Migratory birds usually respond to climate change by modifying breeding and/or wintering areas, as well as by reproducing earlier. In addition, changes in winter habitat use or breeding phenology could have important carry-over effects on subsequent breeding success. Here, we studied age- and sex-dependent carry-over effects from wintering to the breeding stage of a small aerial insectivorous long-distance migratory bird, the barn swallows (Hirundo rustica) breeding in Denmark during 1984–2013. First, we used stable isotope analyses combined with ringing recoveries to identify wintering areas. Second, we found that environmental conditions as inferred by Normalized Differential Vegetation Index (NDVI) have deteriorated at the wintering grounds. Third, we used confirmatory path analysis to quantify the indirect effect of winter conditions on subsequent breeding success. Females advanced onset of breeding, laid more eggs and raised more fledglings in the first brood when ecological conditions during the previous winter improved. This response was age dependent, since yearlings did not respond to this environmental cue but the response was increasingly stronger as females aged. Males showed a similar response to winter conditions, although not statistically significant. These results highlight the importance of studying carry-over effects within the context of climate change, especially in relation to age of individuals.
Many species of birds, especially those breeding at high latitudes, have adapted to perform regular annual movements between breeding and non-breeding areas, often involving thousands of kilometers1. Migratory behavior is important because, in an ecological context, the previous history and experience of a given individual may explain its performance in the current situation (i.e. “carry-over effect”)2. Hence, in migratory birds, what occurs during winter and the migratory journey could have consequences for subsequent reproductive success back at the breeding areas, which in turn, may result from a response to previous environmental conditions3,4. Furthermore, carry-over effects from the winter to the breeding stage may change according to age and sex classes5,6,7,8,9,10,11. Environmental conditions experienced in winter should first affect physical condition during winter, then timing of departure from wintering sites and ultimately timing of arrival to breeding areas12,13. In addition, arrival time to breeding areas usually determines the start of breeding, and this in turn influences the number of offspring produced14,15,16. Because all these variables flow sequentially in a time series fashion, Structural Equation Models (hereafter: SEM) provide a powerful and elegant approach to investigate carry-over effects in migratory birds17,18,19,20,21.
Differential migration among age and sex groups may arise because of the different role of each sex in reproduction, timing of other events during the annual cycle (e.g. moult) or factors related to body size and behavioral dominance1. For instance, males usually arrive at breeding areas earlier than females, a consequence of sexual selection through female choice1. Because females usually select partners that are established in the most suitable nesting habitats, competition for earlier arrival to breeding areas is higher among males, and they can develop different strategies for this purpose22,23,24. For example, males can stay closer to breeding areas during winter, advance departure date from winter areas, travel more quickly during spring migration or compete more aggressively for food1. In addition, young adults breeding for the first time usually arrive later to the breeding areas and have lower breeding success than more experienced adults. Age-related migratory behavior may differ for several reasons. For example, dominant (e.g. older) individuals could displace subordinates (e.g. younger) from better winter or stopover sites. In such cases, older individuals require less time to acquire fuel reserves or to complete moult, allowing older individuals to depart earlier from the wintering areas or to speed up the return journey (the so-called dominance hypothesis)1,25. On the other hand, individuals should gain competence through experience as they age, allowing them to perform the migratory journey better or to respond more eficiently against environmental conditions found in the wintering areas (the so-called constraint hypothesis)26,27. There is evidence supporting both hypotheses that are not mutually exclusive1,27. Alternatively, a third explanation for variation in migratory behavior among age classes is the disappearance from the population of individuals according to their quality (the so-called selection hypothesis)26,27. For example, an improvement with age in migratory performance at the population level (e.g. earlier arrival to the breeding areas) could be explained by the disappearance from the population of later arriving individuals (than average) as they age (i.e. removal of poorer quality individuals in relation to migratory performance).
Many different climate variables (e.g. temperature and precipitation) are changing rapidly worldwide and are predicted to continue to change in future years28. The use of biotic indices are especially powerful to understand and predict the indirect effect of climate change on organisms; such is the case of the Normalized Differential Vegetation Index (hereafter: NDVI)29,30. NDVI integrates the effect of climate on vegetation and vegetation dynamics in turn are linked to animal distribution and performance. For example, NDVI is linked to abundance and diversity of insects31,32,33. Associations between NDVI and climatic variables shift throughout the world. For instance, in tropical latitudes lower temperature and higher rainfall are usually correlated with higher NDVI34. Therefore, lower temperature and higher rainfall may result in greener vegetation, higher abundance of insects31,33,35,36, and thus favorable conditions for aerial insectivores wintering in the tropics, whereas higher temperatures and lower precipitation may result in the opposite5,7,37,38,39.
The barn swallow (Hirundo rustica) is a small passerine (c. 20 g) that breeds semi-colonially and feeds on insects while flying7,18. This species is a long-distance migrant that usually breeds north of the Tropic of Cancer and winters south of this latitude throughout the world40,41. For example, the Danish breeding population covers >6000 km to reach Central and South Africa40. Barn swallows provide excellent longitudinal data because once recruited to the breeding population, individuals almost always return to the same colony to breed every year5,8,23. In addition, first reproduction occurs at the first year, and therefore, a high trapping effort each year allows the estimatation of age with accuracy in this species5,6,8.
In this study, first we used both stable isotope analyses and ringing recoveries to determine the winter areas used by our study populations of barn swallows breeding in Denmark. Second, we tested if winter conditions based on satellite-derived NDVI have changed during the last three decades corresponding to the data available on our barn swallow study populations (i.e. 1984–2013). Third, we focused on how environmental conditions experienced in wintering areas, based on NDVI, affected subsequent breeding date and reproductive success measured as clutch size and number of fledglings in the first brood. We were interested to determine how the response to external factors affected these life history traits during the lifetime of known-aged individuals over the 30-year period (see Table 1 for sample sizes).
To explain age-related difference in response to conditions found at winter areas and subsequent carry-over effects during the breeding period, we set a priori predictions from established hypotheses as described above. Many previous studies have emphasized that earlier arrival to a breeding colony is associated with earlier breeding date and higher reproductive success of migratory birds14,15,23,24. Individuals may thus depart earlier from the wintering grounds in order to arrive earlier at the breeding areas12,42,43. To date it is well known that timing of migration depends on external environmental cues such as photoperiod44, weather conditions1,45 or winter habitat quality12,13. For instance, Coppack and collaborators44 experimentally analyzed the onset of migratory activity in the European pied flycatcher (Ficedula hypoleuca). These authors simulated different wintering latitudes by exposing birds to different photoperiods, and they found that a moderate increase in wintering latitude from 10° to 20°N resulted in a pronounced advancement in the termination of molt, initiation of migratory activity and gonadal growth. As another example, American redstarts (Setophaga ruticilla) wintering in Jamaica start spring migration as soon as they acquire the adequate body condition to withstand the migratory journey, which in turn depends on the quality of the habitat chosen for wintering and also on internal factors such as behavioral dominance among age and sex classes12,13. If we assume that individuals depart from their wintering areas as soon as possible after acquiring the appropriate body condition to start the migratory journey, then under the competition hypothesis we should expect a smaller age-related difference in breeding date in good compared to poor years. This outcome is expected because during good years competition is relaxed, and there should be less difference in breeding date amongst individuals of different ages. If it is the inexperience of young individuals that causes a delay in the onset of breeding, then environmental conditions experienced during winter should be less important, and we would expect similar age-related differences in breeding date with individuals breeding earlier as they age (i.e. constraint hypothesis).
All the processes we followed to combine ringing data with stable isotope values are summarized in Fig. 1. The prior surface derived from movement directions effectively narrowed the potential moult origin of our population of barn swallows. Specifically, Fig. 1E represents the potential origins based only on the multivariate assignment of δ2H, δ13C and δ15N, but after having incorporated the prior surface within the calculation of the spatially explicit probability density maps for each individual (fx), the potential origins for the population are narrowed from 10°E to 30°E (Fig. 1G,H). Wintering areas were largely spread through Cameroon, Central African Republic, Gabon, Congo, Democratic Republic of Congo, Angola, Zambia, Zimbabwe and some patches across Botswana and South Africa (Fig. 1H). Total surface of the wintering area was 1,370,090 km2, 492,257 km2 and 97,621 km2 for our different spatial scales (respectively pixels assigned to be the origin for >50, >60 and >70% of our sample). African biomes dominating the wintering areas were mainly equatorial broadleaf forest, tropical savannah and open shrubland46.
On one hand, these winter areas that we identified matched adequately with previous ringing recoveries for barn swallows breeding in Denmark47. On the other hand, we believe that our method to identify potential wintering areas is robust against any bias in ringing effort across Africa. As an example, we identified moult origins in the western portion of South Africa (Fig. 1G), but ringing recoveries in this country are located in the centre and in the east (Fig. 1D), which is in agreement with another study based on NDVI correlation analyses39.
Temporal trends in environmental conditions from wintering areas
Temporal trend in NDVI averaged across all the wintering area showed a decline after 1991 (Fig. 2). We found a positive main effect of year and a negative quadratic effect of year on NDVI values extracted from pixels that were assigned to be the likely origin for >50% of our sample (year: estimate = 0.747, SE = 0.359, p = 0.046; year2: estimate = −0.0002, SE = 0.0001, p = 0.046; R2 = 0.189), >60% of our sample (year: estimate = 0.829, SE = 0.386, p = 0.040; year2: estimate = −0.0002, SE = 0.0001, p = 0.040; R2 = 0.211) and 70% of our sample (year: estimate = 0.879, SE = 0.393, p = 0.033; year2: estimate = −0.0002, SE = 0.0001, p = 0.033; R2 = 0.2).
In order to determine differences in environmental conditions within our identified wintering area, we repeated the above analysis for each African biome. Temporal trend in NDVI of savannahs from our wintering area mirrored the average trend found across different biomes (Fig. S1 in Supplementary Materials). Specifically, we found a positive main effect of year and a negative quadratic effect of year on NDVI values extracted from pixels that were assigned to be the likely origin for >50% of our sample in woody savannahs (year: estimate = 0.682, SE = 0.293, p = 0.027; year2: estimate = −0.0002, SE = 0.0001, p = 0.026; R2 = 0.493) and savannahs (year: estimate = 0.308, SE = 0.148, p = 0.046; year2: estimate = −0.0001, SE = 0.0001, p = 0.045; R2 = 0.365). By contrast, NDVI of evergreen broadleaf forests from our wintering area showed a linear increase through the years (estimate = 0.003, SE = 0.001, p = 0.002, R2 = 0.268). Finally, NDVI of open shrublands from our wintering area did not show any significant temporal trend (year: estimate = −0.033, SE = 0.085, p = 0.701; year2: estimate < 0.0001, SE < 0.0001, p = 0.7; R2 = 0.029).
Winter conditions vs subsequent breeding phenology and reproductive success
Environmental conditions from the winter areas carried over to affect subsequent reproductive traits, with effects stronger in females compared to males (Table 2; Fig. 3). Confirmatory path analysis revealed that winter conditions affected breeding phenology of females differently according to age class. The direct effect of the interaction term between age and NDVI on breeding date of females was significant [estimate (SE) = −0.712 (0.323), p = 0.028]. In addition, breeding date negatively affected clutch size of females [estimate (SE) = −0.289 (0.022), p < 0.001]. Furthermore, clutch size positively affected the number of fledglings produced by females [estimate (SE) = 0.484 (0.019), p < 0.001]. Because all these direct effects were significant, the effect of the interaction between age and NDVI on breeding date was ultimately translated to affect the number of fledglings raised, and this indirect effect was calculated by multiplying the standardized path coefficients connecting the former and latter variable. Thus, females advanced laying date (Fig. S2), laid more eggs (Fig. S3) and produced more fledgings (Fig. 4) when NDVI during the previous winter was higher and the slope of this response was increasingly stronger as they aged. For example, considering a shift on NDVI values from bad winter years (i.e. NDVI = 0.5) to good winter years (i.e. NDVI = 0.7), yearling females advanced breeding date only 3 days (from day 144 to day 141), whereas five year-old females advanced breeding date by about 18 days (i.e. from day 108 to day 90). Likewise, for yearling females the path analysis predicted a shift of 0.07 eggs (i.e., from 4.95 to 5.02) and 0.06 fledglings (i.e., from 4.12 to 4.18) when comparing predicted values from bad winter years to good winter years, whereas for five year-old females the path analyses predicted a shift of 0.34 eggs (i.e., from 5.64 to 5.98) and 0.32 fledglings (i.e., from 4.76 to 5.09). On the other hand, setting winter conditions constant, the shift in breeding date across age was much stronger. Specifically, the predicted values obtained for breeding date in response to the age of females (i.e. from 1 to 5 years) accounted for a difference in 36 days (i.e., from day 144 to day 108) under bad winter conditions (i.e. NDVI = 0.5), whereas the difference in breeding date accounted for 50 days (i.e., from day 140 to day 90) under good winter conditions (i.e., NDVI = 0.7). Likewise, setting winter conditions constant, predicted values obtained for age-related change in clutch size and number of fledglings also showed a large shift. Specifically, under bad winter conditions our analyses predicted a change of 0.68 eggs (i.e., from 4.95 eggs of one year-old females to 5.63 eggs of five year-old females) and 0.65 fledglings (i.e., from 4.12 fledglings of one year-old females to 4.77 fledglings of five year-old females), whereas under good winter conditions our analyses predicted a larger change of 1 egg (i.e., from 5.02 eggs of one year-old females to 5.98 eggs of five year-old females) and 1 fledgling (i.e., from 4.18 fledglings of one year-old females to 5.09 fledglings of five year-old females).
For males, the direct effect of the interaction term between age and NDVI on breeding date was similar to that found for females, but not statistically significant [estimate (SE) = −0.569 (0.339), p = 0.094]. In addition, breeding date negatively affected the number of fledglings raised by males [estimate (SE) = −0.222 (0.023), p < 0.001]. Consequently, carry-over effects operating in males were similar to that of females, but weaker in strength. Confirmatory path analyses also revaled a significant negative effect of tail length on breeding date for both females [estimate (SE) = −0.060 (0.022), p = 0.006] and males [estimate (SE) = −0.097 (0.022), p < 0.001]. Therefore, barn swallows with longer tail feathers started to reproduce earlier and ultimately raised more fledglings. In addition, body mass was only significantly correlated with breeding date of females (estimate = 0.042, p = 0.029), indicating that females which started to reproduce later were heavier (because most of body mass measures were taken after breeding date).
We used d-separation test to quantify the goodness of fit of our models, which tests the assumption that all variables are conditionally independent48. Both path analyses provided robust fit to data (females: Fisher’s C = 15.3, df = 10, p = 0.121; males: Fisher’s C = 2.2, df = 4, p = 0.699). Thus, we concluded that the hypothesized causal relationships we examined were consistent with the data. By further inspection of “qq plots”, we determined that every single linear mixed model was adequately fitted.
We determined with reasonable precision the wintering areas of barn swallows breeding in Denmark, being distributed from Central to South Africa. This is the first study in which carry-over effects from wintering areas on subsequent reproduction have been analyzed at such large spatial and temporal scales, taking into account age- and sex-dependent variation. Environmental conditions in African wintering areas have changed during the study period showing a steep decline after 1991. We found that barn swallows advanced breeding date, produced larger clutches and raised more fledglings when experiencing good conditions during the previous winter. Interestingly, we found sex- and age-dependent carry-over effects in the annual cycle from wintering to breeding stage. Environmental conditions experienced during winter affected subsequent reproduction more strongly in females than in males, although both sexes similarly responded to this environmental cue. Individuals of different age responded differently to conditions experienced in winter quarters. In general, there was not a clear response of yearlings to conditions experienced during winter, whereas the response gradually increased as individuals aged. Additionally, our results were not confounded by other variables known to affect breeding performance, such as body mass and tail length. Finally, the analyses performed were not biased by the spatial scale used to define winter areas.
There is increasing evidence showing that environmental conditions experienced during winter are related to subsequent spring migration, breeding phenology or reproductive success of migratory birds5,7,9,11,12,17,18,20,38. Such carry-over effects need to be fully understood because they have important consequences for population dynamics and evolutionary processes18. We found that barn swallows advanced breeding date in response to favorable winter conditions, and that individuals responded differently depending on their age. In agreement with our findings, barn swallows breeding in northern Italy have also been shown to advance arrival and breeding in response to good winter conditions when studied at the population level5,38, but the opposite outcome was found at the individual level18. These authors argued that when studied at the population level, individuals on average aquired body condion earlier during good winter years (i.e. high NDVI) and should depart sooner from winter areas or perform faster migration than during poor winter years (i.e. low NDVI). However, when studied at the individual level, individuals may experience different winter conditions within years and those individuals that experienced poor winter conditions (i.e. low NDVI) should depart earlier from winter areas because of seasonal deterioration of winter conditions taking place within years18. Therefore, both spatial and temporal variation in winter conditions could influence carry-over effects between the wintering and breeding stage. Hence, the outcome could differ when the same population of the same species is studied either at the individual or population level.
Ecological conditions from the wintering grounds usually affect subsequent breeding success more strongly in females than in males9,17,18. Specifically, carry-over effects from winter conditions on timing of breeding and reproductive success were found for female but not for male barn swallows18. We found the same carry-over effects operating in both sexes, but in agreement with the latter study, the response was stronger for females. This difference found between sexes could be explained because females may be more constrained than males in the time needed to start breeding after arrival to their breeding grounds. Along this line, the variance in time elapsed between arrival and breeding was smaller for females compared to males18. Alternatively, the acquisition of an appropriate physical condition after arrival to breeding areas could be more important to determine breeding date for females than for males, because females must meet the physiological demands associated with egg production. Otherwise, males are more prone to arrive earlier than females to breeding grounds because males that arrive earlier usually mate sooner and have greater reproductive success24. Because we lacked arrival date in our path analyses, we could not make many inference about the causes of small difference we found in carry over effects between sexes.
It is well known that avian migratory behavior depend on the age of individuals1,22. The carry-over effects we found in this study should have been mediated by the effects of different winter conditions on migratory behavior such as departure date from winter areas, flight speed on route, number and duration of stopovers and arrival date to breeding areas. Most previous studies investigating migratory behavior and carry-over effects5,7,9,11,20 have discriminated between two age classes (i.e. juveniles vs adults) and consequently age-related migratory behavior based on actual known age marked-individuals has been reported less frequently (but see Sergio and colaborators10). Difference amongst age classes in migratory behavior is hence a less studied phenomenon49,50,51.
We found different responses to winter conditions with the age of individuals. Our results support the constraint hypothesis and reject the competiton hypotheses to explain age-related differences in timing of reproduction and breeding success. We found the highest difference in reproductive traits across age when winter conditions were favorable. If the competiton hypothesis was operating in our study system and if older individuals were dominant over young ones, greater difference in reproductive traits across age were expected after experiencing poor winter conditions. This is because competition among individuals should be exacerbated under low resource availability. However, we found the opposite, and we hypothesize that the higher reproductive success of older individuals compared to yearlings could be due to swallows acquiring competence via experience. For instance, gaining experience or skills could help older individuals to select better quality wintering habitats or make better use of them than yearlings27. Our findings are consistent with another study performed with barn swallows in southern Europe, in which experience explained age-related difference in spring migration and reproductive success6. However, we acknowledge that we cannot discard the selection hypothesis as a possible explanation in our case.
Our results have implications for conservation and management of migratory insectivorous birds which are declining52,53,54. In this study, we have linked reproductive success of barn swallows breeding in north Europe with environmental changes occurring in the African wintering areas. Furthermore, we have identified that NDVI from savannahs was the main driver of the general trend in NDVI found for all wintering areas. We suggest that the dramatic decline in NDVI we found for woody savannahs could be buffered by the increase in equatorial broadleaf forests. These different trends in ecological conditions within the wintering areas may help to predict how barn swallows could respond to current global change. Because environmental conditions have deteriorated in savannahs and improved in rainforests, barn swallows may now encounter more suitable habitats at the latter, which is in agreement with the general northwards shift of their wintering areas55. Because African savannahs constitute a highly seasonal environment with periods of heavy rain and periods of severe drought, we can assume that greater rainfall across the wintering areas should be associated with higher values of average NDVI during winter34, and furthermore both NDVI and rainfall are positively correlated with adundance of flying insects31,33,35,36. Indeed, previous studies have generally assumed that higher NDVI during winter imply favorable ecological conditions for aerial insectivores to overwinter5,7,37,38,39. Temperature is predicted to increase across Africa, while precipitation is predicted to increase in central Africa and decrease in south Africa28. Considering that the correlation of NDVI with these climatic variables differs spatially throughout the winter areas34, further complex analyses would be necessary to predict trends in NDVI for these winter areas. Because NDVI has been established as a crucial tool for assessing the effects of climate change on organisms30, and specifically migratory birds5,38, our results are also appropriate for understanding and predicting the potential adaptation to climate change of organisms throughout their life spans.
Summarizing, we identified winter areas of barn swallows breeding in Denmark by combining a multi-isotope assignment with prior information from ringing data. We have also determined, using satellite-derived NDVI values, that environmental conditions at the African wintering grounds have deteriorated in recent years. In addition, we have shown that changes in NDVI during winter were related to subsequent reproductive parameters, and that age and sex determined the strength of these carry-over effects. The increasing collection of long-term individual data could shed light on the ability to predict future adaptation of long-distance migrants to current climate change. We highlight that our findings are compiling evidence of the great importance of age on migratory ecology, with relevant fitness consequences.
We monitored 40 breeding colonies of barn swallows at Kraghede (57.22°N, 9.97°E), Denmark, during 1984–2013. This 45 km2 study area consists of scattered farms and houses interspersed by meadows and fields where the main crops are wheat and potatoes. APM captured barn swallows with mist nets at least once weekly throughout the breeding season, assuring that more than 95% of all adults are captured and measured annually. All individuals were provided with both metallic and colour rings. We assigned an age of 1 year to all barn swallows when captured the first time as adults. This assumption is supported because breeding dispersal is negligible in the study species. Only three of more than 5000 adults ever moved from one breeding colony to another, and then only a short distance of <750 m24. Therefore, we assumed that adult disappearance from the breeding population indicated mortality rather than dispersal5,7. Moreover, age assignment was also supported because age at first reproduction was at the age of one year for 100% of recruits. Fieldwork showed that from 315 recruits, all were captured for the first time as an adult when one year old, implying that no recruit was captured the first time when aged two or more years. Therefore, because capture effort was high every year, all recruits without metal rings should be yearlings coming from outside the study sites.
All captured birds were weighed with a Pesola spring balance to the nearest 0.1 g, and tail length was measured to the nearest mm with a ruler. We estimated breeding date of females as the date when they laid the first egg during the first brood, whereas breeding date of males was the date their mate’s laid their first egg. Laying date for females was assigned during weekly visits to nests assuming that one egg was laid daily. Males were assigned to their female patners by regular observations from a hide of colour ringed swallows at their nests using binoculars (8 × 30 Zeiss). We estimated reproductive success as the number of eggs laid and the number of fledglings raised during the first breeding attempt, because we expected that carry-over effects from the winter period would be easily detected during the first part of the breeding stage. Reproductive success at the first brood correlated well with total reproductive success (i.e. including subsequent broods), and so is a good proxy for breeding performance in this multi-brooded species24.
During the breeding seasons 2000 and 2016, we collected respectively 84 and 119 feather samples for later stable isotope analyses to identify winter origins. Feather samples from 2000 were taken from the red throat badge, whereas feather samples from 2016 were taken from the rectrix (i.e. outermost feather). Barn swallows usually moult their feathers in the African winter areas56,57, and furthermore they moult feathers continuously during almost the whole wintering stay (A. P. Møller, unpublished data from South Africa, Namibia and Ghana). Because feather keratin is metabolically inert after synthesis58, we were confident that our feather samples represent the isotopic values derived from the winter areas.
Stable isotope analysis
All feathers were cleaned of surface oils in 2:1 chloroform:methanol solvent rinse and prepared for δ2H, δ13C and δ15N analysis. Deuterium abundance in the non-exchangeable hydrogen of feathers was determined following standard procedures59 and using three calibrated keratin hydrogen-isotope reference materials (CBS = −197‰; SPK = −121.6‰; KHS = −54.1‰). Deuterium measurement was performed on H2 gas derived from high-temperature (1350 °C) flash pyrolysis of 350 ± 10 µg feather subsamples and keratin standards. Measurement of the three keratin laboratory reference materials, corrected for linear instrumental drift, were both accurate and precise with typical within-run (n = 5) SD values of <2‰. On the other hand, δ13C and δ15N measurements were performed on CO2 and N2 gases resulted from combustion of 0.5–1.0 feather material. Using previously calibrated internal laboratory C and N standards [powdered keratin (BWBIII; δ13C = −20‰; δ15N = 14.4‰) and gelatin (PUGEL; δ13C = −13.6‰; δ15N = 4.73‰)], within run (n = 5) precisions for δ15N and δ13C measurements were ~ ± 0.15‰. Stable isotope ratios are reported in standard delta (δ) notation relative to VSMOW-SLAP (Vienna Standard Mean Ocean Water – Standard Light Antarctic Precipitation) for 2H/H, VPDB (Vienna Pee Dee Belemnite carbonate) for 13C/12C and atmospheric nitrogen for 15N/14N.
Combination of ringing recoveries and isotope analysis to identify winter areas
In June 2018, we requested access to all ringing recoveries of Palearctic barn swallows from The European Union for Bird Ringing (EURING). Every ringing scheme, except for the Copenhagen Bird Ringing Centre, agreed to share their data. We originally obtained a sample 2333 individuals that were ringed and recaptured across the barn swallow range60. We selected both kinds of recoveries of birds that were either dead or alive. We only retained records of birds captured either in the European breeding areas (from 36° to 66°N and from 9°W to 42°E) during May-August, or in the sub-Saharan wintering range (from 15°N to 35°S and from 17°W to 42°E) during November-February. Finally, because directions followed during migration depend on the longitude of breeding location, we filtered our data subset leaving only individuals whose breeding location was between 5°E and 15°E (i.e. a buffer zone in longitude around Denmark). Following this data filtering, our final sample size for ringing data was 253 individuals ringed in Europe and recaptured in sub-Saharan Africa (or vice versa).
It is well known that ringing effort is very uneven across Africa, being biased to regions of higher density of wintering birds or to more developed countries40. This inherent drawback of the banding method for studying migration can be solved by including other sources of information. For instance, ringing recoveries in South Africa indicates that barn swallows breeding in Denmark usually winter in the central and eastern parts of the country, but NDVI correlation analyses point to western South Africa as the main wintering area39. This is because ringing effort has been higher at the largest wintering roosts, which are distributed across the centre and east of South Africa, whereas western roosts are smaller and more variably distributed in space and time. Therefore, we considered that combining ringing recoveries with stable isotope analyses would overcome the strong bias in ringing effort; but at the same time, isotope assignment should be constrained longitudinally thanks to ringing recoveries and so the potential moult areas should be identified with higher precision.
To incorporate ringing recoveries within the probabilistic assignment of origins based on istotope analysis, we followed the method described by Van Wilgenburg & Hobson61. From our final subset of ringing recoveries (Fig. 1D), first we calculated bearings from breeding to wintering location. Second, we extracted the mean (circular) direction and “kappa” (the concentration parameter, an inverse measure of dispersion). Then, using these parameters, we fit a von Mises probability density function to predict the likelihood of any possible direction (from 1° to 359° separating by 2°). These probability densities were normalized with respect to the sum and were used to create a raster (fm) that was used as a prior probability surface (Fig. 1F).
We evaluated the likelihood of origin for a given sample feather by employing a multivariate normal probability density function62,63. By applying Bayes’ Rule61, we incorporated the prior probability surface (fm) in assessing the posterior probabilities of origin (fx) based on feather δ2H, δ13C and δ15N. Geographic locations that were consistent with the upper 67% of the spatially explicit probability density maps for each individual (fx) were coded as 1; all others were coded as 064. Thus, for each individual being assigned to a moult origin we obtained one binary map, and we depicted the population likely origin by summing over all individual binary surfaces (Fig. 1G). Finally, to select the most-probable wintering areas used by our study populations, we clipped the population likely origin surface at three different threshold values, leaving only pixels that were assigned to be the moult origin for >50, >60 and >70% of our sample (Fig. 1H). We used this cutoff criteria to extract NDVI values at three different spatial scales.
Environmental conditions from wintering areas
We extracted NDVI values for the wintering areas previously identified using the public dataset developed by ECOCAST (available at https://ecocast.arc.nasa.gov/data/pub/gimms/3g.v0/). This data base provided bimonthly NDVI measures at 0.83° resolution, during the period 1981–2013. We used as a proxy of environmental conditions during winter of a given year “i”, the averaged NDVI measures from November (year “i-1”) to March (year “i”). The selection of this time window (November-March) was chosen because this is the average period that barn swallows breeding in Denmark spend wintering in sub-Saharan Africa (A. P. Møller, unpublished data). It has been found that winter areas in southern Africa have moved northwards during 1912–2008 at a rate of 9 km/yr for the Palearctic barn swallow55. Consequently, during our study (1984–2013), barn swallow wintering areas may have shifted approximately 290 km northwards.
We were also interested in differences across sub-regions within the identified wintering area. For this purpose, we clipped our wintering area with a land cover classification layer46, thus obtaining average NDVI values for each African biome. We used the land cover classification layer from 2011 at 0.1° resolution (available at http://neo.sci.gsfc.nasa.gov/view.php?datasetId=MCD12C1_T1). Because we have inferred the wintering areas of a large population (~ 4,000 individuals) from moult origins given by a much smaller subset (~ 200 sample feathers), in order to test for subsequent fitness carry-over effects we need to use the average NDVI across the whole wintering area. Nevertheless, determining differences in environmental trends within this defined area should provide complementary information to better interpret our results.
To investigate how NDVI from wintering areas may influence subsequent breeding date and reproductive success, we used a structural equation modelling approach. SEMs are probabilistic models that hypothesize a causal network with multiple variables that can appear as both predictor and response variables71. We used “confirmatory path analysis” or piecewise SEM, based on applications from graph theory48. In piecewise SEM, the causal network is translated to a set of linear equations (e.g. linear mixed models), which are then evaluated individually thus allowing a wide range of distributions and sampling designs. In addition, the goodness of fit of the entire causal network can be quantified by a directed separation test (”d-separation test”), which tests the assumption that all variables are conditionally independent (i.e. that there are no missing relationships among unconnected variables).
We conducted separated confirmatory path analyses for males and females. In this way, different responses across sex are indicated by differences in path coefficients from both causal models. We included eight variables in the model built for females: NDVI from winter areas, age, the interaction term between age and NDVI, tail length, body mass, breeding date, clutch size and number of fledglings in the first brood. For males, we excluded clutch size from the causal model, because this variable should be determined by the female only. Reproductive parameters were considered dependent (endogenous) variables, while the rest were considered independent (exogenous). Every observed variable included in our models had the same sample size (i.e. listwise deletion).
The path analysis built for females was made up of three linear mixed models, and that for males was made up of two linear mixed models (i.e. one for each dependent variable). The structure of the path analyses was designed based on previous knowledge of migratory ecology17,20. The first linear model in both path analyses, tested the effects of NDVI from winter areas, age and their interaction term on breeding date, while controlling for the confounding effect of tail length23,24. The second linear model for females, tested the effect of breeding date on clutch size while controlling for age and tail length. The third linear model for females tested the effect of clutch size on number of fledglings while controlling for age and tail length. The second linear model for males tested the effect of breeding date on number of fledglings while controlling for age and tail length. Individual identity (ring), breeding colony and year were included as cross-random intercepts in all linear models, thus controlling for repeated measures in the dependent variable taken on the same individual, and also for among breeding site and inter-annual variation. We used normal distributions of errors and the identity link function for all response variables. Residuals of each linear model were visually inspected for deviation from normality using normal qqplots. Finally, we also controlled for the confounding effect of body mass on each reproductive parameter8. Our dataset contained measures of body mass before, after or during the first breeding attempt. Thus, in our full dataset, we do not know the direction of causality between body mass and the reproductive parameters. Consequently, to maintain a high sample size, we assumed correlated errors (i.e. a relationship that is bidirectional and assumed to be caused by a shared underlying driver) between body mass and every reproductive parameter.
Finally, we repeated our analyses using the different spatial scales defining wintering areas to determine if spatial scale could influence the results obtained. Thus, we conducted the above described analyses for NDVI extracted at different sizes of wintering areas, as defined by pixels assigned to be the origin for >50, >60 or >70% of our sample. All these analyses performed at different spatial scales provided the same results. Thus, for simplicity, we only present results for NDVI extracted from wintering areas defined as pixels assigned to be the origin for >60% of our sample.
Field sampling was conducted in agreement with the national Danish hunting law. Permit for capture and release of birds was obtained from the Danish Ministry of Agriculture. All efforts were made to ameliorate suffering of animals and minimize handling time according to Guidelines to the Use of Wild Birds in Research (J. Fair, E. Paul, and J. Jones, Eds. 2010. Washington, D.C. Ornithological Council).
The dataset analysed during the current study is available from the corresponding author on reasonable request.
Newton, I. The migration ecology of birds. (Academic Press: London, UK, 2008).
O’Connor, C. M., Norris, D. R., Crossin, G. T. & Cooke, S. J. Biological carryover effects: linking common concepts and mechanisms in ecology and evolution. Ecosphere 5, 28 (2014).
Norris, D. R. & Marra, P. P. Seasonal interactions, habitat quality, and population dynamics in migratory birds. The Condor 109, 535 (2007).
Harrison, X. A., Blount, J. D., Inger, R., Norris, D. R. & Bearhop, S. Carry-over effects as drivers of fitness differences in animals. J. Anim. Ecol. 80, 4–18 (2011).
Saino, N. et al. Ecological conditions during winter predict arrival date at the breeding quarters in a trans-Saharan migratory bird. Ecol. Letters 7, 21–25 (2004).
Balbontín, J. et al. Age-related change in breeding performance in early life is associated with an increase in competence in the migratory barn swallow Hirundo rustica. J. Anim. Ecol. 76, 915–925 (2007).
Balbontín, J. et al. Individual responses in spring arrival date to ecological conditions during winter and migration in a migratory bird: phenotypic plasticity in arrival date and climate. J. Anim. Ecol. 78, 981–989 (2009).
Balbontín, J. et al. Lifetime individual plasticity in body condition of a migratory bird. Biol. J. Linn. Soc. 105, 420–434 (2012).
Drake, A., Rock, C., Quinlan, S. P. & Green, D. J. Carry-over effects of winter habitat vary with age and sex in yellow warblers Setophaga petechia. J. Avian Biol. 44, 321–330 (2013).
Sergio, F. et al. Individual improvements and selective mortality shape lifelong migratory performance. Nature 515, 410–413 (2014).
López-Calderón, C. et al. Wintering areas predict age-related breeding phenology in a migratory passerine bird. J. Avian Biol. 48, 631–639 (2017).
Marra, P. P., Hobson, K. A. & Holmes, R. T. Linking winter and summer events in a migratory bird by using stable-carbon isotopes. Science 282, 1884–1886 (1998).
Studds, C. E. & Marra, P. P. Nonbreeding habitat occupancy and population processes: an upgrade experiment with a migratory bird. Ecology 86, 2380–2385 (2005).
Winkler, D. W. & Allen, P. E. The seasonal decline in tree swallow clutch size: physiological constraint or strategic adjustment? Ecology 77, 922–932 (1996).
Smith, R. J. & Moore, F. R. Arrival timing and seasonal reproductive performance in a long-distance migratory landbird. Behav. Ecol. Sociobiol. 57, 231–239 (2004).
Marzal, A. et al. Malaria infection and feather growth rate predict reproductive success in house martins. Oecologia 171, 853–861 (2013).
Norris, D. R., Marra, P. P., Kyser, T. K., Sherry, T. W. & Ratcliffe, L. M. Tropical winter habitat limits reproductive success on the temperate breeding grounds in a migratory bird. Proc. Royal Soc. Lond. 271, 59–64 (2004).
Saino, N. et al. Sex-dependent carry-over effects on timing of reproduction and fecundity of a migratory bird. J. Anim. Ecol. 86, 239–249 (2017).
Saino, N. et al. Wing morphology, winter ecology, and fecundity selection: evidence for sex-dependence in barn swallows (Hirundo rustica). Oecologia 184, 799–812 (2017).
López-Calderón, C. et al. Environmental conditions during winter predict age- and sex-specific differences in reproductive success of a trans-Saharan migratory bird. Sci. Rep. 7, 18082 (2017).
Briedis, M., Krist, M., Král, M., Voigt, C. C. & Adamík, P. Linking events throughout the annual cycle in a migratory bird—non-breeding period buffers accumulation of carry-over effects. Behav. Ecol. Sociobiol. 72, 93 (2018).
Cristol, D. C., Baker, M. B. & Carbone, C. Differential migration revisited: latitudinal segregation by age and sex class. Current Ornitholy 15, 33–87 (1999).
Møller, A. P. Sexual selection in the monogamous barn swallow (Hirundo rustica). II. Mechanisms of sexual selection. J. Evol. Biol. 5, 603–624 (1992).
Møller, A. P. Sexual selection and the barn swallow. (Oxford University Press, 1994).
Gauthreaux, S. A. The ecological significance of behavioural dominance in Perspectives in Ornithology, Vol. 3. (eds Bateson, P. P. G. & Klopfer, P. H.) 17–54 (New York, Plenum Press 1978).
Curio, E. Why do young birds reproduce less well? Ibis 125, 400–404 (1983).
Forslund, P. & Pärt, T. Age and reproduction in birds—hypotheses and tests. TREE 10, 374–378 (1995).
IPCC. Climate Change 2013: The Physical Science Basis. Contribution of Working Group I to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change (Cambridge University Press, 2013).
Pettorelli, N. et al. Using the satellite-derived NDVI to assess ecological responses to environmental change. TREE 20, 503–510 (2005).
Pettorelli, N. et al. The Normalized Difference Vegetation Index (NDVI): unforeseen successes in animal ecology. Clim. Res. 46, 15–27 (2011).
Lassau, S. A. & Hochuli, D. F. Testing predictions of beetle community patterns derived empirically using remote sensing. Divers. Distributions 14, 138–147 (2008).
Lourenço, P. M. et al. Anopheles atroparvus density modeling using MODIS NDVI in a former malarious area in Portugal. J. Vector Ecol. 36, 279–291 (2011).
Roiz, D., Ruiz, S., Soriguer, R. & Figuerola, J. Landscape effects on the presence, abundance and diversity of mosquitoes in mediterranean wetlands. PLoS One 10, e0128112 (2015).
Ichii, K., Kawabata, A. & Yamaguchi, Y. Global correlation analysis for NDVI and climatic variables and NDVI trends: 1982–1990. Int. J. Remote Sens. 23, 3873–3878 (2002).
Denlinger, D. L. Seasonal and annual variation of insect abundance in the Nairobi National Park, Kenya. Biotropica 12, 100–106 (1980).
Cumming, G. S. & Bernard, R. T. F. Rainfall, food abundance and timing of parturition in African bats. Oecologia 111, 309–317 (1997).
Balbontín, J. et al. Divergent patterns of impact of environmental conditions on life history traits in two populations of a long-distance migratory bird. Oecologia 159, 859–872 (2009).
Saino, N., Szép, T., Ambrosini, R., Romano, M. & Møller, A. P. Ecological conditions during winter affect sexual selection and breeding in a migratory bird. Proc. Royal Soc. Lond 271, 681 (2004).
Szép, T. et al. Searching for potential wintering and migration areas of a Danish Barn Swallow population in South Africa by correlating NDVI with survival estimates. J Ornithol 147, 245–253 (2006).
Ambrosini, R., Møller, A. P. & Saino, N. A quantitative measure of migratory connectivity. J. Theor. Biol. 257, 203–211 (2009).
García-Pérez, B., Hobson, K. A., Powell, R. L., Still, C. J. & Huber, G. H. Switching Hemispheres: A New Migration Strategy for the Disjunct Argentinean Breeding Population of Barn Swallow (Hirundo rustica). PLoS ONE 8, e55654 (2013).
Szép, T., Liechti, F., Nagy, K., Nagy, Z. & Hahn, S. Discovering the migration and non-breeding areas of sand martins and house martins breeding in the Pannonian basin (central-eastern Europe). J. Avian Biol. 48, 114–122 (2017).
Cooper, N. W., Hallworth, M. T. & Marra, P. P. Light-level geolocation reveals wintering distribution, migration routes, and primary stopover locations of an endangered long-distance migratory songbird. J. Avian Biol. 48, 209–219 (2017).
Coppack, T. et al. Can long-distance migratory birds adjust to the advancement of spring by shortening migration distance? The response of the pied flycatcher to latitudinal photoperiodic variation. Glob. Change Biol. 14, 2516–2522 (2008).
Liechti, F. et al. Timing of migration and residence areas during the nonbreeding period of barn swallows Hirundo rustica in relation to sex and population. J. Avian Biol. 46, 254–265 (2015).
Allen, J. Land cover classification (1 year). NASA Earth Observations, http://neo.sci.gsfc.nasa.gov/view.php?datasetId=MCD12C1_T1 (2016).
Bønløkke, J. et al. The Danish Bird Migration Atlas. (Rhodos Press, Humlebæk & Zoological Museum, University of Copenhagen, 2006).
Shipley, B. Cause and correlation in biology: a user’s guide to path analysis, structural equations and causal inference with R. (Cambridge University Press, 2016).
Gromadzka, J. & Serra, L. Differential migration of juvenile and adult grey plovers Pluvialis squatarola at the mouth of the Vistula River, Poland. Ornis Fennica 75, 193–199 (1998).
Yosef, R. L. et al. Differential spring migration of adult and juvenile Levant Sparrowhawks (Accipiter brevipes) through Eilat, Israel. J. Raptor Res. 37, 31–36 (2003).
Markovets, M. L., Zduniak, P. & Yosef, R. Differential sex- and age-related migration of Bluethroats Luscinia svecica at Eilat, Israel. Sci. Nat. 95, 655–661 (2008).
Sanderson, F. J., Donald, P. F., Pain, D. J., Burfield, I. J. & Van Bommel, F. P. Long-term population declines in Afro-Palearctic migrant birds. Biol.Cons. 131, 93–105 (2006).
Nebel, S., Mills, A., McCracken, J. D. & Taylor, P. D. Declines of aerial insectivores in North America follow a geographic gradient. Avian Conserv. Ecol. 5 (2010).
BirdLife International. Hirundo rustica. The IUCN Red List of Threatened Species, http://www.iucnredlist.org/details/22712252/0 (2016).
Ambrosini, R. et al. Climate change and the long-term northward shift in the African wintering range of the barn swallow Hirundo rustica. Clim. Res. 49, 131–141 (2011).
Cramp, S. The birds of the Western Palearctic. Vol. 5. (Oxford University Press, 1988).
Jenni, L. & Winkler, R. Moult and ageing of European passerines. (London, UK: Academic Press, 1994).
Hobson, K. A. Tracing origins and migration of wildlife using stable isotopes: a review. Oecologia 120, 314–326 (1999).
Wassenaar, L. I. & Hobson, K. A. Comparative equilibration and online technique for determination of non-exchangeable hydrogen of keratins for use in animal migration studies. Isot. Environ. Health S. 39, 211–217 (2003).
du Feu, C. R. et al. EURING Data Bank geographical index 2009, http://www.euring.org/edb (2009).
Van Wilgenburg, S. L. & Hobson, K. A. Combining stable-isotope (δ D) and band recovery data to improve probabilistic assignment of migratory birds to origin. Ecol. Appl. 21, 1340–1351 (2011).
Royle, J. A. & Rubenstein, D. R. The role of species abundance in determining breeding origins of migratory birds with stable isotopes. Ecol. Appl. 14, 1780–1788 (2004).
Hobson, K. A. et al. A multi-isotope (δ 2H, δ 13C, δ 15N) approach to establishing migratory connectivity in Palearctic-Afrotropical migrants: an example using Wood Warblers Phylloscopus sibilatrix. Acta Ornithol. 49, 57–69 (2014).
Hobson, K. A., Wunder, M. B., Van Wilgenburg, S. L., Clark, R. G. & Wassenaar, L. I. A method for investigating population declines of migratory birds using stable isotopes: origins of harvested lesser scaup in North America. PLoS ONE 4, e7915 (2009).
R Core Team. The R Project for Statistical Computing. R Foundation for Statistical Computing, http://www.R-project.org (2016).
Hijmans, R. J. Raster: Geographic Data Analysis and Modeling. R package version 2.5-8, https://CRAN.R-project.org/package=raster (2016).
Hijmans, R. J. Geosphere: Spherical Trigonometry. R package version 1.5-5, https://CRAN.R-project.org/package=geosphere (2016).
Sumner, M. D. graticule: Meridional and Parallel Lines for Maps. R package version 0.1.2, https://CRAN.R-project.org/package=graticule (2016).
Lund, U., & Agostinelli, C. CircStats: Circular Statistics, from “Topics in circular Statistics” (2001). R package version 0.2-4, https://CRAN.R-project.org/package=CircStats (2012).
Gross, K. & Bates, D. Mvnmle: ML estimation for multivariate normal data with missing values. R package version 0.1-11, https://CRAN.R-project.org/package=mvnmle (2012).
Lefcheck, J. S. PiecewiseSEM: Piecewise structural equation modelling in R for ecology, evolution, and systematics. MEE 7, 573–579 (2016).
Bates, D., Mächler, M., Bolker, B. & Walker, S. Fitting Linear Mixed-Effects Models using lme4. J. Stat. Soft. 67, 1–48 (2015).
Hobson, K. A. et al. A multi-isotope (δ 13C, δ 15N, δ 2H) feather isoscape to assign Afrotropical migrant birds to origins. Ecosphere 3, 44 (2012).
We are grateful to the European Union for Bird Ringing (EURING), which made the recovery data available through the EURING Data Bank and to the many ringers and ringing scheme staff who have gathered and prepared the data. This study was funded by Junta de Andalucía (research project: P12-RNM-2144). The corresponding author was supported by an operating grant from Junta de Andalucía (P12-RNM-2144). The progam “Erasmus+” provided additional funding necessary for the corresponding author to develop a research stage. We sincerely thank farm owners and volunteers that helped in this long-term field study. Finally, we thank Steven Van Wilgenburg for his comments regarding probabilistic assignment of origins.
The authors declare no competing interests.
Publisher’s note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
About this article
Cite this article
López Calderón, C., Balbontín Arenas, J., Hobson, K.A. et al. Age-dependent carry-over effects in a long-distance migratory bird. Sci Rep 9, 12032 (2019) doi:10.1038/s41598-019-47374-3