Thanks! You know... for some reason I thought I had that issue up in the top known-bug post. But yeah, the terrain noise-generator doesn't like negative coordinates at the moment. Need to fix that :)
I've also been considering making it so that worlds are actually finite in size (still big!) but that they'd loop around if you traveled far enough in both x and y directions. But that would introduce a whole 'nother can of worms; technically, and gameplay-wise.
If you're using something like Perlin Noise for the terrain generator, there are a number of ways for converting a basic noise field into a tiling noise field.
The simplest IMO would be to blend the value for the current coordinates with a copy shifted over by one and work using modulo math.
F'(0) = 1 * F(0) + 0 * F(-1)
F'(0.25) = 0.75 * F(0.25) + 0.25 * F(-0.75)
F'(0.66) = 0.34 * F(0.66) + 0.66 * F(-0.34)
F'(1) = 0 * F(1) + 1 * F(0) // Matches with F'(0) = tileable!
That will keep the planet from having seams at the poles, and you just have to take care with the math when the player is near the pole and looking over.
If the worlds are reasonably large, that shouldn't be a problem for gameplay since even 16 bit coordinates would make a circumnavigation in the buggy an epic week long RL marathon if you don't stop for anything and have more gigs of ram than you can shake a stick at.
Smaller worlds would need the plants to regrow at least, and probably a new drilling rig setup to get more water.
According to the resource_tiles.json, there could be up to an average of 3 seeds per 10 plant tiles dug up. A little less than 1 seed per 5 plants harvested in that case.
If the listed drop chances are sequential and independent rather than cumulative, it could also be a 1 : 18.9 ratio which is close to your number there.
A crafting reaction to turn one plant into one (or more) seeds could be good.
That would let you maintain the farms by saving half your production to use as seed stock for the next harvest.
The main problem would then be the finite supply of water/ice.
