As our premise, let's assume that you can make it to the fruit stage. As such your focus is on an optimum strategy for killing fruit.
The fruit level is characterized by high hitpoint creeps moving relatively quickly and populating the entire map. Range is only a minor issue, since the map will be completely saturated with both creeps and towers.
This particular situation vastly changes the relative strengths of towers. Some towers which would otherwise be quite strong become pointless.
Let's look at pure darkness tower. These have extremely high damage. But overkill will be a problem. You can only have so many of them, lets say you get 6. A worst case scenario will cost you 18749 damage. This means you are operating at 5/6ths for your primary DPS for 1.5 seconds. Pure light has similar problems, and its long range is of no real value in this situation.
Compare this with a blitz tower. A blitz tower does high dps (1200 * 1.11^20, or 9674 DPS at full charge if I am reading the description correctly) and has effectively immediate target switching (negligible overkill). It is, even on these attributes alone, an attractive alternative to pure darkness. Furthermore it costs less than half the price, and the maximum number you can get is dependent only on tile space. Combined with forge towers (which is possible given the required element combination) they are easily the highest DPS in the game, even if you include charged AoE towers, and almost none of their DPS will be wasted.
There are two other towers I considered.
The first is the sub zero tower. Let's be generous and assume double damage is always inflicted, and that the AoE always hits at least two targets. This gives us 1000 DPS to each creep passing through the helix, plus at least 2000 DPS for the AoE. If 7 or more creeps pass through the helix we have matched the haste tower. Since the map is saturated and fruit moves quickly, this may be possible. Unfortunatley the game mechanics remove viability: the helix works best when targets are moving towards it. Towers will try to fire at the closest target to the exit. This means that the tower will fire at targets moving away from it, significantly reducing the number of creeps hit by the helix. A set of towers in one location could be microd to always fire into oncoming fruit, but it is not possible to micro in this manner when the map is saturated with towers. The sub zero tower therefore cannot compete with the blitz.
The second possibility I considered was to line the north-south wall on the eastern edge of the map with flooding towers, and attempt to damage all creeps by some amount before they reach this point. AoE is probably the best way to do this.
There is little point to getting drowning towers, the goal is to use the percentage chance of instant death after dealing some damage earlier in the map with the AoE. This gives us more element points to work with for our DPS.
Assume we manage to get all creeps to 75% life before they reach the flooding towers, and we have 30 flooding towers, each one attacking a creep on average 3 times as it passes. These numbers are very approximate, but good enough to prove a point. The chance for any shot to kill a creep will be about .25. The chance of one creep not dying to a shot is therefore .75. The chance of one creep living through 30 towers firing 3 times is therefore .75 ^ 90, or 5.7e-12! This is, statistically, easily good enough to get over 300 fruit kills if you reach the fruit stage with a couple of dozen lives in hand. The question is, can you deal out 25% damage to all creeps by the time they get to that point? I think this build is a possible winner, but the numbers are very sensitive. Just how much damage can you do before the creeps reach the death trap? How many shots do the flooding towers get at each creep? Anyone tried it?
