Creationists will tell you that the Bible is the 'complete and inerrant' record of the history of life on planet earth and an accurate scientific description of the life on it, and yet it's so inadequate that they've had to supplement it with all manner of fanciful additions to make any sense of it and to make it look slightly more connected to reality.
They then get themselves into a hopeless muddle trying to define a 'kind’, so it conforms to their nonsensical taxonomy. They will tell you one 'kind' can't interbreed with another 'kind' and then need to use a different definition of 'kind' when you mention hybrids, ring species and other evidence of slow divergence over time as barrier to hybridization and genetic remixing evolve, before what science recognises as genetically distinct genera and species.
In any social media group or creationists disinformation site, you'll see a typical double-think approach to evolution: on the one hand we'll be assured that there is no evidence for evolution (based on some childish parody definition of evolution); on the other hand, you'll be assured that, after two of each 'kind' of animal got off the mythical Ark, they underwent a period of warp-speed evolution, producing several, entirely new taxons each generation in different parts of their range to give us the millions of different families, species, sub-species and varieties we see today, which, coincidentally, no-one reported noticing, not even the people who left rock and cave paintings recording what they saw in the world around them several thousand years ago.
But of course, evolution is what it has always been since biologists gained insight into how life changes in response to environmental change - change in the frequency of different alleles in the population over time.
These changes may or may not include new structures or morphology that taxonomists base their classifications on, and they may or may not include changes in coloration (which is a morphological change just as much so as is a new structure).
Here, for example, is the recent discovery by a team lead by Zachariah Gompert of the Department of Biology, Utah State University, Logan, UT, USA, that the stick insect species, Timema cristinae, has had several predictable, repeats of the same pattern of evolution in response to cyclic changes in their environment.
The species occurs in two colour forms - a uniform green, which cryptically camouflages it as leaves of the Californian lilac shrub, Ceanothus spinosus, and a green-striped form which cryptically camouflages it as the leaves of the chamise shrub, Adenostoma fasciculatum.
What the team discovered is the subject of both an open-access paper in Science and a news release from Utah State University:
A longstanding debate among evolutionary scientists goes something like this: Does evolution happen in a predictable pattern or does it depend on chance events and contingency? That is, if you could turn back the clock, as celebrated scientist Stephen Jay Gould (1941-2002) described in his famous metaphor, “Replaying the Tape of Life,” would life on Earth evolve, once again, as something similar to what we know now, or would it look very, very different?
“If you frame it as an either/or question, it’s too simplistic,” says Utah State University evolutionary biologist Zachariah Gompert. “The answer isn’t ‘completely random’ or ‘completely deterministic and predictable.’ And yet, examining short time scales, we can find predictable, repeatable evolutionary patterns.”
Gompert and colleagues report evidence of repeatable evolution in populations of stick insects in the May 24 online edition of the American Association for the Advancement of Science’s journal Science Advances. Contributing authors on the paper include Gompert’s long-time collaborator Patrik Nosil and other researchers from France’s University of Montpelier, Brazil’s Federal University of São Paulo, the University of Nevada, Reno and Notre Dame University. The research is supported by the National Science Foundation and the European Research Council.
The team examined three decades of data on the frequency of cryptic color-pattern morphs in the stick insect species Timema cristinae in 10 naturally replicate populations in California. T. cristinae is polymorphic in regard to its body color and pattern. Some insects are green, which allows the wingless, plant-feeding insect to blend in with California lilac (Ceanothus spinosus) shrubs. In contrast, green striped morphs disappear against chamise (Adenostoma fasciculatum) shrubs.
Hiding amongst the plants is one of T. christinae’s key defenses as hungry birds, such as scrub-jays, are insatiable predators of the stick insects.|
“Bird predation is a constant driver shaping the insects’ organismal traits, including coloration and striped vs. non-striped,” says Gompert, associate professor in USU’s Department of Biology and the USU Ecology Center. “We observed predictable ‘up-and-down’ fluctuations in stripe frequency in all populations, representing repeatable evolutionary dynamics based on standing genetic variation.”
He says a field experiment demonstrates these fluctuations involved negative frequency-dependent natural selection (NFDS), where cryptic color patterns are more beneficial when rare rather than common. This is likely because birds develop a “search image” for very abundant prey.
“At short time scales, evolution involving existing variations can be quite predictable,” says Gompert, who received a National Science Foundation CAREER grant in 2019 to support his research. “You can count on certain drivers always being there, such as birds feeding on the insects.”
But at longer time scales, evolutionary dynamics become less predictable.
“The populations might experience a chance event, such as a severe drought or a flooding event, that disrupts the status quo and thus the predictable outcomes,” Gompert says.
On long time scales, a new mutation in the species could introduce a rare trait, he says. “That’s about as close to truly random as you can get.”
“Rare things are easily lost by chance, so there’s a strong probability a new mutation could disappear before it gains a stronghold,” he says. “Indeed, another species of Timema stick insect that also feeds on chamise either never had or quickly lost the mutations making the cryptic stripe trait. Thus, the evolution of stripe is not a repeatable outcome of evolution at this long scale.”
Gompert notes replicated, long-term studies from natural populations, including research on the famous Darwin’s finches, are rare.
“Because most of this work is restricted to one or few populations, it is difficult to draw inferences on repeatability among multiple evolutionary independent populations,” he says. “Such studies are challenging to implement not only because they take concerted effort, but also because you can’t rush time.”
AbstractIn summary, the population of the different morphs changes according to the strength of the selectors in the environment, in a predictable way, at least in the short-term, predictable being in the way the Theory of Evolution by Natural selection predicts. There is no suggestion here that some other, supernatural, force is involved in the process or that the biologists think the TOE is inadequate for explaining the observations. Quite the contrary, in fact, the TOE is confirmed by the observations.
The extent to which evolution is repeatable remains debated. Here, we study changes over time in the frequency of cryptic color-pattern morphs in 10 replicate long-term field studies of a stick insect, each spanning at least a decade (across 30 years of total data). We find predictable “up-and-down” fluctuations in stripe frequency in all populations, representing repeatable evolutionary dynamics based on standing genetic variation. A field experiment demonstrates that these fluctuations involve negative frequency-dependent natural selection (NFDS). These fluctuations rely on demographic and selective variability that pushes populations away from equilibrium, such that they can reliably move back toward it via NFDS. Last, we show that the origin of new cryptic forms is associated with multiple structural genomic variants such that which mutations arise affects evolution at larger temporal scales. Thus, evolution from existing variation is predictable and repeatable, but mutation adds complexity even for traits evolving deterministically under natural selection.
INTRODUCTION
The extent to which evolution is repeatable and predictable is central to understanding the role of determinism and chance in the history of life, with implications for both basic and applied science (1–4). These ideas are captured in Gould’s famous metaphor of “replaying the tape of life” (5). Gould argued that historical contingency and chance idiosyncrasies would result in different (i.e., non-repeatable) evolutionary outcomes if the history of life was to be replayed over and over again. However, others such as Morris (6) have argued that evolution is inherently predictable, and many examples of deterministic natural selection promoting repeatable and predictable elements to evolution do exist (3, 7). Thus, beyond spurring decades of debate, Gould’s metaphor has been usefully transformed into an empirical research program (1, 8).
Comparative phylogenetic studies of parallel evolution often support repeated outcomes driven by natural selection (1, 7, 9). However, inferring evolutionary processes and their interplay from such retrospective work can be challenging. For example, selection that fluctuates rapidly between time points can be misconstrued as evolutionary stasis if only end outcomes, or a few distant time points, are analyzed. Detailed studies of the fossil record and ancient DNA can help address these issues (10, 11), as can real-time studies of evolutionary dynamics. The latter are exemplified by studies of experimental evolution in replicated laboratory populations of microbes (12–14) and other organisms, such as insects (15, 16). This impressive body of work has revealed not only repeatable patterns of evolution by natural selection but also a role for the contingency of mutation.
In contrast to laboratory experimental evolution studies, replicated, long-term studies from natural populations are rare. Certainly, highly influential long-term studies of the predictability of evolution in the wild do exist, in finches and other birds (17–20), moths and butterflies (21–23), stickleback and guppy fish (24–27), sheep (28), fruit flies (29), and deer (30) (to name a few). For example, rare climatic events affect evolution in the famous Darwin’s finches (19, 20). However, most such work is restricted to one or few populations, making it difficult to draw inferences on repeatability among multiple evolutionarily independent populations (note the distinction between predictability within a population and repeatability among populations, Fig. 1). Thus, replicated studies of evolution in the wild are required to test the generality of the findings from microbes (12–14) and to help bridge now disparate laboratory and field studies. Such studies are challenging to implement not only because they take concerted effort but also because time cannot simply be sped up with more effort.Here, we provide such a study on the basis of compiled data from 30 years of tracking morph frequencies across 10 replicate populations of a stick insect in the wild (Fig. 1). We integrate these data with field experiments, modeling, and genomic data to elucidate the processes driving evolutionary dynamics across timescales. In particular, our genomic analyses allow consideration of the contingency of mutation, which can add complexity and nuance relative to the sole consideration of evolution from standing genetic variation (31). Specifically, mutational dynamics can be related to the genetic architecture of traits, with consequences for the repeatability of evolution at the genetic versus phenotypic level (32, 33). Some predictions are as follows and illustrated in Fig. 2.
Fig. 1. Evolution in replicate long-term field populations of T. cristinae stick insects.
(A) Illustrations of the color-pattern morphs of T. cristinae. (B) Predictions of highly repeatable evolutionary dynamics over time. Each line represents a different population, each exhibiting predictable “up-then-down” fluctuations in trait or gene frequency over time. (C) Predictions of less repeatable evolutionary dynamics over time. In contrast to the panel to the left, each population exhibits different patterns of trait or gene frequency change over time. Note the distinction between predictability within any single time series (i.e., population or replicate) and repeatability among them. (D) Empirical variation in morph frequencies in T. cristinae between 1990 and 2023. The orange line (median) and shading [95% equal-tailed probability interval (ETPI)] represent yearly averages on the host plant Adenostoma. The blue line (median) and shading (95% ETPI) represent yearly averages on the host plant Ceanothus. (E) Population-specific morph frequency variation over time, representing replicate evolutionary dynamics (mean number of years per population = 14). Results are shown for the 10 core populations, i.e., replicates, that this study focuses on.When traits are controlled by simple genetic architectures composed of one or few loci of large effect, then individual mutations can strongly affect evolutionary dynamics (4, 34). In such cases, individual mutations “matter” and mutation can be an important consideration for understanding evolution. For example, such architectures offer few mutational targets, making mutations that improve fitness rare (4, 34). Moreover, if there are strong interactions between loci (i.e., epistasis), then only a few genetic combinations may work well together to increase fitness. In other words, there are few mutational sequences or “paths” that adaptive evolution can actually take (4, 34–36). Thus, traits controlled by few loci may be challenging to evolve de novo and thus show modest repeatability in their origin over time. However, when such traits do evolve, they will have a predictable, repeated genetic basis (i.e., using the same, few loci and mutations that increase fitness). These predictions could be different for more polygenic architectures composed of many loci with smaller effects (37). Here, no individual mutation has a strong effect on a trait, many loci have redundant effects, and path dependency is unlikely to force a particular mutational sequence to be used to evolve higher fitness. Thus, traits controlled by many loci may evolve repeatedly because they can use a wide range of mutational variation. However, when they do so, a different set of loci and mutations is likely to be used in each instance, making evolution at the genetic level not very repeatable (37).Fig. 2. Conceptual diagram illustrating the relationship between the genetic architecture of a trait and the repeatability of evolution at the genetic and phenotypic levels (where the tips of each phylogeny represent a different taxon/species). (A) Hypothetical scenario where a trait (i.e., stripe) depends on a specific mutation. In this case, the evolution of stripe is contingent on the specific mutation occurring and thus occurs only once in the hypothetical evolutionary scenario depicted on the phylogeny. Thus, phenotypic evolution is not repeatable (it would, however, have a repeatable genetic basis if the trait evolved in multiple species). (B) An alternative scenario where stripe depends on a few loci. Here, the evolution of stripe is still contingent on a small number of specific mutations occurring and this might make the trait evolve less repeatedly. Moreover, in such cases, evolution might (or might not) be dependent on the order in which the mutations occur (i.e., some mutations might only be beneficial if another mutation has already occurred). (C) A third alternative scenario where stripe is affected by many mutations (i.e., is polygenic). In this case, stripe does not depend on a specific, unlikely mutation, and is unlikely to depend on the order of mutations. This genetic architecture thus makes the repeated evolution of stripe much more likely, occurring in all four species in the hypothetical example. However, the genetic basis of stripe varies among the species, and, thus, the genetic basis of stripe is not repeatable (different combinations of alleles, i.e., mutations, can generate a stripe). In all diagrams, + and − symbols denote alternative alleles, with + contributing to the stripe phenotype.
We acknowledge that there can be much nuance beyond the stylized ideas and predictions noted above [e.g., see (38)]. However, these ideas, nonetheless, illustrate why it can be informative to study the genetic architecture of traits that are being analyzed for their repeatability, particularly to understand the dynamics of evolution from standing variation versus new mutation. In this context, we here combine our time-series and experimental work with genomic analyses of the genetic architecture of cryptic color pattern. Our results reveal clear similarities but also differences with past work focused on body color (rather than pattern, see below for details). Our integrative approach leads to a more complete understanding of the repeatability of evolution than would be possible by studying selection or mutation in isolation.
Ecology and genetics of Timema stick insects
Our study system is Timema stick insects (Fig. 1), a genus of wingless, plant-feeding insects found throughout southwestern North America (39). We focus primarily (but not exclusively) on Timema cristinae, which exhibits three highly heritable morphs and uses two primary host plant species (40, 41). Two of these morphs have diverged in frequency between host species due to strong divergent selection imposed by visual predators such as lizards and birds (42, 43). Specifically, a green unstriped color-pattern morph (green morph hereafter) is cryptic on the broad leaves of Ceanothus. In contrast, the striped color-pattern morph (a green morph that also bears a white, longitudinal stripe on its dorsal surface) is cryptic on the thin needle-like leaves of Adenostoma. Thus, each morph is generally more common on the host on which it is more cryptic, which led to a body of past work on the potential for divergent adaptation between hosts to drive ecological speciation (41, 44). However, we stress that polymorphism is maintained such that both green and striped morphs occur in most populations. This variation is maintained due to negative frequency-dependent selection (NFDS) and gene flow, creating a mosaic of variation in morph frequencies across the landscape, both within and among populations (45, 46). Specifically, the frequency of the striped morph varies within local populations ranging from low (near zero), to intermediate, to high (near one) but is rarely truly fixed (Fig. 1). This provides the requisite variation to study changes in the frequency of green versus striped morphs over time, which we focus on here. There is also a third darkly colored or melanic morph that is rarer and found at comparable frequencies between the two hosts. The melanic morph is not specifically adapted to either host (40, 47) does not fluctuate strongly in frequency over time and does not appear subject to NFDS (40, 45, 47). Thus, we focus first on testing for repeatable fluctuations in color-pattern morphs but return to melanism later in this study when considering longer timescales. We also return to the relationship between polymorphism and speciation at the end of our study.
Past observations and a manipulative field experiment documented changes in color-pattern morph frequencies that were highly predictable due to NFDS (i.e., a fitness advantage to rare forms) (45). This likely occurs due to birds switching search images to hunt for common prey items (48–51). Thus, increases in the frequency of the striped morph one year were reliably followed by decreases the following year and vice versa. This past work focused on changes in a single 18-year time series (i.e., a single population or “play of the tape”) because substantial long-term data from other populations did not exist at the time. Thus, past work could not address the issue of repeatability or “replays,” i.e., among replicate populations, per se. We thus focus here on the repeatability of evolution among populations.
We report results from 30 years of data collection on morph frequencies in T. cristinae, representing 692 year-by-host-by-locality estimates of morph frequency derived from 48,349 individuals (Fig. 1 and data S1). Most critically, these data represent 10 localities with at least a decade of the data required to test for NFDS (see table S1; mean, 14 years; maximum, 22 years). These 10 localities were chosen to represent replicates, for example, because of little to no gene flow among them such that each locality undergoes independent yearly changes in morph frequency. Supporting this claim, we emphasize that our 10 localities are geographically separated (fig. S1), generally by several kilometers, yet the average per-generation dispersal distance based on a mark-recapture study is only 12 m (52). This makes it very unlikely that there is sufficient migration between our localities to generate detectable morph frequency fluctuations. Molecular data further support limited gene flow between our localities and thus their evolutionary independence (especially in the context of yearly changes in morph frequency). For example, genetic data have shown that even parapatric populations that are directly adjacent to one another exchange only a few migrants per generation, and all of our study localities here are geographically separated from each other such that gene flow among them is even lower (fig. S1) (53, 54). Thus, each locality acts as a replicate or replay for analyzing the repeatability of evolution, particularly because the distance (i.e., kilometers) typically separating each locality make it likely that different bird and lizard individuals hunt insects at each locality.
Notably, T. cristinae is univoltine with nonoverlapping generations. Thus, each year of data represents a generation, with evolution occurring between each pair of years (the insects diapause as eggs through autumn and winter, hatch in spring, and mate and die in early summer, repeating this cycle each generation). Despite these data representing an appreciable effort, the timescales involved are very different from those associated with Gould’s metaphor. We thus also consider mutations affecting color pattern over deeper evolutionary time (55).
Here, the genetic architecture of color pattern is relevant, for reasons outlined above and in Fig. 2. Specifically, we here evaluate the extent to which stripe depends on specific mutations that could make the repeated evolution of the trait less likely. Past work suggests that pattern is controlled by one or few loci on chromosome 8 (chr8 hereafter) (40, 47, 56). However, the number, effect sizes, and physical distributions of the loci involved remains unclear. Specifically, past work described a region of chr8 named the Mel-Stripe locus that harbors complex structural variation that explains most of the variation in body color (a large inversion and deletion distinguish green versus melanic morphs, indicative of suppressed recombination and opening the potential for linked selection) and is also partially associated with color pattern (striped versus green). However, whether one or multiple regions of Mel-Stripe or even other loci are associated with pattern remains unclear due to (i) the focus of past work on color (not pattern) and (ii) the use of a fragmented reference genome. Here, we integrate better chromosome-level genome assemblies with genome-wide association (GWA) mapping to show that color pattern is actually associated with multiple different structural variants within the Mel-Stripe locus, as well as a chromosomal inversion in a region not associated with color or pattern in past work (i.e., the “Pattern” locus, which is reported here for the first time). We conclude our study by discussing what these genetic details tell us about the repeatability of evolution.
Fig. 9. Two solutions to the problem of crypsis on Adenostoma (in T. cristinae versus T. podura) and an experimental test of which solution may offer higher fitness.
(A) Conceptual overview of known patterns of crypsis. Both T. cristinae and T. podura use Ceanothus and Adenostoma as host plants and both exhibit a green unstriped morph that is cryptic on Ceanothus. In contrast, the two Timema species have evolved different morphs that are cryptic on Adenostoma, the striped morph in T. cristinae (left) and a melanistic gray/brown morph in T. podura (right). (B) Block-specific results of the mark-recapture field experiment testing the fitness of striped versus melanistic morphs of T. cristinae on Adenostoma. Shown is the number of each morph recaptured in each block (i.e., replicate), demonstrating higher recapture and thus fitness of the striped morph. (C) Average fitness of each morph across the entire experiment with the posterior distributions summarized in box plots (boxes denote the median, first and third quartiles of the posterior with whiskers extending to the minimum and maximum or 1.5 times the interquartile range) and individual parameter values sampled from the posterior overlain as points.
