83% of all calories eaten come from 10 crops; 66% from 4: corn, wheat, rice, and soy. It's counterintuitive, because typical diets consists of meat, fish, eggs, diary, etc. But cows, chickens, and even now fish are fed corn and soy, so calories can be traced to a few crops.
A crop pandemic has a different character from a human pandemic. Human pandemics spread globally and can hurt the people who released it. Even if engineering not to, they can mutate. A crop pandemic has less effect on anyone wealthy, well connected, from a different country, or with advance knowledge.
Crop pandemics have happened before.1 Irish potato blight killed a million and the Irish population has never recovered. Coffee rust destroyed Sri Lanka's coffee industry and forced them to switch to tea, which is why that is now their main agricultural export. Major powers developed crop weapons in WWII; by nature, it is difficult to prove when they are used: in 1945, Japan suffered mass rice failures from rice blast, and Germany's potato crop was blighted; both coincided with covert Allied research into those diseases, but the connection was never proven. Research escalated during the Cold War, but both sides officially renounced their programs and destroyed their stockpiles in the 1970s. At the time, 25% of the American stockpile was crop pandemics.
Crop pandemics are an extremely covert method of war. A single person can release wheat rust undiscovered and devestate an ecosystem. Targeting food sources is taboo: but these taboos have been broken in every major recent conflict.2 The destruction has been kinetic - blockades, airstrikes, sieges. Deliberate pathogens are technically simpler than human pandemics, offer better deniability, and produce the same strategic effect with less direct attribution.
The existential risk of crop pandemics comes from global instability. If everyone in a region, country or the world knows that only 90% of them can survive, it creates a very bad game theoretic equilibrium. Beyond existential risk, it creates significant suffering, and there are already recurring problems with natural pandemics.
From a national security perspective, the goal of war is to paralyze or topple an enemy state. To topple the government of e.g. Iran, it would take 2 million troops—or maybe just one bad harvest and popular discontent. Similarly, the U.S. develops counters to almost every military threat; no country today has the logistics to field an invasion of the U.S. and even a nuclear war would be devestating but not existential. Which is a greater threat to the stability and continuity of the U.S. government: a military threat or a quadrupling in food prices? The measure of a weapon is how much pain and disruption it can cause at what cost. Crop pandemics carry almost no cost and can deployed anonymously.
Basic mitigations include technologies developed for human pandemics, plus:
Research need not include any capabilities research: the existing natural pandemics are already devestating and easy to release without being discovered. Initial research can target these natural pandemics.
Other examples: French Wine Blight wiped out 40% of Europe's vineyards. Brown spot disease starved 2-3 million. In the 1950s, bananas used to be sweeter, before Panama disease killed the Gros Michel banana, and now Tropical Race 4 now threatens modern bananas. Corn Leaf Blight killed 15% of U.S. corn in 1971 and wheat rust has caused recurring devastation, with Ug99 posing an ongoing threat.↩︎
Ukraine feeds 50-100 million people in import-dependent countries with few good alternatives, and its exports have been targeted. 25 million people—half of Sudan—are facing acute food insecurity.↩︎