This code uses pathway modelling to look at correlations of exotic plant invasion in tropical rainforest remnants and continuous sites. Partial least squares path-modelling looks at correlations between latent variables that are informed by measured variables. The code examines the relative influence of landscape-level fragmentation, local forest disturbance, propagule pressure, soil characteristics and native community composition on invasion. The total native community is examined first. Then subsets of the native community are modelled separately, adult trees, tree saplings, tree seedlings and ground vegetation. The relationship between the native and exotic communities was tested in both directions. Full details about this application can be found at https://doi.org/10.5285/adbf6d29-ee7b-4dd1-9730-11d2308d526c

This resource contains the R code and core results of a study seeking to identify whether there are global patterns in whether larger or smaller bodied species are showing different population trends within communities. This used the global BioTIME database (including community time series from hundreds of studies of mostly covering the last 20-50 years) and several large trait databases to gain a very large sample size (12,956 assemblage time series from 144 studies, incorporating 2,109,593 observations of 10,286 species, of which 7,234 could be linked to at least one size trait). Data resources in this deposit include matched trait values for each species, several population trends of each species, and community level correlations between population trends and body-size trait values. Additionally, html files describing the R markdown code to produce the data resources are included. This resource does not contain the raw population and trait data, which are openly available from various sources that are listed in the supporting documentation. The matching between global databases required a large amount of initial cleaning and filtering steps. Although the data was subject to a number of checks, with tens of thousands of species, it was too large to check all alignments manually, and trait matches assume a single bodysize value for a species across its range. The original purpose was to generate a relative rank within a community, but caution is needed for more fine-grained analyses using this approach. Full details about this application can be found at https://doi.org/10.5285/1f7687de-d68e-4349-b4d3-b7d3a127a7df

This is an application providing code for the non-parametric comparison of soil depth profiles, and testing for significant differences between soil depth profiles, using bootstrapped Loess (local) regressions (BLR). The BLR approach was developed to be able to compare and test for significant differences in potentially non-linear depth profiles of soil properties across land use transitions, which does not need to meet any parametric distribution assumptions, and is intended to be generally applicable regardless of specific contexts of land use and soil type. A small dataset is provided with the code to demonstrate the BLR approach and its outputs. The code was written using the R statistical programming language and provides two examples of the BLR approach. This application was created by the Centre for Ecology & Hydrology at Lancaster in 2015 during the ELUM (Ecosystem Land Use Modelling & Soil Carbon GHG Flux Trial) project, which was commissioned and funded by the Energy Technologies Institute (ETI). Full details about this application can be found at https://doi.org/10.5285/d4f92cd8-43e8-49e4-8f9e-efcc0e3b2478

The speciesRecordTools R package contains functions for examining the distribution of species records, understanding sampling trends and potential biases, and building correlative presence-background species distribution models for prediction of the distribution of species across the landscape. The package is built to work with the Environmental Record Centre for Cornwall and the Isles of Scilly's (ERCCIS) opportunistic species records. Full details about this application can be found at https://doi.org/10.5285/030b49f4-9e1f-46e9-ad98-157d8668a517

This R application is an implementation of state tagging approach for improved quality assurance of environmental data. The application returns state-dependent prediction intervals on input data. The states are determined based on clustering of auxiliary inputs (such as meteorological data) made on the same day. The method provides contextual information to assess the quality of observational data and is applicable to any point-based, daily time series observational data. To use this application, the user will need to input two separate csv files: one for state variables and the other for observations. This work was supported by the Natural Environment Research Council award number NE/R016429/1 as part of the UK-SCAPE programme delivering National Capability. Full details about this application can be found at https://doi.org/10.5285/1de712d3-081e-4b44-b880-b6a1ebf9fcd8

This dataset consists of computer code transcripts for two proprietary flood risk models from a study as part of the NERC Rural Economy and Land Use (RELU) programme. This project was conceived in order to address the public controversies generated by the risk management strategies and forecasting technologies associated with diffuse environmental problems such as flooding and pollution. Environmental issues play an ever-increasing role in all of our daily lives. However, controversies surrounding many of these issues, and confusion surrounding the way in which they are reported, mean that sectors of the public risk becoming increasingly disengaged. To try to reverse this trend and regain public trust and engagement, this project aimed to develop a new approach to interdisciplinary environmental science, involving non-scientists throughout the process. Examining the relationship between science and policy, and in particular how to engage the public with scientific research findings, a major diffuse environmental management issue was chosen as a focus - flooding. As part of this approach, non-scientists were recruited alongside the investigators in forming Competency Groups - an experiment in democratising science. The Competency Groups were composed of researchers and laypeople for whom flooding is a matter of particular concern. The groups worked together to share different perspectives - on why flooding is a problem, on the role of science in addressing the problem, and on new ways of doing science together. We aimed to achieve four substantive contributions to knowledge: 1. To analyse how the knowledge claims and modelling technologies of hydrological science are developed and put into practice by policy makers and commercial organisations (such as insurance companies) in flood risk management. 2. To develop an integrated model for forecasting the in-river and floodplain effects of rural land management practices. 3. To experiment with a new approach to public engagement in the production of interdisciplinary environmental science, involving the use of Competency Groups. 4. To evaluate this new approach to doing public science differently and to identify lessons learnt that can be exported beyond this particular project to other fields of knowledge controversy. This dataset consists of computer code transcripts for two proprietary flood risk models. Flood risk and modelling interview transcripts from this study are available at the UK Data Archive under study number 6620 (see online resources). Further documentation for this study may be found through the RELU Knowledge Portal and the project's ESRC funding award web page (see online resources).

Two scripts for classifying remotely sensed data used to produce maps of peatland distribution and predicted peat thickness, using random forest classification and regression. Written in JavaScript for use with Google Earth Engine. These are versions of the scripts used in Hastie et al. (2022), https://doi.org/10.1038/s41561-022-00923-4. Users should also cite Rodríguez-Veiga et al. (2020), https://doi.org/10.3390/rs12152380 . Full details about this application can be found at https://doi.org/10.5285/e337de58-df5e-4412-8aef-28875870f965

MultiMOVE is an R package that contains fitted niche models for almost 1500 plant species in Great Britain. This package allows the user to access these models, which have been fitted using multiple statistical techniques, to make predictions of species occurrence from specified environmental data. It also allows plotting of relationships between species' occurrence and individual covariates so the user can see what effect each environmental variable has on the specific species in question. The package is built under R 3.1.2 and depends on R packages 'leaps', 'earth', 'fields', 'mgcv', 'stringr', 'gsubfn', 'randomForest' and 'nnet'. Full details about this application can be found at https://doi.org/10.5285/94ae1a5a-2a28-4315-8d4b-35ae964fc3b9

This resource provides code to fit a transmission model of bovine tuberculosis spread to a population of wild badgers in Woodchester Park in the UK. The code produces Markov chain Monte Carlo samples from a model fitted to individual-level badger data from Woodchester Park. The badger data came from and can be requested from the Animal and Plant Health Agency. Example outputs are provided in the Supporting Information. This code was developed as part of a Natural Environment Research Council funded grant (number NE/V000616/1). Full details about this application can be found at https://doi.org/10.5285/fe0f6bd7-ffd2-4e21-8c84-493cf4f3080d

A collection of python and bash scripts to implement, train and deploy a generative adversarial network for population genetic inferences. The networks have been tuned to be deployed to genomic data from Anopheles mosquitoes. However, the general framework can be applied to other species. It requires the input data to be in Variant Call Format (VCF) format and the simulations need to be in msprime format. Full details about this application can be found at https://doi.org/10.5285/3ae572f6-4862-47ae-b4a0-4b9c496b5b54