Ó Springer-Verlag 1999
Behav Ecol Sociobiol (1999) 45: 339±348
ORIGINAL ARTICLE
A.J. Mark Hewison á Reidar Andersen Jean-Michel Gaillard á John D.C. Linnell Daniel Delorme
Contradictory ®ndings in studies of sex ratio variation in roe deer (Capreolus capreolus)
Received: 9 December 1997 / Accepted after revision: 11 November 1998
Abstract Patterns of sex ratio variation and maternal investment reported in the literature are often inconsistent. This could be due to intra- and inter-speci®c variation in social systems, but may also be a result of the a posteriori nature of much of this type of analysis or the testing of models which are inappropriate. Two recent papers reported directly opposed results concerning variation in ospring sex ratios in relation to maternal condition in roe deer, interpreting the results as support for the Trivers and Willard model and for the local resource competition hypothesis, respectively. In this paper, we present data on ospring sex ratios and early juvenile body weight from two long-term studies of this species to test predictions arising from these two models concerning sex biases in litter composition and maternal care. First, we observed no consistent pattern of sex dierences in an index of weaning weight or body weight at 1 month old in either population, indicating a lack of sex bias in maternal care. However, in one population, higher maternal body weight was associated with higher juvenile body weight of daughters, but not of sons. Secondly, we found a negative, but not statistically signi®cant, relationship between maternal body weight and litter sex ratio such that heavier females tended to produce more A.J.M. Hewison (&) Institut de Recherche sur les Grands MammifeÁres Institut National de la Recherche Agronomique BP 27, F-31326 Castanet-Tolosan, France e-mail:
[email protected] Tel.: +33-5-61285123, Fax: +33-5-61735477 R. Andersen á J.D.C. Linnell Norwegian Institute for Nature Research, Tungasletta 2 N-7005 Trondheim, Norway J.-M. Gaillard Universite Claude Bernard, Lyon 1. UMR CNRS 5558 43 boulevard du 11 novembre 1918, F-69622 Villeurbanne, France D. Delorme Oce National de la Chasse, CNERA CervideÂs-Sangliers 85bis Avenue de Wagram, F-75017 Paris, France
daughters and lighter females to produce more sons. These results indicate that roe females which have additional investment potential available do not invest it in sons, as predicted by the Trivers and Willard model. Our results may provide some support that roe deer are subject to local resource competition acting at the level of the individual mother; however, the fact that particular trends in sex ratio data can be explained in functional terms provides no indication that they are actually adaptive. Key words Local resource competition á Maternal care á Roe deer á Sex ratio á Trivers and Willard model
Introduction The adaptiveness of sex ratio variation has been widely considered in recent years for several taxonomic groups (for reviews see Trivers 1985; Clutton-Brock and Iason 1986). Discussion centers on the idea that when genetic return per unit investment diers for the production of male and female ospring, there will be selection pressure to invest more in that sex giving the highest return. Thus, parents should adjust the sex ratio of ospring in response to factors aecting their own future reproductive success and the future reproductive success of sons and daughters (Clutton-Brock et al. 1984; Green and Rothstein 1991). The Trivers and Willard model (TWM) predicts sexbiased investment of an individual mother dependent on the degree of skew in expected reproductive success of sons and daughters and her own ability to invest in those ospring (Trivers and Willard 1973). For species with polygynous breeding systems, a son in better than average condition at the end of the period of maternal investment is expected to have greater reproductive success than a daughter in similar condition, while the daughter is expected to have greater reproductive suc-
340
cess than the son if both are in poorer than average condition. When ospring condition at adulthood is dependent on the level of maternal investment, natural selection should favor maternal ability to adjust the sex ratio of ospring produced and/or to bias subsequent investment towards the sex most likely to be pro®table, according to her ability to invest. So, because reproductive success is expected to be more variable among males than among females in polygynous species, and because additional investment will have a disproportionately greater eect on male reproductive success, mothers in better than average condition should be selected to produce sons. Equally, parents might also be expected to adjust ospring sex ratios according to the dierence in potential future costs of rearing a son or a daughter and the consequent eects on the their own future reproductive success. Thus, the local resource competition (LRC) hypothesis (Clark 1978; Silk 1983) predicts that where non-dispersing progeny pose a competitive threat for resources, mothers in poor condition should produce that sex of ospring more likely to disperse from the natal area. Clark's (1978) original model concerned global patterns in the population sex ratio in response to the average level of resource competition. However, Silk (1983) extended the argument to the individual level, reasoning that those individual mothers more susceptible to resource competition should prefer to produce ospring of that sex which is more likely to disperse. In Silk's model, individual primate mothers were more susceptible to competition because of low social rank; however, an analogous argument may apply in other species with respect to mothers in poor physical condition which are susceptible to resource competition (see Hiraiwa-Hasegawa 1993; Hewison and Gaillard 1996). Thus, where males are the dispersing sex and females are philopatric, mothers in poorer than average condition should prefer to produce sons. Both the TWM and the LRC hypothesis have received support from some empirical studies among ungulates (TWM: Clutton-Brock et al. 1986; Thomas et al. 1989; LRC: Caley and Nudds 1987). However, thus far the evidence supporting adaptive variation of the ospring sex ratio according to maternal body condition is confusing and sometimes contradictory. Reports of biased sex ratios or sex-biased investment are not consistent for a given species (e.g., bison: Rutberg 1986; Green and Rothstein 1991; reindeer: Skogland 1986; Kojola and Eloranta 1989) or across species (ungulates: Byers and Moodie 1990; PeÂlabon et al. 1995). These inconsistencies may be accounted for by heterogeneity within and among ungulate species but, alternatively, may be due to testing of models for which the assumptions are not ful®lled. Recently, two studies of sex ratio variation in relation to maternal condition in roe deer (Capreolus capreolus) have reported directly opposed ®ndings for this species
and interpreted the results as support for the TWM on the one hand (Wauters et al. 1995) and the LRC hypothesis on the other (Hewison and Gaillard 1996). The TWM classically assumes that (1) the reproductive success of males is more variable and may be more strongly aected by physical quality (e.g., body size) than that of females, (2) physical quality of a male during adulthood is dependent on physical quality at the end of the period of maternal care, and (3) mothers in good condition are better able to care for their young than mothers in poor condition. Wauters et al. (1995) found a strong relationship between maternal and ospring condition, suggesting the last of the above assumptions is ful®lled for roe deer. However, the existing data for this species suggest that (1) male reproductive success is more determined by age-related levels of aggression than by body size (Strandgaard 1972), and thus larger bucks do not obtain a disproportionate number of matings, and (2) by the age of 8 months, body weight may be completely independent of body development during the period of maternal care when resources are abundant (Gaillard et al. 1993a). The LRC hypothesis assumes sex dierences in ospring dispersal behavior (or in the degree of sociospatial association with the mother) potentially resulting in competition for resources between a mother and her philopatric ospring. Available data for roe deer indicate that dispersal is more common and occurs at an earlier age for male juveniles (Strandgaard 1972; Bideau et al. 1993). In addition, strong matrilineal associations and a winter clan structure (Kurt 1968; Linnell 1994) create high potential for resource competition between mother and daughter. Both the studies of Wauters et al. (1995) and Hewison and Gaillard (1996) were based on analysis of cull data which may not have been collected expressly for the purpose of testing such models. Here, we present data from long-term ecological studies of two roe deer populations (monitored intensively for 12 and 4 years) on an index of weaning weight (an index of maternal care) and sex ratio variation in relation to maternal condition to test a number of predictions. Based on the TWM and the `modi®ed' LRC hypothesis (at the individual level): (1) roe deer mothers, when extra investment potential is available, provide higher levels of maternal care for sons than for daughters, hence (a) sons are heavier than daughters during and at the end of the period of intensive maternal care and (b) weaning weight increases with increasing maternal body weight more rapidly for sons than for daughters (TWM), or there is no sex bias in levels of maternal care (LRC); (2) roe deer mothers in better than average condition produce more sons than daughters (TWM), or more daughters than sons (LRC). We discuss the model assumptions and stress that studies of sex ratio variation must pay special attention to these assumptions before testing a model for particular species or populations.
341
Methods Study sites Storfosna Storfosna is a small (10.5-km2) island 2 km o the coast of central Norway (63°400 N, 9°250 E). The intensive study area (8 km2) covered the northern part of the island and was made up of a mosaic of moorland (33%), rough grassland (18%), cultivated pasture (35%), and woodland (12%). The climate is characterized by mild winters and cool summers, with 160±180 growing days per year. Total annual precipitation is 1048 mm and the average monthly temperature is 5.8 °C. Winter snowfall is sparse and snow rarely lies on the ground for more than a week at a time. From spring 1991 to spring 1994, between 33 and 138 radio-collared animals were present on the site each year, allowing accurate population density estimates using visual capture-recapture techniques in autumn and spring (Seber 1982). In the study period 1991±1994, there was no legal harvest and attempts were made to reduce poaching to a minimum. This caused an increase in density from 10.0 deer/km2 in 1991 to 19.2, 30.4 and 40.6 deer/km2 in 1992, 1993, and 1994, respectively (Linnell 1994). Trois Fontaines Trois Fontaines is a 13.6-km2 enclosed forest reserve in France (48°430 N, 20°610 E), managed by the Oce National de la Chasse. The forest is predominantly oak (Quercus spp.) and beech (Fagus sylvatica), with a dense understorey dominated by ivy (Hedera helix) and brambles (Rubus spp.) oering high food availability throughout the year (Gaillard et al. 1996). The climate is of a continental type, with rather cold winters (mean daily temperature of 2 °C in January and minimum daily temperature below 0 °C on 70 days per year). The roe deer population has been monitored for 20 years using mark-recapture methods and currently more than 70% of deer older than 1 year are marked with ear-tags and collars carrying numbers for individual identi®cation (Gaillard et al. 1993b). Managers remove between 60 and 100 deer each winter (January±February). Consequently, the population was maintained at an approximately constant size of around 200±250 roe older than 1 year in March and was highly productive (see Gaillard et al. 1993b, 1996). Data collection We located fawns by following radio-collared adult does closely during the fawning period (Storfosna only), or by observation of non-marked does and searches of likely areas (both study sites). We determined the age of fawns from direct observation of birth, the degree of wear of the hoof cartilage, the appearance of the umbilical cord (both sites) or the behavior of the fawn (Trois Fontaines only) (Jullien et al. 1992). Most fawns were 5 days old or less when ®rst found. During this stage, the distance between siblings is usually
