From individual behaviour to population pattern: weather-dependent foraging and breeding performance in black kites

ANIMAL BEHAVIOUR, 2003, 66, 1109–1117 doi:10.1006/anbe.2003.2303

From individual behaviour to population pattern: weather-dependent foraging and breeding performance in black kites FABRIZIO SERGIO

Edward Grey Institute of Field Ornithology, Department of Zoology, University of Oxford, U.K. (Received 21 November 2002; initial acceptance 24 January 2003; final acceptance 20 May 2003; MS. number: 7538)

The links between weather and animal behaviour in population processes have received relatively little attention. I studied the effect of weather conditions on foraging and breeding performance of a medium-sized raptor, the black kite, Milvus migrans. The frequency of prey capture attempts and their likelihood of success increased with temperature and declined with rainfall. Kites used flight styles involving a higher energy expenditure in less favourable weather. Nestling provisioning rates declined during rain spells. More kites hunted during periods of favourable weather and after periods with a high frequency of successful prey capture attempts by conspecifics; this result suggested that individuals may fine-tune their foraging effort to the expected foraging reward. Such compensatory behavioural adjustments may increase species resilience to climate change. The behaviourally mediated effects of weather on prey availability translated into population effects. Yearly weather conditions during the last stage of the prelaying period affected population-level productivity, probably through an effect on female body condition mediated by male provisioning capability and hunting yield. The predicted effects of climate change on kites may already be occurring, with progressively earlier laying and northward range expansion. These results confirm the need to pay greater attention to behaviourally mediated effects of climate on populations, particularly when individuals make compensatory adjustments that may enhance resilience to climate change. 

2003 The Association for the Study of Animal Behaviour. Published by Elsevier Ltd. All rights reserved.

Weather can affect individual behaviour, natality, mortality and dispersal, with potential repercussions on population and community regulation (Newton 1986; Elkins 1988; Sæther et al. 2000; Thompson & Ollason 2001). The effect of weather can be direct, for example by altering metabolic costs (e.g. Machmer & Ydenberg 1990), and indirect, for example by affecting food availability and abundance (e.g. Grant et al. 2000). There is growing evidence of human-induced rapid change of global climates, caused by increased levels of ‘greenhouse gases’ in the atmosphere. Current models predict potentially contrasting trends in different regions, with the highest impacts at northern latitudes (40–70N), where temperature is expected to increase by 0.5–2C in the next 50 years; precipitation may also increase at northern latitudes (Houghton et al. 1996). The consequences of such changes on animal and plant populations have already been demonstrated in numerous taxa (review in Hughes 2000). Most of these studies have documented Correspondence and present address: F. Sergio, Raptor Conservation Research Unit, Trento Natural History Museum, Via Calepina 14, 38100 Trento, Italy (email: [email protected]). 0003–3472/03/$30.00/0



changes in species distribution and phenology, usually interpreted as responses to climate-induced variation in metabolic expenditure, foraging performance and food abundance or availability. However, in-depth knowledge of the effect of weather conditions on both individual behaviour and population ecology is a prerequisite for biologically meaningful predictions of the impact of climate change on animal populations. Unfortunately, for most species to date there is little information on individual or population responses to variations in weather conditions. Furthermore, for those species that have been studied, research has focused on either individual performance or population responses (e.g. Marti 1994; Steenhof et al. 1997; Thomson et al. 1997; Thompson & Ollason 2001). Therefore, most predictions of future climate impacts are purely qualitative and usually based on supposed effects on the phenology and abundance of food resources (e.g. Burton 1995; Moss 1998). Furthermore, in most studies, the mechanisms by which weather affects individuals or populations are inferred rather than explicitly examined (Redpath et al. 2002). In such cases, unfavourable weather

1109 2003 The Association for the Study of Animal Behaviour. Published by Elsevier Ltd. All rights reserved.

ANIMAL BEHAVIOUR, 66, 6

is often suggested to impair foraging performance of parent individuals, thus decreasing food availability and eventually affecting breeding success (e.g. Newton 1986; Steenhof et al. 1997; Dawson & Bortolotti 2000). Despite such long-recognized effects on behaviour, relatively few studies have explicitly examined weather-dependent foraging performance (e.g. Dunn 1973; Grubb 1977; Avery & Krebs 1984; Flemming & Smith 1990), even though there is a large body of literature on optimal foraging (e.g. Stephens & Krebs 1986). Furthermore, few studies have tested whether animals use compensatory behavioural adjustments that allow them to minimize the effect of weather on foraging and reproduction (Grubb 1975; Martin 2000; Redpath et al. 2002). For example, individuals making optimal foraging decisions may be expected to recognize weather conditions that yield poor foraging rewards and to adjust their foraging decisions accordingly (adaptive foraging effort: Daan 1982). Such behavioural adjustments could enhance species resilience to climate change. Therefore, additional focus is needed on the links between weather and animal behaviour in population processes (Rotenberry & Wiens 1991; Post et al. 1999). I used data from a 10-year study of a medium-sized raptor, the black kite, Milvus migrans, to test the following predictions: (1) weather influences foraging success; (2) weather affects foraging flight style and thus potential foraging energy expenditure; (3) individuals minimize the impact of adverse weather through adaptive foraging effort; (4) weather affects parental provisioning rates; (5) behavioural processes at the level of the individual translate into population effects: spatiotemporal variations in population density and breeding success are related to spatiotemporal variations in weather conditions. The black kite is an opportunistic predator that exploits spatiotemporal food concentrations (Forero et al. 1999; Vin ˜ uela 2000; Sergio et al. 2003a, b). Its European breeding populations are declining (Vin ˜ uela & Sunyer 1994; Sergio et al. 2003b), pointing to an urgent need to assess potential future effects of climate change. Studies on black kites have been conducted exclusively at the individual level and have highlighted how unfavourable weather conditions may affect the commencement of incubation and subsequent hatching asynchrony (Vin ˜ uela 2000), the growth rate of nestlings (Hiraldo et al. 1990) and the flight-style and its energy costs during migration (Spaar 1997). Rain is generally suggested to impair or prevent foraging, and the effect of higher temperature is usually considered positive (e.g. Henty 1977; Hiraldo et al. 1990), even though such relationships have never been quantitatively tested. Therefore, the black kite qualifies as a useful model species to investigate the effect of weather conditions on foraging and reproduction in an opportunistic predator. METHODS

Study Site I conducted fieldwork in 1992–2001 in a study plot along the Italian shore of Lake Lugano (4554 N, 855 E). Besides the open water of the lake, the landscape was

Monthly rainfall (mm)

1110

300 280 260 240 220 200 180 160 140 120 100 80 60 40 20 0

J

F

M

A

M

J J Month

A

S

O

N

D

Figure 1. Mean±SE monthly rainfall over 10 years (1992–2001), and recorded at a weather-recording station 4 km north of the study area (courtesy of ‘Meteosvizzera’).

characterized by mountain slopes covered by deciduous woodland, interrupted by cliffs and scarce open areas (Sergio et al. 2003b). Climate is temperate continental (Pinna 1978). During the 10 years of study, cumulative yearly rainfall varied between 1200 and 2100 mm, with two peaks in spring and autumn (Fig. 1). The study plot supported a slightly increasing black kite population of 11–30 territorial pairs (Sergio et al. 2003b), most of them breeding within loose colonies of up to 16 pairs. Black kites settled on territory after spring migration between 17 March and 5 April and started their clutches, on average, on 24 April; the diet was strongly dominated by fish, caught within undefended (communal) foraging grounds (Sergio & Boto 1999). Pairs with a diet dominated by fish had higher than average breeding output (Sergio et al. 2003c). Breeding success was low (Sergio et al. 2003b), and most failures were caused by nonlaying (16.1% of 118 breeding attempts) or failure during incubation (29.5%). Chick survival was high; of 139 nestlings first checked when less than 5 days old (N=72 nests), only 3.6% died before fledging.

Foraging Performance To assess the effect of weather on foraging performance, I conducted systematic observations of fishing kites during the kite nestling period (May–June) of 1994 and 1995 (N=105 h in 19 days). Observation days were randomly selected a priori so as to sample prevailing weather conditions experienced by kites in an average year. All observations were conducted from a fixed station on the lake shore below a colony of nine and 12 pairs in 1994 and 1995, respectively. During each observation session, I recorded all attempts to capture prey (strike attempts) and their outcome. An attempt was classified as successful if a fish was spotted in the talons, or if the bird was seen to feed on the captured item. If the same individual tried to catch prey many times, only one randomly selected attempt was retained for analysis to minimize

SERGIO: WEATHER-BASED FORAGING AND REPRODUCTION

pseudoreplication. Every 20 min, I also recorded the number of kites foraging over a fixed 1.5-km2 portion of the lake. I assumed that scans with a larger collective number of hunting individuals per unit space corresponded to a higher mean per capita foraging effort and define such a variable as ‘foraging effort’. The scanned portion of the lake was identified by natural and purposely placed markers along the shore and in the water. During the observation sessions and opportunistically during other days and on other portions of the lake, I also observed foraging individuals (focal animals) by following them as long as possible through focal animal sampling (Altmann 1974). For these I recorded, with a stopwatch, the number of strike attempts/min, the number of successful strikes/min, and hunting time(s) spent in flapping flight compared to soaring. Only individuals followed for at least 10 min were retained for analysis. I recorded environmental variables after each strike attempt, count of foraging kites and focal animal sampling session. These included temperature (measured with a digital thermometer in shade at 2-m height), presence or absence of rain, wind speed (measured with a hand-held cup anemometer), cloudiness percentage (visually estimated by reference to measured photographs); rippling of the water surface (measured on a Beaufort ordinal scale), and number of fishermen in the area (to control for potential human disturbance). For each strike attempt, I also visually estimated the distance to the shore (in 10-m increments), and I recorded the presence or absence of other kites within 50 m of the striking one to test the potential effect of interference. I did not measure water transparency, because kites capture fish only in the top 5–10 cm of the water column. Black kite density in the study area increased from 51.4 pairs/100 km2 in 1994 to 60.0 pairs/100 km2 in 1995, and the meanSE number of fledged young was higher in 1995 (1.050.22, N=19) than in 1994 (0.470.19, N=17; ANOVA: F1,34 =3.80, P=0.05). The higher density and productivity in 1995 could have been caused by higher food availability or by better weather in that year. However, weather conditions were worse in 1995 for kite breeding and foraging (Sergio 2003, unpublished data), hence, by exclusion, food availability was likely to be higher in 1995. Therefore, I predicted that, independently of weather, foraging performance would be better in 1995 than in 1994. Colonial animals may improve their foraging performance by observing their foraging conspecifics (local enhancement: Buckley 1997). To test this possibility, I hypothesized that foraging effort would increase after periods with a high frequency of successful prey capture attempts, independently of weather conditions. I calculated the number of successful strikes by all the foraging individuals in the observed area in the 10 min before foraging effort was recorded, and included it as an explanatory variable in the model that examined the factors affecting foraging effort. Invariably, foraging individuals passed quickly through the sample area, usually remaining within it a minute at the most; therefore, the individuals counted to estimate foraging effort were never the same as those that captured prey in the previous

10 min; thus, foraging effort was expected to be independent and not necessarily correlated with the number of successful strikes in the previous 10 min unless conspecific attraction was involved. Because kite pairs in my area concentrate their foraging effort within 1 km of the nest (Sergio et al. 2003d), this study design is likely to involve a certain degree of repeated sampling of the same individuals. Therefore, to minimize the effect of pseudoreplication, during the nestling period of 2000, I conducted further systematic observations of fishing kites at 46 observation stations scattered along the shore of Lake Lugano with a minimum distance of 2 km between them. I randomly selected the day that each station was visited a priori, ensuring that neighbouring stations were not sampled during the same day. During each of the 46 observation sessions, I recorded (1) the success or failure of the first observed prey capture attempt; (2) the number of strike attempts/min, the number of successful strikes/min, and the percentage of time spent in flapping flight by the first individual that could be followed through focal animal sampling for more than 10 min; (3) the number of kites foraging over the lake area within 1 km of unobstructed view of the observation station (always recorded 20 min after the beginning of the observation session). This allowed me to rerun all the analyses on a sample of 46 spatially independent records.

Biomass Delivered to the Nestlings Nest watches were conducted at 38 nests, each from a different territory to avoid pseudoreplication. Nests were observed for 6–11 h from vantage points through a 20–60 telescope or from a hide built near the nest. Nests were chosen so that their cup was clearly visible from the observation point, which allowed recognition of all delivered prey items and approximate estimation of their mass. To minimize the effect of chick age on delivered biomass, I observed only nests with well grown chicks, 30–45 days old.

Reproduction Kites were intensively monitored between 1992 and 2001. The study site was visited every 1–4 days throughout the prelaying period. Territorial pairs were censused by watching territorial displays and transfers of nest material. Accessible nests (N=17–25 each year) were checked at least three times to record clutch size, hatching success, brood size at hatching and number of fledged young (Sergio & Boto 1999; Sergio et al. 2003a, b). Yearly variation in mean weather conditions was assessed through daily data from a meteorological station near the town of Lugano, 4 km from the study site. To assess the effect of weather on density and breeding performance, I first subdivided the kite breeding cycle into potentially sensitive periods to test the effect of biologically meaningful weather variables. The breeding cycle of black kites follows the typical raptor pattern, with

1111

1112

ANIMAL BEHAVIOUR, 66, 6

Table 1. Stages of the black kite breeding cycle within which mean weather conditions were estimated, and their predicted effects on density and reproduction Period

Breeding stage

17 March–4 April 5 April–24 April

Settlement on territory 20 days before egg laying

25 April–24 May

Incubation

25 May–4 June

First 10 days after hatching

Predicted effect of weather conditions on

Density of territorial pairs Mean clutch size, hatching success, brood size and mean number of fledged young per breeding pair Mean hatching success, brood size and mean number of fledged young per breeding pair Mean brood size and mean number of fledged young per breeding pair

pronounced division of roles between the sexes (Newton 1979; F. Sergio, unpublished data). During the prelaying period, females depend largely on the males for food through courtship feeding; they become increasingly inactive and accumulate fat reserves, which are crucial for laying eggs and buffering females from potential food shortages during incubation and the early nestling period (Newton 1979; Meijer et al. 1989). Only females that accumulate sufficient body reserves manage to lay and successfully accomplish incubation; the period of 15–20 days before laying is thus considered the most crucial stage of the breeding cycle (Newton 1979; Meijer et al. 1989). During incubation and most of the nestling period, the male provides food for himself, the female and the brood. Chick mortality is highest during the first few days after hatching. Based on this information, I divided the breeding cycle into biologically sensitive periods (Table 1). I did not test the effect of weather in the nestling period because chick mortality in this population is very low. During each of the biologically sensitive periods. I calculated the following variables: rainfall (mm), number of rain days (d0.2 mm of rain), mean daily temperature and mean daily wind speed. I tested the effect of such variables on yearly kite territorial pair density, clutch size, hatching success, brood size and number of fledged young, following the predictions of Table 1.

Data Analyses I examined the effects of environmental variables on foraging performance and provisioning rates through generalized linear models (GLMs, software GLIM 4: McCullagh & Nelder 1983). Multiple and logistic regression GLMs were built by fitting year (as a factor variable), date, time of day and environmental variables (plus their first-order interactions) as explanatory variables, and by using as dependent variables the outcome of a strike attempt (successful or not), the number of strike attempts/min, the number of successful strikes/min, the percentage of time spent in flapping flight, foraging effort and the biomass of prey delivered to the nest/h (see footnote to Table 2). For analyses of provisioning rates, I added as covariates the number of chicks in the nest and the distance of the nest to the lake shore, so as to control for brood size and territory quality (Sergio & Newton 2003; Sergio et al. 2003a, c, d). All GLMs were built by a backward elimination procedure, (Crawley 1993): all

explanatory variables were fitted to the model, extracted one at a time from the maximal model, and the associated change in model deviance assessed by the significance (P≤0.05) of an F test (or a
2 test for GLM logistic regression with binomial errors and a logit link function). For analyses at the population level, I tested the effect of the weather variables in Table 1 on yearly kite territorial pair density, clutch size, hatching success, brood size and number of fledged young by means of linear regression (Sokal & Rohlf 1981). All tests are two tailed, statistical significance was set at c0.05, and all means are given 1 SE. Whenever multiple tests were made on the same data set, I used Bonferroni sequential correction to adjust the significance levels (Rice 1989). RESULTS

Foraging Performance Pooling data from 1994 and 1995, I observed 367 prey capture attempts, 66% of which were successful. On all occasions on which I could observe a successful attempt from nearby (<40 m, N=42), kites captured live fish. Kleptoparasitic attempts were observed after 4.9% of the 244 successful attacks, but none of them was successful. Interference affected striking success, which was lower when there was another kite within 50 m of the attacking one (49%, N=45) than when no other kite was nearby (69%, N=322;
21 =6.77, P=0.009). However, conspecific interference did not enter the multivariate model. The probability that a strike attempt was successful was higher in 1995 than in 1994 increased with temperature and declined with increasing rainfall (Table 2). Both the number of strike attempts/min and the number of successful strikes/min by focal individuals increased with increasing temperature (Table 2). The percentage of hunting time spent in flapping flight decreased with increasing temperature and increased with rain (Table 2). The interaction of rain and temperature was also significant: the effect of temperature was more pronounced during rain and the effect of rain more pronounced at lower temperatures. Foraging effort was higher in 1995 than in 1994, increased at higher temperatures, declined during periods of rain and increased with the number of successful prey capture attempts by other kites in the previous 10 min (Table 2). To examine whether parent individuals hunted more during spells of unfavourable weather to meet

SERGIO: WEATHER-BASED FORAGING AND REPRODUCTION

Table 2. Effect of environmental variables on foraging performance and rate of prey delivery at the nest by black kites in the Italian pre-Alps Variable

Probability of a strike being successful (N=367)* Year† Temperature‡ Rainfall† Constant Number of strike attempts/min (N=90)‡** Temperature‡ Constant Number of successful strikes/min (N=90)‡** Temperature‡ Constant % Hunting time spent in flapping flight (N=76)**†† Temperature‡ Rainfall† Interaction term: temperature×rainfall Constant Foraging effort (N=254)‡** Year† Temperature‡ Rainfall† No. of successful strikes by other kites in previous 10 min‡ Constant Prey biomass delivered to nest/h (N=38)‡** Year† Distance to lake shore‡ Constant

Parameter estimate±SE

t

0.63±0.30 0.17±0.03 −1.40±0.50 −3.19±0.79

10.3§ 6.5§ 7.6§ —

1 1 1

<0.01 <0.05 <0.02 —

0.26±0.09 −0.61±0.27

2.80 —

79

<0.01 —

0.31±0.11 −0.70±0.34

2.69 —

79

<0.01 —

−0.37±0.15 4.01±1.00 −1.26±0.35 1.58±0.45

2.40 4.01 3.60 —

65 65 65

<0.02 <0.001 <0.001 —

0.31±0.08 0.43±0.18 −0.61±0.10 0.30±0.06 −0.75±0.55

3.90 2.46 6.29 4.56 —

241 241 241 241

<0.001 <0.02 <0.0001 <0.001 —

−2.24±0.56 −0.43±0.18 6.33±1.18

4.01 2.41 —

31 31

<0.001 <0.05 —

df

P

% Deviance explained

89.2

8.2 7.6 53.5

35.1

35.2

Explanatory variables entered into GLM analyses: all dependent variables: year, date, time of day, temperature, rainfall, wind speed, cloudiness percentage, rippling of water surface; strike success: number of fishermen, distance to shore, presence of other kites within 50 m; foraging effort: number of fishermen, number of successful strikes by other kites in previous 10 min. Sample sizes are given in parentheses. *GLM logistic regression with binomial errors and a logit link function (Crawley 1993). †Categorical, dichotomous variable. ‡Variable ln transformed. §Tested by means of a χ2 test. **GLM linear regression with normal errors and an identity link function (Crawley 1993). ††Variable converted to a proportion and arcsine square-root transformed.

increased energy demands of the nestlings, I added as covariates to the previous model the average rain and temperature during the previous day and during the previous 2 days. The only significant additional effect was that of rain in the previous 2 days, but its effect was negative (parameter estimate=
0.100.02, t237 =4.20, P<0.001). Thus, kites foraged less during prolonged periods of rain. The effect of rain or temperature in even longer periods was not significant. All these results were confirmed by the analyses based on the 46 observation sessions conducted at 46 widely spaced locations (Table 3). The only difference was the absence of a significant effect on flight style by the interaction between temperature and rainfall (Table 3).

Population-level Reproduction Within the Lake Lugano population, the only significant effect of weather on reproduction, after Bonferroni adjustment of significance levels, occurred during the last 20 days before egg laying. Hatching success was positively related to temperature during this period (Fig. 2a). As a result, the mean number of fledged young per nest was also positively related to mean daily temperature during the last 20 days of the prelaying period (Fig. 2b). This effect was caused mainly by desertion of a large number of clutches after prelaying periods characterized by low temperatures (Fig. 2c). DISCUSSION

Biomass Delivered to the Nestlings The biomass of prey delivered to the nest per h declined with rainfall and with increasing distance to the nearest lake shore (Table 2). This result did not change when I added as explanatory variables mean weather conditions in the preceding 1–5 days.

Weather conditions affected all measures of foraging and provisioning. Higher temperatures allowed kites to hunt mainly by gliding and low soaring, which involve an energy expenditure of at least 11% of that required for flapping flight, for a bird with the body size of a black kite (Pennycuick 1989). Higher temperatures also increased the likelihood of successful capture once an attack was

1113

1114

ANIMAL BEHAVIOUR, 66, 6

Table 3. Effect of environmental variables on foraging performance by black kites in the Italian pre-Alps: data from 46 observation sessions conducted at 46 stations located at 2-km intervals along the shore of Lake Lugano (see Methods) Parameter estimate±SE

t

df

P

0.29±0.12 −10.63±24.74 −4.62±2.40

7.55‡ 16.55‡ —

1 1

<0.02 <0.001 —

0.75±0.12 −1.97±0.33

6.46 —

36

<0.001 —

0.55±0.01 −1.44±0.25

6.13 —

36

<0.001 —

−0.45±0.17 0.40±0.13 1.82±0.52

2.60 3.04 —

36 36

<0.02 <0.01 —

1.05±0.34 −0.52±0.18 0.34±0.12 −3.39±1.17

3.13 2.84 2.93 —

Variable

Probability of a strike being successful* Temperature† Rainfall§ Constant Number of strike attempts/min†** Temperature† Constant Number of successful strikes/min†** Temperature† Constant % Hunting time spent in flapping flight**†† Temperature† Rainfall§ Constant Foraging effort†**‡‡ Temperature† Rainfall§ No. of successful strikes in previous 10 min† Constant

% Deviance explained

49.6

48.6 46.1 33.1

63.1 <0.01 <0.01 <0.01 —

*GLM logistic regression with binomial errors and a logit link function (Crawley 1993). †Variable ln transformed. ‡Tested by means of a χ2 test. §Categorical, dichotomous variable. **GLM linear regression with normal errors and an identity link function (Crawley 1993). ††Variable converted to a proportion and arcsine square-root transformed. ‡‡Measured as number of kites foraging over the lake area within 1 km of unobstructed view of the observer and standardized per unit area (number of kites/km2, N=46).

initiated, eventually resulting in a higher rate of successful strikes per unit time. These direct, positive effects on kite-foraging capabilities might have been enhanced by indirect effects on prey. For example, many fish species lower their vigilance against predators at high temperatures (Matthews 1998). Rain had a more abrupt effect than temperature and usually fully prevented any successful prey capture, probably by impairing flight and prey visibility. Optimal foraging theory predicts that prey exploitation is structured to maximize fitness, with foraging performance often used as an estimate of fitness (Stephens & Krebs 1986). Predators making optimal foraging decisions are expected to recognize conditions of high prey availability and adjust their decision to hunt in response to this expectation (Daan 1982). However, despite its strategic importance, few studies have examined the decision of when to hunt (e.g. Daan 1982; Flemming & Smith 1990). In agreement with theoretical predictions (Stephens & Krebs 1986), kite-foraging effort was highest during periods of maximum hunting reward. Kites hunted in higher numbers at higher temperatures and typically stopped hunting altogether as soon as it started to rain. These weather conditions were tightly related to the expected foraging yield and to flight styles associated with a lower energetic expenditure. Redpath et al. (2002) reported that parent hen harriers, Circus cyaneus, compensated for the increased thermoregulatory needs of nestlings by higher provisioning

rates during adverse weather (low temperature). I found no such effect for black kites. Prey deliveries to the nest did not increase during prolonged spells of adverse weather, and continuous rainfall resulted in decreased foraging effort, as also shown for sparrowhawks, Accipiter nisus (Newton 1978). The difference from the pattern described by Redpath et al. (2002) for hen harriers in Scotland might have been linked to the higher temperatures experienced by nestlings at the more southern latitude of my study. However, results similar to mine were obtained on sparrowhawks at a latitude similar to that of the harrier study (Newton 1978). Alternatively, harriers might have been able to hunt more proficiently than kites in unfavourable weather, with higher rewards despite the likely flight-related energy costs. In my area, spells of bad weather were also generally short. For example, during the nestling period, the mean number of consecutive days with rain in the 10 years of study was 2.3 (range 1–6). Furthermore, during such days, rain was not constant; periods of rain were typically interspersed with some dry spells. Kites were often observed rapidly to exploit such favourable ‘time windows’ for hunting. This short-term decision of when to hunt may have partly prevented the need to compensate for apparently prolonged periods of inclement weather. Finally, the decision to minimize hunting activity during bad weather may be integral to a wider strategy of prey exploitation. Black kites are opportunistic predators whose ecology is based on adaptation to ephemeral, spatiotemporal

SERGIO: WEATHER-BASED FORAGING AND REPRODUCTION

1

(a)

Hatching success

0.9 0.8 0.7 0.6 0.5 0.4

Mean no. of young fledged per breeding pair

0.3 2.5 2.3 2.1 1.9 1.7 1.5 1.3 1.1 0.9 0.7 0.5

9

10

11

12

13

14

10

11

12

13

14

13

14

(b)

9

Proportion of clutches deserted during incubation

0.7 (c)

0.6 0.5 0.4 0.3 0.2 0.1 0

9

10

11 12 Temperature (°C)

Figure 2. Effect of temperature in the 20 days before egg laying on (a) hatching success, (b) mean number of fledged young per breeding pair and (c) the proportion of clutches deserted during incubation by black kites in the Italian pre-Alps (1992–2001). Regression equations (temperature (T) always square-root transformed): (a) hatching success (arcsine square-root transformed)= 1.86T−5.11, F1,8 =47.3, P=0.0002, R2 =0.84; (b) mean number of young fledged per breeding pair (square-root transformed)= 1.37T−3.52, F1,8 =31.9, P=0.0096, R2 =0.80; (c) proportion of clutches deserted during incubation (arcsine square-root transformed)= −1.35T+5.10, F1,8 =50.2, P=0.0001, R2 =0.84.

overabundance of easy prey (Vin ˜ uela 2000). As a consequence of this feeding strategy, periods of food scarcity with no prey delivery to the female or nestlings may be followed by periods of overabundance in which prey accumulate unconsumed in the nest (Vin ˜ uela 2000; F. Sergio, personal observation). Nestling growth rates are probably adapted to such a pattern, and, compared with other raptors, black kite chicks are capable of delayed growth during unfavourable conditions, and rapid growth recovery during subsequent favourable spells (Hiraldo et al. 1990; Veiga & Hiraldo 1990). Nestlings with starvation symptoms and delayed growth may thus

recover and successfully fledge, so that a prolonged period of low food availability may not be indicative of the potential for rearing the brood to fledging (Vin ˜ uela 2000). These dynamics may also explain the low nestling mortality in my study area. Independently of weather conditions, kites also adjusted their foraging effort on the basis of the behaviour of conspecifics; periods with more successful prey capture attempts were followed by higher numbers of foraging individuals. This dynamic was probably caused by local enhancement mechanisms (Buckley 1997). The area where foraging effort was recorded was below two colonies on the opposing shores of the lake stretch. Kites were regularly seen to perch on dead trees and cliffs on parts of the mountain slopes with a clear view of the lake. The birds at most nests had direct good visibility of the lake. In such conditions, it might have been easy for kites to monitor the performance of their conspecifics and adjust their foraging decisions accordingly. This may again be integral to an overall strategy of exploitation of ephemeral food resources. On many occasions (although never during the intensive sessions of observation), I observed kites congregating in large numbers and extremely rapidly at a fish shoal. In one case, 40 hunting individuals congregated within 4 min of the first one starting to plunge repeatedly at a fish shoal; some of them were observed to take off from perches near their nest. Such foraging facilitation must be an important benefit or determinant of colonial behaviour (e.g. Beauchamp 1999). Whatever the mechanism behind it, the capability of kites to recognize weather conditions related to high foraging reward and adjust their effort accordingly is likely to apply to many other species, especially longlived ones, whose allocation of energy during breeding is probably the result of a trade-off between survival and reproduction (e.g. Lessells 1991). For example, regular hunting during unfavourable weather would probably increase the parental cost of reproduction, with potential repercussions on survival and on prospects for further breeding, making it an unviable long-term strategy. Furthermore, members of long-lived species are likely to experience different yearly weather conditions through their lives and to have evolved compensatory mechanisms allowing them to minimize the impact of such conditions. Thus, the behavioural adjustments observed in this study are likely to apply to many other vertebrates, with important repercussions on their resilience to future climate change or on their capability to take advantage of it.

Reproduction The effect of weather conditions on foraging performance of individuals was paralleled by a similar effect on population-level reproduction: mean temperature in the 20 days before egg laying affected subsequent populationlevel hatching success and fledgling production. These results confirm the importance of this stage of the breeding cycle of raptors (Newton 1979). In this period, females depend almost completely on the male for food,

1115

1116

ANIMAL BEHAVIOUR, 66, 6

and only those females that accumulate sufficient body reserves manage to lay eggs. Thus, factors disrupting the increase in male hunting yield at this stage may affect subsequent breeding performance. In agreement with this, low temperatures during the last stage of the prelaying period increased the degree of complete clutch failure during incubation. Such failing females were probably in poor body condition because of low male foraging performance and provisioning rate coupled with increased female thermoregulatory costs. In contrast, higher temperatures in the prelaying period probably increased male hunting yield and female body condition, leading to better breeding performance. Furthermore, some of the major ciprinid fish prey of kites start breeding during the kite prelaying stage, and higher temperatures in this period may have also increased prey availability, with additional positive effects on kite reproduction. A negative effect of inclement weather during the prelaying period on subsequent reproduction has also been found in other raptors (e.g. Newton 1986; Marti 1994; Steenhof et al. 1997), and is likely to apply to other species whose breeding output depends on body reserves accumulated during the preincubation period (review in Drent & Daan 1980; Newton 1998).

Response to Climate Change In agreement with global trends, temperatures are increasing in the Alpine region at both high and low elevations (e.g. Beniston 1994, 1997; Ambrosetti & Barbanti 1997). Predicting species’ responses to such changes is difficult. However, based on my results, black kites are likely to be affected by such changes. Increased temperatures may lead to improved foraging performance, higher nest provisioning rates and increased breeding success. However, some global climate change scenarios also predict a future increase in precipitation (Houghton et al. 1996). Kites seemed capable of compensating for current levels of rainfall, with no populationlevel repercussions on reproduction. Whether increased rainfall will negatively affect black kite populations will probably depend on the magnitude of the increase, its temporal distribution and whether it will coincide with sensitive periods of the breeding cycle (e.g. prelaying). The duration and continuity of rain spells within and between days is likely to be particularly important. Whatever the future predictions are, black kites may be already responding to climate change. In the pre-Alps, they have been laying eggs progressively earlier in the past decade (Sergio 2003). Furthermore, the breeding and wintering ranges of black kites have been gradually expanding northwards in various parts of Europe in recent decades (e.g. Burton 1995; Doumeret 1995; Sara` 1996; Sunyer & Vin ˜ uela 1996). My results highlight the importance of species-specific time windows within which the effects of weather variations may be particularly pronounced. They also confirm the need to pay more attention to behaviourally mediated effects of climate on populations, particularly when individuals make compensatory adjustments that

may enhance their resilience to temporally adverse conditions. Acknowledgments Weather data for this study were kindly provided by Meteosvizzera (Locarno Monti). In particular, I thank V. Ortelli for facilitating the acquisition of meteorological data, V. Donnini Mele for extracting some subsets of data from the archives, and A. Boto and C. Scandolara for help in the field. W. Cresswell, B. Kempenaers, I. Newton, C. M. Perrins and two anonymous referees gave valuable comments on the manuscript. This study was partially funded by Regione Lombardia (Servizio Faunistico) and by the Autonomous Province of Trento (Project Biodiversita`). References Altmann, J. 1974. Observational study of behaviour: sampling methods. Behaviour, 49, 227–267. Ambrosetti, W. & Barbanti, L. 1997. Problems related to physical limnology of the deep southern Alpine lakes. Documenta Istituto Italiano di Idrobiologia, 61, 3–18. Avery, M. I. & Krebs, J. R. 1984. Temperature and foraging success of great tits Parus major hunting for spiders. Ibis, 126, 33–38. Beauchamp, G. 1999. A comparative study of breeding traits in colonial birds. Evolutionary Ecological Research, 1, 251–260. Beniston, M. (Ed.) 1994. Mountain Environments in Changing Climates. London: Routledge. Beniston, M. 1997. Variations of snow depth and duration in the Swiss Alps over the last 50 years: links to changes in the large-scale climatic forcings. Climate Change, 36, 281–300. Buckley, N. 1997. Spatial-concentration effects and the importance of local enhancement in the evolution of colonial breeding in seabirds. American Naturalist, 149, 1091–1112. Burton, J. F. 1995. Birds and Climate Change. London: Christopher Helm. Crawley, M. J. 1993. GLIM for Ecologists. Oxford: Blackwell Scientific. Daan, S. 1982. Timing of vole hunting in aerial predators. Mammal Review, 12, 169–181. Dawson, R. D. & Bortolotti, G. R. 2000. Reproductive success of American kestrels: the role of prey abundance and weather. Condor, 102, 814–822. Doumeret, A. 1995. Milan noir Milvus migrans. In: Nouvel Atlas des Oiseaux Nicheurs de France (1985–1989) (Ed. by D. YeatmanBerthelot & G. Jarry), pp. 160–163. Paris: Socie´te´ Ornithologique de France. Drent, R. H. & Daan, S. 1980. The prudent parent: energetic adjustments in avian breeding. Ardea, 68, 225–252. Dunn, E. K. 1973. Changes in fishing ability of terns associated with wind speed and sea surface conditions. Nature, 244, 520–521. Elkins, N. 1988. Weather and Bird Behaviour. Calton: T. & A. D. Poyser. Flemming, S. P. & Smith, P. C. 1990. Environmental influences on osprey foraging in northeastern Nova Scotia. Journal of Raptor Research, 24, 64–67. Forero, M. G., Dona´zar, J. A., Blas, J. & Hiraldo, F. 1999. Causes and consequences of territory change and breeding dispersal distance in the black kite. Ecology, 80, 1298–1310. Grant, P. R., Grant, B. R., Keller, L. F. & Petren, K. 2000. Effects of El Nin ˜ o events on Darwin’s finch productivity. Ecology, 81, 2442– 2457.

SERGIO: WEATHER-BASED FORAGING AND REPRODUCTION

Grubb, T. C. 1975. Weather-dependent foraging behavior of some birds wintering in a deciduous woodland. Condor, 77, 175–182. Grubb, T. C. 1977. Weather-dependent foraging in ospreys. Auk, 94, 146–149. Henty, C. J. 1977. Thermal soaring of raptors. British Birds, 70, 471–475. Hiraldo, F., Veiga, J. P. & Man ˜ ez, M. 1990. Growth of nestling black kites Milvus migrans: effects of hatching order, weather and season. Journal of Zoology, 222, 197–214. Houghton, J. T., Meira Filho, L. G., Callander, B. A., Harris, N., Kattenberg, A. & Mashell, K. (Eds) 1996. Climate Change 1995: The Science of Climate Change. Cambridge: Cambridge University Press. Hughes, L. 2000. Biological consequences of global warming: is the signal already apparent? Trends in Ecology and Evolution, 15, 56–61. Lessells, C. M. 1991. The evolution of life histories. In: Behavioural Ecology: An Evolutionary Approach (Ed. by J. R. Krebs & N. B. Davies), pp. 32–68. Oxford: Blackwell. McCullagh, P. & Nelder, J. A. 1983. Generalised Linear Modelling. London: Chapman & Hall. Machmer, M. M. & Ydenberg, R. C. 1990. Weather and osprey foraging energetics. Canadian Journal of Zoology, 68, 40–43. Marti, C. D. 1994. Barn owl reproduction: patterns and variation near the limit of the species distribution. Condor, 96, 468–484. Martin, T. E. 2000. Abiotic versus biotic influences on habitat selection of coexisting species: climate change impacts? Ecology, 82, 175–188. Matthews, W. J. 1998. Patterns in Freshwater Fish Ecology. New York: Chapman & Hall. Meijer, T., Masman, D. & Daan, S. 1989. Energetics of reproduction in female kestrels. Auk, 106, 549–559. Moss, S. 1998. Predictions of the effects of global climate change on Britain’s birds. British Birds, 91, 307–325. Newton, I. 1978. Feeding and development of sparrowhawk Accipiter nisus nestlings. Journal of Zoology, 184, 465–487. Newton, I. 1979. Population Ecology of Raptors. Berkhamsted: T. & A. D. Poyser. Newton, I. 1986. Population regulation in sparrowhawks. Journal of Animal Ecology, 55, 463–480. Newton, I. 1998. Population Limitation in Birds. London: Academic Press. Pennycuick, C. J. 1989. Bird Flight Performance: A Practical Calculation Manual. Oxford: Oxford University Press. Pinna, M. 1978. L’Atmosfera e il Clima. Torino: UTET: Unione Tipografico-Editrice Torinese. Post, E., Peterson, R. O., Stenseth, N. C. & McLaren, B. E. 1999. Ecosystem consequences of wolf behavioural response to climate. Nature, 401, 905–907. Redpath, S. M., Arroyo, B. E., Etheridge, B., Leckie, F., Bouwman, K. & Thirgood, S. J. 2002. Temperature and hen harrier productivity: from local mechanisms to geographical patterns. Ecography, 25, 535–540. Rice, W. R. 1989. Analyzing tables of statistical tests. Evolution, 43, 223–225. Rotenberry, J. T. & Wiens, J. A. 1991. Weather and reproductive variation in shrubsteppe sparrows: a hierarchical analysis. Ecology, 72, 1325–1335.

Sara`, M. 1996. Wintering raptors in the central Mediterranean basin. In: Biologi´a y Conservacio´n de las Rapaces Mediterra´neas (Ed. by J. Muntaner & J. Mayol), pp. 345–359. Madrid: Sociedad Espan ˜ ola de Ornitologi´a. Sæther, B. E., Tufto, J., Engen, S., Jerstad, K., Røstad, O. W. & Ska˚tan, J. E. 2000. Population dynamical consequences of climate change for a small temperate songbird. Science, 287, 854–856. Sergio, F. 2003. Relationship between laying dates of black kites Milvus migrans and spring temperatures in Italy: rapid response to climate change? Journal of Avian Biology, 34, 144–149. Sergio, F. & Boto, A. 1999. Nest dispersion, diet, and breeding success of black kites (Milvus migrans) in the Italian pre-Alps. Journal of Raptor Research, 33, 207–217. Sergio, F. & Newton, I. 2003. Occupancy as a measure of territory quality. Journal of Animal Ecology, 72, 857–865. Sergio, F., Marchesi, L. & Pedrini, P. 2003a. Spatial refugia and the coexistence of a diurnal raptor with its intraguild owl predator. Journal of Animal Ecology, 72, 232–245. Sergio, F., Pedrini, P. & Marchesi, L. 2003b. Reconciling the dichotomy between single species and ecosystem conservation: black kites (Milvus migrans) and eutrophication in pre-Alpine lakes. Biological Conservation, 110, 101–111. Sergio, F., Pedrini, P. & Marchesi, L. 2003c. Spatio-temporal shifts in gradients of habitat quality for an opportunistic avian predator. Ecography, 26, 243–255. Sergio, F., Pedrini, P. & Marchesi, L. 2003d. Adaptive selection of foraging and nesting habitat by black kites (Milvus migrans) and its implications for conservation: a multi-scale approach. Biological Conservation, 112, 351–362. Sokal, R. R. & Rohlf, F. J. 1981. Biometry. New York: W. H. Freeman. Spaar, R. 1997. Flight strategies of migrating raptors: a comparative study of interspecific variation in flight characteristics. Ibis, 139, 523–535. Steenhof, K., Kochert, M. N. & McDonald, T. L. 1997. Interactive effects of prey and weather on golden eagle reproduction. Journal of Animal Ecology, 66, 350–362. Stephens, D. W. & Krebs, J. R. 1986. Foraging Theory. Princeton, New Jersey: Princeton University Press. Sunyer, C. & Vin ˜ uela, J. 1996. Invernada de rapaces (O. Falconiformes) en Espan ˜ a peninsular e islas baleares. In: Biologi´a y Conservacio´n de las Rapaces Mediterra´neas (Ed. by J. Muntaner & J. Mayol), pp. 361–370. Madrid: Sociedad Espan ˜ ola de Ornitologi´a. Thompson, P. M. & Ollason, J. C. 2001. Lagged effects of ocean climate change on fulmar population dynamics. Nature, 413, 417–420. Thomson, D. L., Baillie, S. R. & Peach, W. J. 1997. The demography and age-specific annual survival of song thrushes during periods of population stability and decline. Journal of Animal Ecology, 66, 414–424. Veiga, J. P. & Hiraldo, F. 1990. Food habits and the survival and growth of nestlings in two sympatric kites (Milvus milvus and Milvus migrans). Holarctic Ecology, 13, 62–71. Vin ˜ uela, J. 2000. Opposing selective pressures on hatching asynchrony: egg viability, brood reduction, and nestling growth. Behavioral Ecology and Sociobiology, 48, 333–343. Vin ˜ uela, J. & Sunyer, C. 1994. Black kite Milvus migrans. In: Birds in Europe: Their Conservation Status (Ed. by G. M. Tucker & M. F. Heath), pp. 148–149. Cambridge: BirdLife.

1117