Both White knights are on light squares. So are Black’s queen and both of his rooks. The knight on d5 has no forking moves, but the knight on d3 has one: Nxe5, forking the queen and both rooks. Is the square protected? Yes―twice (all defenders must be accounted for!). Start by worrying about the pawn on d6. It turns out to be a non-issue because it is pinned to its queen by White’s rook at g6. That leaves the bishop on g7 to consider. Imagine the board after 1. Nxe5, BxN. What lines would that exchange open, and what checks would you then have? Just one: Rg8. Examine its consequences and you find that they can be summarized in a word: mate! For when you imagine the king fleeing to e7, notice that your d5 knight already attacks the square; and you should be mindful from the beginning of the battery of rooks you have bearing down on the g-file adjacent to the king.
You might as easily have seen the idea the other way around: you observe that your battery of rooks on the g-file nearly is ready to mate on g8, but that the bishop on g7 stands in your way; this means the bishop is pinned (we will see many studies like this in the chapters on pins), and is not really defending e5. Since the e5 pawn’s other defender also is pinned (to its queen), e5 is available to your knight for a possible fork.
In any event, if Black sees all this as well as you do (and you should assume he will), the actual consequences of the forking move Nxe5 still require some thought. If he tries to forfeit “only” the exchange by just moving his queen and letting go of the rook on f7, Black ends up losing more than that; for White can instead take the rook on c4 at no cost. Notice that once White plays Nxe5, the c4 rook is attacked twice by White and protected only once by Black. So the lesser evil for Black in reply to 1. Nxe5 probably is 1. … d6xN; 2. RxQ, RxRc6 losing his queen and a pawn in return for a knight and a rook.