Natural Selection Is Not Predictable
Every once in awhile, biologists argue over whether evolution is predictable. The latest flap over stick insects sticks up for predictability, but flops.
An international group of scientists, publishing in the journal Science, studied stick insects—those long, skinny walking insects that try to blend in with plants by mimicking twigs. In their paper, “Natural selection and the predictability of evolution in Timema stick insects,” they do their best to estimate the trajectory of these bugs. Laura Zahn, however, in a summary of the paper in the same issue of Science, has this to say:
Evolution results from expected effects, such as selection driving alleles toward fixation, and stochastic effects, such as unusual environmental variation and genetic drift. To determine the potential to predict evolutionary change, Nosil et al. examined three naturally occurring morphs of stick insects (see the Perspective by Reznick and Travis). They wanted to determine which selective parameters could be used to foresee changes, despite varying environmental conditions. One morph fit a model of negative frequency-dependent selection, likely owing to predation, but changes in other morph frequencies remained unpredictable. Thus, for specific cases, we can forecast short-term changes within populations, but evolution is more difficult to predict when it involves a balance between multiple selective factors and uncertainty in environmental conditions.
According to Zach Gompert at Utah State University, one of the authors, the predictability is hardly surprising: brown stick insects would be found on brown plants, and green stick insects would be found on green plants. The reason is that birds can more easily see the out-of-place morphs and eat them. This explains why out-of-place insects would be missing, but says little about the arrival of the camouflaged species. A USU press release says that the team analyzed 25 years’ worth of data to try to figure out if evolution is predictable.
“With the green versus green-striped morphs, the cause of selection was simple and well understood facilitation of predictability,” Gompert says. “In contrast, with the melanistic morph, natural selection was more complex and tied to variation in weather and climate, making it harder to predict from past patterns of change.”
The scientists compared their results to better known studies, including Darwin’s finches and the scarlet tiger moth, both of which were also not very predictable.
“Our findings support previous discoveries and suggest evolution of morph frequencies in these stick insects is indeed a result of selection,” Gompert says. “They also suggest poor predictability of environmental variation and how it affects selection, rather than random evolutionary processes, might be the main limits on predicting evolution.”
While we can use the past to predict change, he says, we’re constrained by our lack of knowledge of the future and complex ecological processes that contribute to change.

c. Brett Miller. Used by permission.
It’s hard to characterize any of this data support for evolution being predictable. They’re basically saying, ‘evolution is predictable except when it isn’t.’ Reznick and Travis sum up the results:
Evolution is like population dynamics because evolutionary change over time can be governed by multiple factors, the relative influence of which vary over time. Nosil et al. used a series of observational data taken over 25 years on natural populations in combination with experiments to show that in one case, evolution can be predicted very well, but in another, it cannot. More generally, they show that without deep biological knowledge, we cannot understand either past or future, much less predict the future from the past.
The problem is not just with stick insects. It extends to all of biology:
Questionable predictability is not specific to stick insects. Nosil et al. analyzed data sets for other long-term studies of evolution in various species, including Galapagos finches and the peppered moth, and show that they also offer low temporal predictability. In these cases, the likely cause is also multiple forms of selection the strength of which varies over time.
Interesting that they would present finches and peppered moths, both of which are “icons of evolution” featured in the list by Jonathan Wells, yet say they were subject to ‘multiple forms of selection.’ Why not simple ‘natural selection’ that strikes so many evolutionists as intuitively obvious? Now, we find, things are not so obvious after all. It’s complicated to predict even one thing on which natural selection might act:
These results show that an iconic example of a simple trait subjected to a single agent of strong selection is actually much more complicated. Similar lessons have been taught by other seemingly simple phenomena. For example, the complex ways in which known agents of selection on the color polymorphism of Cepaea snails meant that “each population is subject to a unique explanation”. This is in stark contrast to studies of microbial, viral, and immune system selection, for which evolution seems to be highly predictable. Why this is the case, when it is not so in organisms such as stick insects and others, is a new challenge for evolutionary biologists.
So the environment is unpredictable, selection is unpredictable, and mutations are clearly random. Adding three random factors together does not improve on randomness. After 158 years of Darwinian evolution, what has been accomplished to improve scientific understanding other than to say, “Stuff happens”?
We like to periodically back up our claim that Darwinism reduces to the Stuff Happens Law. It explains everything; it explains nothing. This is how a stupid idea can put on invisible royal robes and masquerade as an emperor of understanding. Look at these proofs of the Stuff Happens Law we presented earlier. Don’t you feel wiser knowing them?
Why the Stuff Happens Law is Scientific
- It is reductive: all events can be reduced to this law.
- It makes predictions: Stuff will happen.
- It is universal: Stuff always happens.
- It is normative, not just descriptive: Given matter in motion, stuff must happen.
- It is falsifiable: If nothing happens, the law has been disproved.
- It is practical: If something happens, you know you will find stuff around.
- Corollaries can be derived from it: e.g., Stuff happens at the worst possible time, Bad stuff happens to good people, Murphy’s Law, etc.
Impressed? Darwin’s laws of nature are about as helpful to the understanding of nature as the Stuff Happens Law. Your science might be healthier with a bit of Cole’s Law (i.e., thinly sliced cabbage).