HIERARCHIES AMONG NORTHERN HARRIER (CIRCUS CYANEUS) HAREMS AND THE COSTS OF POLYGYNY

Transcription

1 Journal of Animal Ecology (1986),, HIERARCHIES AMONG NORTHERN HARRIER (CIRCUS CYANEUS) HAREMS AND THE COSTS OF POLYGYNY BY R. E. SIMMONS*t, P. C. SMITH* AND R. B. MACWHIRTERt *Department of Biology, Acadia University, Wolfville, Nova Scotia, Canada BOP IXO and tdepartment of Biology, Mount Allison University, Sackville, New Brunswick, Canada EOA 3CO SUMMARY (1) A polygynous population of northern harriers was studied for years in New Brunswick to establish the costs and benefits to females within harems. (2) Polygynous harems of up to five females were observed, and clear hierarchies in clutch size and reproductive success were apparent in each year. Alpha females raised more fledglings than any other harem female. (3) Seasonal declines could explain the tendency of polygynous females to lay smaller clutches but not their reduced reproductive success relative to monogamous females. () Later-settling polygynous females suffered higher rates of predation than either monogamous or cr females. This could not be explained by differences in nest site quality; nor could female inexperience be cited, since older females predominated in harems. () Reduced success of polygynous females was best explained by male provisioning rates. Alpha females received significantly more prey items per day than any other harem female. (6) We surmise that the increased predation of later (p/, y, and 6) harem nests was also a result of low male provisioning rates; later-settling females were forced to forage by themselves and to leave their nests unattended. (7) We contend that a skewed sex ratio is not the prime reason for the evolution of polygyny, as suspected in other harrier populations, but that differences in male quality or controlled resources contribute to the crossing of the polygyny threshold. INTRODUCTION Benefits to first-settling (cr) females in polygynous harems are usually obvious, since not only do they mate with bachelor males, but also have first choice among available limiting resources. Present-day studies have correctly focused on possible benefits to later-settling (polygynous) females which choose non-bachelor males. This was not always so. Early polygyny models saw male emancipation from parental care and skewed sex ratios as sufficient reasons for non-bachelor matings (Mcllleny 19; Armstrong 197; Crook 196). However, skewed sex ratios were shown to be the exception (Balfour & Cadbury 1979; Smith, Yom-Tov & Moses 1982), and not the rule (Armstong 19; Selander & Giller 1961; Willson & Pianka 1963; Stobo & McLaren 197; Burke 1979) among polygynous birds. Thus, unmated females are usually not simply faced with a choice between polygyny and non-breeding. Benefits to polygynous females were suggested (Verner 196), modelled (Orians 1969), t Present address: Department of Zoology, University of the Witwatersrand, Johannesburg 21, South Africa. 7

2 76 Harrier harem hierarchies found to be lacking among some species (von Haartman 1969; Downhower & Armitage 1971; Crawford 1977; Weatherhead & Robertson 1977), and remodelled (Elliot 197; Altmann, Wagner & Lenington 1977; Wittenberger 1979, 198, 1981; Garson, Plesczcynska & Holm 1981), and are apparently absent in some populations of two well-studied species: the red-winged blackbird Agelaius phoeniceus (Linn.) and the pied flycatcher Ficedula hypoleuca (Pallas) (Weatherhead & Robertson 1979; Alatalo et al. 1981; Alatalo, Lundberg & Stahlbrandt 1982, 198). Out of this productive period came two prominent ideas: that high quality 'sexy sons' may benefit their reproductively disadvantaged mothers through increased genetic contributions during their own lifetimes (Weatherhead & Robertson 1979, 1981; Heisler 1981), and that males, through polyterritorial behaviour, may deceive females into poorer polygynous situations (von Haartman 1969; Alatalo et al. 1981). What is clear is that all probable (and even improbable) benefits to polygynous females must be evaluated before their situation in the harem can be described as maladaptive or otherwise; inadequate testing can lead to poor and possibly inaccurate results (Simmons 198). Our purpose here is to explore the costs and benefits of harem occupancy in an otherwise well-studied species, the northern (or hen) harrier Circus cyaneus (L.). We detail the effect that harem membership and position has on clutch size and yearly reproductive success, and analyse the effects of nest-site quality, nest position, nest predation, male provisioning rates and male patterns in provisioning on female reproductive success. Since the harem may consist of up to seven females (see Balfour & Cadbury 1979; van der Kraan & van Strien 1969)-unlike many other polygynous species-the position in the harem may become significant. For this reason we use the terms beta (f/), gamma (y) and delta (b) to denote later-settling harem females; these are the polygynous females, since they choose polygynous over monogamous partners. Alpha (a) females are treated separately in the analyses since they choose bachelor males (as do monogamous females), yet later become part of a harem; their position is unique. Therefore, alpha females are 'primary' females as defined by Martin (197), later-settling females are 'polygynous' females, and collectively the a, /, y, and 6 females comprise the 'harem' females. STUDY AREA The Tantramar Marsh of Westmorland County, New Brunswick, Canada (?3'N, 6?2'W), is a large, partially drained freshwater marsh, reclaimed from the Bay of Fundy in the middle 16s (Smith 1967). The 6-km2 study site constituted the major upper portion of this flat, dyked area. Vegetation associations reflect partial (3% by area) drainage and agricultural use and may be divided into four field types: damp grazed pasture consisting of Calamagrostis canadensis (Michx.) and Spartina pectinata Link. grasses; dry, annually hayed timothy Phleum pratense L. fields; open water duck impoundments, bounded by cattails Typha spp.; and both damp and dry abandoned fields supporting Spirea latifolia (Ait.) bushes, vetch Vicia spp., umbelliferans, and grasses such as timothy and S. pectinata. Lightly grazed pasture and damp abandoned fields make up the largest portions and were the most frequently used for nesting by the harriers. METHODS Nest-finding Nest-finding was of prime importance in this study and four methods were used in the

3 R. E. SIMMONS, P. C. SMITH AND R. B. MACWHIRTER 77 first year. In later years, 198 sites were simply revisited; harriers, like many raptors, use the same nest areas annually. Briefly, methods in 198 comprised the following. (a) Watching for prey-carrying male in unusually direct and steady flapping flight-flights invariably terminated in a prey pass to a female. This was the most efficient method since % of sixty-four nests found in the first 3 years were discovered in this way, some from a distance of more than 1 m. (b) Observing females (and initially males) build nests. This was accompanied in the earliest stages by mutual soaring over future sites. Nest-building observations led to the discovery of 17% of sixty-four nests. (c) Watching for females in flight (following a prolonged spell of incubating), head-scratching, defaecating and occasionally foraging over their nesting locations. Following these 'exercise flights', females returned to their nests via a slow hovering descent (in contrast to pounces or glides into roost sites). If a female did not appear within min, a nest location was usually apparent; these observations led to the location of 3% of nests. (d) Nests discovered late in the season following the nestlings' first flight were only pinpointed by monitoring the 'hot and cold' diving responses of the adults defending their nests. Two nests required this method. Although these methods could be used to locate sites prior to egg laying, five nests that were found with - or 1-egg clutches were deserted; visits at this time were therefore later curtailed. Settling dates The status of a female in a harem was based on the date on which her clutch was started (cr = first initiated). No other method could be used, since nest-building and even courtship-feeding were unreliable cues in ascertaining a stable pair bond. Since eggs are normally laid over 2-day intervals and fade from light blue to white in a similar time period (Sealy 1967; Hamerstrom 1969; personal observation), interpolating the date of initiation was relatively simple and accurate. Twenty-five of sixty-four nests found with blue eggs in the nest required this method, while fourteen nests were located containing the first egg. Assuming a 31-day incubation period (Breckenridge 193; Hamerstrom 1969; personal observation), thirteen nests found with recently hatched chicks (about 2 g in weight) were backdated to the day of laying. Eleven other nests first found with feathered nestlings were backdated using measurements of the longest primaries and rectrices to age the chicks (Munoff 1963; Scharf & Balfour 1971). Only one nest, which was destroyed by a predator and contained white eggs, could not be backdated using the above criteria; we assigned a date of initiation from observations of the monogamous female flying in the area and perching close to the (subsequently found) nest. If a female re-nested, her second choice of nest was taken as a separate mate choice. In these circumstances she had the choice of remaining polygynous or re-mating monogamously; two such females remained with their polygynous mates (Simmons 198). Polygyny: ascertaining mated status All males were assumed monogamous until proven polygynous, hence our data on the degree of polygyny is conservative. Few adults were marked because the method of trapping we used (dho-gaza trapping, Hamerstrom 1963) caused partial desertion. Moreover, plumage differences among males proved too unreliable to accurately distinguish all males. Therefore, judgement of a male's mated status necessitated watching for either successive food passes to separate females and/or successive defence of more than one nest (a behaviour we could induce fairly easily). During these assessments males

4 78 Harrier harem hierarchies were observed continuously to avoid possible errors in identification; this was made easier by the fact that the study area was flat, treeless and, hence, two-dimensional in nature. Because of the large number of hours spent observing, we were able to assess male status repeatedly. Polygynous nests were also close to one another and polygynous males vigorously defended their nesting territories against all other males. Boundaries were well defined, and where a monogamous nest was closer to the centre of the harem than an outlying harem nest, the status of the nest was relatively easily assessed by boundary disputes. Nest proximity alone could not be used to assess mated status. Combining these factors, we are confident that the lack of marked males made no difference to our ability to accurately assess their mated status. Occasionally females, which were very site-specific and could be individually recognized (Simmons 198), settled on exactly the same day and could not be categorized as cror / on this basis. In 198 we separated two such females on the basis that the 'a' site was frequented more often by the male while preening, loafing or roosting; the a female was also generally the first of the two females to be fed in the morning-characteristics of all other a females. We stress that these two females were not separated on the basis of provisioning rate. In 1983 two y females could not be separated on these grounds, and so both retained y status. The fifth female in this harem was designated as 6. Nest-site quality Nest-site quality in relation to settling order has been fully explored elsewhere (Simmons & Smith 198), and full details of methods used in assessing quality are given there. In brief, we determined quality by relating combinations of moisture, vegetation and visibility scores of the nest site with subsequent nest success. Moisture was the best predictor of success (correctly classifying 6-82% of nests); successful nests were normally built in wet areas, while up to % of nests in dry areas were predator-destroyed (Simmons & Smith 198). We gave each nest a unique quality score by combining the arcsine transformed percentage successes in each of the categories moisture, vegetation and visibility. The analysis is repeated here with regard to harem status and site quality. Nest positions In each year all nests were plotted on a scaled map drawn from a combination of aerial photographs and commercial 1: maps. Distances between all nests, and in particular harem associations, were thus calculated. In the case of a re-nesting female, only her first nest was used to compute nest proximities. This was not possible for 1983 data, since different aspects of harrier ecology were under investigation. Male food-provisioning rates Nest observations were specifically designed to record the rates at which males provisioned food to their nests in both the pre- and post-hatch periods. Nine nests were intensively watched per year and seasonal provisioning was based on 62 h of observation and 21 h at harem nests in particular. Watches averaged 3- h and the time, size and/or type of delivery were recorded. Observations covered all daylight hours and most weather types. Since rain cut provisioning rates to almost zero (Barnard 1983), watches at these times were excluded from the analyses. Also excluded were watches in which either adult alarm-called in response to our presence-on later analysis all such periods exhibited rates lower than normal. Thereafter, nests were observed from 2 m or more depending on the tolerance of the pair.

5 R. E. SIMMONS, P. C. SMITH AND R. B. MACWHIRTER Provisioning rates alone are compared between females in harems because no discernible differences in biomass were apparent. Provisioning rates collected in 1983 (R. B. M.) are not included here, but data relating to other aspects of the 1983 study are included. Breeding success 79 Throughout this paper, breeding success or productivity refers to the number of young raised to the flight stage (approximately 3 days after hatch) per year; this was easily assessed as flying young stayed close to their nests for 2-3 weeks. RESULTS Frequency ofpolygyny Over the -year study, polygyny accounted for between 11% and 3% of eighty-nine nests observed (Table 1). In 1983, polygyny reached the most extreme levels found thus far in North America, with one harem of five and one of four females among thirty-one nests. In total, approximately one third of the female breeding population bred within harems while 1% of males were polygynously mated (Table 1). Despite these high levels for North America, they are considerably lower than those proportions reported from the Netherlands (Schipper 1978) and Orkney (Balfour & Cadbury 1979; Picozzi 198a). The proportion of females in harems from year to year varied widely (Table 1) and was significantly associated with prey abundance (Simmons et al., in press). Female settling dates By definition, a females settled prior to polygynous females. Early season observations indicated, however, that a females were generally the most aggressive birds, dominating favoured positions. Alpha females tended to lay earlier than #/ females by 8 days (- > P <.7); fl, y and 6 females tended to lay in that order. Intra-harem settling dates were highly variable (Table 2), ranging from simultaneous settlements (two cases) to a and fl TABLE 1. Incidence of polygyny among male and female Northern Harriers over years in New Brunswick Year * Totals Number of 99 and nests 28 (2) 19 (3) 12 (1) 3 (1) 89 (7) (re-nests) Number of 9 9 in harems (%) 11 () 2(11) (33) 13 (3) 3 (3) Number of males Number of polygynous (19) 1 (6) 2 (2) (13) 11 (1) (%) Number of harems of the following sizes: * Thirty-seven nests were found but the mating associations of six were unknown, thus percentages for 1983 are based on thirty-one nests.

6 76 Harrier harem hierarchies TABLE 2. Mean settling dates of monogamous and harem female Northern Harriers over years in New Brunswick, including re-nests (n = sample size). Year Mono 99 a 99 YY 9 9Y9? May (17) 11 May () 16 May () 3 May (3) 8 June (1) May (19) 6 May (1) June (2) May (8) 8 June (2) 12 June (2) May (18) 16 May () 22 May () 27 May (3) 2 June (2) Total* 22 May (62) 17 May (11) 26 May (12) 28 May (6) June (3) * Difference in means between harem females, tested with one-tailed t-tests: ca v. ft (t= 1.63, P<.7); cv. (y+ ) (t= 2.7, P <.1);v. y+ (t= -.99, P> -1). associations arising 26 days apart. No obvious pattern was evident. In relation to monogamous males breeding on the marsh, polygynous males were also variable in the dates on which they settled with additional females. For example, in 1982, a year of extremely low spring vole abundance (Simmons et al. in press) the only polygonous associations arose at the end of the season, while in other years polygyny was spread evenly throughout. In 198 and 1983, two polygynous males had attracted and mated with two females each before other polygynous or monogamous associations had arisen. Furthermore, one male had paired with five females before,/ positions were occupied in two other harems in that year. Quality differences between polygynous males or the resources that they controlled are thus implied. Relative to monogamous females, a females settled earlier in 3 of years;,f females settled earlier in 2 of years, while y and 6 females settled, on average, later than monogamous birds. Individual monogamous females settled later than any other females in 2 of the years studied. In general, then, afemales settled early, y and 6 females settled late and f/ females were intermediate relative to monogamous females. These considerations are important in dealing with clutch sizes and reproductive successes for the various groups. Clutch sizes and yearly success within harems Completed clutches varied in size with respect to (a) year (decreased in food-poor years), (b) laying date (decreased with season) and (c) harem position. Further data on declines due to (a) and (b) will be published elsewhere. Clutch size and productivity among all harem and monogamous females (including those laying full clutches and failing) are illustrated in Fig. 1; mean productivity for successful females only are presented in Table 3. Alpha females did not lay significantly larger clutches but they reared more young (ANOVA F2,2 =.23, P <. all females), than any other female in the harem. This trend was consistent over the years studied. In 2 of the years, a females also laid larger clutches (x =.2 and -) than monogamous females (x =.91 and.2 respectively), and a females averaged fractionally higher clutch sizes than monogamous birds (-6 v..7). The mean clutch size of ft females (3-78) was the smallest of any group over the years (Fig. 1). Individually assessed, the numbers of flying young reared by a (t = 3.6, P <.) and ft females (t = 1-98, P <.) were significantly higher than those reared by y and? females. This relationship held despite there being no significant differences in the clutch sizes of these three groups (ANOVA F2, 23 = 1.67). Moreover, a females reared a greater proportion of their eggs to flying young (61%) than did monogamous females (7%), fl (8%), y females (13%) or 6 females (%). Thus in all respects, a females enjoyed a greater

7 R. E. SIMMONS, P. C. SMITH AND R. B. MACWHIRTER \II -3 3 I I I I Mono a 8 Female status FIG. 1. Mean clutch size and reproductive success (?+ S.E.) for all females in relation to harem status, Small numbers indicate sample sizes; nests failing through research interference are excluded. (O), Eggs; (?), young. TABLE 3. Productivity per year (fledged young expressed as x + 1 S.D. (n)), of successful female Northern Harriers with respecto harem position Year Mono 9 cc 9a / 99 (y+ 6) (1) () (3) None successful (9) - +-(1) None successful * (7) 2. +.(1) 2. +.(1) * (1) (3) (3) (1) Total 3.38?+ 1.1() 3?+ 1.1(9) 2.86?+ 1.7 (7) 2.?+. (1) \ * No females occurred. productivity than any other group of females, and clear hierarchies were present in terms of reared young. We tested the idea that clutch size and yearly success were not inherently greater among a females, but were a function of laying date, by assessing the productivity of each harem female in relation to her egg laying date. Equations (see Tables and ) relating the two variables allowed us to compare observed and expected clutch sizes for each group. No significant differences between clutch sizes of the various groups were apparent; seasonal declines were adequate to explain the decreased clutches of late settling harem females. When a similar analysis was performed for the number of young reared (Table ), different conclusions were evident; seasonal declines could not adequately explain the poorer success of y and 6 females (Fig. 1). Mean observed success for these females was significantly lower than expected (Table ), whereas a and f/ females raised as many as expected. The poorer success of some harem females relative to their predicted output could possibly be attributed to factors such as nest predation (mitigated through inferior nest sites), inexperience in breeding, and/or poor provisioning rates from their mates. Each of these factors is examined below.

8 762 Harrier harem hierarchies TABLE. Comparisons between observed () and expected (E) clutch sizes of harem female Northern Harriers over years; expected values are based on the date of laying* Year x n 1 S.D. tt a99 E (N.S.) E (Y + c) yy (N.S.) E (N.S.) * Based on equations derived from monogamous females laying full clutches only: 198, eggs = (days); 1981, eggs = (days); 1982, eggs = (days), 1983, eggs = (days); where 1 May = day 1. t Difference in means tested with two-tailed t-tests. TABLE. Comparisons between observed () and expected (E) reproductive success of harem female Northern Harriers over years; expected values are based on the date of laying* ct yy (Y + ) y y Year E o E E x n 1 S.D (N.S.) (p <.) tt.6 (N.S.) * Based on equations derived from monogamous females only: 198, young = (days); 1981, young = (days); 1982, young = (days); 1983, young = (days); where I May = day 1. t Difference in means tested with two-tailed t-tests.

9 R. E. SIMMONS, P. C. SMITH AND R. B. MACWHIRTER 763 Female harem position and nest failure Nests of polygynous females suffered somewhat higher levels of predation per attempt than either a or monogamous nests (Fig. 2). One of ten a nests failed through predation, while nine of forty-nine monogamous nests, four of eleven /f nests, and three of five y + 6 nests were destroyed by predators. Differences between these proportions were marginally short of significance (mono v. a v. poly: x2 =., P <.7). The same trend was recently reported for the hen harrier in Orkney (Picozzi 198a). Other forms of failure unrelated to research activity (excluded from this analysis) were also higher among later settling harem females. Three of four natural desertions occurred among 6 females; two had laid full clutches, of which one hatched young prior to desertion. Diseased chicks also accounted for the failure of one y nest. Through these, combined factors, y and 6 females failed in seven of their eight attempts. Harem position and nest site quality Since polygynous females suffered high levels of nest predation, their late settling dates may have been associated with inferior quality nest sites. Quality scores were assigned to each nest in the period and summed, averaged and compared between all female groups (Table 6). No significant differences existed between the means for each group (Table 6), suggesting that greater levels of predation were not linked to site quality. 8- o J - a) - z I 11 Mono a /3 y+ 8 Female status FIG. 2. Proportion of harem nests destroyed by predators, Smaller numbers indicate sample sizes; nests failing through research activity are excluded. TABLE 6. Mean nest site quality scores (QS) for all nests of monogamous, a; fl, and (y + 6) Northern Harrier females over three years (198-82)*. Scores are from Simmons & Smith (198) cr y,f y99 (y+ 6) 99 Mono 99 Number of QS Range * Kruskal-Wallis test for difference in means (crv. l v. (y + 6)): H = -37, df = 2, N.S.

10 76 Harrier harem hierarchies Nevertheless, as expected for females exercising first choice among resources, a and ft females tended to have better quality sites than y or 6 females. Monogamists chose, on average, higher quality sites than later-settling harem females, but lower quality sites than first settling ones (Table 6). Only monogamous females used the entire range of nest sites during the 3 years, and only y and 6 females did not use the highest quality sites. Again, the latter result might be expected for females exercising the last choice among resources. Female age and mated status Females of this species are relatively easy to age in the field by eye colour. Close-diving females defending their nests during our nest visits allowed repeated assessments of age as young (1-2 years) or old (3+ years), depending on brown or predominantly yellow irides respectively (Hamerstrom 1968; Picozzi 1981). Of eighty-nine females that bred, 71% could be aged using this method, or from captures (three females), hide observations (eleven females), or previous bandings (one female). A special effort was made to age harem females; this indicated that the majority were old females (Table 7). Unexpectedly, the lowest proportion of 'experienced' females occurred among monogamists. Therefore, because young females raised slightly more young than old females in New Brunswick (2. v. 2.3 young per female respectively) (Simmons et al., in press) and young females rarely occurred among harems, the lower productivity of late-settling harem females is inexplicable on the basis of inexperience. Senescence, however, cannot be ruled out. Female harem position and male provisioning Harem females were ordered in a distinct hierarchy based on their mate's provisioning rate (Fig. 3). Alpha and monogamous females were provisioned at a significantly higher rate than were polygynous females in 198 and 1981 (ANOVA F2 2 =.72, P <.). On average, a females (x = 7.1 items day-1) obtained over 1. times more prey than ft females (-6 items day-1), and six times as much as y and 6 females (1.3 items day-l). Monogamists, however, received marginally more (8- items day-') than did a females. The same trend was apparent in 1982, despite provisioning rates to harem females being assessed only prior to hatch; a females (x =.9 items day-') received twice as many pre-hatch prey items as,b females (2- items day-'), but still fractionally (by -3 items day-1) less than monogamists. The differences were not significant in 1982, probably due to the small sample of harem females (n = ). In all 3 years in which provisioning rates were studied, therefore, hierarchies in food supply (Fig. 3) and nest predation levels (Fig. 2) were evident among harems, and paralleled declines in female productivity. Male provisioning behaviour and female competition Polygynous males not only provisioned their a females more often than other harem females, but also roosted, preened and loafed more often in the vicinity of these nests. TABLE 7. Age in relation to mated status of female Northern Harriers, Female status ca 99 (y + 6) (2 Mono $9 Number of breeding Number of known age Number of 9 9 >- 3 years (% of known age) 11 (1) 7 (78) 6 (86) 21 (8)

11 R. E. SIMMONS, P. C. SMITH AND R. B. MACWHIRTER 76 I o ' I Mono a,i +8 Female harem position FIG. 3. Male seasonal (pre- and post-hatch) provisioning rates per 1-h day in relation to harem status of females, Rates determined in 1982, despite reflecting similar hierarchies, are not included as they could be determined only prior to hatch (see text). Vertical bars represent + 1 S.D. and small numbers indicate sample sizes. Moreover, a females were the first fed of the harem females at the start of each day. Despite these inequalities, rfemales did not receive their greater share of attention through aggression, but through male choice. This was evident in the behaviour of males flying directly to the nests that they eventually provisioned-nests that were occasionally furthest from their point of prey capture. In 198 and 1982, some males were observed soaring and flapping to heights approaching 1 m, directly after prey capture, and flying over one of their nests to provision another. Only once was direct piracy between intra-harem females observed in approximately 2 h of field observation, and only twice was a male forced to give up his prey (once each by an a and a ft female) when it was obviously destined for delivery at another nest. Surprisingly few intra-harem interactions took place once early season boundary disputes were settled, even though polygynous females tended to forage a great deal in the harem vicinity from an early stage. Male behaviour and not female aggression was thus responsible for the feeding hierarchies observed. Harem position and nest dispersion To determine whether polygynous females might position their nests to solicit food away from the incoming male, nesting locations were plotted for all harem females to assess whether polygynous females were closer to 'their' a females than expected by chance. Such an association was evident; of all polygynous females studied between 198 and 1982, 8% settled closer to their a females than to any other female on the marsh. Simple nest

12 766 Harrier harem hierarchies TABLE 8. Intra-harem nest dispersion among New Brunswick harriers, Status of female* Polygynous 9 fl Loop /f Loop,P Culv. 3 f/ CP2 /f PB / Loop 1 /3 Culv. 3 y Culv. y Alders 6 Hog 2 Alpha 9 Q a Loop 1 a Loop 2 ca Culv. 1 a CP 1 a Culv. 2 a Loop 3 ca Culv. 1 Status of nearest neighbour Intra-harem ca Intra-harem a Intra-harem cr Intra-harem cr Monogamous 8% closer to 'their' cr female Intra-harem cr Intra-harem ca Intra-harem cr Intra-harem a Monogamous Monogamous Extra-harem Intra-harem y Intra-harem /33% closer to 'their' harem females Monogamous Intra-harem f/ Monogamous Alpha females, then polygynous females, are compared against their nearest neighbours to to answer the question: is the harem centered about the cc nest, and is this due to simple nest clumping? See text for details. * Some nest designations were used in more than one year, hence 'Loop 1' in 198 was occupied by a /3 female, and in 1981 by an a female. clumping, as this might suggest, was not the cause since a females were not similarly placed. Less than % of all a females were closer to other females in 'their' harem than to all other sets of females (Table 8). This is a conservative analysis of nest location, since individual polygynous females are assessed with regard to just one other female ('their' a female), while a females are tested against several other females (the rest of 'their' harem); no statistical test seems appropriate to test such a dispersion. The two polygynous females which did not nest close to the crfemale (Table 8) may have had no option; they were the last and penultimate birds in their respective years to settle. An c-biased nest-centredness was thus apparent among the polygynous females studied from 198 to An alternative hypothesis to the idea that these females were attempting to solicit food away from their mates is that they were attempting to exploit the high quality food territories of their a females (cf. Seastedt & Maclean 198), a possibility investigated later. DISCUSSION Costs imposed by harem hierarchies This -year study of hierarchies and their inherent costs among harem females brings to light six points: (i) afemales laid somewhat larger clutches, raised significantly more young and reared a greater proportion of their clutch to flight than any other female in the harem; (ii) clutch sizes were larger among c females than among other harem females because they were laid earlier in the season; however, the reduced annual reproductive success of late polygynous females could not be explained as a seasonal effect; (iii) hierarchies in levels of

13 R. E. SIMMONS, P. C. SMITH AND R. B. MACWHIRTER predation were obvious among harems, affecting a females the least and y and 8 females the most; (iv) these greater predation levels could not be explained in terms of nest site quality differences because no significant variation existed among harem females in this respect; (v) female inexperience (age) could not explain the higher levels of predation and lower productivity of late-settling harem females since harems mainly comprised birds 3 years old and over; and (vi) male provisioning rates were significantly higher to a females than to polygynous females in 198 and 1981, and followed the same trend in 1982-this was maintained through male choice not female aggression. The main finding, then, was that clear hierarchies of male provisioning and predation levels were evident among harems. The decreased productivity of later settling harem females could be explained best, not through seasonal declines in productivity, but through a decrease in male provisioning. This was expected, since strong significant correlations between male rates and subsequent success in rearing young were apparent in all 3 years in which they were studied (Simmons 1983; Simmons et al., in press). If successful, a polygonous female would thus be expected to raise fewer young than monogamous or a females since females cannot compensate for the loss of paternal assistance (Simmons, Barnard & Smith 1987). We contend that the lower provisioning rates to polygynous females also added to the higher levels of predation. Harriers are ground nesters and their nests are potentially open to predation from many sources; a nest left unattended by a female forced to forage for herself should be more vulnerable to both terrestrial and aerial predators. Indeed, two females that had left their nests unattended both lost their clutches to common ravens Corvus corax (Ridgeway), even though their nests were in sites designated as 'concealed' (Simmons & Smith 198). Daeman and Lorij (in Nieboer 1973) also noted that higher predation levels were characteristic of harrier nests left unattended by less vigilant females. Since neither inferior quality nest sites nor female inexperience could account for the greater predation levels, female absence seems the most plausible explanation. In its most extreme form, a poor male food supply may have also led to the desertion of viable nests by 6 females. In these cases, however, the females' own survival may have been at stake. This is likely because females were the least provisioned of any harem females (Fig. 3), a finding supported by data from 1983 (MacWhirter 198). Despite a large number of studies of polygynous harriers, only recently have hierarchies been explored, and then only as far as ft females. Altenburg et al. (1982), studying the European marsh harrier Circus aeruginosus Linn., and Picozzi (198a), working on C. cyaneus in Orkney, found that a females also raised more young per year than fl females. Altenburg et al. (1982) determined that a females laid slightly smaller clutches and A females laid slightly larger clutches than expected on the basis of seasonal differences in clutch size (in our study, both a and / females laid slightly smaller clutches than predicted). The fewer number of young raised by f, female marsh harriers was not a season-related trend, since fewer young were raised relative to concurrently settling monogamous females; our results are identical in this regard. Comparisons with Picozzi's (198a) data are not possible since he did not discuss seasonal productivity trends. Another fitness-related cost noted among Orkney harriers (Balfour 197) was the high incidence of unhatched eggs among harems. In the New Brunswick population, even though a females raised a greater proportion of their young than monogamists, they suffered high infertility rates, along with other harem members. Polygynous males in fact gave rise to significantly more infertile eggs than monogamous males, a result probably due to sperm depletion rather than food stress (R. E. Simmons, R. B. MacWhirter & G. L. Hansen unpublished). 767

14 768 Harrier harem hierarchies Effects of harem membership on alpha females A result common to studies in Orkney (Picozzi 198a), the Netherlands (Altenburg et al. 1982), and New Brunswick was the greater success of a females. This was somewhat unexpected in New Brunswick, because a females received slightly less food from their mates than did monogamous females. Two explanations are possible: either a females received more biomass per item than monogamists (as implied by the results of Altenburg et al for the European marsh harrier), or c females foraged in higher quality food territories and supplied more biomass to their young than monogamists (as also found in the European marsh harrier, Attenburg et al. 1982). The first idea is the least plausible in this study, because no discernible differences in the mean sizes of the chief prey (Microtus pennsylvanicus (Ord.), Barnard 198) provisioned to monogamous and a females were noted between 198 and However, the availability of voles in 198 and 1981, as measured by snap-trap sampling and an assessment of vegetative structure in a -m diameter area about the nest (the females' usual foraging area; Simmons 1983), was significantly different among successful a,, and monogamous nests (Kruskal-Wallis H = 6*29, d.f. = 2, P <.). As expected, a females exhibited much higher prey availability scores (x = 76) than did monogamous (x = 68) or,b females (x = ) (Simmons 1983). These differences may, therefore, explain the differing successes of harem females in conjunction with male provisioning rates. The relationship between polygynous females and 'their' a females in terms of nest positioning might also be a result of the greater food quality of the a territories. Since polygynous females started foraging earlier in their breeding cycle than a females, they may have had the chance of exploiting the a females' resources while these females were still guarding their broods. An additional explanation, in light of the fact that males tend to choose the general location of the nest site (Simmons, Barnard & Smith 1987), is that a male could guard his dependent females more efficiently, both from extra-pair copulations with other males and from predators at the nest. The present study cannot distinguish between these possibilities. Male fitness, hierarchies, and unanswered questions So far we have dealt exclusively with female-related costs and benefits because of the importance of female choice in polygyny studies. However, male strategies, once males have attracted and started a nest, are of interest since males too are assumed to be maximizing their yearly fitness even if the two sexes' goals sometimes conflict. Here we determined that males not only fed more prey items to their first-settled females, but these females were the first fed in the day and the male spent more time near the a nests. This greater investment in the crnest seems a logical tactic. The a nest probably not only has a greater chance of avoiding predators because of the slightly higher quality of such sites (Table 6), but nests started early in the season had inherently greater chances of high reproductive successes (for undetermined reasons). Furthermore, the most aggressive and possibly the fittest females acquired a positions. Pooled, these advantages to the male would probably outweigh advantages in preferentially feeding later-arriving females. However, this raises several questions: (a) why can't polygynous males feed their other dependent females to the same degree as they do their a females?; (b) why don't polygynous females desert their mates on realizing a reduction, or at least, a failure to increase the rate as found in monogamous birds?; and, most important, (c) why do polygynous females mate polygynously at all, assuming that they can assess their mate's quality prior to settling? In studies in Orkney and the Netherlands (Balfour & Cadbury 1979; Altenburg et al.

15 R. E. SIMMONS, P. C. SMITH AND R. B. MACWHIRTER ; Picozzi 198a, b), these questions are thought to be soluble through considerations of the skew of the sex ratio towards a predominance of females. Specifically, unmated females accept their poor chances of success by mating polygynously since the alternative is not to breed. We query this conclusion for two reasons. First, in one study of C. cyaneus in North America, sex ratios of unity are found in conjunction with relatively high degrees of polygyny (Burke 1979), while the reverse-female predominance but no polygyny-is found in one other well-studied raptor (Newton 198). Secondly, two studies (Altenburg et al. 1982; this study) have demonstrated differential quality among males, a condition much more likely to lead to the potential for polygyny (Orians 1969). Here it was determined that some polygynous males had obtained five mates prior to other males becoming polygynous. Hence, a skewed sex ratio alone is unlikely to select for polygyny in harriers. The above questions demand analyses of (a) the constraints on males, (b) information available to settling females, and (c) other options open to females. These aspects, together with analysis of possible cooperation among females and the potential for 'sexy sons' as possible pathways to the evolution of polygyny in Circus cyaneus, are considered elsewhere. ACKNOWLEDGMENTS Field work for this study was made possible through partial financial support from a Natural Sciences and Engineering Research Council of Canada grant to P. C. S., extensive logistical support from the Canadian Wildlife Service to Sackville, New Brunswick, graduate fellowships to R. E. S. from Acadia University and a Mount Allison Research grant to R. B. M. We are grateful to each of these bodies for their aid. We are also indebted to Gay Hansen for agreeing to continue this harrier study, and for helping to collect the 1983 data. The paper was reviewed and improved by Phoebe Barnard and anonymous referees. REFERENCES Alatalo, R. V., Lundberg, G. A. & Stahlbrandt, K. (1982). Why do pied flycatcher females mate with already-mated males? Animal Behaviour, 3, Alatalo, R. V., Lundberg, A. & Stahibrandt, K. (198). Female mate choice in the pied flycatcher Ficedula hypoleuca. Behavioural Ecology and Sociobiology, 1, Alatalo, R. V., Lundberg, G. A. Stahlbrandt, K. & Ulfstrand, S. (1981). The conflict between male polygamy and female monogamy: the case of the pied flycatcher, Ficedula hypoleuca. The American Naturalist, 117, Altenburg, W., Daan, S., Starkensberg, J. & Ziljstra, M. (1982). Polygamy in the marsh harrier, Circus aerunginosus: individual variations in hunting performance and number of mates. Behaviour, 79, Altmann, S. A., Wagner, S. S. & Lenington, S. (1977). Two models for the evolution of polygyny. Behavioural Ecology and Sociobiology, 2, Armstrong, W. A. (197). Bird Display and Behaviour. Lindsay Drummond, London. Armstrong, E. A. (19). The Wren. Collins, London. Balfour, E. (197). Observations on the breeding biology of the hen harrier in Orkney. II. Bird Notes, 27, Balfour, E. & Cadbury, C. J. (1979). Polygyny, spacing and sex ratio among hen harriers Circus cyaneus, in Orkney Scotland. Ornis Scandinavica, 1, Barnard, P. E. (1983). Foraging behaviour and energetics of northern harriers Circus cyaneus (L.). B.Sc. (Hons) thesis, Acadia University. Barnard, P. E. (198). Prey selection and provisioning strategies by northern harriers. Proceedings of the 2nd Symposium of African Predatory Birds (Ed. by J. M. Mendelsohn & C. W. Sapsford), p Natal Bird Club, Durban.

16 77 Harrier harem hierarchies Breckenridge, W. J. (193). An ecological study of some Minnesota marsh hawks. Condor, 37, Burke, C. J. (1979). Effect of prey and land use on mating systems of harriers. M.Sc. thesis, University of Wisconsin. Crawford, R. D. (1977). Polygynous breeding of short-billed marsh wrens. The Auk, 9, Crook, J. H. (196). The evolution of social organization and visual communication in the weaver birds (Ploceinae). Behaviour 1 (Supp.), Downhower, J. F. & Armitage, K. B. (1971). The yellow-bellied marmot and the evolution of polygyny. The American Naturalist, 1,3-37. Elliot, P. F. (197). Longevity and the evolution of polygyny. The American Naturalist, 19, Garson, P. J., Plesczcynska, W. K. & Holm, C. (1981). The polygyny threshold model: a reassessment. Canadian Journal of Zoology, 9, Hamerstrom, F. (1963). The use of great horned owls in catching marsh hawks. Proceedings of the 13th International Ornithological Congress, pp Hamerstrom, F. (1968). Aging and sexing harriers. Inland Bird Banding News,, Hamerstrom, F. (1969). A harrier population study. Peregrine Falcon Populations: their biology and decline (Ed. by J. J. Hickey), pp University of Wisconsin Press, Madison. Heisler, I. L. (1981). Offspring quality and the polygyny threshold: a new model for the 'sexy son' hypothesis. The American Naturalist, 117, MacWhirter, R. B. (198). Breeding ecology, prey selection and provisioning strategies of northern harriers Circus cyaneus L. B.Sc. (Hons) thesis, Mount Allison University. Mcllleny, E. A. (19). Sex ratios in wild birds. The Auk, 7, Martin, S. G. (197). Adaptations for polygynous breeding in the bobolink, Dolichonyx oryzivorous. American Zoologist, 1, Munoff, J. A. (1963). Food habits, growth and mortality in nesting marsh hawks. Kingbird, 13,67-7. Newton, I. (198). Lifetime reproductive output of female sparrowhawks. Journal of Animal Ecology,, Nieboer, E. (1973). Geographic and ecological differentiation in the genus Circus. Published Ph.D. thesis, Free University, Amsterdam. Orians, G. H. (1969). On the evolution of mating systems in birds and mammals. The American Naturalist, 13, Picozzi, N. (1981). Weights, wing lengths, and iris colour of hen harriers in Orkney. Bird Study, 28, Picozzi, N. (198a). Breeding biology of polygynous hen harriers Circus c. cyaneus, in Orkney. Ornis Scandinavica, 1, 1-1. Picozzi, N. (198b). Sex ratio, survival and territorial behaviour of polygynous hen harriers Circus c. cyaneus, in Orkney. Ibis, 126, Scharf, W. C. & Balfour, E. (1971). Growth and development of nestling hen harriers. Ibis, 113, Schipper, W. J. A. (1978). A comparison of breeding ecology in three European harriers (Circus). Ardea, 66, Sealy, S. G. (1967). Notes on the breeding biology of the marsh hawk in Alberta and Saskatchewan. Blue Jay, 2, Seastedt, T. R. & MacLean, S. F. (198). Polygyny and nest spacing in lapland longspurs; a hypothesis. Animal Behaviour, 28, Selander, R. K. & Giller, D. R. (1961). Analysis of sympatry of great-tailed and boat-tailed grackles. Condor, 63, Simmons, R. E. (1983). Polygyny, ecology, and mate choice in the northern harrier Circus cyaneus (L.). M.Sc. thesis, Acadia University. Simmons, R. E. (198). Do northern harriers lay replacement clutches? Raptor Research, 18, Simmons, R. E. (198). Did tree swallows cross a polygyny threshold? The Auk, 12, Simmons, R. E. & Smith, P. C. (198). Do northern harriers Circus cyaneus choose nest sites adaptively? Canadian Journal of Zoology 63, Simmons, R. E., Barnard, P. E. & Smith, P. C. (1987). Reproductive behaviour of Circus cyaneus in North America and Europe: a comparison. Ornis Scandinavica (in press). Simmons, R. E., MacWhirter, R. B., Barnard, P. E. & Hansen, G. L. (in press). The influence of microtines on polygyny, productivity, age and provisioning of northern harriers: a -year study. Canadian Journal of Zoology. Smith, A. D. (1967). Waterfowl habitat productivity and management at Missaquash, Nova Scotia. M.Sc. thesis, Acadia University. Smith, J. N. M., Yom-Tov, Y. & Moses, R. (1982). Polygyny, male parental care, and sex ratio in song sparrows: an experimental study. The Auk, 99, -6. Stobo, W. T. & McLaren, I. A. (197). The Ipswich Sparrow. Nova Scotia Institute of Science, Halifax. van der Kraan, C. & van Strien, N. J. (1969). Polygamie bij de blauwe kiekendief (Circus cyaneus). Limosa, 2, 3-3. Verner, J. (196). The evolution of polygamy in the long-billed marsh wren. Evolution, 18,

17 R. E. SIMMONS, P. C. SMITH AND R. B. MACWHIRTER 771 von Haartman, L. (1969). Nest site and evolution of polygamy in European passerine birds. Ornis Fennica, 6, Weatherhead, P. J. & Robertson, R. J. (1977). Harem size, territory quality, and reproductive success in the red-winged blackbird (Agelaius phoeniceus). Canadian Journal of Zoology,, Weatherhead, P. J. & Robertson, R. J. (1979). Offspring quality and the polygyny threshold: the 'sexy-son' hypothesis. The American Naturalist, 113, Weatherhead, P. J. & Robertson, R. J. (1981). In defence of the 'sexy-son' hypothesis. The American Naturalist, 117, Willson, M. F. & Pianka, E. R. (1963). Sexual selection, sex ratio, and mating system. The American Naturalist, 97,-7. Wittenberger, J. F. (1979). The evolution of mating systems in birds and mammals. Handbook of Behavioural Neurobiology, Vol. 3 (Ed. by P. Marler & J. Vandenbergh), pp Plenum Press, New York. Wittenberger, J. F. (198). Group size and polygamy in social animals. The American Naturalist, 11, Wittenberger, J. F. (1981). Time: a hidden dimension in polygyny threshold model. The American Naturalist, 118, (Received 2 March 198)