Di_Vittorio_et_al_2022_GCB

Doubling protected land area may be inefficient at preserving the extent of undeveloped land and could cause substantial regional shifts in land use

Alan V. Di Vittorio^1*,Kanishka B. Narayan ² ,Pralit Patel ², Katherine Calvin ² & Chris R. Vernon ²

¹ Lawrence Berkeley National Laboratory, Berkeley CA

² Joint Global Change Research Institute, Pacific Northwest National Lab, Washington DC, USA

* corresponding author: avdivittorio@lbl.gov

Abstract

Land change projection is highly uncertain yet drives critical estimates of carbon emissions, climate change, and food and bioenergy production. We use new, spatially-explicit land availability data in conjunction with a model sensitivity analysis to estimate the effects of additional land protection on land use and cover. The land availability data includes protected land and agricultural suitability and is incorporated into the Moirai land data system for initializing the Global Change Analysis Model (GCAM). Overall, decreasing land availability is relatively inefficient at preserving undeveloped land while having considerable regional land use impacts. Current amounts of protected area have little effect on land and crop production estimates, but including the spatial distribution of unsuitable (i.e., unavailable) land dramatically shifts bioenergy production from high northern latitudes to the rest of the world, as compared to uniform availability. This highlights the importance of spatial heterogeneity in understanding and managing land change. Approximately doubling current protected area to emulate a 30% protected area target may avoid land conversion by 2050 of less than half the newly protected extent while reducing bioenergy feedstock land by 10.4% and cropland and grazed pasture by over 3%. Regional bioenergy land may be reduced (increased) by up to 46% (36%), cropland reduced by up to 61%, pasture reduced by up to 100%, and harvested forest reduced by up to 35%. Only a few regions show notable gains in some undeveloped land types of up to 36%. Half of the regions can reach the target using only unsuitable land, which would minimize impacts to agriculture but may not meet conservation goals. Rather than focusing on an area target, a more robust approach may be to carefully select newly protected land in order to meet well-defined conservation goals while minimizing impacts to agriculture.

Data reference

Models and software used

Models used for this experiment include,

Global Change Analysis Model (GCAM) v5.4 with modifications for the protected areas and bio-energy constraints. Latest version of GCAM available here and latest documentation on GCAM available here. The specific model version used for this analysis with all configuration files used is available here.
Moirai land data system v 3.1 Latest version with documentation available on GitHub here

Input data

All GCAM runs are available as R project files within the project_files/ directory.Input xml files for GCAM runs can be made available on reasonbable request. Ancillary mapping files are stored within the repo under folder other_data/. Shapefiles and raster files used to make plots are available here

Paper figures and supplementary material

All paper figures and their associated data are in the paper_figures folder. All supplementary material is in the paper_supplemental folder.

Reproduce our experiment

To reproduce the results and figures shown in Di Vittorio et al.,

Install R here - https://www.r-project.org/
Install R studio from here - https://www.rstudio.com/
Run the rmd file in the root directory called Protected_area_paper.rmd chunk by chunk to generate relevant figures. All outputs (csv and images) will be saved to the outputs/ folder (in separate subfolders).
Run the proc_moirai_land_distribution_public.r script in the scripts/ folder. Outputs will be saved to the outputs/distribuiton_threshold folder.
Run the proc_landavail_thresh_public.r script in the scripts/ folder. Outputs will be saved to the outputs/distribuiton_threshold folder.
Run the proc_avail_threshold_from_initial_public.r script in the scripts/ folder. This output will be saved to the 'paper_figures' folder.
We have made all outputs available in this repo for user convenience.