Data from: Among-species variation in six decades of changing migration timings explained through ecology, life-history and abundance
Data files
Jun 19, 2024 version files 12.29 MB
-
1_2_Null_Models_DRYAD_ARCHIVE.R
45.21 KB
-
3_Distance_Models_DRYAD_ARCHIVE.R
72.42 KB
-
4_Diet_Models_DRYAD_ARCHIVE.R
108.64 KB
-
5_Habitat_Models_DRYAD_ARCHIVE.R
97.05 KB
-
6_BroodNumber_Models_DRYAD_ARCHIVE.R
81.76 KB
-
7_Generation_Length_Models_DRYAD_ARCHIVE.R
92.39 KB
-
8_9_Local_Migratory_Abundance_Models_DRYAD_ARCHIVE.R
109.55 KB
-
AUTUMN_2_step_simulated_data_2_S8_ARCHIVE_RUN1.csv
2.51 MB
-
AUTUMN_2_step_simulated_data_2_S8_ARCHIVE_RUN2.csv
2.51 MB
-
Autumn_quantile_data_64_species.csv
723.70 KB
-
Final_Trait_Data_72_species.csv
7.43 KB
-
README.md
11.17 KB
-
S12_2_Level_Diet_MCMCglmm_DRYAD_ARCHIVE.R
71.04 KB
-
S3_Caclulating_Change_in_LMA_DRYAD_ARCHIVE.R
37.69 KB
-
S5_Correlations_Nesting_EVs_DRYAD_ARCHIVE.R
13.85 KB
-
S6_Multivariate_Analysis_DRYAD_ARCHIVE.R
108.03 KB
-
S7_Abundance_Weighting_DRYAD_ARCHIVE.R
11.21 KB
-
S8.1._Simulation_Models_RUN1_DRYAD_ARCHIVE.R
30 KB
-
S8.2_Simulation_Models_RUN2_DRYAD_ARCHIVE.R
40.52 KB
-
SPRING_2_step_simulated_data_2_S8_ARCHIVE_RUN1.csv
2.44 MB
-
SPRING_2_step_simulated_data_2_S8_ARCHIVE_RUN2.csv
2.44 MB
-
Spring_quantile_data_64_species.csv
720.96 KB
Abstract
Species utilising seasonal environments must now alter timings of key life-history events in response to large-scale climatic changes, thereby maintaining trophic synchronies. Yet substantial among-species variation in cross-decadal phenological changes is observed. Transitioning from basic description of such variation towards prediction of future phenological responses now requires standardised studies that rigorously quantify and explain variation in the direction, magnitude and form of changing timings across diverse species in relation to key ecological and life-history variables. Accordingly, we fitted multi-quantile regressions to 59 years of high-quality multi-species data on spring and autumn bird migration timings through northern Scotland. We demonstrate substantial variation in cross-decadal changes in timings among 72 species, and quantify the degree to which variation can be explained through differences in species ecology, life-history and population trajectories. Consistent with predictions, species with seasonal diets, narrower breeding habitat breadths, shorter generation lengths and capability to produce multiple offspring broods per year advanced their migration timing in one or both seasons. In contrast, species with less seasonal diets, and that produce single annual offspring broods, showed no change. Meanwhile, contrary to prediction, long-distance migrants advanced their migration timings as much as short-distance migrants. Changes in migration timing also varied with changes in local migratory abundance, such that species with increasing seasonal abundance apparently altered their migration timing, whilst species with decreasing abundance did not. These patterns concur with expectation if changing migration timing is adaptive. However, we demonstrate that similar patterns can be generated through numerical sampling processes given changing abundances, implying that apparent phenology-abundance relationships should be carefully validated and interpreted. Overall, our results show that migrant bird species with differing ecologies and life-histories have shown systematically differing phenological changes over six decades contextualised by large-scale environmental changes, potentially facilitating future predictions and altering temporal dynamics of seasonal species co-occurrences.
https://doi.org/10.5061/dryad.31zcrjdth
The main dataset comprises two data files, giving the quantile migration dates for each focal species in 1) spring and 2) autumn in each study year (1960-2018).
There are data for 64 species in each season (which can differ between the seasons).
The datasets also give:
the values of key ecological and life-history covariates for each species
the total abundance of each species in each year and the mean across years through the study period
the slope of abundance on year
These data are sufficient to reproduce the analyses presented in Dale et al.’s manuscript, when combined with phylogeny data which can be downloaded from birdtree.org
In addition, for ease of analysis, we include a separate file of all covariates for all 72 species considered across spring and autumn combined.
Further, there are four datafiles of simulated data for the analyses of changing phenology in relation to changing abundance (two for spring and two for autumn. Each file contains 10 replicates).
Description of the data and file structure
The two datasets named “Spring_quantile_data_64_species.csv” and “Autumn_quantile_data_64_species.csv” give the data for spring and autumn respectively. The dataset named “Final_Trait_Data_72_species” gives the full trait data for all species.
The four datasets named below are the simulated datasets:
SPRING_2_step_simulated_data_2_S8_ARCHIVE_RUN1.csv
AUTUMN_2_step_simulated_data_2_S8_ARCHIVE_RUN1.csv
SPRING_2_step_simulated_data_2_S8_ARCHIVE_RUN2.csv
AUTUMN_2_step_simulated_data_2_S8_ARCHIVE_RUN2.csv
Column names are defined as below:
rowno - datafile row number
species – English species name as given in Fair Isle Database. Factor/Character, 64 levels in each season.
year – Study year from 1960 – 2018
quant5 – Ordinal julian day by which the 5th quantile of the species’ sighting distribution had occurred. Integer from 0 – 366.
quant10 – Ordinal julian day by which the 10th quantile of the species’ sighting distribution had occurred. Integer from 0 – 366.
quant25 – Ordinal julian day by which the 25th quantile of the species’ sighting distribution had occurred. Integer from 0 – 366.
quant50 – Ordinal julian day by which the 50th quantile of the species’ sighting distribution had occurred. Integer from 0 – 366.
quant75 – Ordinal julian day by which the 75th quantile of the species’ sighting distribution had occurred. Integer from 0 – 366.
quant90 – Ordinal julian day by which the 90th quantile of the species’ sighting distribution had occurred. Integer from 0 – 366.
quant95 – Ordinal julian day by which the 95th quantile of the species’ sighting distribution had occurred. Integer from 0 – 366.
year_sum – Total count of individuals sighted of each species in a single year in Spring or Autumn. Integer/Count.
distance (or Migration_Distance) – Species classified as Long or Short Distance migrants. Divided according to whether species predominantly breeds north (short) or south (long) of the Sahara. Factor with 2 levels: Long: Long-distance migrants; Short: Short-distance migrants.
diet (or Diet_Category) – Dominant diet type assigned out of 5 diet categories, following Eltontraits. Factor with 4 levels. Invertebrates: Invertebrate diet; PlantSeed: Plant and Seed/Grain diet; VertFishScav: Vertebrate, fish and carrion diet; Omnivore: Multiple diet types, with no single diet comprising >50% of diet.
diet_diversity_count – Number of constituent diet types out of the 10 diet constituents recorded for the species in the Eltontraits Database. Integer from 1 – 10.
gen_length – Mean age (in years) of breeding adults, sourced from Birdlife International’s IUCN Redlist for Birds (2020). Numeric.
body_mass (or BodyMass_Value) – Body mass (g) sourced from Eltontraits. Given value is an average of body mass.
breeding_hab_range (or Suitable_Breeding_Habitat_Levels) – Total number of habitat types classed as “suitable” or above for each species by Birdlife International in the breeding season.
nonbreeding_hab_range (or Suitable_NonBreedingHabitat_Levels) – Total number of habitat types classed as “suitable” or above for each species by Birdlife International in the nonbreeding season.
BN_Factor – – Species categorised according to their ability to produce multiple broods in a single year. Data source: BTO BirdFacts. Factor with 2 levels: Single: Species restricted to single brood; Multiple: Species able to produce multiple broods in a single year.
mean_abundance_autumn - Mean number of individuals of each species sighted on Fair Isle in Autumn across study period from 1960 – 2018.
mean_abundance_spring – Mean number of individuals of each species sighted on Fair Isle in Spring across study period from 1960 – 2018.
FI_Autumn_Trend_Adj_0.5 – Mean standardised species slope estimate from linear regression models of the total number of autumn sightings in individual years on Fair Isle on year. Numeric.
FI_Spring_Trend_Adj_0.5 – Mean standardised species slope estimate from linear regression models of the total number of spring sightings in individual years on Fair Isle on year. Numeric.
year_scaled - Study year encoded from 0 – 58. E.g. 0 = start, 58 = end. Integer.
Family – Birdlife V3.0 Checklist Family Latin name from Eltontraits. Factor.
Order – IOC V2.7 Checklist Order from Eltontraits. Factor.
fad – Ordinal julian day on which the first species sighting occurred. Integer from 0 – 366.
lad – Ordinal julian day on which the last species sighting occurred. Integer from 0 – 366.
loop_number – Number between 1 – 20 defining the simulation iteration number in which data were generated.
References:
“Eltontraits” refers to: Wilman, H., Belmaker, J., Simpson, J., de La Rosa, C., Rivadeneira, M., & Jetz, W. (2014). EltonTraits 1.0 : Species-level foraging attributes of the world’s birds and mammals. Ecology, 95(7), 2027. https://doi.org/10.1890/13-1917.1
BTO BirdFacts refers to: Robinson, R. A. (2005) BirdFacts: profiles of birds occurring in Britain & Ireland (BTO Research Report 407). BTO, Thetford (http://www.bto.org/birdfacts, accessed on 01/09/2020)
BirdLife International refers to: BirdLife International (2020) IUCN Red List for birds. Downloaded from: http://www.birdlife.org on 01/09/2020.
Sharing/Access information
Data was derived from the full Fair Isle bird migration dataset, which is held by Fair Isle Bird Observatory Trust.
https://www.fairislebirdobs.co.uk/
Code/Software
Codes were created in R version 3.6.1.
There are seven main R code files, which together generate all analyses and figures presented in the main text. File numbers indicate the numbers of the figures generated.
1&2_Null_Models_DRYAD_ARCHIVE.R - Import data and run null models without any covariates to quantify the amount of variance present in changing species phenology. Code generate Figs 1-2 of main manuscript
3_Distance_Models_DRYAD_ARCHIVE.R - Import data and run distance models in both seasons with distance divided into long vs. short. Models run with and without phylogenetic controls. Code generates Fig 3 from main manuscript.
4_Diet_Models_DRYAD_ARCHIVE.R - Import data and run diet models in both seasons with diet divided into 4 categories. Models run with and without phylogenetic controls. Code generates Fig 4 from main manuscript.
5_Habitat_Models_DRYAD_ARCHIVE.R - Import data and run habitat models in both seasons with habitat_breath (breeding_hab_range ) as a covariate. Models run with and without phylogenetic controls. Code generates Fig 5 from main manuscript.
6_BroodNumber_Models_DRYAD_ARCHIVE.R - Import data and run brood number models in both seasons with brood number divided into 2 categories (single and multiple). Models run with and without phylogenetic controls. Code generates Fig 6 from main manuscript.
7_Generation_Length_Models_DRYAD_ARCHIVE.R- Import data and run generation length models in both seasons with gen_length as a covariate. Models run with and without phylogenetic controls. Code generates Fig 7 from main manuscript.
8&9_Local_Migratory_Abundance_Models_DRYAD_ARCHIVE - Import data and run local migratory abundance models in both seasons with FI_Spring_Trend_Adj_0.5 / FI_Autumn_Trend_Adj_0.5 as covariates. Models run with and without phylogenetic controls. Code generates Fig 9 from main manuscript, including results of same analyses on simulated data from Supporting information 8.
There are seven additional R code files, which together generate the main analyses and figures presented in the supporting information. File numbers indicate the supporting information sections.
S3_Caclulating_Change_in_LMA_DRYAD_ARCHIVE.R - Import data and fits regressions for seasonal abundance ~ year. These estimates are mean standardised to provide the covariate values in FI_Spring_Trend_Adj_0.5 & FI_Autumn_Trend_Adj_0.5. Code generates Fig 8 from the main manuscript. Further details are included in Supporting information 3.
S5_Correlations_Nesting_EVs_DRYAD_ARCHIVE.R – Import data and code produces all correlation plots and mosaic plots to show nesting of variable. Code generates outputs included in Supporting information 5.
S6_Multivariate_Analysis_DRYAD_ARCHIVE.R – Import data and run multivariate models with a subset of species in both seasons. Code generates outputs included in Supporting information 6.
S7_Abundance_Weighting_DRYAD_ARCHIVE.R – Import data and trial abundance weighted regressions on changing species phenologies. Code produces all the analyses and graphs in Supporting information 7.
S8.1._Simulation_Models_RUN1_DRYAD_ARCHIVE.R - Imports simulated data from simulation run 1 (which comprised the first 10 iterations/loops 1 - 10) and runs models testing for an effect of changing local migratory abundance on simulated data.
S8.2_Simulation_Models_RUN2_DRYAD_ARCHIVE.R - Imports simulated data from simulation run 2 (which comprised the first 10 iterations/loops 11 - 20) and runs models testing for an effect of changing local migratory abundance on simulated data. Code also combines results RUN 1 and RUN 2 models into a single plot. Code produces all analyses and graphs in Supporting information 8.
S12_2_Level_Diet_MCMCglmm_DRYAD_ARCHIVE.R - Import data and run diet models in both seasons with diet divided into 2 categories. Models run with and without phylogenetic controls. Code produces all analyses and graphs in Supporting information S12.
The data comprise processed field observations of migrant bird timings through Fair Isle, Scotland, in both spring and autumn from 1960-2018.
The data are the 5th, 10th, 25th, 50th, 75th, 90th and 95th quantiles of the raw distributions of species sighting dates.
Covariate data on species ecology and life-history (extracted from other published data sources) are included to facilitate replication of analyses.
There are data from a total of 72 species in spring and/or autumn, comprising selected (relatively common) passerines and non-passerines.