Kibale National Park Malaise trapping 2011-2012 data and analyses

Data on the Malaise trapping of ichneumonids in Kibale National Park, Uganda, in 2011-2012. Includes data on what ichneumonids were caught by Malaise trapping, where they were caught, and analyses (in R).<br><br>This fileset is associated with the article:Hopkins, Roininen &amp; Sääksjärvi (2018): Assessing the species richness of Afrotropical ichneumonid wasps with randomly placed traps provides ecologically informative data. <i>African Entomology</i> <b>26</b>: 350- 358. https://doi.org/10.4001/003.026.0350<br>Contains a bewildering number of files, of which only 'Uganda subfamily analyses 170512.R' will normally be of interest:<br><br><b>Analyses</b><br><br>- Uganda subfamily analyses 170512.R<br>The analyses used in the associated manuscript. Plots a NMDS visualisation of the differences in community composition between sites, estimates expected catches for each subfamily in forest and logged plantation with GLM:s, checks if the forest and logged plantation communities differ with both GLM:s and distance-based methods, and finally estimates the minimum sample size needed to get significant results. These analyses were run in the R software version 3.2.1, and require packages 'vegan' and 'mvabund'. Note that the working directory will need to be changed if used on a different computer.<br><br><i>R functions</i><br><br>- function simper2.R<br>R code for running a modified simper (similarity percentage) analysis. Conventional simper permutes trap sites, then calculates the average contribution of each subfamily to the average Bray-Curtis distance between forest and logged plantation. Modified simper also permutes trap sites, but uses the overall B-C distance (acquired by pooling the data) instead of averages. We found this worked better with our data, since conventional simper was confusing variable subfamilies with strongly contributing subfamilies. The function is used by 'Uganda subfamily analyses 170426.R'.<br><br>- functions p.boot p.sample boot.R<br>R code for three helper functions used by 'Uganda subfamily analyses 170426.R'. These are used to estimate the minimum sample size, and to calculate confidence intervals for the expected catches.<br><br><i>R saved results</i><br><br>Some of the analysis results from the file 'Uganda subfamily analyses 170426.R'. By default, the file loads these results instead of calculating them from scratch; this is because the calculations would take anything from a minute to several hours.<br><br>- Uganda manyglm results.RData<br>The GLM results, in variables 'anova.unadjusted' and 'anova.adjusted'.<br><br>- Uganda simper results.RData<br>The modified simper results, in variable 'si2.trap'. Also includes conventional simper (in variable 'si.trap') for comparison.<br><br>- Uganda resampling results.RData<br>The minimum sample size results, in variables 'n', 'p', 'n2', 'p2'.<br><br><b>Data</b><br><br>- Uganda individuals 170411.csv<br>The 1212 ichneumonid individuals caught by Malaise trapping, with each individual on a separate row. Does not include 36 individuals with missing or unclear information on their trap location. This file contains the actual original data; all other data files (i.e. the community matrices used in the analyses) are based on this data.<br><br>- Convert individual data to community matrices.R<br>R code for reformatting the data in 'Uganda individuals 170411.csv' into community matrices.<br><br><i>Community matrices</i><br><br>All values are count data. Note that Excel may misplace the first row; the files open OK in R however.<br><br>- Uganda subfamilies per sample 170411.csv<br>433 rows of samples, and 17 subfamilies in columns. Not used in the associated manuscript.<br><br>- Uganda subfamilies per trap 170411.csv<br>80 rows of trap sites, and 17 subfamilies in columns.<br><br>- Uganda subfamilies per site 170411.csv<br>8 rows of sites, and 17 subfamilies in columns.<br><br><i>Forest classes</i><br><br>Data on which trap site, site, successional class and forest class each row of the community matrices belongs to. The analyses mainly use the forest class, which is simply a classification into forest or logged plantation.<br><br>-Uganda forestclasses of each sample 170411.csv<br>Classification of the 433 samples. Not used in the associated manuscript.<br><br>-Uganda forestclasses of each trap 170411.csv<br>Classification of the 80 trap sites.<br><br>-Uganda forestclasses of each site 170411.csv<br>Classification of the 8 sites.<br><br>