Right, but they didn't do that. It's a meta-analysis, so they took the value that each study got for a given crop in a specific country and then weighted all of the values by the share of global production that that country is responsible for. So if we pretend that the only three countries are Estonia, Latvia, and Lithuania, they did the following:
Found three studies from Estonia, two from Latvia, and two from Lithuania
Averaged the values of the three Estonian studies
Did the same for the two Latvian ones and the two Lithuanian ones
Found that Estonia is responsible for 60% of the world's beef, Latvia 25%, and Lithuania 15%
Took their three national averages and weighted them 0.6 for Estonia, 0.25 for Latvia, and 0.15 for Lithuania to get the final value for beef
Repeat for each other crop
The dataset was 1530 studies across 39,000 farms in 119 countries