Data analysis
Sample-size-based rarefaction curves with extrapolation (Chao et al., 2014) were created using the iNEXT package in R (Hsieh et al., 2016) to examine species richness across the three iDNA datasets and from the camera trapping data. For the iDNA data, we converted species read abundance to species incidence data (presence/absence) at the sample level. Due to differences in the number of amplified samples across the three iDNA sources, species richness was extrapolated to the iDNA source with the greatest number of samples following Chao et al. (2014). For the camera trapping data, we measured species richness across the total number of camera days from all deployed cameras.
We used a relative abundance index (RAI) to compare the biodiversity found using iDNA metabarcoding data and camera trap data. For metabarcoding data, the RAI is equal to the sum of occurrences for species i divided by the total number of pooled samples. The RAI measured from camera trapping data is calculated as the number of species i events divided by the total number of camera trap nights. A single species event from camera trap data was defined by any individual (or group of individuals) from one species captured by a camera over a one-hour time frame. We then used RAI to visualize the gamma diversity according to both iDNA and camera trapping methods by binning species according to important life history traits to compare the effectiveness of the biomonitoring methods in describing the gamma diversity of different taxa groups. To compare RAI across iDNA and camera traps, we only included sites that were sampled by both methods.
After comparing diversity from iDNA and camera traps, we then compared the iDNA sources to each other to assess the similarities, efficiencies, and potential biases of each sampler type. We used a Bray-Curtis dissimilarity matrix of sites x species and applied a PERMANOVA test using the adonis function to test for significance in the separation of the vertebrate community based on iDNA sampler. We then used the vegan package in R (Oksanen et al., 2013) and themetaMDS function to visualize dissimilarities in species composition at sites using non-metric multidimensional scaling (NMDS). Each site was designated three rows in our abundance matrix: one for each iDNA type (carrion fly, sandfly, or mosquito). We also utilized theordiellipse function to visualize the spread of the data based on iDNA source with ellipses drawn to two standard deviation (95% confidence intervals) from the group mean. Lastly, significant species vectors were calculated using the envfit function.
To answer whether an iDNA sampler could replace another iDNA sampler in profiling alpha diversity, we calculated RAI of species at the site level for each of the iDNA datasets and compared species occurrences at sites from the different iDNA samplers. We visualized the results in scatterplots. If two iDNA datasets showed similar species composition at sites, most data points would fall along an isometric line. If two iDNA samplers revealed dissimilar species composition across sites, more data points would fall along the x-axis or y-axis revealing occurrence of a species in one of the iDNA datasets and an absence of that species in the other iDNA dataset. Finally, to further explore the different feeding ecologies associated with the insect taxa used as iDNA samplers, we compared the RAI at sites for the most abundant species in each iDNA dataset using boxplots.