Why is bishop g5 a blunder in this position?

I was reviewing my game (which I did eventually win) and I noticed the engine had this as my only "blunder" in the game. 

In this position, I went bishop g5. I'm not defending it as a good move, but the engine analysis dropped it from 0.4 in my favor to a -4.0 against! Suggesting instead that bishop d2 was the best choice. 

Again, obviously not arguing with the engine, but I can't see what was so awful about the move I did make. I didn't leave anything hanging, I didn't put anything in danger, I don't see any clear advantage I gave my opponent. Is it just because it leaves open the potential for him to pressure the bishop with f6 or h6? Again, I agree it's a "miss" or "inaccuracy", but I'm failing to see what causes it to be a full blown blunder here. Or is it just because I failed to play bishop d2, the better move?

Anyway, I usually see the blunders in post-game analysis and understand what I missed, but this one just seems like a big negative, and I don't fully see it. I'd love some insight! Thanks

after Bxe4, there is a discovered attack on the bishop so after takes back the queen takes to bishop so you blundered a bishop

Oh! I see that now! Thankfully my opponent didn't lol but thank you for the help.

I just used stockfish lol

When you get to my level, you'll find that you have to pay much closer attention to see potential mistakes from your opponent.

Everyone blunders in pretty much every game they play. However, 700-800 marks the point where people start to rarely make blunders that are blatantly obvious; and when they do, it's usually a risky move that they don't realize has a very obvious counter. Generally speaking, they tend to require a bit more thinking to see, like the one shown here.

I probably wouldn't have noticed that either if that was me playing black; maybe on a good day, but overall it's not very likely.

As you had the engine running, you were just one click away from discovering it yourself. Don't stare at Game review, it won't help you in any way. Instead, click to "Analysis" and all engine lines will be revealed.

More importantly, if the recapture on e4 is done by the f4 pawn, fxe4 uncovers an attack by the White queen on the h5 knight, but Qxg5 captures the bishop on g5 and protects that knight.

For future, there is a self-analysis function in chess.com, so you can see why is it a blunder by following computer's moves. And if you think that YOUR move is better, and can't see why is your move bad, you can see what will be next after your move

Ah yeah, I'm still not great at reading the analysis system yet so I think it's time to wrap my head around that properly! I didn't actually realize it showed the whole line following a move, and not just the better move. I'll definitely get that sorted and start using analysis instead of game review.
Thanks everyone for the feedback!

Actually it's not that simple. Yes, the issue with Bg5 is that the bishop is unguarded, and Black wins because of a subsequaent defensive move ...Qd5.

In the final position White has a couple of pawns for the piece, , but he is underdeveloped and with no active play, while he will lose time protecting his Queen after a few moves.

I saw that line, yeah interesting.

Though, I would probably play Qd8 instead of Ke7. That way, black doesn't lose the second pawn and basically threatens queen exchange, that is more or less forced. Only Qb7 can be played to not allow exchange of queens, but that wastes further time, as after Nd7, black is developing more.

So even though the engine says Ke7 is fractionally a better choice for black, from human perspective, I think Qd8 is simpler for black, and much better if white decides not to exchange queens.

That's the least of it. You can also try any move, and it shows you all the best answers and their evaluations. And you can do it as deep as you want, go back and forth as you wish.

Analysis is amazingly useful, while Game review is shockingly useless and dumb.