Current food trade helps mitigate future climate change impacts in lower-income nations

respective climate model (see S1 Fig). We then assumed crop yields at 1˚C global mean warm- ing relative to pre-industrial as the reference as this roughly represents current global warming (exceedance years for 1˚C warming range between 1999 and 2021 across our suite of climate models). Then we estimated gridded climate impacts as the percentage yield change with addi- tional 1˚C and 2˚C future warming compared to our reference. For each country y, crop i, crop-climate model CCM, and degree Celsius of additional global warming w, we estimated domestic production impact DP∗ ðy;i;CCM;wÞ=P∗ ðy;i;CCM;wÞ as pro- duction-weighted national average impacts denoted in percentage change terms (Eq 5; S1 Dataset): DP∗ ðy;i;CCM;wÞ P∗ðy;i;CCM;wÞ ¼ Xl n¼1 CIði;w;CCM;nÞ∗ Aði;nÞ P AðiÞ ; ð5Þ where CI(i,w,CCM,n) is crop-model simulated climate impact (yield changes in percentage) for crop land pixel n and A(i,n) is the total crop production in tonnes for pixel n from Monfreda et al. 2008 [25]. We assumed that harvested areas are not affected by climate change. We esti- mated domestic production impact for each country, crop and crop-climate model separately. Then we calculated country- and crop-specific ensemble median, and 20th and 80th percentile estimates (confidence intervals) across the 60 crop-climate model impact estimates. Note these are national level averages and so different impacts in different parts of the country could offset each other. We estimated crop aggregated domestic production impact ( � DP∗ ðy;q;wÞ= � P∗ ðy;q;wÞ) for each

country y, percentile q and global warming w, as the weighted mean impact of warming on production for domestic use of the three crops (Eq 6; S1 Dataset): � DP∗ ðy;q;wÞ � P∗ ðy;q;wÞ ¼ X3 i¼1 DP∗ ðy;i;q;wÞ P∗ðy;i;q;wÞ ∗pðy;iÞ; ð6Þ where p(y,i) is production-share weights equal to the share of each crop to the total calories pro- duced for domestic use from all three crops. Consumption and import impact As with domestic production impact, we estimated the consumption and import impacts on supply for each country based on the three crops separately and for the crops aggregate. Con- sumption impact (DCðy;i;q;wÞ=Cðy;i;q;wÞ) and import impact (DIðy;i;q;wÞ=Iðy;i;q;wÞ) for a country are the weighted average production impacts across all supplying countries, including production for domestic-use for the former but excluding it for the latter (Eq 7 and Eq 8; S1 Dataset). DCðy;i;q;wÞ Cðy;i;q;wÞ ¼ Xmþ1 x¼1 DP∗ ðx;i;q;wÞ P∗ðx;i;q;wÞ ∗ zðxjy;iÞ P zðy;iÞ ; ð7Þ DIðy;i;q;wÞ Iðy;i;q;wÞ ¼ Xm x¼1 DP∗ ðx;i;q;wÞ P∗ðx;i;q;wÞ ∗ zðxjy;iÞ P zðy;iÞ ; ð8Þ where m is the number of countries that consumer country y imports from, and m+1 includes the supply from domestic production of the country in question. z(x|y,i) are the calories contrib- uted by a supplying region towards the total supply of the country in question. Next, we computed crop-aggregated consumption ( � DCðy;q;wÞ=�Cðy;q;wÞ) and import impacts (�Iðy;q;wÞ=�Iðy;q;wÞ) for each country using a similar weighted average approach as for crop- PLOS ONE Current food trade helps mitigate future climate change impacts in lower-income nations

PLOS ONE | https://doi.org/10.1371/journal.pone.0314722 January 3, 2025 4 / 16