The future is not like the past
Author: Peter Morrison
Introduction
When researchers study ecosystem services they are often interested in how land uses have changed in the past and how they might change in the future, and as a result how the services provided shift in location and amount. More generally, just as researchers choose the spatial extent and grain for their studies, they also choose the temporal extent and grain, asking questions such as: How far back in time should I go? Do I only care about annual changes, or do I want to know about changes within years?
Some ResNet projects have studied historical changes over several decades or centuries (e.g., Venter et al., 2016 [7]; Sherren et al., 2021 [8]), and in one case back five thousand years (Hillis et al., 2022 [9]). Similarly, some projects are focusing on possible future changes in ecosystem services resulting from human actions. For example, one research team is studying restoration of vacant lots in Quebec City (e.g., Mendes et al. 2024 [10]). Another group is studying the possibility of agriculture in the Northwest Territories 40-60 years in the future (Krishna KC et al., 2021 [11]; Lemay et al., 2021 [12]).
The tools and approaches for assessing ecosystem services in the past are not the same as those for projecting into the future – and the methods for studying the recent past or the near future are not the same as those for understanding times more distant from the present. For example, global carbon emissions are well known in the recent past, but the future trajectory is very uncertain and strongly depends on human actions. Especially over large areas, data from satellite sensors has been essential to understanding changes over time, yet accessible satellite data only extend back to 1972 (NASA). Before that, other data sources such as aerial photos and historical records, and methods, such as participatory GIS methods with knowledge holders (e.g., Brown and Fagerholm, 2015 [13]), must be used. And when studying the social aspects of ecosystem services, for example through interviews, we can only obtain input from people who are still alive.
This creates significant challenges when scaling up in time, that is, moving from a short duration study to a longer time period, and when combining assessments of past and future services.
Case study
The following case study illustrates some of the challenges with scaling up in time, and the differences between scaling up information from the past and predictions for the future.
People obtain ecosystem services when they harvest wild plants (such as berries or mushrooms) from forested land. Such harvests go back hundreds of years in Canada, first by Indigenous peoples, then by settlers and now by a mix of different users (Kumar et al., 2019 [14]; MacKinnon et al., 2015 [15]; Turner, 2020 [16]). With the adoption and spread of industrial agriculture, and the reduction in forested land, the relative importance of such food sources in Canadian diets has dropped.
Here, we create a “snapshot” of the present – to understand the ecosystem services from harvesting wild plants found in Canada’s forests. For this case study, we have focussed just on the Montérégie region of Quebec, but have considered the methods used to scale up in time – either back into the past or forward into the future. [Note: this case study excludes trees harvested for timber or fuel wood, and maple syrup.]
This region has evolved dramatically over the last five centuries, with changes in control of land from Indigenous people to settlers, growth in human population, shifts in land use, forest clearing, establishment of increasingly industrialized agriculture, urbanization and development of an extensive road network.
Best practices and opportunities
This case study highlights the potential role for several best practices. These include:
- Finding ways to combine information from multiple sources. This will be particularly important when scaling up to longer durations, when a greater variety of data sources and methods need to be used. For example, unconventional sources (e.g. historical photographs, old herbarium records) may be helpful. Bayesian methods can help assign weights to different sources, while retaining estimates of the uncertainty and flexibility in terms of the evolution of functional relationships.
- Using mapping techniques, such as participatory GIS methods, to identify where current beneficiaries are in relation to the locations and services they value.
- Recognizing that the future is not simply a linear extension from the past. This might mean deliberately including decision makers and beneficiaries of ecosystem services in research related to the future, through techniques such as scenarios. It might also include modelling human learning and adaptive management as part of predictions. Some researchers have recommended considering longer time extents than the focus of current decisions (e.g., Lyon et al., 2022 [17]).