AND COLOUR SELECTION PREFERENCES IN LESSER SNOW GEESE by
F. COOKE and C. M. McNALLY 1) (Department of Biology, Queen's University, Kingston, Canada) (With I Figure) (Rec. 10-V-1974)
INTRODUCTION role of early learning in species recognition has interested many ethologists. When a species is polymorphic for some externally recognizable feature, such as plumage, there may be some confusion between recognition of morph and recognition of species. In an attempt to investigate the importance of early learning of plumage colour to the behaviour of the dimora long term field and exphic Lesser Snow Goose (Anser caerulescens) was set in perimental program up 1968. The Lesser Snow Goose occurs in two colour phases, a white morph and a grey-blue morph (referred to hereafter as the blue morph) which are quite distinct from one another, although the blue morph is variable in terms of the amount of white on the belly. Non-random mating occurs with like phases pairing more frequently than would be expected by chance (COOCH & BEARDMORE, 1959). The plumage dimorphism is inherited as a single gene (or co-adapted gene complex) difference (COOKE & MIRSKY, 1972). It was postulated by COOKE & CoocH (1968) that Snow Geese select mates which have a plumage colour similar to that of one or both of their parents. Early learning of parental colour by the young birds, which in a monomorphic species would ensure that the birds would always pair with a member of their own species, would lead in a dimorphic species to apparent positive assortative mating and to the maintenance of the polymorphism (KALMUS & MAYNARD SMITH, 1966; SEIGER, 1967). It is difficult in field studies to The
1) We would like to thank the National Research Council of Canada and the School of Graduate Studies, Queen's University, Kingston, Ontario for financial assistance in of the Ontario carrying out this research; Mr R. N. JONESand Mr W. N. CARRICK Waterfowl Research Foundation for their assistance in housing the geese; the Canadian Wildlife Service for assistance in collecting eggs from the wild; Mrs. E. Sinclair for for looking after the geese and Mr H. BOYD,Dr P. COLGANand Mr R. ROCKWELL helpful advice in the preparation of this manuscript.
152 elucidate the role of colour in determining mate selection and group associations,. However, a series of experiments have been carried out which the of sib and role self, investigate parental colour on the behaviour of test birds. The Snow Goose is particularly for a number of advantageous reasons. Pair bonds are usually maintained throughout the life of the birds (PREVETT, 1972). Young usually remain with their parents for a year in the wild and even longer in captivity. The young are precocial and can readily be fostered to adult parents with plumage colours to suit the needs of the experimenter. COOKE, MIRSKY & SEIGER (1972) showed that goslings tested in a choice situation at 6 and 14 weeks of age showed a significant preference for goal birds with a plumage colour similar to that of their foster parents. They also showed that when the colour of a foster parent was changed, either by dyeing the foster parent pink or by providing a new foster parent of a different colour, the preference for the colour of the original foster parent is lost. The investigations reported here were designed to determine whether: a) sibling colour also,has an effect on colour preference b) the preferences shown in a test situation are demonstrated in a free choice situation and in mate selection c) the responses to parental plumage colour change with time d) auditory stimuli modify the visual responses in a test situation. METHODS Each year since ig6g, eggs collected from the wild at the La P6rouse Bay, Manitoba Snow Goose colony have been sent to the Ontario Waterfowl Research Foundation at Guelph, Ontario where they were incubator hatched. Four even-aged flocks, designated A, B, C and D, have been established which were hatched in ig6g, i97o, Ig7I and 1972 respectively. Experiments described in this paper were carried out with birds from flocks A, B and D. Rearing of birds. On hatching, eggs were transferred to hatching trays where each gosling was kept in visual, but not auditory, isolation. At approximately io hours of age, the goslings were placed in family groups in outdoor pens. Each pen contained a brooder, a group of evenaged goslings (hereafter referred to as a peer group) and except for flock A, one or two foster parents. The peer groups were formed according to the time of hatch and were not necessarily related to the original clutches. Burlap surrounding the lower 3 feet of the pens kept the families in visual but not auditory isolation. Except for daily maintenance, the families were left undisturbed. The groups remained isolated for four weeks for flocks A and B, and for six weeks for flock D. The families were then combined in a 6-acre 'rearing' field. In the rearing field, families remained intact and families of a particular year class usually moved as a single flock. The family structure differed in each year class. Flock A was raised in 3 peer groups, with family sizes of 30-40. These families were exposed to two parental plumaged birds only from week 2 to week 4. Flock B consisted of II peer groups with
153 family sizes of 9-39, each with one foster parent which remained with the young birds for one year. Birds of this flock were either blue phase, white phase or white phase dyed pink. In each family the foster parent was a different colour from the peer group (e.g. a blue foster parent with white goslings). Details of the family sizes and composition are given in COOKEet al. (1972). In flock D, the family size was much smaller (4-1 I) and each family had two even-aged adult plumaged foster parents from flock A. In all but one pen, the foster parents were a mated pair, though none had ever successfully nested. for colour _ preferences. Testing To determine whether young birds acquire a preference for a particular colour as a result of earlier associations, the following tests were performed. Birds from flock D were tested 9 and I weeks after placement in the rearing field. One third of the flock was tested on the first trial and the remainder on the second. No bird was tested more than once. Birds were tested in an outdoor circular area, 35 ft in diameter and 8 ft in height. A plan of the arena is shown in Fig. 1. The divisions consisted of bars on the ground and did not impede movement within the arena. Four caged adult birds corresponding to the four parental types, white male, white female, blue male and blue female, were placed at equal intervals around the periphery of the arena. These goal birds were rearranged at io trial intervals to counter pen bias due to light and shade effects and to mix parental colour combinations. The goal birds were of comparable docility and were unknown to the young birds which were being tested, thus precluding the possibility of individual recognition. The sides of the arena were covered in burlap and the observer who viewed the activity through a small hole was not visible to the birds.
Fig. i. Test arena with four goal sections containing cages (C) for adult plumaged geese. The test goslings were released from a closed box (B) in the centre. Arena diameter was 35 ft. Test goslings were released from a closed box in the centre of the arena and their preferences scored. A choice was considered to have been made if a test bird spent a predetermined time in any goal section. The time prerequisite varied from 1.5 consecutive minutes if the test bird lay down immediately adjacent to a goal bird, to 3.0 minutes if it stood away from the adult cage but remained within the goal section. A bird making no clear choice within 5 minutes was removed and not retested. Birds from flock B had been tested and the results of these tests were reported by COOKEet al. (1972). One third of these birds, however, were retested two years later to see if the colour preferences shown by the geese as goslings were maintained up to the age of pair formation. The experimental procedure was identical to that described in the earlier paper and differs from that used for flock D only in the size of the arena and the fact that 3 goal birds, one white, one blue and one dyed pink, were used.
154 Nearest analysis (KUMMER, ig6g). neighbour certain affinities Within a particular flock, birds do not associate randomly, but for other birds based on family relationships and colour preferences. In order to measure these affinities within the flock, nearest neighbour analyses were conducted. The flocks usually remained separate from one another except at feeding time and they spent a considerable time loafing. Each bird was individually recognisable through coloured bands on the legs. The procedure for the nearest neighbour analysis was to observe a particular bird and record its three nearest neighbours. A second bird was selected and its nearest neighbours recorded and so on, until every bird in the flock had been recorded. This was time consuming but was usually completed within a single day. This procedure was repeated at three day intervals for several weeks during the spring and summer. We realize that observations are not independent in the strictest sense since we sampled from a finite population without replacement. Flock B and flock D were analysed in this way. Towards the end of the period, some birds in flock B had selected their mates. Mate selection. To obtain information on the nature of mate selection, birds which formed pairs during the study were recorded. Behaviour which characterizes the pair bond is distinct from other goose behaviour. Criteria which were used to determine pair formation were triumph ceremonies (FisCHER, 1965), territorial defense and the continuous association of the two birds during feeding, preening and walking so that the two seemed to be moving with one accord. No one criterion was considered as sufficient evidence of pair formation. Birds from flock A and B formed pairs during the study. RESULTS Family
Within a day of being introduced into the family pens the goslings from flock D exhibited an almost continuous following response to the rearing foster parents. Generally, the initial parental response was poor with the adults fleeing and trampling the goslings when approached by the experimenter. This is readily understood as the birds were unfamiliar with their and inexperienced with goslings. Within three weeks, howsurroundings ever, most of the foster parents were exhibiting good parental responses as vocalcharacterized by brooding, defensive posturing and corresponding Characterizations when the family was approached by the experimenter. istically, these behavioural sequences occur in the Snow Goose only during incubation and early rearing of the young. The development of a good parental response to the imposed goslings was probably due to the fact that mated pairs were used as foster parents. In the one instance where a non-mated pair was used, although the birds tolerated their imposed association, the parental response was the poorest. The response of the mated foster parents was also much improved over that exhibited in the earlier experiment where single adult birds were used as surrogate parents (COOKE et al., 1972).
the families were combined in the rearing field, they remained as intact units. Each family fed, preened and moved as a group, together within the flock structure. usually The observations both in the family pens and in the rearing field suggest that despite artificial rearing conditions, good family structure resulted which was strongly reminiscent of that found under natural conditions. Approach
These tests were applied to flock D goslings and to birds of flock B, which had been tested 2 years earlier and reported by COOKE et al. (r97z). In the earlier tests, where one parental bird was present with 9-39 siblings. in each family (except a control family of pink birds) the parental colour differed from the peer group colour. Birds from flock B when tested at the ages of 6 and 14 weeks showed a significant preference for unknown birds of the same colour as their parents. After this initial test the birds from flock B were kept with their parents for one year after which the parents were removed to simulate conditions in the wild. After the removal, the peer groups remained together for one more year before being tested again. As in the previous test the birds were presented with a choice between the colours white, blue and pink. Results of the present test are recorded in Table I. As in the previous run, the birds chose preferentially but instead TABLE Colour pre f erences of one
third of the
Sibling choices are underlined. W o - no clear choice.
1970 flock tested at two years of age
white phase, B -
blue phase, P -
of showing parental colour preferences, they chose on the basis of peer colour. 16 Of the birds making a choice, II chose peer group colour. group A binomial test was performed and a highly significant probability of .004 obtained. Responses are not random but directed to birds with sibling or self colouration. The total number of birds responding was low, only 16 of 83 (19%) making a choice. There are a number of possible explanations
156 for this. Flock B test birds were mature but not in the breeding stage and there may have been little incentive to approach a strange bird. The lack of responses could also indicate that in most birds the parental colour preference was lost and not substituted by an alternate preference. There may have been a conflict of preference with the result that no choice was made. Family groups in flock D were smaller and consisted of two parental birds. Thus some of the weaknesses inherent in the design of the experiments with flock B were eliminated. On the other hand, because of shortage of facilities the sample sizes were smaller. Family categories and peer group sizes are listed in Table 2. Families where parental and gosling
TABLE FamiLy categories and the sizes of the
2 peer groups formed
for Flock D
* Due to the shortage of Blue X Blue mated pairs this family received two non-mated birds. colour are the same and where parental and gosling colour differ from one another were set up to discover the relative importance of parental and and later mate selection. sibling colour in determing colour preferences Families where one parent was blue and the other white were used to assess the relative importance of maternal and paternal colour on gosling preferences. Replicates of each of the six family combinations were made. Results of the tests are shown in Table 3. To test whether the goslings showed a significant preference for a goal bird with the same colour as its foster parent, the families denoted in the first four rows of Table 3 were used. (The final two must be omitted since these goslings had parents of both colours). From these families, of the 31 birds making a choice 25 chose parental colour. A Z test with continuity correction was performed giving a value Z = 3.20 which is significantly greater than would be expected by
3 birds colour preferences of young for the trials 9 and ii weeks after flock formation
W - white phase, B - blue phase, 0 - no clear choice. The underlining refers to the expected choice if parental colour is chosen. = .0007). Thus under test conditions, most of change alone (probability the goslings which show a response move to a goal bird with the same colour as their foster parents. While the responses made are strongly in the direction of parental colour, a large number of birds showed no preference. This reduced the sample size available for statistical analysis, and makes further rigorous analysis impossible. The results are, however, suggestive. When parental and peer group colours are the same, the birds chose that choices to 2 incorrect. When colour almost without exception, 17correct peer group and parental colour differ, the choice is still in the direction of so, 8 correct choices to 4 incorrect. parental colour, but not significantly From this experiment alone no conclusions about the relative importance of parental and peer group colour are warranted. However, in the earlier experiment of COOKE et al. (1972) where also parental colour and peer group colour differed, it was found that 69 birds chose parental colour and 36 peer group colour. Even allowing for differences in the experimental design, it seems that choice of parental colour is stronger when the peer group is also of the same colour. Thus sibling and/or self colour appears to modify the preference choice. When peer group and parent colour are the same the probability of making a correct choice is increased. When the two colours differ the preference for parental colour is less pronounced. In the cases where families were composed of a mixed pair and mixed goslings, the response is random. That is, there is no preference for either
158 blue or white. This does not mean that there has been no earlier learning as a parental and/or peer group as either choice would be interpreted response. The colour preference test was repeated when the goslings were r month old. They had been in family groups in a large field throughout and Table 4
at z months
Underlining refers to expected choice if parental colour is chosen. shows that the proportion of birds responding and the types of response made were similar to the earlier test. Although the pattern of response was similar in the two experiments, individual birds did not always respond the same way in the two experiments. In the approach response test reported in Table 3, 2 male and 2 female birds were used as goals. Of the 48 goslings making a choice, 24 chose a male goal and 24 a female. There is no indication therefore that the sex of the goal bird was important to the test goslings. In this second test, an attempt was made to see if an auditory stimulus associated with the goal bird would modify the response of the test bird. Speakers emitting calls recorded from white phase birds were placed behind white phase goal birds and speakers with calls from blue phase birds behind blue phase goal birds. To the human ear there is no indication that the sounds of the two colour phases differ systematically from one another. 72 of the test birds were tested with the auditory response but the proportion of birds responding and the types of response made did not differ from those made when no auditory response was applied. Under the conditions of the experiment, the auditory cues did not modify the behaviour of the birds, but sample size was small and this experiment must be regarded as preliminary.
analysis. neighbour The results of the nearest neighbour observations during the first year for flock D are given in Table 5 as number of associations with peer group, replicate peer group and goslings from other families. From these observations it is clear that the young birds are associating primarily with birds within their own families. This confirms the casual observations of intact family units. Three pairs of families occasionally mixed. Both all-white families associated with one another as did both all-blue families. One of the families with blue parents and white goslings associated with one of the families with a blue male and white female parent and mixed goslings. Thus, although the associate sample is small, there is a tendency for families to preferentially with other families of a similar colour composition, white with white, blue with blue, and mixed with mixed. TABLE
Flock D. Nearest neiglabour analysis of goslings; number of observations where nearest neighbour was a) a member of the birds own peer group b) a gosling from the replicate peer group or c) a gosling from another family; together with expected nuznber assuming random associations
Nearest neighbour observations were conducted on flock B as one and two year olds. The types of birds with which the birds associate fall into three categories for the purpose of this analysis: i) birds from the same family (SF) which, since in flock B all siblings within a family were of the same colour, are of necessity the same colour as the test bird, 2) birds from a different family but of the same colour as the birds own colour (SC) and
160 the colour of his siblings, and 3) birds from a different family and also a different colour (DC). It should be recalled that the foster parents were removed from these families at one year of age. Table 6 presents the three nearest neighbour observations as the percentage of observations spent with all i are with birds of the same family SF, SC and DC. Of the I families, TABLE
with neighbour analysis of Flock B as percentage of observations colour and other peer group (SF), replicate group (SC), fayvtilies (DC) at one and two years of age
* Blue families 1-3 had white parents. Pink families i and 2 had pink parents; 3 and 4 had white parents. White families i and 2 had blue parrents; 3 and 4 had pink parents. less frequently at two years of age than at one year of age. A sign test shows that the probability of obtaining such a difference by chance is less than .006. There is, therefore, an increasing tendency as the birds mature to spend time with non-family members. They nevertheless still spend most of their time with birds of their own family. This tendency may be important in that it would increase the probability of choosing a mate from a different family and decrease the possibility of inbreeding if these were natural sibships. Table 6 showed that as the birds became older they spent less time associating with birds of their foster family. Table 7 presents a statistical analysis of these associations in terms of family relationship and colour. The questions asked are whether the birds are associating randomly with other members of the flocks. First, how much time do they spend associating
7 B at two years of age. Nunaber off Nearest neighbour analysis of Flock observations of associations with birds of the same family (SF), same colour but different family (SC) and different colours and different family (DC). Relative numbers of associations compared
with members of their own family in comparison with non-family members, which are the same colour as themselves and their family members? Second, when they are associating with non-family birds, do they associate randomly in respect to the colour of the non-family birds? For each family two X î values were obtained, first by comparing SF with SC to answer question i and second by comparing SC with DC to answer question 2. The X 1 values are calculated as follows: expected values are calculated by assuming that a test bird chooses family and non-family birds randomly within a colour. Thus a particular bird in blue family i has 9 members of his peer group with whom he could associate since the family size is 10, and 42 blue nonfamily birds since blue families 2 and 3 contain 23 and 19 birds respectively. Since there were 873 observations and 51 birds with which to associate, one would have expected that 9/51 would be with SF birds and 42/51 with SC birds. Expected values thus calculated are compared with observed and Xi values recorded in the final column of Table 7. The X i values are highly significant for each family and show that the birds associate with their peer group much more frequently than would be expected by chance. The data also show that when the birds associate with birds other than their siblings they associate much more frequently with birds of a similar colour to that of their peer group. A mechanism based on learning of familial plumage colour would account
162 for the large numbers of associations of these birds with birds the same colour as their siblings. The evidence for such colour learning has been presented in this study and by COOKE et al. (1972). If colour preferences influence mate selection then the strong attachment to family members as well as the tendency to associate with birds of a colour similar to family members will have an important influence on mate selection. selection.
Mate Flock A.
The results of pair formation in flock A at the ages of two, three and four years respectively are given in Table 8. Expected values reflect the TABLE Pair
formation in flock A at two, three and four years of age. Observed and expected values on the assumption of random pairing
number of each class of pair expected assuming the paired birds had selected mates at random. The small number of blue X blue pairs reduces the value of the X2 tests. The data cannot be lumped since many of the same birds are involved in successive years. Changes in the number and frequency of pairs reflects a) increasing number of birds becoming paired with age b) decreasing number of birds due to mortality. Since one of the classes in each year class was below 5, a probability value of 0.01 was chosen to indicate significance. X2 tests performed on the data to determine whether or not pairing was random with respect to colour gave values of 0.38, p > 0.5; 0.31, p > 0.55 and o.89, p > 0.25 for the three age classes respectively. As there is no between the observed and expected results on the significant difference of random assumption mating, it can be concluded that pairing in the flock is random in terms of plumage colour. These results contradict those found for wild populations where mating is positively assortative (CoocH & BEARDMORE, 1959).
163 Flock B. The results of pair formation in Flock B are presented in Table 9. To test whether or not pairing in this flock was random with respect to colour, conwere performed and a X 5 value of 23.6; p < .ooi was tingency X tests obtained. This value is significantly greater than would be expected by chance alone. TABLE
Pair formation in flock B at three years of age. Observed and expected values assuvning random mating
to account for the non-random are two possible explanations pairing. The birds may be selecting mates on the basis of a preference for for sib/self colour. The experiment was parental colour or alternatively that if birds such chose mates on the basis of the colour originally designed of their foster parents, blue X white and pink X white crosses would predominate, whereas if birds chose mates on the basis of selb/sib colour blue X blue, white X white and pink X pink crosses would predominate. There was a significant excess of matings based on sib/self colour. Combining all sib/ self matings on the one hand and the remainder on the other, a contingency a value of 19. I ; p G .oo was obtained. Thus, birds choosing mates were choosing predominantly according to sib/self colour. As both sib and self colouring were the same, this experiment cannot measure the relative importance of these two parameters. However, the strong evidence against self colour influencing mate selection in flock A suggests that it is sib colour which flock B birds are preferentially choosing. There
' The present study Lesser Snow Geese. response. The first approach goal birds
DISCUSSION examines three separate responses exhibited by captive These are the approach, the association and the pairing refers to the tendency of birds in a test situation to of a colour similar to that with which they are most
164 familiar, the second to the tendency of birds to associate in groups with birds of a colour similar to that with which they are most familiar, and the third to the tendency of birds to choose mates with the preferred plumage colour. The approach response test on flock D in conjunction with the earlier work of COOKE et al. (1972) provides evidence of colour as a recognition cue. Under the stress of a test situation, young goslings have been shown to approach unknown birds of plumage colour similar to that of their foster parents more frequently than expected if chance alone accounted for their choice. The test Snow Geese had been exposed in the rearing field to both colour phases as well as a variety of other stimuli. The presence of these alternate stimuli did not negate the importance of parental colour. Parental colour is the single most important factor determining colour preferences in the approach response situation. When parental and sib colour are the same the colour preference is most pronounced. When parental and sib colour differ parental colour preference still exists but is diminished in strength. While the adults with which the gosling has been most closely associated are This is not unimportant, siblings also influence the colour preference. expected. The development of a preference requires an emotional attachment of exceptional strength for the preferred object (IMMELMANN, 1970). In Snow Geese strong bonds develop not only between parents and young, but also among siblings. Triumph ceremonies, intimately involved in maintaining pair and family cohesiveness and enhancing individual recognition, are performed by all family members. Therefore, it is to be expected that both parents and siblings should influence colour preferences. In the families in which parents were a mixed pair and young consisted of both blue and white goslings, no preferences were found. Choices were random with respect to parental colour and sex. These results suggest that neither parental sex was more important in determining colour preferences. The results differ from those of ScHUTZ (1963) who found the foster mother to be the most effective imprinting object in mallards (Anas platyrhynchos), but are readily explained by the nature of Anser family structure. While in the Mallard only the females raise the young, in Snow Geese both male and female are involved, which suggests that in this species both male and female are equally effective imprinting objects. This was verified by MIRSKY (I97I) who found that goslings with a single male foster parent responded in colour preference tests equally as well as goslings with a single female foster parent. Also the fact that male and female Snow Geese are similar in appearance reduces the likelihood of one of the sexes being more important in determining colour preferences (MANNING, 1967). Whether this holds true at pair formation remains to be shown.
165 The testing of flock D has shown evidence for colour preferences as a result of early association with foster parents and peer groups. For this early preference to be reflected in the selection of mates, it must be maintained until sexual maturity. Stability
from all One of the reasons why imprinting was considered different other types of learning was that it appeared to have longterm effects, determining not only the objects toward which the young animal directed its filial response but also those to which its sexual responses were directed. Imprinting was considered irreversible, restricted to a sensitive period in early life and dependent on the primacy of the imprinting stimuli (LORENZ, 1937; HESS, 1959, 1964). The irreversible nature of early learning has undergone considerable criticism and an increasing quantity of evidence now suggests that such "imprinting" may indeed be reversible at least in certain instances (COFOID & HONIG, 1961; BOYD & FABRICIUS, 1965; KLOPFER, 1965). COOKE et al. (1972) have shown that the colour preferences of goslings can be altered if the colour of the foster parent is changed. The results of the present study indicate that filial preferences of Snow Geese may be even more labile than has previously been supposed. Birds which as goslings preferentially approached parental colour in a test situation when retested at 2 years of age had changed their preferences to peer group colour. This is readily explained by the fact that these birds had been recently associating with their sibs and with birds of sib colouring, but not with their parents who had been removed from the families after one year. It appears then that colour preferences can be changed not only at an early age, but also as a result of later adolescent with other species of birds support this experience. Several experiments finding (GOODWIN, 1948; GUITON, 1961, 1962; MOLTZ, 1960; WALLER & of the colour preferences of Snow WALLER, 1963). In the determination Geese recency supersedes primacy under the conditions in which the experiments were carried out. A change in the imprinting stimulus before sexual maturity could result in an alteration of preferences. Therefore, the continuous association of the young with their parents, sibs or birds with a similar plumage would be a prerequisite for maintaining a colour preference and for choosing mates according to the colour of the parents (and/or sibs). The unusual family structure in flock B where parental and offspring colour differed would rarely occur in nature due to the inheritance of colour in the Snow Goose. It would be unlikely for a gosling not to have at least one parent of its own colour. In most families parents and goslings are the
166 bond which occurs same colour, and the breaking of the parent-offspring after one year would probably not produce a change in colour preferences. In wild populations the association with parents through the early sensitive period and early adolescence and with birds of the same colour as parents and sibs in later adolescence (CoocH, 1958; CoOKE & CoocH, 1968; COLGAN COOKE & SMITH, 1974) would be expected to play an important role in maintaining the birds' preferences for parental colour to sexual maturity. The outcome of this parental colour preference at pairing would be assortative mating. Random
mating. Thus far, the discussion has assumed that the assortative mating pattern in the Snow Goose is the result of an acquired preference for a mate with plumage similar to the bird's parent. There is the possibility that the learned filial preferences do not measure mate preferences. LORENZ (1937) pointed out the possibility of unlearned preferences affecting mate selection. For Mallards, ScHUTZ (1963) found that females mated with their own species regardless of their early experience. He suggested that the preference of the female Mallard for males of her own Snow Goose where both species is innate. In the sexually monomorphic parents raise the young, there is ample opportunity for acquiring a sexual preference and inherent preferences seem less likely and perhaps less necessary. The mate selection results of flock A support this hypothesis. Flock A was kept with foster parents only from week 2 to 4. After this time the goslings were reared together in a large field and behaved as a single large family. In colour preference tests at seven months of age they showed no preferences (MIRSKY, 1971). In view of this, it appears that the constant association over a period of time is required for the development and maintenance of a colour preference. Flock A either did not develop colour preferences during the brief association with foster parents, or lost them during the lengthy period without foster parental influence. This was also reflected in their pairing behaviour. White pairing in this flock does not reflect the assortative mating pattern found in wild populations, it does rule out selfing as an explanation for assortative mating. If birds chose mates of the same colour as themselves, assortative mating would result regardless of rearing procedure. This was not found. In the absence of constant association with parents and distinct peer groups, pairing was random. This is significant for it shows that in the Snow Goose the observed pattern of assortative mating is not inherent. were acquired and In flock B on the other hand, colour preferences
167 assortative mating took place. However, the pattern of assortative mating was not as expected. Instead of choosing mates according to parental colour, the birds chose according to peer group colouring. Both the nearest neighbour analysis and the approach response tests showed that birds from flock B, when making a choice, showed a distinct preference for peer group colour. This was also true for mate selection. It appears then that the recency of the in determining colour preferences, and experience was most important these birds chose mates according to the colour of their most recent associates. The apparent lack of support for the parental hypothesis in this experiment is probably an artifact due to the nature of the experimental design. In wild populations where parental and gosling colour are usually the same, the association of sibs after the break of the parental-offspring bond (PREVETT, 1972) would reinforce the preference for parental and sib colour. There are a number of reasons for believing that both parents and sibs influence mate selection in wild populations. One reason is the nature of the family structure. The development of intense social bonds would seem a for the determination of long lasting sexual preferences. In prerequisite estrildine finches which develop strong bonds between parent and young, but weak sib attachments, mates are chosen from the foster parent species even when siblings have been conspecifics. In Bullfinches, on the other hand, sib bonds are exceptionally strong and preferences are determined primarily through the presence of sibs (IMMELMANN,1970). Snow Geese seem to be somewhere in between. Strong bonds develop between adults and young and among sibs. This was shown in both the approach and association responses. In flock B the parental and sib hypothesis for mate selection are mutually exclusive, but for natural populations they are not. It seems likely that both parents and sibs influence mate selection. However, the relative importance of either component is not known. The pairing of flock D should provide some answers to this question as well as to the question of the relative importance of paternal and maternal colour on mate selection. SUMMARY The assortative mating which has been found in the dimorphic Lesser Snow Goose, Anser caerulescens caerulescens, could be explained if birds chose mates according to the colour of their parents and/or sibs. Three distinct captive flocks were tested for colour preferences in terms of: 1) approach response, 2) association preferences, and 3) mate selection. 1) Approach responses. Young birds placed in a choice situation had a significant preference for birds of the parental colour. Sib colour when different from parental colour appeared to modify their
168 choice. If parents were removed during adolescence the early colour preferences could be altered, the most recent associations determining the preferences. The pattern of response did not change in tests carried out at the age of 3 months and repeated at 11 months. The presence or absence of auditory cues did not alter the pattern of response. No differences were detected in the responses of gosling to maternal versus paternal colour. 2) Association preferences. In an open field situation, birds usually associate with their peer group at both one and two years of age. The degree of association with the peer group is less at two years than at one year. When birds associate with non-peer group birds they show a distinct tendency to associate with birds which are the same colour as their peer group. 3) Mate selection. In the flock which was raised as a single large group, with virtually no parental contact, pair formation did not depart from randomness in terms of colour, suggesting that non-random mate selection in Snow Geese is not an inherent property but is a function of their prepairing experience. In the flock which consisted of families where the foster parents were of one colour and the sibs of a different colour, and in which the foster parents were removed after one year, the pairing at two and three years of age was non-random; pair formation reflected preference based on sib/self colouring. It is concluded that familial plumage colour does influence colour preferences in terms of approach response, association preferences and mate selection. However, continuous association with birds of parental plumage colour is a prerequisite for this colour to influence mate selection. If the parent is removed (as happens in the wild) the colour preference may be altered but it is more likely that the preference will be maintained through association with birds of familial plumage colour. Thus, directly or indirectly, parental colour will influence mate selection. REFERENCES E. H. & Observations on the incidence of following of visual FABRICIUS, (1965). BOYD, and auditory stimuli in naive mallard ducklings (Anas platyrhynchos). - Behaviour 25, p. 1-15. COFOID,D. A. & HONIG,W. K. (1961). Stimulus generalization of imprinting. - Science 124, p. 1692-1697. COLGAN, P., COOKE,F. & SMITH (1974). The analysis of group composition in assortatively mating populations. - Biometrics 30, p. 693-696. COOCH,F. G. (1958). The breeding biology and management of the Blue Goose. Ph.D. Thesis, Cornell, Ithaca, N.Y. -& BEARDMORE, J. A. (1959). Assortative mating and reciprocal difference in the Blue-Snow Goose complex. - Nature 183, p. 1833-1834. COOKE,F. & COOCH,F. G. (1968). The genetics of polymorphism in the goose Anser caerulescens. - Evol. 22, p. 289-300. -& MIRSKY,P. J. (1972). A genetic analysis of Lesser Snow Goose families. Auk 89 (4), p. 863-871. & SEIGER,M. B. (1972). Colour preferences in the Lesser Snow Goose and -,their possible role in mate selection. - Can. J. Zool. 50, p. 529-536. FISCHER,H. (1956). Das Triumphgeschrei der Graugans Anser anser. - Z. f. Tierpsychol. 22, p. 247-304. GOODWIN,D. (1948). Some abnormal sexual fixations in birds. - Ibis 90, p. 45-48. GUITON,P. (1961). The influence of imprinting on the agonistic and courtship responses of the Brown Leghorn cock. - Anim. Behav. 9, p. 167-177. The development of sexual response in domestic fowl in relation to the - (1962). concept of imprinting. - Symp. Zool. Soc. Lond. 8, p. 227-234.
169 HESS, E. H. (1959). Imprinting. - Science 130, p. 133-141. (1964). Imprinting in birds. - Science 146, p. 1128-1139. K. (1970). The influence of early experience upon the development of IMMELMANN, social behaviour in estrildine finches. - Proc. XVth Int. Ornith. Cong. SMITH, S. (1966). Some evolutionary consequences of pegKALMUS,M. & MAYNARD matypic mating systems. - Amer. Natur. 100, p. 619-636. P. H. (1965). Imprinting: A reassessment. - Science 147, p. 302-303. KLOPFER, KUMMER,H. (1968). Social organization of Hamadryas baboons. - University of Chicago Press, Chicago. LORENZ,K. (1937). Über die Bildung des Instinktbegriffes. - Naturwis. 25, p. 289-300, 307-318, 324-331. MANNING,A. (1967). An introduction to animal behaviour. - Addison-Wesley Publishing Co. MIRSKY,P. J. (1971). Colour dimorphism in the Lesser Snow Goose (Anser caerulescens caerulescens). -- M.Sc. Thesis, Queen's University, Kingston, Ontario. MOLTZ,H. (1960). Imprinting: Empirical basis and theoretical significance. - Psych. Bull. 57, p. 291-314. PREVETT, J. P. (1972). Family behaviour and age-dependent breeding of the Blue Goose, Anser caerulescens. - Ph.D. Thesis, University of Western Ontario, London, Ontario. SCHUTZ,F. (1963). Objektfixierung geschlechtlicher Reaktionen bei Anatiden und Hühnern. - Naturwissenschaften 50, p. 624-625. SEIGER,M. B. (1967). A computer simulation study of the influence of imprinting on population structure. - Amer. Natur. 101, p. 47-58. WALLER,P. F. & WALLER,M. B. (1963). Some relationships between early experience and later social behaviour in ducklings. -- Behaviour 20, p. 343-363. ZUSAMMENFASSUNG Die Wahl des Partners nach seiner Farbe, die bei der dimorphen Schneegans, Anser caerulescens caerulescens, beobachtet wurde, könnte dann erklärt werden, wenn die Vögel ihren Geschlechtspartner gemäss der Farbe ihrer Eltern und/oder ihrer Geschwister wählten. Drei verschiedene Schwärme wurden in der Gefangenschaft auf Farbpräferenzen getestet hinsichtlich 1) der Annäherung, 2) der Präferenzen in bezug auf das Zusammenleben und 3) der Wahl des Geschlechtspartners. 1) Annäherung. Jungvögel, die in eine Auswahlsituation gebracht wurden, zeigten eine auffallende Bevorzugung von Vögeln der elterlichen Farbe. Unterschied sich die Farbe der Geschwister von der der Eltern, dann schien dies ihre Wahl zu modifizieren. Wenn die Eltern während der Reifezeit entfernt wurden, konnten die frühen Farbpräferenzen verändert werden, wobei die jüngsten Verbindungen in bezug auf das Zusammenleben die Präferenzen bestimmten. Das Reaktionsmuster änderte sich nicht in Tests, die an Vögeln im Alter von 3 Monaten vorgenommen und im Alter von II Monaten wiederholt wurden. Auch das Vorhandensein oder Fehlen von akustischen Signalen veränderte das Reaktionsmuster nicht. Es konnten keine Unterschiede zwischen den Reaktionen des Jungvogels auf mütterliche und väterliche Farbe festgestellt werden. 2) Präferenzen in bezug auf das Zusammenleben. In der freien Natur leben die Vogel während der ersten zwei Jahre gewöhnlich mit ihren Geschwistern zusammen. Im zweiten Jahr sind die Tiere jedoch weniger mit ihren