Food allocation rules vary with age and experience in a cooperatively breeding parrot

Although it is known that parents can differ in their optimal resource allocation to offspring in size-structured broods, the mechanisms determining differences in allocation rules of carers are not yet clarified. In cooperatively breeding species, breeders and non-reproductive helpers often differ in their fitness payoffs of providing care and in their breeding experience. Cooperative breeders thus provide an appropriate system to examine two hypotheses originally proposed to explain differences in food allocation among parents: (i) food allocation between carers differs because of the distinct cost-benefit ratio of selective feeding (i.e. breeders and helpers are expected to differ in food allocation) and (ii) carers differ in their ability to feed selectively (i.e. differences in food allocation are expected between experienced adults and inexperienced yearlings). We compared feeding rates with which breeders, old helpers and yearling helpers provisioned nestlings of different hatching rank. The influence of experience upon food allocation was further assessed by comparing food allocation of yearlings early and late during nesting. We show that allocation rules differ between age classes because breeders and old helpers fed the youngest chicks most, whereas yearlings showed the opposite pattern. The role of experience was supported by the fact that yearlings adjusted food allocation to that observed in experienced adults during the breeding season. We thus suggest that food allocation in El Oro parakeets depends either on differential skills of adults to transfer food to the youngest chick or on their ability to recognize nestling needs.     

Do stress hormones suppress helper reproduction in the cooperatively breeding red-cockaded woodpecker (<Emphasis Type="Italic">Picoides borealis</Emphasis>)?

In the cooperatively breeding red-cockaded woodpecker (Picoides borealis), male helpers are subordinate to male breeders and do not mate with females, even when unrelated to the breeding female within their group or through extra-group matings, yet exhibit reproductive hormone profiles similar to those of breeders. We investigated whether reproduction might be suppressed in helper males via high levels of the stress hormone corticosterone. We also examined effects of group size and season on corticosterone levels by comparing baseline and maximal plasma levels of corticosterone between helper males and breeding males, and among helper males and breeders of both sexes living in groups of different sizes throughout the reproductive cycle. We also measured plasma levels of estrogen, progesterone, and testosterone to examine other potential hormonal differences between helpers and breeders. Male status did not explain variation in any hormones; therefore, our data do not support the hypothesis that helper males are reproductively suppressed via corticosterone or the other hormones investigated. However, the presence of two or more helper males in a group tended to reduce baseline corticosterone in breeding and helper males, but not breeding females, suggesting that helper males reduce parental effort of other male group members. Seasonally, maximal corticosterone peaked during the nestling provisioning phase for breeding and helper males, but not breeder females, suggesting that males show an increased response to stressors posing a potential threat to survival of offspring.     

Seasonal variation in male-female competition,cooperation and selfish hoarding in a monogamous songbird

Both cooperation and conflict between the sexes are commonplace in monogamous mating systems. However, little is known about how cooperation and competition varies seasonally in monogamous species that maintain permanent territories. We presented territorial pairs of male and female New Zealand robins (Petroica australis) with a large supply of insect prey at monthly intervals for 2 years. Behavioural observations after food presentation were then made to quantify seasonal and sexual differences in aggressive interactions over prey, prey acquisition rates, mate provisioning, offspring provisioning, selfish food hoarding and cache retrieval. Data were used to evaluate sex-specific behavioural strategies of mediating competition for food. Results showed that males aggressively excluded females from experimental food sources year-round. Females only accessed food sources when males left them unattended. Consequently, females acquired fewer prey than males. After controlling for differences in prey acquisition, both sexes consumed similar amounts of prey in the non-breeding season. Even though males aggressively excluded females from accessing food sources directly, males fed large amounts of prey to females during the breeding season. Both sexes provisioned young at similar rates. Males cached less prey than females in the breeding season but more prey than females in the non-breeding season. Females showed similar caching intensities year-round. Although males tried to defend their hoards, females frequently retrieved male-made caches. Overall, results showed that although New Zealand robins cooperate to raise offspring during the breeding season, conflict between the sexes occurs year-round. Males and females display different behavioural strategies to gain access to experimental food sources, which appear to lessen male–female competition for food and evenly distribute food resources between the sexes.     

Who cares? Individual contributions to pup care by breeders vs non-breeders in the cooperatively breeding banded mongoose (<Emphasis Type="Italic">Mungos mungo</Emphasis>)

Knowledge of the investment rules adopted by breeders and non-breeders, and the factors that affect them, is essential to understanding cooperative breeding as part of a life-history tactic. Although the factors that affect relative contributions to care of young have been studied in some cooperative bird species, there is little data on mammals, making coherent generalisations within mammals and across taxa difficult. In this study, we investigate individual contributions to pup escorting, a strong predictor of offspring provisioning, in the banded mongoose (Mungos mungo), a cooperatively breeding mammal in which reproductive skew is low. Contributions by those under a year old (which virtually never breed) increased with age and body weight but were generally low. Among older age classes (yearlings and adults), individuals that had not bred in the current litter generally contributed less to escorting than those that had bred (with the exception of yearling males). In addition, females that did not breed reduced their investment if they were heavy presumably because such females are more likely to breed in the following event and benefit from saving resources for this. The generally greater contributions by breeders in banded mongooses contrast with the recent findings in meerkats (Suricata suricatta), another obligatorily cooperative mongoose with similar group size but wherein reproductive skew is high. Our results suggest that relative contributions by breeders vs non-breeders are not dependent on group size but on the ratio of breeders to carers and the probability that non-breeders will breed in the near future.     

Intra-group relatedness affects parental and helper investment rules in offspring care

In any system where multiple individuals jointly contribute to rearing offspring, conflict is expected to arise over the relative contributions of each carer. Existing theoretical work on the conflict over care has: (a) rarely considered the influence of tactical investment during offspring production on later contributions to offspring rearing; (b) concentrated mainly on biparental care, rather than cooperatively caring groups comprising both parents and helpers; and (c) typically ignored relatedness between carers as a potential influence on investment behavior. We use a game-theoretical approach to explore the effects of female production tactics and differing group relatedness structures on the expected rearing investment contributed by breeding females, breeding males, and helpers in cooperative groups. Our results suggest that the breeding female should pay higher costs overall when helpful helpers are present, as she produces additional offspring to take advantage of the available care. We find that helpers related to offspring through the breeding female rather than the breeding male should contribute less to care, and decrease their contribution as group size increases, because the female refrains from producing additional offspring to exploit them. Finally, within-group variation in helper relatedness also affects individual helper investment rules by inflating the differences between the contributions to care of dissimilar helpers. Our findings underline the importance of considering maternal investment decisions during offspring production to understand investment across the entire breeding attempt, and provide empirically testable predictions concerning the interplay between maternal, paternal and helper investment and how these are modified by different relatedness structures.     

A critical analysis of ‘false-feeding’ behavior in a cooperatively breeding bird: disturbance effects,satiated nestlings or deception?

‘False feeding,’ where helpers arrive at nests with food but fail to provision the young, has been reported in several cooperative species. This and other potentially ‘deceptive’ behavior has been interpreted as indicating that helping may operate as a signal within such social groups. We critically examine these phenomena in the provisioning behavior of the bell miner Manorina melanophrys. Excessively close observation distances can artificially elevate the rate of false feeding in this (and other) species, but once this had been accounted for, there was little evidence for any ‘deceptive’ behavior by helpers or breeders. Natural and experimentally induced variation in the presence of a potential conspecific audience at the nest did not have any consistent influence upon the rate of false feeds, which was low at 7.94% of 6,880 nest visits. Instead, encountering unexpectedly low levels of brood demand provided a more parsimonious explanation for those visits where helpers failed to feed nestlings or ate the food themselves. Failure to completely transfer a load to nestlings was more likely when the load contained a high proportion of sticky lerp, indicating a simple prey-transfer problem. Finally, individuals that arrived at nests without prey were often members of neighboring breeding pairs, suggesting that these few non-feeding visits may instead involve an information-gathering function. We, therefore, suggest that future studies explicitly exclude the possibility of observer disturbance and all aspects of normal provisioning behavior before applying the terms ‘false feeding’ or ‘deceptive’ and inferring anything more than straightforward helping at the nest.     

Nestling provisioning in polygynous great reed warblers (Acrocephalus arundinaceus): do males bring larger prey to compensate for fewer nest visits?

Most studies of social polygyny in birds have examined male provisioning on the basis of the number of feeding visits. This may be misleading if males compensate for infrequent visits by bringing larger prey at each visit. We investigated nestling provisioning in the socially polygynous great reed warbler, Acrocephalus arundinaceus, in south Central Sweden in 1996–1997. We collected data on rate of feeding visits, prey size and the amount of biomass delivered by males and females. Males had lower rates of feeding visits and provided smaller prey to nestlings in secondary than in monogamous and primary nests. Secondary females had higher rates of feeding visits and brought larger prey than monogamous and primary females. These results confirm that secondary females face a potential cost of polygyny through a lower rate of male feeding, and that this cost was reinforced by the significantly lower male provisioning rate (biomass h^–1) at secondary nests. Secondary females compensated for the lack of male assistance by increasing their rate of feeding and bringing larger prey. As a result, offspring in nests of secondary females received as much food as did those in nests of primary females. Prey load size increased with the parent’s proportion of feeding visits, suggesting that parents use different feeding strategies depending on their amount of responsibility for nestling provisioning. We suggest that parents which take the main responsibility for nestling feeding have to forage further away from the nest, and based on optimal-foraging theory, they should then on average bring larger prey to their nest. Received: 4 April 1999 / Received in revised form: 12 October 1999 / Accepted: 23 October 1999     

Cooperative breeding in purple gallinules: the role of helpers in feeding chicks

Summary The participation of breeders and helpers in the feeding of 21 broods of chicks was studied in a population of cooperatively breeding purple gallinules (Porphyrula martinica). In the breeding group, all birds over the age of 2 months fed chicks. Female breeders fed chicks at the highest rate, followed by male breeders and adult helpers, old juvenile helpers, and young juvenile helpers. The amount of food breeders fed chicks was independent of the number of helpers in the breeding group. However, breeders made fewer feeding visits when they had helpers. Male and female breeders spent similar amounts of time feeding chicks. Helpers spent significantly less time feeding chicks than did breeders. As helpers grew older they fed chicks at a faster rate, made more feeding visits and spent more time feeding chicks. Analysis of variance was used to determine which variables explained the variation in the brood feeding rate (amount of food delivered to an entire brood during one observation). Age of chicks had a significant nonlinear effect, and size of brood and number of helpers had significant linear effects on the brood feeding rate. Chicks in groups with helpers received more food and were accompanied for longer periods of time than chicks in groups without helpers; either or both of these factors may have led to increased chick survival.     

The role of helpers in feeding chicks in cooperatively breeding green (red-billed) woodhoopoes

Summary Observations were made of ten green (red-billed) woodhoopoe Phoeniculus purpureus flocks during the breeding season in order to quantify the relationship between flock size and the amount of food delivered to chicks. The study period was kept short specifically to minimize the effects of environmental stochasticity. Neither woodhoopoe feeding visit rates nor the total amount of food brought to chicks increased with flock size. Although nonbreeders did not increase the net rate of food provisioning to chicks, they reduced parental input in chick rearing, and hence energy expenditure by the breeding pair. However, over an 8-year study period, which includes data for 144 flock years, this did not result in increased breeding frequency or enhanced survival of breeders. There is thus no evidence that helpers' feeding contributions to young per se influence the indirect fitness of helpers.     

Incubation feeding by helpers influences female nest attendance in the green woodhoopoe,<Emphasis Type="Italic"> Phoeniculus purpureus</Emphasis>

In many bird species, the female alone incubates the eggs, but the male provides her with some food during the incubation period. In cooperatively breeding species, helpers might be expected to assist the breeding male in provisioning the female, but their contribution has been generally ignored. Here, I show that in the green woodhoopoe, Phoeniculus purpureus, the breeding male and helpers of both sexes bring food to the incubating female. The helpers did not increase the overall amount of provisioning the female received: groups of all sizes delivered a similar amount of food per hour. Helpers did, however, reduce the workload of the breeding male. Male and female helpers provisioned incubating females equally, as expected in a species where both sexes are likely to derive equal benefits from their helping behaviours. Female nest attendance was positively related to the level of group provisioning, but only in the short term. Thus, the female-nutrition hypothesis, which suggests that incubation feeding can provide an important source of energy to the breeding female, is supported in the green woodhoopoe.Communicated by I. Hartley     

Cooperatively breeding carrion crows adjust offspring sex ratio according to group composition

In cooperatively breeding species, parents should bias offspring sex ratio towards the philopatric sex to obtain new helpers (helper repayment hypothesis). However, philopatric offspring might increase within-group competition for resources (local resource competition hypothesis), diluting the benefits of helper acquisition. Furthermore, benefits of offspring sex bias on parents’ fitness may depend on different costs of production and/or different breeding opportunities of sons and daughters. Because of these counteracting factors, strategies of offspring sex allocation in cooperative species are often difficult to investigate. In carrion crows Corvus corone corone in northern Spain, sons are more philopatric and more helpful at the nest than daughters, which disperse earlier and have higher chances to find a breeding vacancy. Consistent with the helper repayment hypothesis, we found that crows fledged more sons in groups short of subordinate males than in groups with sufficient helper contingent, where daughters were preferred. Crow females also proved able to bias primary sex ratio, allocating offspring sex along the hatching sequence in a way that provided the highest fledging probability to sons in the first breeding attempt and to daughters in the following ones. The higher cost of producing male offspring may explain this pattern, with breeding females shifting to the cheapest sex (female) as a response to the costs generated by previous reproductive attempts. Our results suggest complex adjustments of offspring sex ratio that allowed crows to maximize the value of daughters and sons.     

Helper contributions to reproductive success in the splendid fairy-wren (Malurus splendens)

Summary In Malurus splendens, helpers were present in 65% of 226 group-years with at least one helper female in 37% of group-years. Most females helped for only one year, while many males did so for at least two years. Most were offspring of one or both present breeders, and in 53% of helper-years, helped both parents. For 159 helpers of known age and parentage, the mean coefficient of relatedness to the offspring was 0.47. Novice females with or without helpers produced fewer fledglings per season than females with one year breeding experience and the same level of help. Helpers did not affect production of fledglings per year by females with one year of experience. Females with two or more years experience and at least two helpers produced more fledglings than equivalent birds with one or no helpers. Experience and helpers have little effect on production of fledglings per nest but they lead to more females renesting after a first brood has been raised. Fewer than 20% of novices renest after fledging one brood, while for females with at least two years experience, the percent renesting after success is 40% with no help, 56% with one helper and 69% with 2 or more helpers. Experienced females begin their first clutch earlier than novices, and helpers reduce the time to renest after success from 66 days for an experienced female with no helpers to 50 days for females with at least two years experience and two or more helpers. Breeding females with helpers survive better (76%) than those with no helpers (55%), and helpers thus gain future indirect fitness. Despite their close relatedness to breeders and offspring, in only 19% of group-years did helpers increase their indirect fitness from an increase in productivity.     

Reproductive suppression and inbreeding avoidance in wild populations of co-operatively breeding meerkats (Suricata suricatta)

Meerkats live in co-operatively breeding familial groups in which reproduction is monopolised by a dominant pair of breeders. Offspring of the breeders are behaviourally subordinate, and typically remain in their natal group as sexually mature, non-breeding helpers. In this study, we investigated the proximate factors limiting subordinate reproduction. Evidence for reproductive suppression by dominants was investigated by comparing life history, behaviour and hormonal profiles of dominants and subordinates. Baseline levels of plasma luteinising hormone (LH) were significantly higher in dominant than in subordinate females. However, following an exogenous injection of gonadotrophin-releasing hormone (GnRH), both categories had comparable concentrations of circulating LH. There were no significant differences in pre- or post-GnRH challenge LH levels in dominant or subordinate males. Reproduction in both dominant and subordinate females rarely occurred in the absence of unrelated males. Given that groups typically comprise parents and offspring, lack of suitable mates emerged as the primary constraint on subordinate reproduction. When this constraint was removed, subordinates typically bred but at a lower rate than dominants. This difference in reproduction may be attributed to intrasexual competition manifested through direct interference by dominant females through subordinate evictions, infanticide and the abandoning of subordinate litters. We argue that differences in reproductive regulation within mammalian co-operative breeding systems may be explained by differences in the mating strategy (inbreeding versus outbreeding) and the probability that subordinates in obligate outbreeding species will encounter unrelated opposite-sex partners. Received: 19 April 2000 / Accepted: 17 July 2000     

Positive correlation between helpers at nest and nestling immune response in a cooperative breeding bird

In cooperatively breeding species, helpers may contribute to the success of the brood by increasing the number of independent offspring, but also, they may affect offspring condition and, hence, their survival and recruitment into the breeding population. This second type of benefits is rarely included in theoretical models or assessed in field studies. Immune response is a good proxy of individual quality and fitness, and there is good evidence that the performance of the immune system of chicks during the nestling phase is related to their chances of survival and future reproduction. However, no study has so far explored whether helpers at the nest might contribute to enhance immune functioning of nestlings in species with a cooperative breeding system. Here we investigate this issue in the azure-winged magpie (Cyanopica cyanus) and found for the first time a positive correlation between the number of helpers at nest and the cell-mediated immune response of nestlings. This effect was not explained by a general improvement of body condition of chicks because it was independent of individual variation in body mass or tarsus length. Our results suggest that helping can have subtle effects on the quality of offspring that may influence their survival and future reproduction.     

Kin associations and direct vs indirect fitness benefits in colonial cooperatively breeding sociable weavers Philetairus socius

Indirect fitness benefits are believed to be an important force behind the evolution of cooperative breeding. However, helpers may associate with their relatives as a result of delayed dispersal, hence, kin associations might be a consequence of demographic viscosity rather than active choice. In addition, recent studies showed that helpers may have access to reproduction therefore direct benefits might also play an important role. Here, we investigate the possible roles of direct genetic benefits and kin associations on helping behavior in the sociable weaver Philetairus socius, a colonial and cooperatively breeding passerine. We used a microsatellite-based genotyping method to describe the genetic structure within nests and colonies. Within a colony, we found considerable genetic structure between males but not females. Sociable weaver colonies have several nests that are simultaneously active, giving individuals a choice of associating with a range of first-order kin to unrelated individuals. Helpers were significantly more related to the young in the helped nests than in other nests of the colony, suggesting an active choice for associating with kin. The helpers were generally offspring or first-order relatives of one (50%) or both (43%) breeders, although more infrequently, seemingly unrelated individuals also helped (7%). We found no supporting evidence of extrapair parentage and hence no direct genetic gains from helping in our population. This strong reproductive skew is contrary to theoretical models predicting conflicts over reproduction in stepfamilies. We discuss whether female decisions and/or other direct benefits of remaining in kin associations or helping might explain the high skew observed.     

Cooperative breeding and immunity: a comparative study of PHA response in African birds

Cooperatively breeding birds might be expected to suffer from higher costs of parasitism than pair-breeding species because of two aspects of their ecology which should facilitate horizontal transmission and possibly select for higher parasite virulence: first, they interact regularly with more individuals than pair-breeding species, and second, these individuals are commonly close relatives that could share similar resistance alleles. This hypothesis predicts that cooperative breeders should invest relatively more in immune defence than closely related species which breed in pairs. I tested this prediction comparatively in African birds by examining the response of the immune system to the mitogenic lectin, phytoahemagglutinin (PHA response) in relation to cooperative breeding. Among 66 species, of which 18 breed cooperatively, PHA response was significantly higher in cooperatively breeding species. This association appeared not to be confounded by body size, clutch size, nest position, coloniality or similarity owing to common phylogenetic descent. These results suggest that cooperatively breeding birds may have been selected to invest more than pair-breeders in defences against parasites. If so, then additional costs of philopatry and helping behaviour might be imposed on breeders, helpers and offspring.     

Cooperative breeding by the Galápagos mockingbird,Nesomimus parvulus

Summary The costs and benefits of helping behavior were analyzed for 36 pairs of the Galápagos mockingbird, Nesomimus parvulus, and their associates. Helping at the nest is usually done by sons or males suspected to be offspring of the breeders. Costs and benefits to breeders were assessed by comparison of pairs with and without helpers, and costs and benefits to helpers were assessed by comparison of birds which help and those which establish themselves as novice breeders.Helping behavior benefits breeders by increasing fledging success and by reducing the adult energy load in territory defense and feeding of nestlings. Breeders assisted by helpers may also benefit by decreased nest predation. Helpers enhance their inclusive fitness by helping, and gain directly by increasing their chances of securing a territory. Helpers do not appear to gain any fitness advantage from the experience of assisting, nor do they increase their survivorship by remaining on natal territory.Ecological and demographic features such as saturated territories and low territory turn-over rates due to high adult survival may be primarily responsible for the evolution of the helping behavior, with kinselection reinforcing it. Associated features of this system are a male-biased population sex ratio, a greater energetic benefit to breeding males than to breeding females in having helpers, earlier dispersal and breeding by females than by males, and much more frequent helping by males than by females. These are interpreted as consequences of brothersister aggression that indirectly minimizes the chances of inbreeding.     

Multiple paternity in the cooperatively breeding fish <Emphasis Type="Italic">Neolamprologus pulcher</Emphasis>

In cooperative breeders, mature males may compete for fertilizations. In this study, we measured the degree of multiple paternity in a natural population of a cooperatively breeding fish. Neolamprologus pulcher (Perciformes: Cichlidae) is a highly social cichlid endemic to Lake Tanganyika. We used highly variable microsatellite loci to survey 12 groups with an average number of 10.6 brood care helpers per group and a total of 43 offspring (mean 3.6 per brood). In 11 of 12 groups, all young were assigned to the dominant female. The dominant male sired all offspring in three groups, part of the offspring in four groups, and in five groups, he had no paternity at all. In total, 44.2% of young were not fathered by the current male territory owner. Multiple paternity was found in 5 of 12 broods (41.7 %), with 8 of 35 young (22.9 %) being sired by males other than the respective territory owners. This is an exceptionally high rate of extra-pair paternity among cooperatively breeding vertebrates. Neither helpers present in these territories during collection nor neighbouring males were unequivocally assigned to have sired these extra-pair young. However, behavioural observations suggest that male helpers may have produced these young before being expelled from the territory in response to this reproductive parasitism. We discuss these results in the light of reproductive skew theory, cooperative breeding in vertebrates and alternative reproductive tactics in fish.     

Effects of helpers on breeder survival in the red-cockaded woodpecker (Picoides borealis)

Helpers can gain future indirect fitness benefits by increasing the survival of breeders that produce offspring related to the helper. Helping augments group size through the helper's presence and, in some cases, by increasing fledging success. Breeders may then experience enhanced survivorship because of the benefits of living in large groups. Helping may also reduce the workload of the breeder, which in turn may increase the likelihood that the breeder will survive to breed again. We used Cox's proportional hazards model to examine whether breeders' survival in two populations of the red-cockaded woodpecker (Picoides borealis) was enhanced when group size was increased in the presence of (1) the helper itself, or (2) "extra" fledglings (fledglings produced by the breeder because of helping behavior). We found that in the presence of helpers, the risk of a breeder dying declined by 21–42% for males and 0–14% for females. Our results suggest reduced breeder workload as one mechanism to explain reduced breeder mortality in the presence of helpers: breeders spent less time incubating and provisioning nestlings when assisted by helpers. The risk of a breeder dying declined by 16–42% in males and 26–43% in females in the presence of "extra" fledglings. We speculate on possible mechanisms by which fledglings might affect breeder survival. Our results support the hypothesis that helpers gain future indirect benefits by reducing breeder mortality. Electronic Publication     

Helpers-at-the-nest have the same provisioning rule as parents: experimental evidence from play-backs of chick begging

Despite the success of kin selection in explaining helping-at-the-nest among communally breeding birds, we know almost nothing about how helpers regulate their chick-feeding effort. This is especially interesting given how much we now know about parental provisioning `rules-of-thumb' and the evolution of chick begging as an honest signal of `need'. This study explores the provisioning rules of helpers and parents in Arabian babblers (Turdoides squamiceps), using tape play-backs to supplement chick-begging signals and increase apparent brood demand. In all eight groups tested, both helpers and parents fed older, noisier broods at higher rates. Total provisioning rates to nests increased during begging play-back days compared to control days. Absolute provisioning rates by helpers and the scale of their responses to play-backs were statistically indistinguishable from those of parents. In both helpers and parents, increases in nest visits during play-backs were associated with reductions in foraging distance from the nest and increases in size of prey delivered. Older birds of both sexes delivered slightly larger prey items, possibly reflecting differences in foraging ability due to experience. These results are consistent with the idea that, like the parents, helpers-at-the-nest in Arabian babblers provision nestlings as part of a strategy of investment, irrespective of helper age, dominance status or sex. In this species, high relatedness within groups may provide parents and helpers with similar kin-selected fitness benefits, although the mutualistic advantages to helpers from simply augmenting group sizes cannot be ruled out. Received: 17 June 1997 / Accepted after revision: 28 February 1998