Data and code for: "Natural History Collections at the Crossroads: Shifting Priorities and Data-Driven Opportunities"

# Wombat Value of Information (VOI) Analysis This repository contains the code and data for the manuscript "Natural History Collections at the Crossroads: Shifting Priorities and Data-Driven Opportunities" Owen Forbes1*, Peter H. Thrall1, Andrew G. Young1, Cheng Soon Ong2 1 CSIRO National Research Collections Australia, Canberra, ACT 2601, Australia 2 CSIRO Data61, Canberra, ACT 2601, Australia *Corresponding author: [e-mail] ## Repository Overview This project applies a Value of Information (VOI) analytical framework to wombat occurrence data from GBIF. It demonstrates a novel research prioritisation approach that integrates: - **Value of Information (VOI)**: Expected information gain from additional observations- **Need for Information (NFI)**: Habitat condition/loss metrics- **Cost of Information (COI)**: Remoteness areas affecting sampling feasibility The analysis uses a binomial model as a simple species distribution model, to evaluate the research value of new wombat observations across different locations in NSW and ACT, Australia. ## Repository Structure The repository consists of one main Quarto (.qmd) script and associated data files: - `Wombats_VOI.qmd`: Primary analysis script containing data cleaning, modelling, and visualisation ## Data Files ### GBIF Wombat Occurrence Data- `0014630-250127130748423.csv` - GBIF wombat occurrence export (download available from: https://www.gbif.org/occurrence/download/0014630-250127130748423 ) ### Environmental and Administrative Data- `HCAS31_AHC_2020_2022_NSW_50km_3577.tif` - Habitat condition raster for NSW (from CSIRO Habitat Condition Assessment System - subset of full dataset available from https://data.csiro.au/collection/csiro%3A63571v7 )- `RA_2021_AUST_GDA2020` - ABS Remoteness Areas 2021 shapefile (folder containing multiple files - available from https://www.abs.gov.au/statistics/standards/australian-statistical-geography-standard-asgs-edition-3/jul2021-jun2026/access-and-downloads/digital-boundary-files Specficic shapefile URL: https://www.abs.gov.au/statistics/standards/australian-statistical-geography-standard-asgs-edition-3/jul2021-jun2026/access-and-downloads/digital-boundary-files/RA_2021_AUST_GDA2020.zip ) ## Requirements - R (version 4.3.2 or later)- Required R packages: - tidyverse (v2.0.0) - for data manipulation and visualization - sf (v1.0-16) - for spatial data handling - lubridate (v1.9.3) - for date handling - raster (v3.6-26) - for raster data handling - ozmaps (v0.4.5) - for Australian state boundaries - terra (v1.7-71) - for efficient raster processing - exactextractr (v0.10.0) - for exact extraction from raster to polygons - patchwork (v1.2.0) - for combining plots - corrplot (v0.92) - for correlation matrix visualization - RColorBrewer (v1.1-3) - for color palettes - gridExtra (v2.3) - for arranging multiple plots - grid - for low-level graphics (base R) Install these packages before running the script. ## Analysis Process The analysis follows these key steps: 1. **Data Cleaning** - Filter to NSW and ACT wombat records - Remove records with high uncertainty (>10km) - Create a 50km grid system across NSW and ACT 2. **Temporal Analysis** - Calculate presence/absence of wombats in each grid cell by year - Determine empirical observation probability for each grid cell - Analyse temporal stability using 5-year sliding windows 3. **Value of Information Calculation** - Calculate expected information gain through KL divergence - Simulate adding new observations using the binomial model - Map VOI across NSW and ACT 4. **Need for Information Integration** - Import and process habitat condition data - Convert to habitat loss metric (NFI) - Integrate with VOI analysis 5. **Cost of Information** - Process remoteness area data as a proxy for sampling cost - Map remoteness areas across the study region 6. **Combined Analysis** - Create quadrant analysis combining VOI and NFI - Generate comprehensive visualisations of all three metrics ## How to Use 1. Download this repository to your local machine.2. Set your working directory to the location of the script.3. Ensure all required R packages are installed.4. Run the script in RStudio or your preferred R environment. ## Outputs The primary output is `combined_quadrant_analysis_100425.png`, which displays four panels:1. Value of Information (VOI) - Expected information gain across the study area2. Need for Information (NFI) - Habitat loss percentiles3. Cost of Information (COI) - Remoteness areas as a proxy for sampling cost4. VOI vs NFI Quadrant Analysis - Relationship between information value and need ## Citation If you use this code or methodology, please cite: Forbes, O., Thrall, P.H., Young, A.G., Ong, C.S. (2025). Natural History Collections at the Crossroads: Shifting Priorities and Data-Driven Opportunities. [Journal information pending]# 袋熊信息价值（Value of Information, VOI）分析 本仓库包含论文《Natural History Collections at the Crossroads: Shifting Priorities and Data-Driven Opportunities》对应的代码与数据集。 欧文·福布斯1*, 彼得·H·思罗尔1, 安德鲁·G·杨1, 郑顺旺2 1 澳大利亚联邦科学与工业研究组织（Commonwealth Scientific and Industrial Research Organisation, 简称CSIRO）国家馆藏研究中心，堪培拉，澳大利亚首都领地2601；2 澳大利亚联邦科学与工业研究组织Data61数据科学实验室，堪培拉，澳大利亚首都领地2601；*通讯作者：[e-mail] ## 仓库概览 本项目将**信息价值（Value of Information, VOI）**分析框架应用于来自全球生物多样性信息设施（Global Biodiversity Information Facility, 简称GBIF）的袋熊出现数据，展示了一种整合以下三类要素的新型研究优先级确定方法： - **信息价值（Value of Information, VOI）**：额外观测带来的预期信息增益 - **信息需求（Need for Information, NFI）**：生境状况/损失指标 - **信息成本（Cost of Information, COI）**：影响采样可行性的偏远区域等级 本分析采用二项分布模型作为简易物种分布模型，用于评估澳大利亚新南威尔士州（New South Wales, 简称NSW）与澳大利亚首都领地（Australian Capital Territory, 简称ACT）内不同区域的新增袋熊观测数据的研究价值。 ## 仓库结构 本仓库包含1个主Quarto（.qmd）脚本与配套数据集文件： - `Wombats_VOI.qmd`：核心分析脚本，涵盖数据清洗、建模与可视化流程。 ## 数据集文件 ### GBIF袋熊出现数据 - `0014630-250127130748423.csv`：GBIF袋熊出现数据导出文件（下载链接：https://www.gbif.org/occurrence/download/0014630-250127130748423） ### 环境与行政数据 - `HCAS31_AHC_2020_2022_NSW_50km_3577.tif`：新南威尔士州生境状况栅格数据（源自澳大利亚联邦科学与工业研究组织生境状况评估系统，完整数据集子集可从https://data.csiro.au/collection/csiro%3A63571v7获取） - `RA_2021_AUST_GDA2020`：澳大利亚统计局（Australian Bureau of Statistics, 简称ABS）2021年偏远区域矢量shapefile（包含多个文件的文件夹，可从https://www.abs.gov.au/statistics/standards/australian-statistical-geography-standard-asgs-edition-3/jul2021-jun2026/access-and-downloads/digital-boundary-files获取，具体shapefile下载链接：https://www.abs.gov.au/statistics/standards/australian-statistical-geography-standard-asgs-edition-3/jul2021-jun2026/access-and-downloads/digital-boundary-files/RA_2021_AUST_GDA2020.zip） ## 依赖要求 - R（版本4.3.2及以上） - 所需R包： - tidyverse（v2.0.0）：用于数据处理与可视化 - sf（v1.0-16）：用于空间数据处理 - lubridate（v1.9.3）：用于日期数据处理 - raster（v3.6-26）：用于栅格数据处理 - ozmaps（v0.4.5）：用于获取澳大利亚州界矢量数据 - terra（v1.7-71）：用于高效栅格处理 - exactextractr（v0.10.0）：用于栅格数据到多边形的精准提取 - patchwork（v1.2.0）：用于图形拼接 - corrplot（v0.92）：用于相关矩阵可视化 - RColorBrewer（v1.1-3）：用于配色方案设置 - gridExtra（v2.3）：用于多图排版 - grid（基础R包）：用于底层图形绘制 运行脚本前请先安装上述依赖包。 ## 分析流程 本分析遵循以下核心步骤： 1. **数据清洗** - 筛选新南威尔士州与澳大利亚首都领地的袋熊观测记录 - 移除定位不确定性大于10km的记录 - 在新南威尔士州与澳大利亚首都领地范围内构建50km格网系统 2. **时序分析** - 按年份计算每个格网单元内袋熊的出现/未出现情况 - 计算每个格网单元的经验观测概率 - 采用5年滑动窗口分析时序稳定性 3. **信息价值计算** - 通过KL散度计算预期信息增益 - 基于二项分布模型模拟新增观测数据 - 绘制新南威尔士州与澳大利亚首都领地范围内的信息价值空间分布图 4. **信息需求整合** - 导入并处理生境状况数据 - 转换为生境损失指标（即信息需求NFI） - 与信息价值分析进行整合 5. **信息成本分析** - 将偏远区域数据作为采样成本的代理变量进行处理 - 绘制研究区域内的偏远区域空间分布图 6. **综合分析** - 构建整合信息价值与信息需求的象限分析图 - 生成三类指标的综合可视化成果 ## 使用方法 1. 将本仓库下载至本地设备 2. 将工作目录设置为脚本所在路径 3. 确认所有依赖R包已安装完成 4. 在RStudio或其他偏好的R环境中运行该脚本 ## 输出成果 核心输出成果为`combined_quadrant_analysis_100425.png`，该图包含四个子面板： 1. 信息价值（VOI）：研究区域内的预期信息增益分布 2. 信息需求（NFI）：生境损失百分位数分布 3. 信息成本（COI）：作为采样成本代理变量的偏远区域分布 4. 信息价值与信息需求象限分析：信息价值与信息需求的关联关系 ## 引用说明 若使用本代码或研究方法，请引用以下文献： Forbes, O., Thrall, P.H., Young, A.G., Ong, C.S. (2025). Natural History Collections at the Crossroads: Shifting Priorities and Data-Driven Opportunities. [Journal information pending]