On the dependence of speciation rates on species abundance and characteristic population size ANASTASSIA M MAKARIEVA* and VICTOR G GORSHKOV Theoretical Physics Division, Petersburg Nuclear Physics Institute, 188300, Gatchina, St.-Petersburg, Russia Corresponding author (Fax, +7-812-7131963; Email, [email protected]) The question of the potential importance for speciation of large/small population sizes remains open. We com- pare speciation rates in twelve major taxonomic groups that differ by twenty orders of magnitude in characteristic species abundance (global population number). It is observed that the twenty orders of magnitude’s difference in species abundances scales to less than two orders of magnitude’s difference in speciation rates. As far as species abundance largely determines the rate of generation of intraspecific endogenous genetic variation, the result obtained suggests that the latter rate is not a limiting factor for speciation. Furthermore, the observed approximate constancy of speciation rates in different taxa cannot be accounted for by assuming a neutral or nearly neutral molecular clock in subdivided populations. Neutral fixation is only relevant in sufficiently small populations with 4Nen < 1, which appears an unrealistic condition for many taxa of the smaller organisms. Further research is clearly needed to reveal the mechanisms that could equate the evolutionary pace in taxa with dramatically dif- ferent population sizes. [Makarieva A M and Gorshkov V G 2004 On the dependence of speciation rates on species abundance and characteristic population size; J. Biosci. 29 119–128] 1. Introduction On the other hand, the majority of studies in conserva- tion biology are currently concentrated on the problem of Current views on the relationship of speciation rate on maintaining sufficient intraspecific genetic variation, population size remain to be largely shaped by modelling which is thought to be vital for species survival and fur- studies. Shifting balance models envisage speciation ther (adaptive) evolution (Cohn 1986; O’Brien 1994). Here events as shifts between high-fitness adaptive peaks (spe- population size together with generation time emerge as cies) separated by low-fitness valleys. This implies that important factors governing the rate of generation of gene- only if the population size is sufficiently small, the geno- tic variation in a species. It can be thought that species typic composition of the population is able to drift ran- composed of a large number of individuals with short domly through the valley of low fitness with a considerable generation time should evolve more quickly, because they probability (Wright 1931; Mayr 1954; Barton and Charles- will sooner generate beneficial mutations needed for fit- worth 1984). Other modelling studies based on the as- ting into a new environment. The experimental studies sumption of existence of ‘ridges’ of highly fit genotypes aimed at revealing in vitro evolution in abundant, rapidly (Dobzhansky 1937; Gavrilets et al 1998) that connect reproducing organisms (e.g. Papadopoulos et al 1999) are neighbouring adaptive peaks, suggest either independence apparently theoretically grounded in such an approach. of speciation rates from total population size (Orr and Orr So far, no large-scale analysis of the available empiri- 1996) or higher speciation rates in large populations with cal data on speciation rates versus characteristic popula- large geographic ranges (Gavrilets et al 1998). tion sizes of species has been ever performed, despite the Keywords. Extinction; fixation; molecular clock; neutral theory; population size; species duration J. Biosci. | Vol. 29 | No. 1 | March 2004 | 119–128 | © Indian Academy of Sciences 119 120 Anastassia M Makarieva and Victor G Gorshkov potential importance of such an analysis for the speci- Values of S and E remain close to each other when ation theory. To a large degree, this may reflect difficul- considered for particular taxonomic groups as well, even ties in obtaining estimates of characteristic population when the calculations are made for the initial period of sizes of different species (Nei and Graur 1984). In this the group’s diversification. During this time the increase paper we compare rates of speciation between organisms of the number of species within the group is most rapid that differ by up to twenty orders of magnitude in species and the difference between ts and td values is the largest abundances (i.e. global species population numbers). It possible. Based on the data of Stanley (1979), Sepkoski proves that the twenty orders of magnitude’s difference (1998) plotted the per-species rate of diversification, S – E, in species abundances scales to less than two orders of versus the extinction rate, E, for twelve major groups of magnitude’s difference in speciation rates. We further taxa. For nine of these groups, the ratio (S – E)/E is less show that the observed pattern cannot be accounted for than unity, implying a less than two-fold difference bet- by the neutral molecular clock approach. ween S and E. The largest ratio, (S – E)/E ~ 4, is obser- ved for planktic foraminifers. It means that the speciation rate in these organisms is five times that of extinction 2. Methods rate, and that ts is five times less than td. These figures show that even in the extreme cases of most rapid diversi- 2.1 Estimates of speciation rates fication, the orders of magnitude of ts and td still coincide. This makes species duration, t , a reasonable estimate of t According to the fossil record, morphologically distinct d s in all cases. groups of organisms (morphospecies) succeed each other It has been suggested that the increase in biodiversity to dominate ecological space. We are interested in the can sometimes be better fitted by logistic rather than by mean time, t , between two successive speciation events s exponential curves (Courtillot and Gaudemer 1996). The in a given evolutionary lineage. If the number of species proposed logistic models account for the periods of rela- remains constant or changes only slowly, this time can be tive stasis in the number of species. Such periods are in- estimated by the mean time of species duration, t (Sep- d terrupted by mass extinctions, which, in their turn, are koski 1998), i.e. the time elapsed between the origin and followed by a rapid regain in the species number. When extinction of a species. the number of species remains constant, species duration, Life has existed on Earth for about T ~ 4×109 years t , is by definition equal to the reciprocal of speciation (Hayes 1996). Increase of species numbers within sepa- d rate, i.e. to t . On the other hand, the direct estimates of rate taxa as well as the global increase in biodiversity is s the (S – E)/E ratio made for the periods of active growth sufficiently well described by exponential curves (Bush of biodiversity, see above, show that t and t remain et al 1977; Benton 1995). One can thus write d s close to each other during these periods as well. This shows n = e(S – E)T, (1) that, on an average, one species during its lifespan gives where n ~ 107 is the current number of species in the bio- rise to no more than a few species, even when the species number grows most rapidly. This being the case, the con- sphere (May and Nee 1995); S º 1/ts is the mean speci- clusion that td and ts are close to each other, is independent ation rate; E º 1/td is the mean extinction rate. Using these values of n and T and taking the average species of the model of biodiversity growth applied (exponential, 6 logistic or other). duration to be about td ~ 4×10 years (Raup 1991a), we obtain from eq. (1): Species durations in marine diatoms, planktic and ben- thic foraminifers, higher plants and vertebrates [marked t with (L) in table 1] were estimated by Stanley (1985) and (S - E) / E = d ln n ~ 0× 016. (2) T Bush et al (1977) using the Lyellian approach. At any given time point in the past, a Lyellian curve depicts the This agrees with the result obtained by Raup (1991b) percentage of species within the considered fossil biota who noted that more than 99% of species that have ever that survives to the present. Mean species duration is esti- lived are now extinct. One can see from eq. (2) that, on mated from the absolute value of the slope of the curve at average, the time between two successive speciation zero. The number of species analysed in such studies never events ts º 1/S can be estimated by species duration, falls significantly below one hundred (Stanley 1979). td º 1/E, to the accuracy of less than 2%. Species durations in dinoflagellates and diatoms [mar- This conclusion is very robust. Even if one assumes ked with (V) in table 1] were calculated by us from the that life originated during the Phanerozoic and use in eq. slopes of survivorship curves (plots of the proportion of 8 (2) TPh = 6×10 years instead of the whole time of life the original sample that survives for various intervals) existence T, one still has (S – E)/E ~ 0×11, that is, about constructed by Van Valen (1973). For most taxa survi- 10% difference between ts and td. vorship curves are approximately log linear (Van Valen J. Biosci. | Vol. 29 | No. 1 | March 2004 Speciation rate and population size 121 1973; Stanley 1985). Mean species duration can be esti- and genetic characteristics of bacteria, that close but dif- mated from the constant linear slope of the curve. Num- ferent bacterial species display less than 97% of sequence ber of species analysed ranged from several hundreds in identity (Stackebrandt and Goebel 1994). To obtain an dinoflagellates to several thousands in diatoms.