Guest post: Natural selection in real time via road kill

by Greg Mayer

A new paper in Current Biology by Charles & Mary Brown with the folksy title, “Where has all the road kill gone?”  reports evidence for rapid evolution of wing length in cliff swallows (Petrochelidon pyrrhonota) nesting on highway overpasses in Nebraska. (See also this news piece on Science‘s website.) For those evolution-deniers who demand to see natural selection in “real time,” this is one bit of evidence.

During the course of a 30-year field study, the Browns found that the number of road killed birds declined, from about 20 per year, to about 4 per year (panel A in the figure). They could rule out most of the obvious possibilities: the bird population size had not declined (panels A & D), and they had not changed their survey effort or methods. They then compared the wing length of the road killed birds to a sample of the population at large (obtained from netting fatalities, but corroborated by released birds), and found that wing length had declined by several millimeters (panel B) and that the change was cumulative over the 30 years, with the road-killed and population at large birds slowly diverging (panel C; I must say I’m a little perplexed that the dead birds keep getting bigger).

A. Roadkill declines, but the population increases. B. Road-kill birds have shorter wings than the population at large. C. The difference in wing size increases over time, D. Again, the population size goes up.

A. Road kill declines, but the population increases. B. Road kill birds have longer wings than the population at large. C. The difference in wing size increases over time, D. Again, the population size goes up. (From Brown & Brown, 2013, Current Biology)

Wing shape in birds is well known to relate to specific functional abilities, and shorter-winged birds are better at vertical take off and pivoting. The Browns suggest that as the birds moved from their pre-industrial nesting spots (on cliffs) to bridge abutments and highway overpasses, the ability to avoid speeding cars conferred a selective advantage (the swallows frequently land in the road).

The authors acknowledge that other selective factors may influence wing size in the birds, and allow that a decrease in road kills could be due to learning.

[JAC note: one problem here is the lack of demonstration that the changes in wing length really were due to genetic as opposed to purely environmental causes (for example, perhaps temperature changed over the years in a way affecting wing length). A genetic basis for the change is, of course, essential for showing that the short-term change really did reflect evolution. Given that the dead versus live birds did not change in the same direction, however, one can tentatively rule out some environmental factor affecting all birds the same way. Nevertheless, results like these must always remain tentative until genetic work—ideally breeding under constant conditions in captivity—is performed. See Greg’s caveat below.]

This study joins a growing list of observations of evolution-in-action over short time periods in birds. These include the pioneering studies of selective mortality in house sparrows by Hermon Bumpus,  and the now classic, decades-long studies by Peter & Rosemary Grant and their colleagues on Galapagos finches.

Bumpus’ work, which Matthew posted about recently, was one of the very first studies of natural selection, and his data has been much analyzed (see the data, bibliographies, and discussions posted by the Field Museum, Clark University and Pearson College). Like Bumpus’s study (but unlike the Grants, who also had quantitative genetic data), the Browns’ study is of phenotypic selection, and does not demonstrate the genetic basis of the observed changes (although a several millimeter change in wing length begins to approach low-level taxonomic importance).

One unusual aspect of this study is that there appears to be an increase in the size of the population. Sir Ronald Fisher showed 80 years ago that in simple, but fairly general, models of natural selection, the effect of selection is to increase mean population fitness, something he called the “fundamental theorem of natural selection“. In laboratory populations, this is actually not infrequently observed: a newly established population of flies or a culture of bacteria will increase in equilibrium population size or reproductive rate as the population adapts to the new laboratory conditions. But in nature, populations are subject to control by a wide variety of factors (e.g. predators, competitors, climate), so that populations may evolve genetically (increasing their mean “fitness”) without changing in size (because the carrying capacity is set by these other ecological factors). Alternatively, size changes that do occur may be in response to these ecological factors and not to changes in fitness. In the swallows, viability with respect to road kills (a component of fitness) is seen to quite directly increase (i.e., the mortality rate declines), and the population size correspondingly increases. I think it certain that many factors influenced the increase in population size of the swallows, and it would be hard to partition out the effect of decreased car-collision mortality. Nevetheless, in this case an increase in mean fitness due to selection among individuals appears to be reflected in overall population size.

Darwin would, I think, be gratified by all the evidence for evolution by natural selection that has accumulated since the Origin was published in 1859, but I believe nothing would have astounded him more than the now-abundant evidence for evolution occurring on the timescale of a single human life.


Brown, C. R., and M. B. Brown. 2013. Where has all the road kill gone? Current Biology 23:R233-R234.

Bumpus, H.C. 1899. The elimination of the unfit as illustrated by the introduced sparrow, Passer domesticus. Biological Lectures from the Marine Biological Laboratory Wood’s Holl, Mass. 1898: 209-226. (BHL)

Grant, P.R. and Grant, B. R. 2008. How and Why Species Multiply. The Radiation of Darwin’s Finches. Princeton University Press, Princeton, New Jersey.


  1. marksolock
    Posted March 30, 2013 at 6:14 am | Permalink

    Reblogged this on Mark Solock Blog.

  2. Posted March 30, 2013 at 6:25 am | Permalink

    Wow. I covered the same story from a more dumbed-down angle because I thought it was necessary for ease of communication. You’ve proved me wrong. This is excellent work, well communicated without pandering. I think I’m going to have to wean my readers onto a better caliber of writing. Thanks, and keep up the excellent work.

  3. Torbjörn Larsson, OM
    Posted March 30, 2013 at 8:23 am | Permalink

    I must say I’m a little perplexed that the dead birds keep getting bigger

    I suck at thinking in multivariate terms (which is why I do physics and/or software for a living =D), but to me it suggests there is another factor under selection that alleviates the problem with long wings. For example, increased alertness could mitigate it partially, hence more extreme outliers are being culled as time passes.

    • Torbjörn Larsson, OM
      Posted March 30, 2013 at 8:25 am | Permalink

      Oops, I forgot: To test that, the frequency of roadkills needs to go down with time – and it has.

  4. Posted March 30, 2013 at 8:28 am | Permalink

    I’m pretty sure the megafauna of Tennessee are becoming faster when crossing Whites Creek Pike due to the unforeseen selection pressure of TNT Motorsports.

  5. David Cohen
    Posted March 30, 2013 at 8:53 am | Permalink

    Speaking of road kill, I have always wondered why more intelligent squirrels haven’t evolved. They always seem to run right in front of cars, and often end up as road kill.

    I guess the ones that survive are just lucky, and I suppose that there are enough of them that the ones that are run over are negligible in number.

    • Gregory Kusnick
      Posted March 30, 2013 at 9:49 am | Permalink

      Same with bugs. You’d think there’d be enormous selection pressure on flying insects to avoid freeways at all costs. Yet my windshield still ends up splattered with corpses.

      My theory is that special strains of freeway-seeking bugs are being bred in secret labs run by car-wash conglomerates.

    • uva3021
      Posted March 30, 2013 at 11:27 am | Permalink

      Why aren’t there squirrels that have evolved the ability to deconstruct and re-manifest themselves at a molecular level as a car passes right through them

    • RFW
      Posted March 30, 2013 at 2:25 pm | Permalink

      It may be a trait common among the Sciuridae: I once had the unpleasant experience of running over a specimen of Tamiasciurus douglasii, the western red squirrel who ran right out in front of the car; stopped; hesitated; seemed to turn back; then ran on and crunch under the wheel. Not a nice thing to have happen, esp. considering that unlike most squirrels, the Vancouver Island race of T. douglasii is notoriously shy.

    • Lurker111
      Posted March 31, 2013 at 6:21 am | Permalink

      Actually, over the 25 years that I’ve lived on my street, I -have- noticed that the number of squirrelcakes on the street has noticeably declined. Squirrels get run over because they tend to move quickly for a short distance, then stop and look around dumbfoundedly, then move again. Obviously, stopping in the middle of crossing the street is not a survival-positive trait. Many of the squirrels on my street now cross in one motion.

      What I want to know is, how come I never see any starlings in two dimensions?

  6. Thanny
    Posted March 30, 2013 at 9:27 am | Permalink

    The caption says that roadkill birds have shorter wings than the population at large.

    That should be reversed.

    • whyevolutionistrue
      Posted March 30, 2013 at 9:47 am | Permalink

      Thanks– fixed!


  7. Gregory Kusnick
    Posted March 30, 2013 at 9:42 am | Permalink

    Sport-utility vehicles, which have probably increased during our study and offer a greater surface area for collision (relative to sedans that were more common in the 1980s), might contribute to changing bird mortality. However, road kill decreased as the larger vehicles became more common.

    On the other hand, larger vehicles have larger grilles for dead birds to get stuck in and carried away from the kill site. The authors don’t seem to have considered this possibility.

  8. Jim Thomerson
    Posted March 30, 2013 at 9:45 am | Permalink

    I have recently seen a deer stop and look both directions before crossing a road.

  9. Tan
    Posted March 30, 2013 at 10:33 am | Permalink

    Very interesting. Who would have thought. Thanks for sharing.

  10. BilBy
    Posted March 30, 2013 at 10:47 am | Permalink

    For another taxon there is an unfortunate other factor influencing road kill. Yesterday I watched someone swerve deliberately to run over a black racer snake – the snake was already dead due to a previous car.

  11. s.k.graham
    Posted March 30, 2013 at 3:06 pm | Permalink

    Why the dead (road-killed) birds are getting bigger as average average size of population as a whole gets smaller:

    (1) because the population is still in transition under selection pressure, there is more variability in size overall, allowing for some birds to still be in the same range of sizes from the population 30 years ago. Naturally these are much more likely to be killed.

    (2) The birds will undoubtedly have been evolving sensory, neurological, behavioral adaptations which also help avoid cars. Even if wingspan was not changing, it would be the largest birds which would be least able to benefit from faster reflexes, etc., pushing selection bias of the dead (road-killed) birds toward sizes even larger than the original population average.

  12. W.Benson
    Posted March 30, 2013 at 4:21 pm | Permalink

    I have an ad hoc hypothesis for increasing size in road-killed swallows after 1990. In the 1980s birds were naïve and killed unselectively. During the 1990s, birds started to wise up (natural selection) and increasingly evaded vehicles, only longer-winged birds were less efficient at dodging (more natural selection). However, only two birds seemed to have been killed after 2007, and excepting these, the points after 1995 run flat (i.e., b=0). There is little evidence that dead birds get larger after 1995.
    Eyeballing the points suggests that after 1995 the average size of killed birds leveled off at about 110 mm, and of living birds at about 107 mm. 110 mm may be the cutoff size for avoiding cars on Brown and Brown’s bridges. If smaller birds are progressively more agile, there is the interesting possibility that bird wing-length may adapt to vehicle speed (and speed limits) on bridges (assuming that birds return to nest at the same bridge each year).
    NB: my comment overlaps with those of Larsson (#’s 3 +4) and s.k.graham (# 11).

    • Marella
      Posted March 30, 2013 at 8:00 pm | Permalink

      So you could test that by putting the speed limit up by 5mph and seeing if the birds got even smaller.

      • Gregory Kusnick
        Posted March 30, 2013 at 11:20 pm | Permalink

        You’re making the unwarranted assumption that drivers actually pay attention to speed limits.

      • John Scanlon, FCD
        Posted April 1, 2013 at 8:23 am | Permalink

        Interesting suggestion. In discussion about the same paper, the following story was told recently on a palaeontology mail list to which I subscribe. Apologies for length of quote:

        “Many years ago, back in the mid 1960s, I was a curator at the University of Nebraska State Museum at Lincoln, Nebraska. Part of my job was to act as the vertebrate paleontology highway salvage person and therefore I drove widely across the state, mostly on back roads. At one point the state raised the maximum speed limit from 55 to 65 mph, the change to be made effective on a particular Monday.

        “I was driving across the northern edge of Nebraska on that day, and of course drove the new speed limit. Over and over and over I begin to hit birds which were sweeping across the road in front of my car.

        “It is a common strategy for birds in that part of the prairie to wait for cars to startle inserts into the air, and then pick them off as the car drives by. (All of these dead birds were collected, because I had a federal license to pick up such animals, which were turned into study skeletons at the museum)

        “I struck an unconscionably large number of birds during the first part of that week, but during the following weeks, I seldom hit a bird. The frequency was as it was before the speed limit increase.

        “The Darwinian observation was very clear. I had changed the parameters of the game, and those birds that could not learn to time themselves relative to the now much faster cars, were selected from the population.”

        Clair Russell Ossian, PhD
        Professor Emeritus, Geology
        Tarrant County College

        Of course, this needs to be replicated. Try to find out when and where rural speed limits are to be increased by various increments, and get out there for multiple, time- and weather-matched before-and-after transects at the relevant speeds.

        • Gregory Kusnick
          Posted April 1, 2013 at 10:25 am | Permalink

          The Darwinian observation was very clear. I had changed the parameters of the game, and those birds that could not learn to time themselves relative to the now much faster cars, were selected from the population.

          This conclusion seems dubious to me. All we know is that at the start of the learning curve, when luck dominated, lots of birds died; later in the curve, when skill dominated, fewer birds died. This is exactly what you’d expect even if all the birds were identical in their learning abilities.

          There may well have been some Darwinian selection pressure operating here, but this observation is not sufficient to make that “very clear”.

          Also, if the guy knew he was killing “unconscionably large” numbers of birds, why didn’t he slow down? Or maybe he did, and that accounts for the decline in bird fatalities.

        • infiniteimprobabilit
          Posted April 1, 2013 at 11:25 pm | Permalink

          Or maybe it just shows that birds learn quickly and after a week they’d got used to the faster cars.

          Of course, that could be just one factor, in conjunction with the others mentioned.

  13. Thomas
    Posted March 30, 2013 at 5:30 pm | Permalink

    “I must say I’m a little perplexed that the dead birds keep getting bigger ”

    I wondered if maybe there are other adaptations going on (birds getting smarter?, cluier?, faster?), and thus it takes longer wings than it used to for a bird to tend to get squashed

    Conversely, a really stupid short-winged bird will still get runned over.

  14. Cremnomaniac
    Posted March 30, 2013 at 6:58 pm | Permalink

    I’m struggling with basic genetics here, maybe someone could enlighten me.

    “one problem here is the lack of demonstration that the changes in wing length really were due to genetic as opposed to purely environmental causes..”

    So what your saying is, there are potential environmental factors that effect phenotype. The problem (or confusion) I have is in the attributing change demonstrated over 40 years to any environmental factor. For instance, the example of temperature is given. In that case would it not be reasonable to expect a similar change across many related species if the environmental factor is that general?

    Intuitively speaking, it just doesn’t seem likely that such long term trends are occurring due to environment due to variability. Genetic change seems most likely.

    • Cremnomaniac
      Posted March 30, 2013 at 7:00 pm | Permalink

      Correction? Maybe its more appropriate to call it genetic expression.

  15. infiniteimprobabilit
    Posted March 31, 2013 at 12:54 am | Permalink

    Isn’t there a paradox inherent in deducing characteristics from the roadkilled birds? Namely, that the population of swallows living in close proximity to the highway could be expected to get better at avoiding cars; but apparently the birds the experimenters were examining were precisely those birds which had failed to do that thing – so, atypical of the ‘new’ swallow population.

    Or do I just like paradox too much?

  16. HaggisForBrains
    Posted March 31, 2013 at 5:19 am | Permalink

    In my lifetime of driving (nearly 50 years) I have covered many miles on the unfenced minor roads of the Highlands of Scotland. Most of the land there is sheep farming land, and it seems to me that nowadays the sheep are far less likely to wander out into the road in front of my car than they were in my youth. I also see far fewer dead sheep by the roadside nowadays. This in spite of a far greater number of cars on these roads now, faster, although perhaps with better brakes. Obviously this is purely subjective, but I have often wondered if I have been observing natural selection in action.

    • infiniteimprobabilit
      Posted April 1, 2013 at 1:50 am | Permalink

      Here in New Zealand I’ve noticed the same thing. Used to be, that if you drove through a paddock, sheep 50 yards from the road would make a desperate dash to reach the road ahead of you so they could then run, panic-stricken, down the middle of the road six inches ahead of the car. Why this behaviour I have no idea. Did they see my car as some off-road-capable predator and the road as their escape route, or what?
      But in recent years there seem to be more sheep who just mind their own business and ignore my car.

      Goats, in contrast, always did ignore cars. You often see them tethered by the roadside to keep the grass down.

      Back to birds – one species which is notoriously car-savvy is the Indian mynah bird (an introduced species here, though I suspect they actually just stowed away on a boat and didn’t bother about a formal introduction). We see them all the time on the roads (alive!), presumably looking for insect-roadkill; it is impossible to run them over. They see your car coming, and walk (not run) the necessary couple of feet to take them clear of your line. Always.

%d bloggers like this: