Great tits lay increasingly smaller clutches than selected for: a study of climate- and density-related changes in reproductive traits

Journal of Animal Ecology 2009, 78, 1298 1306 doi: 10.1111/j.1365-2656.2009.01596.x Great tits lay increasingly smaller clutches than selected for: a study of climate- and density-related changes in reproductive traits Markus P. Ahola*, Toni Laaksonen, Tapio Eeva and Esa Lehikoinen Section of Ecology, Department of Biology, University of Turku, FI-20014 Turku, Finland Summary 1. The phenology of temperate environments and therefore timing of breeding has advanced in a number of bird species due to climate warming. Few studies, however, have examined the mechanisms behind the observed changes, the role of natural selection in them or the determinants of the selection. In other traits such as clutch size, even changes over years have been rarely studied. 2. We studied patterns and trends in timing of breeding, clutch size and fledgling production in the great tit Parus major in South-West Finland during 1953 2008, as well as natural selection on the timing and clutch size, based on fledgling production. We also examined connections between these parameters and a number of climatic and population intrinsic factors. 3. Laying date was earlier when the pre-breeding period was warm and tended to be earlier when breeding density was high, but it did not show any temporal change during the study period despite temporal increases in both explanatory factors. Number of fledglings decreased through declines in both mean clutch size and fledging success. Fledging success was better with higher breedingtime temperature and larger clutch size. Both the clutch size and fledging success were lower at higher breeding density. 4. Selection on laying date did not change through time, but there was a selection for early laying with high breeding-time temperature and high breeding density. Interestingly, in contrast to the decrease in reproductive output, the selection for larger than average clutch size strengthened with time, which was not explained by any tested factor. 5. We suggest that increasingly favourable conditions in winters have enhanced the survival and resulted in the observed increase in great tit breeding density. This may have most concerned young and otherwise low-quality individuals, which also most likely end up breeding in the increasingly occupied low-quality territories. This hypothesis was indicatively supported by increased withinyear variation in both laying date and clutch size. The changes could also explain the lack of advancement in laying date as well as the increasing selection for large clutch sizes as the fittest individuals most likely occupy the best territories and lay largest clutches. Key-words: density dependence, long-term study, reproductive success, selection differential Introduction At temperate latitudes, seasonality makes the timing of reproduction especially important (Lack 1968; Immelmann 1971; van Balen 1973; Barba & Gil-Delgado 1990). Poor timing of breeding can lead to a mismatch with the peak food availability and thus to low breeding success (van Noordwijk, McCleery & Perrins 1995; Visser et al. 1998; Both & Visser 2001; Both et al. 2006). Due to climate warming, the phenology of temperate environments has advanced during the last few decades (Parmesan 2006). Timing of breeding has also *Correspondence author. E-mail: mapeah@utu.fi advanced in a number of bird species (Crick et al. 1997; Dunn 2004). The mechanism behind the change has, however, been little studied. The population-level change in the laying date has been found to be sufficiently covered by individual phenotypic plasticity in the collared flycatcher Ficedula albicollis (Przybylo, Sheldon & Merilä 2000) as well as in some great tit population (Charmantier et al. 2008), but obviously not in all (Visser et al. 1998). Despite concern over how the timing of breeding relates to environmental phenology, surprisingly few studies have examined whether the selection on the timing of breeding or other life-history traits in fact varies in relation to phenology of the environment or temperatures during breeding (Ahola et al. 2004; Visser & Both 2005). In general, Ó 2009 The Authors. Journal compilation Ó 2009 British Ecological Society

Great tits lay increasingly smaller clutches 1299 evidence for a direct connection between selection and observed change at the population level has only seldom been found (Gienapp et al. 2007). Avian clutch size typically decreases with later laying date in seasonally constrained environments (von Haartman 1969; Klomp 1970) and both the timing of breeding and clutch size are affected by parental condition (Rowe, Ludwig & Schluter 1994). Indeed, because the clutch size depends on laying date, the selection on clutch size is linked to the selection on laying date (Sheldon, Kruuk & Merila 2003), but there are cases where the selection on clutch size has found to be independent from that of the laying date (Charmantier et al. 2006). In addition to environmental conditions, the number of competitors affects the availability of food, nest sites and other resources both during breeding and non-breeding seasons (Gustafsson 1987). High breeding densities decrease the clutch size and fledgling production in at least some bird species (Dhondt, Kempenaers & Adriaensen 1992; Both 1998, 2000). Breeding density is mainly altered by reproductive success in previous years and by survival. In temperate breeding areas, resident species may benefit from climate warming if milder winters increase their survival over this critical period (Berthold et al. 1998). Here we report patterns and trends in the timing of breeding, clutch size, and reproductive success of the great tit (Parus major L.) using a long-term data-set from South- Western Finland. We examine if these reproductive traits are affected by climate change through local temperatures or breeding densities of great tits. We further examine patterns of selection on laying date and clutch size on the basis of fledgling production and test whether the direction and strength of selection depend on the environmental and population intrinsic variables. Our aim is a better understanding of the mechanisms behind changes appearing at the population level. We also wanted to shed light on the factors behind the temporal phenotypic changes, as well as to those behind natural selection as they have been often omitted in recent studies related to climatic changes. Materials and methods DATA We used breeding data from two nest-box studies in SW Finland. The first data set was collected by Lars von Haartman during his long-term study in Askainen (60 30 N, 21 45 E). This data set included breeding data on great tits during 1953 1994 (with missing years 1958, 1990 and 1992). The number of nest-boxes studied each year varied from 79 (1993) to 155 (1983) (median = 134, n = 39 years). The second data set was collected by T.E. and E.L. in Harjavalta (61 20 N, 22 10 E) during 1991 2008. From the Harjavalta data, we excluded all nesting attempts that were within 3 km of a polluting industrial complex known to affect the breeding parameters of tits during early 1990s (Eeva, Lehikoinen & Pohjalainen 1997; Eeva & Lehikoinen 2000). Further from the industrial complex, there were from 131 (2001) to 599 (1995) nest-boxes studied per year (median = 462Æ5, n = 18 years). The two study areas are c. 95 km apart. Both consisted of spruce- and pine-dominated coniferous forest, and mixed forests with birches as main broad-leafed species, which are typical for SW Finland. In both areas, nest-boxes were checked at least once a week, and usually more often during the egg-laying period. First egg laying dates were determined from the number of eggs observed during the laying period, assuming a laying frequency of one egg per day. Clutches were considered as completed when incubation began. The number of fledglings was deduced from the number of healthy chicks seen in the nest (at age of 10 days or older) subtracting the possible dead chicks remained in the nest after fledging. Fledging success was measured as the number of fledged young per the number of eggs in a clutch. Predated nests were excluded from the data. Only nests considered to be the first attempt per female each year, where both the clutch size and the number of fledged offspring were known, were used. Proportion of the first clutches from all nesting attempts varied between 58% and 96% and it showed no trend over years (P = 0Æ56). Eventually, the data contained information on 1373 great tit nests. We tested the similarity of laying dates and clutch sizes in the two study areas using data from three common years (1991, 1993 and 1994). A two-way anova showed that there was much more variation among years than between locations in both laying dates (year: F 2,315 =34Æ24, P <0Æ0001; location: F 1,315 =3Æ73, P =0Æ054, n = 66 for Askainen and n = 255 for Harjavalta) and clutch sizes (year: F 2,315 =5Æ16, P =0Æ063; location: F 1,315 =1Æ80, P =0Æ19, n = 64 for Askainen and n = 255 for Harjavalta). In the model for the laying date, there was also a significant year location interaction (P =0Æ0002). This was because both degree and direction of the difference in mean laying dates between the sites varied during the 3 years. Only in 1991 was the between-location difference significant (average laying was 3Æ55 days later in Harjavalta than in Askainen; F 1,86 =11Æ36, P =0Æ0011). During that spring, the nest-boxes were originally placed in the area. Therefore, there were many firsttime breeders (Eeva, Lehikoinen & Sunell 1997), and even the more experienced individuals likely found the nest-boxes later in the spring. In the other 2 years, the between-location differences were much smaller and non-significant (in 1992, 0Æ67 days later and, in 1994, 1Æ39 days earlier in Harjavalta; F 1,114 =1Æ17, P =0Æ28 and F 1,115 =2Æ81, P =0Æ10 respectively). Based on this, we conclude that there was no reason to consider the two data sets systematically different in these traits, but safe to combine. VARIATION AND DETERMINANTS OF BREEDING PARAMETERS We examined temporal trends in yearly mean laying date, clutch size, number of fledglings and fledging success of the great tit. In addition, we examined with multiple regression models the relationships of these variables to a number of environmental and population intrinsic factors. For laying date, we considered as candidate explanatory variables: year, pre-breeding temperature and great tit breeding density. The pre-breeding temperature was the mean temperature of period 2 April 7 May, which is the period that most affects the mean laying date of great tit (Ahola et al. 2007). The breeding density was measured as the percentage of nest-boxes occupied by great tits. This measure was used because, in Askainen, we did not know the exact locations of all nest-boxes, and it was thus not possible to calculate the density per area. We, however, knew that, in both areas, several separate sets of nest-boxes were placed along paths at c. 40 m intervals and the separate sets were tens to hundreds of metres apart. We consider the availability of nest-holes rather than space around them

1300 M. P. Ahola et al. to be the limiting resource in this environment and thus the occupation rate of nest-boxes to be relevant in measuring resource limitation due to within-species density and competition. As the locations of the nest-boxes were the same, meaning constant density, from year to year, the measure is also comparable between consecutive years. For clutch size, the effects of the same variables plus laying date and relative laying date were tested. As the laying date measures the timing of breeding in relation to the progress of photoperiodicity and thus to calendar, the relative laying date was used to measure the timing of breeding in relation to the progress of spring temperatures, equivalent to environmental phenology. It was calculated as mean laying date minus the date when thermal sum 50 was reached. The thermal sum is the cumulative sum of daily mean temperatures exceeding +5 C from the beginning of March. The thermal sum can be considered a useful indicator of the phenology of the environment, and thus that of the species prey items, as the phenology of the main deciduous tree in the study area, the silver birch (Betula pendula Roth), strictly follows the thermal sum (Häkkinen, Linkosalo & Hari 1998). Thermal sum 50 was chosen because it is generally reached at the overall average laying date of the great tit. For mean fledging success, clutch size and breeding-time temperature were added to the set of candidate explanatory variables. Breeding-time temperature is the mean temperature of the almost 6-week period that was pinpointed to have the strongest correlation with the yearly mean fledging success (the number of mean fledged per the number of mean eggs layed; 8 May 25 June; see Ahola et al. 2007 for the pinpointing method). NATURAL SELECTION ON LAYING DATE AND CLUTCH SIZE AND THEIR DETERMINANTS We measured selection on the laying date and clutch size of the great tit on the basis of fledgling production. We calculated yearly standardized selection differentials (Falconer & Mackay 1996) for the laying date (SSD LD ) by subtracting yearly average laying date (LD) from the average laying date weighted for the number of fledglings produced in each nest (LD W ), and dividing this difference by the standard deviation of the laying date (StdevLD), i.e. SSD LD = (LD W ) LD) StdevLD. Respective standardized selection differentials were calculated for the clutch size: SSD CS = (CS W ) CS) StdevCS. We used the selection differentials based on the number of produced fledglings (instead of recruits) because adult great tits (especially males) were insufficiently captured in our study population, leading to low recapture probability for the ringed recruits. It must be noted that the fledgling-based selection differential measures the selection that is expressed until fledging, and it does not take into account the possible effects of post-fledging success of the offspring or long-term fitness of the parents themselves. Unfortunately, we did not either have enough measurements on the offspring, and thus we were not able to measure their fledging condition, which could have indicated their probability to survive after nestling period. However, we assume that our fledgling-based approach is broadly usable. Although the approach does not take into account the possible effect of biased post-fledging survival on natural selection, it, on the other hand, avoids the biases caused by differentials in natal dispersal distance that concerns selection measures based on recruit-production. It is also obvious that the number of fledglings produced by an individual strongly determines the number of its offspring reproducing in further generations. To identify which factors determine the selection on laying date and clutch size in this population, we built up multiple regression models for the selection differentials. The candidate explanatory factors for the standardized selection differential for the laying date were year, mean laying date, relative laying date, great tit breeding density, pre-breeding temperature and breeding-time temperature. For the standardized selection differential for clutch size, the selection differential for laying date and yearly mean clutch size were also added in the set of candidate explanatory variables. ANALYSES In all multiple regression models, we reduced the number of explanatory factors by dropping out the clearly non-significant factors (P >0Æ1) one by one, starting from the least significant explanatory variables. Also the marginally non-significant factors (0Æ05 P 0Æ1) were included in the final models. Estimates are given with their standard errors. We used adjusted R 2 values (Adj R 2 ) to estimate how much the model explains variation in the response variable. The non-significances of the variables dropped at earlier steps of the model reduction were confirmed by adding them one by one in the final model. We also checked the normality of the model residuals, and examined whether there were problems of collinearity on the basis of variation inflation and tolerance values of the parameter estimates. In all analyses, we used the running day numbers, set in such a way that the vernal equinox was always day number 80. The results, however, are reported as calendar dates based on a vernal equinox on 20 March. sas 9.2 (SAS Institute Inc., Cary, NC, USA) was used for all analyses. Results There was no significant change in yearly mean laying date over the study period 1953 2008 (b = 0Æ0013 ± 0Æ039, t =0Æ03, P =0Æ97; Fig. 1a), although the pre-breeding temperature increased (b = 0Æ047 ± 0Æ012, t = 4Æ00, P = 0Æ0002; Fig. 1b). The breeding-time temperature again did not show any trend (b =0Æ0034 ± 0Æ011, t =0Æ31, P = 0Æ75; Fig. 1c). Clutch size decreased by one egg (b = )0Æ021 ± 0Æ0042, t = )4Æ94, P <0Æ0001; Fig. 1d) and the mean fledging success (proportion of eggs that were raised to fledglings) tended to decrease (b = )0Æ0022 ± 0Æ0013, t = )1Æ77, P =0Æ08; Fig 1e). As a result, the mean number of fledglings per brood decreased by almost two fledglings (b = )0Æ032 ± 0Æ012, t = )2Æ57, P =0Æ013; Fig. 1f). The date when thermal sum 50 was reached advanced during the study period (b = )0Æ182 ± 0Æ0564, t = )3Æ22, P = 0Æ0022) and, as a consequence, the relative laying date was significantly delayed (b = 0Æ183 ± 0Æ041, t = 4Æ44, P < 0Æ0001; Fig. 1g). Great tit density had the highest individual values in early 1970s, but still showed a significant linear increase during the whole period (b = 0Æ119 ± 0Æ054, t =2Æ19, P =0Æ033, Fig. 1h). Laying was earlier in years with high pre-breeding temperature (Table 1a) and tended to be earlier also when breeding density was high (Table 1a). After these effects were controlled for, there was a delaying effect of year on the laying date (Table 1a). Mean clutch size decreased with increasing density and it also decreased during the study period (Table 1b). In addition, the mean clutch size tended to decrease both with

Great tits lay increasingly smaller clutches 1301 (a) (b) (c) (d) (e) (f) (g) (h) Fig. 1. Yearly values of great tit breeding parameters and phenological variables in SW Finland during 1953 2008: (a) mean first egg laying date (running day number from 1 January), (b) mean temperature ( C) of the pre-breeding period (2 April 7 May), (c) mean temperature ( C) of the breeding-time period (8 May 25 June), (d) mean clutch size (eggs), (e) mean fledging success (number of fledglings clutch size), (f) mean number of fledglings produced, (g) mean relative laying date (the mean laying date minus the date when thermal sum 50 was reached each year, see Materials and methods) and (h) breeding density measured as the percentage of all nest-boxes occupied by great tits. Table 1. Effects of the candidate explanatory variables on yearly mean (a) laying date, (b) clutch size and (c) fledging success in a great tit population of SW Finland Response variable Explanatory variable d.f. Adj R 2 b ±SE t P (a) Laying date Pre-breeding temperature )2Æ6070 ± 0Æ2763 )9Æ97 <0Æ0001 Year 0Æ1375 ± 0Æ0275 6Æ45 <0Æ0001 Great tit density 1,50 0Æ632 )0Æ1188 ± 0Æ0594 )1Æ00 0Æ051 (b) Clutch size Great tit density )0Æ0284 ± 0Æ0103 )2Æ74 0Æ0086 Year )0Æ0128 ± 0Æ0049 )2Æ60 0Æ012 Relative laying date )0Æ0248 ± 0Æ0134 )1Æ85 0Æ071 Laying date 1,49 0Æ396 )0Æ0240 ± 0Æ0143 )1Æ68 0Æ10 Pre-breeding temperature 1,48 0Æ384 0Æ0169 ± 0Æ1071 0Æ16 0Æ88 (c) Fledging success Breeding-time temperature 0Æ0720 ± 0Æ0114 6Æ34 <0Æ0001 Great tit density )0Æ0055 ± 0Æ0024 )2Æ26 0Æ028 Clutch size 1,50 0Æ501 0Æ0571 ± 0Æ0268 2Æ13 0Æ038 Year 1,49 0Æ498 )0Æ0010 ± 0Æ0011 )0Æ86 0Æ39 Relative laying date 1,49 0Æ491 0Æ0002 ± 0Æ0031 0Æ08 0Æ94 Pre-breeding temperature 1,49 0Æ491 0Æ0004 ± 0Æ0098 0Æ04 0Æ97 Laying date 1,49 0Æ491 )0Æ0008 ± 0Æ0035 )0Æ22 0Æ82 Bolded variables formed the final models. At each line below them, parameters are from a model where an earlier removed candidate explanatory variable (the one at the line) is added to the final model. The lines are in reverse order in which the candidate explanatory variables were dropped out from the model and in the order of absolute values of t within the final models.

1302 M. P. Ahola et al. delaying relative laying date and calendar laying date, but these effects were marginally non-significant (Table 1b). High breeding-time temperature had a positive effect on fledging success, whereas the effect of high breeding density was negative (Table 1c). In addition, the mean fledging success decreased with the decreasing mean clutch size in the population (Table 1c). As the number of fledglings is a function of both clutch size and fledging success, we took the final models for clutch size (density + year + relative laying date + laying date) and for fledging success (breeding-time temperature + density + clutch size), and compared which better explains the number of fledglings. The latter model was found to better explain the variation in the number of fledglings produced (the latter model: Adj R 2 =0Æ593; the first model: Adj R 2 =0Æ210). Selection differentials for the laying date did not change during the study period (b = )0Æ0015 ± 0Æ0011, t = )1Æ40, P = 0Æ17; Fig. 2a). Standardized selection differential for the laying date decreased (indicating change towards selection for early laying) with increasing breeding density and breeding-time temperature (Table 2a). Selection differentials for the clutch size indicated increasing selection for larger than average clutch size over time (b =0Æ0024 ± 0Æ001, t =2Æ14, P =0Æ037; Fig. 2b). There were 3 years (1955, 1982 and 1988) with exceptionally high values that could not be considered as outliers. These three data points were removed from the analyses to check for their effect on the results and to get the model residuals distribute normally (the exceptionally high values were not clearly explained by exceptional values in any of our candidate explanatory variables). The removal of the three data points further clarified the trend of strengthening selection for larger than average clutch size (b =0Æ0028 ± 0Æ0006, t =4Æ63, P <0Æ0001). These data points are yet presented in Fig. 2b. The standardized selection differential for clutch size was not significantly affected by any of the tested explanatory variables besides the temporal trend (Table 2b). (a) (b) Fig. 2. Trends over years in the standardized selection differentials for (a) laying date (SSD LD ) and (b) clutch size (SSD CS ) on the basis of fledgling production (see Materials and methods). The dotted lines show the level at which the SSDs indicate no directional selection. We found that, on the basis of fledgling production, the selection on clutch size changed to the opposite direction than the phenotypic mean of the trait. This kind of lack of response to selection may be explained by environmental change (e.g. Kruuk, Merilä & Sheldon 2003), which enables individuals, that are not favoured under selection on the measured trait, also to reproduce successfully. Unfortunately, we were not able to examine selection on the genetic component of the phenotype. Instead, we examined one potential explanation caused by changed environmental conditions. Increasing breeding density despite lower reproductive success suggested increased survival due to more favourable environmental conditions (see e.g. Laaksonen, Korpimäki & Table 2. Effects of the candidate explanatory variables on (a) standardized selection differential for the laying date (SSD LD ) and (b) for the clutch size (SSD CS ) Response variable Explanatory variable d.f. Adj R 2 b ±SE t P (a) SSD LD Great tit density )0Æ0061 ± 0Æ0025 )2Æ44 0Æ018 Breeding-time temperature 1,51 0Æ154 )0Æ0303 ± 0Æ0127 )2Æ39 0Æ021 Year 1,50 0Æ146 )0Æ0008 ± 0Æ001 )0Æ73 0Æ47 Relative laying date 1,50 0Æ145 )0Æ0021 ± 0Æ0030 )0Æ70 0Æ48 Laying date 1,50 0Æ137 )0Æ0004 ± 0Æ0038 )0Æ11 0Æ92 Pre-breeding temperature 1,50 0Æ141 )0Æ0052 ± 0Æ011 )0Æ48 0Æ63 (b) SSD CS Year 1,49 0Æ290 0Æ0028 ± 0Æ0006 4Æ63 <0Æ0001 Great tit density 1,48 0Æ302 )0Æ0020 ± 0Æ0015 )1Æ35 0Æ18 SSD LD 1,48 0Æ281 )0Æ0495 ± 0Æ0808 )0Æ61 0Æ54 Clutch size 1,48 0Æ275 )0Æ0028 ± 0Æ0200 )0Æ14 0Æ89 Relative laying date 1,48 0Æ287 0Æ0020 ± 0Æ0022 0Æ91 0Æ37 Pre-breeding temperature 1,48 0Æ276 )0Æ0016 ± 0Æ0073 )0Æ22 0Æ83 Breeding-time temperature 1,48 0Æ275 )0Æ0006 ± 0Æ0083 0Æ07 0Æ94 Laying date 1,48 0Æ281 0Æ0013 ± 0Æ0022 0Æ62 0Æ54 Details as in Table 1.

Great tits lay increasingly smaller clutches 1303 (a) (c) (e) (b) (d) (f) Fig. 3. Trends in within-year variation of breeding parameters in the study population of great tits in SW Finland represented as standard deviations (Stdev; upper figures) and 10th and 90th percentiles (P10: filled dots and P90: open dots; lower figures) of clutch size (CS = number of eggs in clutch; a, b), laying date (LD = first egg laying date, 1 = 1 January; c, d) and fledging success (FS = number of fledglings CS; e, f). Hakkarainen 2002). The increased survival supposedly most concerns low-quality individuals that also have low productivity. We thus hypothesized a posteriori that our contrasting results on realized yearly mean clutch size and selection on clutch size could be explained by an increasing proportion of breeders being low-quality individuals or individuals having to occupy low-quality territories (Dhondt et al. 1992). Both of these could potentially be seen as a change in the clutch size distribution. We therefore examined whether the variation in clutch size has changed during the study period. We found a temporal increase in the standard deviation of clutch size (b =0Æ0058 ± 0Æ0025, t =2Æ28, P =0Æ027; Fig. 3a), which was mostly due to a decrease of the lower end of the clutch size range (10th percentile P10: b = )0Æ0264 ± 0Æ0071, t = )3Æ70, P =0Æ005; Fig. 3b), whereas the upper end of the distribution did not change to the same extent (90th percentile P90: b = )0Æ0134 ± 0Æ0050, t = )2Æ67, P =0Æ010; Fig. 3b). Thus, although the mean clutch size of great tit in this population has decreased, variation in it has increased largely due to the increasing proportion of small clutches, supporting our hypothesis on increased proportion of low-quality breeders. Similarly, if the proportion of low-quality breeders in the population has increased, increased variation in the laying date and fledging success can be expected. We found that the standard deviation of laying dates significantly increased during the study (b =0Æ0329 ± 0Æ0091, t =3Æ63, P =0Æ0007; Fig. 3c), as well as the difference between early (P10) and late (P90) breeders (P90 P10: b = 0Æ076 ± 0Æ027, t = 2Æ87, P = 0Æ0059), although the separate changes in them were statistically only weak tendencies (P90: b = 0Æ0445 ± 0Æ0417, t = 1Æ07, P = 0Æ29; P10: b = )0Æ0319 ± 0Æ0381, t = )0Æ84, P = 0Æ41; Fig. 3d). There was no temporal change in standard deviation of fledging success (b = )0Æ0005 ± 0Æ0007, t =0Æ61, P =0Æ54; Fig. 3e). Because the fledging success is a proportion, the P10 and P90 often got values 0 and 1, respectively, and thus assumptions for residual normality were not met for them. Therefore, we calculated Spearman s rank correlations between the two percentiles and year. Neither of the correlations were significant (P10 vs. year: r =0Æ17, P =0Æ21; P90 vs. year: r = )0Æ14, P =0Æ32; Fig. 3f). Discussion The mean laying date of the great tit did not change in relation to calendar dates, and there was no apparent change in the selection on laying date during the last five decades. However, in relation to spring phenology, the laying date markedly delayed (relative laying date) because there was an increase in early spring temperatures. It remains to be examined why the laying dates of the great tit have not followed the advancement of spring phenology, especially as they are largely determined by the pre-breeding temperature in this and other populations (Slagsvold 1976; Dhondt & Eyckerman 1979; McCleery & Perrins 1998; Visser et al. 1998; Ardia, Cooper & Dhondt 2006; Ahola et al. 2007). One factor contributing to this is that according to our analyses, the prebreeding temperature did not directly affect the direction of selection on laying date. Instead, the breeding-time temperature had such an effect: there was stronger selection for early laying when the temperature during the incubation and nestling period was high. This suggests that, especially in warm summers, the late broods are in a disadvantageous situation. They may suffer from food depletion because the development of their main food items, caterpillars, is accelerated when it is warm (see e.g. Visser et al. 1998). In our study environment,

1304 M. P. Ahola et al. however, it is questionable if the food peak has such importance as in Central Europe. In Finland, we do not have good data on the availability of moth caterpillars, but existing information suggests that they, together with substitutive food items like saw-fly caterpillars, are available longer during the breeding season than in Central-European broadleafed forests (van Balen 1973; Eeva et al. 1997; Eeva, Ryömä & Riihima ki 2005). High breeding density additionally strengthened the selection for early laying. This indicates that food depletion during the breeding season may be one reason why early breeders are favoured in a high-density situation. On the other hand, the earliness may partially be a result of favourable choice of territory, as individuals that have occupied the best territories may reach the best breeding condition and thus start their breeding earlier. These early individuals in high-quality territories would also have better breeding success than individuals with lower-quality territories. The territory choice again turns to individual quality in withinspecies competition. Major changes have taken place in the reproductive output of great tits in this population. The mean clutch size decreased by one egg and the mean reproductive output decreased by almost two fledglings during five decades. The fledging success was better with high temperature during breeding, which could stem from both favourable food conditions and lower energy expenditure in warm summers. This effect of temperature is important to notice when estimating the potential effects of climate warming on great tits. Warming summers would favour great tits on average, but with emphasis on early breeding individuals (see above). So far, however, the breeding-time temperatures have not risen in this area over years. Density dependence of clutch size has been found in earlier studies on the great tit. Results of Dhondt et al. (1992) supported the idea of habitat heterogeneity to cause the negative effect of density on clutch size, whereas Both (1998) found evidence for individual adjustment hypothesis to be the reason behind that. The habitat heterogeneity hypothesis also fits to our results, but we cannot separate the variation in the quality of habitats from that of individuals. These probably are cumulative as the poor-quality individuals most probably end up occupying the poorest territories. In addition to density effect on clutch size, we also found the increasing density to decrease population mean fledging success, which suggests that either the individual adjustment has been inadequate or does not exist in this case. With the rather sparse but even distribution of the nest-boxes through our study, we suggest that the great tits do not need to adjust their clutch size to the food limitation by other individuals. Supposedly, the most critical resource here is the nest-boxes, which also determine in what kind of habitats the great tits can settle to breed. The density effect on reproductive output is nevertheless only a partial explanation to the change in clutch size, as the trend over years remained when the effects of density and other factors were controlled for. Interestingly, although clutch size decreased during the study, selection (on the basis of fledgling production) on clutch size linearly changed towards favouring larger than average clutches. Similarly to our study, divergent changes in phenotypic mean and selection on a trait were found in the collared flycatcher on the island of Gotland (Kruuk, Merilä & Sheldon 2001; Merila, Kruuk & Sheldon 2001). In that population, the mean relative body weight of fledglings decreased over time despite selection favouring heavier chicks, and an increase in population mean breeding value at the genotype level (Merila et al. 2001). Also the tarsus length was under directional selection, but there was no response to that selection either (Kruuk et al. 2001). As proposed in those studies (Kruuk et al. 2001, 2003; Merila et al. 2001), this kind of pattern probably appears due to changes in the environment that obscure the effect of the selection on the phenotype (Merila et al. 2001). This can also be the case in our population, but in this population both the decrease in the phenotypic mean clutch size and the increase in the selection for larger than average clutch size might be at least partially caused by the same environmental change. The decreased productivity of our study population did not, as might be expected, result in lower breeding density. Therefore, it is likely that survival during non-breeding season has increased compensating the effect of reduced productivity on population size. Another explanation might be that the increasing number of immigrating individuals from outside of the nest-box areas compensate for the decreased productivity. We, however, suppose that, in our study area, warmer winters together with increased additional food supply by humans have improved the winter survival of the great tits (von Haartman 1973). Survival may have especially improved among young and otherwise lower-quality individuals, increasing their contribution to the breeding population. As these new survivors probably have later laying date and lower reproductive output than such individuals that used to survive already in the past, this would explain the decreasing trends in the mean clutch size and fledging success at the population level, as well as the lack of advancement in the mean laying date with increased pre-breeding temperatures. This idea was, in fact, supported by the temporal increases in standard deviation of both clutch size and laying date. For the standard deviation of clutch size, the increase came mostly due to increasing proportion of small clutches indicating increased proportion of low-quality breeders, while for the standard deviation of laying date it arose due to symmetric widening of the distribution: the early breeders tended to advance and the late ones to delay their breeding time, which fits to both addition of the low-quality individuals and to advancement of breeding in high-quality individuals. Furthermore, although the contribution of the low-quality breeders in the population increases, the fitness of the high-quality individuals may remain the same and still their better fledgling production should be emphasized more in relation to average individuals, just as the increasing selection for larger clutch size showed. It must nevertheless be remembered that our measure for selection is based just on the fledgling production and we were not able to take into account the effects of post-fledging sur-

Great tits lay increasingly smaller clutches 1305 vival of the offspring and or the survival or the lifetime reproductive success of the parents, not to mention the productivity of the offspring. These fitness measures might modify the results on selection, as there is evidence that offspring from larger clutches have lower recruitment rate than those from small clutches (Tinbergen 2005). However, lower recruitment to the natal (study) area does not necessarily mean lower fitness for their parents, as that might just be a result of longer natal dispersal distance of the offspring from large clutches (Nilsson 1989; Tinbergen 2005). Thus, it is possible to get flawed estimates with both recruit- and fledgling-based selection measures. On the other hand, increasing survival of both adults and young may favour a strategy of producing less offspring during one breeding attempt (Dijkstra et al. 1990). This may be at least a partly reason why also the size of the largest clutches (P90) slightly decreased in our study. Selection on clutch size has only seldom been found to be separate from selection on laying date, as the two traits are commonly considered as being linked to each other. Charmantier et al. (2006) found a trend of selection for increasingly larger clutch size in the mute swan (Cygnus olor), which was probably due to human-induced additional food supply and decreased predation risk in the population. Although the enhanced conditions for the mute swans have obviously both turned selection towards larger clutch size and increased mean clutch size in the population (Charmantier et al. 2006), the decrease in the mean clutch size and fledging success in our great tit population could be interpreted as a sign of weakened average condition or decreased reproductive investment of the breeding individuals. Furthermore, this has probably caused the emphasized selection for larger clutch sizes in this population. These results support the suggestion that the favoured phenotype under natural selection must be proportioned to the rest of the population (Cooke et al. 1990). Even if the clutch size favoured under selection would be fixed, the standardized selection differential may still change along with a change in the phenotypic mean of the trait. Acknowledgements We greatly appreciate the contribution of Professor Lars von Haartman, whose lifelong interest in hole-breeding passerines and dedication to data-collection made our study possible. We thank Risto A. Väisänen and the Finnish Museum of Natural History for entrusting the data to us and a number of people for participating in the field work at Harjavalta-area. Christiaan Both, Tore Slagsvold, Pälvi Salo and two anonymous referees gave valuable comments on the manuscript. Robert Thomson kindly checked the language. Finnish Meteorological Institute provided the temperature data. This study was funded by Academy of Finland (Project 8119367 to EL), Kone Foundation (grants to MA and EL) and Emil Aaltonen Foundation (a grant to TL). References Ahola, M., Laaksonen, T., Eeva, T. & Lehikoinen, E. (2007) Climate change can alter competitive relationships between resident and migratory birds. Journal of Animal Ecology, 76, 1045 1052. Ahola, M., Laaksonen, T., Sippola, K., Eeva, T., Rainio, K. & Lehikoinen, E. (2004) Variation in climate warming along the migration route uncouples arrival and breeding dates. Global Change Biology, 10, 1610 1617. Ardia, D.R., Cooper, C.B. & Dhondt, A.A. (2006) Warm temperatures lead to early onset of incubation, shorter incubation periods and greater hatching asynchrony in tree swallows Tachycineta bicolor at the extremes of their range. Journal of Avian Biology, 37, 137 142. van Balen, J.H. (1973) A comparative study of the breeding ecology of the great tit Parus major in different habitats. Ardea, 61, 1 93. Barba, E. & Gil-Delgado, J.A. (1990) Seasonal variation in nestling diet of the great tit Parus major in orange groves in Eastern Spain. Ornis Scandinavica, 21, 296 298. Berthold, P., Fiedler, W., Schlenker, R. & Querner, U. (1998) 25-year study of the population development of Central European songbirds: a general decline, most evident in long-distance migrants. Naturwissenschaften, 85, 350 353. Both, C. (1998) Density dependence of clutch size: habitat heterogeneity or individual adjustment? Journal of Animal Ecology, 67, 659 666. Both, C. (2000) Density dependence of avian clutch size in resident and migrant species: is there a constraint on the predictability of competitor density? Journal of Avian Biology, 31, 412 417. Both, C. & Visser, M.E. (2001) Adjustment to climate change is constrained by arrival date in a long-distance migrant bird. Nature, 411, 296 298. Both, C., Bouwhuis, S., Lessells, C.M. & Visser, M.E. (2006) Climate change and population declines in a long-distance migratory bird. Nature, 441, 81 83. Charmantier, A., Perrins, C.M., McCleery, R.H. & Sheldon, B.C. (2006) Evolutionary response to selection on clutch size in a long-term study of the mute swan. The American Naturalist, 167, 453 465. Charmantier, A., McCleery, R.H., Cole, L.R., Perrins, C., Kruuk, L.E.B. & Sheldon, B.C. (2008) Adaptive phenotypic plasticity in response to climate change in a wild bird population. Science, 320, 800 803. Cooke, F., Taylor, P.D., Francis, C. & Rockwell, R. (1990) Directional selection and clutch size in birds. The American Naturalist, 136, 261 267. Crick, H.Q.P., Dudley, C., Glue, D.E. & Thomson, D.L. (1997) UK birds are laying eggs earlier. Nature, 388, 526. Dhondt, A.A. & Eyckerman, R. (1979) Temperature and date of laying by tits Parus spp. Ibis, 121, 329 331. Dhondt, A.A., Kempenaers, B. & Adriaensen, F. (1992) Density-dependent clutch size caused by habitat heterogeneity. Journal of Animal Ecology, 61, 643 648. Dijkstra, C., Bult, A., Bijlsma, S., Daan, S., Meijer, T. & Ziljstra, M. (1990) Brood size manipulations in the kestrel (Falco tinnunculus): effects on offspring anf parent survival. Journal of Animal Ecology, 59, 269 285. Dunn, P. (2004) Breeding dates and reproductive performance. Advances in Ecological Research, 35, 69 87. Eeva, T. & Lehikoinen, E. (2000) Recovery of breeding success in wild birds. Nature, 403, 851 852. Eeva, T., Lehikoinen, E. & Pohjalainen, T. (1997) Pollution-related variation in food supply and breeding success in two hole-nesting passerines. Ecology, 78, 1120 1131. Eeva, T., Lehikoinen, E. & Sunell, C. (1997) The quality of pied flycatcher (Ficedula hypoleuca) and great tit (Parus major) in an air pollution gradient. Annales Zoologici Fennici, 34, 61 71. Eeva, T., Ryo mä, M. & Riihima ki, J. (2005) Pollution-related changes in diets of two insectivorous passerines. Oecologia, 145, 629 639. Falconer, D.S. & Mackay, F.T.C. (1996) An Introduction to Quantitative Genetics. Longman Higher Education, Essex. Gienapp, P., Teplitsky, C., Alho, J., Mills, J. & Merila, J. (2007) Climate change and evolution: disentangling environmental and genetic responses. Molecular Ecology, 17, 167 178. Gustafsson, L. (1987) Interspecific competition lowers fitness in collared flycatchers Ficedula albicollis: an experimental demonstration. Ecology, 68, 291 296. von Haartman, L. (1969) The nesting habits of Finnish birds I. Passeriformes. Commentationes Biologicae, 32, 1 187. von Haartman, L. (1973) Talgmespopulationen på Lemsjo holm. Lintumies, 8, 7 9. Häkkinen, R., Linkosalo, T. & Hari, P. (1998) Effects of dormancy and environmental factors on timing of bud burst in Betula pendula. Tree Physiology, 18, 707 712. Immelmann, K. (1971) Ecological aspects of periodic reproduction. Avian Biology (eds D.S. Farner & J.R. King), pp. 341 389. Academic Press, New York.

1306 M. P. Ahola et al. Klomp, H. (1970) The determinants of clutch size in birds: a review. Ardea, 58, 1 125. Kruuk, L.E.B., Merila, J. & Sheldon, B.C. (2001) Phenotypic selection on a heritable trait revisited. The American Naturalist, 158, 557 571. Kruuk, L.E.B., Merila, J. & Sheldon, B.C. (2003) When environmental variation short-circuits natural selection. Trends in Ecology & Evolution, 18,207 209. Laaksonen, T., Korpima ki, E. & Hakkarainen, H. (2002) Interactive effects of parental age and environmental variation on the breeding performance of Tengmalm s owls. Journal of Animal Ecology, 71, 23 31. Lack, D. (1968) Ecological Adaptations for Breeding in Birds. Methuen & co LTD, London. McCleery, R.H. & Perrins, C.M. (1998)...temperature and egg-laying trends. Nature, 391, 30 31. Merilä, J., Kruuk, L.E.B. & Sheldon, B.C. (2001) Cryptic evolution in a wild bird population. Nature, 412, 76 79. Nilsson, J.-Å. (1989) Causes and consequences of natal dispersal in the marsh tit, Parus palustris. Journal of Animal Ecology, 58, 619 636. van Noordwijk, A.J., McCleery, R.H. & Perrins, C.M. (1995) Selection for the timing of great tit breeding in relation to caterpillar growth and temperature. Journal of Animal Ecology, 64, 451 458. Parmesan, C. (2006) Ecological and evolutionary responses to recent climate change. Annual Review of Ecology, Evolution, and Systematics, 37, 637 669. Przybylo, R., Sheldon, B.C. & Merila, J. (2000) Climatic effects on breeding and morphology: evidence for phenotypic plasticity. Journal of Animal Ecology, 69, 395 403. Rowe, L., Ludwig, D. & Schluter, D. (1994) Time, condition, and seasonal decline of avian clutch size. The American Naturalist, 143, 698 722. Sheldon, B.C., Kruuk, L.E.B. & Merila, J. (2003) Natural selection and inheritance of breeding time and clutch size in the collared flycatcher. Evolution, 57, 406 420. 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 Scandinavica, 7, 127 145. Tinbergen, J.M. (2005) Biased estimates of fitness consequenses of brood size manipulation through correlated effects on natal dispersal. Journal of Animal Ecology, 74, 1112 1120. Visser, M.E. & Both, C. (2005) Shifts in phenology due to global climate change: the need for a yardstick. Proceedings of the Royal Society of London. Series B: Biological Sciences, 272, 2561 2569. Visser, M.E., van Noordwijk, A.J., Tinbergen, J.M. & Lessells, C.M. (1998) Warmer springs lead to mistimed reproduction in great tits (Parus major). Proceedings of the Royal Society of London. Series B: Biological Sciences, 265, 1867 1870. Received 22 December 2008; accepted 22 June 2009 Handling Editor: Jonathan Wright