Next Generation Biomonitoring of change in ecosystem structure and function

Our vision is to develop and test a generic NGB approach that will detect ecosystem-wide change more rapidly, sensitively and cheaply than current biomonitoring. Using a unique combination of Next-Generation Sequenced DNA data and Machine Learning, NGB will reconstruct species interaction networks to identify change in ecosystem properties, revolutionising both our understanding of ecosystems and our ability to predict and mitigate global change.

A Revolution in Biomonitoring

Machine Learning
Machine learning offers huge potential for reconstructing ecological networks from available data. Statistical inference and logic-based machine learning approaches have been widely used in the social sciences and for molecular and genetic interaction networks, but have only recently been applied in ecology. The idea behind these machine-learning methods is simple; embedded in a dataset is the imprint of the recent processes and interactions that created the data, and this information can be recovered to reconstruct networks.
Stacks Image 35
DNA as data
We have shown that it is possible to quickly learn valid ecological networks from classical ecological sample data (functional network above right). Machine learning methods can therefore do science and greatly reduce the costs of building networks over traditional means. However, to use network learning for biomonitoring in any ecosystem, it is necessary to move away from classical sampling data to a generic source of ecological data.
Next Generation Sequencing, or NGS, describes a number of similar protocols for generating large numbers of nucleic acid sequences for the identification of species (operational taxonomic units or OTUs) and functions. The great beauty of these methods is that nucleic acids data are common to all life forms and ubiquitous. In principle, NGS can be applied to the identification of OTUs and functions in environmental samples from any biome, habitat, and environment and any source material with minimal change in protocol. Our argument is that if coupled together, machine learning and NGS data could serve as the foundation for a global, generic, and rapid network-based biomonitoring system that requires relatively little refinement to fit the environmental context in which it is deployed.
Stacks Image 42

Impact and Benefits

Real-time NGB would be far more sensitive than current approaches, and the window for identifying when a stressor first elicits a response could be greatly foreshortened. The approach could solve the three key problems, of accuracy, cost and generality of current approaches to biomonitoring. By broadening biomonitoring from single indicators, the approach enrichens the evaluation of change in ecosystem structure and function. The costs of time and effort in reconstructing networks and, as a consequence, detecting change in an ecosystem are markedly reduced. In turn, NGB is general because it is based on sampling ubiquitous nucleic acids.
Importantly, fusing NGS and machine learning allows us to learn Ecology. The functional ecological network, above, demonstrated that machine learning can be used for ‘hypothesis-test’ science about ecosystem structure. The fusion should therefore allow reconstruction of networks for ecosystems about which we have very little understanding and discover novel interactions within the ecosystems we already work in. NGB will also work well with existing large-scale monitoring approaches, such as remote sensing of the Earth’s environments.
Traditionally, there has been relatively little exchange and cross-fertilization between the disciplines of biomonitoring and ecology. A shift towards a large-scale biomonitoring approach that measures change in terms of network structure would provide richer, more ecological information that is moreover comparable between ecosystems. NGB data would add to our knowledge explicitly and foster a revolution in ecological understanding of ecosystem change.


French language website