The reproductive benefits of livestock farming in barn swallows Hirundo rustica: quality of nest site or foraging habitat?

Authors


Correspondence author. E-mail: martin.gruebler@vogelwarte.ch

Summary

1. In many parts of the world, changes in agricultural land-use have led to significant declines of bird species, including aerial insectivores such as barn swallows. In particular, barn swallow populations have been declining across Europe where mixed and livestock farming have been replaced by arable farming.

2. A positive association between livestock farming and barn swallow reproductive success is well documented but the specific roles of micro- and macroenvironment, which are not mutually exclusive, remain unclear. A positive effect of livestock on swallow breeding performance might be due to improved feeding conditions associated with dung around cattle farms (macrohabitat). Barn swallows also might profit from raised and more constant temperatures at the nest site in stables housing farm animals (microhabitat).

3. We analysed data on barn swallows breeding across Switzerland to quantify the effects of livestock farming at the micro- and macrohabitat on the reproductive success of single- and double-brood pairs. We focus on the effects of nest temperature (expressed as presence of livestock) and food availability around the nest (quantified by the number of manure heaps providing large number of flies).

4. The presence of livestock in the building with the nest and large numbers of manure heaps around nest sites increased nestling survival in double-brood but not in single-brood pairs. Furthermore, the presence of livestock tended to increase the probability of pairs rearing double broods and increased the annual output of double-brood pairs by 0·8 chicks. Both factors of livestock farming combined increased the annual output by 1·6 chicks.

5.Synthesis and applications. The productivity of barn swallows depends on the characteristics of the micro- and the macrohabitat. Since changes in farming systems, grazing patterns, landscape heterogeneity and climate may have different effects on micro- and macrohabitats, respectively, they affect productivity of declining bird species in a complex way. Measures designed to enhance habitat quality in aerial insectivores should improve microclimatic conditions at the nest and increase the number of food patches providing airborne insects. In general, habitat improvements should include both spatial scales, namely suitable sites for nesting and accessible food resources on the foraging grounds.

Introduction

In many parts of Europe, agricultural intensification and the associated changes in agricultural land-use have resulted in a significant decline of many bird species, including aerial insectivores such as barn swallows Hirundo rustica (Vickery et al. 2001; Benton et al. 2002; Benton, Vickery & Wilson 2003; Burfield & Van Bommel 2004; Newton 2004; Donald et al. 2006). In particular, European barn swallow populations declined during the second half of the last century (Tucker & Heath 1994; Evans & Robinson 2004), including Swiss populations (Weggler & Widmer 2000; Schmid et al. 2001); however, country wide declines had lessened or even reversed by the end of the century (Turner 2006). Since the 1960s, agricultural intensification has favoured the polarization of arable and pastoral farming systems so that the production of arable crops has become segregated from livestock farming in different regions (Robinson & Sutherland 2002). As a consequence, where arable farming predominates, the number of barn swallow breeding pairs continued to decline as livestock farming was abandoned (Møller 2001; Ambrosini et al. 2002a; b; Robinson, Crick & Peach 2003; Evans & Robinson 2004).

In barn swallows, animal husbandry, in particular cattle farming, is associated with high numbers of breeding pairs, better offspring survival and increased reproductive success compared to other farming systems (Møller 1983, 2001; Ambrosini et al. 2002a). Livestock farming affects barn swallow habitats at two different spatial scales, the microhabitat of the nest site, and the macrohabitat of foraging grounds. Their relative contribution to the positive association between swallows and the presence of livestock is unknown (Møller 2001; Ambrosini et al. 2002a; Ambrosini & Saino 2010). There may be an increase in insect prey availability at and around cattle farms; alternatively, barn swallows might benefit from livestock creating warmer and more constant temperatures at the nest site inside farm buildings. A correlative analysis of reproductive output for different combinations of nest site quality and foraging habitat quality will not show causation. Nevertheless, it may yield information on the relationships between characteristics of the microhabitat, attributes of the macrohabitat and reproductive parameters of barn swallows. Natural and anthropogenic changes in the environmental conditions might affect nest sites and foraging habitats differently. Therefore, conservation measures should be applied at the right spatial scale, depending on the relationships between habitat characteristics and reproductive parameters and on the spatial scale of habitat degradation.

In seasonal environments, multi-brood bird species (species that invest in more than one brood per breeding season) usually breed earlier (and later) than the optimal breeding time for single-brood species (Crick, Gibbons & Magrath 1993; i.e. before and after the seasonal peak in food availability). Swallow food availability peaks in mid-summer because high temperatures favour insect flight activity and long days allow for prolonged foraging sessions (Grüebler, Morand & Naef-Daenzer 2008). In central Europe, 60 to 83% of barn swallow pairs produce more than one brood, depending on the weather conditions (Turner 2006). Some pairs are thought to be limited to a single brood because they begin reproduction late and the survival probability of a subsequent brood would be very low (Grüebler & Naef-Daenzer 2008); however, this one brood coincides with the most favourable breeding conditions in the season (Turner 2006; Grüebler, Morand & Naef-Daenzer 2008). Additional improvements at nest sites and on foraging grounds through changes in farming practice might therefore have only minor effects on reproductive output of single-brood pairs, because energetic constraints are weak. In contrast, double-brood pairs arrive earlier at the breeding sites. They start egg laying when food availability and temperatures are still low and days are short. Their last broods fledge when prevailing conditions deteriorate again (Grüebler & Naef-Daenzer 2008). Under these conditions, high-quality food patches and increased temperatures at the nest site might be crucial factors affecting reproductive output. We therefore predict that the two potential benefits of livestock farming on reproductive performance (increased food availability and enhanced nest site quality) will have the greatest effect on double-brood pairs.

We studied the reproductive output of barn swallows across Switzerland to quantify the effects of livestock farming at the nest sites (i.e. presence of livestock in the stable) and at foraging habitats (i.e. number of manure heaps around farms within the feeding range of barn swallows) on (1) the start of breeding, (2) single-brood versus multi-broodedness, (3) annual fecundity (number of eggs) and (4) nestling survival to fledging.

Materials and methods

Study areas and agricultural parameters

We studied barn swallows from 1997 to 2004 in 13 areas distributed throughout Switzerland (see Table S1, Supporting information). Study sites were situated north and south of the Alps (denoted as regions) and at different altitudes. The number of farms within each study area varied from 8 to 53. The majority of farms practised mixed farming systems with priority given to dairy farming, using manure from livestock as fertilizer for grass and crop production. However, many different farming systems were represented, including pig farming, arable farming, meat production (beef and veal) and chicken farming. The distribution of farms ranged from single farms surrounded by mixed agricultural fields to clusters of several farms within small villages. Thus, the number of manure heaps around farms could be greatly enhanced by neighbouring active livestock farms, even for swallow pairs breeding at farms (or in buildings) without livestock. For each farm, we mapped all manure heaps within a radius of 500 m, since barn swallows forage mostly within this distance of the nest (Turner 2006). Manure heaps provide high-quality food patches to foraging barn swallows, because the density of aerial insects above dung and pastures is very high, in particular in adverse weather conditions (Evans, Wilson & Bradbury 2007; Grüebler, Morand & Naef-Daenzer 2008). Thus, the number of manure heaps and the area of pasture is a good proxy for high-quality foraging habitats (Loske 1994). We recorded the altitude above sea level of each farm. Within farms, barn swallows often bred in different parts of the farm building complex. Nests were situated in nearly closed rooms, in open buildings or even outdoors. Ambrosini & Saino (2010) showed that rooms housing livestock are significantly warmer than rooms without, and that this difference in the thermal environment affects nestling phenotype rather than differences in room size and structure. In this study, nest sites were classified as occurring in the presence of livestock if nests were located in a room with small openings to provide barn swallows access and if cattle or pigs occupied the room for at least half the day. All other nest sites were classified as being remote from livestock.

Reproductive parameters and adult phenotypes

In all study areas, volunteers and field assistants visited the potential breeding sites every week throughout the breeding season (April to September) to record the number of eggs and nestlings in occupied swallow nests. 17 899 nestlings were ringed 5 to 15 days after hatching. The start of laying was calculated, assuming that females laid one egg per day, that incubation started on the day of clutch completion, and that it lasted for 14 days. The number of nestlings alive on the last visit before fledging was considered as the number of fledglings, i.e. the reproductive output of a brood. This number was adjusted if dead nestlings were found in the nest after the brood had fledged. Adults were trapped and ringed when roosting with their 5- to 15-day-old chicks on the nest or nearby (= 2251 ringed adult individuals). The length of the two outermost tail feathers (tail streamers), a phenotypic quality trait in barn swallows (Møller et al. 1998), was measured to the nearest 0·5 mm. The longer of the two measurements was used in the analysis. We assumed that unringed individuals were immigrating yearlings breeding for the first time from outside our study area (Saino et al. 2002). Adults were trapped during the night to ensure that most individuals could be assigned as the functional parent of a given brood. Parent birds were caught using a special hand net. After putting the opening of the net slowly over the roost, birds were chased into it (overall capture rate = 0·65, with large differences between regions and years). Subsequent broods were assigned to the breeding pairs of the first brood either if adults were trapped in both broods or by assuming that breeding pairs used the exact same nest again for subsequent broods and the onset of the second brood was within 3 weeks from fledging of the first brood. Whenever there was uncertainty, the adults were retrapped. For each breeding pair we calculated annual productivity as the total number of eggs and fledglings produced in a given year.

Statistical analyses

We focused on the effects of livestock farming on fitness correlates of barn swallows, i.e. on fecundity (annual number of eggs) and survival of these eggs up to fledging, while simultaneously controlling for confounding effects, i.e. the effects of date, altitude, parental phenotypic quality traits and age, colony size and differences between regions. In particular, the analyses included two agricultural characteristics associated with livestock farming: the frequent presence of livestock in the room of the nest and the number of high-quality food patches (i.e. the number of manure heaps) in the close vicinity of the nest site. Presence of livestock and number of manure heaps were significantly correlated (Wilcoxon U-test: < 0·001). However, the median number of manure heaps did not differ between rooms with and without livestock present, and the overlap in the numbers of manure heaps between the two room types was large. We chose to retain the two correlated predictors in our models, because recent research suggests that the coefficients in linear models will return unbiased estimates with the inclusion of correlated variables (Smith et al. 2009).

The reproductive performance of swallow pairs could be influenced by individual characteristics of the nest site (and the pair) itself or by more general characteristics of the farm that would affect more than one breeding pair. We therefore used a Linear Mixed Model fit by maximum likelihood (for analysing effects on laying date) and Generalized Linear Mixed Models fit by the Laplace approximation (for analysing effects on multi-brooding, fecundity and survival) in the program R (version 2.8.1, R Development Core Team 2008, libraries nlme and lme4, respectively) to account for the inter-correlation of the observations from the same farms by including the factor farm as a random effect. Moreover, the factor year was added as random effect. Since some individuals bred in several years, we included male and female identity as random effects into the models, if these data were available. We used a Generalized Linear Mixed Model with binomial error distribution and logit-link function to analyse effects on multi-brooding. A Poisson error distribution was assumed and the log-link function used for analysis of the effects on the number of eggs. A binomial error distribution and logit-link function was used to analyse the effects on survival probability. Because there was over-dispersion in the model analysing nestling survival, Markov chain Monte Carlo methods (function MCMCglmm from the library MCMCglmm) instead of Laplace approximation were used to fit the model parameters. This Bayesian method allows extra-variation in the data to be estimated and accounts for this variation when calculating uncertainty of predictions.

It is well known that the laying date of the first clutch and the number of broods per season are major factors affecting reproductive output of barn swallows (Turner 2006; & references therein). Parental age and phenotypic quality of individuals are the main determinants of arrival date, mating success, start of reproduction and the number of annual broods (Balbontín et al. 2007). However, these reproductive traits might depend also on agricultural characteristics. Because data on the individual characteristics of parents were only available for a reduced data set (= 1530 pairs from 278 farms; Table S1, Supporting information), we used a four-step approach in the analyses. In the first step, we investigated whether the start of breeding and multi-breeding were related to agricultural characteristics associated with livestock farming. These analyses used the reduced data set and included age of male and female as well as their tail streamer length as predictors, and their identity as random factors. To account for differences among years and between regions and to control for the effect of altitude, we included region as factor, altitude as covariate and year as random factor. In the analysis of multi-breeding, we used also laying date and brood loss (factor specifying whether all broods of a pair survived to fledging or not) as controlling variables, because the probability of having multiple broods decreases with advancing laying date and increases if the first brood fails to fledge (Møller 1994). In a second step, we investigated factors affecting the annual fecundity and nestling survival by using the reduced data set. In these models, we included the same variables as in the analyses of the first step (except brood loss). We fitted the models by including all two-way interactions between multi-breeding or date and other variables (except region). Non-significant interactions were omitted from the models. In a third step, we omitted parental traits and parental identity from the analyses in a model reduction to estimate the effect of parental traits and pseudo-replication on effect sizes and significance of the focus variables (but note that the use of stepwise modelling has been criticized lately, Whittingham et al. 2006). In the final step, we applied the reduced model of step three to the full data set, which allowed the results of the reduced and the full data set to be compared. Predictions were based on the full data set and calculated for 1000 random samples of the fixed effect parameters (function sim() from library ‘arm’) and the 2·5 and 97·5% quantiles of these were used as the limits of the 95% confidence intervals (CI) for the predictions (Gelman & Hill 2007). Predictions from the model of nestling survival (Bayesian methods) were calculated as the median of their posterior distributions and the 2·5 and 97·5% quantiles gave the 95% credible interval (CrI). For predictions, we defined low-quality foraging habitats as one manure heap within 500 m of the nest, whereas high-quality foraging habitat was denoted by six manure heaps on the nesting farm.

Results

Descriptive statistics of breeding parameters and traits of the farms for the reduced and the full data set are given in Appendix S1 (Supporting information).

Start of reproduction and multi-broodedness

Individual characteristics of the parent birds were the most relevant factors affecting laying date of the first clutch (Table S2, Supporting information). In both sexes, older individuals and individuals with longer tail streamers started reproduction earlier than yearlings and short-tailed individuals (Table S2). Age as well as tail feather length contributed to the variation in laying date. The most important factor was female age with older females starting reproduction more than 10 days earlier than yearling females. Moreover, differences between individual females explained a high proportion of the variance (variance females = 42·48 ± 6·52 SD; residual variance = 84·37 ± 9·19). Breeding started approximately 5 days earlier at an altitude of 500 m compared to 1200 m above sea level (Table S2). Livestock presence at the nest site and the number of manure heaps around farms showed no significant effect on laying date (Table S2).

The most relevant factor associated with the occurrence of multiple broods was laying date of the first clutch. Thus, multi-broodedness was linked to an early start of reproduction (Table S2). Late starting pairs showed a significantly reduced probability of being multi-brood in comparison to early pairs. In addition, the frequency of multiple broods was significantly increased if the first brood was lost, if the male of a pair was older than 1 year, and if the nest was at a farm at higher altitudes. Further, multi-broodedness tended to be positively related to the presence of livestock at nest sites (Table S2). There was no significant difference in the frequency of multi-broodedness between regions or between females of different age. Neither tail streamer lengths nor the number of manure heaps within 500 m around the nest were significantly associated with the frequency of multi-broodedness.

Fecundity and nestling survival

In the analysis of the factors affecting annual fecundity we found no differences in the size of effect between the different models based on the reduced or the full data set (Table S3, Supporting information). Neither the exclusion of parental traits nor that of parental identity changed effect sizes of the focus variables. Using the full data set only decreased P-values of significant effects due to the larger sample size. Fecundity showed no differences between regions. Neither the number of manure heaps within 500 m around the nest, nor livestock presence at the nest site had any significant effect on fecundity. Females of multi-brood pairs laid significantly more eggs in a season than females of single-brood pairs. In both, single-brood and multi-brood pairs, fecundity declined if the laying date of the first clutch was delayed (Fig. S1 and Table S3, Supporting information). However, this decline was stronger in multi-brood than in single-brood pairs (Fig. S1), indicated by the significant interaction between multi-broodedness and laying date (Table S3). Furthermore, females of pairs breeding at high altitudes produced slightly more eggs than females of pairs breeding in the lowlands (altitude 500 m asl: 6·63 eggs; altitude 1200 m asl: 7·00 eggs).

In the analysis of the factors affecting nestling survival, the stepwise exclusion of parental traits and parental identity did not change effect sizes and P-values of the focus variables, although male age showed a significant effect on nestling survival (Table S4, Supporting information). Moreover, using the full data set showed that effect sizes were within the 95% credible interval of effect sizes of the model using the reduced data set. However, P-values increased most probably due to the higher sample size of farms in the full data set (Appendix S1, Table S1 and Table S4, Supporting information). Therefore, we used effect sizes from the analyses of the full data set to calculate predicted values for the focus variables (Appendix S2, Supporting information). The model contained four important interactions between multi-broodedness and other variables (Fig. 1, Table S4). The effects of livestock presence, number of manure heaps within 500 m around the nest, laying date and altitude of the nesting farm on nestling survival differed significantly between single-brood and multi-brood pairs. Nestlings of multi-brood pairs showed a higher survival probability when they were raised in the presence of livestock and on farms surrounded by a large number of manure heaps compared to when they were raised in the absence of livestock or on farms with few manure heaps (Fig. 1a, b). However, in single-brood pairs nestling survival was neither associated with the presence of livestock at the nest nor with the number of manure heaps around nest sites (Fig. 1a, b; Table S4). Moreover, the positive effect of livestock presence at nesting sites was considerably higher in late pairs (both single-brood and multi-brood) than in early pairs. The seasonal decline in nestling survival was considerably steeper in multi-brood than in single-brood pairs. This resulted in a higher nestling survival of multi-brood pairs at the beginning of the season, but no differences in nestling survival between single- and multi-brood pairs at the end of the season (Table S4). South of the Alps, nestling survival was reduced compared to breeding areas in the north.

Figure 1.

 Nestling survival up to fledging (model estimates) of single- and multi-brood pairs (a) for pairs breeding in rooms with no livestock presence (closed circles) and frequent livestock presence (open circles), and (b) for pairs with few manure heaps (one heap, closed circles) and many manure heaps (six heaps, open circles) around the nest site. Vertical bars represent 95% credible intervals.

Annual reproductive output

The average predicted annual output of barn swallow pairs computed by multiplication of the predicted number of eggs and predicted nestling survival was 4·40 fledglings (single-brood pairs: 2·92 fledglings; multi-brood pairs: 7·20 fledglings). In multi-brood pairs, predicted values of annual output were increased by 1·61 fledglings if nests were located at high-quality nest sites (i.e. in the presence of livestock) surrounded by high-quality foraging habitats (i.e. large numbers of manure heaps) compared to nests located at low-quality sites with low-quality foraging habitats (Fig. 2). However, these positive effects of livestock farming were not present in single-brood pairs (Fig. 2). The presence of livestock accounted for an increase of 0·8 fledglings in the annual reproductive output of double-brood barn swallows. Moreover, pairs at high-quality sites (including both nest site and surrounding foraging habitat) showed a higher probability to be multi-brood (= 0·971) compared to low-quality nesting sites (= 0·946), leading to an average annual output of 8·31 fledglings at high-quality nesting sites compared to 6·56 fledglings at low-quality nesting sites. Thus, average annual output at high-quality sites was 26% higher than at low-quality sites.

Figure 2.

 Annual output (model estimates) of single- and multi-brood pairs in different combinations of foraging habitat quality (one or six manure heaps) and nest site quality (livestock presence or absence). Vertical bars represent 95% confidence intervals.

Discussion

We have demonstrated that the annual output of barn swallows was related to two different aspects of livestock farming. First, the presence of animals in the room of the nest was positively related to nestling survival and, by trend, to the pairs’ probability of producing more than one brood in a season. Secondly, high numbers of manure heaps in the foraging area around the nest (within 500 m) were associated with enhanced nestling survival. The results of this correlative study suggest that both the microhabitat of the nest site and the macrohabitat of the foraging grounds are associated with the annual reproductive output. In multi-brood pairs, the annual reproduction in top quality habitat (including micro- and macrohabitats) increased by more than 1·5 fledglings per annum compared to sites of lower quality. In contrast, single-brood pairs showed no difference in their annual output between sites of different quality, indicating that agricultural factors affected the reproductive success of single- and double-brood pairs in different ways.

Hitherto, it was difficult to quantify the effects of habitat characteristics around farms associated with differential farming practices on the reproductive performance of barn swallows because (1) the critical evaluation of the hypotheses requires many farms, as individual farms represent statistically independent observations (Møller 2001), and (2) it was critical to disentangle the effects of agricultural characteristics of the nest site from those of the surrounding habitat (Ambrosini et al. 2002a). We overcame these problems by using a large data set from a sufficiently large sample of farms over several years across Switzerland. We also made use of the fact that topography has limited Swiss agriculture to small-scale enterprises as compared to neighbouring countries. Thus, barn swallows nesting on farms without livestock often have access to manure heaps at neighbouring farms. In contrast, birds nesting in rooms with livestock may use only one or even no manure heap if the farm is isolated. Nevertheless, in a correlative study like ours it is impossible to show clear evidence of causation. Therefore, we cannot tell whether the quality of sites affects the reproductive output of barn swallows or, alternatively, whether specific kinds of birds recruit to high-quality sites. Livestock presence and number of manure heaps were not related to the parents’ tail streamer length or age. However, we cannot exclude the possibility that birds at high- and low-quality sites differ in other traits affecting their reproductive output.

Our results provide evidence that habitat characteristics around farms (macrohabitat) are important determinants of the reproductive performance of barn swallows. However, manure heaps can be a proxy for habitat characteristics of livestock farms, such as soil quality, field types and landscape structure. Consequently, the direct effect of manure heaps on the birds’ reproduction remains unclear. On the other hand, it is well known that manure produces and attracts insects around the farmyards close to the nests of barn swallows (Loske 1994; Vickery et al. 2001). Recent studies on the availability of flying insects in agricultural landscapes identified the following areas where the main food of barn swallows accumulate even in adverse weather conditions: pastures and manure heaps, windbreaks such as hedgerows, orchards and single trees, and open standing or flowing water bodies provide considerably more insects than arable and silage fields (Vickery et al. 2001; Evans, Bradbury & Wilson 2003; Evans, Wilson & Bradbury 2007; Grüebler, Morand & Naef-Daenzer 2008). We suggest that such high-quality food patches in the foraging habitat of breeding barn swallows are important to ensure nestling survival in adverse weather situations. In rainy and cold weather, parent barn swallows have to trade self-maintenance against feeding of the brood, which results in decreased feeding rates (Jenni-Eiermann et al. 2008). Broods may even be abandoned during long-lasting periods of inclement weather. Any remaining food patches of high insect availability close to the nest site are therefore likely to increase the probability of nestling survival.

The positive effect of cattle farming on abundance and reproduction of barn swallows has mainly been attributed to the increased food supply at and around farms (Møller 2001; Ambrosini et al. 2002a). However, our results provide evidence that the presence of livestock in the immediate vicinity (the same room) of the nest is crucial for nestling survival and increases the probability of producing more than one brood in a season. Although livestock presence is also associated with other characteristics of the nesting site (e.g. presence of rats, cats and humans), the effects detected are likely to be linked to the higher temperature and the constant microclimatic conditions in the neighbourhood of livestock. Our results are in agreement with recent studies on hirundines showing that heating nests during incubation results in females adjusting their incubation investment and parents’ feeding nestlings at higher rates (Spencer & Bryant 2002; Pérez et al. 2008; Ardia et al. 2009). Nest temperature during incubation even affected nestling condition, suggesting carry-over effects from incubation to the nestling period (Pérez et al. 2008). Moreover, ambient temperature was an important determinant of growth, development and survival of nestlings (McCarty & Winkler 1999; Dawson, Lawrie & O’Brian 2005). We therefore suggest that breeding in rooms that house livestock allows females the opportunity to devote more time to foraging and less time to warming her offspring, and nestlings may allocate more energy to growth and less to thermoregulation.

As predicted, the positive effects associated with livestock presence and manure heaps were most apparent in pairs producing more than one brood. This suggests that single-brood pairs may be less vulnerable to poor weather conditions than double-brood conspecifics. Single-brood birds raise their young at the annual peak of insect availability and in the period with the lowest probability of bad weather. In contrast, multi-brood pairs are prone to experience critical periods of inclement weather either when rearing the first or the second brood, depending on the start time of reproduction. Therefore, in seasonal environments double-brood pairs are especially constrained by the unpredictable conditions at the beginning and at the end of the breeding season. In particular, constraints on multi-brooding should be aggravated if the breeding season is short, as in higher latitudes and higher altitudes. This was corroborated by our result that offspring of multi-brood pairs at high altitudes showed a larger seasonal decline in survival than offspring of single-brood pairs. Further support is provided by the effects of laying date on nestling survival differing between single- and multi-brood pairs. Late double-brood pairs experienced a steeper seasonal decline in nestling survival than late single-brood pairs, probably because the second (and third) brood was exposed to poor climatic conditions and short day lengths at the end of the season.

Fecundity was not directly related to the presence of livestock or the number of manure heaps. However, whether a pair raised one or more broods in a season was an important factor associated with fecundity and since this tended to depend on the presence of livestock at the nest site, animal farming may show an indirect effect on female fecundity. This is in line with recent studies showing that cattle farming increases the proportion of pairs producing second broods (Møller 2001). The underlying mechanism might be that the reproductive value of a second brood is increased by the presence of livestock, since the nestlings show higher survival probabilities and enhanced development compared to nestlings in the absence of livestock.

Thanks to the topography of our study area, we were able to quantify the effects of altitude on different reproductive parameters. In barn swallows, such data are still scarce. As in many species, reproduction started later and nestling survival was reduced at higher altitudes in the Alps compared to the Swiss midlands, most probably due to the shorter breeding season at higher elevations. These results suggest that barn swallow populations at high altitudes experiencing a curtailed breeding season are particularly susceptible to cessation of livestock farming and deterioration of the surrounding foraging habitats. Female fecundity and the rate of multi-broodedness increased with altitude. This supports our finding of an increase in clutch size for higher latitudes (Royama 1969; Møller 1984). It has been argued that this may be because of a combination of long days and abundant food. Since at higher altitudes, as in high latitudes, the growing season is shorter, but days are longer, our results suggest that the observed reproductive patterns are linked to an increased food supply rather than to longer days.

Management implications

The results of this study suggest that measures to improve productivity of barn swallow populations should focus on both the characteristics of the nest sites and the quality of the surrounding foraging habitats. In the feeding grounds, i.e. a radius of 500 m around the nest sites (Turner 2006), landscape heterogeneity should be high and should include features encouraging airborne insects such as manure heaps, pastures, permanent meadows and water bodies (streamlets, ponds or slops). Moreover, structures where flying insects accumulate even in adverse weather conditions such as hedgerows, orchards or single trees should be available (Grüebler, Morand & Naef-Daenzer 2008). At least part of the population decline of barn swallows (Tucker & Heath 1994; Turner 2006) is probably due to the intensification of farming practices, accentuated by changes from animal husbandry to arable farming. Nest sites in constantly warm microclimatic conditions are likely to be most favourable for barn swallow breeding, particularly during cold weather. We suspect that stables with a high air exchange rate leading to balanced temperatures within and outside are inferior for barn swallows compared to poorly ventilated stable buildings. Thus, changes in the architecture of animal husbandry are likely to affect the reproductive output of bird species breeding in these buildings. However, further research is needed to gain insights into how nest site quality could be enhanced in modern livestock stables, in shelters for horses or cattle, and in arable farms without livestock.

In general, more attention should be given to the microhabitat conditions at the nesting sites of threatened or vulnerable bird species, in particular those associated with buildings or in holes or niches. Optimizing the energetic constraints in adults and juveniles make nest site choice an important parental decision. Hence, adequate nest sites with optimal thermal conditions are crucial to high reproductive output in many bird species.

Our results suggest that the reproductive output of multi-brood birds depends on factors enhancing breeding and developmental conditions at the beginning and at the end of the breeding season, when breeding environment is suboptimal. In multi-brood birds, the decision to breed early may be constrained by the microhabitat of the chosen nest site. The benefits of an early start might be counteracted by a poor-quality nest site unprotected from low outside temperatures. This is particularly true when the breeding season is short, for example at high altitudes. Livestock presence at the nest site and predicted climate change towards warmer springs and an extended breeding season in autumn should improve the reproductive output of birds breeding in buildings. Thus, climate change may relax the dependence of these species on livestock farming, at least in the lowlands of central Europe.

Acknowledgements

We are very grateful to all volunteers who collected the data on breeding biology and individual characteristics of barn swallows. Without their effort, this study would not have been possible. We thank J. Balbontín, L. Jenni, L. Schifferli and two anonymous reviewers for valuable comments on the manuscript. Financial support was provided by the Botanisch-Zoologische Gesellschaft Liechtenstein-Sarganserland-Werdenberg, Lotterie romande, Karl-Mayer Stiftung, Migros-Genossenschaft, Museo cantonale di storia naturale di Lugano, Natur- und Landschaftskomission des Kantons Baselland, Region Sarganserland-Walensee, Sarganserländische Talgemeinschaft and Stiftung für Suchende.

Ancillary