COMMENTARY

Measuring the evolution of body size in

P. David Polly1 Departments of Geological Sciences, Biology, and Anthropology, Indiana University, Bloomington, IN 47405

n Tempo and Mode of Evolution,an A B early classic of the evolutionary syn- I thesis, George Simpson argued that, although many aspects of evolution can be studied in living populations, rates Time and patterns of evolutionary change are Time best documented with the fossil record (1). The fact that fossils can be placed into phylogenetic and stratigraphic frameworks makes it possible to calculate rates of change directly rather than extrapolating them from the inferred history of living Body Mass Body Mass . Although Simpson’s logic is sound in principle, the task of measuring Fig. 1. The relationship of lineage evolution to CMR. (A) A clade consists of many evolving and bi- furcating lineages, only one of which is the largest member of the clade at any one time. (B) CMR is evolutionary rates from fossils is laborious calculated from only the largest members, which are represented by the largest points on the broken in practice: Fossils must be taxonomically black line in each time interval. assigned to and then assembled into a phylogenetic by using methods that distinguish true ancestors from sister lineage, but of how fast a clade pushes its thus better known than their most groups; the sediments in which fossils are maximum size boundary (Fig. 1). A clade medium-sized relatives. found must be correlated lithologically or consists of many evolving lineages, only Evans et al. (3) found that, in terrestrial biostratigraphically and then packages of one of which is the largest at a given time; mammals, it takes approximately 1.6 mil- rocks must be correlated into a global furthermore, the same lineage is not likely lion generations for a 100-fold increase in timescale or otherwise given absolute ages; to be the largest at different times. CMR size, approximately 5.1 million generations and finally, the numbers of generations is thus calculated from members of several for a 1,000-fold increase and approxi- represented by the intervals of time sepa- lineages, each of which happens to be mately 10 million generations for a 5,000- rating the fossils must be estimated (2). biggest at a particular time; it is therefore fold increase. Rates of maximum size Because of the herculean task of gathering a measure of how fast a clade can reach evolution in , the largest of mam- these basic data, most paleontological record size, regardless of how many line- mals, were approximately twice as fast. fi studies of evolutionary rates are narrow in ages are competing for that record. Even The latter nding was expected because scope, confined to a few lineages or clades, though CMR does not fit the standard the remarkably rapid increase in size in too piecemeal to say much about the fi early whales has long been noted (5), but de nition of an evolutionary rate, it is an fi broad-scale patterns that are normally the intuitively useful metric, because it is often less expected was the nding that maxi- mum body size decreases occur, on aver- strength of the paleontological record. the largest organisms—, whales, age, much faster than increases: 10 times In PNAS, Evans et al. (3) develop sequoias, Titanoboas—whose rate of evo- faster, in fact. Evans et al. (3) discuss a “shortcut” that allows evolutionary rates lution is of special interest. CMR radically possible biological reasons for this asym- to be compared across all mammals simplifies the task of calculating rates, over the entire Cenozoic Era (last 65 My), metry, including the possibility that short- because phylogeny within each clade does ening the period of growth before sexual giving a very broad picture indeed. They not have to be known. Given that the fi maturity (i.e., pedomorphosis) is an evo- nd some unexpected things, including an constituency of most mammalian orders is evolutionary bias in which body mass ap- lutionary easy thing to do, whereas obvious but their internal phylogenetic lengthening the period of growth comes pears to decrease faster than it increases. fi details are not, the simpli cation allows with physiological and reproductive costs. Importantly, their paper puts numbers broad patterns to emerge from numerous on the amount of evolutionary time re- The faster rates of maximum size decrease and complicated data. Alternative strate- quired for body mass changes of different may be related to the commonness of gies for dealing with the complexity of the magnitudes. island dwarf species such as foxes, red deer, fossil record, such as sidestepping it by The breadth of Evans et al.’s analysis (3) mammoths, and hippos, compared with was achieved first by focusing exclusively calculating rates based on phylogenies of island giants such as the Orkney vole and – on body size, a trait common to all mam- living mammals (4), are innovative but speckled rattlesnakes (6 9). Interestingly, mals (indeed, to all organisms), and less satisfying, because they substitute an- the rates of maximum size evolution found second by focusing not on the rates of cestral reconstructions for the empirical by Evans et al. (3) are slower than those evolution of individual lineages, but on temporal data provided by the fossil re- predicted from studies of individual line- changes in the maximum body mass of cord. To call CMR a shortcut is doing it ages, in which per-generation rates of larger clades. The result is a rather ab- a disservice because the amount of data evolution suggest that the 1,000-fold in- stract measure called the clade maximum assembled by Evans et al. (3) is massive; crease in body mass between a small cat rate (CMR), the rate by which the maxi- their exercise in data reduction avoids a mum body size in a clade increases or paralyzing morass of taxonomic and tem- decreases over time. CMR appears to be poral uncertainties and allows them to Author contributions: P.D.P. wrote the paper. one step removed from a true evolutionary focus on well studied points. After all, the The author declares no conflict of interest. rate because it is a measure not of the rate biggest prehistoric have received See companion article 10.1073/pnas.1120774109. of change within an ancestor-descendant considerable scientific attention and are 1E-mail: [email protected].

www.pnas.org/cgi/doi/10.1073/pnas.1201030109 PNAS Early Edition | 1of2 Downloaded by guest on September 29, 2021 and an elephant could, in principle, hap- elapsed since the last common ancestor whales, but that slower rates were associ- pen in fewer than 1 million generations (2, of the two lineages. As Evans et al. (3) ated with restricted environmental varia- 10). The apparent disjunction between point out, this phenomenon means that tion within adaptive zones, preventing body predictions based on rates calculated in lineage-specific rates within the clade are size from diverging as rapidly as it did with microevolutionary studies and the empiri- the initial colonizations of the adaptive cal evidence presented by Evans et al. (3) Maximum body size zone. Despite the different perspectives of raises questions both mathematical these two studies, their findings are similar and biological. decreases occur, on in that despite rapid changes over short One possible explanation for why CMR periods, the overall rate of divergence in shows slower rates of evolution than pre- average, much faster mammalian body size appears to be con- dicted from within-lineage studies is the strained over long time periods, even when fact that it is an unusual kind of rate, and than increases: 10 times the scaling bias in rate measurement has so is not comparable to ordinary rates of been taken into account. The explanations evolution. Although it is true that CMR is faster, in fact. for this phenomenon offered by the two not a typical rate of evolution, it is statis- studies are also broadly compatible in that tically related to typical rates: If a trait they both view selection as moving more evolves in a Brownian motion pattern, as expected to be higher than CMR. Still, the often in one direction than another, but are it would if the intensity and direction of logic presented in the previous paragraph focused on different levels of explanation. selection was randomly distributed across suggests that a bias related to time interval These new empirical studies invite time and lineages, then the SD across cannot easily explain the comparatively renewed investigation of the relative roles lineages at a given time will increase with slow macroevolutionary rates observed by of selection and constraint in evolution, the square root of the number of gen- Evans et al. because CMR should still and of the sources of difference between erations elapsed (11). As a trait’s maxi- scale directly with within-lineage rates. broad macroevolutionary patterns and mum is related to its SD—both are metrics Having ruled out these possible sources smaller microevolutionary or “mesoevolu- of range—CMR is expected to be a func- of bias, the relatively slow rates of CMR tionary” scales. Renewed attention should tion of the average rate of body size evo- are suggested by Evans et al. (3) to result also be given to the way rates of evolution lution in the clade’s constituent lineages from factors that favor body size decreases are measured and to the biological and should be comparable to them. over increases. They do not attempt to theory of how microevolutionary patterns Another possible reason for slow mac- assess these factors directly, but point are expected to scale into macroevolution- roevolutionary rates is that CMR is biased to their finding that decreases in maximum ary ones, if at all. Hansen and colleagues because it is measured over what could body size are 10 times faster on average recently pointed out that heritability, be interpreted as the wrong intervals. A than increases. They suggest that factors a variance-scaled parameter, is unlikely to rate is the amount of change divided by the like pedomorphosis, the physiological dif- be a good predictor of long-term evolution interval over which the change occurred. ficulties associated with large body size, because population variance is highly Stochastic reversals in trait evolution re- and a bias toward selection favoring small context sensitive (15). Most evolutionary sult in a bias because the reversals in- body size and fast reproduction are candi- rates, including CMR, are measured as creasingly cancel out change over longer date explanations. Interestingly, Uyeda and the change in the trait as a proportion of periods of time: rates measured over colleagues recently analyzed a similar data the within-species variance in the trait, longer intervals appear to be lower than set of body masses and found that short, making them an example of variance- those measured over short intervals (12, rapid bursts of evolution occurred infre- scaled parameters. The apparent mis- 13). CMR is measured over deceptively quently within lineages, but body size evo- matches between rates on large and small long intervals, longer than the time bin in- lution appeared to be constrained over scales may deserve a fresh look in light of tervals used in its calculation, because the longer intervals (14). Uyeda and colleagues Hansen and colleagues’ observations. true evolutionary interval between clade argued that the short bursts were caused CMR provides yet another way of looking maxima is not the interval of time between by lineages moving from one adaptive zone at the evolution of traits, providing even the bins, but the total phylogenetic time to another, such as with the origin of more incentive for renewed attention.

1. Simpson GG (1944) Tempo and Mode in Evolution (Co- 7. Haynes S, Jaarola M, Searle JB (2003) Phylogeography 11. Felsenstein J (1981) Evolutionary from gene fre- lumbia Univ Press, New York). of the common vole (Microtus arvalis) with particular quencies and quantitative characters: finding maxi- 2. Polly PD (2001) Paleontology and the comparative emphasis on the colonization of the Orkney archipel- mum likelihood estimates. Evolution 35:1229–1242. method: Ancestral node reconstructions versus ob- ago. Mol Ecol 12:951–956. 12. Gingerich PD (1993) Quantification and comparison of served node values. Am Nat 157:596–609. 8. Weston EM, Lister AM (2009) Insular dwarfism in hip- evolutionary rates. Am J Sci 293A:453–478. 3. Evans AR, et al. (2012) The maximum rate of evo- pos and a model for brain size reduction in Homo flor- 13. Gingerich PD (2009) Rates of evolution. Annu Rev Ecol lution. Proc Natl Acad Sci USA, 10.1073/pnas.1120774109. esiensis. Nature 459:85–88. Evol Syst 40:657–675. 4. Cooper N, Purvis A (2010) Body size evolution in mammals: 9. Meik JM, Lawing AM, Pires-daSilva A (2010) Body size 14. Uyeda JC, Hansen TF, Arnold SJ, Pienaar J (2011) The Complexity in tempo and mode. Am Nat 175:727–738. million-year wait for macroevolutionary bursts. Proc 5. Uhen MD (2010) The origin(s) of whales. Annu Rev evolution in insular speckled rattlesnakes (Viperidae: – Earth Planet Sci 38:189–219. Crotalus mitchellii). PLoS ONE 5:e9524. Natl Acad Sci USA 108:15908 15913. 6. Lister AM (1989) Rapid dwarfing of red deer on Jersey 10. Gingerich PD (2001) Rates of evolution on the time scale 15. Hansen TF, Pélabon C, Houle D (2011) Heritability is not in the last interglacial. Nature 342:539–542. of the evolutionary process. Genetica 112-113:127–144. evolvability. Evol Biol 38:258–277.

2of2 | www.pnas.org/cgi/doi/10.1073/pnas.1201030109 Polly Downloaded by guest on September 29, 2021