Theoretically pontoon bridges with rope anchors keeping them to the bottom would work against wind and flow, overcoming the problem jhabbot mentioned in his answer (same as train length limit - stretching force).
In practice these come with more problems of their own.
They drift on water surface and as result, rise and fall with water waves. The larger the body of water they span, the higher the waves; at certain point in stormy weather the bridge would just launch the vehicles into the air.
The anchoring isn't exactly simple if it's to withstand such forces. You could just as well go with pillars, these don't add much to complexity.
They are only a short way above the surface. Waves could roll over them washing vehicles off. Also, they stay level to the local surface - a wave wouldn't need to roll over it - it could just flip a vehicle sideways by twisting the bridge.
Since segments need to be mobile relative to each over, their joints will be uneven, forcing a severe speed limit.
Unless they float freely, they'd be very limited with water level. If you anchor them firmly, water rising (even due to a storm) may submerge them. And yet again, as the segments need to be at least partially mobile, a longitudal force stretching one side of the bridge may lead to the other side to stack segments against each other.
We have construction technologies that are extremely durable against longitudal forces (stretching, compressing) - reinforced concrete, steel ropes etc. But add lateral forces, and the design becomes much harder to keep strong; buckling, twisting, and loss of stability become very severe. With keeping the bridge suspended you keep lateral forces limited to wind. If the bridge is partially submerged, this goes out of the window.
The primary upside of pontoon bridges is the simplicity - they can be deployed in matter of hours, and as such they play huge role in the military. But since they are vulnerable against weather and due to the slew of problems they create - especially with increasing span = size of the body of water = mechanical influence of water conditions, they make very poor permanent bridges, and so firmly supported bridges are simply superior.
...also, where the ratio of vehicle traffic demand versus length of the span of water to cross is too low, the right solution is a ferry. Ferries can take many vehicles on board and cross the distance (and depths!) not viable for any bridge, and of course their cost is a tiny fraction of cost of the bridge covering that distance.
Note pontoon bridges are okay as temporary solution (say, in place of a bridge washed off by flood, or for duration of construction of permanent bridge, or in locations where building a permanent bridge would be overly expensive or difficult), but they are always considered a poor man's substitute - and while they are okay for crossing a moderately sized river, the engineering problems scale up to insurmountable levels as the distance covered rises - they are really unsuitable for very long spans of water.
