Modeling Almond Distribution

This work is part of a broader project examining the potential for agricultural expansion of almonds under climate change. It was published in the International Journal of Biometeorology and the full paper can be accessed here.

What we did

Delineating locations viable for cultivating high-value crops such as almonds provides key information for economic and land use planning and water management. Three modeling approaches were used to identify the potential distribution and key thermal constraints on almond cultivation across the southwestern United States (US), including two empirical species distribution models (SDMs) – one using commonly-used bioclimatic variables (Traditional SDM) and the other using more physiologically relevant climate variables (Nontraditional SDM) – and a mechanistic model (MM) developed using published thermal limitations from field studies.

What we found

While all models showed comparable results over the majority of the domain, including over existing croplands with high almond densities, the MM suggested the greatest potential for the geographic expansion of almond cultivation, with frost susceptibility and insufficient heat accumulation being the primary thermal constraints in the southwestern US.

"Species

SVI for each phenological phase modeled in the MM. While cold hardiness and GDD can be modeled independently, frost risk relies on phase timing. Consequently, grey areas indicate locations where subsequent heat accumulation (GDD) did not allow for the stage to be reached and therefore frost risk is not applicable.

SVI for each phenological phase modeled in the MM. While cold hardiness and GDD can be modeled independently, frost risk relies on phase timing. Consequently, grey areas indicate locations where subsequent heat accumulation (GDD) did not allow for the stage to be reached and therefore frost risk is not applicable.

The Traditional SDM over-predicted almond suitability in locations shown by the MM to be limited by frost, whereas the Nontraditional model showed greater agreement with the MM in these locations, suggesting that incorporating physiologically relevant variables in SDMs can improve predictions.

Difference in SVI between the MM and each of the SDMs. Blue hues indicate that the SDM over-predicts SVI relative to the MM, while red hues indicate SDM under-prediction. SVI differences of +/- 0.1 are masked with white.

Difference in SVI between the MM and each of the SDMs. Blue hues indicate that the SDM over-predicts SVI relative to the MM, while red hues indicate SDM under-prediction. SVI differences of +/- 0.1 are masked with white.

Finally, opportunities for geographic expansion of almond cultivation under current climatic conditions in the region may be limited, suggesting that increasing production may rely on agronomical advances and densifying current almond plantations in existing locations.

Average SVI over croplands with varying almond densities. Low SVI in croplands where almonds are not currently grown may indicate limited opportunity for geographic expansion of almond plantations across the Southwest, meaning that current almond locations would need to densify to meet any increases in demand.

Average SVI over croplands with varying almond densities. Densities are given as percent of 4sq-km area covered with almonds. Almond coverage data comes from the USDA-NASS Cropland Data Layer for 2015.

Earlier work on using species distribution models to estimate the potential distribution of almonds was done for the WSU CEREO Interdisciplinary Poster Session. For that poster click here.

Advertisements