Effective population size (Ne) determines the rate of genetic drift and the relative influence of selection over random genetic changes. While free-living protist populations characteristically consist of huge numbers of cells (N), the absence of any estimates of contemporary Ne raises the question whether protist effective population sizes are comparably large. Using microsatellite genotype data of strains derived from revived cysts of the marine dinoflagellate Pentapharsodinium dalei from sections of a sediment record that spanned some 100 years, we present the first estimates of contemporary Ne for a local population in a free-living protist. The estimates of Ne are relatively small, of the order of a few 100 individuals, and thus are similar in magnitude to values of Ne reported for multicellular animals: the implications are that Ne of P. dalei is of many orders of magnitude lower than the number of cells present (Ne/N ∼ 10−12) and that stochastic genetic processes may be more prevalent in protist populations than previously anticipated.
A crucial concept in evolutionary biology is that only some individuals obtain genetic representation in subsequent generations. This number of ‘successful’ individuals may be approximated by the contemporary effective population size (Ne), which, strictly defined, is the size of an ‘ideal’ genetic population experiencing the same amount of random genetic changes as the target population [1,2]. Our use of contemporary Ne refers to a local population and therefore is different to the ‘long-term Ne’ that represents an expected (given some estimate of mutation rate) amount of neutral genetic variation maintained within a species at mutation-drift equilibrium . Estimates of Ne predict the rate of loss of genetic diversity and the efficacy of selection, with the action of selection compared with random genetic drift favoured in large effective populations [1,2]. Further insights can be gained from the ratio Ne/N, where N is the number of potentially reproductively active individuals, which quantifies the extent of departure from idealized population genetic models and the susceptibility of apparently large populations to stochastic evolutionary processes [3–5].
The potentially huge population sizes of free-living, unicellular eukaryotes (protists) would appear to render their populations somewhat unsusceptible to random genetic changes. However, we are not aware of any estimates of contemporary Ne in any free-living protist. This lack of data raises fundamental questions about the magnitude of contemporary protist effective population sizes as well as the extent to which estimates of Ne are less than the absolute numbers of cells present.
Obtaining reasonable estimates of contemporary Ne in field populations is notoriously difficult [1–6]. The most widely used genetic methods to estimate Ne exploit the association between population size and the magnitude of (stochastic) temporal changes in allele frequencies [1,6–8]. Such temporal estimates of Ne require genotypes from two or more samples separated by several generations, which are not available for most taxa unless there have been sufficient foresight and resources to establish a long-term sampling programme and maintain appropriate material archives.
Obtaining temporally separated samples can be straightforward for many marine protists, such as numerous dinoflagellates and diatoms, when they produce resting stages (cysts) that accumulate in benthic sediments. Protist cysts may remain viable for up to 100 years [9,10], thereby providing a source of naturally archived historic material. We revived cysts of Pentapharsodinium dalei, a spring-blooming marine dinoflagellate that inhabits polar, sub-polar and cold-temperate coastal regions, from samples spanning some 100 years to make the first estimates of contemporary Ne in a free-living protist population.
2. Material and methods
All viable P. dalei cysts were recovered and cultured from four slices of sediment core taken from Koljö Fjord, Sweden (58°13′ N, 11°34′ E). Estimated dates for the core slices were 2006, 1985 (±3 years s.e.), 1960 (±5 years s.e.) and 1922 (±12 years s.e.) [9,10], representing changes in population composition over approximately a century (84 ± 12 years s.e.); these slices yielded 41, 31, 33 and 29 clonal strains, respectively, with each strain representing a single haploid cell isolated from a cell culture of a germinated cyst. Observations indicate that the cysts of P. dalei are hypnozygotes (diploid cells resulting from sexual fusion), and thus part of the sexual cycle . The frequency of sexual reproduction has not been determined for this species, but as cysts are considered to be survival stages that are produced most frequently at the end of bloom periods, it is a reasonable assumption that P. dalei reproduces sexually once a year.
Genomic DNA from monoclonal cultures was extracted using a CTAB method . Samples were genotyped at six microsatellite loci . PCRs consisted of 0.3 µM forward and reverse primers (forward primers 5′-labelled with 6-FAM, NED or HEX fluorophores; Applied Biosystems), 10 µl HotStarTaq Master Mix (Qiagen) and published thermal cycling conditions . PCR products were pooled with Genescan LIZ500, separated by capillary electrophoresis on an ABI3730xl and sized using GeneMapper (Applied Biosystems).
Contemporary estimates of Ne (and 95% CI) for all sample pairs were made using two temporal methods. First, MLNE was used to estimate Ne using a maximum-likelihood (ML) estimator  that does not assume a particular mating system or require genetic equilibrium, and applies to small and large populations. Second, we used NeEstimator v. 2  to calculate Ne using Waples' moment method  based on Nei & Tajima's  estimator of standardized variance in allele frequency change. Our calculations assumed a 1-year generation time (the timing of the sexual cycle), a closed population (Koljö Fjord has limited water exchange with elsewhere (see electronic supplementary material)) and a population not at migration–drift equilibrium.
The total abundance of cells (N) in Koljö Fjord was estimated using data on the densities of P. dalei in the neighbouring Gullmar Fjord and the bathymetry of Koljö Fjord (see electronic supplementary material) and are estimates of the numbers of the vegetative (increasing by cell division) stages at a given time.
All loci were polymorphic, yielding a total of 80 alleles (see electronic supplementary material). Genotyping revealed some haplotypes originating from the same cyst, which left 21, 27, 20 and 18 different haploid genotypes for analysis. One locus failed to amplify alleles in one to two clones per sample. There was no significant difference among samples in the frequency of non-amplifying genotypes (Kruskal–Wallis χ2 = 0.602, d.f. = 3, p = 0.896) and the presence of null alleles would have little impact upon estimates of Ne [8,14]; treating missing data as distinct (null) alleles does not affect the basic results (see electronic supplementary material).
None of our estimates of Ne in P. dalei was particularly huge. The ML estimates of contemporary Ne of P. dalei among pairs of samples in Koljö Fjord lay between 179 and 815, with a maximum upper 95% CI of 2489; comparable, albeit slightly higher, estimates of Ne were derived using the moment estimator (Ne = 178–1183, with a maximum upper 95% CI = 5545; table 1). A moderate Ne remained when we examined the potential effects of error associated with sediment dating (see electronic supplementary material).
Based on a conservative value of 1500 cells l−1 in the water column above the halocline, the number of vegetative cells of P. dalei in Koljö Fjord was estimated as approximately 2.9 × 1014 (see electronic supplementary material). While these calculations are approximate, the Ne for P. dalei represents a tiny fraction of the number of cells present (N), with an Ne/N ∼ 10−12 (and between approx. 10−11 and approx. 10−13 based on the maximum and minimum 95% CIs).
Palstra & Ruzzante's  meta-analysis yielded an average contemporary Ne ∼ 260 for wild populations. Our first estimates of contemporary Ne in a free-living protist population are robust to error associated with sediment dating and are mostly of the same order of magnitude as those of macroscopic species. A corollary of the numerical abundance of protist cells is an extremely low Ne/N ratio of approximately 10−12 in P. dalei, which is the lowest value for any taxon that we are aware of. For comparison, median estimates of Ne/N for animals are around 0.1–0.14 [3,4], but this ratio typically is lower (approx. 10−3 to 10−5) in marine teleosts [4,5,15,16] and approaches approximately 10−8 in freshwater copepods ; the single estimate of Ne/N in an abundant marine macroalga, Fucus serratus, was between 10−3 and 10−4 .
The ecological and life-history processes behind P. dalei's low Ne/N are not known but are probably a consequence of temporal fluctuations in population size, as well as variation among strains in achieving genetic representation  and extensive vegetative cell division (that increases N). Also, our study population may have violated some of the model requirements. For example, Koljö Fjord is not completely closed and estimates of Ne based on an assumption of no gene flow can be biased compared with estimates of Ne that account for gene flow [8,18], with the direction and magnitude of the bias determined by the genetic composition of the source population(s) [4,14] and the amount of gene flow ; for example, under very high gene flow (i.e. where all individuals within a local population, for example in Koljö Fjord, are derived from the global population), temporal estimates of Ne for a local population are expected to be biased downwards by up to 50% owing to an additional variance component . Future studies of protists should consider this effect by quantifying spatial genetic structure and by jointly estimating Ne and rates of migration . The impact upon our estimates of Ne of other aspects of protist life history, such as a lack of separate sexes or overlapping generations owing to possible revival of buried cysts, is not clear. However, the long interval between slices of sediment core should provide temporal estimates of Ne with increased precision and reduced bias (owing to any effect of overlapping generations) . Low Ne/N ratio may reflect the action of selection, with hitchhiking effects expected to be important when there is substantial clonal reproduction , and insufficient sampling is expected to be an issue when Ne is very large [4,5]. Irrespective of the underlying driver(s), an extremely low Ne/N ratio appears to be characteristic of numerically abundant multi-cellular aquatic organisms [4,5,14–17] and aquatic protists apparently are no exception. The implication of low contemporary Ne is that stochastic genetic processes are more prevalent than anticipated from a population census.
One key issue with the concept and interpretation of Ne in wild populations is that it may be defined in many ways [1,2,5], including a long-term Ne (defined in the Introduction). Although estimates of long-term Ne are perhaps less relevant to understanding contemporary evolutionary processes , the few data for protists provide evidence for comparatively large [19,20] and small  long-term Ne. Given the diversity of protist life histories, we expect that additional studies on protists will find comparable inter- and intra-specific variation in contemporary Ne: understanding the drivers behind any diversity in Ne presents an exciting challenge. With this in mind, contemporary Ne can be estimated from a single genetic sample [1,22], providing the advantage that every sample gives an estimate of Ne ; moreover, some single sample estimators of Ne outperform temporal estimators of Ne when population sizes are small and there are large sample sizes . By contrast, temporal estimators of Ne are appropriate for characterizing large populations, when there is a long time interval between samples , as here, and when there is high gene flow . We did not use single sample estimators of Ne because (i) they suffer from poor statistical power when few loci are used  and (ii) an analysis of haploid data represented an important departure from their underlying model assumptions. Nonetheless, for protists that can be genotyped at many loci and during a diploid phase, single sample estimators of Ne could also be used . For most protists, however, there remains an apparent need for temporally separated samples to obtain accurate estimates of Ne, and this should not present a major obstacle for species that produce cysts as they are amenable to the palaeogenetic approach used here (see also [9,10,23]).
Genotypes archived with Dryad: doi:10.5061/dryad.221t6.
The authors are part of PRODIVERSA, a NordForsk-funded network on protist population genetics. Laboratory work was financially supported by the Danish Research Council (2111-04-0011) and fieldwork by Göteborg University Marine Research Centre. S.R. holds a postdoctoral fellowship from the Carlsberg Foundation (2011_01_0337).
We would like to thank Jinliang Wang, David Tallmon and Robin Waples for their helpful comments about estimating Ne with haploid samples, and also Adam Eyre-Walker, Daniel Ruzzante and the editors for their insightful critiques of the manuscript.
- Received October 2, 2013.
- Accepted November 7, 2013.
- © 2013 The Author(s) Published by the Royal Society. All rights reserved.