There is a great diversity of mating systems in anurans, but the factors driving this diversity remain uncharted. Here, using phylogenetic comparative methods, we explore which factors are related to the presence of mating systems in anurans and examine evolutionary patterns. We collected data for 717 anuran species and evaluated the effects of climate, temporal breeding pattern, sex ratios, terrestriality, and fecundity on their mating systems. Scramble competition and simultaneous polyandry occur more in species with aquatic eggs and oviposition sites, which have larger clutches, in habitats with high temperature seasonality, and low annual temperature/precipitation. Direct benefits occur often in species with terrestrial larger eggs and smaller clutches, which have terrestrial or sheltered oviposition sites, and in habitats with lower temperature seasonality. Only lek and polygyny were correlated with sex ratios. Most mating systems are strongly influenced by shared evolutionary history and are linked to the evolution of reproductive modes. Climate and sex ratios seem to play a role in the plasticity of mating systems, as a species can have more than one. Finally, our study represents a considerable advance toward understanding the anuran mating system evolution.

We conducted a systematic review of anuran mating systems and reproductive behavior through literature searches in Web of Science (WoS), Scopus, and Scielo, spanning all years on record (Appendix1: Figure S1). The first search was performed on 29 March 2020, using the following keywords (Appendix1: Figure S1a): (("Female choice" OR "Female preference" OR "Male attendance" OR "Endurance rivalry" OR "Chorus attendance" OR "Polyandry" OR  "lek" OR "resource defense" OR "mating system" OR  "scramble competition" OR  "breeding ecology" OR "breeding activity" OR "breeding behavior" OR "Reproductive biology" OR " courtship behavior" OR "Sexual selection") AND (Anura*  OR *frog*  OR toad)). In addition to mating systems, we collected information related to aggressive behavior, territorial defense, satellite behavior, male endurance rivalry, and parental care. To compile more specific behaviors, we conducted three additional searches in WoS on 23 August 2021, with the following keywords (Appendix1: Figure S1b): (i) (("male") AND ("nest" OR "constructed basins" OR "chamber*") AND ("frog*" OR "anuran*" OR "treefrog")); (ii) (("mating ball" OR "multiple amplexus" OR "multiple mating" OR "multiple-male mating" OR "multiple spawning" OR "group spawning") AND ("frog*" OR "anuran*" OR "treefrog")); and (iii) (("satellite" OR "fight*" OR "territorial defense") AND ("frog*" OR "anuran*" OR "treefrog")). Also, we expanded our literature review by searching for complementary information from other sources (e.g. books, thesis, reports, etc.; Appendix1: Figure S1c).  For particular species missing in our database (species that had vague information about the mating system), we finally conducted a series of specific searches in Google Scholar using the species name along with the following keywords: “mating system” OR “reproductive biology” OR “reproduction” (Appendix1: Figure S1c). All these searches resulted in a total of 2801 identified records (Appendix1: Figure S1). We screened the abstract, title, and keywords of each identified article and selected 525 for subsequent full-text assessment. According to the relevance of the information reported, we finally retained 440 articles that were used to retrieve data and compose our final database.

We conducted a systematic review of OSR using complementary literature searches in Web of Science (WoS), Scopus, and Scielo, spanning all years on record. The search was performed on 21 September 2020, with the following keywords: ("Sex ratio" OR "OSR" OR "Operational Sex Ratio") AND (Anura* OR frog* OR toad*). We followed previous studies (Elmberg 1990) by including articles that calculated the sex ratio during the breeding season, assuming that both females and males were sexually active within the period. We found 694 papers, but only 94 reported OSR or sex ratio during the breeding season. In this systematic review, we also included books and articles found in additional sources (n=24). We standardized OSR as the percentage of sexually active males of the breeding population: Number of sexually active males / Number of sexually active males and females. This calculation is the most suitable ratio, ranging from 0 to 1 (Kvarnemo and Ahnesjö 1996). Finally, we calculated a species-level average if the species had multiple OSR data.

To obtain additional information about the mating systems and reproductive behavior of particular anuran species and develop a consistent global database, minimizing missing values, we conducted a survey among international experts (N=32) in specific taxa and regions, with proven experience in the natural history of anurans. An online form composed of six closed-ended questions  was shared with the researchers who provided answers for a range of 1-50 species (mean ± SD = 7.46 ± 14.04) based on their knowledge and experience. The form was specifically designed to provide an easy and interpretable link between the expert answers and each category of the studied variables, using a decision tree . Thereby, we obtained information for 219 species through the expert survey, totaling 798 species in our database. We were able to fill gaps for the species already present in our database (18.80 %, n=150) as well as adding new species (8.5%, n=68).

For this study, species’ mating systems were categorized into: (i) General mating system, with three general categories (promiscuity [PGA], exclusive polyandry [PA], or polygyny [PG]; Table 1), and (ii) Specific mating system, with five specific categories (simultaneous polyandry [PASI], sequential polyandry [PASE], lek, scramble competition [SC], and direct benefits [DB]; table 1). For this study, we used social mating systems, the social and behavioral associations between males and females (Zuk and Simmons 2018). Promiscuous mating systems are those in which males and females mate with more than one partner, whereas polyandrous mating systems are those in which females mate with more than one male (Mobley 2014) and take place in the form of (i) simultaneous polyandry (concurrent mating with several males and presence of sperm competition; Roberts and Byrne 2011) or (ii) sequential polyandry (non-simultaneous mating with several males throughout the breeding season; Byrne and Roberts 2012). Polygynous mating systems are those in which males mate with more than one female and take place in anurans in the form of (i) lek (when males gather in an exhibition arena and are chosen by the female, so that males do not offer any type of resources or parental care; Bradbury and Gibson 1993), (ii) scramble competition (when males actively search for females; Wells 2007) and (iii) direct benefits system, which encompasses the resource defense system itself (when males defend territories and/or resources essential for reproduction; Wells 2007), but also those cases in which there is paternal care and other male behaviors that benefit the offspring/female, such as nest building. Paternal care was used as a proxy for direct benefits. As we found only two monogamous anuran species (both males and females have only one partner throughout the reproductive season; Emlen and Oring 1977), we excluded them from the analysis. We also excluded the species Limnonectes palavanensis, as it presents a reverse sexual mating system (Goyes Vallejos et al. 2017; Vallejos et al. 2018), incompatible with the explored categories.

As explanatory variables, we also extracted information related to fecundity (clutch size maximum, offspring size maximum, and female size) and temporal breeding patterns from several sources, including the articles retained in the literature searches. Most of the fecundity data were obtained from published databases (Han and Fu 2013; Nali et al. 2014; Oliveira et al. 2017; Vági et al. 2019a; Pincheira-Donoso et al. 2021; Furness et al. 2022) and selected the highest value for fecundity variables with multiple values. Adult Sex Ratio (ASR) was extracted from the Vági et al. (Vági et al. 2020) database, and the reproductive pattern was defined based on the original publications and published databases (Nali et al. 2014; Trochet et al. 2014). Regarding terrestriality data, we collected the following information: general reproductive mode (aquatic egg or terrestrial egg) and oviposition site (aquatic, terrestrial, arboreal, or sheltered). General reproductive mode data were obtained from published databases (Gomez-Mestre et al. 2012; Vági et al. 2019b; Nunes-De-almeida et al. 2021). The definition of aquatic and terrestrial eggs followed from the studies differed only in the cases of eggs deposited in foam nests in the aquatic environment. We use the definition that considers them terrestrial (Vági et al. 2019a). When there was a conflict of information between studies, we excluded the data. The oviposition site was treated as a categorical variable (aquatic, terrestrial, arboreal, and sheltered). Data were obtained from previously published databases (Silva et al. 2020; Nunes-De-almeida et al. 2021; Pincheira-Donoso et al. 2021; Furness et al. 2022). We followed the definition of the categories according to Silva et al., (Silva et al. 2020), and species that contained conflicting records between the databases were excluded.

We also determined the climate conditions across the geographic range of each studied species. Based on occurrence data of the species in our dataset, available at the International Union for Conservation of Nature (IUCN, IUCN 2023) and other sources (Batista et al. 2013; Souza et al. 2019), and overlaid on a world grid with 1º of resolution (latitude and longitude) to obtain the presence and absence matrix of species. We retrieved four climatic variables from interpolated layers in WorldClim expressed at a spatial resolution of 2.5 min (BIO1- Annual Mean Temperature; BIO4- Temperature Seasonality; BIO12- Annual Precipitation; BIO15- Precipitation Seasonality; www.worldclim.org; Fick and Hijmans 2017). We assigned each species a single value per predictor, calculated as the average of all values obtained for each variable of each species.

For mating system data, the general mating systems (PA, PG, PGA) and specific mating systems (PASI, PASE, SC, LEK, DB; for abbreviations see Table 1) were included as binary variables (presence or absence). Considering the predictor variables, general reproductive mode (aquatic eggs and terrestrial eggs) and temporal breeding pattern (explosive or prolonged) were binary variables, and oviposition site was a categorical variable (arboreal, aquatic, sheltered, and terrestrial). The other variables were all continuous: Offspring size, clutch size, ASR, OSR, mean temperature, mean precipitation, temperature seasonality, and precipitation seasonality. The models with fecundity variables (i.e., offspring size and clutch size), were performed with and without controlling for female size. Female body size was used because of the relationship between female body size and offspring size (Monroe et al. 2015).