Breeding of the Peregrine Falcon Falco peregrinus: 111. Weather, Nest Quality and Breeding Success

Breeding of the Peregrine Falcon Falco peregrinus: 111. Weather, Nest Quality and Breeding Success PENNY D. OMEN' & JERRY OLSEN~ I Divirion of WiIdLge & Ecology, CSIRO, PO. Box 84, Lyneham, A.C.T. 2602 RMB 1705 Read Road Sutton, MS. W 2620 Received 1 November 1987, accepted 6 July 1988 Summary Olsen, P.D. & Olsen, J. (1989). Breeding of the Peregrine Falcon Falco peregrinus: 111. Weather, nest quality and breeding success. Emu 89,6-14. Peregine Falcons laid clutches at 75% of temtories annually and fledged young from 58%. Sixty-seven per cent of pairs fledged young each year. Brood size at fledging was 2.16, equivalent to 1.44 young per pair, or 1.23 young per temtory. For the population persistent rain and low temperatures in the three to four months between egg laying and to a week or so after hatching had an adverse effect on breeding success; in years when the number of raindays was high fewer pairs bred, more clutches failed to hatch and some very young nestlings may have died. July-September raindays accounted for 60% of the variation in young fledged per territory while mean maximum temperature between August and October accounted for 74% of the variation in the percentage of territories from which young fledged and raindays in those months accounted for 58%. On the other hand, brood size was larger in years when it was wet during the nestling period. Overall, severe drought enhanced the breeding success of the population because more pairs bred successfully. Almost all of the adverse effect of rain could be attributed to nest quality: (1) in cold, wet years Falcons with a choice of nest, or with a welldrained, well-sheltered nest, bred as successfully as they did in dry years, while those without such nests performed poorly; and (2) the average number of young fledged from temtories with a choice of nest was about 50% greater than that from territories with only one nest. The advantage of holding a territory with a high quality nest(s) as oppposed to a poorer nest, has clear selective value in the Darwinian sense. Introduction In the past two decades much interest has centred on the reproductive performance of various populations of Peregrine Falcon Falco peregrinus (e.g. Hickey 1969; Cade & Fyfe 1970; Ratcliffe 1970). While variation in the output of young has been linked with the degree of contamination by pesticides (Newton 1979; Ratcliffe 1980), other factors influencing reproductive success have received little attention. Breeding success of the Peregrine Falcon in Australia is generally high if their relatively small clutch size, compared with Falcons elsewhere in the world, is taken into account (Olsen & Olsen 1979; Pruett-Jones et al 1981; Olsen 1985). A population of resident Falcons near Canberra was studied for 12 years. Virtually free from the deleterious effects of pesticides, predators and disturbance, the population suffered marked declines in reproductive success in wet years. Lowered occupancy of breeding territories by pairs and a shortened egg laying season were associated with wet weather and resulted in fewer pairs breeding. This lowered occupancy appeared to be due mainly to flooding of nest sites (Olsen & Olsen 1988b, 1989). Newton (1979, 1986) found that the number of raindays in April accounted for 46% of the variation in the production of young European Sparrowhawks Accipiter nisus; he argued that lowered breeding performance in raptors in wet years is due largely to a reduction in hunting efficiency of the breeding birds which, in turn, influences laying date, clutch size and survival of young. In this paper we look at the effect of inclement weather on the breeding success of Falcons near Canberra. Some preliminary results were reported in (Olsen & Olsen 1988a). Methods Over a 12 year period, between 1975 and 1986, several traditional nest sites, centred in a territory, were monitored yearly. Occupancy of the temtories by a pair or single Peregrine Falcon remained high (82-100%) throughout the study (Olsen & Olsen 1988b). All pairs nested on cliffs. Each nest site was visited at least once in November and a return visit made if necessary. The presence or absence of birds and fresh signs of occupation were noted. Reproductive success was recorded - a successful pair was one fledging at least one young. Fresh shell fragments, eggs or nestlings indicated that the pair had laid a clutch; some clutches may have been lost without trace but, because this was not the case at more closely monitored sites, any such loss was considered negligible. In the non-breeding season, cliff and nest characteristics, including total height, from the ground, of nest cliff, height of nest ledge (a flatish shelf, perhaps overhung above but more-or-less unsheltered from the side) or pot-hole (a small cave), adequacy of drainage of the floor of the nest (a litre of water was tipped onto the nest scrape and the presence or absence of a pool of water after

1989 OLSEN & OLSEN: BREEDING OF PEREGRINE FALCON Kl 7 five minutes recorded) and number of alternative nest sites, was noted. Monthly meteorological information was obtained for Canberra for each month between June and November, it included mean minimum daily temperature, highest rainfall, number of days of fog and frost, number of raindays and total rainfall. A rain intensity index was calculated as the amount of rain in mm divided by the number of raindays. The relationship between breeding performance and monthly, two- and three-monthly and annual (both actual year and previous year) totals or means of rain, raindays, mean maximum and mean minimum temperature was investigated. Bi- or tri-monthly totals are designated, for example, as August-October, which is the the total or mean, where appropriate, for those three months. Only significant relationships are reported except where a particular point is being made about the lack of a relationship. Chi-square, multiple regression, ANOVA and paired t-test were used to test for differences and correlations. TABLE 1 Yearly breeding success of Peregrme Falcons near Canberra, 1975-1 986 Year Younglpair Younglsuccessful Young1 Young1 % temtories pair territory clutch successful Results Overall breeding success Over 12 years, 232 sites visits were made to 30 sites where breeding success was recorded (Table 1). All Peregrine Falcons occupying territories were in adult plumage. A 1 I I I I 24 28 32 36 40 July/September raindays I I I I 24 28 32 36 40 August/October raindays FIGURE 1 The relationship between breeding success and rain each year 1975-1986: (bottom) the percentage of sites (territories) from which young fledged and raindays between August-October; (top) the percentage of clutches failing to hatch and ra~ndays in July-September. See Table 2 for details of regression lines. clutch was laid at a yearly average of 74.5% (64-96 %) of sites and one or more young fledged from 57.5% (41-78 %). Fifteen and a half percent (0-30 %) of clutches failed to hatch; 66.8% (52-88 %) of pairs bred successfully each year. Yearly average brood size at fledging was 2.2 (1.9-2.6), or 1.8 (1.5-2.6) young per clutch, 1.4 (1.2-1.8) young per pair and 1.2 (0.9-1.6) young per temtory. For all years combined, young were fledged from 57% of temtories and from 83% of clutches. The number of young per successful pair (brood size) was 2.1, this was equivalent to 1.8 young per clutch, 1.4 young per pair, and 1.2 young per territory. Weather and breeding success Failure of clutches to hatch An increased number of clut- ches failed to hatch when the weather was wet during the main egg laying and incubation period (August-September). The percentage of pairs losing eggs was correlated with highest daily rainfall in both August and September (r = 0.75, P < 0.01 and r = 0.69, P < 0.05, respectively). Similarly, the percentage of clutches that failed to hatch was correlated with highest daily rainfall in August (r = 0.70, P < 0.05) and with the number of raindays in July- September (r = 0.59, P < 0.05) (Fig 1). Ten per cent of clutches failed to hatch in dry seasons (July-September raindays less than the long-term average of 31 days) compared with 22% in wet seasons (> 31 raindays, ~2 = 3.97, P < 0.05). Apart from rain the only other detected cause of failure to hatch was the usurping of a pot-hole by a Brush-tailed Possum Trichosum vulpecula that prevented incubation. Loss of nestlings Fewer young fledged in wet seasons than in dry. Although nests were usually not monitored around hatching time, correlations between various mea-

8 OLSEN gr OLSEN. BREEDING OF PEREGRINE FALCON III EMU 89 sures of reproductive success and weather indicate that some chicks died during hatching or in the fvst week after hatching and that mortality after that time was rare. No correlations were found between weather and the number of young fledged per clutch. The number of young per clutch varied quite widely over the 12 years while the number of young per successful pair (brood size) did not (Table 1). While most rain effects found were negative, large brood size was associated with high rainfall. The number of young per successful pair (brood size) was positively correlated with October rain (r = 0.80, P < 0.01, Table 2), the total amount of rain in September-October (r = 0.74, P < 0.01), August-October rain (r = 0.66, P < 0.05) and TABLE 2 Equations relating various measures of reproductive success to weather; P and percentage of variance shown. Temperature is in O C. Youngltemtory (ynglterr) with (i) July-Sept. raindays (J-S rd): ynglterr = 2.15-0.03 J-S rd, P < 0.01,60% (ii) Aug.-Sept. raindays (A-S rd): ynglterr = 2.02-0.04 A-S rd, P < 0.05,44% (iii) Aug.-Oct. raindays (A-0 rd): ynglterr = 2.05-0.03 A-0 rd, P< 0.05,50% (iv) Aug.-Oct. raindays (A-0 rd) and mean maximum temperature (A-0 t): ynglterr = 1.25-0.02 A-S rd + 0.04 A-S t, P< 0.05, 52% Young/successful pair (ynglsucc pr) with Oct. rain (0 r, cm): yng/succ pr = 1.96 + 0.03 0 r, P< 0.01,64% % of pairs fledging young (% pr succ) with (i) July-Aug. raindays (J-A rd) and Aug.-Sept. mean maximum daily temperature (A-S t): % pr succ = 0.1 1-0.69 J-A rd + 5.35 A-S t, P< 0.01, 64% (ii) July-Sept. raindays (J-S rd) and August-September temperature (A-S t): % pr succ = 17.08-0.70 J-S rd + 4.10 A-S t, P< 0.01, 70% (iii) Aug.-Oct. raindays (A-0 rd) and mean maximum daily temperature (A-0 t): % pr succ = 26.47-0.68 A-0 rd + 3.57 A-0 t, P< 0.01,68% % of clutches failing to hatch with July-Sept. raindays (J-S rd): % clutches failing = -12.56 + 0.95 J-S rd, P < 0.05, 35% % of sites fledging young (% sites succ.) with (i) Aug.-Oct. mean maximum temperature (A-0 t): % sites succ. = -75.91 + 7.78 A-0 t, P< 0.001,74% (ii) Aug.-Oct. raindays (A-0 rd): % sites succ. = 101.60-1.42 A-0 rd, P < 0.01,58% TABLE 3 Brood size of Peregrme Falcons near Canberra in wet and dry Octobers. A wet October has a total rainfall greater than the long term mean of 69 mm; a dry October has less. Brood size (% of Broods) 1 2 3 Mean Wet 4 (10%) 19 (49%) 16 (41%) 2.31 Dry 27 (24%) 63 (56%) 22 (20%) 1.96 Chi-square (wet vs dry) - 8.27, P < 0.05 September-November rain (r = 0.64, P< 0.05). There were relatively more broods of three in wet years than in dry years when broods of two predominated (Table 3). The distribution of brood sizes did not differ from a normal distribution in dry years ( ~2 = 2.20, n.s.) but did in wet ( ~ 2 = 7.52, P < 0.05). The number of young per pair was negatively correlated with July-September raindays (r = -0.75, P < 0.01) while August and September raindays alone accounted for 5 1 % of the variation in younglpair (r = -0.71, P < 0.01). Weak negative correlations were found with August-October and July-August raindays (r = -0.57, P < 0.05 and r = -0.55, P < 0.1, respectively). More young were fledged in dry years and warm years. July-September raindays accounted for 60% of the variation in the number of young fledged per temtory (r = -0.77, P < 0.01). August-October raindays were also negatively correlated with the number of young per temtory (r = - 0.7 1, P < 0.0 1). August-September raindays accounted for 44% of the variance (r = -0.66, P < 0.05) and mean maximum temperature for those two months accounted for 39% (r = -0.62, P < 0.05). Other correlations for young1 territory were with: July-August raindays (r = -0.63, P < 0.05); September-October raindays (r = -0.58, P < 0.05); September-November raindays (r = -0.62, P < 0.05); and August-October mean maximum temperature (r = 0.61, P < 0.05). The addition of temperature to any of the equations relating raindays to young per territory only marginally improved the correlations (Table 2). More pairs bred successfully in wet and warm years; August (early egg laying period) and October (hatching of most clutches) were the most critical months. The percentage of pairs that laid eggs was unrelated to any of the weather variables examined. For example, in years when raindays in July-September were > 30, 82% of pairs laid eggs compared with 80% in dry years. However, the percentage of pairs breeding successfully was related to maximum daily August and October temperature (r - 0.64, P < 0.05 and r = 0.66, P < 0.05, respectively) and the mean maximum daily temperature in August-September accounted for 60% of the variation (r = 0.77, P < 0.01).

1989 OLSEN & OLSEN: BREEDING OF PEREGRINE FALCON III 9 August-October raindays accounted for 58% of the variation in the percentage of pairs breeding successfully (r = -0.76, P < 0.01); the percentage of pairs breeding successfully was also related to July-August raindays (r = -0.65, P < 0.05), the inclusion of September increased this to r = -0.73 (P< 0.01). In warm, dry years young were fledged from more territories - October weather had the greatest influence followed by July-August. October temperature accounted for 65% of the variation in the percentage of temtories from which young fledged, and October rainfall 37% and rain intensity 42% (r = -0.62 and -0.65, respectively, both P< 0.05). September-October raindays accounted for 47% of the variation in the percentage of sites from which young fledged (r = -0.68, P < 0.05); adding August raindays increased this to 58% (r = -0.76, P < 0.01) (Fig. 1). Other correlations included August-September rain (r = -0.68, P < 0.05), July-August raindays (r = -0.61, P < 0.05), August-September and September-October temperature (r = 0.68, P < 0.05 and r = 0.77, P < 0.01, respectively). Temperature and rain were not independent; wet years were usually also cold. For August to October, the correlation between the number of raindays and mean maximum temperature was r = -0.77, P < 0.01 and that between total rain and temperature r = -0.88, P < 0.001. The number of raindays and total rainfall was negatively correlated with temperature in August, September, October and November (for raindays r = -0.75, -0.84, -0.75 and -0.83, respectively, and for rain r = -0.42, -0.59, -0.76, -0.75, respectively). There was no correlation between laying date and the number of young fledged per clutch laid, and the number of young fledged by late laying pairs was not significantly different from that of early laying pairs (Olsen & Olsen 1989). Hence, the deleterious effect of wet weather on breeding success was not due, to any great extent, to the slightly later laying observed in wet years. The population was monitored again in 1987. Based on the number of raindays and calculated from the equations in Table 2, the predicted number of young per territory should have been 1.19 and 59% of territories should have fledged young. The actual figures were 1.04 and 48%, respectively; lower than expected. However, rain was similar to 1983 and also came from an unusual direction as it did in that year when the equivalent figures were almost identical to 1987, at 1.03 and 47%. Effect of drought During the study there were five drought years (1976, 1977, 1979, 1980, 1982). In 1982, a severe drought year (annual rainfall 261 mm, long-term mean 623 mm), Peregrine Falcons near Canberra fledged more young than in any other year. This was due mainly TABLE 4 Breeding success of Peregrine Falcons near Canberra with only one nest site compared with those with two or three alternatives, according to the number of raindays between August and October. Wet years have more than the long term average of 32 raindays while dry years have < 32. Younglsite is annual mean. A successful site is one from which young fledge; number of sites in brackets. Significant differences are indicated. Clutches are divided according to raindays in the egg laying period (August-September raindays > 21 or < 21). No Alternative Alternative(s) Wet Dly Wet Dly Youngisite 0.66#$ 1.25$ 1.68# 1.43 %of sites successful 35%*t (72) 61%* (74) 75%t (51) 75% (56) % of clutches 30% 18% 18% 6% failing to hatch t Chi square (df: I) - 9.23, P < 0.01. * Chi-square (d.f 1) - 5.18, P < 0.05. # F (df: 1,lO) = 16.93, P< 0.01. $ F (df: 1,lO) = 12.26, P< 0.01 to the greater number of pairs breeding successfully and, to a lesser extent, to a relatively high number of young fledging per clutch laid (Table 1). Overall, for the drought years, young fledged from 65% of temtories compared with 53% in non-drought years; the corresponding figures for the number of young per temtory were 1.4 and 1.1. Weather, nest characte~tics and breeding success A choice of nest site In 12 of 30 temtories the Peregrine Falcons used two or three alternative nest sites, while at the remainder onlv one nest site was available. In wet vears territories witi alternative ledges had significantly getter reproductive success than territories where Falcons had no choice; in dry years there was little difference between the two groups (Table 4). In wet years, at territories with alternatives, Falcons most often nested at the site most protected from rain (Table 5). Futhermore, the number of young per temtory was negatively correlated with August- October raindays at sites without alternative ledges (r = -0.74, P< 0.01, Fig. 2) while there was no such correlation at sites with alternatives (r = -0.38, n.s.). Similar relationships were found with July-September raindays. Brood size, on the other hand, was unaffected by nest characteristics. Brood size was larger when October rainfall was high in both territories with and those without alternative nests, while in dry years both had smaller broods (Table 6). Type of nest site Peregrine Falcons nested either on ledges or in pot-holes; in wet years those nesting in potholes fledged more young than those on more exposed

10 OLSEN & OLSEN: BREEDING OF PEREGRINE FALCON m EMU 89 TABLE 5 Some characteristics of the nest sites available to pairs with one or more alternative ledges. The nest sites used in wet years (July-September raindays > 31) are underlined. Characteristics were assessed as follows. Aspect is the compass direction the nest site opened towards; for all of these nest sites except seven much of the bad weather came from the south-west (1 80-270'). In wet years the Peregrine Falcons avoided the nest site with that aspect and preferred the more easterly and, particularly, north-easterly facing nest site, which would have received the winter sun. Nest type: 1 - ledge, 2 = pot-hole. Drainage: 0 = didn't drain; 1 = some drainage; 2 = good drainage. Overhang: 0 = no overhang above nest; 1 = some overhang; 2 = well overhung, Substrate: a subjective assessment of whether there was no soil or debris on the nest floor (0), some substrate (I), or ample substrate (2); water drained more readily from sites with deep substrate and more windswept sites had less substrate. Shelter: a subjective assessment of whether the site was poorly protected from the elements (O), was moderately protected (1) or was well protected (2). Site no. Aspect Nest type Drainage Overhang Substrate Shelter - 181135 all 120 1341325 3201320 2001168 45/45 3471347 2941308 1251116 154112 2511150 1451120 c O 20 24 28 32 36 40 August/October raindays TABLE 6 Brood size of Peregrine Falcons near Canberra nesting in territories with either one nest (no alternative) or a choice of nests, according to October rainfall. A wet October had total rainfall greater than the long-term mean of 69 mrn; a dry October had less. Brood size (% of broods) Wet 1 2 3 Mean Alternative nest 3 (12%) 12 (48%) 10 (40%) 2.28 No alternative nest 1 ( 6%) 7 (50%) 6 (43%) 2.36 Dry Altemative nest 16 (29%) 29 (50%) 12 (21%) 1.93 No alternative nest 11 (20%) 35 (63%) 10 (18%) 1.98 FIGURE 2 The number of young fledged per nest site (territory) and August-October raindays each year 1975-1 986. Territories where there was a choice of nest (closed circles) are compared with sites where only one nest was available (open circles); the regression line for territories with no alternative nest is shown. For territories with no alternative nest the equation relating the number of young per territory (y) to August-October raindays (rd) was y - 2-0.03 (rd). ledges (Table 7). Lowered breeding success was not due to lowered occupancy or to smaller brood size. In dry years both types of site were approximately equal in breeding success. If only ledges are considered, then the ability of the nest site to drain becomes important and those pairs with poorly TABLE 7 Breeding success of Peregrine Falcons nesting on ledges compared with those nesting in a pot-hole, according to August to October raindays. A wet year is one in which the number of raindays in those three months exceeds 31. Mean number of young per pair each year, the percentage of sites occupied, and the annual mean number of young per successful pair (brood size). Wet Dry Ledge (n = 23) 0.91* 87% 2.00 1.43* 90% 1.98 Pot-hole (n = 6) 1.59# 86% 2.27 1.42# 89% 1.96 # F (d$ 1,27) = 4.19, P almost significant at 5% level. * F(d$ 1,45) = 28.79, P< 0.01.

1989 OLSEN & OLSEN: BREEDJNG OF PEREGRINE FALCON III 11 TABLE 8 The effect of rain on reproductive success of Peregrine Falcons using ledges on cliffs, rather than a pot-hole, according to whether the ledge was well-drained or not (see methods and Table 5). A wet year is one in raindays in August- October exceeded the longterm average of 31 ; a dry year has less than 31 raindays in that period. a. young per site; b. young per successful pair; c. the percentage of sites occupied by a pair and, in brackets, number of nest sites. Significant difference indicated. Nest ledge Wet (n - 6) Dry (n - 6) a. b. c. ( n) a. b. c. ( n) Drains 1.04*2.29 81% (13) 1.39 2.09 97% (13) Does not O X * 2.47 92% ( 8) 1.43 2.31 86% ( 8) drain drained sites produced fewer young in wet years (Table 8). This was not due to any decrease in occupancy or in brood size and so was probably due to failure of eggs to hatch or to loss of very young nestlings. Nest height from ground In wet years, some low nest sites appeared to stay wetter for a longer time than higher sites. On the other hand, higher sites were more windswept. Driving rain can enter even well-protected nest sites. However, while nine low nest sites (< 15 m) produced fewer young (1.0 younglsite) in wet years than in dry (1.3) and 19 high sites (> 15 m) did not (1.3 in both wet and dry years), none of the differences was significant. Further, the four highest nest sites (over 30 m), like the lowest, also produced fewer young in wet years (0.8) compared with dry (1.3). Any differences were unrelated to occupancy, which was similar for both low and high nest sites (Wet years: low nests 81%, high nests 87%. Dry years: low nests 92%, high nests 86%). Cliff height showed similar relationships. Aspect of nest Most inclement weather came from the south-west. However, at a few sites low mountain ranges deflected the weather and it arrived from a north-westerly direction. Only three of 29 nest cliffs faced the direction from which most inclement weather amves; they produced a yearly average of 0.7 young in wet years (August- October raindays > 3 1) compared with 1.4 in dry years. The equivalent figures for the other nest sites were 1.0 young in wet years and 1.4 in dry. Best vs worst nests Of the three best territories, with 100% occupancy and producing an average of two or more young a year, two had nests in pot-holes (one of which had an altemative nest on a ledge) and the third had two nests on ledges: all were well-protected from the prevailing weather and the ledges were well-drained. The four worst territories did not necessarily have lower occupancy. One territory was centred on a cliff with no ledge suitable for a nest but was occupied in most years. Another pair occupied a territory with no real nest site: they laid and failed every year until we altered a ledge for them so that it was larger and well-drained and they raised chicks for the first time. Neither of the two other worst territories had an alternative nest site, the nests at both did not drain and one faced the prevailing weather. Discussion Breeding success was high in the Canberra Peregrine Falcon population. During a two-year study in Victoria (Pmett-Jones et a1 1981) brood size at fledging was 2.14, similar to 2.16 for Canberra, but at least one young fledged from only 71% of clutches laid compared with 85% at Canberra so that overall production of young was lower. The annual number of young per pair was 1.1 for Victoria and 1.4 for Canberra. Allowing for the relatively small clutch of about 2.9 (Olsen & Olsen 1979), breeding success also compares favourably with that of Falcons that are elsewhere in the world little affected by pesticides. For example, Falcons in Scotland usually have a mean clutch size of about 3.6 and a brood size of 2.4 (Newton 1979; Ratcliffe 1980). Wet, cold years resulted in lowered production of young by the Canberra Peregrine Falcon population. Failure of clutches to hatch was associated with a high number of raindays during the egg laying period (July to September), while the number of young fledging was also adversely affected by a high number of raindays during this period and October, the month during which most broods are hatching. On the other hand, high October rain may have benefited those pairs that did not lose eggs or young chicks as the brood size of successful pairs was greater in wet than in dry Octobers. Rain, especially, and low temperatures affected breeding performance by: (1) destroying some nest sites; (2) lowering the number of pairs closely occupying sites possibly due to flooding of nests (Olsen & Olsen 1988b); (3) limiting the length of the egg laying period and the number of pairs laying at least partially because some nests are unavailable (wet) (Olsen & Olsen 1989); (4) increasing mortality of embryos (clutches) and young broods; and (5) increasing the chances of survival of nestlings that have escaped the direct adverse effects of bad weather. In general then, although exposure may have caused significant egg or chick loss in wet years, some aspect of food availability or quality may have resulted in larger surviving broods so that overall production of young by the population was not, on average, greatly below that of dry years

12 OLSEN & OLSEN: BREEDING OF PEREGRINE FALCON m EMU 89 (average difference of 18% between wet and dry years). The Peregrine Falcons thus had a breeding strategy that led to a fairly stable annual production of young. Cold temperatures contributed to losses possibly by adding to the likelihood of wet eggs or young being chilled (killed) and increasing the food needs of the chicks. The amount of rain had little effect on overall breeding performance (heavy rain caused clutches to fail but brood size was higher in years of high rainfall), but the number of raindays did. Perhaps persistent rain over a long period wets some nests (and eggslchicks) and they are unable to dry out. Prolonged rain also caused one nest site to collapse. Once broods have survived beyond the first few days after hatching their chances of fledging seem good. Possibly rain gave the adults, with surviving young, the advantage when hunting; Peregrine Falcons seem to have relatively waterproof plumage compared with many of their prey (e.g. Starlings Sturnus vulgarir and parrots) and may more easily capture rain-soaked prey. As fewer pairs bred successfully in wet years there may also have been less competition for food during the nestling period. Food quality may also be better in wet years, perhaps with high hormone levels (Olsen & Olsen 1987a). By improving food quality Weaver & Cade (1986) increased the proportion of fertile eggs they obtained from captive falcons by over 20%: chick vitality also improved markedly. When annual rainfall was low (drought) Peregrine Falcons in this study produced more young than in wet years. Baker-Gabb (1984) found that Falcons were one of the few raptors to breed in the severe drought of 1982 in his semi-arid, Mildura study area. In both study areas nest sites are scarce, and nest sites and social factors limit the size of the breeding population to well below the carrying capacity of the habitat, i.e., food may not be a proximate limiting resource. There was no change in the number of young fledged by a population of Canadian arctic Peregrine Falcons in a year of exceptionally high precipitation, compared to two drier years (Court 1985), indicating that food was not more limiting in the wetter year. These Canadian Peregrine Falcons had an abundance of large nests compared with the Canberra population so that flooding of nests may not be a problem there. Conversely, other raptors have suffered breeding failure thought to be attributable to limited food, both low prey numbers and low availability (e.g, long grass affords protection to prey) (Newton 1979; Ridpath & Brooker 1986). Black Eagles Aquila verreauxi generally bred more successfully in dry years than in wet (Gargett 1977) but breeding subsequently declined in some severely dry years (Gargett pers. comm.). Few or no Wedge-tailed Eagles A. audar bred in various parts of arid Western Australia during drought (Ridpath & Brooker 1986). Peregrine Falcons take a wide variety of aerial, avian prey which, in general, tends to be K-selected (in the sense of MacArthur & Wilson [19671) and less subject to dramatic fluctuations in numbers than the more r selected mammals and, occasionally, birds eaten by some other raptors like these large eagles. Ratcliffe (1980) could find no evidence that falcons vary seasonally, in number or breeding success, with their food supply as do, for example, Gyr Falcons Falco rusticoh. The adverse effect of weather on Peregrine Falcons has been commented on previously. Hagar (1969), for example, found that storms, particularly around hatching time, caused breeding failure. Ratcliffe (1980) noted the deaths of broods connected with cold, and often wet, springs in Britain and, in the Aleutians, White (1975) found fewer pairs breeding and more desertions in seasons when the weather was severe. McGrath (1987) reported that bad weather contributed to poor breeding performance in Irish peregrines. Newton (1979, 1986) and others attribute the greater part of the adverse effect of rain on raptors generally to food shortage due to lowered hunting efficiency and prey availability. This may be the case for some raptors nesting in stick nests that drain easily but there is growing evidence that raptors nesting on cliffs are vulnerable to the direct effects of bad weather. Prairie Falcons Falco mexicanus nesting on high, exposed buttes suffered high, apparently weather-related, breeding losses (Runde & Anderson 1986). Peregrine Falcons at two sites in the German Democratic Republic lose clutches in wet years and nest boxes have been installed to overcome the problem (Anon. 1986). In Victoria, Falcons fledged significantly more young from protected nest sites than from exposed sites but this was not linked to bad weather (Pruett-Jones et al 1981). At Rankin Bay, Canada, tundra Peregrine Falcons reused the same nest on only three of 68 nesting attempts and then never in successive years (Court 1985). Tasmanian Peregrine Falcons also had ample choice of nest site (Mooney & Brothers 1987). Near Canberra, in only about onethird of territories was there a choice of nest and nests were often of lower quality than would be used in areas where cliffs are abundant; the effects of weather were thus made more apparent. If rain-induced change in food supply was the primary factor influencing breeding success then pairs at all nest sites should have been affected similarly. However, it was mainly pairs at nest sites affording poor protection from the weather and having no choice of nest site in their territory that were affected and it was these that accounted for much of the variability in annual reproductive output by the population associated with weather. It could be argued that that adults at poor sites become rain soaked and their hunting impaired as a result. As males do most of the hunting to sustain the breeding attempt and have their own roost, and as brood size was not adversely affected by rain, it seems unlikely that food shortage was the major problem. No weather related brood reduction was detected and no deaths could be attributed to starvation in nestlings over a week old. The main effect of bad weather then seemed to be proximate, as a stochastic mal-

entity (Andrewartha & Birch 1984). Presumably weather, food and the availability of a dry or otherwise useable nest can all act as proximate factors influencing breeding and overriding each other in importance depending on local conditions. In this study, nests at medium height, in pot-holes, welldrained and facing away from the prevailing weather are the most secure from the adverse effects of weather. Nevertheless, even well-protected nests can be flooded if the rain is blown in by wind from an unusual direction and these sheltered nests probably dry slowly. Local weather is affected by landform so that characteristics affording protection from the extremes of weather at one site may not be appropriate at another. After the severe drought, ending in 1983, the run-off from ground bare of vegetation and baked hard by the sun seemed to pour into some Canberra nests in greater quantities than the same amount of rain would have caused following wetter years; in spite of there being only an average number of raindays many pairs failed to breed. The relationship of weather to breeding success is clearly important but complex. Preference for a particular nest or nest type may be implied if high numbers of young are fledged from these nests and they are used more often than poorer sites. There was no correlation between the number of young per territory and occupancy (percentage of territories occupied by a pair) or between occupancy and breeding success (percentage of territories from which young fledged) suggesting that there was no preference for high quality nests. However, because adults were not marked, we have no information on turnover and cliffs were so scarce they would be unlikely to be abandoned regardless of quality. Olsen & Olsen (1987b) proposed that in raptors such as the Peregrine Falcon, the females compete for a scarce resource, i.e. a male who is a good provider with a territory. Females paired with a male in possession of good quality, all-weather nest@) would have a selective advantage over those with poor nests. Natural selection dictates that individuals vary in their genetic contribution to future generations (Darwin 187 1); life-time reproductive rate is thought to give a good estimate of this biological fitness (Clutton Brock 1986). Falcon pairs holding a territory with alternative nests would have fledged, on average, 17 young during this 12 year study; those with only one nest would have fledged 11.5. Similarly, those nesting in a pot-hole would have fledged 16 young, those on a ledge 12. Acknowledgements A number of friends assisted with field work, they include Robert Bartos, David Mallinson, Peter Ormay, Tony Ross, Jack Thompson and Jane and Don Richardson. The CSIRO Science and Industry Fund and N.S.W. ANPWS Wildlife Foundation partially funded the study, Frank Knight prepared the figures and Peter Jarman commented on the manuscript. We are grateful to them all. References Andrewartha, H.G. & Birch, L.C. (1984). The Ecological Web. University of Chicago Press, Chicago. Anon. (1986). Raptor conference in the GDR. Newsletter of the World Working Group on Birds of Prey & Owls 5,9-11. Baker-Gabb, D. J. (1984). The breeding ecology of twelve species of diurnal raptor in north-western Victoria. Aust. WUL Res. 11,145-160. Cade, T.J. & Fyfe, R.W. (1970). The North American peregrine survey, 1970. Canadian Field Nut 84,231-245. Clutton-Brock, T.H. (1986). Reproductive Success. University of Chicago Press, Chicago. Court, G.S. (1985). Some Aspects of the reproductive Biology of Tundra Peregrine Falcons. MSc thesis, University of Alberta. Darwin, C. (1871). The Descent of Man and Natural Selection in Relation to Sex. Murray, London. Gargett, V. (1977). A 15-year study of the Black Eagles of the Matopos, Rhodesia, 1964-1976. Ostrich 48, 17-27. Hagar, J.A. (1969). History of the Massachusetts Peregrine Falcon population, 1935-1957. In: Peregrine Falcon Populations: Their Bioloa and Decline (ed. J.J. Hickey) pp. 123-132. University of Wisconsin Press, Madison. Hickey, J.J. (ed.) (1969). Peregrine Falcon Populations. Their Biology and Decline. University of Wisconsin Press, Madison. MacArthur, R.H. & Wilson, E.O. (1967). The Theory of Island Biogeography. Princeton University Press, Princeton. McGrath, D. (1987). The Peregrine Falcon in south-east Ireland, 1981-1986. Irish Birds 3,377-386. Mooney, N.J. & Brothers, N.P. (1987). The Peregrine Falcon Falco peregrinus macropus S., in Tasmania. I. Distribution, abundance and physical characteristics of nests. Aust. WiM Res. 14, 81-94. Newton, I. (1979). Population Ecology of Raptors Poyser, Calton. Newton, I. (1986). The Sparrowhawk Poyser, Calton. Olsen, P.D. (1982). Ecogeographic and temporal variation in the eggs and nests of the peregrine, Falco peregrinus (Aves: Falconidael in Australia. Aust. Wild1 Res. 9.277-292. Olsen, P.D. (1985).'~o~ulation studies of the Peregrine in Australia. ICBP Tech. Pub1 No. 5,381-388. Olsen, P.D. & Olsen,J. (1979). Eggshell thinning in the Peregrine, Falco peregrinus (Aves: Falconidae), in Australia. Aust. Wildl Res. 6,217-226. Olsen, P.D. & Olsen, J. (1987a). Are hormone levels in prey proximate timers for breeding in raptors? Australasian Raptor Association News 8,49. Olsen, P.D. & Olsen, J. (1987b). Sexual size dimorphism in raptors: Intrasexual competition in the larger sex for a scarce breeding resource, the smaller sex. Emu 87,59-62. Olsen, P.D. & Olsen, J. (1988a). Population trends, distributional and numerical status of the Peregrine in Australia. In: Peregrine Falcon Populations: Their Management and Recovery (eds T. Cade, C. White, J. Enderson & C. Thelander). The Peregrine Fund, New York. Olsen, P.D. & Olsen, J. (1988b). Breeding of the Peregrine Falcon Fako peregr'nus: I. Weather, nest spacing and territory occupancy. Emu 88, 195-201. Olsen, P.D. & Olsen, J. (1989). Breeding of the Peregrine Falcon Fako peregrinus: 11. Weather, nest quality and the timing of egg laying. Emu 89, 1-5. Pruett-Jones, S.G., White, C.M. & Devine, W.R. (1981). Breeding

OLSEN & OLSEN: BREEDING OF PEREGRINE FALCON JII EMU 89 of the Peregrine Falcon in Victoria, Australia. Emu SO, supply in arid Western Australia. Ibis 128, 177-194. 253-269. Runde, D.E. & Anderson, S.H. (1986). Characteristics of cliffs and Ratcliffe, D.A. (1970). Changes attributable to pesticides in egg nest sites used by breeding Prairie Falcons. Raptor 20, breakage frequency and eggshell thickness in some British 21-28. birds. J. AppL Ecol 7,67-107. Weaver, J.D. & Cade, T.J. (1986). Falcon Propagation. A Manual Ratcliffe, D.A. (1980). The Pereg&e Falcon Poyser, Calton. on captive Breeding. The Peregrine Fund, New York. Ridpath, M.G. & Brooker, M.G. (1986). The breeding of the White, C.M. (1975). Studies on Peregrine Falcons in the Aleutian Wedge-tailed Eagle Aquila audux in relation to its food Islands. Raptor Research Report 3,33-50.