The need for near-term ecological forecasts  

Weather forecasts influence the decisions of millions of people every day, yet there are few forecasts of ecological phenomena to assist environmental decisions. While early weather forecasts inspired little confidence, over the past half a century or more, meteorologists have been producing forecasts and continually improving their precision, accuracy and duration of the forecast period. Adopting this iterative cycle of making forecasts and confronting them with new data has facilitated this learning and improvement. Ecology needs to do the same.

In a time of planetary-scale environmental crises, more than ever, managers and policymakers require evidence to inform their decisions, and tools to learn from the outcomes. Few decisions are once-off (should you burn, cull, harvest, restore, implement or retract a policy, etc.), and most are made repeatedly at some time-step (e.g. hourly, daily, monthly, seasonally, annually, decadally). While one should always evaluate and learn from the outcome of your decision, this does not always happen…

Iterative near-term ecological forecasts, with fully specified uncertainties, provide a framework to inform decisions and learn from outcomes. By collecting new data to evaluate the outcomes of decisions against the original model forecasts, we can update our prior knowledge and improve our ability to forecast and make decisions in the next time step. The new data can tell us whether our forecasts or decisions were any good, and whether we need to refine or replace our model, re-evaluate our assumptions or ecological understanding, consider additional scenarios or inputs, etc.

While early iterations of new forecasts may have low skill, the learning cycle ensures they will rapidly improve, and there are many other benefits. Forecasts aside, just the bioinformatics pipeline required to support iterative forecasts should be valuable to decision-makers in that they provide rapid access to the most recent data. This also adds value for long-term research and monitoring programmes in that new data are interrogated as they are collected, allowing platform managers to rapidly identify issues with equipment or data collection protocols. Ultimately, as the forecasts improve, they can support adaptive monitoring, highlighting key data needs required to reduce uncertainty in the forecasts.

An ecological forecasting community of practice  

The Ecological Forecasting Initiative (EFI; ecoforecast.org) was established in the USA in 2018 as an international grassroots consortium aimed at building and supporting an interdisciplinary community of practice around near-term (daily to decadal) ecological forecasts. EFI chapters have since been established in Canada, Oceania and Europe, and now in Africa.

Ecoforecast Africa (ecoforecast.africa), the recently formed African chapter, is in its founding phase, but has already grown to >60 members from across Africa and the world in just a few months (see ecoforecast.africa/join). Ecoforecast Africa had its first community call in May and ran its inaugural training event in Cape Town, South Africa, 21–25th July. The short course on iterative near-term ecological forecasting included 27 participants from South Africa, Kenya, Nigeria, Zimbabwe, Japan, Switzerland and the USA, including students, academics, decision-makers and consultants from several universities, conservation agencies, NGOs and environmental consultancies.