Are there observable changes in the squirrel (or skunk, etc.) gene pool, as a result of street-crossing behavior? Do we have evidence that roadkill is making the population smarter or more agile or having quicker reflexes?
Any such changes would be far too small to observe in the 100 or so years since it’s been an issue. Evolutionary adaptation simply doesn’t operate over such small timescales for complex organisms like squirrels.
In my corner of NE Texas, the cars are winning. Possums still stop in front of your car.
SSG Schwartz
I’m not so sure about that. I was told once that hedgehogs in the UK are learning to run away when in danger, since rolling up into a ball doesn’t work very well against an oncoming car.
Surely it’s been an issue for thousands of years, with carts, waggons, and the like?
Carts and wagons likely weren’t moving in excess of 10 mph. Your question is similar to asking why humans haven’t yet evolved to dodge bullets fired at them, what with guns having been around for centuries.
Roadkill in NZ is mostly 3 unfortunate creatures; possum (Australian), hedgehog (British) and Pukeko (NZ bird). Of the 3 perhaps hedgehogs have been exposed to traffic the longest. Possums have had MAYBE 200 years to get used to traffic but every Pukeho acts like it is the first one ever to cross the road, but has at least 150 years to get its act together.
When you travel around the country you tend to see more hedgehogs, then possums and then pukeko as roadkill…is it backwards evolution???
Are pukekos very common? When I was in NZ (South Island) road kill was mainly rabbits.
But that’s just ordinary learning behavior. All animals with a brain do that to some extent or other. The OP is discussing actual, observable evolutionary changes, not simply learned behavior modification.
From the OP:
How do you determine improvements in “smartness” other than by observing changes in behaviour? If hedgehogs are more likely to run away now than they were 50 years ago, why shouldn’t that be down to an evolutionary increase in their intelligence?
I’m not so sure about that. That’s about 100 generations for squirrels, which would correspond to about 15,000 - 20,000 years for humans.
The issue here is not so much the timescale, but whether or not the environmental factor (being killed by cars while crossing a street) is common enough as to affect the gene pool one way or the other. IOW, what percent of squirrels are subject to this environmental pressure? I don’t think we have a factual answer to that question, but my guess is that it’s not much a factor. Most street crossing for squirrels is going to be done in residential neighborhoods where cars aren’t going all that fast.
I’m sure I’m missing some mathematical subtlety here, but by my calculations, that makes a human generation 150-200 years long. I’m pretty sure that’s not right. Nevertheless, I understand and agree with your larger point.
You seem to have left out this part of the OP:
John Mace raises a good point, but either way you look at it, we’re not going to see any fundamental genetic changes in the squirrel population as a result of roadkills. Learning does not correspond to increasing intelligence. The one occurs in individuals as a result of exposure to new experiences, the other only occurs, generally, as result of selection pressure in populations.
Hey, what’s a simple factor of 10 between friends? 
Yeah, I did the math wrong. Assuming 15-20 years per generation of humans, then 100 generations would be 1,500 - 2,000 years. But I’m still going with time being less of an issue than the % of the squirrel population subject to the environmental pressure of having to dodge cars. We’ve seen moth populations adapt to changes in coloration in just a few decades, but they were pretty much all subject to the environmental pressure to do so.
I have seen quite a few city dogs who (apparently) understand how to wait (for a red light) to cross a buy street. Of course, you can argue that the stupid ones get killed. But my aunt had a dog (Rusty) who travelled up to 16 miles from home (he had lots of girlfriends). One day, she got a call from a homeowner in an adjacent town-Rusty was in her yard, swimming in the kid’s wading pool!
Domestic dogs are trained to be cautious about crossing the street. Even if they are not purposely trained, they pick up a lot of stuff from just hanging around with us humans. I’m certain, though, that natural selection hasn’t shaped dog behavior so that they instinctively know what a red light means.
I thought that possums would top the list, since everyone I know in New Zealand seems to aim for them deliberately.
Well I live in Auckland and Pukeko are VERY common. I drive to work via the motorway and they are a very common sight…squished is also a common sight. 
Cunctator Possums are probably hit deliberately sometimes (not by me! I nearly caused an accident avoiding a starling who wouldn’t leave his lunch!). Possums have caused HUGE damage in NZ and we have been taught that they are bad bastards!
I think they are cute though Cute but BAD.
It is a short timescale, but it’s not as if the animals in question have to develop their entire flight/caution instincts from scratch.
If the pressure was consistent and sufficiently severe (without being so severe as to cause extinction), adaptation could probably happen fairly quickly - behavioural modification has been achieved through artificial selection in domestic animals, but that’s quicker because it involves concerted effort on the part of the humans in charge - so it would probably be too optimistic to expect natural selection to have produced anything very noticeable yet.
I think possums just aren’t at street level as much, more likely to be climbing through trees.
Poor buggers.
Natural selection certainly cannot operate over those short (~100 year) timescales.
But you’re talking about something called artificial selection. A great example of artificial selection acting over extremely short timescales is the Heiki Crab from Japan. There was a Japanese legend a few hundred years ago about a samurai massacre, where they threw the bodies of the slain samurai into a lake. Subsequently, when fishermen flung nets in that lake and caught Heiki Crabs, they threw back the ones that had marks on the back of their shells that resembled samurai faces. And so in the absence of any natural preference, the crabs with those marks on their shells were selected for survival, artificially.
Now, I still don’t have enough information about squirrels etc. to know if there is a population among them that is better at not becoming roadkill. But if there is, they will certainly be selected for. The majority of my experience with animal behaviorists (of which I am not one) is that squirrels, pigeons, and other urban animals are most commonly regarded as pests, and weren’t well-enough studied 100 years ago to draw conclusions based on that.
And if you want some anecdotal evidence from someone who spent 5 years in Florida, all armadillos still jump when they see oncoming cars. This way they’re sure to run right into your grill, ensuring their death. So for that case, the answer is no.