Data and code from: Breakdown in seasonal dynamics of subtropical ant communities with land-cover change
Data files
Sep 27, 2023 version files 2.67 MB
-
okiAntsTempVar_data.zip
-
README.md
Oct 03, 2023 version files 2.67 MB
-
okiAntsTempVar_data.zip
-
README.md
Abstract
Concerns about widespread human-induced declines in insect populations are mounting, yet little is known about how land-use change modifies the dynamics of insect communities, particularly in understudied regions. Here, we examine how the seasonal activity patterns of ants—key drivers of terrestrial ecosystem functioning—vary with anthropogenic land-cover change on a subtropical island landscape, and whether differences in temperature or species composition can explain observed patterns. Using trap captures sampled biweekly over two years from a biodiversity monitoring network covering Okinawa Island, Japan, we processed 1.2 million individuals and reconstructed activity patterns within and across habitat types. Forest communities exhibited greater temporal variability of activity than those in more developed areas. Using time-series decomposition to deconstruct this pattern, we found that sites with greater human development exhibited ant communities with diminished seasonality, reduced synchrony, and higher stochasticity compared to sites with greater forest cover. Our results cannot be explained by variation in regional or site temperature patterns, or by differences in species richness or composition among sites. Our study raises the possibility that disruptions to natural seasonal patterns of functionally key insect communities may comprise an important and underappreciated consequence of global environmental change that must be better understood across Earth’s biomes.
README: Data and code for: Breakdown in seasonal dynamics of subtropical ant communities with land-cover change
Access these datasets on Dryad
This archive contains the data files used in the analyses for "Breakdown in seasonal dynamics of subtropical ant communities with land-cover change", published in Proceedings of the Royal Society B: Biological Sciences (DOI: 10.1098/rspb.2023.1185).
NOTE: All datasets (from Dryad and Zenodo) must be put into a single folder called /data
within the main directory of the R analysis package before running any code.
activity_orig.rds
: A tibble with the ant activity data (before processing), collected by the Okinawa Environmental Observation Network (OKEON) with Sea, Land, and Air Malaise (SLAM) traps on Okinawa Island in Japan from 2016-2018 at 24 sites sampling different environments across the island (data downloaded from OKEON database in November 2022). The field "count" refers to how many individuals were collected during a collection event, but the interpretation of this count is a quantity called "activity", further described in the paper.
microclim_data.rds
: Tibbles with in situ climate measurements and associated metadata per site with temporal resolution of 15 minutes. These in situ climate data (site air (1.5 m above-ground) and soil (10 cm below-ground) temperatures) were used to determine if site differences in climate contributed to observed community temporal variability patterns. Temperatures are in Celcius. Some values for soil_temperature in the original dataset are NA due to occasional equipment malfunction, but the final table mc.summary
(created in tempvar_analysis.Rmd
, see below) averages these values over 2-week time bins and ignores these NA values. Thus, the final table used for analysis does not contain NAs.
samples_FY2016.csv
and samples_FY2017.csv
: Tables describing the exact dates that collection events occurred. These data were used to attribute the sampling events by time bin. Standardized time bin attribution was necessary because some sampling events began or ended on different days.
okinawa_ant_invasive_status.csv
: Table describing invasive status per species collected, decided by expert opinion by co-authors M. Yoshimura, F. H. Garcia, and G. Fischer.
Several other data files are included in a Zenodo supplemental information archive with the CC-By 4 license.
lc.rds
: Raster data describing land-cover classifications for Okinawa Island, developed by co-author K. L. Dudley and originally described in Ross et al. 2018 (reference below). The land-cover data was used to characterize each sampling site. We did this by calculating land-cover proportions within a 1-km buffer, then reducing dimensionality with a PCA to derive two principal component axes: PC1 describing the developed-forested gradient, and PC2 describing the rural-urban gradient. These axes (PC1 and PC2) were used as predictor variables in our models of temporal variability. In this dataset,
jma_data.rds
and jma_meta.rds
: Tibbles with Japan Meteorological Agency (JMA) regional climate data with temporal resolution of 1 day. The downloaded data was processed to retain only dates between 2016-03-22 and 2018-04-01 for the analysis, Japanese station site names were removed while English ones were kept, and several column names were changed for ease of analysis. Like the in situ site climate data, the JMA regional climate data was used to determine if site differences in climate (here, coarser spatial scale) contributed to observed community temporal variability patterns. Temperatures are in Celcius and precipitation is in millimeters. Similar to the in situ climate dataset, some values in the original dataset are NA but the final table jma.summary
used for analysis (created in tempvar_analysis.Rmd
, see below) averages over 2-week time bins and ignores these NA values. Thus, the final table used for analysis does not contain NAs.
Description of the data and file structure
R object files with extension .rds can read in R
using readRDS()
, and .Rdata files can be read with load()
.
NOTE: All datasets (from Dryad and Zenodo) must be put into a single folder called /data
within the main directory of the R analysis package before running any code.
Code/Software
The analysis code for the paper is found here structured as a simple R package. Run the file okiAntsTempVar.Rproj
to open an R Studio session to easily access the code.
Main analysis script
tempvar_analysis.Rmd
: A documented R Markdown script file with the code for the main analysis. This script run all analyses presented in the paper, creates and saves all figures to .pdf or .png (some of which were then edited in Adobe Illustrator), saves the model results table to a .csv file, and also saves a copy of the R environment as a .Rdata file after the analysis is complete. This .Rmd has parameters in the YAML header (between the ---
lines at the top), and these are changed to rerun different iterations in analysis/run_sensitivity_analysis.R
(explained below) to perform a sensitivity analysis (Fig. S6 in the paper). All custom package functions used in this .Rmd are found in the scripts below that specify them.
Scripts found in /R
okeon_activ.R
: Functions for processing the ant activity data.
okeon_bd.R
: Functions for calculating community temporal beta diversity ("compositional diversity" in the paper) using the R package adespatial
.
okeon_cv.R
: Functions for calculating community temporal coefficient of variation ("functional diversity" in the paper).
okeon_div.R
: Functions for calculating species richness metrics, including extrapolations, using the R package iNEXT
.
okeon_gis.R
: Function for making spatial objects for the site points using the R package sf
.
okeon_jma.R
: Functions for processing the JMA climate data.
okeon_lc.R
: Functions for processing the land cover data.
okeon_mods.R
: Functions for building models, calculating AICc, and performing model selection.
okeon_rar.R
: Functions for performing different rarefaction analyses on the ant activity data (sampling site-level and land-cover group-level).
okeon_ts.R
: Functions for implementing time-series regression models to decompose the ant activity time-series data.
okeon_util.R
: Utility functions to abbreviate text strings for species and site names.
utils-pipe.R
: ROxygen code to allow the use of the R package magrittr
"pipe" command.
Scripts found in /analysis
run_sensitivity_analysis.R
: This code iterates different parameterizations for the main analysis script tempvar_analysis.Rmd
by changing the count threshold and trap/site level. The count threshold (parameter threshold.val
) governs the maximum ant count value retained in the dataset, and all values above this are constrained to the threshold (the default is 500). The trap/site level (parameter threshold.lvl
) governs the spatial level at which the threshold is applied (either by trap or by site, where there are 3 traps per site). The purpose of these iterations is to conduct a sensitivity analysis for the effects of these parameter changes on the results, and the outcomes are shown in Fig. S6 (none were found to be extreme).
References
Ross, S. R. J., Friedman, N. R., Dudley, K. L., Yoshimura, M., Yoshida, T., & Economo, E. P. (2018). Listening to ecosystems: data-rich acoustic monitoring through landscape-scale sensor networks. Ecological Research, 33, 135-147.
Methods
The ant activity data used in the analysis was collected with Sea, Land, and Air Malaise (SLAM) traps on Okinawa Island in Japan from 2016-2018, and was processed using the code provided in the Zenodo archive (see README for links and the paper for references). Other datasets come from the Japan Meterological Agency (JMA), in situ climate variables for sampling stations measured on-site, and land-cover data for Okinawa developed by our team. The data used in the analysis is included in the data upload (with JMA and land-cover data files in a Zenodo supplemental information under the CC-By 4 license), and all the analysis code is included in the R package provided in a separate Zenodo software archive.
NOTE: All datasets (from Dryad and Zenodo) must be put into a single folder called `/data` within the main directory of the R analysis package before running any code. Please consult the README for further details on the data and code.
Usage notes
All analysis were conducted in R.