Plant Evolution Modeled in Computer
Simulation games are popular on computers. Darwinian biologists seem to like them, too. What they cannot go back in time to observe, they sometimes try to recreate in silico, inside the silicon chips of a computer. Karl J. Niklas (Cornell) tried to simulate plant evolution, and wrote about it in Annual Review of Earth and Planetary Sciences.1 He feels his contribution was to demonstrate that plants had to be multitasking specialists: they optimized competing interests in a dynamic environment, rather than achieving perfection with any one structure. This involved tradeoffs; a horizontal stem might provide the best light-gathering stance, for instance, but puts the plant at the burden of having to fight gravity’s leverage:
Indeed, when viewed with a biophysical or engineering perspective, none of the basic biological tasks plants perform can be maximized without decreasing or imperiling the performance of another necessary task. In this sense, the relationships among organic form-function generally involve optimization rather than maximization. But differently, single-tasked devices can perform their ascribed functions perfectly, at least in theory. In contrast, multitasked devices, whether organic or inorganic, invariably involve compromises and tradeoffs—they perform all of their ascribed tasks reasonably well, but no task perfectly.
(For more on evolutionary tradeoffs, see 05/11/2004 entry.)
The origin of land plants “sparked one of the most dramatic bursts of diversifying evolution in the history of life,” he claims, indicating the motivation for this project. In just 46 million short years from the Silurian through the Devonian, these pioneering plants had “diversified phyletically and structurally to encompass all of the major land plant lineages and the full spectrum of organizational grades represented in present-day floras, with the exception of flowering plants.” He lists 11 innovations they introduced, from branching stems to leaves to stomata with guard cells to seeds and wood. They employed sexual reproduction with alternation of generations and diversified into an enormous number of morphologies, from mosses and ferns to pines and giant redwoods. “Why plant evolution was so rapid during the Late Silurian-Devonian time interval remains problematic,” he admits. “Lessons drawn from evolutionary theory provide limited insights” Thus, computer modeling to the rescue.
For his model, Niklas used a principle proposed by Sewall Wright in 1931: the fitness landscape, a “heuristic device” that visualizes evolution as “a series of walks over fitness landscapes with adaptive hills and maladaptive valleys.” On this landscape, Niklas placed his digital plants and gave them four competing problems to solve: (1) water conservation, (2) mechanical stability, (3) spore dispersal, and (4) light interception. He defined the fitness of each combination and set the plants on their “adaptive walk” on the fitness landscape (peaks on the fitness landscape imply high fitness and good adaptation, and valleys imply poor adaptation and low fitness). First, he used a stable fitness landscape, then he ran it again with a dynamic landscape, which would reflect a more realistic environment changing over time. He found that overall fitness levels dropped considerably in the dynamic fitness landscape. How does one decide when to vary the landscape? “Unfortunately, there are no a priori rules for how or when a particular landscape changes,” he says. “Therefore, the number of permutations of shifting landscapes is literally astronomically large.” So he looked to the fossil record for guidance, and also tried to learn from repeated trials what seemed to match natural history.
In a brief aside, he compared his results to the predictions of Zimmerman’s telome theory – the idea that all of the diverse morphologies of plants can be reduced to the action of five developmental processes – planation, overtopping, reduction, recurvature, and webbing – acting on branched points (telomes) and unbranched points (mesomes). But telome theory is far from a complete story:
The telome theory has been criticized, and rightly so, for a variety of reasons (Niklas 2000, Kaplan 2001). One obvious problem with the theory is its vagueness regarding the developmental mechanisms responsible for overtopping, planation, etc. Indeed, these terms are descriptive rather than explicative in nature. Another criticism is that the telome theory never explains why certain morphological transformations occur as opposed to others, nor does it stipulate the sequence of processes foreshadowing the appearance of a particular morphology. Why should planated and webbed lateral branch systems evolve? Are the leaves of ferns or seed plants functionally adaptive in terms of light interception or some other biological requirement? Did these megaphylls [broad leaves] evolve as the result of the simultaneous operation of reduction, overtopping, planation, and webbing, or did planation and webbing occur after reduction and overtopping? Questions such as these can be answered retrospectively (and only in small part) by examining the fossil record, but the telome theory sheds little light on them.
So why use it? Because the terminology is useful: “Zimmerman’s ideas are nevertheless useful because they provide a lexicon of terms for the morphological transformations observed in the fossil record and for those identified by the computer simulations presented here. In turn, these simulations suggest the adaptive significance of the transformations envisioned by the telome theory.”
Niklas produced some digital plants that succeeded in adapting to his fitness landscape, but warned against overinterpreting the results. In his concluding “Caveats and Desiderata,” he said,
Computer models such as the ones presented here are heuristic tools. They provide an opportunity to test assumptions about how a particular biological or physical system operates or behaves. Their validity can be evaluated by comparing predicted with observed behavior. When observation and prediction disagree, the assumptions upon which a model rests are either incorrect or incomplete. However, the obverse is not true. When predicted and observed behavior agrees, the assumptions upon which a model rests cannot be said to be sufficient and necessary. The reason is simple — model can describe the behavior of a system for the wrong reasons. This caveat is important, because the only rigorous test of a computer model is to experimentally manipulate the system it purports to describe and to see if the model predicts the outcome for each manipulation.
Niklas did not perform any such rigorous experimental tests with real plants. He explains why, but still claims his model had merit:
Unfortunately, we cannot experiment with history. We can only observe it. For this reason, the most conservative interpretation of the simulations presented here is that six general properties emerge logically (mathematically) from the assumptions made about early vascular plant evolution. These properties are as follows: (a) the number of equally fit morphological variants is predicted to increase as the number of functional tasks subject to selection increases; (b) the relative fitness of these phenotypes decreases as the number of tasks increases; (c) therefore, morphological diversification is easier on complex as opposed to simple fitness landscapes; (d) constraints on how morphology can be developmentally altered do not a priori limit the number of equally fit variants that can be reached by adaptive walks; (e) however, the relative fitness of these variants is significantly lower than the phenotypic optima that can be reached by unfettered adaptive walks; and (f) adaptive walks on shifting fitness landscapes (used to mimic changes in the focus of selection) identify morphological optima that often differ significantly from those on stable fitness landscapes (used to mimic constant selection).
He points to a few living vascular plants as confirmations of these general predictions, and concludes that the six properties also make biological sense. Feeling thus justified, he concludes,
Computer simulations of morphological evolution are still very much in their infancy, especially in terms of constructing morphospaces and understanding the developmental mechanisms that permit or confine phenotypic transformations in them (see Thomas & Reif 1993, McGhee 1999, Niklas 2003). However, as conceptual tools, they provide opportunities to explore the logical consequences of popular metaphors for evolution, such as Sewall Wright’s adaptive walks on fitness landscapes, and by so doing, quantify the possible biological structure and dynamics of opportunistic historical events that distinguish some evolutionary episodes as more adaptive than others.
1Karl J. Niklas, “Computer Models of Early Land Plant Evolution,” Annual Review of Earth and Planetary Sciences, May 2004, Vol. 32, pp. 47-66 (doi:10.1146/annurev.earth.32.092203.122440).
You can prove anything on a computer. This is so oversimplified, so narrow-minded, so dumb, it’s a wonder any journal would publish such tripe. It’s only because biologists have offered their brains in sacrifice to Darwin’s image that they cannot see the illogic of their own positions. Niklas came close, and had a gem of insight here or there,2 but failed to see the worthlessness of his simulation. His fake plants evolved because he made them evolve. We’ve seen this so many times before with other computer models. It is not evolution, it is intelligent design. These modelers set the fitness goals, define the criteria for success, and reward the ones that get there. Natural selection has no such guiding intelligence.
2(One perceptive insight he shared was that agreement with predictions does not necessarily make the assumptions of one’s model sufficient and necessary. But then, contrary to his requirement for rigorous experimental testing, he failed to deliver any.)
To get a sense of the futility of this model, imagine my writing a simulation about the evolution of computers. I define fitness scores for screen visibility, mouse responsiveness, keyboard ergonomics and other factors I deem worthy, then start some high-tech devices evolving and reward those that succeed in terms of outward conformance to my specifications. Maybe I wind up with a variety of objects that look like palmtops, laptops and desktops. But I describe nothing about programmers or users, nor the thinking required to make a computer. Is such a result worth anything more than a cheap science-fiction game for kids?
Let’s understand something important here. Plants have DNA. They are adapted because they have complex, specified information in their genes. Watching little digital organisms evolve branches and leaves and other structures might be cute, but says absolutely nothing about how the genetic information and developmental pathways achieved the structures, and more importantly, says nothing about the intricate cellular processes, like photosynthesis and cell division and sexual reproduction and regulation of stomates originated and employed the morphologies. Niklas halfheartedly admitted as much. He should know that plants at the cellular level are fantastically complicated factories of molecular machines. Which is more intricate: the cover of your computer or the chips and software inside? If I only pay attention to the morphology of the external parts, I have missed the whole point of what is required to make a computer, or a plant.
The “fitness landscape” metaphor is only a metaphor, but it is actually a more accurate metaphor than a ramp. Early evolutionists mistakenly pictured Darwinism like a ramp, on which organisms marched onward and upward. “Progressivists” viewed natural selection as a “fitness ratchet” leading inevitably to bigger and better things. Knowledgeable evolutionists today realize this view was simplistic. There are peaks and valleys of fitness (whatever that is; see Fitness for Dummies, 10/29/2002). Picture marbles rolling around on a surface constantly in motion, with peaks and pits forming and reforming at random locations. Real marbles might roll uphill for short periods, but gravity will ensure they tend to inhabit the valleys and pits most of the time. The gravity in the fitness landscape is the second law of thermodynamics – the inviolable trend toward entropy. Evolutionists want us to believe that natural selection will overcome this entropic gravity and force the marbles to the tops of the peaks. Trouble is, even if they got there and stayed there, they would be stuck, unable to evolve further without dropping down into the valley again and losing what fitness they had. Now realize that the adaptive peaks are like Devils Towers and Space Needles: the exquisite engineering seen in whale flippers and cormorant eyes and spider silk and the other things the biomimetic engineers marvel at are so improbable as to be unthinkable for undirected processes to achieve (despite Richard Dawkins’ claims that chance and natural selection can “climb Mt. Improbable” another fallacy “proved” by worthless computer simulations). No matter, it’s just a metaphor, and metaphors bewitch you. Engineering doesn’t emerge without design except in the imagination of evolutionists.
Niklas said, “Unfortunately, we cannot experiment with history. We can only observe it.” When was the last time you observed history? If you watched a historical event like 9/11, you observed it in the present. Did you observe the fall of the Roman empire, or the building of Stonehenge? Did Niklas observe the origin of land plants? We don’t observe history. We believe eyewitness accounts and examine artifacts. Even recordings (artifacts viewed in the present) can have biases, and there were no videotapes of the origin of plants, anyway. We can observe fossils as they are in the present, but can only infer how they got there; piecing scattered fossils into a sequence is even more fraught with difficulties (see 05/21/2004 headline). There is no observable history of evolution. There is a story imposed on the artifacts seen in the present. An eyewitness account from a credible witness that can be corroborated by observation is superior to a story weaved out of personal bias and propped up by circumstantial evidence.
By the way, there is a credible eyewitness account that fits the evidence. It will tell a biologist all he needs to know about the emergence of plants and how they achieved their high levels of adaptive fitness. It explains not only the outward morphology, but the programming and developmental constraints that maintain fitness. It’s in that best seller in your hotel room drawer. Let’s start at the very beginning, a very good place to start. (Nothing comes from nothing; nothing ever could.)