Path analysis reveals combined winter climate and pollution effects on the survival of a marine top predator
Data files
Jul 19, 2024 version files 22.45 KB
-
Covariate_Dat_forDryad.csv
-
Individ_Dat_forDryad.csv
-
README.md
Abstract
Marine ecosystems are experiencing growing pressure from multiple threats caused by human activities, with far-reaching consequences for marine food webs. Determining the effects of multiple stressors is complex, in part, as they can affect different aspects of biological organisation (behaviour, individual traits, demographic rates). Determining the combined effects of stressors, through different biological pathways, is key to predicting the consequences for the viability of populations threatened by global change. Due to their position in the food chain, top predators such as seabirds are considered more sensitive to environmental changes. Climate change is affecting the prey resources available for seabirds, through bottom-up effects, while organic pollutants can bioaccumulate in food chains with the greatest impacts on top predators. However, knowledge of their combined effects on population dynamics is scarce. Using a path analysis, we quantify the effects of climate change and pollution on the survival of adult great black-backed gulls, both directly and through the effects of individuals’ body mass. Warmer ocean temperatures in gulls’ winter foraging areas in the North Sea were correlated with higher survival, potentially explained by shifts in prey availability associated with global climate change. We also found support for the indirect negative effects of organochlorines, highly toxic pollutants to seabirds, on survival which acted, in part, through a negative effect on body mass. The results from this path analysis highlight how, even for such long-lived species where variance in survival tends to be limited, two stressors still have had a marked influence on adult survival and illustrate the potential of path models to improve predictions of population variability under multiple stressors.
README: Path analysis reveals combined winter climate and pollution effects on the survival of a marine top predator
https://doi.org/10.5061/dryad.t4b8gtj9k
Description of the data and file structure
The data needed to run the analyses in the study (CMR-path model and model selection in RMark) are in two files;
(1) "Individ_Dat_forDryad.csv" containing individual-level capture histories and contaminant data.
(2) "Covariate_Dat_forDryad.csv" containing annual climate covariates.
File (1) includes; individual capture histories ("ch", one individual per row) and also that individual's sex ("sex", M = male, F = female), scaled weight in grams ("Weight_sc", mean centred and standard deviation = 1), the first principal component from the PCA of all log-transformed contaminant levels (ng g-1 wet weight), and the scaled value of the log-transformed oxychlordane (see Methods Biometric and pollutant data).
File (2) includes; year (from and to year, which is relevant for covariates including months covering year t and year t+1) and all covariates of different relevant periods. "mslp" = mean sea level pressure (hPa), "sst" = sea surface temperature, °C), "airT" = air temperature, °C, "nao" = North Atlantic Oscillation (no units), "aw" = Atlantic water inflow (10^6 m3 s−1) . "ns" = North Sea, "hor" = Hornøya, for covariates where values were extracted from specific areas (see Supplementary Information S2 Figure S3). Seasons were separated into; "EWin" = early winter, "LWin" = later winter, "spr" = spring, "aut" = autumn, and "win" = winter.
Sharing/Access information
Demographic data are collected as part of the SEAPOP program (https://seapop.no/en/).
Geolocation data for visualising seabirds' core non-breeding areas are available through the SEATRACK program (https://seapop.no/en/seatrack/).
Code/Software
The model script to run the CMR-path model (run in JAGS via JAGSUI in program R) is included as "Path-CMR_Model_Script".
The path model was implemented in JAGS (Plummer 2003), via the program JAGSUI, version 1.5.2 (Kellner 2015), and implemented in R version 4.2.2 (R Core Team 2023). Three chains were run of length 200,000, where the first 50,000 was discarded as burn-in.
Visualisation of model outputs (Figures 3-4) was produced using the R package MCMCvis.
Methods
Capture-mark-resight data
We used individual-level capture-mark-resight (CMR) data collected between 2001 and 2017. Only mark-recapture data from individuals with known levels of OCs (N=158), i.e., caught and sampled in either 2001 or 2002, were included in the analysis. During the first capture, birds were marked with a unique numbered metal ring and individually coded colour ring for future resighting, which is possible with a telescope or binoculars. Each year, visual searches were made for marked birds, predominantly at the same colony where initial capture and ring-marking took place.
Biometric measures (weight)
Breeding adults were caught using nest traps during the incubation period (May–June) in 2001 and 2002. Once caught, birds were weighed, and the head and bill length were measured. GBB gulls were sexed using measurements of head and bill length as they are size dimorphic (see Bustnes et al. 2008a for details). Individuals’ body mass (weight) was used as a measure of individual quality.
Pollutants
Blood concentrations of several organochlorines (OCs) were measured from 158 individuals during field campaigns in 2001 and 2002.
Climate covariates
We selected climate variables reflecting non-breeding season conditions that may affect survival rates, which were available at different spatial scales; North Atlantic Oscillation, Atlantic Water inflow, land temperature, sea surface temperature, and mean sea level pressure. North Atlantic Oscillation (NAO) and Atlantic Water inflow (AW) were available as monthly values. The NAO index reflects large-scale weather patterns specifically cyclone activity in the North Atlantic and is widely used as a proxy for indirect effects of climate conditions on seabirds via changes in prey distributions and abundances (Stenseth et al. 2003). Annual monthly values of the NAO index are openly available from www.cpc.ncep.noaa.gov. In the Barents Sea, ocean conditions are largely determined by fluctuations in the inflow of warm and saline Atlantic water (AW; Loeng 1991; Ingvaldsen, Asplin & Loeng 2004), again affecting seabirds through changes in prey availability (Barrett, Erikstad & Reiertsen 2017). Cyclones, in part, modulate the influx of Atlantic water and so this can be related to NAO (Heukamp et al. 2023). Data of AW were available at a monthly scale from www.thredds.met.no/thredds/catalog/nansen-legacy-ocean/SVIM/catalog.html. NAO and AW were aggregated for two periods: winter (November–February) and spring (March–April). Land temperature (airT), sea surface temperature (SST), and mean sea level pressure (MSLP) were available as gridded data and were extracted from within seabirds’ seasonal foraging ranges (Appendix S2, Fig. S2) and were available from the European Centre for Medium-Range Weather Forecasts (ECMWF) Re-Analysis Interim Project (‘ERA-Interim’). ERA-Interim is a gridded model dataset at a resolution of 0.70° or approximately 79km based on data assimilation of meteorological station data, and satellite data, among others (Dee et al., 2011; Mesquita et al., 2015). Values of MSLP, SST and airT were extracted from areas reflecting the utilisation distributions, for the relevant seasonal periods (Supplementary Information S2 Figure S3). Annual covariates were aggregated to four periods: Autumn (September–October, Aut), early Winter (November–December, Ewin), late Winter (January–February, Lwin), and Spring (March–April, Spr), for the years 2001–2017. Values were extracted from the North Sea region for Autumn, early Winter, and late Winter (NNS/SNS, Figure S3). For spring, values were extracted from around the breeding colony at Hornøya (HRN, Figure S3). See Supplementary information S2 Figure S4 for plots of the annual time series.