Multidimensional natal isotopic niches reflect migratory patterns in birds

Naturally occurring stable isotope ratios in animal tissues allow estimation of species trophic position and ecological niche. Measuring multiple isotopes of migratory species along flyway bottlenecks offers the opportunity to sample multiple populations and species whose tissues carry information at continental scales. We measured δ2H, δ18O, δ13C, δ15N in juvenile feathers of 21 bird species captured at a migratory bottleneck in the Italian Alps. We examined if trends in individual isotopes reflected known migratory strategies and whether dietary (δ13C–δ15N) and spatially-explicit breeding origin (δ2H–δ18O) niche breadth (NB) differed among long-distance trans-Saharan (TS), short-distance (IP) and irruptive (IR) intra-Palearctic migrants, and whether they correlated with reported populations long-term trends. In both TS and IP groups, species δ2H declined with capture date, indicating that northern populations reached the stopover site later in the season, following a Type-I migration strategy. Values of δ2H indicated that breeding range of TS migrants extended farther north than IP and IR migrants. The breeding season was longer for IP migrants whose δ13C and δ15N values declined and increased, respectively, with time of capture. Average species dietary NB did not differ among migratory groups, but TS migrants displayed wider breeding origin niches, suggesting that long-distant migration is linked to broader ecological niches. Isotope origin NB well reflected species geographic range extent, while dietary NB did not correlate with literature accounts of species’ diet. We found no relationship between species breeding NB and population trends in Europe, suggesting that conditions in the breeding grounds, as inferred by stable isotopes, are not the only determinant of species’ long-term persistence. We demonstrate that ringing activities and isotopic measurements of passerines migrating through a bottleneck represents a unique opportunity to investigate large-scale life-history phenomena relevant to conservation.

). TSM Trans-Saharan migrants, IPM Intra-Palearctic migrants, IRM Irruptive migrants. www.nature.com/scientificreports/ and 2. Migratory phenology of each species during the post-breeding passage is shown in Fig. S1 (Supplementary Material), where the temporal patterns in the number of species captured is also shown. Temporal changes in δ 2 H (reflecting species breeding latitude) supported the hypothesis that both TSM and IPM groups display a Type-I (chain) migration, whereby northern populations (with lower δ 2 H values) reached the Alpine passage later in the season (Fig. 3). This pattern appeared consistent among species in each group, with a significant effect of capture day in the lme model (p < 0.01 for both groups). Mean δ 2 H values were evidently lower for TSM, indicating that species in this migratory group tend to breed at higher latitudes than IPM. No temporal trend in δ 2 H was observed in IRM species.
A trend in δ 15 N was evident for IPM and IRM species with a positive effect of capture day (lme p < 0.01). This trend was not observed in TSM, which displayed slightly higher mean δ 15 N values (Fig. 3). Similarly, a significant negative trend in δ 13 C was observed in IPM (p < 0.01), which was not significant in either TSM and IRM species. Figure 3 also shows that most TSM species reached the capture site by day 280, whereas the season appeared longer for IPM and IRM. The difference in phenology is also clear from Fig. S1.
Average species dietary isotopic NB did not differ significantly among migratory groups (Fig. 4). Conversely, the average isotopic origin niche was significantly broader for TSM, about 30% broader than IPM (GLS model; p = 0.03). However, there was relatively large variation in origin NB within TSM. IRM displayed a much narrower origin NB (Fig. 4). Excluding the apparent outliers from the analysis of dietary NB (δ 13  www.nature.com/scientificreports/ In addition, migration distance was related to the breadth of species origin niche, although there was considerable variation at larger distances ( Fig. 5; GLS; p = 0.01). No relationship was observed between dietary isotopic NB and migration distance (Fig. S2).
We compared our estimates of species dietary and origin isotopic niche with those recently derived from species distributional data. We found that the origin NB well matched measures from range maps (Fig. 6), with a positive linear correlation (p = 0.02). In addition, climate niche breadth derived from range maps well reflected the upper distribution of the empirical isotope origin niche. This is evident from the significant quantile regression at q = 0.8 (p < 0.01). Conversely, the dietary isotopic NB was not related to diet specialisation, as derived from the literature.
Finally, we appraised whether long-term population trends were related to the empirical isotope niches. Neither dietary nor origin isotopic niche breadth were related to species population trends as reported in the PECBMS portal (Fig. S3).  www.nature.com/scientificreports/

Discussion
We used juvenile feathers of multiple bird species intercepted along an autumn migratory flyway as 'remote carriers' of isotope niche information from Western Palearctic breeding areas. This allowed quantifying niche parameters from species that evolved different migratory strategies. In addition, despite potential confounding factors related to variance in local isotopic baselines, including individuals from different populations and communities allowed the quantification of species niches that were less influenced by local biotic interactions (e.g. competition), and likely more representative of the realised niche dimensions of the species. Our results generally supported our initial hypotheses, but also revealed a poor association between isotope niche dimensions and literature-derived niche parameters and long-term population trends.
Temporal trends in feather δ 2 H indicated that individuals breeding at higher latitudes (displaying lower δ 2 H values) reached the sampling locations later in the season, with patterns that appeared consistent among species and across the two major migratory groups (TSM, IPM). This type of migration, often called Type I (as opposed to Type-II, whereby northern populations leap-frog southern populations during migration), has also been observed across a range of Nearctic-Neotropical migrants 9,38 , and appears common in Afro-Palearctic migrants as well 39,40 . Different migration timing of conspecific populations contributes to limit competition among individuals and likely occurs at the species level too. Assessing temporal segregation in the use of migratory stopover helps identify the species most likely to co-occur and compete locally for resources. In our study site, a few species showed complete segregation (Fig. S1), but high overlap was otherwise evident with up to nine migratory species captured in a single day. Bottlenecks can indeed concentrate large numbers of individuals and species during a short period, which not only heightens intra-and inter-specific competition, but also increases vulnerability to localised events and disturbance. The identification and conservation of migratory stopover and bottlenecks are therefore key for the persistence of many declining populations [7][8][9] .
Temporal trends observed in both δ 15 N and δ 13 C in IPM could be explained by different non-mutually exclusive mechanisms. For instance, the lower latitude breeding range of IPM, as well as IRM (compared to TSM), implies that early captures were dominated by individuals from the Alpine/central-Europe/Carpathian areas at high elevations. This could be associated with higher δ 13 C 19 also due to lower precipitation compared to northern Europe 41 . Similarly, it is also possible that the extent of anthropogenic nitrogen input from agriculture is lower in mountainous regions, which could contribute to lower δ 15 N in early captures. It is known, in fact, that anthropogenic sources such as fertilisers can alter and mostly increase nitrogen isotopic baseline values 23 .
Alternatively, seasonally rising values of δ 15 N could be associated with the increased availability of adult insects (as opposed to larvae or other arthropods with lower δ 15 N values) as the longer breeding season of IPM progresses. Conversely, the shorter season of TSM and their strictly insectivorous diet would result in more seasonally stable δ 15 N values.
We found that species breeding origin niche was, on average, broader in long-distance trans-Saharan compared to intra-Palearctic migrants. Long-distance migration is generally thought to have evolved from ancestor species with broad habitat niches: individuals able to exploit a wider range of conditions during migration or www.nature.com/scientificreports/ at wintering grounds should be selectively advantaged over more specialised ones [33][34][35] . Empirical support for this hypothesis is limited 36,42,43 , and our data thus provide some additional evidence. However, the link between niche breadth and migration patterns remains unclear, with some evidence suggesting the opposite view that narrower habitat or diet niche may have favoured long-distance migrations 44 , because specialists are forced to leave an area when conditions become unfavourable or resources scarce. Yet, our data do not show any correlation between migration distance and dietary niche breadth, providing no support for this alternative hypothesis. It has to be kept in mind, however, that our estimates of ecological niches reflect the breeding diets of species (i.e. isotopes from juvenile feathers grown at nest), and the inclusion of information from wintering grounds would have provided a more complete and likely different picture. This would require measuring isotope ratios in body tissues with different turnover or growth rates 45 , or collecting feathers grown in winter 29 . However, for most passerine species, information on wintering locations is still limited, especially for the Afro-Palearctic migration system 46,47 . Similarly, the extent to which migratory species track their climate and habitat niche between breeding and wintering grounds is poorly understood and likely to vary among species 37,47 . Most assessments of seasonal niche shifts during migration relied on large-scale climatic and land-use data 37,48 , while those employing tissue isotopes are rare and mostly focussed on a few key species to appraise migratory connectivity 10,45,49 . Therefore, it is difficult to evaluate whether and to what extent our estimates of isotope niche breadth would have differed had we included wintering isotope signals. We used recently published data from Reif et al. 35 to examine the extent to which isotope niches from our sampled populations reflected diet specialisation and climate niche breadth reported at the species level. As expected, dietary isotopic niche did not match the breadth or diversity of consumed food items reported in the literature. This further supports the notion that the breadth of dietary isotope niche reflects the isotopic variance of the resources and should not be considered synonymous with species diet 23 . Conversely, estimates of spatially-explicit breeding origin niche correlated significantly, albeit weakly, with literature data. In particular, measures of climatic niche from range maps appeared to better predict the upper distribution (quantile: 0.8) of our isotope niche, rather than its central response. A possible interpretation is that niche estimates derived from range maps necessarily reflect the maximum expected value for a given population. In other words, if we assume that a correlation exists between spatially-explicit origin niche inferred from isotope and range maps, any sampled populations should display isotope niches that are either equal or narrower than those derived from species full range maps. This provides additional evidence that species bi-dimensional δ 2 H-δ 18 O isotope space is a good proxy of their geographic and climatic niche dimensions 30 .
Among the key motivations for appraising niche requirements of migratory species, is that their populations appear to be declining more rapidly than residents 4,5,50,51 . The mechanisms underpinning these trends are still unclear and we examined whether narrow isotope breeding niches were associated with stronger population declines. We found no correlation between either dietary or origin breeding niche and long-term population trends as reported in the Pan European Common Bird Monitoring Scheme. This is not surprising as there is increasing evidence that environmental conditions and dynamics in non-breeding areas are key for the longterm persistence of populations 50,52,53 . In particular, habitat change and degradation in tropical wintering areas appear to be a major cause of decline in migrant species. As the conditions in wintering grounds carry-over to influence breeding success 54,55 , quantifying both breeding and wintering niche parameters would provide insight not only on basic ecological questions such as the degree of niche tracking, but also on conservation requirements of long-distance migrants 45 .
We used multiple stable isotopes to quantify and compare ecological niche parameters between short-and long-distance migrants. We showed that post-breeding migratory bottlenecks, such "Bocca di Caset" and "Passo del Brocon" in the Italian Alps, not only represent important stopovers for many migrants of the Western flyway 40 , but also strategic natural laboratories to examine migratory patterns across multiple species. However, not every species or populations are equally likely to use these locations as stopover. For instance, some central-European populations could fly over the Alps during the first migratory flight without stopping. The extent to which the potential selectivity of our study area influenced the results depends on the heterogeneity in isotope profiles within populations, which, however, can be large in some cases 55,56 .
Our results illustrate that long-distance trans-Saharan migrants, mostly breeding at higher latitudes than intra-Palearctic migrants, reach the migratory stop-over sooner. This is in line with the notion that early departure allows trans-Saharan migrants to reach the Sahel zone at the peak of vegetation greenness, when feeding conditions are optimal 57,58 . Moreover, in both migratory groups, the onset of breeding for northern populations appeared delayed, likely due to climatic constraints, so that the timing of migration was proportionally shifted later in the season. The delayed timing of northern populations can apparently carry over to the entire annual cycle 39 , and likely contributes to limit intra-specific competition. This type of migration pattern appears fairly common in many Nearctic-Neotropical migrants too 9,38 .
Our data also provide some support to the hypothesis that broad ecological niches are linked to long-distance migration, although this was only evident for the breeding origin niche dimensions, and requires further testing.
The use of multiple isotope ratios in animal tissues has allowed the quantification of both dietary and spatiallyexplicit niche aspects. However, the relation between the isotopic ecological niche and other descriptors of species niche (e.g. based on feeding habits, distribution patterns), or conservation status remains vague 23,30 . Here, we found that origin isotopic niche was consistent with the climatic range extent of the species, illustrating how δ 2 H and δ 18 O measurements can provide insight into the bio-climatic and geographic niche dimensions. However, the breadth of isotope niches was not related to the long-term population trends of the species at the European scale. Appraisal of the complete multi-seasonal niche requirements of long-distance migrants is therefore needed to link ecological information to conservation actions for these declining species. The study is part of the long-term post-breeding monitoring project 'Progetto Alpi' 60 . We only included species with at least 15 sampled individuals (median individuals per species = 26) to provide a representative measure of species isotopic niches. The remaining species were represented by a median of 3 individuals and were excluded (Fig. S4). Migratory species pass through the Italian Alps at different times during autumn migration 60 . Three migratory groups were considered: long-distance trans-Saharan migrants (TSM), which are fully migratory and leave their breeding ranges to winter in sub-Saharan Africa; short-distance intra-Palaearctic migrants (IPM), which can be fully or partially migratory and which have their breeding and wintering ranges within the Western Palaearctic; irruptive intra-Palaearctic migrants (IRM), which are partial migrants and residents, but show marked invasive and nomadic movements during non-breeding seasons. Date of capture was also recorded for each individual. Data analysis. We gathered information on migratory behaviour and distance for each bird species from the literature 65 . Migration distance was estimated as the distance between the breeding and wintering areas. Specifically, distance was estimated as the mean of the two latitudes during breeding minus the mean of the two latitudes during winter 66 .

Scientific Reports
To compare empirical estimates of dietary and origin isotopic niche breadth with literature data, we used recently published measures of diet specialisation and climate niche breadth that were derived from literature data and breeding range maps of European birds 35 . Diet specialisation was expressed as the coefficient of variation in the presence/absence of eight food types in the diet of each species (foliage, fruit, grain, insects, other invertebrates, terrestrial vertebrates, water vertebrates, carrion). Climate niche breadth was measured by overlying the breeding range map of each species to climatic data and calculating the covered temperature range 35 .
Information on long-term population trends for each species at the European scale was gathered from the Pan European Common Bird Monitoring Scheme (PECBMS) (https:// pecbms. info/ trends-and-indic ators/ speci es-trends/).
Dietary isotopic niche breadth (NB) was expressed as the Bayesian Standard Ellipse Area corrected for small sample size (SEAc) using the isotopic space defined by δ 13 C-δ 15 N. Similarly, spatially-explicit origin NB was expressed as SEAc from the δ 2 H-δ 18 O space. The SIBER R-package was used for these calculations 30,67 .
Differences in NB among migratory groups were examined using generalised least squares (GLS) allowing the variance to differ among groups (with varIdent function in R). Seasonal trends in δ 2 H, δ 13 C and δ 15 N within each migratory group were assessed using linear mixed-effect models including species as random factor to account for different isotope values and trends across species. We used the lme function in R allowing both intercept and slope to differ among species. To examine if origin NB was related to migration distance, we used GLS to regress the mean NB values for each species (across all measured individuals) against the estimated migration distance, allowing the variance to differ among species (with varFixed).
Although migration timing can vary within species depending on age and sex, our analyses only included juveniles born in the year of sampling, which were likely fed a similar diet regardless of sex. In fact, no particular differences in isotope patterns were observed between male and female for the species that we were able to sex (Figs. S5, S6) www.nature.com/scientificreports/