Identifying Patterns of Evolutionary Change
Outgroups: Polarizing Ancestral and Derived Traits
The Principle of Parsimony and the Construction of Phylogenetic Trees
Phylogenetic Inferences on Adaptation
Were Dinosaurs Dead-beat Dads?
Evolution of male parental care in sea horses and pipefish
Evolution of aposematic coloration and gregariousness
Sensory Bias and Sexual Selection
Throughout the readings we have focussed on a microevolutionary perspective regarding the evolution of behaviors. In our consideration of mechanisms of natural and sexual selection, we focussed on the behavioral traits being shaped by the success of individual organisms. The mechanisms of natural and sexual selection are referred to as microevolutionary processes. The power of the natural and sexual selection can lead to rapid evolution in a trait that is favored by adaptation. Adaptation is subject to constraints that limit the direction that evolution can take a species. Constraints on evolutionary change can arise at a number of levels. We have already focussed on constraints on adaptation that arise from the design of sensory structures, as well as the physical environment around the organism -- these are the constraints of proximate mechanisms.
A number of authors have advocated the idea that a species is constrained by its own phylogenetic history. During the evolution of a group, certain features were developed that served to limit the possible development of other features. For example, once birds developed wings and powered flight, this adaptation makes it necessary to limit the number of eggs that a female is carrying at any point in time to only a few eggs. Thus, most birds labor under the constraint that they cannot produce the eggs all at once like a lizard and they lay eggs in sequence one at a time. This constraint then makes it necessary to alter many other reproductive behaviors to make allowances for the fact that hatchlings might be of different age. Such phylogenetic constraints surely exist but they are difficult to identify. If birds evolved flight only once, then this gives evolutionary biologists on a single example to analyze, not a very large sample size. In addition, proof of the existence of such constraints relies on the comparative method. In most cases it is very difficult to experimentally manipulate the trait of interest. Such higher order processes that might limit or channel evolutionary change in certain directions are referred to as macroevolutionary processes.
We have already put the comparative method into practice when we ask questions concerning differences in male parental care among: sticklebacks, pipefish, and seahorses. We identified the changes in environment that may have been influential in promoting the evolution of male biased care. In the case of mating systems, any environmental factor that leads to a male versus a female biased operational sex ration will favor sexual selection on males versus females. What we lacked for a complete consideration of the comparative method was an idea of the history of evolutionary change in these groups. To understand the history of change, we have to develop a notion of phylogeny and cladistic relationships.
The fundamental concept underlying all phylogenetic analysis is the phylogenetic tree. The idea of a tree representing the evolutionary history of a group was so compelling to Darwin that he included a figure in his book the Origin of Species. Darwin was not one to include many pictures, as this is the only figure in the entire book. A phylogenetic tree traces genalogical relationships among species much like a family tree traces genalogical relationships among individuals. The formation of new lineages or species takes place by the Speciation Mechanisms that were described earlier.
Let us considered the topology and names for a few features on the tree to develop the working vocabulary of phylogenetic analysis. A branch point where it thought that two separate lineages arose is termed a node or the common ancestor of the members of the lineage. The group of species that are all descendants of the common ancestor are called a clade. Thus, the phylogenetic tree is also referred to as a cladogram in that it describes all ancestor-descendant relationships in a graphical form. The vertical axes for a cladogram reflects time.
Evolutionary biologists have many ways of assigning time to the vertical axis. The most intuitive component of time that we might imagine is real time as drawn from the fossil record. For example, the common ancestor of all homonid lineages is thought to be Australopithecus ramidus , an ape that appeared in the fossil record 4.6 million years ago. Australopithecus ramidus gave rise to the lineages of Australopithecines of which its most famous member, a skeleton called "Lucy", is thought to be a member of the species, Australopithecus afaraensis, that gave rise to all lineages of Homo as well as other branches of Australopithecus. In addition to fixing the vertical axis based on time, molecular biologists can calibrate the number of amino acid substitutions in a protein or the number of nucleotide base pairs in a gene with what is known as the molecular clock. If change in these molecules is constant over time and for all members of a clade, one could calibrate the molecular clock or rate of change in amino acid residues or base pairs using a single fossil that marked the divergence of two clades. Where the calibration is not available, information on genetic distance is used as a measure of "time".
The steps involved in any phylogenetic analysis are as follows:
The concept of an outgroup is crucial for interpreting the changes that might have occurred in a lineage. For example, an appropriate outgroup for hominid lineages would be the great apes. If you were to look at the behavior of the great apes (e.g., the genus Pan or chimpanzees) and found aspects of their behavior that is similar to modern representatives of the hominid lineages (just Homo sapiens), one might infer that such behaviors were also found in all extinct forms of the hominid lineages. The outgroup of chimpanzees presumably arose from the common ancestor in the remote past, and both humans and chimps share this common ancestor. Under Charles Darwin's paradigm of "descent with modification", one might assume that there was no modification in the transmission of behaviors. For example, consider the tool using abilities of humans. Jane Goodall has reported that chimpanzees use small twigs to get termites out of tree holes -- a clear indication of tool use. Is there evidence of tool use in fossils. Yes, good tools have been found associated with the fossils of Homo habilis and all later hominids. However, evidence for tool use in the Australopithecines is sketchy at best. But also notice that chimps use twigs, wooden tools, that are unlikely to fossilize, and even more unlikely for us to recognize them as tools even if we were to find them.
The hominid example serves to illustrate the limitations of any phylogenetic analysis. You cannot necessarily increase your sample size -- there is only one extant species of homo from which to draw information. However, there are more outgroups, that could be used to refine the information. For example, humans are thought to be most closely related to chimps. Gorillas are the species of great apes with which we share moderate relationships, and orangutans are furthest along. One could in prinicple date the divergence of gorillas, which do not use tools, from chimps and homo to perhaps date the origin of tool use in the family pongidae.
The phylogenetic argument implies that all Australopithecines used some kind of tools because tool use is shared by living members of clade (us) and our nearest outgroup -- the non-hominid chimpanzees. This is an argument based on inference, not direct observation of the ancestors of hominids. In many cases the condition of traits in the common ancestor are reconstructed from information provided by the outgroup. It is often assumed that the outgroup has more primitive traits than the clade of interest. Indeed, this feature of outgroup choice is often crucial to phylogenetic inference. This is because we are interested in two kinds of changes in a lineage. We are interested in clades that have:
A shared ancestral character is found in the outgroup, and in those members of the clade that have not experienced any modification of the trait from the state found in the common ancestor. This is because we infer that if the outgroup species shares the trait and it is found in some clades, the specis in these clades must have received the trait in an unmodified form from the common ancestor. An opposable thumb is found in all members of the pongidae, and it is generally thought that an opposable thumb is an important requirement for a tool using hominid (Note that other animals have evolved tool use, birds manipulate twigs with their beaks to obtain termites in much the same way as chimpanzees. Birds have found an alternative evolutionary pathway to tooluse.)
A shared derived character is ideally found in some subset of the clade and in nearly all members of that sub-clade. By inference we might hypothesize that those members of the clade with the shared derived character possess that character because the character arose once, in the common ancestor of the found at the node of the clade of interest. For example tool use is shared by chimps and homo, but it is not found in outgroups more removed from this sub-clade (gorillas or orangutans).
In the absence of fossil information (which is true for most species on the planet), how on earth do you make trees? You use the information from shared derived versus shared ancestral traits. The same principle applies to all kinds of information be it derived from molecules or morphology.
Willi Hennig is credited with coming up with a simple rule for reconstructing the evolutionary changes that have occured in a clade that has revolutionized the way comparative biology is carried. First let me contrast phylogenetic analysis before and after Hennig. In the dark ages, a professor would study a group, indeed they might even study all species in a group, and after such lengthy lifetimes work, they would draw a tree. The construction of the phylogenetic tree did not take place with any formal rules in mind. To learn how to draw such trees, students would enter into a lengthy apprenticeship of sorts and pick some smaller group of the large clade. In many cases the systematists would consider the concept of shared derived characters in the construction of their tree but no formal rule was used.
Willi Hennig formalized the use of shared derived characters by devising the principle of parsimony. The tree in which the fewest evolutionary steps are required to connect the different branches of a tree is considered the most parsimonious tree. In making such an assumption to connect branches of the tree, we assume that evolution is conservative and that evolutionary change does not occur all over the tree. Remember that any tree is our best guess as to the actual pattern of evolution underlying the phylogenetic relationships between species. A phylogenetic tree is a hypothesis of the pattern of evolution. Inferences is used in the construction of trees.
Let us examine a hypothetical example of how we would draw a the most parsimonious tree from the following traits and let us assume that ancestral is scored as 0 and advanced is scored as 1:
| trait 1 | trait 2 | trait 3 | trait 4 | |
| species A | 0 | 0 | 1 | 0 |
| species B | 1 | 1 | 1 | 1 |
| species C | 0 | 1 | 1 | 1 |
| outgroup | 0 | 0 | 0 | 0 |
The simplest way to construct the most parsimonious tree by "hand" is to identify pairs of species with the most derived set of characters as they have changed quite a bit, and to identify the species that have the most primitive set of characters. For example, species A only differs from the outgroup in a single trait, trait 3. In addition, the other two species also differ from the outgroup in trait 3, but they also differ in a number of other traits. Thus trait 3 distinguishes our clade of species A, B, and C from the outgroup and it also tells us that the branch from 1 to the outgroup should be closer to the ancestral node compared to the branches for species 2 and 3.
Now let us look at species 2 and 3. Species C has three derived traits, and species B has four derived traits. Again species C is closer to the node than species B, but it is farther from the node than species A. Voila, we have a tree based on the four traits and this tree has minimized the changes.
We can map the changes onto the tree with a "notch" and count that four changes are required. You can draw any other topology and you will require more changes than the four we see. Draw some other ancestor descendant relationship and test this out.
In practise, real data and real trees have conflicts between characters. One character suggests a different phylogeny than if we consider a different character. What if the distribution of traits among species was as follows:
| trait 1 | trait 2 | trait 3 | trait 4 | |
| species A | 0 | 0 | 1 | 0 |
| species B | 1 | 1 | 0 | 1 |
| species C | 0 | 1 | 1 | 1 |
| outgroup | 0 | 0 | 0 | 0 |
I have highlighted the single change in bold. Trait 3 and trait 1 provide a different set of trees. Now we have three primitive charcters for both species B and C and we cannot definitively place the node for these two species on the tree. There are two "most parsimonious trees", each of which requires 5 evolutionary steps. These two trees are, however, better than all other possible trees.
Such problems typically arise from either:
convergence in which the same character arises more than once on the tree (an abhorent thought to a strict adherent of the principle of parsimony)
or perhaps from evolutionary reversals where a derived character states reverts back to a more primitive state.
There are kinds of parsimony which allow for reversals in evolution so this is not much of a problem. However, multiple evolution of a characters on various branches of the tree is difficult for for the principle of parsimony. Consequently, the characters that people tend to choose are ones that minimize evolutionary reversals and the frequency of multiple evolution of characters which we refer to as homoplasy.
While, multiple evolution of character states or homoplasy, is bad from the point of view of constructing trees, it is a good thing from the point of view of the analysis of adaptations. If a trait evolves a number of times on a tree we have a much larger sample size to use in our tests of the conditions that drive the evolution of a trait. The more independent events that we observe the more data we have on the conditions that might favor the evolution of a behavioral trait.
This rule is simple to put into practise for a student of behavior. Do not use behavioral traits to make the tree. Do use behavioral traits as the focus of the study of adaptations.
Let's begin to put our ideas into operation with a not-so-hypothetical example taken from paleontology.
Having now studied some simple "+, 0" kinds of phylogenetic trees, let's look at a more interesting example involving dinosaurs. On page 478 in the Campbell textbook you will find a cladogram a la Jurassic Park. Take a look at the various characters which have evolved in each kind of dinosaur. Notice particularly that some of the traits are found in all dinosaurs on the tree (the presence of a trait is denoted by a + sign just as in our previous examples) while other traits are present in some dinosaurs but absent in others.
How can we use the presence or absence of a trait to infer the evolutionary processes that led to the dinosaur lineage? Just as before, we begin by using traits common to all individuals, that is, shared derived characters or synapomorphies. We can place the first character (hip socket with a hole) at the very base of the tree since all the dinosaurs in the example share the trait. Now we look to the next most commonly shared character and notice a split in the evolutionary trajectory. This split leads to one group of three individuals sharing a common trait and another group of four individuals sharing a different common trait. We continue through the list of characters keeping in mind the principle of parsimony, and group dinosaurs according to shared characters.
Now that we've finished mapping our characters onto the phylogentic tree
we can look at the evolutionary relationships among the various dinosaurs
we are studying. But BEWARE, this tree is only our best estimate of evolutionary
history based on parsimony. Since we were not around to witness dinosaur
evolution, this tree is only our best approximation of how the world works.
The vast majority of fish have no parental care, and simply squirt their gametes into the water leaving zygotes to develop on their own. This reproductive mode was present in one of the first organisms that we would have called a fish. We refer to this original condition as the ancestral state (Brooks and McLennan 1991). Derived states would be modifications of the ancestral condition that entail the evolution of new adaptations. For example, sticklebacks, pipefish, and seahorses belong to the same order of fish, Gasterosteiformes, and are noteworthy in the animal kingdom for evolving highly-advanced system of male care. Care from a male parent is derived relative to the ancestral mode of 'broadcast spawning' found in most fish.
The care in sticklebacks is limited to nest defense, a relatively common occurrence in the animal kingdom (e.g., many families of fish, birds, mammals, insects, etc. have males that guard the nest). Some species of male pipefish, Nerophis ophiodon, develop a brood patch to which eggs are glued, and the male carries the eggs until they hatch. Other species of male pipefish, Sygnathus typhle, have developed an elaborate brood pouch during the reproductive season, into which the female oviposits her eggs. The pouch splits open when the eggs hatch, releasing the newly developed fry. Males seahorses have evolved elaborate vascularization in the pouch where nutrients are transferred to the eggs much like a female mammal transfers micronutrients across the placenta. The sticklebacks, pipefish and seahorses have family tree of sorts, a phylogenetic tree, that describes the order in which each species split off during evolution (Brooks and McLennan 1991). Sticklebacks split off earlier and also have the least derived level of male care in which a nest is defended. Pipefish like Nerophis ophiodon that possess a brood patch split off later, while pipefish with a more derived brood pouch, Syngnathus typle, split off later. Seahorses with a brood pouch that completely encases the embryos, the most derived state, split off last. The phylogenetic history for the Gasterosteiformes shows a clear evolutionary progression for more derived traits that enhance male care.
Figure 10.2. Phylogeny for fish in the order Gasterosteiformes illustrating the more derived care that has evolved in more recently evolved pipefish and sticklebacks. Male care in a nest is a derived condition relative to the more ancestral condition of all fish that broadcast spawn gametes (not shown). A simple brood patch is found in pipefish Nerophis ophiodon, brood pouch with folds in pipefish Syngnathus typhle, and a completely enclosed pouch in the seahorse, Hippocampus whitei. (stickleback from (Drickamer and Vessey 1986), other fish drawings from Vincent et al 1992).
We considered an experimental test of the predators propensity to learn whether prey were aposematically colored within a very short time frame, and whether being solitary or gregarious gave the evolution of Aposematic coloration a selective boost. Recall Ronald Fisher's original arguments:
Ronald Fisher observed that many aposematic forms tend to also be quite gregarious and congregate in the same locale. Fisher speculated that kin selection may favor such aggregations. An individual may die during the lesson required to teach a naive predator that the color also results in a bad experience. However, because the predator leaves the remaining kin alone, the inclusive fitness of the dead aposematling is positive because the cost of individual death is balanced by the surviving kin that are left alone. Gregariousness can easily result from kin groups (e.g., a localized clutch), and such kin groups greatly enhance the probability that aposematic coloration will spread even though brightly colored individuals attract attentions of naive predators.
Sillen-Tullberg studied the evolution of aposematic coloration by mapping both the morphological trait (bright color as a proxy for unpalitability) and the behavioral trait (gregariousness) onto trees. She was more interested in whether the origin of gregariousness was contigent upon unpalatability and aposematic coloration evolving first. The evolution scenario for her arguments go as follows:
Sillen-Tullberg tested these ideas out on several clades of caterpillars. Unfortunately unpalatability is very difficult to score (you have to make a lot of birds barf) -- so she used aposematic coloration as a conservative index of unpalatability. She found that unpalatability did preceed gregariousness every time. Gregariousness is a behavior of the female -- she decides to lay one or many eggs on a plant.
Theories of sensory bias postulate that the evolution of a sexually selected male character arises in a group in which females have a pre-existing phylogenetic bias for certain kinds of signals, and those signals are the ones that males evolve. The basic phylogenetic distribution for the female preference and male trait is as follows:
The principle of parsimony is crucial because the argument above assumes that evolution occurred in the smallest number of changes.
 |
Hypothetical relationship between a male and the phylogeny. Notice that the male trait does not occur in the outgroup. The mapping of a single evolution for the male trait assumes that it only evolved once -- it could have evolved twice, once in each lineage. |
| Hypothetical relationship between a male trait and female preference that would suggest that preference evolved prior to the male trait. (Note, a non-parsimonious reconstruction could have preference evolve three times! | |
| Hypothetical relationship between a male trait and female prefernce in which the hypothesis of correlated evolution of the trait and preference cannot necessarily be ascribed to a pre-existing bias. |
Basalo looked a genus of Sword-tailed fish, Xiphoporus, which have elongate swords to investigate the role of sensory bias in channeling sexual selection. A phylogeny of Xiphophorus indicates that most recent members of the "clade" have swords. One member of the genus, the most "ancestral" type lacks a sword. Females mate with males without swords.
Basalo asked whether females from this ancestral species preferred males of their own species which lack a sword, or males of their own species with swords tied on. The overwhelming choice was for males that had a Sword!!! She interpreted these results to imply that there existed an "ancestral" bias, for swordedness in these fish, that in turn led to a Runaway Process.
Recall the conditions for the evolution of hostplant toxicity and the sympatric speciation of their herbivores. This example serves us with a classic case of coevolution. Coevolution in insects and plants relates to the "endless evolutionary arms race" which leads to (Erlich and Raven, 1964 cited in Farrell and Mitter 1994):
I have cartooned the idealized clades for hostplant and herbivore below.
Note that there is a perfect correspondence between clades in the ideal
case. Nature presents us with some near perfect examples.
I will discuss Farrel's data on the coevolution between beetles and their milkweed hosts. The beetles have a tight association with their hosts which is required for true coevolution and eggs in that larvae and adults require the milkweed.
We can contrast the beetle associations with butterflies, that only feed on nectar as adults. The offspring of butterflies are raised on milkweed, but because adults are not necessarily dependent on the milkweed for food, the coevolution is not seen in the butterfly/milkweed phylogenies.
Farrell, B. D. and C. Mitter., 1994. Adaptive radiation in insects and plants: time and opportunity. Amer. Zool. 34:57-69.