Biological Conservation 143 (2010) 946–951

Contents lists available at ScienceDirect

Biological Conservation

journal homepage: www.elsevier.com/locate/biocon

Ecological effects on estimates of effective in an annual plant

E.K. Espeland a,*, K.J. Rice b a USDA ARS Pest Management Research Unit, 1500 N. Central Avenue, Sidney, MT 59270, USA b Graduate Group, Department of Plant Sciences, Mail Stop 1, University of California, Davis, One Shields Avenue, Davis, CA 95616, USA article info abstract

Article history: Effective population size (Ne) is a critical indicator of the vulnerability of a population to allele loss via Received 13 April 2009 , and it can also be used to assess the evolutionary potential of a population. While some Received in revised form 28 December 2009 plant conservation plans have focused on outcrossing through cross-pollination as a way to increase esti- Accepted 3 January 2010 mated N , variance in reproductive output determined by ecological factors such as can also Available online 25 January 2010 e strongly affect estimated Ne. We examined the effects of intraspecific and interspecific competition, stressful soils, and local on estimates of Ne in an annual plant species. While ecological influ- Keywords: ences on plant growth rate variance have been predicted to influence estimates of N /N, we found a sig- e nificant effect on the estimate of N /N, but no significant ecological effects on growth rate variance. Lower Competition e Effective population size survivorship on stressful soil was the most important effect reducing estimates of Ne/N. If stochastic mor- Genetic drift tality is greater in environments that are abiotically stressful, then populations in these stressful environ- Plantago erecta ments may be slower to adapt because of lower census sizes and reduction of Ne/N. In populations of

Local adaptation conservation concern, increasing survivorship may be of greater benefit for maximizing Ne than the Serpentine reduction of variability in reproductive output among surviving adults. Conservation genetics Ó 2010 Elsevier Ltd. All rights reserved.

1. Introduction (Lande, 1988; Frankham, 1995; Hedrick, 1995; Frankham et al., 2003), many of which are ecological in nature.

Effective population size (Ne) can be used to assess the evolu- Published estimates of Ne in plant populations have most often tionary potential of a population because it is an important indica- been used in a descriptive manner to characterize population sub- tor of the vulnerability of a population to allele loss through division and differentiation (Tremblay and Ackerman, 2001; Latta, random processes (genetic drift). A population with a very low 2008). Conservation biologists working with both plants and ani-

Ne is more susceptible to genetic drift and less able to respond to mals have used Ne estimates to explore the importance of popula- selection. This is because in small populations there is less genetic tion bottlenecks on (Frankham, 1995; Amos and variation for to act upon, and there is a higher Harwood, 1998). Other conservation uses of Ne estimates have fo- probability that beneficial alleles will not be maintained by selec- cused on reconstructions of historical anthropogenic effects on tion and will instead be lost from the population because of gene flow and genetic diversity (Levy and Neal, 1999; Morris random drift effects (Willi et al., 2007). Modeling population ge- et al., 2002). netic processes in conservation has become relatively widespread Ne can be estimated in a number of ways, and there are three (Halley and Manasee, 1993; Higgins and Lynch, 2001; Obioh and commonly-used types of Ne. Inbreeding Ne describes the probabil- Isichei, 2007; Pertoldi et al., 2007; Palstra and Ruzzante, 2008), ity of mating among relatives, therefore lowering genetic diversity and when Ne is incorporated into minimum viable population size of the population. Variance Ne describes the probability that indi- (MVP) models, MVP as censused may need to be substantially viduals will pass on their genes to the next generation, with in- more than 5000 individuals (Obioh and Isichei, 2007). Rare and creased variance in offspring number leading to a decrease in Ne. endangered plant populations in particular often have very low or eigenvalue Ne describes the rate of loss of heterozy- census sizes (N) and even lower estimated Ne (Chung et al., 2007; gosity, with a calculation of an asymptotic Ne as the outcome. Zietsman et al., 2008). Plant and animal populations of conserva- Although estimates of the three types of Ne often have the same re- tion concern tend to have multiple factors acting to reduce Ne sult (Vitalis and Couvet, 2001), they can differ, so the choice of which type to calculate may revolve around the genetic process

of interest (Caballero, 1994). Inbreeding Ne is often estimated via * Corresponding author. Tel.: +1 406 433 9416; fax: +1 406 433 5038. genetic methods by calculating the percent heterozygosity and E-mail addresses: [email protected] (E.K. Espeland), kjrice@ucdavis. comparing it to the expected heterozygosity in the population if edu (K.J. Rice).

0006-3207/$ - see front matter Ó 2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.biocon.2010.01.003 E.K. Espeland, K.J. Rice / Biological Conservation 143 (2010) 946–951 947

it were in equilibrium (Hartl and Clark, 2007). Inbreeding Ne can distribution reduces the CV of plant sizes as densities in- also be estimated as the geometric mean of population sizes crease (Koide and Dickie, 2002). This would lead to a relative in- through time (Caballero, 1994). Variance Ne is estimated as the var- crease in estimated Ne /N when seed output is positively iance in reproductive outputs among individuals in a population correlated with plant size.

(Heywood, 1986). Published estimates of Ne /N for annual plants In this experiment we used populations of the California native tend to be between 5% and 30%, with ratios most often below annual forb, Plantago erecta (E. Morris), to examine environmental 10% (Heywood, 1986; Husband and Barrett, 1992; Goldringer and evolutionary influences on variance in reproductive output et al., 2001; Siol et al., 2007). and estimated Ne. We designed the experiment to test the relative Recent work by Siol et al. (2007) suggests that relative repro- importance of density-dependent interactions and interspecific ductive variance must be taken into account in order to correctly competition on reproductive hierarchies, and how stressful soils estimate Ne. As understanding of the influence of reproductive var- and local adaptation may alter estimates of Ne/N. iance on Ne estimation has increased, evidence has also accumu- lated to demonstrate that ecological factors can be extremely important in determining this variance (Van Kleunen et al., 2001, 2. Methods and materials 2005). In fact, ecological models of density-dependence have al- ready been theoretically extended to the estimation of Ne (Rice, P. erecta is an annual plant with a native range extending from 1990; Koide and Dickie, 2002; Van Kleunen et al., 2001, 2005). Baja California and Arizona north through the California Floristic Other ecological factors such as herbivory (Doak, 1992) can affect Province to southern Oregon. Although completely self-compati- the variance in reproductive outputs in plants and thus influence ble, some outcrossing is likely in this species (E. Espeland, unpub- estimates of Ne/N. Although Ne is often estimated from the geomet- lished data). Populations of P. erecta typically occur at high ric mean of population sizes of above-ground plants, seed banks in- densities and are found in shallow, low fertility soils (e.g., serpen- crease Ne above this estimate because the seed bank is often not tine outcrops, road cuts) as well as in deep, more fertile grassland part of the censused population size (Nunney, 2002). Thus, ecolog- soils. ical factors that reduce the seed bank such as granivory and fungal The experiment was conducted at McLaughlin Reserve infection will act to reduce Ne. (38.52°N, 122.24°W), located within the California North Coast The linkage between variation in reproductive output and Range (Hickman, 1993) and operated by the University of Califor- estimates of Ne in annual plants was explored theoretically by nia Natural Reserve System. Seeds were collected from the field Heywood (1986) and can be expressed by the following in spring 2004 from four serpentine populations and four non-ser- relationship: pentine populations at McLaughlin Reserve. Serpentine soil is a nutrient-limited, ultramafic soil type characterized by low calcium, Ne 1 very high levels of exchangeable magnesium (Jurjavcic et al., 2002), ¼ ð1Þ N ½ð1 þ FÞðr2=l2Þþ1 and toxic heavy metals: a stressful environment for plant growth (Kruckeberg, 1984). Serpentine areas are also often drier than where the inbreeding coefficient F ranges from 0 (complete out- non-serpentine areas (Macnair et al., 1989). Serpentine popula- crossing) to 1 (complete inbreeding), r2 = variance in reproductive tions of P. erecta were identified by the presence of serpentine en- output, and l = mean reproductive output. It follows that if an an- demic plant species. Serpentine grasslands occur in a mosaic with nual plant population with a large census size has a few individuals loam grasslands at this site. Four populations on each soil type that produce the majority of the seed, reproductive variance is high within a geographic range of three linear kilometers were selected and estimated Ne/N is small. In contrast, Ne is larger and estimated to span a range of productivities (estimated by total above-ground Ne/N is close to 1 in an annual plant population of large census size ; E. Espeland, unpublished data) and were spatially that has a variance in reproductive output that approximates a Pois- blocked into serpentine/loam paired plots. Each population was son distribution. While the absolute value of Ne is more important about 0.5 ha in size. Seeds from 30 families from each population than this ratio, we can use this ratio to determine the environmen- were bulked by soil type (serpentine or non-serpentine) before tal influences that have the most important effect on estimates of replanting into field collection locations. To increase germination,

Ne. Because N in plant populations is often determined by larger- dry seeds were chilled at 4 °C for 10 days prior to sowing. The scale factors that we cannot influence, such as rainfall patterns seeds were then planted into 16 circular 38.5 cm2 plots in a

(Zietsman et al., 2008), the estimated Ne/N ratio is more useful as 1m 1 m area at each of the eight locations on October 10 and a relative index to apply to the evaluation of conservation practices 11, 2004. A factorial combination of two levels of interspecific that may maximize Ne. competition, two sowing densities, and two seed sources (serpen- In annual plant species, size is highly correlated with reproduc- tine, non-serpentine) was replicated twice at each location. The 16 tive output, and competition is one of the main determinants of plots at each location comprised a randomized complete block plant size (Harper, 1977). Models of competitive interactions not design. only predict relative plant sizes, but also the distribution of plant For the high sowing density treatment, 15 seeds were affixed to sizes in a population and thus the coefficient of variation in plant a single 0.2 cm2 piece of tissue paper with water-soluble glue. This sizes. In annual plants, the coefficient of variation in plant sizes group of seeds was placed in the center of the plot and buried at a is directly related to the coefficient of variation in reproductive depth of 2 mm. For the low density sowing treatment, each seed outputs (Heywood, 1986). A relevant model of above-ground was glued to a single toothpick and 10 toothpicks were inserted asymmetric competition is where large plants shade small ones into the soil of each plot so that the seed was buried at a depth and the coefficient of variation (CV) of plant sizes increases with of 2 mm. The 10 toothpicks were evenly spaced over the plot area. density (Weiner and Thomas, 1986). This larger CV of plant sizes Interspecific competition was manipulated experimentally. For in high density populations decreases estimated Ne/N (see Eq. (1)). half the plots, only P. erecta germinating from the resident seed A facilitative model of plant–plant interactions can also affect bank was removed throughout the experiment. For the other half the CV of reproductive outputs. Resource sharing via common of the plots, all plants except for the planted P. erecta plants were mycorrhizal networks can assist resource capture in smaller plants removed throughout the fall and winter. This resulted in a range and increase their growth rates relative to larger plants (Shumway of biomass of plant species other than Plantago in the plots (here- and Koide, 1995; Selosse et al., 2006). This type of equalizing after referred to as ‘‘interspecific biomass”). 948 E.K. Espeland, K.J. Rice / Biological Conservation 143 (2010) 946–951

Survivorship was tracked over the growing season as well as the were close to normality (Shapiro-Wilk >0.88), and no transforma- height of each P. erecta plant in our plots. Plants were harvested tion of CV or relative reproductive variance was necessary. prior to seed ripening at each site in spring 2005 (March 10, To test for fitness effects of local adaptation across life history 2005 to June 7, 2005). Seed output was estimated by multiplying stages, we examined the interaction of planting soil and source soil the number of flowers + buds by two, as there are two ovules per on the number of flowers produced per seed sown using a Tukey flower and P. erecta rarely aborts ovules (61% abortion rate, E. HSD test. Our seed collections were bulked across interspersed ser- Espeland, unpublished data). P. erecta flowers synchronously at pentine and non-serpentine sites, and we are therefore confident each location, much more synchronously than congeners found that any differences between serpentine and non-serpentine elsewhere such as P. aristata and P. patagonica (E. Espeland, per- sources are due to the influence of soil type and not site specific sonal observation). Because there is a relatively long period of time factors. Because we used field collected seed, the influence of seed between flowering and seed ripening (3–5 weeks, E. Espeland, per- source on fitness may be due to maternal environmental effects, sonal observation), we are confident that all plants that would heritable , or both. have flowered that year had done so by harvest time. To ensure we did not underestimate the reproductive output of any plants 3. Results that may have begun flowering later than the main cohort, we in- cluded unopened buds as well as open flower numbers in our Plants from non-serpentine sources produced significantly reproductive output calculation. This is a measure of maternal fit- flowers per seed sown on serpentine soil (p < 0.05). Serpentine ness; we did not attempt to measure paternal fitness in this exper- sources did not differ significantly (p > 0.05) in flower production iment. At harvest, all interspecific plant biomass was harvested between the two soil types, although the trend for serpentine from each plot, placed in a drying oven at 65 °C for 6 days, and then sources was to produce more flowers than non-serpentine sources weighed. on serpentine soil and fewer flowers than non-serpentine sources Because we measured , not germination, in the field on non-serpentine soil (Fig. 1). experiment, we conducted a growth chamber experiment to exam- There were no significant treatment effects on the CV of plant ine inherent differences in germination rates between serpentine sizes at the end of the growing season (Table 1, left side). Planting and non-serpentine populations. Of the eight populations studied, soil type was the only treatment factor that had a significant effect seeds were collected from three serpentine and three non-serpen- on relative reproductive variance (Table 1, right side). Plants grow- tine populations during March 2008. These seed collections repre- ing on non-serpentine soil exhibited 14% lower relative reproduc- senting different seed source soil types were stored at room tive variance than plants growing in serpentine soil (Table 2, left temperature until August 26, 2008. On that date, three Petri plates, side). fitted with moistened filter paper, and each containing 25 seeds Germination totals and rates did not differ significantly be- each per population were placed in a dark germination chamber tween serpentine and non-serpentine collections in the 2008 ger- (12 h at 4 °C and 12 h at 16 °C for 14 days). Plates were scored mination trial (p > 0.7). Over the 14-day trial period, germination semi-weekly for germination, and germinants were removed at in non-serpentine sources averaged 92 ± 13% SD while germination each census. for serpentine sources averaged 86 ± 13% SD.

Using Eq. (1), we estimated Ne/N ratios for significant treatment effects on relative reproductive variance. Estimated N /N values 2.1. Data analysis e were 8% smaller in plant populations growing on serpentine soil compared to populations growing on non-serpentine (Table 2, We sowed P. erecta at two extremely different densities to test if right side). extremes in intraspecific density affect the response to the other Because the results for CV and relative reproductive variance experimental treatments. Effects of sowing density, interspecific differed, this indicated that survival may play a key role in esti- biomass, location soil type (serpentine or non-serpentine soil at mates of N /N. Therefore, we performed an analysis of treatment ef- planting location), and seed source soil type on plant size CV and e fects on survivorship. We applied the same statistical model we relative reproductive variance were assessed. The number of P. used for CV and relative reproductive variance to arcsine trans- erecta that emerged in each plot was used as a covariate. CV was formed percent survivorship. Only the interaction between inter- calculated as the coefficient of variation (standard deviation/mean) specific biomass and planting soil type was close to significant in in plant sizes at the end of the experiment. This estimates the var- determining survivorship (p = 0.07). Competition from interspecif- iance in plant growth rates within the population (Rice, 1990). Rel- ics had no significant effect on survivorship on non-serpentine soil, ative reproductive variance was calculated as the CV2 (variance/ but increased interspecific biomass decreased survivorship on ser- mean2) of estimated seed output for plants that germinated in pentine soil (arcsine% survivorship = 1.25 0.55 g interspecific the plot, including zeros for plants that died. This calculation of rel- biomass, R2 = 0.08, p < 0.02). ative reproductive variance relates directly to estimated Ne/N (see Eq. (1)). Because plant size is highly correlated with reproduction in this species (Espeland and Rice, 2007), the two dependent vari- 4. Discussion and conclusions ables, CV and relative reproductive variance, should differ primar- ily by the inclusion of zeros in the relative reproductive variance In our experiment, non-serpentine seed sources are more calculation, thus partitioning the relative effect of survivorship strongly adapted to the less stressful non-serpentine soil. Seeds on estimated Ne/N. from serpentine sources do not perform significantly differently We tested a subset of interactions of interest in our statistical on the two soil types. Although we found evidence for local model. Source soil by planting soil was tested to determine the ef- adaptation, local adaptation does not appear to have an impor- fect of possible local adaptation. To determine the effect of soil tant influence on relative reproductive variance and estimated type on inter- and intra-specific interactions, the interactions of Ne/N. In addition, if ecological factors that influence variance in (1) planting soil by interspecific biomass and (2) planting soil by plant growth rates also influence relative reproductive variance, sowing density were tested. To determine the effect of local adap- we would expect parallel results in our CV and relative tation on each of these two-way interactions, source soil was reproductive variance analyses. However, plant growth rate var- added to test the significance of three-way interactions. The data iance (CV) was unaffected by our experimental treatments, and E.K. Espeland, K.J. Rice / Biological Conservation 143 (2010) 946–951 949

2 A AB 1.5 AB B

1

0.5

Number of flowers per seed seed sown per flowers of Number 0 Planting soil Nonserpentine Serpentine Nonserpentine Serpentine

Source soil NONSERPENTINE SERPENTINE

Fig. 1. Non-serpentine sources produce the most flowers per seed sown in home soil, suggesting local adaptation to soil type (bars with different letters are significantly different, p < 0.05). Serpentine sources show a trend towards local adaptation, but this difference is not significant.

Table 1 Statistical tables of (a) treatment and covariate effects on coefficient of variation and on relative reproductive variance, (b) growing soil and source soil on number of flowers per seed sown, p < 0.05 in bold.

Source Coefficient of variation Relative reproductive variance DF SS F-ratio P > F SS F-ratio P > F (a) #P. erecta emerged 1 0.423 3.29 0.072 0.072 1.28 0.261 Block 3 0.816 2.11 0.103 0.587 3.44 0.019 Sowing density (D) 1 0.300 2.33 0.130 0.028 0.498 0.482 Interspecific biomass (B) 1 0.052 0.41 0.524 0.010 0.178 0.674 Soil 1 0.306 2.39 0.125 0.274 4.83 0.030 Source 1 0.204 1.58 0.211 0.062 1.10 0.297 Soil source 1 0.274 2.13 0.147 0.152 2.68 0.105 Soil B 1 0.002 0.02 0.896 0.009 0.16 0.687 Source soil D 1 0.047 0.37 0.546 0.0005 0.01 0.922 Source soil D 1 0.423 3.29 0.072 0.036 0.64 0.427

Source DF SS F-ratio P > F (b) Soil 1 1.064 1.591 0.210 Source 1 0.141 0.212 0.646 Soil source 1 4.441 6.641 0.011

Table 2 Average for relative reproductive variance (RRV) ± one standard deviation, plus is unlikely, and that inbreeding will decrease heterozygosity. The estimated Ne/N by location soil. census size of plant populations is positively correlated with fit- ness measures (Reed, 2005), and in a recent meta-analysis Leimu RRV Ne/N and Fischer (2008) showed that populations of small census size Non-serpentine 0.59 ± 0.20 0.74 Serpentine 0.69 ± 0.29 0.68 are less likely to be locally adapted. This meta-analysis is sup- ported by individual studies showing a positive ability for larger populations to adapt to new environments (Willi and Hoffman, 2009). When lowered survival not only makes census sizes small relative reproductive variance was significantly influenced only but also makes the estimated Ne/N ratio small, this suggests that by planting soil type and block. Stressful serpentine soil in- the population will also be less likely to retain any beneficial ge- creased relative reproductive variance compared to non-stressful netic variation because of genetic drift. Plant populations growing soil. Although we would expect below-ground competition on on stressful soil types have a twofold obstacle to local adaptation: stressful soil types to decrease plant size hierarchies and de- reduced population growth rates and reduced genetic variation crease plant growth rate variance, this was not the case. Plant (low Ne/N) less able to respond to selection (Willi and Hoffman, growth rate variance was not affected by planting soil, and rela- 2009). Plants and other sessile organisms that are limited in their tive reproductive variance was higher, not lower, in populations ability to sample the environment may thus suffer doubly in on the stressful soil type. The lack of parallel between CV and stressful environments. Plant populations then face a more difficult relative reproductive variance indicates that survivorship plays evolutionary hurdle in adapting to stressful environments com- a significant role in the observed relationship between soil type pared to non-stressful environments. and estimated Ne/N. In contrast, populations growing under less stressful abiotic When survivorship is the main driver of estimated Ne/N, stress- conditions, with larger census sizes and larger estimated Ne /N, ful environments may intrinsically have lower estimated Ne/N as may both generate and harbor more genetic variation. Greater evo- well as lower census sizes. Low census sizes in plant populations lutionary potential in populations with larger estimated Ne /N mean that the accumulation of new variation through should allow them to more effectively respond to natural selection 950 E.K. Espeland, K.J. Rice / Biological Conservation 143 (2010) 946–951 and adapt to local conditions. While stress-adapted species and References populations are commonly documented (Chapin et al., 1993), adaptation to stressful environments may be less common than ex- Amos, W., Harwood, J., 1998. Factors affecting levels of genetic diversity in natural populations. Philosophical Transactions of the Royal Society of London Series B pected because lower census sizes and lower estimated Ne/N may – Biological Sciences 353, 177–186. slow the adaptation process. Within stressful environments, the Caballero, A., 1994. Developments in the prediction of effective population size. balance between strong selection regimes and the evolutionary Heredity 73, 657–679. capacity of populations to respond to selection is an important Chapin, F.S., Autumn, K., Pugnaire, F., 1993. of suites of traits in response to environmental stress. American Naturalist 142, S78–S92. consideration in the management of populations of conservation Chung, M.Y., Park, C.W., Chung, M.G., 2007. Extremely low levels of allozyme concern. variation in southern Korean populations of the two rare and endangered In small populations of conservation concern, factors that re- lithophytic or epiphytic Bulbophyllum drymoglossum and Sarcanthus scolopendrifolius (Orchidaceae): implications for conservation. and duce estimated Ne will also decrease the ability of a population Conservation 16, 775–786. to retain beneficial alleles. These factors may be ameliorated with Doak, D.F., 1992. Lifetime impacts of herbivory for a perennial plant. Ecology 73, appropriate management practices. In this case, with a common 2086–2099. Espeland, E.K., Carlsen, T.M., MacQueen, D., 2005. Fire and dynamics of granivory on plant species, estimated Ne/N was well above 0.5, indicating that a rare California grassland forb. Biodiversity and Conservation 14, 267–280. reproductive variance is not the main influence on genetic drift Espeland, E.K., Rice, K.J., 2007. Facilitation across stress gradients: the importance of processes in these populations (Heywood, 1986). Ecological influ- local adaptation. Ecology 88, 2404–2409. Frankham, R., 1995. Effective population size/adult population size ratios in ences on survivorship, not growth rates, were the main factor wildlife: a review. Genetics Research 66, 95–107. determining estimated Ne/N in P. erecta. No studies that we know Frankham, R., Ballou, J.D., Brisco, D.A., 2003. Introduction to Conservation Genetics. of have examined the effect of differential survival on estimated Cambridge University Press, Cambridge, UK. Goldringer, I., Enjalbert, J., Raquin, A.-L., Brabant, P., 2001. Strong selection in wheat Ne /N estimates in plants, however, survivorship is an extremely populations during ten generations of dynamic management. Genetics important factor in estimating Ne /N in marine species (Waples, Selection Evolution 33, S441–S463. 2002). Maximizing survivorship within semelparous species of Halley, J.M., Manasee, R.S., 1993. A population-dynamics model for annual plants conservation concern may be more important than reducing the subject to inbreeding depression. Evolutionary Ecology 7, 15–24. Harper, J.L., 1977. Population Biology of Plants. Academic Press, New York, NY. variability among surviving breeding adults in these species. Hartl, D.L., Clark, A.G., 2007. Principles of . Sinauer Associates, To ensure that plant populations have the potential to adapt to Sunderland, MA. changing environmental conditions, we need to further investigate Hedrick, P.W., 1995. Elephant seals and the estimation of a population bottleneck. Journal of Heredity 86, 232–235. ecological drivers of relative reproductive variance that include Heywood, J.S., 1986. The effect of plant size variation on genetic drift in populations survivorship, within plant populations in order to limit the effects of annuals. The American Naturalist 127, 851–861. of genetic drift within these populations. Although some plant con- Hickman, J.C., 1993. The Jepson Manual: Higher Plants of California. University of California Press, Berkeley, CA. servation plans have focused on promoting cross-pollination as a Higgins, K., Lynch, M., 2001. extinction caused by mutation way to increase estimated Ne (Ingvarsson, 2002; Porcher et al., accumulation. Proceedings of the National Academy of Sciences 98, 2928–2933. 2004), variance in reproductive output can often have a much Husband, B.C., Barrett, S.C.H., 1992. Effective population-size and genetic drift in tristylous Eichhornia paniculata (Pontederiaceae). Evolution 46, 1875–1890. greater effect on estimated Ne than breeding system (Heywood, Ingvarsson, P.K., 2002. A metapopulation perspective on genetic diversity and 1986). Plant conservators are beginning to prioritize reducing differentiation in partially self-fertilizing plants. Evolution 56, 2368–2373. reproductive variance in order to maintain genetic diversity Jurjavcic, N.L., Harrison, S., Wolf, A.T., 2002. Abiotic stress, competition and the distribution of the native annual grass Vulpia microstachys in a mosaic (Yonezawa et al., 1996; Vencovsky and Crossa, 2003). Our results environment. Oecologia 130, 555–562. show that mortality may strongly influence reproductive variance, Koide, R.T., Dickie, I.A., 2002. Effects of mycorrhizal fungi on plant populations. Plant thus reducing estimated Ne/N. The source of mortality in our exper- and Soil 244, 307–317. iment appeared to be fungal infection at the seedling stage. In Kruckeberg, A.R., 1984. California Serpentines: Flora, Vegetation, Geology, Soils and Management Problems. Univ. of California Press, Berkeley. some cases plant conservationists may be able to increase survi- Lande, R., 1988. Genetics and demography in biological conservation. Science 241, vorship in populations of concern, for example by excluding preda- 1455–1460. tors in select years (Espeland et al., 2005). Latta, R.G., 2008. Conservation genetics as applied evolution: from genetic pattern to evolutionary process. Evolutionary Applications 1, 84–94. Effective population size may need to be managed very differ- Leimu, R., Fischer, M., 2008. A meta-analysis of local adaptation in plants. PLoS ONE ently for plants and animals. Colonizing species such as annual 3, e4010. doi:10.1371/journal.pone.0004010. plants produce large numbers of progeny with the same genetic Levy, F., Neal, C.L., 1999. Spatial and temporal genetic structure in chloroplast and allozyme markers in Phacelia dubia implicate genetic drift. Heredity 82, 422–431. background, and not all of these progeny will survive. In these col- Macnair, M.R., Macnair, V.E., Martin, B.E., 1989. Adaptive in Mimulus –an onizing species, mortality may play a larger role in genetic drift ecological comparison of Mimulus cupriphilus with its presumed progenitor processes compared to species whose population growth rates Mimulus guttatus. New Phytologist 112, 269–279. Morris, A.B., Baucom, R.S., Cruzan, M.B., 2002. Stratified analysis of the soil seed are determined by intraspecific competition, such as large herbi- bank in the cedar glade endemic Astragalus bibullatus: evidence for historical vores (Waples, 2002). Environmental factors can very strongly af- changes in genetic structure. American Journal of Botany 89, 29–36. Nunney, L., 2002. The effective size of annual plant populations: The interaction of a fect Ne /N estimates and thus environmental factors may be seed bank with fluctuating population size in maintaining genetic variation. The primary determinants of a population’s ability to retain genetic American Naturalist 160, 195–204. variation through time. Obioh, G.I.B., Isichei, A.O., 2007. A population viability analysis of serendipity berry (Dioscoreophyllum cumminsii) in a semi-deciduous forest in Nigeria. Ecological Modelling 201, 558–562. Palstra, F.P., Ruzzante, D.E., 2008. Genetic estimates of contemporary effective Acknowledgements population size: what can they tell us about the importance of genetic stochasticity for wild population persistence? Molecular Ecology 17, 3428– This work was performed at the McLaughlin Reserve of the Uni- 3447. Pertoldi, C., Bijlsma, R., Loeschcke, V., 2007. Conservation genetics in a globally versity of California Natural Reserve System (UCNRS) and partially changing environment: present problems, paradoxes and future challenges. funded by a Mildred E. Mathias Grant for work at the UCNRS to EKE Biodiversity and Conservation 16, 4147–4163. and a Packard Foundation Interdisciplinary Science Grant (2000- Porcher, E., Gouyon, P.H., Lavigne, C., 2004. Dynamic management of genetic resources: maintenance of outcrossing in experimental of a 01607) to both K.J.R. and E.K.E. Thanks to S. Mueller for field assis- predominantly inbreeding species. Conservation Genetics 5, 259–269. tance, to M. O’Mara for lab assistance, and to S. Harrison, K. Moore, Reed, D.H., 2005. Relationship between population size and fitness. Conservation S. Elmendorf, M. Schlesinger, K. Jones, J. Harding, the Big Science Biology 19, 563–568. Rice, K.J., 1990. Reproductive hierarchies in Erodium – effects of variation in plant- Lab, D. Grace, J. Weiner, S. Sultan, A. Caballero, J. Gaskin, and E. Le- density and rainfall distribution. Ecology 71, 1316–1322. ger for comments on early versions of the manuscript. E.K. Espeland, K.J. Rice / Biological Conservation 143 (2010) 946–951 951

Selosse, M., Richard, F., He, X., Simard, S.W., 2006. Mycorrhizal networks: des Vitalis, R., Couvet, D., 2001. Estimation of effective population size and migration liaisons dangereuses? Trends in Ecology and Evolution 21, 621–628. rate from one- and two-locus identity measures. Genetics 157, 911–925.

Shumway, D.L., Koide, R.T., 1995. Size and reproductive inequality in mycorrhizal Waples, R.S., 2002. Evaluating the effect of stage-specific survivorship on the Ne/N and nonmycorrhizal populations of Abutilon-theophrasti. Journal of Ecology 83, ratio. Molecular Ecology 11, 1029–1037. 613–620. Weiner, J., Thomas, S.C., 1986. Size variability and competition in plant Siol, M., Bonnin, I., Oliveri, I., Prosperi, J.M., Ronfort, J., 2007. Effective population monocultures. Oikos 47, 211–222. size associated with self-fertilization: lessons from temporal changes in allele Willi, Y., Van Buskirk, J., Schmid, B., Fischer, M., 2007. Genetic isolation of frequencies in the selfing annual Medicago truncatula. Journal of Evolutionary fragmented populations is exacerbated by drift and selection. Journal of Biology 20, 2349–2360. 20, 534–542. Tremblay, R.L., Ackerman, J.D., 2001. Gene flow and effective population size in Willi, Y., Hoffman, A.A., 2009. Demographic factors and genetic variation influence Lepanthes (Orchidaceae): a case for genetic drift. Biological Journal of the population persistence under environmental change. Journal of Evolutionary Linnean Society 72, 47–62. Biology 22, 124–133. Van Kleunen, M., Fischer, M., Schmid, B., 2001. Effects of intraspecific competition Yonezawa, K., Ishii, T., Nomura, T., Morishima, H., 1996. Effectiveness of some on size variation and reproductive allocation in a clonal plant. Oikos 94, 515– management procedures for seed regeneration of plant genetic resources 524. accessions. Genetic Research in Crop Evolution 43, 517–524. Van Kleunen, M., Fischer, M., Schmid, B., 2005. Three generations of low versus high Zietsman, J., Dreyer, L.L., Esler, K.J., 2008. Reproductive biology and ecology of neighborhood density affect the life history of a clonal plant through differential selected rare and endangered Oxalis L. (Oxalidaceae) plant species. Biological selection and genetic drift. Oikos 108, 573–581. Conservation 141, 1475–1483. Vencovsky, R., Crossa, J., 2003. Measurements of representativeness used in genetic resources conservation and plant breeding. Crop Science 43, 1912–1921.