Impervious surface cover and number of restaurants shape diet variation in an urban carnivore
Data files
Dec 12, 2024 version files 40.48 MB
-
Data_Table_1.csv
91.42 KB
-
Data_Table_2.csv
91.34 KB
-
Data_Table_3.xlsx
14.71 KB
-
Data_Table_4.xlsx
18.10 KB
-
Data_Table_5.xlsx
15.43 KB
-
Data_Table_6.xlsx
19.79 KB
-
PERMANOVA_sample_FOO.rds
17.70 MB
-
PERMANOVA_sample_RRA.rds
17.77 MB
-
README.md
8.82 KB
-
SF-Coyote-Diet-Metabarcoding-main.zip
4.75 MB
Abstract
In the past decade, studies have demonstrated that urban and nonurban wildlife populations exhibit differences in foraging behavior and diet. However, little is known about how environmental heterogeneity shapes dietary variation of organisms within cities. We examined the vertebrate prey components of diets of coyotes (Canis latrans) in San Francisco to quantify territory- and individual-level dietary differences and determine how within-city variation in land cover and land use affect coyote diet. We genotyped fecal samples for individual coyote identification and used DNA metabarcoding to quantify diet composition and individual niche differentiation. The highest contributor to coyote diet overall was anthropogenic food followed by small mammals. The most frequently detected species were domestic chicken, pocket gopher (Thomomys bottae), domestic pig, and raccoon (Procyon lotor). Diet composition varied significantly across territories and among individuals, with territories explaining most of the variation. Within territories (i.e., family groups), the amount of dietary variation attributed to among-individual differences increased with green space and decreased with impervious surface cover. The quantity of anthropogenic food in scats also was positively correlated with impervious surface cover, suggesting that coyotes consumed more human food in more urbanized territories. The quantity of invasive, human-commensal rodents in the diet was positively correlated with the number of food services in a territory. Overall, our results revealed substantial intraspecific variation in coyote diet associated with urban landscape heterogeneity and point to a diversifying effect of urbanization on population diet.
README: Impervious surface cover and number of restaurants shape diet variation in an urban carnivore
SF-Coyote-Diet-Metabarcoding-main.zip
The zip file contains raw data files and scripts used to generate results for the above manuscript, which has been accepted in Ecosphere (December 5th, 2024). Below we provide descriptions of the files within each directory.
1. Sequence-Processing
Trimming-Reads.Rmd: script to trim raw fastq files with cutadapt. High performance computing required.
DADA2.Rmd: script correct amplicon errors and infer amplicon sequence variants (ASVs) with DADA2 denoising algorithm. High performance computing required.
2. Assign-Taxonomy
BLAST.Rmd: script to create a local BLAST database of 12SV5 sequences of vertebrate genera recovered from pilot studies. Use the blastn feature of BLAST+ to assign ASVs via the custom database and append or correct as needed with the nucleotide database of NCBI available online. 12S_reference_lib.fasta is the custom reference library of 12SV5 sequences of all vertebrate genera recovered from pilot studies.
3. Filtering-and-QC
FilteringReads.Rmd: script to manually filter denoised data. The input files batch1_reads.csv and *batch2_reads.csv * are ASV count tables output by DADA2 during sequence processing with taxonomy assigned by BLAST+ (output of step 1).
- SampleID: fecal sample identifier
- ASV: amplicon sequence variant identifier
- Sequence: amplicon sequence variant
- Kingdom/Phylum/Class/Order/Genus/Species: taxonomy
- FinalName: highest-resolution taxonomic classification for each ASV
- Flag: amplicon sequence variants flagged for removal
- pident: percent identity (from BLAST)
- qcovs: query coverage (from BLAST)
All_Metadata.csv provides the metadata associated with the samples.
- SampleID: fecal sample identifier
- Replicate: whether or not the sample is a PCR or extraction replicate
- Name: field collection identifier
- Site: location sample was collected
- Initials: initials of the sample collector
- Year/Month/Day: time of sample collection
- Lat/Long: sample collection location
- Sus.sp: suspected species origin as determined in the field
- Genotype: whether or not genotyping was successful
- Geno.Sp: species identified via genotyping
- Category: whether or not the sample comes from a coyote in San Francisco
- Individual: individual coyote identification
- Cytb_Sp: species identified via Cyt b analysis
NoCanis_Scats.csv: list of samples without Canis reads
Species_FunctionalGroups.csv: list of diet items categorized into functional groups
4. Data Visualization and Statistical Analyses
Creating figures or conducting statistical analyses requires that filtering and QC steps have been run as denoised and filtered data frames are required for visualizing and analyzing diet data.
Diet-Plots
Diet-Plots.Rmd: generate Figures 1 and 2 in the main text to visualize relative amounts of diet items in the population diet and among biological seasons, territories, and individuals. 12S_species_categories.csv categorizes diet items into general categories of species. IDs.csv assigns individual coyotes in family groups and sexes.
Regression-Analyses
Regression-Analyses.Rmd: script to generate (1) correlation matrix of land cover and land use covariates; (2) beta regression for RRA and quasibinomial GLM for FOO to test the effect of percent cover of impervious surfaces on the proportion and frequency of anthropogenic food in each coyote territory and the number of food services on the proportion and frequency of nuisance rodents in the diet in each coyote territory; and (3) generate the figure for the manuscript showing the correlation between land cover/land use and diet items.
Territory_All_Covariates.csv provides territory-level covariates
- Territory: territory name
- ISA: mean percent impervious surface cover in 1km buffer
- Urban: mean percent urban cover in 1km buffer
- Food.Services: number of food services in 1km buffer
- Pop.Den.2020: mean human population density in 1km buffer
- Housing.Den.2020: mean housing density in 1km buffer
iNEXT
iNEXT.Rmd: script to generate rarefaction curve plots and calculating diversity metrics and sample coverage for coyote territories and individuals.
nMDS
nMDS.Rmd: script to construct dissimilarity matrices and ordinate with non-metric multidimensional scaling to visualize dietary differences among biological seasons, territories, and individuals. To save time, model output is provided (sp.FOO.ind.rds, sp.FOO.rds, sp.RRA.ind.rds, sp.RRA.rds).
PERMANOVA
PERMANOVA.Rmd: script to conduct permutation-based multivariate analysis of variance tests to investigate differences in diet as a function of biological season and territory as well as among individuals and family groups.
SIMPER
SIMPER.Rmd: script to conduct similarity percentage analysis to assess which diet items contributed the most to observed differences in coyote diets among territories.
Replication-Analyses
Replication-Analyses.Rmd: script to calculate correlations between extraction replicate and PCR replicate sample pairs to assess the repeatability of results. Replicates_set1.csv and Replicates_set7.csv provide sequence reads for PCR replicates from sets 1 and 7 and **sp.RRA.clean.replicates.csv **provides filtered and cleaned diet data containing duplicate samples from extraction and PCR replication.
RInSp
Diet-Specialization.Rmd: script to calculate BIC/TNW and PSi metrics with the R package RInSp.
Model Output Files
PERMANOVA_sample_FOO.rds and PERMANOVA_sample_RRA.rds are model output from the PERMANOVA.Rmd script in the zip file described above. Because the model can take a long time to run, the model output is provided. Both files contains the results from 1,000 PERMANOVA trials following random subsampling of scats down to one sample per individual, for both frequency of occurrence (FOO) and relative read abundance (RRA) data, respectively. See the description in PERMANOVA.Rmd for further details.
Data Tables
Data_Table_1.csv
Organized results from 1,000 PERMANOVA trials following random subsampling of scats down to one sample per individual based on frequency of occurrence data (csv file of model output provided in PERMANOVA_sample_FOO.rds).
- Trial: nth trial of 1,000
- Covariate: biological season or territory
- F: Pseudo-F statistic
- R2: R2 value
- Pr(>F): p-value
Data_Table_2.csv
Organized results from 1,000 PERMANOVA trials following random subsampling of scats down to one sample per individual based on relative read abundance data (csv file of model output provided in PERMANOVA_sample_RRA.rds).
- Trial: nth trial of 1,000
- Covariate: biological season or territory
- F: Pseudo-F statistic
- R2: R2 value
- Pr(>F): p-value
Data_Table_3.xlsx
Contribution of dietary functional groups to coyote diet composition across territories in San Francisco, CA, summarized with relative read abundance and frequency of occurrence. Functional groups include: anthropogenic food, small mammals, medium-sized mammals, birds, herptiles, and marine mammals. Sheet 1 presents the data and sheet 2 lists the diet items included in each functional group.
Data_Table_4.xlsx
Contribution of diet items sorted into 26 categories to coyote diet composition across territories in San Francisco, CA, summarized with relative read abundance and frequency of occurrence. Sheet 1 presents the data and sheet 2 lists the diet items included in each category.
Data_Table_5.xlsx
Contribution of dietary functional groups to coyote diet composition across individual coyotes in San Francisco, CA, summarized with relative read abundance and frequency of occurrence. Functional groups include: anthropogenic food, small mammals, medium-sized mammals, birds, herptiles, and marine mammals. Sheet 1 presents the data and sheet 2 lists the diet items included in each functional group.
Data_Table_6.xlsx
Contribution of diet items sorted into 26 categories to coyote diet composition across individual coyotes in San Francisco, CA, summarized with relative read abundance and frequency of occurrence. Sheet 1 presents the data and sheet 2 lists the diet items included in each category.
Code/Software
R is required to run all Rmd scripts, except for Trimming-Reads.Rmd, which requires cutadapt, and BLAST.Rmd, which requires BLAST+. Scripts were created using R version 4.2.1 and BLAST+ version 2.13.0. Microsoft Excel* *can be used to view all data tables.