Data from: Experimental increases in fecundity are offset by reduced juvenile survival, limiting growth of a songbird population

Title: “Experimental increases in fecundity are offset by reduced juvenile survival, limiting population growth of a songbird”
Authors: Sarah D. Mueller, Daniel J. Mennill, Hayley A. Spina, Sarah L. Dobney, Nathaniel T. Wheelwright, Amy E. M. Newman, Stéphanie M. Doucet, Joseph B. Burant, Greg W. Mitchell, Bradley K. Woodworth, D. Ryan Norris
Journal: Ecology (in review)
Year of publication: (in review)

Study system: Savannah sparrows (Passerculus sandwichensis)
Study site: Kent Island, New Brunswick, Canada (44.58°N, 66.76°W)
Timespan: 1987-2025

Summary

The repository includes the data and R scripts to analyze whether experimentally increasing nesting success using predator exclosures drives population growth in an island population of Savannah sparrows (Passerculus sandwichensis). We used exclosures to protect nests in four years (2021-2024) and compared demographic data from these years to the preceding years in which predator exclosures were not used (1987-2019). The data come from a long-term study of the demography and breeding biology of a population of Savannah sparrows on Kent Island, New Brunswick, Canada (44.58°N, 66.76°W).

Datasets
(1) events.csv: contains mark-resighting data of individual Savannah sparrows, with one row per individual per year
(2) cjs_output.csv: contains estimates of annual juvenile and adult survival of Savannah sparrows generated by the Cormack-Jolly-Seber model fit in the script exclosure_CJS_model.R (see below)
(3) population_data.csv: contains annual demographic data for the population including population size and reproductive metrics
(4) nests_sub.csv: contains records of all nests within the study area from 1987 to 2025 (note that 2025 nests have only male and female IDs and no reproductive data)

Other files
(1) adult_SAVS.png: illustration of Savannah sparrow used in Figure 1 (copyright Sarah Mueller)
(2) Three_islands.kml: GIS file containing polygon borders of the Three Islands Archipelago (Kent and surrounding islands) for map in Figure 1
(3) canada_provinces folder: contains shapefile (lpr_000b21a_e.shp) of provinces of Canda for inset map in Figure 1. Downloaded from Government of Canada - Statistics Canada website. Full citation: Statistics Canada, 2021. Census Boundary files (province-level). https://www12.statcan.gc.ca/census-recensement/2021/geo/sip-pis/boundary-limites/index2021-eng.cfm?year=21.
(4) us_states folder: contains shapefile (cb_2018_us_state_500k.shp) of US states for inset map in Figure 1. Downloaded from US Census Bureau website. Full citation: United States Census Bureau, 2018. Cartographic Boundary Files - Shapefile (state-level). https://www.census.gov/geographies/mapping-files/time-series/geo/carto-boundary-file.html.

Variables

In events.csv:
year = integer; the year of the capture occasion
band_id = factor; unique aluminum band number (Canadian Wildlife Service) assigned to each individual bird
birth_year = integer; the year in which an individual bird was born
banding_age = factor; the age at which an individual bird was banded: AHY = after hatch year (adult, but exact age unknown), ASY = after second year (adult, born at least two years ago), N = nestling, SY = second year (adult, born the preceding year)
detect = binary; whether the individual was detected in a given year (0 = not detected, 1 = detected)
initial.age = factor; the individual’s age on its first capture occasion (0 = juvenile, 1 = adult)
n_captures = integer; the number of capture occasions on which an individual was detected

In cjs_output.csv:
year = integer; the focal year (year t)
phi_juvenile = numeric; the probability of a juvenile bird surviving from the focal year (year t) to the following year (year t + 1) generated by a Cormack-Jolly-Seber model (see the R script exclosure_CJS_model.R)
lcl_juvenile = numeric; the lower bound of the 95% confidence interval of the juvenile survival estimate, phi_juvenile
ucl_juvenile = numeric; the upper bound of the 95% confidence interval of the juvenile survival estimate, phi_juvenile
phi_adult = numeric; the probability of an adult bird surviving from the focal year (year t) to the following year (year t + 1) generated by a Cormack-Jolly-Seber model (see the R script exclosure_CJS_model.R)
lcl_adult = numeric; the lower bound of the 95% confidence interval of the adult survival estimate, phi_adult
ucl_adult = numeric; the upper bound of the 95% confidence interval of the adult survival estimate, phi_adult

In population_data.csv:
year = integer; the focal year (year t)
prop_fledged = numeric; nest success, the proportion of nests in the focal year that fledged at least one young
total_fledglings = integer; the total number of fledglings produced from all nests in the focal year
mean_clutch_size = numeric; the mean clutch size across all nests in the focal year
n_nests = integer; the number of nests in the focal year
n_females = integer; the number of breeding females observed in the focal year
n_males = integer; the number of breeding males observed in the focal year
n_db_females = integer; the number of females that double-brooded in the focal year
fecundity = numeric; the average number of fledglings produced per female in the focal year (equal to total_fledglings / n_females)
population_size = integer; the number of breeding adults observed in the focal year (equal to n_females + n_males)
prop_db = numeric; the proportion of females that double-brooded in the focal year (equal to n_db_females / n_females)
caged = factor; whether or not predator exclosures were used in the focal year (“exclosures” or “no exclosures”)
lambda = numeric; the population growth rate, equal to the population size in the year following the focal year (year t + 1) divided by the population size in the focal year (year t). Lambda > 1 indicates the population grew from the focal year to the following year; lambda < 1 indicates the population shrank; and lambda = 1 indicates constant population size.

In nests_sub.csv:
nest_key = factor; an alpha-numeric code assigned to uniquely identify each nest
year = integer; the year in which the nest attempt occurred
eggs_laid = integer; number of eggs laid in the nest (clutch size)
female_id = factor; unique aluminum band number (Canadian Wildlife Service) assigned to the female parent of the nest
male_id = factor; unique aluminum band number (Canadian Wildlife Service) assigned to the male parent of the nest
found_date = date (mm/dd/yy); calendar date on which the nest was found
hatch_date = date (mm/dd/yy); calendar date on which the first egg hatched
eggs_hatched = integer; the number of eggs in the nest that hatched (brood size)
fledglings = integer; the number of young that fledged from the nest
fledge_date = date (mm/dd/yy); calendar date on which the first young fledged from the nest
fate = factor; the fate of the nest: AB = abandoned while building, AE = abandoned with eggs, AN = abandoned with nestlings, DN = dead nestlings, FL = fledged, KN = killed nestlings, P = predated, PE = predated with eggs, PN = predated with nestlings, R = experimental egg removal, U = unknown, XE = nest accidentally stepped on
fail_date = date (mm/dd/yy): calendar date on which the nest failed, if applicable
experiment = factor; nests that received a predator exclosure in 2021-2024 have the experiment code “excl.” Other codes are: C = control nest in 1989; CE, CEGN, CN, EXCL, GE, GECN, GEGN, GN refer to various gull model experiments (see Wheelwright et al. 1997, Animal Behaviour); Con = control nest in 2008-2009; CORT and CORT* = corticosterone fed to nestlings in first week of life; daily = daily weighing of chicks; E1, E2, E3 = estimated brood number (estimated first, second, or third brood), Enl = experimentally enlarged clutch size; Lys = lysozyme treatment, which had no effect; none = no experimental treatment; PBS = treatment with PBS, which had no effect; Red = experimentally reduced clutch size; SOSP = eggs swapped with song sparrow (Melospiza melodia) nest, temp = temperature probe in nest
Notes = notes about the nest

Code

The R scripts provided include all the code required to import the data and perform the analyses in the manuscript. The script exclosure_CJS_model.R is run first, to generate CJS model estimates of adult and juvenile survival. These estimates are saved in the file cjs_output.csv and are used in the script exclosure_analysis.R.

exclosure_CJS_model.R
- Construct capture history from event data (events.csv) and fit CJS model
- Generates cjs_output.csv, with adult and juvenile survival estimates and 95% CIs from CJS model

exclosure_analysis.R
- Uses the datasets cjs_output.csv, population_data.csv, and nests_sub.csv
- Fits a path analysis model using piecewiseSEM to examine effects of predator exclusion, juvenile survival, adult survival, and fecundity on the population growth rate
- Generates the figures presented in the manuscript and supporting material

R version: v. 4.5.2

R package versions:
RMark v. 3.0.0
tidyverse v. 2.0.0
dagitty v. 0.3.4
piecewiseSEM v. 2.3.1
semEff v. 0.7.2
reformulas v. 0.4.3.1
DiagrammeR v. 1.0.11
emmeans v. 2.0.1
DHARMa v. 0.4.7
performance v. 0.15.2
ggeffects v. 2.3.2
sf v. 1.0.23
cowplot v. 1.2.0