Part 2 can be done by extending that optimal strategy:
i know i can get 1000 apples to B with 4000 at A by splitting it up into two lots of 2000. However if there exists a better method, it must be one where the apples at A are between 4000 and 3000 since 3000 gives a max that is less than 1000. Hence we can write the number of apples at A as 3000+a.
that means initially there are 4 apples being eaten per km if we drag across the 4 loads. This occurs for a distance

until the a apples are eaten. Hence:

Now the next 1000 apples must be eaten such that we eat 3 apples per km until we get to position

:
hence:
3=1000)
The final 1000 apples must be eaten for a distance

, 2 per km this time:
2=1000<br />)
Solving these equations backwards to find a gives

Hence

apples at A.
Your version can be done by applying this too.