Using automated web scraping across 11 Australian e-commerce platforms between 2020 and 2023, we identified 3,822 plant species for sale and categorized them by regulatory, naturalisation, and native status, flagging emerging invasive species risks and the online platforms most in need of monitoring.
Analyzing roughly 14,000 online sales listings covering 235 parrot and songbird species, we found 55% of traded parrot species and 64% of traded songbird species were non-native to Australia, with most at extreme risk of establishing invasive populations. Traded abundance was best predicted by breeding and handling requirements, availability, and appearance.
Despite state and territory bans, invasive plants are widely available on Australian e-commerce platforms. Across 235,162 plant advertisements collected over 12 months, we found 155 plant taxa being sold that are prohibited from trade in at least one Australian state or territory.
The online trade of live invertebrates in Australia is largely unregulated and unmonitored. Tracking listings over 12 months, we found 264 traded invertebrate species spanning 71 families and 168 genera, led by the native stick insect Extatosoma tiaratum, with stick insects, ants, and tarantulas the most popular groups.
Using web scrapers across 12 Australian e-commerce websites over 14 weeks, we identified 1,192 species being traded as pets within Australia, including 667 non-native species and 81 threatened species, providing the first broad snapshot of the scale and composition of Australia’s online pet trade.
We searched roughly 2 million dark web advertisements over five years using ~7,000 wildlife trade-related search terms, finding 153 species traded across 3,332 advertisements. The large majority of these listings were for species used as recreational drugs (e.g. psychedelic mushrooms), rather than for the live pet or wildlife-product trade typically associated with the dark web.
We compiled a dataset consisting of all the species involved in the illegal wildlife trade along with the reason (i.e., use-type) they were being traded. In total, the dataset includes c. 4.9k distinct taxa representing c. 3.3k species and contains c. 11k taxa-use combinations from 110 unique use-types. Our dataset can be used to conduct large-scale broad searches of the Internet to find illegally traded wildlife.
The Internet can be vast source of data for the wildlife trade. However, data collected from the Internet is often numerous and messy, making data cleaning a task the requires a lot time and effort. Here, we tested if text classification can be used to speed up the process of data cleaning in relation to online data collected on the wildlife trade. We found that text classification models can predict with great accuracy relaxant advertisements, including the taxonomy of relevant species, using the text found in online advertisements. We recommend using text classification as a method to make data cleaning more efficient. Future efforts should try to pair text classification with image classification for improved efficiency.
The internet is a vast source of wildlife trade data. Here, we present an accessible guide for Internet‐based wildlife trade surveillance, which uses a repeatable and systematic method to automate data collection from relevant websites. Our guide is adaptable to the multitude of trade‐based contexts including different focal taxa or derived parts, and locations of interest.