Evolution of spatial indicators of soil health in response to climate change in Québec and Ontario

The surface area of soil with good agricultural potential is estimated at 2.3 Mha in Quebec (1.4%) and 7.8 Mha in Ontario (7.2%). Climate change is likely to lead to an intensification of agricultural production on current acreages and to an expansion of agriculture into regions where the current climate does not allow it. This intensification of land use will inevitably lead to increased mineralization of soil organic matter and degradation of soil structure.

In addition, soil erosion and runoff (sediment, nutrients and pesticides) into watercourses, as well as greenhouse gas emissions, are likely to increase significantly. However, the impact on crop productivity remains uncertain. It is therefore important to anticipate the nature and extent of changes in soil health and to implement soil conservation practices that will help mitigate these impacts and ensure soil productivity.

The primary aim of the project was to gain a better understanding of changes in soil health indicators and their spatial variability under the impact of climate change. Another goal was to validate soil conservation practices that will help mitigate the impacts of climate change and ensure the productivity of agricultural soils in Quebec and Ontario. To achieve these objectives, the research was conducted in three prongs: 

• Modelling the change in soil organic matter content and its spatial distribution in a future climate, until 2100, across Quebec and Ontario; 

• Developing and spatializing a soil health index for Ontario based on historical soil property data, from which the risks of soil degradation under the effecta of climate change will be quantified and mapped; and

• Measuring the effectiveness of conservation practices for controlling water erosion under current and future edaphic and climatic conditions, using rain simulation trials.

• First prong: Use of different SOM modelling approaches using deterministic machine learning based on soil analysis data from a private laboratory, baseline and future climate data provided by Ouranos, land use data (area under cultivation and crop type) from Financière agricole du Québec and Agriculture and Agri-Food Canada, as well as digital terrain model data. 

• Second prong: Soil maps from six Lake Erie regions were studied in greater depth to test and validate different approaches using the Disaggregation and Harmonisation of Soil Map Units Through Resampled Classification Trees (DSMART) algorithm to disaggregate their map units into soil series on a 30 m grid and apply the method to Ontario. 

• Third prong: Trials in experimental plots were carried out using rainfall simulators at two intensities, 61 and 71 mm/h, to represent reference and future climate  rainfall. Six treatments involving various soil covers, soybeans, tillage or no tillage and intercropping or no intercropping were compared.

Different machine learning or linear models predicted (based on an independent dataset) the distribution of SOM across Quebec with R2 values ranging from 0.42 to 0.49. The Random Forest model used to make these predictions includes, in order of importance, texture, the ratio of perennial crops, climatic indicators, and, lastly, digital terrain model related variables. In the early 2100s, soils in the colder regions of Quebec will experience a more significant decline in soil organic matter content, and the surface area planted with perennial crops will decrease (Figure 1). However, decreases in soil organic matter in these regions will be attributed more to global warming than to changes in crops. Soil texture also plays an important role. As organic matter content is generally lower in sandy soils than in clay soils, future climate dynamics will affect them differently from region to region. 

The different anticipated GHG emission scenarios (RCP 4.5 and 8.5) will have less effect. In fact, they will be of the same order of magnitude as the change in land use from perennial to annual crops. The application of the climate model on the scale of Ontario did not produce such credible results in terms of initial SOM contents and changes in soil organic matter due to poorly represented climatic domains and different data sources for the assessment of the ratio of perennial crops.

Figure 1: Effects of texture (a) and agroclimatic zone (b) on projected soil organic matter levels in Quebec from 2010 to 2100 under two GHG scenarios (RCP 4.5 and 8.5) and the ratio of perennial crops.

Lastly, work on the disaggregation of soil map units at the Ontario-wide scale has produced a unified map of geolocated soil series and a probable map of soil organic matter content on a 250 m grid with an acceptable level of uncertainty.

Demonstration trials conducted on experimental plots using rainfall simulators showed that climate change will significantly increase surface runoff (48%), soil erosion (48%) and nutrient losses in soils (from 42% to 86% depending on the element). In soybean production, soil conservation practices (no-till and cover crop intercropping) have the potential to control increased soil and nutrient losses in a future climate.

This project is funded by the MAPAQ under the Programe Innov'Action agroalimentaire.

• Université du Québec à Chicoutimi

• Agriculture and Agri-Food Canada

• Ministère de l’Agriculture, des Pêcheries et de l’Alimentation du Québec

• Ontario Ministry of Agriculture, Food and Rural Affaires

• Agro Enviro Lab

• Ausable Bayfield Conservation Authority

550028