Not at all. If, for instance, you had a dominant trait is always lethal before sexual maturity, then that trait will be completely eliminated within a single generation, even if it’s very rare (and hence the number of individuals killed by it is a very small fraction of the total population). You only need to kill a large number if you’re trying to eliminate a widespread trait.
Not at all. There just has to be a differential in the number of offspring produced by individuals that cross roads vs those that avoid crossing roads. Road-crossing individuals could have a higher reproductive rate because they obtain access to more food than those that don’t, which enables them to have enough offspring to compensate for the fact that they have a somewhat higher mortality rate by being hit by cars.
Selection doesn’t have to be anywhere near as drastic as killing a majority of the population to have a strong effect over time. It’s just that in this case the number of individuals killed isn’t large enough to have a strong enough selective effect in a short time.
Now if you want to see strong selective effects in a short time you generally do have to practice such extreme selection as you describe. This is what humans do in selectively breeding animals and plants: the individuals selected to breed from comprise only a tiny part of the population, and such severe selection is imposed for many generations. But selection in nature is rarely this strong.
And cars (or, rather, their drivers) presumably have a lot more incentive to avoid collisions with elephants than with corvids or pheasants. 
I think we’re discounting the animals run over on purpose. There are no shortage of sadists who delight in running over easily avoidable critters like turtles. Just saw a dead one today that was squished on the shoulder where no vehicle is supposed to go (and there are usually telltale signs when an animal was moved there posthumously).
I think the time factor is key, but also maybe the imperative to roam may be stronger or more beneficial than the risk of crossing roads.
Also, when a road is built through an area, accompanying the road are many other factors besides being hit by a car related to human activity (or even the road itself) that may be killing off local wildlife.
Previous thread on this topic.
That’s by “How often I notice the species of a dead animal as I’m driving by”. The ones I don’t see I don’t see. However, two more dead raccoons today, so I’m bumping raccoons up from 40% to 50%.
The OP asks a good question, even if it’s not perfectly original. I recall watching a documentary about raccoons in Toronto that suggested city raccoons were smarter than country raccoons though I don’t remember how they estimated raccoon intelligence. The documentary also noted that city raccoons rarely crossed busy streets. That suggests raccoons’ urban environment was selecting some trait (perhaps intelligence) that helps them survive traffic better. So maybe traffic is contributing to the evolution of smarter raccoons that recognize busy streets and learn to avoid crossing them.
ETA: I think this was the documentary: https://www.thestar.com/news/gta/2011/02/11/documentary_reveals_the_secret_lives_of_torontos_raccoons.html
Moderator Note
The above post was posted to an old thread from 2008 on the same subject. I have moved it to the current thread.
Colibri
General Questions Moderator
No, species are the result of natural selection not the cause.
Evolution is about genes. Genes have no aspirations or morality or consciousness or goals. They’re just molecules that produce copies of themselves. It’s a mechanical process.
Genes influence the behavior of the organisms in which they are located. This genetic influence may change the likelihood of the organism reproducing, which provides a new location for the genes. Genes that are useful for producing new locations like this become more common. Genes that are detrimental to producing new locations become less common. But the genes don’t know or care if they’re common or uncommon. Like I said, they’re just molecules. Everything a gene does is just a chemical reaction.
Evolution occurs when something changes. It might be the gene or it might be the environment. A new gene might influence new behavior that is more useful than the behavior produced by existing genes. Or a change in the environment might make existing behavior more likely to cause a organism to reproduce than it used to be. In either event the gene becomes more common. Or the opposite effect can happen. A new gene might produce detrimental effects or an environmental change might make a genetic effect less useful than it was. And then the gene will become less common and might even disappear entirely. But again, the genes don’t know or care.
There’s no greater plan at work. It’s just random chemical reactions. That’s why evolution can take so incredibly long to produce any noticeable changes.
Thanks for moving my post, Colibri.
For what it’s worth, prey species do not respond to a perceived predator threat from a known direction by running laterally, since that reduces the component of their velocity that’s heading away from the predator, and allows the predator to cut the corner diagonally to intercept them. The optimal response (and their evolved instinct) is to run directly away from the direction from which the predator is approaching, and dodge sideways only at the very last minute if the predator closes on them. Thus, if an approaching car is seen as a predator threat, the animal will tend initially to run along the road directly away from the car. I have certainly seen sheep in the U.K. behave this when when a car approaches on a road across open moorland.
I don’t know what you mean by “replacement rate” but that is not what matters. It’s the difference in fitness between genotypes/phenotypes. In other words, the proportion of genes contributed to the next generation by each genotype.
The first requirement for natural selection to take place is that variation in genotype & relevant phenotype must exist in the population. In this case, there must exist at least two genotypes in the population, one of which has a phenotype “more likely to cross roads” and the other “less likely to cross roads” (or “worse/better car avoidance reaction”). It may be that this variation in genetically-determined behavior simply does not exist.
There is a valid underlying point that the mechanism of evolution is differential rates of death among genotypes/phenotypes - and that rapid evolution in response to an environmental change requires strong selection pressure, meaning a lot more death by new means. Depending on the manner of reproduction, introducing a new strong selection pressure may result in a population decline as the less fit genotype is killed off.
One bit I do wonder about, is that cars aren’t the first threat of their type. Plenty of prey animals have evolved different responses to different threats: A rabbit, for instance, will run from a fox, but hide from a hawk. But there are naturally-occurring threats that behave in about the same way as a car: Coming right towards you, but just coincidentally, and harmful if they hit you but won’t try to chase if you get out of the way. Consider a stampeding herd, for instance. You’d think that animals would have evolved an appropriate response to a stampede, in the eons that they’ve lived with large herd animals. And once you already have that behavior established, adapting to cars would only require slotting cars into the appropriate category, which should be much easier than developing an entirely new response.
Apparently there is research that indicated prey animals (such as moose) congregate near roads in order to avoid bears.
http://www.nature.com/news/2007/071010/full/news.2007.155.html
So, it may be that roadkills of certain types of wild animal happen because the roads are considered relatively “safer”.
For the smaller fur-bearing mammals, many cross the roads underneath in the culverts … go out into the country and find a culvert, look for the animal trails leading in and out … usually these trails are very easy to spot … this is a prime location to set a foothold trap for the “mixed bag” trap line, super easy access by car, generally a nice assortment of catch … the main downside is the catch and traps make for easy stealing and there’s always the danger of catching the occasional skunk …
To nitpick, as I said above the mechanism is not necessarily the rate of death, but differential reproductive rates between genotypes. Obviously if an animal is killed before it has realized its reproductive potential it will have a lower reproductive rate than other individuals, all other things being equal. But it is entirely possible for a genotype to have a higher death rate and also a higher reproductive rate, if it is able to produce more offspring before dying.
Well, if you’re going to nitpick, don’t quote me selectively, omitting the part where I defined “fitness” more generally. In the case of roadkill, the mechanism would obviously be differential rate of death, not variation in reproductive rate, and it would be unlikely for any strong selection pressure causing rapid evolution to operate otherwise.
Well, no. As I’ve pointed out above, animals that cross roads could well have a higher reproductive rate because they obtain access to more food and resources even if they have a higher death rate from cars. Focusing just on the death rate misses the point. Animals would not only have to have a higher death rate, they would also have to not gain a compensating advantage from crossing roads.
In species where males compete for mates, males may have a higher death rate but the winners have a much higher reproductive rate.