1.08.2.1 Variation Between Classes
Size and structure of the central part of the nervous system differ clearly between the major animal taxa. From the nerve ring of nematodes to the cerebral ganglion of insects to the brains of cephalopods, birds and mammals, there are major discontinuities in the size and organization of the central organs. If we were to statistically partition the variance in nervous system characteristics over all animals, we would likely find that most of it lies at very high taxonomic levels. Neurons themselves are highly conserved in all animals, as are synaptic processes. For example, memory appears to be based on similar rules of long-term potentiation via glutaminergic synapses from Aplysia (Bailey et al., 2000; Pittinger and Kandel, 2003) and cephalopods (Hochner et al., 2003) to humans. The basic building blocks of learning and cognition might thus show a strong constancy throughout evolution.
The way these neurons are organized in the brain and how many of them are available for more complex information processing also appear to be conserved within major taxa. A mammalian brain is, on average, larger than an avian brain and features a laminar cortex, while avian brains are organized in discrete nuclei (Karten, 1997). In turn, bird brains are larger than cephalopod brains, which are organized in supra- and subesophagal lobes. These average differences have sometimes led to a scala naturae vision of brain evolution, where more recent lineages are seen as more ence-phalized on average than older ones. At least two hypothetical mechanisms could produce such a trend. The first is the possibility of an evolutionary arms race. The oldest animals on earth had no nervous system at all. These were followed by animals with neurons that are linked by a central chain, then by animals with neurons linked to a central organ. Odontocetes and apes have the largest brains and are relatively recent (23 and 34 My respectively; Marino, 2002; Marino et al., 2004). The assumption here is that the bigger the brain, the more information it can store, the faster it can change behavior in response to environmental contingencies (Sol, 2003), and the more complex a behavioral repertoire it can program (Changizi, 2003). Animals that can do more of all this are assumed to have an advantage over those that can do less, once the costs of an enlarged brain are taken into account. A further positive feedback effect of the behavioral flexibility associated with larger brains may add extra pressures for encephalization via social channels. The more flexible the behavior, the trickier it is for other animals to predict and the more useful is a large brain to make such predictions and change behavior quickly in response to the rapid change of others (Byrne and Whiten, 1988). The consequence of these mechanisms is that a competitive arms race might then follow, as it does for sexual selection, limited by the costs of increasing the size of the organ.
The second evolutionary phenomenon that would lead to encephalization over time is behavioral drive (Wyles et al., 1983; Wilson, 1985). All other things being equal, animals that come into more frequent contact with environmental conditions likely to provide a selective context for randomly occurring mutations should be characterized by faster evolutionary rates. Opportunism, generalism, and invasiveness are three traits that will increase the rate of contact with new selective pressures. If these traits are associated with larger brains (see below), then encephalization should also correlate positively with rate of evolution. Because general-ism and invasiveness also make animals more likely to range farther, they should also increase the probability of allopatric speciation. If larger-brained taxa beget more descendant species than smaller brained ones, then the average size of the brain should increase over time. The fact that individuals from larger-brained species tend to have fewer descendants per unit time than those from smaller-brained taxa will, to a certain degree, counteract the positive effects of an arms race and behavioral drive.
Discontinuities in encephalization over major clades prompted comparative psychologists in the 1960s to ask whether certain learning differences paralleled neuroanatomical trends. Overall, the results of these programs tended to show that there are quantitative differences in learning performance in the direction predicted by encephalization differences. Some researchers however, have questioned the heuristic value of these findings. One comparative learning researcher, Riddell (1979, p. 95), ironically summarized his experience:
The comparative psychologist often appears to know little more than a grade school child who would rather have a pet dog than bird, or bird than fish, or fish than worm, simply because they make better friends, as they can be taught more.
Beyond these problems, there are two other limits to comparisons between classes: small sample sizes and the overlap between the encephalization distributions of the taxa. Because each class is n = 1, comparative statistics cannot be used to test predictions about the costs, benefits, evolutionary history, ecological associations, and behavioral correlates of encephalization. Comparing the average fish to the average bird to the average mammal has a sample size of 3. Some birds are as or more encephalized than some mammals. If a crow has a larger brain and more complex cognition than does a tenrec, is it useful to think of an average mammal versus an average bird? Intraclass variance might be biologically as important as interclass variance and the question of whether similar patterns govern intra-class or order variation in different taxa might be the more useful one to ask.
The study of intraclass variation solves the statistical problem mentioned above (classes Aves vs. Mammalia: n = 2; variation between avian species: n = 10000). It also increases the validity of cognitive comparisons by measuring animals with more similar sensorimotor worlds. Comparing the results of several within-class or within-order analyses might thus be a good way of finding general patterns in encephalization.
The taxonomic distribution of a trait as well as its co-occurrence with other traits can be due to two types of processes: ancestral descent and repeated independent evolution. Ancestral descent may represent simple inertia or it may be the source of an important adaptive radiation. To separate ancestral descent and repeated independent evolution, we must know something about the phylogeny of the taxon and control for its effects on the distribution of apparently co-evolved traits. Most phylogenies today are based on differences in molecular sequences of either nucleic or mitochondrial DNA. When a well-resolved (a complete, well-differentiated tree at all levels) and robust (different parts of the genome lead to similar phyletic conclusions) molecular phylogeny is not available for a given taxon, a classical taxonomy, based either on Linnean characters or cladistics is still useful, but some degree of resolution will be sacrificed, usually yielding nonbranching elements and/or equal branch lengths (i.e., with no known genetic distance or estimated time of divergence).
Evolutionary biologists have long been concerned that interpreting correlations between traits in extant species as adaptive consequences of co-evolution might be biased by two sources of type 1 error (Felsenstein, 1985; Harvey and Pagel, 1991). First, two species might show similar values on two traits because they are closely related, not because of independent evolutionary events. This violates the assumption of data point independence for correlations and inflates the sample size via pseudoreplication. The similar values might thus be the result of inertia from an ancestral state, and cannot be considered the result of adaptive co-evolution. Techniques such as independent contrasts have since been routinely applied to trait correlations to deal with such problems. Contrasts are nodal differences between estimated ancestral values of the traits we are interested in. The nodes represent hypothetical ancestors, whose values are assumed to be averages of the trait values for the two branches descending from the node, often weighted for genetic distance. While the trait values of a given pair of taxa may not be independent, the difference between them can be assumed to represent independent evolution. Imagine that we have data for relative brain size and for diet breadth on 100 species and that the independent contrasts yield a nonsignificant correlation between the two traits. Comparative biologists will usually conclude that the null hypothesis for adaptive co-evolution has not been rejected. Imagine now that you examine the taxonomic distribution of your two traits and find that of the two subgroups in your taxon, one contains 85 large-brained species whose diet varies from three to ten food types, and a second one contains 15 small-brained species whose diet varies from one to four food types. What can we conclude?
That there has been no repeated co-evolution of generalist diets and large brains in this order? In all likelihood, yes. That diet and brain size cannot be proven to have had a selective effect on each other in this group? Yes. However, if we run a normal regression (nonphyletically corrected) on the 100 extant species and find a highly significant brain-diet correlation, we would run the risk of type 2 error if the nonsignificant contrast analysis leads us to conclude that the observed pattern contains nothing of evolutionary interest. Clearly, species that combine large brains and a generalist diet are or have been in the recent past quite successful. How they got the combination of the two traits is what a nonsignificant regression on the independent contrasts tells you: they inherited it from their ancestors and the combination did not appear through repeated independent evolution.
Despite its difficulties, the phylogenetic approach has two advantages: the occurrence of repeated independent events is a much more stringent test for adaptative co-evolution than is a single ancestral event. Phyletic trees, combined with molecular clocks, can also generate hypotheses on evolutionary sequences and timescales. If the large-brained generalist combination occurs in six widely separated clades and there is more brain size variance at older phyletic levels, (corresponding, say, to 100 My BP) and more diet breadth variance at more recent levels (say, 20MyBP), then we can hypothesize that large brains in general allow the evolution of broader diets, because variation in brain size precedes variation in diet. For example, most of the variance in avian brain size is at high phyletic levels like the parvorder. The molecular data of Hedges et al. (1996) suggests that divergence of extant birds at this level is 100-125 My old and may coincide with episodes of continental splitting. In contrast, the variance in avian innovative feeding (see below) is highest at much more recent levels of divergence, e.g., the species. The hypothesis that brain size divergence preceded feeding divergence thus follows and can be tested with statistical techniques such as path analysis.
Was this article helpful?