First case of male alloparental care in amphibians: tadpole stealing in Darwin’s frogs

Alloparental care, i.e. care directed at non-descendant offspring, has rarely been described in amphibians. Rhinoderma darwinii is an Endangered and endemic frog of the tem perate forests of Chile and Argentina. This species has evolved a unique reproductive strategy whereby males brood their tadpoles within their vocal sacs (known as neomelia). Since 2009, the National Zoo of Chile has developed an ex situ conservation programme for R. darwinii, in which during reproduction, adults are kept in terraria in groups of 2 females with 2 males. In September 2018, one pair engaged in amplexus, with one of the males fertilizing the eggs. Twenty days later, the second male (the alloparent) ingested all of the 8 viable, recently hatched tadpoles, releasing 2 froglets 9 wk later. This unusual behaviour observed in captivity is complemented by observations in the field of male R. darwinii brooding tadpoles of markedly different sizes, possibly reflecting the brooding of tadpoles from different ovipositions. To our knowledge, this is the first documented case of male alloparental care in amphibians and an example of captive breeding programmes contributing important knowledge about the natural history of threatened species.


INTRODUCTION
Parental care is any parental behaviour or adaptation that increases offspring fitness, often at some cost to the care-giver (Riedman 1982). At least 10% of amphibian species exhibit some kind of parental care, from egg attendance to more complex behaviours such as food provisioning, viviparity and brood-ing (Crump 2015, Schulte et al. 2020). An outstanding case among amphibians is the brooding of larvae within the vocal sac of the male (also known as neomelia), a type of parental care that is exclusively known in Darwin's frogs Rhinoderma rufum and R. darwinii. Darwin's frogs are endemic to the temperate forests of Chile and Argentina, and both species have recently experienced rapid and severe ABSTRACT: Alloparental care, i.e. care directed at non-descendant offspring, has rarely been described in amphibians. Rhinoderma darwinii is an Endangered and endemic frog of the temperate forests of Chile and Argentina. This species has evolved a unique reproductive strategy whereby males brood their tadpoles within their vocal sacs (known as neomelia). Since 2009, the National Zoo of Chile has developed an ex situ conservation programme for R. darwinii, in which during reproduction, adults are kept in terraria in groups of 2 females with 2 males. In September 2018, one pair engaged in amplexus, with one of the males fertilizing the eggs. Twenty days later, the second male (the alloparent) ingested all of the 8 viable, recently hatched tadpoles, releasing 2 froglets 9 wk later. This unusual behaviour observed in captivity is complemented by observations in the field of male R. darwinii brooding tadpoles of markedly different sizes, possibly reflecting the brooding of tadpoles from different ovipositions. To our knowledge, this is the first documented case of male alloparental care in amphibians and an example of captive breeding programmes contributing important knowledge about the natural history of threatened species.
KEY WORDS: Allopaternal care · Rhinoderma darwinii · Neomelia · Cross-fostering · Ex situ conservation · Captive breeding OPEN PEN ACCESS CCESS population declines (Soto-Azat et al. 2013, Azat et al. 2021. R. rufum has not been observed since 1981 and is categorized as Critically Endangered (Possibly Extinct) by the IUCN Red List of Threatened Species (Azaz et al. 2021). Although R. darwinii survives across much of its historical range, its populations are small and severely isolated, and the species is currently classified as Endangered by the IUCN (Azat et al. 2021). Males of both species incorporate newly hatched tadpoles into their vocal sacs, leading to the appearance of being pregnant. Neomelia is completed upon metamorphosis in R. darwinii (Jorquera et al. 1972, Busse 2003, 2004. R. darwinii is a fully terrestrial, diurnal and forestspecialist frog. During the reproductive season (September to March), females of this species approach calling males in their retreat sites to engage in amplexus (Busse 2003(Busse , 2004. Females lay a small clutch of eggs (typically 4−10 eggs, Busse 2004, Valenzuela-Sánchez et al. 2014; but up to 34 have been recorded, Wilhelm 1932). Observations from the field and captivity indicate that neither males nor females of R. dar winii attend egg clutches (Busse 2004, Valenzuela-Sánchez et al. 2014, Serrano et al. 2020. Around 20 d after oviposition, the male parent returns to the egg clutch to collect the newly hatched larvae (Jorquera et al. 1972, Busse 2004. The movement of embryos within the eggs apparently serves as a stimulus for males to aid tadpole hatching by biting off the egg membrane and swallowing the tadpoles (Busse 2003). Males usually brood between 5 and 8 tadpoles at a time, although a maximum of 22 has been reported (Wilhelm 1927). After a period of generally 6 to 8 wk, metamorphosis is completed and froglets are released into the forest environment by the males (Busse 2004).
An alloparent is an individual, other than the genetic parent, that provides care for conspecific offspring (Riedman 1982). This type of care has been well studied in birds, mammals, fish and inver tebrates (Riedman 1982, Eggert & Müller 1992, Lar gia dèr et al. 2001. Rarely reported in amphibians, alloparental care in this group has thus far been documented only in females; however, alloparental care is likely underestimated in amphibians (Schulte et al. 2020). In eastern North America, four-toed salamanders Hemidac tylium scutatum and marbled salamanders Ambys toma opacum build communal nests in which a small proportion of females stay to protect the eggs (including non-related offspring) from predation and dehydration (Harris et al. 1995, Croshaw & Scott 2005. In Brazil, Rodrigues et al. (2011) re ported an anecdotal case of tadpole brood mixing and female interspecific alloparental care between criolla frogs Leptodactylus latrans and pointed belly frogs L. podicipinus, possibly as a result of failed recognition. The best documented case of alloparental care in amphibians is for the caecilian Boulengerula taitana from the Taita hills in Kenya, where female parents provide their own skin as a food source for their offspring, and young from different litters have shown genotypes not matching the guarding female (Kupfer et al. 2008).
The possibility of male alloparental care in R. darwinii has been raised before (Valenzuela-Sánchez et al. 2014), but evidence to support this has re mai ned elusive until now. In the present study, based on observations made of frogs in captivity, we report the successful brooding of tadpoles by a foster male. We discuss hypotheses that could explain this observed behaviour.

MATERIALS AND METHODS
Since 2009, the National Zoo of Chile has developed a Darwin's frog ex situ conservation programme, aiming to safeguard the genetic diversity of Rhinoderma darwinii, with a plan to re-introduce the species using captive-bred animals in areas where this species has recently disappeared and to assist R. rufum conservation if this species is re-discovered.
Founder R. darwinii for this project were 16 adult individuals (8 collected in 2009 and 8 in 2012), and 86 offspring have been born since 2009. Adult R. darwinii were maintained in groups of 2 males and 2 females in glass terraria of 50 × 50 × 50 cm with a 4 cm layer of sphagnum moss. Supplied water was filtered in sediment carbon filters, treated in an inverse osmosis system and reconstituted. Water was provided through a spray pump system, simulating rain to keep the humidity at a constant ~90%. During most of the year (September to June), mean air temperature was maintained between 12 and 15°C. During the peak of the cold season (July and August) frogs were kept in a refrigerator at 5°C to induce hibernation. Food consisted of small invertebrates, including live fruit flies Drosophila melanogaster, crickets Acheta domesticus, woodlice Porcellio scaber, weevils Si to philus garnarius and larvae of tenebrids Tenebrio molitor, given 3 times per week. Each frog was individually identified using its unique ventral colouration patterns (Soto-Azat et al. 2013), but different colouration on the flanks and gular area also allowed easy individual recognition from the outside of the terrarium (Fig. 1B,C). When frogs in a terrarium showed signals of reproductive activity, they were monitored daily between 09:00 and 17:00 h by direct observation, with their activities and behaviours recorded on a standardized data sheet. The behaviour reported here is complemented by 12 yr of ob servations of R. darwinii in the field.

RESULTS
On 2 September 2018, we observed 2 males (M83 and M89) that started calling and competing to engage in amplexus with the same female (F86). The se cond female (F87) did not show reproductive behaviour during the reproductive season. On 27 September, we observed that amplexus took place between individuals F86 and M89 (hereafter the 'genetic parent'), while M83 (hereafter the 'alloparent') consistently attempted to interfere. These interruptions occurred on at least 5 occasions with variable intensity, and consisted of attempts to embrace the amplectant couple and perform jumps and/or calls right next to the couple. The egg clutch, fertilized by the genetic parent, consisted of 14 eggs that were deposited on one side of the terrarium into a hole within moss and under a fern (Fig. 1). After oviposition, the reproductive female (F86) retreated and was never observed near the egg clutch again. For the next 21 d post-oviposition, both males were observed occasionally in proximity to the egg clutch: the alloparent made 5 approaches and the genetic parent made 18. Most egg approaches by males occurred between Days 18 and 21 (14 approaches), when visible em bryos started moving within the eggs. On Day 20, the male allo parent positioned itself over the egg clutch, exhibiting defensive behaviour and preventing the genetic parent from approaching ( Fig. 1B,C). On Day 21, the alloparent began to ingest the tadpoles into its vocal sac, a process that the genetic parent attempted to interrupt several times (Fig. 1D-F). At this point, we counted 8 viable embryos, all ingested by the allo parent within 6 h (see Video S1 in the Supplement at www. int-res. com/ articles/ suppl/ n045 p331_ supp/). The genetic male parent then withdrew and started calling from a new retreat site. After 62 d, the male alloparent completed brooding and released 2 froglets.

DISCUSSION
Ex situ conservation programmes can contribute relevant biological knowledge that otherwise would be difficult to obtain from field studies (Harding et al. 2016). To our knowledge, our observations are the The unusual case of alloparental tadpole brooding exhibited by male Rhinoderma darwinii differs from other types of alloparental care reported in nature. Conspecific brood parasitism consisting of parasitic laying of eggs in a conspecific nest without providing parental care is well described among birds, and less commonly in fish and arthropods (Riedman 1982, Lar giadèr et al. 2001. Cooperative breeding with non-breeding helpers (usually females) has also been reported in many species of social birds and mammals (see Clutton-Brock 2002), and adoption of orphaned newborns has been described in birds and mammals (Riedman 1982).
The causes and consequences of alloparental care remain poorly understood. Although this behaviour has been claimed to be inconsistent with classic evolutionary theory (Riedman 1982), it has also been proposed that selective benefits, along with various environmental pressures or reproductive mistakes, may collectively promote the evolution of alloparental care and adoption (Wisenden 1999). What could drive this unusual behaviour in R. darwinii is unknown, but based on knowledge of alloparental care in other taxo nomic groups, 2 hypotheses can be proposed: (1) fostering unrelated offspring could increase pa rental care experience in the alloparent male and hence promote its future breeding success (Riedman 1982, Largiadèr et al. 2001; and (2) be cause of their apparent breeding success and thus high individual quality, the alloparent male could also enhance its probability of finding a mate for the next breeding event (Clutton-Brock 2002). It is possible that R. darwinii females may exhibit a preference for brooding males for mating, thus potentially forcing non-brooding males to steal unrelated eggs to become 'pregnant', as has been reported in some fish species with male egg attendance that actively steal eggs to increase reproductive success (Largiadèr et al. 2001). However, female midwife toads Alytes ob stetricans in captivity have shown no preference between egg-brooding and non-egg-brooding males (Bush 1996). Although 'pregnant' males continue to vocalize after the onset of brooding, calling ceases after several days later as tadpoles grow (Serrano et al. 2020). Our observations in captive R. darwinii have not shown sexual engagement of 'pregnant' males, despite being kept in groups with additional females.
Offspring stealing and adoption have sometimes been attributed to reproductive mistakes (Riedman 1982, Rodrigues et al. 2011. Male alloparental care in R. darwinii, as we witnessed in captivity, might also occur in the wild. In almost all cases (captive and wild), 'pregnant' males brood tadpoles of similar size, but observations in the field have sporadically noted concurrently brooded tadpoles with marked differences in size (Busse 1970, Crump 2002, Valenzuela-Sánchez et al. 2014; Video S1). This might be the result of either different speeds of development among tadpoles of the same brood, or the ability of some males to swallow newly hatched tadpoles from different oviposition events. Even when they are not exhibiting territorial behaviour, such as egg defence, the high home range overlap between adult males (84%; Valenzuela-Sánchez et al. 2014) allows males to be in contact with non-related conspecific eggs, providing an opportunity for tadpole stealing. Also, the low number of successfully brooded tadpoles reported herein has been observed previously in captivity (O. Cabeza-Alfaro pers. obs.). Only 2 of the 8 ingested tadpoles completed metamorpohosis. The fate of the other 6 tadpoles is unknown, but possibilities are: (1) tadpoles passed to the gastrointestinal tract where they were digested, and (2) dead tadpoles in the vocal sac served as food for the brooding tadpoles. However, the second option is less likely, as the oral structures of R. darwinii tadpoles are not adapted for adelphophagy. These issues should be addressed in future research.
Dramatic population declines have recently been observed in both species of Darwin's frogs (Soto-Azat et al. 2013). Similarly, the gastric-brooding frogs Rheo batrachus vitellinus and R. silus from eastern Australia are considered to have been extinct since the 1980s. As with gastric-brooding frogs, the potential extinction of Darwin's frogs would imply the loss of a unique reproductive behaviour before it is fully understood. Our study provides important information about the reproductive biology of Rhinoderma, which is highly valuable for efforts to conserve these Endangered amphibians.