ECOGRAPHY 25: 705–713, 2002

Different responses to cold weather in two pied flycatcher populations T. Eeva, E. Lehikoinen, M. Ro¨nka¨, V. Lummaa and D. Currie

Eeva, T., Lehikoinen, E., Ro¨nka¨, M., Lummaa, V. and Currie, D. 2002. Different responses to cold weather in two pied flycatcher populations. – Ecography 25: 705–713. We compared how breeding parameters differ according to prevailing weather conditions between a marginal, subarctic (69°N) and temperate (61°N) population of the pied flycatcher Ficedula hypoleuca, a small migratory insectivorous passerine. We predicted that the effects of weather on breeding performance (clutch size, hatching success, nestling growth, fledging success) would be greater at northern latitudes, where the weather conditions are more extreme and unpredictable. We found that the breeding parameters, except clutch size, were not, however, inferior in the north. Northern birds, unlike the southern ones, responded to colder conditions by laying smaller clutches and maintaining a larger energy reserve (indicated by higher female body mass and higher levels of subcutaneous fat). If a cold spell occurred during the nestling period, southern flycatchers had 5 – 10% lower fledging success than the northern ones. Our results indicate that in the north, the breeding individuals coped with cold and variable weather better than the individuals in the southern population. This could be adaptive, because at high latitudes there is a higher probability of cold weather at the time of breeding. T. Ee6a ([email protected]), M. Ro¨nka¨ and E. Lehikoinen, Dept of Biology, Uni6. of Turku, FIN-20014 Uni6. of Turku, Finland. – V. Lummaa, Dept of Zoology, Uni6. of Cambridge, Downing St., Cambridge, U.K. CB2 3EJ. – D. Currie, Birdlife Seychelles, P.O. Box 1310, Victoria, Mahe, Republic of Seychelles.

Food and weather conditions vary geographically. For example, breeding seasons of marginal populations at subarctic latitudes are shorter and colder (Ja¨rvinen 1983) and, consequently, the availability of invertebrate food at breeding time for insectivorous birds may be worse than at more southern temperate latitudes. Due to varying local food and weather conditions, breeding strategies of birds – especially migratory insectivorous birds – may differ geographically between their northern and southern populations (Sanz 1997, 1998, Young 1996). Phenotypic plasticity in reproductive traits observed in many bird populations can be viewed as a mechanism by which individuals attempt to time their reproduction and adjust their clutch size to maximize their breeding success (Perrins 1991, Kawecki and Stearns 1993). There are, however, only few studies that have

attempted to analyse how successful birds are in optimizing their breeding strategies in local populations occupying climatically different areas (Ja¨rvinen 1989, 1993, Sanz 1995, 1998, Young 1996). The pied flycatcher Ficedula hypoleuca Pallas is a small insectivorous migrant species that breeds from temperate through to subarctic conditions (Cramp and Perrins 1993). In this paper we compare the breeding strategies of the pied flycatcher between two areas, which differ markedly in their environmental conditions: one population in a southern temperate habitat (61°N) versus the other in the subarctic zone (69°N). Rather than attempting to compare breeding parameters between the two populations, we compare the breeding responses of the two populations to the markedly different prevailing environmental conditions (temperature and rainfall). In particular, we look for

Accepted 14 February 2002 Copyright © ECOGRAPHY 2002 ISSN 0906-7590 ECOGRAPHY 25:6 (2002)

705

evidence of local adaptation by individuals breeding at northern, typically colder and more variable, weather conditions. If birds were not able to optimize their timing of breeding and clutch size according to local conditions, we would expect the nestling growth and survival to be worse in the north. If, however, birds in the north were locally adapted to the more harsh weather, we expect to find no differences in growth or survival between the two populations.

Materials and methods Study sites The study area in the subarctic, Kevo, was established in 1982 –1986, when ca 1200 nest-boxes were put up mainly in the Utsjoki valley, N Finland (Eeva et al. 1989). Nest-boxes were distributed over a large area (ca 90 km in north-south direction), usually in groups of four, with 500 m distance between the groups. Most nest-boxes were situated near the Kevo Subarctic Research Station (69°45%N, 27°01%E). The study area lies by the northern forest limit of Scots pine Pinus syl6estris, which forms mixed stands with the mountain birch Betula pubescens spp. czerepanowii. Ficedula hypoleuca nests were predominantly (81%) located in pine dominated or mixed woods with the mean altitude of 114 m a.s.l. Our southern study area was established in 1991 in Harjavalta, SW Finland (61°20%N, 22°10%E). Eleven study sites, with 30 –50 nest-boxes in each, were selected so that their habitat characteristics were as similar as possible (Eeva et al. 1997). Four new sites were added in 1992 and 1994 and the number of nest-boxes varied from 540 to 845 according to year. The distance between the nest-boxes was ca 40 m within each site, and the altitude of the sites was ca 20 – 30 m a.s.l. The forests in the area are dominated by Scots pine, which forms mixed stands with Norwegian spruce Picea abies and birch Betula spp. The nest-box sites were originally established to study the effects of aerial emissions from a copper smelter on birds (Eeva et al. 1997). To ensure that pollution effect did not bias the geographic comparison, we omitted the sites closer than 3 km from the polluting factory complex, as we had not observed any detrimental effect beyond this distance (see Eeva and Lehikoinen 1996). The breeding success in the area is similar to those of other southern Finnish populations (Lundberg and Alatalo 1992, Sanz 1997). The nest-boxes were checked weekly to gather breeding data including laying date, clutch size, hatching date, numbers of nestlings and fledglings and nestling mass (for the years 1982 –1994 in Kevo; 1991 –1998 in Harjavalta). Females were trapped at the nest during incubation and their wing length (mm) and body mass (g) were measured. Subcutaneous fat was scored follow706

ing Busse and Kania (1970). Nests where incubation did not start and replacement nests were omitted from all analyses in order to remove the variation that might be caused by other factors than weather conditions. Nests that failed because of predation or human disturbance were also omitted since we expected these failures not to be related to weather conditions. Only a small number (Harjavalta 7.7%; Kevo 5.4%) of females bred more than once in our study areas and B 1% of females used the same territory in two years. This ensures that pseudoreplication does not cause a significant bias in our analyses. The breeding density of F. hypoleuca was much lower in Kevo (X( = 4.19 0.6 pairs km − 2, n =13 yr; in average 8% of the nest-boxes were occupied) than in Harjavalta (X( = 72.095.9 pairs km − 2, n= 8 yr; in average 37% of the nest-boxes were occupied). A recent study has shown that density does not have any adverse effect on reproductive output of F. hypoleuca (Both 2000).

Weather The weather data for the northern area were obtained from Utsjoki Kevo Meteorological Station (69°45%N, 27°01%E) and for the southern area from Peipohja Meteorological station (61°16%N, 22°15%E). We calculated for each nest the mean daily temperatures and mean daily precipitation sums for the pre-laying period (5 d), laying period (variable according clutch size), incubation period (15 d) and nestling period (15 d). The average temperatures and precipitation sums during the four reproductive phases are presented for both areas in Table 1.

Statistical analyses The effects of weather on laying date, clutch size, nestling body mass and female condition were analysed by using homogeneity of slopes models of GLM procedure in SAS, with type III SS (Anon. 1989). The basic model consisted of the main effects of the study area as a class variable, the weather variables as continuous variables and their interactions. The normality of residuals was tested with Shapiro-Wilk’s test (in UNIVARIATE procedure of SAS). In the analyses of clutch size, we first corrected the values for the seasonal decline of the clutch size that is typical of this species (Cramp and Perrins 1993). This was done by calculating yearly the residuals of the clutch size from the regression between the clutch size and laying date. The residuals were then added to the yearly mean to retain the yearly variation in our data. Since there was a general tendency of females to gain weight during the incubation period we added the stage of incubation (days before hatching) as a further covariate in the models. ECOGRAPHY 25:6 (2002)

The effect of weather on nestling body mass was analysed by using ANCOVA models. The nestling mass data were collected in years 1987 –1994 at Kevo and 1991–1998 at Harjavalta. A logistic growth curve was first fitted to the combined data of the two areas: M(x)=A/{1+[(A−I)/I]× exp(−K×x)}, where M(x) =mass at age x, A = asymptotic mass, I = initial mass, K=rate constant (see Ricklefs 1983). Fixed initial mass of 1.8 g was used in the curve fit. Thereafter we calculated the proportional (%) deviation from the best fit growth curve (A =14.3, K =0.44, R2 = 0.86) for each brood (within the age range of 5 –14 d). Mean brood mass residuals were then used as a dependent variable in the ANCOVA models. Study area was used as an independent class variable. Mean daily temperature and mean daily rainfall during the nestling period were used as covariates. The effects of weather on hatching and fledging success were analysed using generalized linear models (GENMOD procedure of SAS, type 3 analysis). As binomial response variables we used the proportions: hatchlings/clutch size (hatching success) and fledglings/ hatchlings (fledgling success). The explanatory variables were the same as in the models above. In these models binomial probability distribution and logit link function were used.

Results Weather The mean temperatures during the F. hypoleuca breeding period were 2.2 –4.7°C lower in Kevo than in Harjavalta (Table 1). The probability of a daily temperature minimum B5°C during the nestling period was 80% in Kevo and 45% in Harjavalta. The variance of mean temperatures was about the same in both areas for pre-laying, laying and nestling periods, but was higher in the southern area during the incubation period (Table 1). However, temperature differences

among the laying, incubation and nestling phases were greater in the northern area, reflecting the faster advance of spring in the north. Similarly, daily variation within each period was, on average, higher in the north (ANOVA for variances: p B 0.0001 for each period). The among-years variation in mean temperature of June was similar in the two areas (Levene’s test, F1,19 = 0.77, p =0.39). The predictability of temperature was analysed using Spearman rank correlations among the weekly mean temperatures. Correlations with future temperatures were calculated for each week of the main breeding period (9 weeks: 17 May – 18 July) over 8 yr (Kevo: 1987–1994; Harjavalta: 1991 –1998) using time-lags of 1– 5 weeks. Time-lag 1 denotes the correlation with next week’s temperature, time-lag 2 denotes the correlation with the following week and so on. The mean correlation coefficients between pairs of weeks through the whole nine-week breeding season were then compared between the two study areas. Short-term temperature correlation coefficients (time-lag 1 week) were positive and significantly higher in south than in north (Fig. 1). Longer-term correlation coefficients tended to be close to zero or negative in both areas (Fig. 1). Therefore, although temperature predictability was slightly better in the southern area, this only applied over a short period of time, i.e. until the following week. Birds started to breed earlier in the north in relation to the phenology of climate and vegetation: the average thermal sum (TS, degree days [dd], base 5°C) of laying date was 44 dd in Kevo (n = 13 yr) and 157 dd in Harjavalta (n = 8 yr). Ficedula hypoleuca started to lay at or slightly before birch leafing in Kevo and always after birch leafing in Harjavalta. In Kevo, the mean temperature rose significantly faster between pre-laying and nestling periods than in Harjavalta. There was significantly more rain in Harjavalta than in Kevo (Table 1). The amount of rainfall varied more in Harjavalta than in Kevo during the pre-laying, laying and

Table 1. The mean daily temperature (°C) and precipitation sum (mm) during the pre-laying, laying, incubation and nestling periods of F. hypoleuca. SD=standard deviation. A one-way ANOVA was used for the site means and Levene’s test for equality of variances, respectively. Harjavalta

Temperature Pre-laying Laying Incubation Nestlings Rainfall Pre-laying Laying Incubation Nestlings

n

x1

1094 1094 1094 1076

10.7 13.5 14.0 14.2

1094 1094 1094 1076

7.3 9.8 35.7 38.4

ECOGRAPHY 25:6 (2002)

Kevo SD 3.48 3.58 2.64 1.96 9.3 11.3 19.5 25.1

H0: s21 =s22

H0: x1 =x2

n

x2

516 516 516 529

6.0 9.8 10.9 12.0

516 516 516 529

4.8 6.9 18.4 33.0

SD 3.27 3.58 1.92 2.03 6.2 7.4 14.9 29.6

F

p

F

p

672.8 378.5 559.3 450.6

0.0001 0.0001 0.0001 0.0001

2.92 0.00 128.2 1.19

0.088 0.99 0.0001 0.28

30.6 69.1 419.4 63.0

0.0001 0.0001 0.0001 0.0001

29.6 50.7 37.1 24.3

0.0001 0.0001 0.0001 0.0001

707

Fig. 1. The predictability of temperature in the two study areas during the main breeding period of F. hypoleuca (9 weeks: 17 May –18 July). The Spearman rank correlation coefficients (r) were calculated among mean weekly temperatures during 8 yr (Kevo: 1987 –l994; Harjavalta: 1991 – 1998). The means were compared with one-way ANOVA between the two study areas (* = pB0.05; n.s. =not significant). The horizontal line denotes significance of correlation coefficients at a-level of 95%.

incubation periods, whereas the opposite was true during the nestling period (Table 1).

Breeding parameters Laying date The mean laying date was a week later (Table 2) and more synchronous in Kevo than in Harjavalta (s2Kevo = 27.2, s2Harjavalta =44.6; Levene’s test, F1,1647 =38.9, p B 0.0001). In both areas the mean laying date for the latest clutches was about the same, i.e. 17 June. On average, the difference between the earliest and latest laying dates was 17 d in Kevo (n =13 yr) and 29 d in Harjavalta (n =8 yr). This probably means that the ‘‘time window’’ to start laying was narrower by nearly two weeks in north, i.e. birds breeding in Kevo were

Fig. 2. The relationship between the annual mean temperature during the 5-d period before laying and annual mean laying date in Harjavalta ( ) and in Kevo (). Numbers denote years.

more constrained in their timing of breeding. Yearly variation in mean laying dates did not, however, differ between the areas (Levene’s test, F1,19 = 0.015, p = 0.90). There was a strong negative correlation between temperature before laying and laying date in both areas (Kevo: r = − 0.59, n = 13, p B 0.05; Harjavalta: r = − 0.73, n = 8, pB 0.05; Fig. 2). This means that a bird’s temperature threshold to start laying is lower in late than in early seasons. Clutch size Clutch size was, on average, 0.34 eggs larger in Harjavalta than in Kevo (Table 2.). Yearly variation in the mean clutch size was similar in the two areas (Levene’s test, F1,19 = 1.04, p =0.32). Clutch size (corrected for the effect of the advancing season) increased with laying-time temperature in Kevo but not in Harjavalta (Table 3). During warm periods, large clutches were laid in both areas, but when the average temperature

Table 2. The breeding parameters of F. hypoleuca in Kevo (1982–l994) and in Harjavalta (1991–1998). Only those nests with incubation started were included. Predated and otherwise destroyed nests were omitted, as well as replacement nests. An ANOVA was used for the site means and generalized linear models for the proportional measures. Harjavalta

1

Laying date Clutch size Hatchling number2 Fedgling number

Kevo

n

x1

SE

n

x2

SE

F

P

1099 1094 1084 1033

151 6.16 5.68 4.81

0.20 0.025 0.040 0.056

550 516 389 541

158 5.82 5.11 4.52

0.23 0.039 0.080 0.086

537 59.0 49.4 22.0

0.0001 0.0001 0.0001 0.0001

Kevo

Harjavalta n Hatching success (%) Fledging success (%) 1

Day 1=1 January.

708

2

2

H0: x2 =x2

1084 1014

x1 92.1 85.1

SE

n

0.52 0.80

385 375

H0: x1 =x2 x2 88.2 85.2

SE

x2

1.12 1.59

22.3 0.13

p 0.0001 0.72

Hatchling number was recorded in Kevo only for years 1986–1994. ECOGRAPHY 25:6 (2002)

Table 3. The homogeneity of slopes models for the effects of temperature and rainfall on breeding parameters of F. hypoleuca. The laying-date effect was removed from the analyse of clutch size. Effect

Area Area×temperature Area×rainfall Temperature Rainfall

Clutch size

Hatching success

Nestling growth

Fledging success

F5,1604

p

x

p

F5,671

p

x2

p

31.5 18.5 0.00 7.23 0.09

0.0001 0.0001 0.97 0.0073 0.76

0.11 0.036 0.24 45.3 13.5

0.74 0.84 0.62 0.0001 0.0002

0.77 0.42 1.19 7.36 0.05

0.38 0.52 0.28 0.0069 0.83

13.5 3.99 11.1 21.9 69.1

0.0002 0.046 0.0009 0.0001 0.0001

2

during the laying period was low, northern birds laid smaller clutches than southern ones (Fig. 3a). Rainfall during the laying period had no effect on clutch size (Table 3). To examine further why birds laid fewer eggs during cold weather at northern latitudes, we checked whether small clutches laid in cold weather were actually late nests. Nests were divided into those laid in cold temperature ( 510°C) and those laid in warm weather ( \ 10°C). There was no difference in laying-dates between the groups (ANOVA, F1,548 =2.33, p =0.13). Similarly, nests were classified as phenologically early (TS at layingBmean TS) and late nests (TS \mean TS). Both groups showed a similar positive relationship between clutch size and temperature (indicated by non- significant

interaction term in ANCOVA: F1,512 = 2.08, p = 0.15). Thus, there was a proximate effect of temperature on clutch size, independent on calendar day or phenology. Hatching success The number of hatchlings was 0.57 chicks larger in Harjavalta than in Kevo which can be explained by larger clutch size and slightly better (3.9%-units) hatching success in the south (Table 2). Yearly variation in mean number of hatchlings was similar in the two sites (Levene’s test, F1,15 = 0.74, p = 0.40). Hatching success was positively related to temperature and negatively related to rainfall during the incubation period (Table 3). Relationships between hatching success and weather

Fig. 3. The effect of temperature on four breeding parameters of F. hypoleuca showing the predicted values and their 95% confidence limits from the homogeneity of slopes models ( — Harjavalta; ----Kevo). For clarity, the rainfall effect was omitted from these models (cf. Table 3). The mean temperature for the analyses were calculated to correspond to each breeding phase: (a) laying period for clutch size; (b) incubation period for hatching success; (c) nestling period for chick body mass and (d) fledging success. The clutch size values are corrected for laying date effect (see Materials and methods). ECOGRAPHY 25:6 (2002)

709

Table 4. Female body mass (g), wing length (mm) and fat score in the two study areas. An ANCOVA was used for estimating the effect of mean temperature of the incubation period on body mass. Stage of incubation (days before hatching) was used as a covariate. Harjavalta

Body mass Wing length Fat score

Kevo

n

x1

SE

n

x2

SE

F

p

648 649 595

14.9 78.4 1.33

0.03 0.06 0.04

182 185 177

15.8 78.1 1.81

0.08 0.11 0.06

130.8 4.05 38.5

0.0001 0.045 0.0001

F

p

20.4 9.29 0.01 8.92 6.78

0.0001 0.0024 0.91 0.0029 0.0094

Body mass1 Effect

DF

Area Area × temperature Area × stage Temperature Stage of incubation 1

Error MS=0.76, DF =824.

1 1 1 1 1 2

Fat score2

F

p

15.6 2.21 3.86 1.49 1.01

0.0001 0.14 0.049 0.22 0.32

DF 1 1 1 1 1

Error MS= 0.75, DF =766.

were similar in the two areas, as indicated by non-significant interactions (Table 3, Fig. 3b). Chick weight Nestlings grew equally well in Kevo and in Harjavalta (Table 3): the growth coefficient of the logistic growth curve was 0.44 for both areas. Nestling body mass was positively affected by temperature in the nestling period but there was no effect of rainfall (Table 3). The temperature effect was similar in both study areas (a non-significant interaction, Table 3, Fig. 3c), and the nestling body mass was higher by 9.3%-units in the warmest (18°C) than in the coldest (7°C) nestling period in combined data of Harjavalta and Kevo. Fledging success The number of fledglings was 0.29 chicks higher in Harjavalta than in Kevo, but there was no difference in average fledging success between the two areas (Table 2). The yearly variation in fledgling number was slightly higher in the north, but the difference was not significant (Levene’s test, F1,19 =3.19, p =0.090). Fledging success was positively related to temperature and negatively related to rainfall during the nestling period (Table 3). The negative effect of rainfall on fledging success was stronger in Kevo than in Harjavalta (a significant interaction, Table 3), but the negative effect of low temperature on fledging success was stronger in Harjavalta (Table 3). Fledging success was high in both areas in warm temperature, but when the average temperature during the nestling period was low (ca B13°C) southern birds suffered higher nestling mortality than northern ones (Fig. 3d), which could indicate a better adaptation of northern birds and/or their food to low temperatures. Female body mass and fat score During incubation females were 0.9 g heavier in Kevo 710

H0: x1 =x2

than in Harjavalta, although they were somewhat smaller based on wing length difference (Table 4). The higher mass of northern birds was due to higher subcutaneous fat reserves (Table 4). Female body mass was not affected by temperature during the incubation period independent of a change in individual fat scores. Fat reserves decreased in southern birds in cold weather but remained the same in northern birds in all temperatures (Table 4). Female body mass correlated with clutch size in both areas (Kevo: r = 0.31, n = 182, p = 0.0001; Harjavalta: r =0.14, n =649, p = 0.0002). Although females seemed to be energetically in better condition in Kevo they laid smaller clutches than in Harjavalta. In Kevo, the clutches laid in cold weather were smaller than those laid in warm weather (see above). To test whether this was due to energy-limitation, we compared body masses of females which had laid in cold weather ( 510°C) with those which had laid in warm weather ( \ 10°C). There was no difference in female body mass between the groups (ANOVA: F1,180 = 0.50, p = 0.48), indicating that females laying in cold weather were not in worse condition than females which laid in warm weather.

Discussion In general, cold weather had negative effect on all of the breeding parameters studied (clutch size, hatching success, nestling growth and fledging success) both in the northern (Kevo) and the southern (Harjavalta) populations. In addition, rainfall also had a negative effect on hatching and fledging success but no detrimental effect was observed on nestling growth. Contrary to expectations, the nestling survival in the north was not noticeably inferior to that in south. Clutch size ECOGRAPHY 25:6 (2002)

was smaller in Kevo, and hatching success was also marginally lower than in Harjavalta. However, despite lower average temperatures and higher probability of cold spells in the northern latitudes, nestling growth and fledging success were the same in both areas. Virolainen (1984) found a strong negative correlation between yearly nestling mortality and a mean temperature for a 10-d period after hatching in southern Finland (60°, n=13 yr) whereas Ja¨ rvinen (1983) did not find this kind of correlation in northern Finland (69°, n= 16 yr). Our results support the idea that flycatchers breeding at northern latitudes cope better with cold temperatures than flycatchers in the south. Northern females responded to low temperatures by laying fewer eggs (see also Ja¨ rvinen 1989) and suffered lower nestling mortality during cold periods than southern females. Furthermore, northern females maintained larger energy reserves (subcutaneous fat) during incubation period. Consequently, they were successful in later phases of breeding, as shown by good nestling growth and fledging success in the north. Southern birds did not respond to cold weather probably because in most years there was no need to do so. However, if a cold spell occurred during the nestling period they paid a price for being unresponsive, and in cold weather (mean temperature in nestling period was below 13°C) the fledging success was 5 –10% lower in the southern than in the northern population. Why do northern birds seem to respond differently to cold weather? One possibility is that smaller clutch size (and egg size: see Ja¨ rvinen and Va¨ isa¨ nen, 1984) is a proximate response to the difference in food availability between the two areas in the beginning of breeding. We found that females which started to lay eggs in cold weather were heavier and had larger fat stores in Kevo than in Harjavalta. Several studies have shown that larger energy reserves may actually indicate poor feeding conditions. For example, Carrascal et al. (1998) compared two great tit Parus major populations in localities with different ambient temperature, and found that the body condition was better at the locality with more severe environmental conditions. Similarly, Romero et al. (1998) found that redpolls Carduelis flammea breeding at a high Arctic site had larger fat reserves and body mass than birds in a low Arctic site. Measurements of caterpillar abundance in our two study areas suggest that the availability of caterpillars is lower in Kevo than in Harjavalta during the laying period of F. hypoleuca (Veistola et al. 1995, Eeva et al. 1997). Birds in Harjavalta may less often face problems with food availability since they breed phenologically later than birds at Kevo. An explanation to higher energy reserves might be that northern birds have a strategy of saving energy for future use, i.e. they respond to a cold climate by laying fewer eggs and maintaining larger fat reserves. An analogical explanation is presented for wintering tits ECOGRAPHY 25:6 (2002)

(Parus spp.), which are shown to reduce the threat of starvation by maintaining higher energy reserves in unpredictable environmental conditions (Lehikoinen 1986, Hurly 1992, Bednekoff and Krebs 1995, Gosler 1996). As indicated by our data, northern birds may also have limited possibilities of fitting their laying-time according to ambient weather (see also Slagsvold 1976), which may favour females to maintain higher energy reserves at the beginning of breeding. Alternatively, birds might not maintain a higher energy reserve against a bad spell in future but it would be a consequence of a proximate effect of weather and food conditions. Such a situation would be possible if invertebrates in the north are better adapted against cold weather than invertebrates in south. If invertebrates in the north are well adapted to cold weather, their mobility and availability to birds may remain better in cold weather. Better food conditions would thus result in better body condition and relatively good breeding success in the north even during the cold spells. This hypothesis does not, however, explain why clutch size of, F. hypoleuca decreases in the north if weather is cold. High energy reserves are required in the north for successful incubation of eggs (Ja¨ rvinen and Va¨ isa¨ nen, 1983, 1984) and could explain the relatively good hatching success in Kevo, despite the lower average temperatures. Ja¨ rvinen and Va¨ isa¨ nen (1984) found that during cold spells lean females interrupted incubation easier than fatter ones. Silverin and Wingfield (1998) compared two Swedish F. hypoleuca populations breeding in temperate and subarctic conditions. They found that northern birds were locally adapted to cold weather by having higher stress hormone (corticosterone) levels and lowered stress hormone response, i.e. the threshold for a northern female flycatcher to abandon her nest in cold weather was higher than for their southern counterparts. Ficedula hypoleuca females with high corticosterone levels can increase their body mass and enhance their own survival in harsh conditions (Silverin 1990). A further study is needed to find out whether hormonal differences would explain the higher body mass of F. hypoleuca females in Kevo than in Harjavalta. Clutch size of F. hypoleuca decreases with increasing altitude by 0.08 eggs 100 m − 1 (Lundberg and Alatalo 1992, Sanz 1997) or even slightly more (Ja¨ rvinen 1993). The difference in altitude between Harjavalta and Kevo is, however, only 90 m, and cannot explain the difference in clutch size between the two areas. Clutch size also decreases during the breeding season, in Kevo the average decrease was 0.042 eggs d − 1. The difference in timing might therefore explain some but not all of the difference between our study sites, since there was a 7-d difference in mean laying date between Harjavalta and Kevo. The rest of the difference might 711

be caused by the food availability difference for the laying females in the beginning of breeding. The ultimate explanation for the smaller clutch size at northern latitudes is most likely the time constraint associated with a short breeding season (e.g. Siikama¨ ki et al. 1994). For both study areas, the weather models predicted stronger detrimental effects on fledging success when both rain and low temperature occurred simultaneously. Moreover, there was an interesting difference between the two areas as to how the two weather variables occurred in concert. In the north, rainfall varied independently on temperature (no correlation: r= −0.059, p = 0.19, n =497 nestling periods), while in the south heavy rainfall took place most often in warm weather (a positive correlation: r = 0.14, p B 0.0001, n =1073 nestling periods). We suppose that this explains why the negative effect of rain on breeding was stronger in the north, despite that rainfall during the breeding was greater in the south. In general, temperature is known to be more variable in the north than in the south (Skre 1971). The meteorological data showed, however, that mean temperatures did not vary more in Kevo than in Harjavalta during the breeding season. Although the yearly amplitude in temperature is clearly larger in the north, its variation is at its smallest in mid-summer (Myers and Pitelka 1979). Furthermore, Skre (1971) stresses the importance of coastal air masses on the temperature variation, and states that the effect of Atlantic air masses lowers the standard deviation in coastal areas. This indicates that the proximity of sea may have a greater importance on the variability of the weather than the latitude. The fact that Kevo area is not far (ca 100 km) from the Arctic Ocean probably diminishes the variation to the level of much more southern areas like Harjavalta. This observation is in accordance with the fact that breeding parameters of F. hypoleuca did not vary markedly more in the north than in the south (but see Ja¨ rvinen 1989). In conclusion, although the patterns found in our study are purely correlational, they show that F. hypoleuca does, on average, rather well in its northern breeding grounds. There was a lack of predictability in future weather conditions at time of breeding in the northern (69° N) and the southern (61° N) population, but our results indicate high plasticity by individuals breeding in north. Northern birds had some means to cope with low temperatures and they seemed to be able to breed ‘‘optimally’’ in an average season in Kevo. Acknowledgements – We thank Simo Veistola, Andy F. Russell, Joe¨ l Beˆ ty and Erkki Haukioja for valuable comments on the manuscript. Lasse Iso-Iivari did the fieldwork in 1983 – 1986 and dozens of other people participated in data collection in both study areas during the years. This study was financially supported by Emil Aaltonen foundation, Turku Univ. foundation, Academy of Finland (project number 50332), KONE foundation and Section of Ecology at Turku Univ.

712

References Anon. 1989. SAS/STAT user’s guide, 4th ed. Vols 1 – 2. – SAS Inst., Cary, NC, USA. Bednekoff, P. A. and Krebs, J. R. 1995. Great tit fat reserves: effects of changing and unpredictable feeding day length. – Funct. Ecol. 9: 457 – 462. Both, C. 2000. Density dependence of avian clutch size in resident and migrant species: is there a constraint on the predictability of competitor density? – J. Avian Biol. 31: 412 – 417. Busse, P. and Kania, W. 1970. Operation Baltic 1961 – 1967 – working methods. – Acta Ornithol. (Warsaw) 12: 179 – 208. Carrascal, L. M. et al. 1998. Interactions among environmental stress, body condition, nutritional status, and dominance in great tits. – Auk 115: 727 – 738. Cramp, S. and Perrins, C. M. (eds) 1993. The birds of the Western Palearctic. Vol. VII. – Oxford Univ. Press. Eeva, T. and Lehikoinen, E. 1996. Growth and mortality of nestling great tits (Parus major) and pied flycatchers (Ficedula hypoleuca) in a heavy metal pollution gradient. – Oecologia 108: 631 – 639. Eeva, T., Lehikoinen, E. and Veistola, S. 1989. Habitat preferences and their effects on nesting in northern hole-nesting passerines. – Ornis Fenn. 66: 142 – 150. Eeva, T., Lehikoinen, E. and Pohjalainen, T. 1997. Pollutionrelated variation in food supply and breeding success in two hole-nesting passerines. – Ecology 78: 1120 – 1131. Gosler, A. G. 1996. Environmental and social determinants of winter fat storage in the great tit Parus major. – J. Anim. Ecol. 65: 1 – 17. Hurly, T. A. 1992. Energetic reserves of marsh tits (Parus palustris): food and fat storage in response to variable food supply. – Behav. Ecol. 3: 181 – 188. Ja¨ rvinen, A. 1983. Breeding strategies of hole-nesting passerines in northern Lapland. – Ann. Zool. Fenn. 20: 129 – 149. Ja¨ rvinen, A. 1989. Geographical variation in temperature variability and predictability and their implications for the breeding strategy of the pied flycatcher Ficedula hypoleuca. – Oikos 54: 331 – 336. Ja¨ rvinen, A. 1993. Spatial and temporal variation in reproductive traits of adjacent northern pied flycatcher Ficedula hypoleuca populations. – Ornis Scand. 24: 33 – 40. Ja¨ rvinen, A. and Va¨ isanen, R. A. 1983. Egg size and related reproductive traits in a southern passerine Ficedula hypoleuca breeding in an extreme northern environment. – Ornis Scand. 14: 253 – 262. Ja¨ rvinen, A. and Va¨ isa¨ nen, R. A. 1984. Reproduction of pied flycatchers (Ficedula hypoleuca) in good and bad breeding seasons in a northern marginal area. – Auk 101: 439 – 450. Kawecki, T. J. and Stearns, S. C. 1993. The evolution of life histories in spatially heterogeneous environments: optimal reaction norms revisited. – Evol. Ecol. 7: 155 – 174. Lehikoinen, E. 1986. Is fat fit? A field study of survival and fatness in the great tit Parus major L. – Ornis Fenn. 63: 112 – 119. Lundberg, A. and Alatalo, R. V. 1992. The pied flycatcher. – T. and A. D. Poyser. Myers, J. P. and Pitelka, F. A. 1979. Variation in summer temperature patterns near Barrow, Alaska: analysis and ecological interpretation. – Arct. Alp. Res. 11: 131 – 144. Perrins, C. M. 1991. Constraints on the demographic parameters of bird populations. – In: Perrins, C. M., Lebreton, J.-D. and Hirons, G. J. M. (eds), Bird population studies. Oxford Univ. Press, pp. 190 – 206. Ricklefs, R. E. 1983. Avian postnatal development. – In: Farner, D. S., King, J. R. and Parkes, K. C. (eds), Avian biology. Vol. VII. Academic Press, pp. 1 – 83. Romero, L. M., Soma, K. K. and Wingfield, J. C. 1998. The hypothalamus and adrenal regulate modulation of corticosterone release in redpolls (Carduelis flammea) – an arctic-breeding song bird. – Gen. Comp. Endocrinol. 109: 347 – 355. ECOGRAPHY 25:6 (2002)

Sanz, J. J. 1995. Environmental restrictions on reproduction in the pied flycatcher Ficedula hypoleuca. – Ardea 83: 421 – 430. Sanz, J. J. 1997. Geographic variation in breeding parameters of the pied flycatcher Ficedula hypoleuca. – Ibis 139: 107 – 114. Sanz, J. J. 1998. Effects of geographic location and habitat on breeding parameters of great tits. – Auk 115: 1034 – 1051. Siikama¨ ki, P., Hovi, M. and Ra¨ tti, O. 1994. A trade-off between current reproduction and moult in the pied flycatcher – an experiment. – Funct. Ecol. 8: 587 – 593. Silverin, B. 1990. Testosterone and corticosterone and their relation to territorial and parental behavior in the pied flycatcher. – In: Balthazart, J. (ed.), Hormones, brain and behaviour in vertebrates. 2. Behavioural activation in males and females. Social interaction and reproductive endocrinology. Vol. 9. Karger, Basel, pp. 129 –142. Silverin, B. and Wingfield, J. C. 1998. Adrenocortical responses to stress in breeding pied flycatchers Ficedula

ECOGRAPHY 25:6 (2002)

hypoleuca: relation to latitude, sex and mating status. – J. Avian Biol. 29: 228 – 234. Skre, O. 1971. Frequency distributions of monthly air temperature and their geographical and seasonal variations in northern Europe. – Meldinger fra Norges landbrukshøgskole 50: 1 – 51. Slagsvold, T. 1976. Annual and geographical variation in the time of breeding of the great tit Parus major and the pied flycatcher Ficedula hypoleuca in relation to environmental phenology and spring temperature. – Ornis Scand. 7: 127 – 145. Veistola, S., Eeva, T. and Lehikoinen, E. 1995. Seasonal occurrence of arthropods as a source of food for birds in Finnish Lapland. – Entomol. Fenn. 6: 177 – 181. Virolainen, M. 1984. Breeding biology of the pied flycatcher Ficedula hypoleuca in relation to population density. – Ann. Zool. Fenn. 21: 187 – 197. Young, B. E. 1996. An experimental analysis of small clutch size in tropical house wrens. – Ecology 77: 472 – 488.

713