The cost of polygyny (multiple matings) and the
evolution of female care in poison

Kyle Summers (Dept of Biology East Carolina University)
David J D Earn (Dept. of Zoology, University of Cambridge).

Parental care is a central feature in the life histories of many animals.  How and why one type of parental care evolves from another is a controversial issue of central importance in evolutionary ecology.  Research on a variety of taxa suggests that several factors may affect the trajectories and equilibria of parental care evolution, including the cost of parental care on male mating success, the effect of parental care on female fecundity, the relationship between parental care and confidence of paternity and costs of male polygyny to female reproductive success.
Weygoldt addressed the evolution of female parental care in poison frogs of the genus Dendrobates in which males or females carry tadpoles from terrestrial oviposition sites to small pools of water.  Most of his research involved captive frogs.  He hypothesised that males in species with bi-parental care might frequently deposit eggs from one mate in pools containing tadpoles from a previous mate resulting in cannibalism and a cost of polygyny to the second female.  He also hypothesised that the high cost of polygyny-could cause the evolution of female care in a lineage derived from bi-parental care.
NATURAL HISTORY
The poison frogs, a Neotropical group of toxic diurnal terrestrial frogs exhibit a variety of parental care strategies.  The D. tinctorius group which includes D. auratus, D. azureus, D. tinctorius, D. leucomelas and D. truncatus deposits small (2-8 eggs) clutches in the leaf-litter that are attended by the male.  Once the eggs have developed into mature tadpoles they are typically carried by the male to small pools of water that form in tree holes.  Additional tadpoles may be placed in the same pools and these may be cannibalised by other tadpoles.  In the D histrionicus group which includes D. histrionicus, D. speciosus, D. pumilio, D. granuliferus and D. arboreus, females deposit the young in small accumulations of water that form in the leaf or stem axils of plants such as bromeliads.  The female then returns to the pools periodically and lays infertile eggs for the tadpole.  This form of feeding is obligatory in D. pumilio, the tadpole will neither grow nor survive if not provisioned with trophic eggs.  Although the male provides some egg attendance in some of these species (i.e. D. pumilio) the female is the main care provider.  In this paper we will classify these species as having female parental care.  In captive D. ventrimaculatus Zimmermann and Zimmemann observed that both parents attend the clutch that is oviposited above a pool in the leaf axil of a bromeliad.  The male then carries the tadpoles (usually one at a time) to a different pool and returns periodically to call thus attracting his mate who lays infertile eggs to nourish her young.  Recent field research suggests that bi-parental care does not occur in D. ventrimaculatus in Amazonian Ecuador.  Briefly, the mating system is promiscuous, unrelated individuals oviposit in the same axil and females do not return to pools to feed tadpoles.  The frogs studied by the Zimmermanns were from Peru and may eventually be classified as belonging to a different species from those in Ecuador.
Bi-parental care has been observed in captive D. reticulatus and in other similar species from Amazonia, and in the field in D. vanzolinii.  In this species males carry tadpoles individually to small pools of water in tree holes and females feed tadpole with nutritive eggs.  Males and females maintain an intimate association or pair bond throughout the period of parental care.  Studies of birds have revealed extraordinary variation within and between populations of a single species in the mating system and parental roles and it is possible that similar levels of variability occur in poison frogs, so there may be variation in parental roles even among populations of the same species.
THE EVOLUTION OF PARENTAL CARE.
Male parental care was hypothesised to be the ancestral state in Dendrobates based on studies of parental care in species of Phyllobates, Minyobates and Epipedobates.  Four of the five species of Phyllobates have been studied in captivity and all exhibit male parental care.  Fourteen of 22 species of Epipedobates have been studied and all exhibit male parental care.  In Minyobates minutus recent research has demonstrated that this species also exhibits male parental care.  The genus Colostethus is thought to be basal to the toxic dendrobatids.  In most Colostethus, the males provide care for the tadpoles although females provide care (tadpole transport) in a few species.  Tadpoles are transported to streams rather than axil pools and tadpoles are never fed nutritive eggs.  The genus Aromobates is believed to be the most basal taxon and it is not known whether there is parental care.  The most parsimonious interpretation of the information currently available is that male parental care is the primitive state within Dendrobates.
It is suggested that the D. ventrimaculatus group (with bi-parental care) is intermediate between the male and female care group in Dendrobates and suggested that parental care evolved in a transition series from male to bi-parental to female parental care.  Weygoldt suggested mat D. ventrimaculatus males may mate with more than one female but will use the eggs from some of their mates to feed tadpoles from previous mates.  He said that they do this by mating with the second female on bromeliad leaves above the axil where previously carried offspring were placed.  The male then pushes the eggs or embryos into the axil pool where they are eaten by the tadpole.  He argued that secondary females of this species experience a high cost of polygyny and that this may have favoured the evolution of female parental care.  The Zimmermanns constructed a phylogenetic hypothesis for 32 species of poison frogs using behaviour observed in terraria as characters using which they suggest that female parental care evolved from bi-parental care which in turn evolved from male parental care.  However some of the behavioural characters that they used were not independent of the behaviours of interest (parental care) and some behaviour apparently differ between the field and ten-aria.  To derive an independent estimate of the evolution of parental care in poison frogs, molecular methods were used to construct a phylogenetic hypothesis for representative species of these frogs which yielded a phylogeny, which differs from, that represented by the Zimmermanns.  This hypothesis supports the monophyly of female care but suggests that female and bi-parental care evolved independently from male parental care.  Hence the pattern of parental care in the genus Dendrobates is somewhat controversial.
THE COST OF POLYGYNY, INTERSEXUAL AND PARENT-OFFSPRING CONFLICT
Selection often favours multiple mating by males but polygyny may impose a cost on some or all of the males' mates resulting in inter-sexual conflict.  In order to evaluate Weygoldt's hypothesis it is important to describe how polygyny can impose costs on female poison frogs.  The cost of polygyny is likely to derive from the deposition of multiple tadpoles in a pool or from the deposition of eggs and/or embryos with tadpoles.  The cost arises because the tadpoles are cannibalistic and because they have a negative effect on each other's growth rate.  The most obvious cost rises when a male mates with two females and deposits one or more offspring from each female in the same pool.  If the tadpole of one female eats its half-sibling, then the direct fitness of the victim's mother is reduced and her indirect (inclusive) fitness is not increased because she is unrelated to the cannibal.  Such cannibalism might be in the interest of the father depending on the trade-off between increased growth and survivorship of the cannibal and reduced number of offspring.  However because the male controls the placement of tadpoles, cannibalism is likely to coincide with the inclusive fitness interests of the father given constraints on pool availability.
A less obvious cost of polygyny may occur if tadpoles refrain from eating half-siblings which may engender an unusual type of parent-offspring conflict.  Usually parent offspring conflict occurs where offspring take more resources than parents are able to provide or take a higher proportion than is in the parent's interest.  This conflict may occur over siblicide if it is in an offspring's interest to kill its sibling but not in the interest of a parent.
A different type of parent-offspring conflict may occur if offspring are more altruistic than is in the parent's interest.  Even if tadpoles could recognise half siblings and the inclusive fitness of tadpoles that do not eat their half-sibs is higher than those that do not there is still likely to be a cost to the mother.  In species with male parental care it is unlikely that females could influence tadpole interactions directly as eggs are not deposited near tadpole deposition sites and the males transport and select those sites.  In those species the females only opportunity to reduce such conflict may be to prevent the male from mating with other females.  Mate guarding has been observed in the field in D. auratus and D leucomelas.
It is not necessarily the case that polygyny imposes a cost on females.  For example if the value of feeding tadpoles is very high and the cost of polygyny is evenly distributed then some degree of it might actually represent a net benefit to females.  In poison frogs it is possible that enhanced growth and survivorship experienced by tadpoles in pools provisioned by eggs and embryos from two females to outweigh costs to each female that result from 'uncertainty of maternity'.  This might occur if there is a premium on fast growth.  By breeding synchronously each of two females may ensure that at least some of her off spring have a chance of reaching a large size quickly (by cannibalising pool-mates) before any other eggs, embryos or tadpoles can be placed in the pool by other individuals.  In order for a cost of polygyny to produce selection for some females to care on their own, there must be no better option for those females.  They must achieve higher reproductive success by caring for their own offspring than by leaving them to be cared for by an already mated male (who might not spend as much time caring for his recent mate's young) or by an unmated male (who may have low territory quality or low parental ability).
 Weygoldt (1987) suggested that the cost of polygyny is highest in members of the D. ventrimaculatus group with bi-parental care and that it predisposed these species towards the evolution of female parental care.
Many thanks to the authors for their permission to use this paper.  This is an abridged version that will be concluded later.