Bioacoustics of Rhinella bella

D I SC USSI O N Our analyses reveal that populations from western Ecuador represent an independent lineage from all other known species of the R. marina group. We base this conclusion on the congruence observed between genetic, bioacoustic, morphological, and climate envelope information. Western Ecuadorian populations are highly different genetically from other species of the group. This mtDNA divergence is much higher than that observed between uncontroversial species pairs of the same group like R. arenarum and R. icterica or R. diptycha and R. poeppigii . Differences in advertisement calls are additional, clear indications that those populations represent a candidate species within the R . marina group. We provide conclusive evidence, based on an integrative approach combining different sources of information, that there are two, independent lineages of R . marina group toads west of the Andes in northern South America and Central America. One, R . horribilis , ranges north through Central America as far as southern Texas, whereas the other, R. bella , occurs in Ecuador and probably as far north as southwestern Colombia and as far south as extreme northern Peru . Past studies have shown two genetically distinct species within R. marina from the east and west sides of the Venezuelan Andes ( Slade and Moritz 1998 ), including the recent work of Acevedo et al. (2016) , who suggested that the R. horribilis name should be revalidated for the western Andes and Central American populations. Additionally, Mulcahy et al . (2006) , Pramuk (2006) , and Vallinoto et al. (2010) had previously found important levels of genetic divergence between Central American and Ecuadorian samples; however, incomplete sampling (only very few were included from western Ecuador ) and the lack of bioacoustic, morphological (but see Pramuk 2006 ), and ecological information, limited their conclusions. We provide novel evidence showing an independent evolutionary lineage consistent with a separate species within the R . marina species group, confirming that the western Ecuadorian populations ( R. bella ) are a new species distinct from the populations of Central America ( R . horribilis ), and from the populations from the Amazon Basin ( R . marina ). In contrast to the cryptic morphological diversity notable throughout the genus Rhinella ( Graybeal 1997 , Pramuk 2006 , Acevedo et al . 2016 ), including the R. marina species group, our morphometric analyses (including the geometric morphometric-based analysis) suggest that R. bella (western Ecuadorian populations) is overtly distinguishable from R . horribilis (Central American populations) and from R . marina (Amazon Basin populations) as currently understood. Isolated from all other R . marina group toads except for R . horribilis by the Andes, R. bella appears to be among the most geographically isolated and morphologically distinctive of this group of species. We report significant acoustic differences between R . horribilis , R. marina , and R. bella (see Figs 4–6 ) despite the generally conservative structure of advertisement calls among species of the R. marina group ( Maciel et al. 2007 ). The distinctly lower number of notes per call in R. bella relative to other species within the R . marina group is significant. Similarly, the higher dominant frequency of R. bella calls in comparison with calls of R . horribilis and R . marina is remarkable as this call trait is among the few not strongly dependent on operational temperature ( Fig. 6 ), and has been proven to be useful to assess species limits ( Köhler et al . 2017 ). This acoustic distinctiveness of R. bella among R . marina group species, and especially from R . horribilis , suggests that calls may have diverged at faster rates compared to morphology in this group, as has also been shown in other anuran species groups (e.g. Padial et al. 2008 , Angulo and Icochea 2010 , Funk et al . 2012 ). These patterns could be due to sexual selection promoting rapid divergence in both male traits and female preferences, or strong selection on species recognition ( Panhuis et al. 2001 ). This may be particularly relevant to the calls of R. bella vs. R . horribilis in that vocal character displacement, where these two species meet (probably in southern Colombia ), could promote mating isolation. Alternatively, mate-recognition traits as calls may diverge faster than morphology due to a strong stabilizing selection on morphological traits ( Funk et al. 2012 ), which may be imposed by several ecological factors ( Bickford et al. 2007 ). Previous studies have not fully evaluated call differences within the R . marina species group, although acoustic information seems to be particularly useful for the delimitation of cryptic species ( Funk et al. 2012 ). The comparison of climatic envelopes among studied populations provided important complementary evidence for species separation (see Fig. 11 ). Toad populations occurring in western Ecuador ( R. bella ) have a climatic envelope with lesser temperature seasonality than populations of R . horribilis occurring further north in South America and Central America. Temperature seasonality can be a major climatic factor limiting dispersal of amphibians ( Wiens et al. 2006 ), and Neotropical anurans demonstrate climatic specializations along temperature seasonality axes ( Graham et al. 2004 ). We would not expect notable and substantial morphological divergence between R. bella and R . horribilis if speciation occurred mainly via geographic isolation related to climatic components unless climate imposed selection on morphology. However, the relative importance of climatic factors in population divergence and speciation of these lineages deserves further investigation. We also found that R. marina from the Amazon Basin remains non-monophyletic, as was previously found by Vallinoto et al. (2010) and Rivera et al. (2022) . Samples of R. marina are nested within only one well supported clade (Clade E in Fig. 2 ) as the sample MJH 3678 of ‘ R . marina ’ from Clade D is probably misidentified and most likely corresponds to R . poeppigii . Clade E is composed of ‘ R. poeppigii ’ from Bolivia , R. diptycha and R. marina from the eastern side of the Ecuadorian Andes, and R. marina from the Amazonian Basin. The sample MNCN/ ADN6044 of R. poeppig ii, however, is likely misidentified and corresponds to either R . marina or R . diptycha . Furthermore, corroborating Vallinoto et al . (2010) and Ron et al . (2015) , our results placed R . crucifer as sister to the R . marina group according to mtDNA sequence information. Although R. crucifer is not formally considered to be in the R. marina species group ( Pramuk 2006 ), morphological similarities and a close cytogenetic and phylogenetic relationship between it and members of the R . marina group have been noted previously ( Blair 1972 , Cei 1972 , Duellman and Schulte 1992 , Baldissera et al . 1999 , Pauly et al. 2004 , Pramuk 2006 ). The discrepancy between phylogenetic relationships and the taxonomic classification of species within the R. marina species group could be influenced by several factors. First, the lack of an integrative taxonomy (i.e. species delimitation based on multiple sources of evidence; Padial et al. 2010 ) that provides a more effective solution to refine the systematics of the R. marina species group. Second, DNA sequences of some lineages (e.g. some populations previously known as ‘ R . horribilis ’) available for only a small number of samples and/or with limited representation of the geographic distribution ( Maciel et al. 2010 , Vallinoto et al. 2010 ). Third, the evident lack of prominent diagnostic morphological characters for some lineages in this group can result in the misidentification of specimens. For instance, Venegas and Ron (2014) reported on 10 specimens of R . poeppigii deposited at Museo de Zoología , Pontificia Universidad Católica del Ecuador (QCAZ) that had previously been misidentified as R . marina . Finally , genetic introgression resulting from interspecific hybridization, which is known to occur among numerous closely related species of bufonid toads in sympatry (e.g. Masta et al. 2002 , Smith and Green 2004 ), may confound phylogenetic reconstruction by obscuring topology and divergence estimates. This could be the case for some populations of R. marina and R. diptycha in the Amazon Basin ( Sequeira et al . 2011 ) and will likely continue to present challenges for delineating species boundaries within the entire R. marina group east of the Andes ( Rivera et al . 2022 ). West of the Andes , however, the genetic differentiation between R. bella and R. horribilis is clearly demonstrable, although the nature and extent of any putative contact zone between the two species, possibly in southwest Colombia and/or extreme northwest Ecuador , remains unknown. Cane toads are economically important and invasive outside of their native range, having been introduced on many Caribbean and Pacific islands ( Easteal 1981 , 1985 , Lever 2001 , Shine 2010 , 2018 ), where they are often now considered pests. The impact of such a large, voracious, prolific, and toxic amphibian on the ecology of northern Australia , for example, has been profound ( Shine 2018 ). Most of the evidence and historical record indicates that the cane toads currently in the Pacific originated from native populations in Venezuela , via Puerto Rico , Cuba , and Florida ( Slade and Moritz 1998 ). Considering the extent of cryptic taxonomic diversity of R. marina species toads, however, and the probability that new species such as R. bella remain to be discovered within the complex, it is possible that the cane toads infesting Australia and many Pacific Islands may not actually be R. marina . If this is the case, some aspects of our current knowledge on global biological invasions of these species (e.g. Tingley et al. 2014 ) would deserve reconsideration.

Distribution, ecology, and conservation status This species is restricted to the western side of the Andes, from sea level up to 2900 m a.s.l. Among Ecuadorian amphibians, it has one of the widest elevation ranges and occurs in six out of ten natural regions: Chocó Tropical Rainforest, Coastal Deciduous Forest , Coastal Dry Scrubland, Inter-Andean Scrubland, Western Montane Forest , and Western Piedmont Forest (regions as defined by Ron et al . 2022 ). It is most frequently found in artificial open areas including agricultural fields, grasslands, and urban and suburban habitats. It occurs infrequently in undisturbed forests. Reproduction takes place in ponds, rivers, and swamps. Reproductive activity (calling males or amplectant pairs) have been observed mainly in April, June, and July. As in other species of the R. marina species group, eggs are deposited in strings in water. Tadpoles are aquatic and have a high resistance to heat, up to 42.5 °C ( Pintanel et al . 2022 ). The complete distribution of R. bella remains unknown, mainly due to lack of distributional information in the Colombian Chocó Bioregion. Ecuadorian localities in northern Ecuador are close to the border with Colombia . Nearby Colombian records of ‘ R. horribilis ’ in iNaturalist (e.g. observation numbers 36168677 in Llorente , San Andrés de Tumaco and 102430734 in Tumaco ) morphologically resemble R. bella and we consider them as unconfirmed records of R. bella in Colombia . Moreover, the limited seasonal temperature variation of the Colombian Chocó Bioregion matches the climatic envelope of R. bella , further suggesting its occurrence in southern Colombia (see results of climatic envelopes; Fig. 11 ; Table 6 ). This species also occurs in northwestern Peru , south to 6°S of latitude (reported as ‘ R. horribilis ’ by Armijos-Ojeda et al. 2021 ). The new species occurs in the following protected areas: Reserva Ecológica Manglares Churute, Reserva Ecológica Los Ilinizas, Parque Nacional Machalilla, Reserva Ecológica Mache Chindul, Reserva Tesoro Escondido, Bosque Protector Mindo-Nambillo, Centro Científico Río Palenque, Bosque Petrificado de Puyango, Bosque Integral Otonga, Estación Biológica Bilsa ( Páez-Rosales and Ron 2022 as ‘ R. horribilis ’). The new species may be abundant in artificial open areas. Its tolerance of habitat disturbance, large distribution range along a wide elevation gradient, and occurrence in diverse natural regions (six different regions in Ecuador , one of the highest among Ecuadorian anurans) allow us to recommend its assignment to the Least Concern category in the Red List (based on the IUCN guidelines).

Diagnosis Rhinella bella is morphologically similar to R . horribilis and R . marina . However, R. bella differs from populations of R . horribilis by having a smaller body size, along with a proportionally larger femur, wider head, and larger tibia ( Fig. 8 ; Supporting Information, Tables S3 , S8). Differences among species were also evident in the skull shape (see results of morphometric analyses). Acevedo et al. (2016) reported cranial morphology differences between populations of the east ( R . marina ) and west ( R . horribilis ) side of the Venezuelan Andes, mainly around the pre-maxillar, nasal bones, and occipital region. We also found skull differences between populations on the eastern ( R . marina ) vs. western ( R. bella ) sides of the Ecuadorian Andes, mainly concerning the pre-maxillar and nasal bones. However, the non-monophyletic status of the eastern and Amazon Basin populations ( Vallinoto et al. 2010 ; this paper) precluded us from making further comparisons with the findings of Acevedo et al. (2016) . The new species is distinguishable from R . horribilis by a depressed area on the quadratojugal-squamosal contact region, and the elongation of frontoparietal bones at their most distant region ( Fig. 10 ; Table 5 ). Shape variables CV I CV II x1 -29.5047 105.6930 y1 -137.1518 101.0467 x2 95.0278 -95.6869 y2 95.6099 -63.0279 x3 -12.6571 26.8945 y3 42.8372 -5.6094 x4 -103.9906 -19.3548 y4 -25.4385 9.6817 x5 19.2644 -18.6325 y5 -19.4825 -22.6826 x6 72.4379 17.1427 y6 -22.4482 14.6958 x7 93.5268 0.6927 y7 78.0845 9.5599 x8 -37.4333 -17.7027 y8 17.0133 -48.1208 x9 -35.4839 -36.8079 y9 2.8802 -18.1410 x10 26.1319 55.7962 y10 -19.5041 -12.9742 x11 -63.8547 15.6947 y11 11.8286 8.6709 x12 -47.9645 -10.7240 y12 8.2034 18.2051 x13 24.5003 -23.0049 y13 -32.4319 8.6958 % of variance explained 53.85 46.15 Cumulative % 53.85 100 Rhinella bella differs in environmental envelope from R . horribilis and R . marina ( Fig. 11 ; Table 6 ; for details see climatic analysis section from Material and methods and Results) and is separated by high genetic distances from both species as well ( Fig. 3 ; Table 1 ; see molecular phylogenetic analysis from Material and methods and Results). Advertisement calls of R. bella are significantly shorter, with fewer notes and a shorter rise time than those of R . horribilis and R . marina ( Fig. 4 ; Table 2 ). The number of notes per call of R. bella (midrange = 23; range 6–40) is also lower than those reported in other species of the R . marina group [e.g. R . arenarum (midrange = 84, range 61–107; Straneck et al . 1993 ); R . marina (midrange = 176.5, range 18–335; present work); R . poeppigii (midrange = 27.5, range 10–45; De la Riva et al . 1996 ); and R . diptycha (midrange = 36.5, range 33–40; Köhler et al . 1997 )]. Furthermore, R. bella produce calls with higher dominant frequency than R . horribilis and R . marina ( Fig. 6 ; Tables 2 , 3 ).