• We are sorry, but NCBI web applications do not support your browser and may not function properly. More information
Logo of geneticsGeneticsCurrent IssueInformation for AuthorsEditorial BoardSubscribeSubmit a Manuscript
Genetics. Sep 2004; 168(1): 425–434.
PMCID: PMC1448113

Population Genetic Evidence for Rapid Changes in Intraspecific Diversity and Allelic Cycling of a Specialist Defense Gene in Zea


Two patterns of plant defense gene evolution are emerging from molecular population genetic surveys. One is that specialist defenses experience stronger selection than generalist defenses. The second is that specialist defenses are more likely to be subject to balancing selection, i.e., evolve in a manner consistent with balanced-polymorphism or trench-warfare models of host-parasite coevolution. Because most of the data of specialist defenses come from Arabidopsis thaliana, we examined the genetic diversity and evolutionary history of three defense genes in two outcrossing species, the autotetraploid Zea perennis and its most closely related extant relative the diploid Z. diploperennis. Intraspecific diversity at two generalist defenses, the protease inhibitors wip1 and mpi, were consistent with a neutral model. Like previously studied genes in these taxa, wip1 and mpi harbored similar levels of diversity in Z. diploperennis and Z. perennis. In contrast, the specialist defense hm2 showed strong although distinctly different departures from a neutral model in the two species. Z. diploperennis appears to have experienced a strong and recent selective sweep. Using a rejection-sampling coalescent method, we estimate the strength of selection on Z. diploperennis hm2 to be ~3.0%, which is approximately equal to the strength of selection on tb1 during maize domestication. Z. perennis hm2 harbors three highly diverged alleles, two of which are found at high frequency. The distinctly different patterns of diversity may be due to differences in the phase of host-parasite coevolutionary cycles, although higher hm2 diversity in Z. perennis may also reflect reduced efficacy of selection in the autotetraploid relative to its diploid relative.

MOLECULAR population genetic surveys of plant defense genes are providing novel insight into the evolutionary history of plant defenses and, by extension, plant-enemy interactions (reviewed in DeMeaux and Mitchell-Olds 2003). On the basis of the limited number of genes studied to date, defenses active against one or perhaps few enemies, i.e., specialist defenses, appear on average to experience stronger selection than defenses active or potentially active against a broad array of enemies, i.e., generalist defenses. Four specialist defenses, including three NBS-LRR genes from Arabidopsis thaliana (Rpm1, Rps2, Rps5, and Rpp13; Caicedo et al. 1999; Stahl et al. 1999; Tian et al. 2002; Mauricio et al. 2003; Rose et al. 2004) and detoxifying enzyme hm2 in Zea mays ssp. parviglumis (Zhang et al. 2002) have patterns of intraspecific diversity inconsistent with expectations under a neutral model. Only one specialist defense, hm1 in Z. mays ssp. parviglumis, has a pattern of intraspecific diversity consistent with expectations under neutrality. In contrast, the majority of putative generalist defenses surveyed have patterns of intraspecific diversity consistent with a neutral model [chitinases A. thaliana chiB (Kawabe and Miyashita 1999); Z. mays ssp. parviglumis chiA, chiB, and chiI; and Z. diploperennis chiB and chiI (Tiffin 2004) and proteinase inhibitors A. thaliana Atti1, A. lyrata Atti1 and Atti2 (Clauss and Mitchell-Olds 2003), and Zea wip1 (Tiffin and Gaut 2001a)] while three show evidence of recent positive selection [A. thaliana Atti2 (Clauss and Mitchell-Olds 2003) and possibly chiA (Kawabe et al. 1997; Kawabe and Miyashita 2002) and Z. diploperennis chiA (Tiffin 2004)].

A second pattern that appears to be emerging from molecular population genetic surveys of plant defense genes is that selection is more likely to maintain diversity at specialist than at generalist defense genes. Four of the five specialist defense genes, A. thaliana Rpm1, Rps2, Rps5, and Rpp13 show evidence for selection having maintained diversity in populations—i.e., the genes have been under balancing selection (Caicedo et al. 1999; Stahl et al. 1999; Tian et al. 2002; Mauricio et al. 2003; Rose et al. 2004). In contrast, only a single generalist defense, A. thaliana mam2 at the GSL-ELONG locus, shows evidence of balancing selection (Kroymann et al. 2003). Evidence for balancing selection in specialist but not generalist defenses suggests that specialist defenses are more likely to evolve in a manner consistent with balanced-polymorphism or trench-warfare models of defense (reviewed in Bergelson et al. 2001). These models predict that functionally distinct defense alleles will be maintained in a population by some form of frequency-dependent selection.

The emerging pattern of balancing selection maintaining diversity at specialist but not generalist defenses must be viewed with caution given that the genes showing evidence for balancing selection come from only one species, inbreeding A. thaliana. The identification of balancing selection, as opposed to selective sweeps, in A. thaliana may be favored for two reasons. First, polymorphisms in A. thaliana are biased toward low frequency (e.g., Purugganan and Suddith 1999; Wright et al. 2003), perhaps due to selfing and population subdivision (Sharbel et al. 2000). Low-frequency variants are also apparent after recovery from selective sweeps (Tajima 1989), and hence it may be difficult to distinguish demographic effects from recent selective sweeps in this system (Clauss and Mitchell-Olds 2003). Second, low effective recombination rates in selfers ensures that sites under balancing selection are in linkage disequilibrium with sites in close physical proximity (Nordborg et al. 2002), leading to a pronounced peak of variation around the site under balancing selection (Kreitman and Hudson 1991; Tian et al. 2002) that may be relatively easy to detect.

The emerging pattern of stronger balancing selection at specialist defenses also may be biased by the selection of genes that have been surveyed. The four specialist defenses surveyed in A. thaliana (Rpm1, Rps2, Rps5, and Rpp13) were all identified because they harbor resistant and susceptible genotypes in contemporary populations (Kunkel 1996; Bittner-Eddy et al. 1999). Loci that experience recent selective sweeps are much less likely to exhibit phenotypically detectible polymorphisms. The GSL-ELONG locus, the one generalist defense to harbor a pattern of balancing selection, was also investigated because it was identified as a candidate gene for a QTL associated with intraspecific variation in glucosinolate production (de Quiros et al. 2000). Because QTL will be detectable only if parental lines have functionally distinct alleles, molecular population genetic analyses of QTL candidates will be biased toward finding evidence of genes under balancing selection.

The only specialist defenses studied from a taxon other than A. thaliana are the Zea hm1 and hm2 genes. Hm1 and hm2 code for nitrate reductases that inhibit the HC toxin produced by the fungal pathogen Cochliobolus carbonum (syn. Helminthosporium maydis), thereby protecting plants from infection (Johal and Briggs 1992; Multani et al. 1998). Work in contemporary populations of Z. mays ssp. mays (maize) indicates that hm1 is primarily responsible for defense against C. carbonum, although hm2 also confers partial resistance (Nelson and Ullstrup 1964). Molecular population genetic analyses of diversity in maize and the teosinte Z. mays ssp. parviglumis indicate that hm2 is the likely target of an ongoing selective sweep in this species (Zhang et al. 2002). Because this apparent sweep is ongoing, it is unclear if the positively selected allele will go to fixation, as predicted by classic arms-race models, or whether the apparent selective advantage associated with this allele will weaken as the allele increases in frequency, as predicted by dynamic-polymorphism/trench-warfare models.

Surveying hm2 diversity in taxa closely related to Z. mays ssp. parviglumis may provide further insight into the evolutionary dynamics of this specialist defense. If hm2 is experiencing similar evolutionary dynamics in closely related species, then sampling closely related species may provide a snapshot of this dynamic at different points in a coevolutionary cycle or selective sweep. Of course, hm2 in closely related taxa may be experiencing different coevolutionary dynamics, but this too would be informative, providing insight into the variation in selection experienced by defense genes in closely related taxa. Two taxa, Z. diploperennis and Z. perennis, are particularly promising for a comparative population genetic survey of hm2. These species are morphologically similar, perennial species with restricted geographic distributions in the state of Jalisco in southwestern Mexico (Doebley 1990; Sanchez et al. 1998). However, these taxa are clearly distinct because Z. diploperennis is diploid and Z. perennis is a polyploid, which morphological and molecular data indicate originated from a Z. diploperennis-like progenitor (Iltis et al. 1979; Doebley et al. 1987; Buckler and Holtsford 1996; Tiffin and Gaut 2001b). Because these species have similar geographic distributions and life histories they may be likely to be exposed to similar pathogen pressures. Like other Zea species, all of which are native to Mexico and Central America, Z. diploperennis and Z. perennis are both wind-pollinated outcrossing taxa. Comparing hm2 diversity in Z. perennis and Z. diploperennis is also facilitated by these species having similar patterns of DNA diversity at apparently neutral genes (Tiffin and Gaut 2001b).

The primary objective of this research is to examine the evolutionary history of hm2 in two closely related species Z. diploperennis and Z. perennis, to gain a greater understanding of the long-term evolution of plant defense. We were particularly interested in determining if hm2 would provide support for the apparent pattern that specialist defenses are more likely to experience stronger selection than generalist defenses and whether the pattern of selection acting on specialist defenses is more likely to be balancing than positive. Examining intraspecific diversity in closely related taxa also provides an opportunity to examine whether defense genes have similar evolutionary dynamics in closely related taxa.

Because we are interested in comparing the selective histories of specialist and generalist defenses we also investigated intraspecific diversity at two putative generalist defenses, the protease inhibitors mpi (maize protease inhibitor) and wip1 (wound induced protein). Evidence for plant protease inhibitors having a primary role in defense against herbivores and pathogens (Ryan 1990) includes induction following pathogen infection and/or herbivore damage (Rohrmeier and Lehle 1993; Cordero et al. 1994; Tamayo et al. 2000), reduced growth and reproduction of some herbivores and pathogens when fed a diet or grown on medium that contains protease inhibitors (Tamayo et al. 2000), and increased resistance in plants expressing transgenic protease inhibitors (Hilder et al. 1987; Johnson et al. 1989). Because proteases are digestive enzymes present in most if not all herbivore guts, are secreted by many parasites, and are encoded by many viruses, protease inhibitors are potentially active against a broad array of enemies. No defense genes have been previously investigated in Z. perennis and only chitinase genes had been previously investigated in Z. diploperennis (Tiffin 2004).


Sampling DNA sequences:

We PCR amplified ~630 bp of mpi, 660 bp of wip1, and 1450 bp of hm2 from eight accessions of Z. diploperennis, six accessions of Z. perennis, and one accession of Tripsacum dactyloides (see appendix). PCR conditions for mpi were 35 cycles of 1 min at 94°, 1 min at 50°, and 2 min at 72°; conditions for hm2 were similar except the annealing temperature was 55° and 1 m betaine (N,N,N-trimethylglycine, Sigma, St. Louis) was added to each reaction. Wip1 alleles, including the entire 280-bp coding region, a 90-bp intron, and 290 bp of flanking sequence, were amplified using primers and conditions described previously (Tiffin and Gaut 2001a). Mpi primers (F, ctgcagtgttgctatctgttc; R, attagtgagaattcacacatcc) amplified the entire coding region (220 bp in Zea) and ~280 and 60 bp of upstream and downstream flanking regions, respectively. hm2 primers (F, tagcagtgaagtgcaggtg; R, attatgagacatggctggag) amplified a region extending from 12 bp 3′ of the atg start site to ~250 bp 3′ of the predicted stop codon. This region extends ~100 bp 5′ and 300–425 bp 3′ (depending on indels) beyond the region investigated by Zhang et al. (2002). Oryza sativa and Sorghum bicolor sequences used in relative rate tests were obtained from GenBank [mpi: AC079022, O. sativa (genomic) and BE917718, S. bicolor (EST); wip1: AP002526, O. sativa and AW680689, S. bicolor]. Sequences new to this study have been submitted to GenBank (mpi, AY549598, AY549599, AY549600, AY549601, AY549602, AY549603, AY549604, AY549605, AY549606, AY549607, AY549608, AY549609, AY549610, AY549611, AY549612, AY549613, AY549614, AY549615, AY549616, AY549617, AY549618, AY549619, AY549620, AY549621, AY549622, AY549623, AY549624, AY549625, AY549626, AY549627; wip1, AY52550, AY52551, AY52552, AY52553, AY52554, AY52555, AY52556, AY52557, AY52558, AY52559 and AY549628, AY549629, AY549630, AY549631, AY549632, AY549633, AY549634, AY549635, AY549636, AY549637, AY549638; and hm2, AY320258–320280; appendix).

Analyses of evolutionary history are sensitive to the frequency of segregating sites, especially unique single-base-pair variants (singletons). To ensure that all singletons in our data set represented true variants and did not result from misincorporation of a nucleic acid during Taq amplification, all DNAs that yielded alleles with singletons were used as templates in one or more subsequent PCRs (a minimum of two for tetraploids) and the products of those reactions were cloned and sequenced. For the tetraploids, at least six clones from each reaction were sequenced. If the tetraploid contained four distinct alleles, sampling 12 isolates results in >95% probability of resampling the allele that initially contained the singleton, assuming no amplification bias (Tiffin and Gaut 2001b). Singletons not confirmed by this approach were assumed to have resulted from polymerase error and were excluded from the analyses.

Sequence analyses:

Two measures of genetic diversity, the number of segregating sites (S) and Watterson's θ (Watterson 1975), were calculated separately on silent (synonymous and intron) and replacement sites. To determine if the diversity estimates for the defense genes that were the subject of this investigation were different from diversity at apparently neutral genes, we calculated maximum likelihood multilocus estimates of θ using the method of Wright et al. (2003) and data from previously investigated genes (adh1, c1, glb1, and waxy; Tiffin and Gaut 2001a). This method assumes no intragenic recombination, free recombination between loci, and a constant mutation rate. We also performed several tests of neutral evolution including Tajima's D (Tajima 1989), Fu's Fs (Fu 1997), McDonald-Kreitman (MK; McDonald and Kreitman 1991), and HKA (Hudson et al. 1987). For wip1 and mpi, MK and HKA tests were conducted with a T. dactyloides sequence as an outgroup. Despite repeated attempts, we were unable to obtain an hm2 sequence from T. dactyloides and therefore MK and HKA tests on hm2 samples were conducted using the ssp. parviglumis sequences analyzed by Zhang et al. (2002) as an outgroup. The significance of MK tests was evaluated using a G-test with a William's correction. Confidence intervals (95%) around estimates of θ ([theta w/ hat]), the probability of obtaining the number of sampled haplotypes (H), and the significance of Fs were estimated by running 1000 coalescent simulations of the neutral model. Simulations based on either S or [theta w/ hat] produced similar results and only results from simulations run with fixed S are shown. To be conservative in determining confidence intervals and testing for departures from a neutral model, all simulations were run with no recombination. Measures of polymorphism, tests of neutral evolution, and coalescent simulations were calculated using DnaSP v.3.53 (Rozas and Rozas 1999). Relative rates tests were conducted using the method of Fitch (1976) and Tajima (1993) as implemented by MEGA2 (Kumar et al. 2000).

To estimate the selection coefficient, s, and fixation time, T, measured in Ne generations, we used a rejection-sampling method based on selective sweeps simulated using the coalescent model of Przeworski (2003). This method employs three summary statistics estimated from the data (Tajima's D, S, and H) as well as assumed values of the mean mutation rate μ: the mean recombination rate per base pair c; the distance in base pairs, d, between the sequenced region and the site under selection; and the mean diploid population size N. In brief, this method simulates selective sweeps and then samples from the posterior distribution of T and s to obtain a sample that is consistent with the summary statistics calculated from the data. We assumed μ = 6.5 × 10−9 mutations/site/year (Gaut et al. 1996), c = 4 × 10−7 (Wang et al. 1999), and d = 1 or 1000 and calculated N from the multilocus estimate of θ (= 4Nμ) based on neutral loci (Figure 1). Because these estimates, N, c, and μ, may be imprecise, uncertainty in their values is modeled by sampling from prior distributions of these parameters, as per Przeworski (2003).

Figure 1.
Maximum likelihood estimates of θ, derived from apparently neutrally evolving genes in Z. diploperennis and Z. perennis. Arrows indicate the estimates of θ at Hm2, which was not used in making the likelihood estimates. The horizontal line ...


Genetic diversity in mpi and wip1:

For mpi [theta w/ hat]'s fall into the 95% credibility interval (CI) of the multilocus likelihood [theta w/ hat] calculated using data from four other apparently neutrally evolving genes, adh1, c1, glb1, and waxy (Figure 1). Moreover, tests of nonneutral evolution that rely on intraspecific diversity—Fu's Fs and Tajima's D—were not significant when applied to mpi data (Table 1). In contrast, [theta w/ hat] in Z. diploperennis wip1 is slightly higher than the 95% CI of the multilocus likelihood estimate. Nonetheless, Fu's Fs and Tajima's D tests were not significant when applied to wip1 (Table 1), indicating that wip1 has not experienced a recent selective sweep or been subject to balancing selection.

Number of alleles (N), haplotypes (H), and segregating sites (S), measures of genetic diversity calculated on silent and replacement (θN) sites, and results from Tajima'sD tests

Results from relative rates and MK tests are also consistent with mpi having evolved neutrally (all P > 0.45). Similarly, MK tests conducted on wip1 data are consistent with a neutral-equilibrium model (P > 0.05). In contrast, relative rates tests conducted on wip1 sequences were significant for both species (P < 0.01 for all sequences), indicating rapid evolutionary change within the lineage leading to Zea. The significant relative rates tests are consistent with wip1 having evolved in response to previous episodes of positive selection (Tiffin and Gaut 2001a), although the evidence of selection is no longer evident in the frequency spectrum of intraspecific polymorphisms. Consistent with earlier analyses of apparently neutrally evolving loci (Tiffin and Gaut 2001b), [theta w/ hat]'s for wip1 and mpi do not differ significantly between Z. diploperennis and Z. perennis (Table 1).

Natural selection on hm2:

In contrast to the mpi and wip1 data, patterns of hm2 intraspecific diversity are inconsistent with neutrality. In Z. diploperennis, seven of the sampled alleles are identical and four others differ from these seven at only a single site (Figure 2). Moreover, hm2 diversity in Z. diploperennis is an order of magnitude lower than that in any other Z. diploperennis gene sampled to date (Figure 1) and lower than diversity at any other gene within any of the teosintes (summarized in Zhang et al. 2002), with the exception of a Z. diploperennis chitinase gene (chiA) that appears to have been subject to strong positive selection (Tiffin 2004).

Figure 2.
The hm2 exon-intron structure and sequences from Z. diploperennis and Z. perennis. The shaded regions represent exons, the position of polymorphic sites is shown at the top, and dashes indicate indels.

HKA tests on Z. diploperennis hm2 data, conducted with adh1, c1, glb1, waxy, mpi, and wip1 as the second locus and ssp. parviglumis as an outgroup, all had P-values <0.07, and all but two were significant at P < 0.03. In contrast, no HKA test among the other genes was significant (all P > 0.3). These results suggest that hm2 or a closely linked region has evolved in response to selection within Z. diploperennis. Other tests of nonneutral evolution [Tajima's D, Fu's Fs, (Table 1), and Fay and Wu's H (Fay and Wu 2000); data not shown] were not significant but this may not be surprising given that these tests rely on the frequency of segregating sites to identify departures from a neutral model. Although we sequenced >1450 bp for each of 11 hm2 alleles, we detected only three segregating sites, providing little power for rejecting a neutral model.

The pattern of hm2 polymorphism in Z. perennis is distinctly different from the pattern in Z. diploperennis but also appears indicative of nonneutral evolution. In contrast to Z. diploperennis, in which there were only 3 segregating sites among 11 sampled alleles, the 12 alleles sampled from Z. perennis contained 38 segregating sites distributed among 3 haplotypes (excluding two apparent recombinants per3b and per3c, Figure 2). The probability of ≤3 haplotypes in a sample with 38 segregating sites is extremely low under the standard neutral model (P < 0.002), and the result holds when the two recombinant haplotypes are included (P < 0.02 of ≤5 haplotypes). In contrast, haplotype numbers at six other genes sampled from the same collection of DNAs were consistent with neutral expectations (all P > 0.25). Interestingly, the coding regions of the two main haplotypes differ from one another primarily at replacement sites, suggesting that the haplotypes may be functionally diverged; although, dN:dS is not significantly >1 (8 replacement vs. 1 synonymous difference, dN:dS = 2.8, P = 0.27).

Although this pattern of diversity is suggestive of nonneutral history, Tajima's D was not significant when calculated on the entire Z. perennis hm2 data set (Table 1). However, one of the three haplotypes is represented by only a single sequence (per7b, Figure 2), resulting in a high proportion of singletons, which could strongly affect tests of neutrality. Eliminating the third haplotype from the data eliminated all singletons and resulted in a substantially lower [theta w/ hat] and significantly positive values of D (Table 1) and Fs (Fs = 8.1, P < 0.001). Fs was also significant when all sequences were included (Fs = 6.97, P < 0.01) although HKA tests were not significant when conducted on either full or reduced Z. perennis data sets.

Two aspects of the hm2 data indicate this gene has diverged between taxa in an atypical manner. First, 95% confidence intervals around [theta w/ hat] in Z. diploperennis and Z. perennis do not overlap (Table 1), unlike other genes from these taxa (Figure 1; Table 1). Second, hm2 appears to be responsible for significant heterogeneity in the distribution of fixed differences and shared polymorphisms among loci (G = 36.2; P < 0.001, Table 2). To determine which locus was responsible for the significant result, we jackknifed contingency tests, removing one locus at a time. Six of these tests were significant at P < 0.05 (all remained significant after a Bonferroni correction), but the hypothesis of homogeneity was not rejected when hm2 was removed (P > 0.2).

Number of sites fixed and polymorphisms shared betweenZ. diploperennis andZ. perennis

Simulations of selection on hm2:

Statistical tests indicate that hm2 has evolved nonneutrally in Zea taxa and the pattern of diversity at hm2 in Z. diploperennis appears to be consistent with a recent selective sweep. To estimate the strength of selection that acted on hm2 during this apparent sweep, as well as the time when the sweep occurred, we examined the posterior distributions of these parameters (T and s, the selection coefficient) that are consistent with the pattern of diversity we found in our sample. Assuming that the distance between the sequenced region and the site under selection equaled one base, a sample from the posterior distribution of T clearly favors a recent selective sweep in hm2 (Figure 3), with the mode of the distribution T < 0.05 and 77.9% of the support on T < 0.2 a value that, following Przeworski (2003), we used as an arbitrary cutoff value consistent with a strong selective sweep. In contrast, when this method was used on an apparently neutral gene, adh1, only 3.4% of the support was on T < 0.2, and the mode of the distribution was T > 0.9—consistent with no sweep or a very old sweep that is no longer evident in a sample of alleles from present day populations. In addition to providing clear evidence for a strong selective sweep in Z. diploperennis hm2, these analyses provide a basis for estimating the strength of selection (ŝ) that acted on hm2. From the joint posterior distribution of s and T, mean s is 0.032 for hm2 T < 0.2, suggesting the selection coefficient during the hm2 sweep was ~3%. Similar results were obtained when we assumed that d = 1000 (hm2, 65.6% support for T < 0.2, ŝ = 0.035; adh, 9.0% support for T < 0.2).

Figure 3.
Results of selective sweep simulations and rejection sampling. (A) Sample from the posterior distribution of T, the time of the fixation event, based on hm2 and adh1 data (based on 10,000 and 6781 simulations, respectively). (B) The joint posterior distribution ...

We also applied the rejection-sampling method to Z. perennis hm2 data. Ten days of computer time produced no posterior data that were consistent with summaries of Z. perennis hm2 data. (In contrast, hundreds of data sets were simulated for the Z. diploperennis data in 1–2 days.) These results suggest that the Z. perennis hm2 data do not fit a selective sweep model for any values of s and T. Moreover, because a sample taken from a neutral locus should, on average, have values of T = 1 and s = 0, the inability of the algorithm to produce any posterior data consistent with the pattern of diversity we found in Z. perennis suggests that these results are also inconsistent with the neutral model.


In this study we investigated the evolutionary history of three plant defense genes, one a specialist defense active against the fungal pathogen C. carbonum and two that are potentially active against a wide array of plant enemies. The two generalist defenses (wip1 and mpi) show no evidence of having evolved in response to recent selection, and estimates of diversity at these genes fall within the range of diversity found at other presumably neutrally evolving loci in Z. diploperennis and Z. perennis (Figure 1). These data also corroborate our earlier finding that these two species harbor similar levels of genetic diversity (θ) at loci with patterns of intraspecific diversity that are consistent with a neutral equilibrium model (Tiffin and Gaut 2001b). It should be noted, however, that the expectation of θ is 4Nμ for diploids, where N is population size and μ is mutation rate, but 8Nμ for tetraploids. Thus, although θ is similar between the two taxa, these estimates suggest that the long-term effective population size of the tetraploid is roughly half that of the most closely related extant diploid. Nonetheless, there is little indication, in either the level of sequence diversity or the values of Tajima's D, that tetraploid formation included a severe bottleneck.

In contrast to the generalist defenses, the specialist defense hm2 appears to have evolved in response to powerful and recent selection in both Z. diploperennis and Z. perennis. This difference is consistent with an emerging pattern of specialist plant defenses showing stronger evidence for selection than generalist defenses. The patterns of diversity and evidence for selection are, however, distinctly different in the two species. In Z. diploperennis, hm2 appears to have experienced a recent selective sweep as evident by extremely low diversity (Figure 1; significant HKA tests and coalescent simulations, see Figure 3). The selection that acted on hm2 in Z. diploperennis appears to have been strong with an estimated selection coefficient, s, of ~3%, assuming conservatively that the selected site is very near the sequenced region. To corroborate the estimate of s obtained using Przeworski's (2003) rejection-sampling method, we also estimated s using the equation ŝ = 100dc (Kaplan et al. 1989). Assuming c = 4 × 10−7 (Wang et al. 1999), ŝ ranges from 0.029 to 0.058, depending on whether the selective site is assumed to be in the middle (d = 725) or the end (d = 1450) of the sequenced region. Thus, lower-range estimates of ŝ from both methods are ~0.03. Moreover, both methods produce estimates of ŝ that are similar to estimates of selection on tb1 (Wang et al. 1999), a gene responsible for major differences in plant architecture that differentiate maize from the teosinte Z. mays ssp. parviglumis (Doebley et al. 1995, 1997; Clark et al. 2004), that were made using the equation of Kaplan et al. (1989). Although estimates for both genes are based on many assumptions that are difficult to verify, this comparison makes the important point that selection on disease resistance genes in the wild can be on the order of the strength of selection on artificially selected genes like tb1.

In contrast to Z. diploperennis hm2, which appears to have experienced a recent and strong sweep, diversity at hm2 in Z. perennis is characterized by the presence of three alleles, each of which differs from alleles in the other classes by at least 27 sites (~2%). The distinct alleles segregating at hm2 as well as positive values of Tajima's D and Fu's Fs and incompatibility with both selective sweep and neutral simulations (see results) could indicate a long-lived polymorphism in Z. perennis. The diversity at Z. perennis hm2 is not, however, consistent with expectations of a balanced polymorphism in which allele frequencies have been stable. Simulation studies (Innan and Tajima 1999) show that when two haplotypes are maintained at stable frequencies through balancing selection the sum of the within-haplotype [theta w/ hat]'s is expected to be equal to θ estimated using all data. We find, however, that the sum of the within-class [theta w/ hat] = 0, well below the 95% confidence intervals around [theta w/ hat] calculated from the Z. perennis data. The lack of intrahaplotype polymorphism suggests that the two main allelic classes have not been maintained as a long-term stable polymorphism but rather that both alleles have recently increased in frequency due to positive selection. Positive selection could also explain the comparatively high interspecific divergence of hm2, relative to other loci (Table 2). It may be possible that the Z. perennis data also reflect population structuring but this seems unlikely given that Z. perennis is wind pollinated with a geographically limited range (Sanchez et al. 1998). Moreover, the six other loci we have examined from this same collection of Z. perennis DNAs reveal no obvious patterns that indicate population structure.

Because Z. diploperennis and Z. perennis are closely related with similar life histories and geographic distributions it seems likely that the distinct patterns of diversity found at hm2 in these species reflect a common underlying host-parasite coevolutionary processes—with the two species being at different phases of coevolutionary cycles. These patterns appear inconsistent, however, with two-allele models that predict regular cyclic fluctuations in gene frequencies (Jayakar 1970; Seger 1988, 1990; Stahl et al. 1999). These data do, however, appear consistent with three-allele models that exhibit highly irregular fluctuations in gene frequencies (Seger 1988) and may more appropriately describe allelic variation at many defense genes in natural populations. In Z. diploperennis, hm2 may have low diversity because it is at the peak of a cycle whereas an apparently long-lived polymorphism may be detected in Z. perennis because hm2 is experiencing simultaneous increase of two alleles (Figure 4). The apparent ongoing sweep detected at hm2 in Z. mays ssp. parviglumis may also be consistent with these more complicated models of host-parasite coevolution.

Figure 4.
Hypothetical dynamics of three defense alleles coevolving with parasite avirulence alleles (after Seger 1990). Hm2 may exhibit very low diversity in Z. diploperrenis because it has been the object of a recent selective sweep (up arrow). Two common alleles ...

This explanation for the distinctly different patterns of diversity found in Z. diploperennis and Z. perennis is clearly speculative. Moreover, many patterns of diversity are consistent with these three-allele coevolutionary models, depending on the strength of selection, the amplitude of cycles, and the phase of a cycle from which alleles were sampled. As such, it is unclear what data would provide definitive evidence against these models. We note that other dynamic-polymorphism/trench-warfare models suffer similar drawbacks. Nevertheless, if these three-allele models correctly describe the evolutionary dynamics of defense genes we might expect that other defense genes will also harbor distinctly different patterns of diversity in closely related species or isolated populations, just as we have documented here. Sequence polymorphism at the recently sampled A. thaliana RPP13 locus is also inconsistent with simple two-allele models of balancing selection, but does appear consistent with multiallele frequency-dependent selection (Rose et al. 2004).

One caveat in interpreting our data is that the higher diversity in Z. perennis may be related to beneficial alleles spreading more slowly through autopolyploid than through diploid species. Theoretical models show that the number of alleles that segregate at meiosis affects the rate at which selectively favored alleles spread through a population (Hill 1971; Otto and Whitton 2000). Z. perennis is an autotetraploid and thus segregates four rather than two alleles per locus. Therefore, beneficial alleles will be slower to fix in the autotetraploid Z. perennis and, at loci that experience strong positive selection, this species may harbor greater diversity than the diploid Z. diploperennis (assuming other things are equal, such as population size). The hm2 data showing higher diversity in Z. perennis than in Z. diploperennis are consistent with the prediction that autotetraploids “mask” beneficial alleles from selection. Nevertheless, under a simple masking model there is no reason to expect that the allele sweeping through the population would be highly diverged from other alleles found within the species—the selectively advantageous allele might differ from other alleles only at recent mutations that altered the fitness effects. Moreover, the alleles that are masked from selection would be expected to be under relaxed selective constraint and therefore accumulate mutations. Neither of these expectations is met in Z. perennis in which each of the three haplotypes differ from one another at ~2% of sites, there are no polymorphic sites within either haplotype that is represented by more than a single sequence, and the relatively high dN:dS between the two main haplotypes suggests that they may be diverged functionally. Given the inconsistencies between our data and the expectations of tetravalent inheritance slowing the spread of positively selected alleles, we think selective pressure imposed by the parasite against which hm2 is active is at least partially responsible for the patterns of diversity we see at this locus. Unfortunately direct tests of this hypothesis may be difficult, given that the parasite genotypes potentially responsible for past selection on hm2 may no longer be common or even present in contemporary populations.

We know of only two other studies that have examined defense gene diversity in closely related taxa; Clauss and Mitchell-Olds (2003) examined intraspecific diversity of duplicated trypsin inhibitor genes in the closely related A. thaliana and A. lyrata ssp. petraea and Tiffin (2004) examined diversity of three chitinase genes in Z. diploperennis and Z. mays ssp. parviglumis. Both of these studies revealed evidence for interspecific differences in evolutionary histories; in particular, both detected evidence for positive selection in one taxon, but neutral patterns of diversity in the second taxon. Unlike Z. diploperennis and Z. perennis, the taxa investigated in Clauss and Mitchell-Olds (2003) and Tiffin (2004) differ in life history traits and geographic ranges and it is therefore unclear whether interspecific differences in diversity found in those taxa are due to selective forces acting on the defense genes or due to demographic forces with genome-wide effects. Here we show that defense genes in closely related taxa with similar life histories and geographic ranges may have substantially different evolutionary histories and levels of sequence diversity. Moreover, the hm2 data suggest that simple two-allele models are unlikely to adequately capture the evolution of all defense genes in natural populations.

Regardless of the evolutionary mechanisms responsible for the higher hm2 diversity in Z. perennis, if greater diversity at functionally important resistance genes is a general phenomenon in polyploids, then this may, in part, explain ecological observations that tetraploids are often more resistant to pathogens and herbivores than are their diploid relatives (Levin 1983; Nuismer and Thompson 2001).


Source and GenBank accession numbers forZ. diploperennis andZ. perennis sequences that are new to this study

Z. diploperennis
AY549598, AY549599AY52550AY320270 (1)
 9476AY549600AY52551AY320271 (2)
 10003AY549601AY52552, AY52553AY320272, AY300273 (3a,b)
 Ames 2317AY549602AY52554AY320274 (4)
 PI 441932AY549603, AY549604AY52555, AY52556AY320275 (5)
 PI 462368AY549605, AY549606AY52557AY320276, AY300277 (6a,b)
 Ames 21884AY549607, AY549608AY52558AY320278 (7)
 PI 441931AY549609, AY549610AY52559AY320279, AY300280 (8a,b)
Z. perennis
 Ames 21869AY549611, AY549612, AY549613AY320258 (1)
 Ames 21870AY549614, AY549615,AY549629, AY549630AY320259 (2)
 Ames 21873AY549617, AY549618,AY549632, AY549633AY320260, AY320261 (3a,b)
AY549619AY320262, AY320263 (3c,d)
 Ames 21874AY549620, AY549621AY320264, AY320265 (4a,b)
AY549622, AY549623
 Jal-88AY549624AY549634, AY549635, AY549636AY320268, AY320269 (6a,b)
 Mo10AY549625, AY549626AY549637, AY549638AY32026, AY320267 (7a,b)

hm2 sample numbers used in Figure 2 are in parentheses following the GenBank number.


We thank Molly Przeworski for her generous help with estimating s and T; Eli Stahl, Deborah Charlesworth, and an anonymous reviewer for comments that improved the manuscript; and John Doebley for providing seed that made this project possible. This research was supported by a U.S. Department of Agriculture Award 99-35301-8076 to P.T. and National Science Foundation grants DEB 9815855 to B.S.G. and DEB 0235027 to P.T.


Sequence data from this article have been deposited with the EMBL/GenBank Data Libraries under accession nos. AY320258–320280, AY52550, AY52551, AY52552, AY52553, AY52554, AY52555, AY52556, AY52557, AY52558, AY52559, and AY549598, AY549599, AY549600, AY549601, AY549602, AY549603, AY549604, AY549605, AY549606, AY549607, AY549608, AY549609, AY549610, AY549611, AY549612, AY549613, AY549614, AY549615, AY549616, AY549617, AY549618, AY549619, AY549620, AY549621, AY549622, AY549623, AY549624, AY549625, AY549626, AY549627, AY549628, AY549629, AY549630, AY549631, AY549632, AY549633, AY549634, AY549635, AY549636, AY549637, AY549638.


  • Bergelson, J., G. Dwyer and J. J. Emerson, 2001. Models and data on plant-enemy coevolution. Annu. Rev. Genet. 35: 469–499. [PubMed]
  • Bittner-Eddy, P., C. Can, N. Gunn, M. Pinel, M. Tor et al., 1999. Genetic and physical mapping of the RPP13 locus in Arabidopsis responsible for specific recognition of several Peronopora parasitca (downy mildew) isolates. Mol. Plant-Microbe Interact. 12: 792–802. [PubMed]
  • Buckler, E. S., and T. P. Holtsford, 1996. Zea systematics: ribosomal ITS evidence. Mol. Biol. Evol. 13: 612–622. [PubMed]
  • Caicedo, A. L., B. A. Schaal and B. A. Kunkel, 1999. Diversity and molecular evolution of the RPS2 resistance gene in Arabidopsis thaliana. Proc. Natl. Acad. Sci. USA 96: 302–306. [PMC free article] [PubMed]
  • Clark, R. M., E. Linton, J. Messing and J. F. Doebley, 2004. Pattern of diversity in the genomic region near the maize domestication gene tb1. Proc. Natl. Acad. Sci. USA 101: 700–707. [PMC free article] [PubMed]
  • Clauss, M. J., and T. Mitchell-Olds, 2003. Population genetics of tandem trypsin inhibitor genes in Arabidopsis species with contrasting ecology and life histories. Mol. Ecol. 12: 1287–1299. [PubMed]
  • Cordero, M. J., D. Raventos and B. San Segundo, 1994. Expression of a maize proteinase inhibitor gene is induced in response to wounding and fungal infection: systemic wound-response of a monocot gene. Plant J. 6: 141–150. [PubMed]
  • de Meaux, J., and T. Mitchell-Olds, 2003. Evolution of plant resistance at the molecular level: ecological context of species interactions. Heredity 94: 343–352. [PubMed]
  • de Quiros, H. C., R. Magrath, D. McCallum, J. Kroymann, D. Scnabelrauch et al., 2000. Alpha-keto acid elongation and glucosinolate biosynthesis in Arabidopsis thaliana. Theor. Appl. Genet. 101: 429–437.
  • Doebley, J., 1990. Molecular evidence and the evolution of maize. Econ. Bot. 44: 6–27.
  • Doebley, J., A. Stec and C. Gustas, 1995. Teosinte branched 1 and the origin of maize: evidence for epistasis and the evolution of dominance. Genetics 141: 333–346. [PMC free article] [PubMed]
  • Doebley, J., A. Stec and L. Hubbard, 1997. The evolution of apical dominance in maize. Nature 386: 485–488. [PubMed]
  • Doebley, J. F., W. Renfroe and A. Blanton, 1987. Restriction site variation in the Zea chloroplast genome. Genetics 117: 139–147. [PMC free article] [PubMed]
  • Fay, J. C., and C.-I Wu, 2000. Hitchhiking under positive Darwinian selection. Genetics 155: 1405–1413. [PMC free article] [PubMed]
  • Fitch, W. M., 1976 Molecular evolutionary clocks, pp. 160–178 in Molecular Evolution, edited by F. J. Ayala. Sinauer Associates, Sunderland, MA.
  • Fu, Y. X., 1997. Statistical tests of neutrality of mutations against population growth, hitchhiking and background selection. Genetics 147: 915–952. [PMC free article] [PubMed]
  • Gaut, B. S., B. R. Morton, B. M. McCaig and M. T. Clegg, 1996. Substitution rate comparisons between grasses and palms: synonymous rate differences at the nuclear gene Adh parallel rate differences at the plastid gene rbcL. Proc. Natl. Acad. Sci. USA 93: 10274–10279. [PMC free article] [PubMed]
  • Hilder, V. A., A. M. R. Gatehouse, S. E. Sheerman, R. F. Barker and D. Boulter, 1987. A novel mechanism of insect resistance engineered into tobacco. Nature 330: 160–163.
  • Hill, R. R., 1971. Selection in autotetraploids. Theor. Appl. Genet. 41: 181–186. [PubMed]
  • Hudson, R. R., M. Kreitman and M. Aguade, 1987. A test of neutral molecular evolution based on nucleotide data. Genetics 116: 153–159. [PMC free article] [PubMed]
  • Iltis, H. H., J. F. Doebley, R. Guzman and B. Pazy, 1979. Zea diploperennis (Gramineae): a new teosinte from Mexico. Science 203: 186–188. [PubMed]
  • Innan, H., and F. Tajima, 1999. The effect of selection on the amounts of nucleotide variation within and between allelic classes. Genet. Res. 73: 15–28.
  • Jayakar, S. D., 1970. A mathematical model for interaction of gene frequencies in a parasite and its host. Theor. Pop. Biol. 1: 140–164. [PubMed]
  • Johal, G. S., and S. P. Briggs, 1992. Reductase activity encoded by the HM1 disease resistance gene in maize. Science 258: 985–987. [PubMed]
  • Johnson, R., J. Narvaez, G. An and C. A. Ryan, 1989. Expression of proteinase inhibitors I and II in transgenic tobacco plants: effects on natural defense against Manduca sexta larvae. Proc. Natl. Acad. Sci. USA 86: 9871–9875. [PMC free article] [PubMed]
  • Kaplan, N. L., R. R. Hudson and C. H. Langley, 1989. The “Hitchiking Effect” revisited. Genetics 123: 887–899. [PMC free article] [PubMed]
  • Kawabe, A., and N. T. Miyashita, 1999. DNA variation in the basic chitinase locus (ChiB) region of the wild plant Arabidopsis thaliana. Genetics 153: 1445–1453. [PMC free article] [PubMed]
  • Kawabe, A., and N. T. Miyashita, 2002. DNA variation in the acidic chitinase locus (ChiA) region in Arabis gemmifera and its related species. Genes Genet. Syst. 77: 167–175. [PubMed]
  • Kawabe, A., H. Innan, R. Terauchi and N. T. Miyashita, 1997. Nucleotide polymorphism in the acidic chitinase focus (ChiA) region of the wild plant Arabidopsis thaliana. Mol. Biol. Evol. 14: 1303–1315. [PubMed]
  • Kreitman, M., and R. R. Hudson, 1991. Inferring the evolutionary histories of Adh and the Adh-dup loci in Drosophila melanogaster from patterns of polymorphism and divergence. Genetics 127: 565–582. [PMC free article] [PubMed]
  • Kroymann, J., S. Donnerhacke, D. Schnabelrauch and T. Mitchell-Olds, 2003. Evolutionary dynamics of an Arabidopsis insect resistance quantitative trait locus. Proc. Natl. Acad. Sci. USA 100: 14587–14592. [PMC free article] [PubMed]
  • Kunkel, B. N., 1996. A useful weed put to work: genetic analysis of disease resistance in Arabidopsis thaliana. Trends Genet. 12: 63–69. [PubMed]
  • Kumar, S., K. Tamura, I. Jakobsen and M. Nei, 2000 MEGA: molecular evolutionary genetics analyses. Pennsylvania State University/Arizona State University, University Park, PA/Tempe, AZ.
  • Levin, D. A., 1983. Polyploidy and novelty in flowering plants. Am. Nat. 122: 1–25.
  • Mauricio, R., E. A. Stahl, T. Korves, D. C. Tian, M. Kreitman et al., 2003. Natural selection for polymorphism in the disease resistance gene Rps2 of Arabidopsis thaliana. Genetics 163: 735–746. [PMC free article] [PubMed]
  • McDonald, J. H., and M. Kreitman, 1991. Adaptive protein evolution at the Adh1 locus in Drosophila. Nature 351: 652–654. [PubMed]
  • Multani, D. S., R. B. Meeley, A. H. Paterson, J. Gray, S. P. Briggs et al., 1998. Plant-pathogen micro evolution: molecular basis for the recent origin of a disease in maize. Proc. Natl. Acad. Sci. USA 95: 1686–1691. [PMC free article] [PubMed]
  • Nelson, O. E., and A. J. Ullstrup, 1964. Resistance to leaf spot in maize. J. Hered. 55: 195–199.
  • Nordborg, M., J. O. Borevitz, J. Bergelson, C. C. Berry, J. Chory et al., 2002. The extent of linkage disequilibrium in Arabidopsis thaliana. Nat. Genet. 30: 190–193. [PubMed]
  • Nuismer, S. L., and J. N. Thompson, 2001. Plant polyploidy and non-uniform effects on insect herbivores. Proc. R. Soc. Lond. Ser. B Biol. Sci. 268: 1937–1940. [PMC free article] [PubMed]
  • Otto, S. P., and J. Whitton, 2000. Polyploid incidence and evolution. Annu. Rev. Genet. 34: 401–437. [PubMed]
  • Przeworski, M., 2003. Estimating the time since the fixation of a beneficial allele. Genetics 164: 1667–1676. [PMC free article] [PubMed]
  • Purugganan, M. D., and J. I. Suddith, 1999. Molecular population genetics of floral homeotic loci: departures from the equilibrium-neutral model at the APETALA3 and PISTILLATA gene of Arabidopsis thaliana. Genetics 151: 839–848. [PMC free article] [PubMed]
  • Rohrmeier, T., and L. Lehle, 1993. WIP1, a wound-inducible gene from maize with homology to Bowman-Birk proteinase inhibitors. Plant Mol. Biol. 22: 783–792. [PubMed]
  • Rose, L. E., P. D. Bittner-Eddy, C. H. Langley, E. B. Holub, R. W. Michemore et al., 2004. Maintenance of extreme amino acid diversity at the disease resistance gene RPP13 in Arabidopsis thaliana. Genetics 166: 1517–1527. [PMC free article] [PubMed]
  • Rozas, J., and R. Rozas, 1999. DnaSP version 3: an integrated program for molecular population genetics and molecular evolution analysis. Bioinformatics 15: 174–175. [PubMed]
  • Ryan, C. A., 1990. Protease inhibitors in plants: genes for improving defenses against insects and pathogens. Annu. Rev. Phytopathol. 28: 425–449.
  • Sanchez, G. J. J., T. A. K. Yamakake, M. A. Sanmiguel, J. M. H. Casillas, A. L. Rodriguez et al., 1998 Distribucion y characterizacion del teocintle. Instituto Nacional de Investigaciones Forestales, Agricolas, y Pecuarias, Mexico City.
  • Seger, J., 1988. Dynamics of some simple host-parasite models with more than two genotypes in each species. Philos. Trans. R. Soc. Lond. B 319: 541–555. [PubMed]
  • Seger, J., 1990 Evolution of exploiter-victim relationships, pp. 3–25 in Natural Enemies: The Population Biology of Predators, Parasites and Diseases, edited by M. J. Crawley. Blackwell Scientific, London.
  • Sharbel, T. F., B. Haubold and T. Mitchell-Olds, 2000. Genetic isolation by distance in Arabidopsis thaliana: biogeography and postglacial colonization of Europe. Mol. Ecol. 9: 2109–2118. [PubMed]
  • Stahl, E. A., G. Dwyer, R. Mauricio, M. Kreitman and J. Bergelson, 1999. Dynamics of disease resistance polymorphism at the Rpm1 locus of Arabidopsis. Nature 400: 667–671. [PubMed]
  • Tajima, F., 1989. Statistical method for testing the neutral mutation hypothesis by DNA polymorphism. Genetics 123: 585–595. [PMC free article] [PubMed]
  • Tajima, F., 1993. Simple methods for testing the molecular evolutionary clock hypothesis. Genetics 135: 599–607. [PMC free article] [PubMed]
  • Tamayo, M. C., M. Rufat, J. M. Bravo and B. San Segundo, 2000. Accumulation of a maize proteinase inhibitor in response to wounding and insect feeding, and characterization of its activity toward digestive proteinases of Spodoptera littoralis larvae. Planta 211: 62–71. [PubMed]
  • Tian, D., H. Araki, E. A. Stahl, J. Bergelson and M. Kreitman, 2002. Signature of balancing selection in Arabidopsis. Proc. Natl. Acad. Sci. USA 99: 11525–11530. [PMC free article] [PubMed]
  • Tiffin, P., 2004. Comparative evolutionary histories of chitinase genes in the genus Zea and family Poaceae. Genetics 167: 1331–1340. [PMC free article] [PubMed]
  • Tiffin, P., and B. S. Gaut, 2001. a Molecular evolution of the wound-induced serine protease inhibitor wip1 in Zea and related genera. Mol. Biol. Evol. 18: 2092–2101. [PubMed]
  • Tiffin, P., and B. S. Gaut, 2001. b Sequence diversity in the autotetraploid Zea perennis and the closely related diploid Z. diploperennis: insights from four nuclear loci. Genetics 158: 401–412. [PMC free article] [PubMed]
  • Wang, R. L., A. Stec, J. Hey, L. Lukens and J. Doebley, 1999. The limits of selection during maize domestication. Nature 398: 236–239. [PubMed]
  • Watterson, G. A., 1975. On the number of segregating sites in genetical models without recombination. Theor. Popul. Biol. 7: 188–193. [PubMed]
  • Wright, S. I., B. Lauga and D. Charlesworth, 2003. Subdivision and haplotye structure in natural populations of Arabidopsis lyrata. Mol. Ecol. 12: 1247–1263. [PubMed]
  • Zhang, L., A. S. Peek, D. Dunamas and B. S. Gaut, 2002. Population genetics of duplicated disease-defense genes, hm1 and hm2, in maize (Zea mays ssp. mays L.) and its wild ancestor (Zea mays ssp. parviglumis). Genetics 162: 851–860. [PMC free article] [PubMed]

Articles from Genetics are provided here courtesy of Genetics Society of America
PubReader format: click here to try


Related citations in PubMed

See reviews...See all...

Cited by other articles in PMC

See all...


  • EST
    Published EST sequences
  • Gene (nucleotide)
    Gene (nucleotide)
    Records in Gene identified from shared sequence links
  • MedGen
    Related information in MedGen
  • Nucleotide
    Published Nucleotide sequences
  • PopSet
    Published population set
  • Protein
    Published protein sequences
  • PubMed
    PubMed citations for these articles
  • Substance
    PubChem Substance links
  • Taxonomy
    Related taxonomy entry
  • Taxonomy Tree
    Taxonomy Tree

Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...