used in the experiment were haphazardly selected from these containers and added to experimental bins as described below.
Laboratory experiments. Experiments were conducted using plastic tanks (26 x 38 x 23 cm), each filled with 23 L water and located in a covered building exposed to ambient temperatures. At the start of the experiment, we added a 2 cm layer of soil substrate and 3 g of algal pellets to each bin. Tadpoles varied in sizes and developmental stages at the beginning of the experiment (see Table 1).
Our experiment consisted of six treatments: (1) 5 larvae of B. j. formosus, (2) 15 B. j. formosus, (3) 50 B. j. formosus, (4) 25 B. j. formosus plus 25 B. j. japonicus, (5) 25 B. j. formosus plus 25 R. japonica, and (6) 25 B. j. formosus plus 25 R. ornativentris. The experiment was a complete randomised block design, with 5 replicate tanks per treatment. We recorded the number of B. j. formosus to metamorphose from each tank (survival), as well as the larval period, and length (snout to urostyle length=SUL) and mass of each B. j. formosus metamorph from each tank. Treatments 1, 2 and 3 allowed us to assess the effect of intraspecific competition on B. j. formosus, whereas treatments 3, 4, 5, and 6 allowed us to assess the strength of interspecific versus intraspecific competition at standardised density.
Because it is not possible to visually distinguish between metamorphs of B. j. formosus and B. j. japonicus, we used a Loop-Mediated Isothermal Amplification (LAMP) assay to distinguish between these two subspecies in the interspecific competition experiments. LAMP is a genetic method which detects the presence/absence of a specific DNA sequence in the tested sample22. Total DNA of each metamorph was extracted from frozen tissue using the DNeasy Blood and Tissue Kit (QIAGEN) with standard protocols. Following extraction, each sample was tested by LAMP assay in two independent systems―assays with B. j. japonicus-positive and B. j. formosus-positive primer sets. For the primer set used and the experimental conditions, we followed the methods with slight modification23. The reaction mixtures were incubated at 63-65˚C for 90 min and then heated at 95˚C for 2 min to terminate the reaction.
Statistical Analyses. We analysed treatment effects on larval period, metamorph SUL and metamorph mass using linear mixed effects models (ANOVA), with treatment and spatial block as fixed effects, and tank as a random effect (JMP 9.0, SAS Institute, Cary, NC, USA). When the overall ANOVA gave a significant result, we performed post hoc Turkey’s HSD tests for pairwise comparison of treatments. We analysed survival to metamorphosis as a binomial response (alive, dead24) using ANOVA, with treatment and spatial block as fixed effects (package carData25,26). Survival analyses were based on the quasi-binomial distribution to account for overdispersion of data. Alpha level was set at p = 0.05 in all analyses.
Ethics approval. All procedures were approved by Rakuno Gakuen University Animal Care Committee (permit #DH21D6). The study was carried out in compliance with the ARRIVE guidelines.