Am. J. Bot. Plant Physiology
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow Submit a response
Right arrow Alert me when this article is cited
Right arrow Alert me when eLetters are posted
Right arrow Alert me if a correction is posted
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via ISI Web of Science (1)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by McIntire, E. J. B.
Right arrow Articles by Waterway, M. J.
Right arrow Search for Related Content
PubMed
Right arrow Articles by McIntire, E. J. B.
Right arrow Articles by Waterway, M. J.
Agricola
Right arrow Articles by McIntire, E. J. B.
Right arrow Articles by Waterway, M. J.
(American Journal of Botany. 2002;89:642-654.)
© 2002 Botanical Society of America, Inc.

Ecology

Clonal structure and hybrid susceptibility to a smut pathogen in microscale hybrid zones of northern wetland Carex (Cyperaceae)1

Eliot J. B. McIntire2 and Marcia J. Waterway3

Department of Plant Science, Macdonald Campus, McGill University, 21,111 Lakeshore Road, Ste. Anne de Bellevue, Quebec, Canada H9X 3V9

Received for publication June 22, 2001. Accepted for publication October 9, 2001.


   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 LITERATURE CITED
 
Interspecific hybrid taxa, especially those with the potential for clonal spread, may play important roles in community dynamics and plant–pathogen interactions. This study combines the mapping of clonal structure for two rhizomatous sedges (Carex limosa, C. rariflora) and their nearly sterile interspecific hybrid with an investigation of the relationship between these taxa and a nonsystemic floral smut pathogen (Anthracoidea limosa) in six subarctic fens in Nouveau-Québec, Canada. We used allozyme polymorphisms in 14 of 18 putative loci to confirm hybrid identification and to distinguish among genotypes for mapping. The incidence of A. limosa was 5–20 times greater on hybrids than on parental taxa across all sites at two spatial scales (intensive extent = 10.5 m2, extensive extent = entire fens). Spatial autocorrelation was detected in smut incidence; however, its statistical removal did not alter the strong association between hybrids and smut infection. Smut incidence on both C. limosa and hybrids was greater when they were growing in areas of high hybrid density. Our study provides evidence that disease can help maintain boundaries between species. We suggest explanations for hybrid susceptibility and provide evidence for a model in which hybrids act as a source for reinfection for all three taxa during subsequent years.

Key Words: allozymes • Anthracoidea limosa • Carex limosa • Carex rariflora • clonal growth • environmental heterogeneity • hybridization • spatial autocorrelation


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 LITERATURE CITED
 
Recent detailed studies of hybrid plant taxa suggest that they play important roles in interactions between plants and their pathogens, herbivores, and pests (Whitham, 1989 ; Aguilar and Boecklen, 1992 ; Lebrun et al., 1992 ; Graham, Freeman, and McArthur, 1995 ). In early studies, improved resistance and consequently fitness were thought to be generally characteristic of hybrids because matings between species would produce maximal heterozygosity (Brown, Kwan, and Shero, 1995 ), leading to superior individuals (e.g., Ginzberg, 1979 ; Mitton and Grant, 1980 ; Grant, 1981 ; review by Mitton, 1994 ). Numerous studies, however, now illustrate a diverse array of hybrid–herbivore and hybrid–parasite associations, refuting this hybrid vigor generalization in natural systems (Boecklen and Spellenberg, 1990 ; review by Strauss, 1994 ; Orians and Floyd, 1997 ; Orians, 2000 ). Fritz, Nichols-Orians, and Brunsfeld (1994) proposed four hypotheses for the responses of hybrids to herbivores: hybrid resistance, hybrid susceptibility, additivity, and dominance. Other studies have demonstrated that the response of hybrids to pathogens may follow similar patterns (Strauss, 1994 ; Fritz et al., 1996 ; Hjältén, 1998 ). Whether all of these hybrid responses are widespread in natural systems, however, is not clear, particularly for pathogens.

Alexander et al. (1996) noted that in many agricultural plant–pathogen systems, individual genotypes (or cultivars) determine the extent of damage caused by the disease. However, we know very little about the genetic component of variation in resistance within natural hybrid zones, or more specifically, how individual hybrid genotypes differ among themselves with respect to the severity of pathogen attack. In examples in which hybrids differ from their parents in resistance or susceptibility to a pathogen—due to breakdown in coadapted genes for example (Whitham, 1989 )—there is the potential for differential resistance or susceptibility depending on the hybrid genotype. If this is the case, then relating genetic composition of host populations with pathogen distribution becomes important (Alexander et al., 1996 ). In addition, the importance of determining genotype when analyzing the relationship between plant and pathogen dynamics in a hybrid zone is greatest when the plant in question spreads clonally because of the large potential for growth and subsequent monocultures. When ramets are used as the independent units of analysis, genotype-based plant–pathogen interactions will be overlooked, and incorrect conclusions may be drawn about spatial patterns of the disease.

Growth patterns of clonal rhizomatous plants may be influenced by environmental heterogeneity (e.g., Silander, 1979 ; Schmid, 1990 ; Macdonald and Lieffers, 1993 ; Wijesinghe and Hutchings, 1997 ). Salzman (1985) observed clear demonstrations of clone-specific foraging patterns along salinity gradients, while others have shown that environmental heterogeneity can be a determinant of clonal structure in populations (Gray, Parsell, and Scott, 1979 ). To our knowledge, no studies have addressed the relationship between environmental gradients and clonal structure within hybrid zones. Of particular interest with nearly sterile hybrid taxa is the nearly obligate use of vegetative propagation to reproduce and spread within populations.

Finally, ecological studies addressing disease dynamics must directly examine not only the host genotype and taxon, but also the physical proximity of inoculum to new uninfected plants (Burdon, 1987 ). Because the occurrence of the pathogen is likely to be limited by its ability to disperse from one place to another and by other environmental factors such as water depth and humidity, spatial relationships of the smut pathogen may be more important than host genotype (Jarosz and Burdon, 1988 ). Thus, spatial pattern must be examined as an alternative hypothesis to genetically based associations that can occur between plant and pathogen (Jean and Bouchard, 1993 ; Johnson, 1996 ).

In this study we investigated the microscale (<5 m) clonal structure of two rhizomatous sedge species and their interspecific, nearly sterile hybrid, the occurrence of a nonsystemic floral smut pathogen that infects all three taxa, and the effects of local environmental heterogeneity on the distribution of the plants and the pathogen. The study had three objectives: (1) to determine the clonal structure of hybrids and their parental taxa in mixed populations within heterogeneous wetlands; (2) to test whether the three taxa or different clones within taxa occupied significantly different microenvironments within the mixed populations; and (3) to test for differential incidence and severity of a nonsystemic floral smut fungus on the three taxa or on clones within taxa, taking spatial autocorrelation into account. The third objective was tested at two spatial scales to differentiate between local and broader scale effects.


   MATERIALS AND METHODS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS
 DISCUSSION
 LITERATURE CITED
 
Study species
Carex rariflora (Wahlenb.) J. E. Smith, C. limosa L. (both Sect. Limosae) and their interspecific, nearly sterile hybrid (C. limosa x rariflora Kükenthal) are rhizomatous sedges forming potentially large clones in northern mires (Bernard, 1990 ). Carex rariflora has a circumpolar distribution in subarctic and arctic latitudes, while the range of C. limosa overlaps that of C. rariflora but also extends much farther south across the boreal forest (Porsild and Cody, 1980 ). Carex limosa is an early colonizing species of open shallow pools and open, mud-bottom sites in a variety of boreal and subarctic mire types, including bogs, poor fens, and minerotrophic fens (e.g., Vitt, Achuff, and Andrus, 1975 ; Slack, Vitt, and Horton, 1980 ; Vitt and Bayley, 1984 ; Foster, King, and Santelmann, 1988 ). It often acts to facilitate the colonization of other vascular plant and bryophyte species into formerly open pools (Ruttner, 1963 ; Kazuo and Kazuyuki, 1997 ; M. J. Waterway, personal observation). Carex rariflora occurs commonly in peaty tundra habitats, at the edge of tundra ponds or streams, and along the margins of fens in which C. limosa grows (Vellend and Waterway, 1999 ).

In mixed populations in suitable heterogeneous habitats where the geographic ranges overlap, hybrids between C. rariflora and C. limosa have been reported (Kükenthal, 1909 ; Holmberg, 1929 ; Scoggan, 1978 ; Jermy, Chater, and David, 1982 ). A broad survey of fens in the Schefferville region revealed the presence of several hybrid populations (N. Djan-Chékar and M. J. Waterway, McGill University, unpublished data), but the extent and abundance of hybrid clones at the study sites were not determined. Little is known about the ecology of this interspecific hybrid. The parental taxa, especially C. limosa, spread vegetatively by rhizome extension but also reproduce sexually by seed. The hybrids also spread vegetatively, but they have low pollen fertility and apparently rarely set seed (N. Djan-Chékar and M. J. Waterway, McGill University, unpublished data).

Anthracoidea limosa (H. Syd.) Kukk. (Ustilaginales) is a nonsystemic, ovaricolous smut fungus that annually infects host Carex plants, enabling the production of fungal teliospores at the expense of host seed development (Ericson, Burdon, and Wennström, 1993 ; Salo and Sen, 1993 ). A successful infection results in the replacement of the ovary by a black teliospore sorus. This sorus remains intact on the dead flowering shoot over winter (Ericson, Burdon, and Wennström, 1993 ); teliospores germinate in spring to form short hyphae that produce infectious basidiospores. More detailed descriptions of life cycles of species within this genus can be found in Kukkonen and Vatanen (1968) . Infection occurs by wind- or insect-borne basidiospores landing on the extended stigmas of the host plants and year-to-year correlations in incidence of Anthracoidea smut on individual plants are usually high (Ingvarsson and Ericson, 1998 ).

Study area and field sampling
Extensive scale
We selected six minerotrophic fens where hybrids had previously been collected in the Schefferville region, at the border of northern Quebec and Labrador, Canada (Capricorn Fen, 54°46' N, 66°47' W; Goodream Fen, 54°56' N, 66°06' W; Greenbush Fen, 55°00' N, 67°13' W; Astray Fen, 54°40' N, 66°36' W; Iron Arm Fen, 54°53' N, 66°38' W; Railroad Fen, 54°54' N, 66°51' W). In each fen, we randomly sampled a single fertile ramet (i.e., one with an inflorescence that had flowers or immature fruits) at intersections of a grid with squares ranging from 10 x 10 m to 20 x 20 m, depending on the size of the fen, to provide relatively equal coverage of the entire fen (Capricorn, N = 187; Goodream, N = 71; Greenbush, N = 86; Astray, N = 149; Iron Arm, N = 146; Railroad, N = 158). Each sample was collected without regard to which of the three taxa it belonged or whether the inflorescence was infected with Anthracoidea teliospores. The samples were stored dry for later identification and determination of smut incidence and severity.

Intensive scale
In three of the six fens (Astray, Iron Arm, and Railroad), we also selected a site 5.25 x 2 m apparently containing both hybrids and the two parental species for small-scale sampling and clone mapping based on multilocus allozyme phenotypes. Within the 10.5-m2 rectangle, we sampled a single ramet at intersections of a grid with squares 0.25 x 0.25 m, creating plots of 22 x 9 plants (198 intersections per site). At the three intensive sites, plants of these taxa were common; all points except two contained a ramet, giving 592 ramets for allozyme analysis. Because the environmental affinities of the two parental species differ, sites were also selected to have a degree of environmental heterogeneity, including hummocks of Sphagnum moss as well as open mud bottoms, so as to encompass all three taxa. A single ramet for allozyme analysis was selected closest to the point of each grid intersection, regardless of taxon or presence of smut. If a fertile ramet was present within 10 cm of the point, it was selected preferentially over a vegetative ramet since the focus of this study was the distribution of the floral smut fungus. Thus, the relative abundance of fertile shoots vs. vegetative shoots will be an overestimate of that in the actual population. In addition, the relative abundance of any given taxon may be over- or underestimated if the proportion of fertile ramets differs among taxa.

Allozyme analysis
As in previous studies (e.g., Rhebergen, Theeuwen, and Verkleij, 1988 ; Parker and Hamrick, 1992 ; McClintock and Waterway, 1993 ), allozyme polymorphisms were used to determine the multilocus genotype of all collected ramets from our intensive sampling regime and to map the clonal structure of the populations. Plant samples for allozyme analysis were stored at 4°C for up to 3 wk prior to extraction. Small amounts of green tissue were homogenized in about 300 µL of extraction buffer (Gottlieb, 1981 ) and absorbed onto filter paper wicks that were stored at –80°C until electrophoresis was carried out. Each ramet was assayed for 14 enzyme systems. Standard techniques of starch gel electrophoresis, as previously described (Wendel and Weeden, 1989 ; Schell and Waterway, 1992 ; McClintock and Waterway, 1993 ), were used to separate electromorphs and stain for enzyme activity. Alcohol dehydrogenase (ADH, E.C. [Enzyme Commission] 1.1.1.1), glucosephosphate isomerase (GPI, E.C. 5.3.1.9), aspartate aminotransferase (AAT, E.C. 2.6.1.1), acid phosphatase (ACP, E.C. 3.1.3.2), triosephosphate isomerase (TPI, E.C. 5.3.1.1) were run on a lithium-borate gel system at pH 8.1. Diaphorase (DIA, E.C. 1.8.1.4), phosphoglucomutase (PGM, E.C. 5.4.2.2), aldolase (ALDO, E.C. 4.1.2.13), shikimic acid dehydrogenase (SKDH, E.C. 1.1.1.25), and 6-phosphogluconic acid dehydrogenase (6-PGD, E.C. 1.1.1.44) were assayed using a histidine buffer system, pH 6.5. A morpholine-citrate gel system, pH 6.1, was used to assay peroxidase (PRX, E.C. 1.11.1.7), malic dehydrogenase (MDH, E.C. 1.1.1.37), and glucose-6-phosphate dehydrogenase (G-6-PDH, E.C. 1.1.1.49). Menadione reductase (MDR, E.C. 1.6.99.3) and DIA (rerun for clarity of interpretation) were run on a tris-citrate gel system, pH 5.7. Peroxidase was also rerun for clarity on a histidine-citrate gel system, pH 5.7. Recipes for staining enzymes were used, with minor modifications, from Soltis et al. (1983) and Wendel and Weeden (1989) .

Hybrid plant identification
The three taxa are difficult to distinguish using only vegetative characters, but C. limosa and C. rariflora can be easily differentiated by inflorescence morphology. Hybrid individuals were confirmed by analysis of the allozyme phenotypes as described below. In a previous study conducted on these three taxa and in the same wetlands, hundreds of individual fertile ramets were randomly collected and genotyped for several loci using the same allozyme techniques used here (N. Djan-Chékar and M. J. Waterway, McGill University, unpublished data). Every ramet was identified using a combination of inflorescence morphology and allele data. Based on this study, four loci were found to be diagnostic of parental status (TPI-2, SKDH, 6PGD, and DIA-1). These loci differed consistently between the two parents in all ramets collected. In our intensive-scale sampling strategy, a plant was determined to be a hybrid when its zymogram showed the expected heterozygous banding patterns at these diagnostic loci. In the previous study, a few individuals that were clearly of hybrid origin based on the other diagnostic alleles and morphology were homozygous at some loci for alleles that were almost exclusive to one parent or the other (e.g., SKDH bb and cc and 6PGD bb), suggesting that at least some F2 hybrids and backcrosses were present. Thus, it was apparent before our study that the hybrids in these sites formed part of hybrid swarms and were not all F1 hybrids.

In the extensive-scale sampling strategy, where sample size was too large to make allozyme analysis feasible, the three taxa were identified based on floral characters established from the previous study (N. Djan-Chékar and M. J. Waterway, McGill University, unpublished data) and from careful study of 100 inflorescences from ramets collected at the intensive scale, whose identification was known based on allozyme data. We did a blind testing of identifications on the remaining 492 inflorescences whose identities were known based on the allozyme study at the intensive scale. Our identifications in this blind test were correct more than 95% of the time, and we assumed a similar level of accuracy for the extensive scale study where the identifications could not be checked by allozyme analysis. Parental species were easily differentiated based on scale color, spike size, and perigynium shape. The two most important traits used to distinguish hybrids were intermediate color of the perigynium scales (C. limosa, light golden; C. rariflora, dark purple; hybrid, medium reddish gold) and larger female spikes (number of perigynia per spike ± SD: C. rariflora = 5.4 ± 1.9; C. limosa = 7.7 ± 2.9; hybrid = 9.0 ± 3.7).

Clonal statistics
At the intensive scale, we used allozyme data to map the occurrence of individual Carex clones. For each taxon and within each fen, we calculated N (number of ramets sampled), G (number of genotypes found), and G/N (genotype discovery rate). We also calculated the maximum number of possible genotypes (Ng; calculated as in Parker and Hamrick, 1992 ) at each site for each taxon, a number based on all possible combinations of observed enzyme-coding alleles.

Environmental heterogeneity
At the intensive scale, we measured rooting height above (or below) the water table at each sampled ramet as a measure of the immediate environment for each plant. The plants were generally rooted in a floating peat substrate so the actual water depth in the fen was usually much greater than the values we recorded. For simplicity, we will refer to this measurement of position of rooting in the peat substrate in relation to the upper surface of the water as "water depth" throughout the paper. To allow us to quantitatively test for among-site environment differences, we performed a single-factor ANOVA on water depth at the three sites. To determine whether there were differences in water-depth affinities among taxa and among clones within taxa, we calculated a three-way nested ANOVA (site, taxon [site], and clone [taxon, site]) on water depth. We used a nested ANOVA because a fully factorial analysis was not possible; C. limosa was not represented in the samples from Railroad and only one ramet of C. rariflora was sampled at Iron Arm. "Site" was included in the model to account for the fact that each site was unique and the sites could not therefore be considered true replicates of one another. Only multiramet clones were used in the analysis to allow for variance estimations for each clone.

To examine whether individual clones of hybrids and parents differed in the range of water depths at which each clone was rooted, we calculated the mean, the range, and the variances of water depth for each species at each site and for each multiramet clone.

Pathogen incidence and severity
In all pathogen-related calculations performed on plants sampled at the intensive scale, vegetative ramets were removed from the analyses because the fungus infects only floral parts. Because only fertile ramets were collected during the extensive-scale sampling, all ramets were used in those analyses.

For each fertile ramet sampled, we counted the total number of pistillate flowers (perigynia) and total number of smut sori that replaced perigynia. We distinguished between disease incidence and disease severity: disease incidence was the number of fertile ramets (inflorescences) infected with smut sori, and disease severity was the proportion of perigynia infected per infected ramet. Thus, disease severity was a measure of the intensity of infection where it occurred and could only be analyzed for infected ramets.

To determine whether the three taxa were differentially infected with the smut fungus and whether scale of sampling caused the apparent infection level and rate to differ, we used a three-way nested general linear model on disease incidence (scale, site [scale], host, and interactions; family = binomial; Math Soft, 1995 ) and a similar three-way nested GLM on disease severity (scale, site [scale], host, and interactions; family = binomial; Math Soft, 1995 ). As above, "site" was included in the model because the six sites could not be considered true replicates. Rather than include all three taxa separately, we present statistical analyses with C. limosa and C. rariflora as a pooled "parent" class to increase the power of contrasts and to more formally test for hybrid disadvantage because preliminary tests showed that hybrids were more infected than either parent. We present raw data with parents separated to allow visual inspection of the differences between parents. To visually examine the spatial pattern of smut incidence, we mapped the occurrence of smut onto both the intensive-scale and the extensive-scale grids.

To examine whether different genotypes suffered different rates of infection, we tested for among-clone differences in disease severity. We calculated a one-way GLM (clone; family = binomial; Math Soft, 1995 ) on disease severity using all hybrid clones with two or more ramets at Railroad Fen. Comparisons of disease severity could not be made for individual clones at Astray or Iron Arm fens due to the small number of clones identified at those sites. Similarly, the small number of clones of the parental species at each site precluded comparisons of disease severity or intensity among clones of the parental species. We could not evaluate among-clone variability in disease incidence because of the overall low number of clones in our sampling (at Astray and Iron Arm fens) and the near-zero variability of smut incidence at Railroad fen where virtually all hybrid ramets were infected.

Finally, because the distribution of smut fungus could either be determined by taxon (genetic factors) or by dispersal limitation (spatial factor) we tested smut distribution for spatial autocorrelation using spatial correlograms of both Moran's I and Geary's c (Cliff and Ord, 1981 ) as implemented in PASSAGE (Rosenberg, 2000 ). We calculated these correlograms for both scales of analysis using ten distance classes in each case. Where significant spatial autocorrelation was detected, we used CRH modified t tests (Clifford, Richardson, and Hémon, 1989 ) to test associations between each species and the presence of smut infection, while adjusting the degrees of freedom for spatial autocorrelation. Because flowers are the only available sites for fungal infection, only fertile ramets were used in these analyses.


   RESULTS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
DISCUSSION
 LITERATURE CITED
 
Allozyme analysis and clonal statistics
Thirteen of the 18 putative loci showed clear banding patterns and could be consistently scored. Nine of these were polymorphic in C. limosa (ACP, 6PGD, SKDH, DIA-1, PGM, GPI-2, AAT-1, AAT-2, PRX), six were polymorphic in C. rariflora (ACP, SKDH, DIA-1, GPI-2, AAT-1, PRX), and ten were polymorphic in the hybrids (ACP, 6PGD, SKDH, DIA-1, PGM, TPI-1, GPI-2, AAT-1, AAT-2, PRX). Genotype discovery rates (G/N) ranged from high values in C. rariflora (0.67, excluding the single ramet in Iron Arm Fen) to very low in hybrids (0.05) (Table 1). The maximum number of possible genotypes (Ng)—calculated based on all possible combinations of observed enzyme-coding alleles—ranged from 1296 (C. rariflora in Astray Fen) to 104 976 genotypes (C. limosa in Astray Fen). These values are based on the relatively small proportion of the mating population found within our intensive grids and are likely to be underestimates of the actual numbers of possible genotypes. Another way to evaluate the accuracy of clonal discrimination is to calculate the probability of encountering the most likely multilocus genotype based on the product of the most common alleles in the mating population. The probability of sampling the most likely hybrid genotype in Astray fen, based on allelic frequencies of the parental species within the intensive sampling grids, was only 0.005 and that of sampling the large hybrid clone we discovered in Astray fen (labelled ‘a’, Fig. 1) was 0.0017, but these probabilities could be even lower if allelic frequencies from the entire mating population were available. We do not report probabilities for the other two sites because so few parents were sampled, but these are also likely to be low based on our limited data from a parent at each site (<0.05 and <0.03). Because a high number of genotypes could potentially be detected with the observed level of allozyme variability in these relatively small plots (10.5 m2) and the probabilities of encountering two or more different individuals of a particular genotype were very small, we assumed that individual ramets with identical multilocus genotypes formed part of the same genet.


View this table:
[in this window]
[in a new window]
  		Table 1. Clonal diversity at the three intensive sites. G is the number of genotypes found; N is the number of ramets sampled; G/N is the genotype discovery rate, a number that indicates the size of the clones, where larger G/N indicates smaller clones and vice versa. Ng is the maximum number of genotypes possible, based on all possible combinations of observed enzyme-coding alleles at each site

 


View larger version (58K):
[in this window]
[in a new window]
  		Fig. 1. Maps of the 2 x 5.25 m grids (intensive sites) from three fens in the Schefferville region, Quebec, Canada. (A) Within each taxon (Carex rariflora, hybrids, and C. limosa), ramets that had identical multilocus genotypes are given the same letter. The letters are different for each grid and do not correspond to those in any other grid. Each taxon is drawn on a separate grid for clarity only. The single blank spaces at Iron Arm Fen and Astray Fen represent locations where no ramet was available to collect. (B) Maps of pathogen incidence. White squares = C. rariflora, light gray squares = hybrid, dark gray squares = C. limosa. Numbers represent presence (1) or absence (0) of Anthracoidea limosa. Squares with no number contained ramets without flowers. An X indicates where no ramet was available to sample

 
Clone maps of the three sites based on shared multilocus genotypes are shown in Fig. 1A. Hybrids have the lowest G/N ratios in each site (Table 1), indicating that they often occur as large clones (Fig. 1A), although hybrid G/N at Railroad Fen was substantially higher than at the other two sites. At Astray and Iron Arm fens, 77% and 40%, respectively, of the sampled ramets belonged to single, large, hybrid clones. Carex rariflora had high G/N ratios and C. limosa had intermediate G/N ratios, indicating that they generally occurred as very small genets and small-to-medium genets, respectively (Table 1). The largest clone of C. limosa was found at Iron Arm fen, represented by 26% of the ramets sampled. No clone of C. rariflora was larger than three ramets or 1.5% of the points in any site.

Although we selected the intensive sites with the intention of including all three taxa within the sampling belt, allozyme analysis revealed only one ramet of C. rariflora at Iron Arm Fen and no individuals of C. limosa at Railroad Fen. Three factors may have contributed to these exclusions. First, we could not do a detailed examination of the plants before sampling due to the destructive nature of sampling, so sites were chosen based on observation of all three taxa in the immediate vicinity. Second, it was difficult to distinguish among vegetative shoots of the three taxa in the field. Finally, because we preferentially sampled fertile ramets, any differential flowering among taxa would contribute to oversampling of taxa with higher flowering rates and undersampling of taxa with relatively lower flowering rates.

Environmental heterogeneity
Water-depth patterns at the three sites are shown in Fig. 2. The water depth at each site differed significantly (F2,585 = 40.7, P < 0.0001), with Railroad Fen having the lowest overall water depth and Astray and Iron Arm fens being statistically indistinguishable (Fig. 2). Results from the nested ANOVA on water depth suggest that within each site, the mean water depths occupied by the taxa varied significantly (Table 2, Fig. 3). Pooled across all sites, C. rariflora had the highest mean rooting depth above the water table, C. limosa the lowest, and hybrids grew most commonly at an intermediate water depth (Figs. 3 and 4). At each site and across all sites, hybrids had the largest range of water depths (Figs. 3 and 4), although this may be a function of disproportionate sampling of hybrids and site selection. This analysis demonstrates, nevertheless, the broad environmental range of hybrid growth. In addition, the effect of clone in the ANOVA was significant, indicating that the mean water depths occupied by particular clones varied significantly (Table 2). Examples of hybrid clones that were rooted at different depths above or below the water table were clone c at Railroad Fen with many ramets rooted below the water table (N = 58, –0.37 ± 0.36 cm, mean ± 1 SE) and hybrid clone t at Railroad Fen that rooted above the water table, on average (N = 20, 1.40 ± 0.78 cm). Variation within and among clones of parental species was also large, with C. rariflora clones g and d at Astray Fen differing by more than 5 cm and each exhibiting a large variance in water depth (clone g: N = 2, 0.5 ± 0.5 cm; clone d: N = 2, 6.0 ± 2.0 cm).



View larger version (94K):
[in this window]
[in a new window]
  		Fig. 2. Contour plots of rooting depth above water table for each site. Heights are in centimeters; negative heights indicate plants were rooted below the water table. Mean water depths (± SE) by site were: Astray = 3.7 (± 0.3) cm, Iron Arm = 4.0 (± 0.3) cm, Railroad = 0.74 (± 0.22) cm

 

View this table:
[in this window]
[in a new window]
  		Table 2. Summary from ANOVA on water depth (N = 548). All single-ramet clones were removed from the analysis

 


View larger version (26K):
[in this window]
[in a new window]
  		Fig. 3. Occurrence of each taxon by water depth at each site and pooled across all sites. Due to unequal numbers of ramets, values are given as a proportion of ramets of each taxon at the site.

 
Pathogen incidence and severity
Incidence of pathogen infection is mapped at both scales to allow visual inspection of the disease pattern on the taxa (Figs. 1B and 5). Examination by scale, site, and taxon demonstrates that hybrid clones had the highest incidence of fungal infection at all sites (Fig. 6). Overall, incidence of smut on hybrids at the extensive sites was 35.6%, over five times higher than on C. limosa and over 20 times higher than on C. rariflora (Fig. 6). At the intensive scale, incidence was 84.5%, 55.2%, and 5.9% for the three taxa, respectively (Fig. 6). At both Iron Arm Fen and Railroad Fen, over 90% of fertile hybrid ramets were infected with smut sori. These differences in levels of disease incidence between scales, between hybrids and parents at each site, and among sites at each scale were all highly significant (Table 3). The interaction between host and site was not significant (P = 0.08), indicating that the pattern of disease incidence on parents and hybrids was not significantly affected by site differences. The interaction between scale and taxon was significant at P = 0.05, however, indicating that the pattern of smut incidence differed between the two scales (Fig. 6).



View larger version (22K):
[in this window]
[in a new window]
  		Fig. 6. Smut incidence on parental taxa and hybrids from both intensive and extensive scale sampling. Incidence indicates the proportion of fertile ramets in a population that had smut sori.

 

View this table:
[in this window]
[in a new window]
  		Table 3. Summary of three-way nested GLM (family = binomial) on smut incidence (N = 1105). Carex limosa and C. rariflora were pooled ("host") to more formally test for hybrid disadvantage

 
Analysis of disease severity among infected plants, however, showed a different trend. Disease severity was significantly greater at the intensive scale than the extensive scale (intensive = 0.52, extensive = 0.39; Table 4) and among sites at each scale, but not between host types (parent vs. hybrid; Table 4). Though in each site hybrids were more severely infected (Table 5), the inclusion of "site" and "scale" in the analysis accounted for most of the variation. Similarly, the among-clone analysis of disease severity (hybrids at Railroad only) was nonsignificant (N = 10, df = 9, deviance = 2.00, P = 0.99) indicating that the intensity of infection did not differ among different genetic individuals.


View this table:
[in this window]
[in a new window]
  		Table 4. Summary of three-way nested GLM (family = binomial) on severity of smut infection among diseased ramets (N = 333). Carex limosa and C. rariflora were pooled ("host") to more formally test for hybrid disadvantage

 

View this table:
[in this window]
[in a new window]
  		Table 5. Summary of disease severity at both intensive and extensive sampling scales. Severity represents the average proportion of flowers infested with smut using only infested ramets (with standard error). N indicates the total number of ramets sampled at each site. N (with smut) indicates the number of inflorescences that have at least one flower replaced by a fungal sorus

 
Finally, spatial correlograms revealed that smut distribution had significant spatial autocorrelation. Moran's I correlograms were significant at Astray and Railroad fens (P < 0.05), and Geary's c correlograms were significant at Iron Arm and Railroad fens (P < 0.05). Patch size (the distance up to which there is significant positive spatial autocorrelation, at P < 0.05) based on Moran's I and Geary's c were 0.8 m (I) and 1.8 m (c) at Iron Arm Fen, and 1 m (I) and 1.3 m (c) at Railroad Fen. There was no significant patch size at Astray Fen. At the extensive scale, all sites with smut had significant Geary's c correlograms (P < 0.05) and all but Iron Arm Fen had significant Moran's I correlograms (P < 0.05). With the presence of this spatial autocorrelation, we performed CRH modified t tests. Results of the CRH modified t tests at the intensive scale are shown in Table 6. At all three sites, we detected a significant correlation between infection with Anthracoidea smut fungi and the presence of hybrid host plants, despite spatial nonindependence. At the extensive scale sampling, all sites also showed a highly significant association between hybrid presence and smut presence (P < 0.001) despite reductions in sample sizes due to spatial autocorrelation.


View this table:
[in this window]
[in a new window]
  		Table 6. The results of a CRH (Clifford, Richardson, and Hémond, 1989) modified t test of smut incidence vs. species presence. Effective sample size (N) reflects the loss of degrees of freedom due to spatial autocorrelation. A significant effect based on the effective sample size indicates that the correlation between the species and smut incidence is significant, despite spatial autocorrelation

 

   DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
LITERATURE CITED
 
Clone structure and environmental analysis
The levels of G/N found in this study are low compared with those in studies of other rhizomatous Carex species (McClintock and Waterway, 1993 ; Jonsson, Jónsdóttir, and Cronberg, 1996 ; Stenström et al., 2001 ), indicating that the sampling structure (i.e., every 0.25 m) was sensitive to intermingling of individual clones. The G/N statistics demonstrate that hybrid clones at these sites can become larger than either parent, providing an example of hybrid vigor with respect to growth. Examples such as these with few, large clones being dominant may result from competition among the intermingling genets observed in this study or from stochastic mortality without seedling recruitment. Our observations that hybrid clones are larger may simply be due to the location of sampling, which encompassed an intermediate zone between the normal parental habitats. Other studies have shown that C. limosa can produce very large clones, usually, however, in wetter places than those sampled in the present study (Novo and Waterway, 1996 ; M. J. Waterway and K. McClintock, McGill University, unpublished data). Thus, the dominance of hybrids observed in these intensive-scale sites might be an artifact of sampling in areas where the two parental species co-occur rather than in the central part of the range of each. Nonetheless, our results demonstrate substantial growth potential among hybrid clones.

In nonclonal hybrid plants, where sexual reproduction is necessary to establish new individuals, hybrid persistence and increase in population size must rely on repeated hybridization events. However, in these rhizomatous sedges, despite low levels of sexual reproduction (due to near sterility and the apparently high incidence of smut infestation), hybrid clones were clearly able to grow to comparatively large sizes and maintain large populations. One important corollary to clonal growth, especially with little sexual reproductive input, is the potential formation of large monocultures. In such cases of reduced genetic diversity, hybrids may be more susceptible to disease infection such as we found here.

The large number and relatively small size of the hybrid clones at Railroad Fen in comparison to the other two fens may be a result of the age of the sites. Because these plants are clonal, their size can be used as an estimate of the relative age of genets (e.g., Hartnett and Bazzaz, 1985 ; Schmid, 1990 ; Parks and Werth, 1993 ). Thus, the large size of the hybrid clone in Astray Fen suggests that it has occupied this site for a long time. Railroad Fen, whose drainage was disturbed about 45 yr before our sampling by construction of railroad tracks on both sides, has numerous areas of open, muddy bottoms and a generally low water table (Fig. 2). Colonization by the larger number of smaller clones of C. limosa and the hybrids suggests that this site may be in an earlier stage of development than the other two intensive sites. Thus, this study provides a natural example of Watkinson and Powell's (1993) simulation model, which is based on disturbances. To understand the patterns of change during the development of clonal plant populations, Watkinson and Powell (1993) modeled the survival of seedlings and the maintenance of clonal diversity. They demonstrated that young populations with recent recruitment began with large numbers of genets that were slowly lost over time, eventually leaving populations dominated by few, large clones. This temporal progression from young to old parallels the differences observed between the three sites in our study: Railroad Fen, Iron Arm Fen, and Astray Fen, respectively. Astray Fen and Iron Arm Fen both have few, generally large clones. The lack of open muddy bottom at either place suggests that these sites have already been colonized and may be in a process of self-thinning due perhaps to limited seedling establishment or competition among clones (Gray, Parsell, and Scott, 1979 ; Watkinson and Powell, 1993 ). Because C. limosa is often one of the first vascular species to colonize open water or open mud, and it facilitates other species, low species numbers would also be indicative of recently colonized areas. Vegetation data gathered from our intensive sites support this line of reasoning; the mean total number of vascular plant species per 100 cm2 circular quadrat centered at each sampling point was lowest at Railroad Fen (mean = 3.2 species, N = 198), intermediate at Iron Arm Fen (mean = 4.6 species, N = 198) and highest at Astray Fen (mean = 7.0 species, N = 198), suggesting the site in Railroad Fen is at an earlier colonization stage than those at either Iron Arm Fen or Astray Fen (E. J. B. McIntire, unpublished data).

Microhabitat selection by different genetic individuals has been demonstrated previously. In glasshouse experiments, for example, Macdonald and Lieffers (1993) showed the foraging tactics of a clonal grass species given a heterogeneous environment with respect to light, competition, and temperature. In this study, we demonstrated that within the intensive sites, water-depth affinities among the taxa varied significantly, indicating that C. limosa, C. rariflora, and their hybrid grew in different microhabitats. At Astray Fen, Iron Arm Fen, and pooled across all sites, hybrids had the greatest variance and greatest range of water depths (Fig. 4). The same analysis demonstrated differential water-depth affinities among clones (Table 2), suggesting that individual clones occupied different parts of the water-depth gradient. The simplest explanation for the large hybrid water-depth variance is that the ecological range of the hybrids is additive with respect to the ranges of its parents and, because our sites were selected to encompass the ranges of both parents, the sites themselves encompass intermediate ranges. This is analogous to the additive hypothesis concerning hybrid resistance to herbivores and pests (Fritz, Nichols-Orians, and Brunsfeld, 1994 ) and is a generally observed pattern in hybrid zones with respect to ecological amplitude (Barton and Hewitt, 1985 ; Boecklen and Spellenberg, 1990 ). In our sites, it appears that while there was a tendency for different genetic individuals to occupy different parts of the water-depth gradient, individual hybrid clones did not span greater or smaller ranges of water depth than individual parental clones did.



View larger version (18K):
[in this window]
[in a new window]
  		Fig. 4. Mean rooting depth (±SE) above (+) or below (–) water table. Maximum and minimum samples are each indicated by an asterix. L represents C. limosa ramets, H represents hybrid ramets, and R represents C. rariflora ramets. Numbers in parentheses are sample sizes.

 
Hybrid–pathogen interactions
While smut incidence showed significant spatial autocorrelation, the CRH modified t test on smut and hybrid association demonstrated that the spatial factors influencing smut occurrence (e.g., smut dispersal, water table) were not sufficient to account for the high infection rate among hybrids (Table 6). Two other factors could explain the increased incidence of smut infestation on the hybrids: differences in genetically based resistance or a difference in flowering phenology. Hybrids could be more susceptible to smut infection because of the breakdown of coadapted resistance genes that may exist in the parental species (Whitham, 1989 ) or due to the loss of alleles for resistance (Thompson and Burdon, 1992 ). However, we did not observe differences in disease incidence or severity among the hybrid clones as would be expected if the differential susceptibility were genetically based. Because we did not experimentally test for genetically based differences in resistance among taxa in controlled conditions (e.g., Hjältén, 1998 ), we are not able to exclude this as a hypothesis. However, field observations on flowering phenology offer an alternative explanation.

The second mechanism, known as the phenological sink model (Floate, Kearsley, and Whitham, 1993 ), is more plausible and has been observed in some systems including other Carex–Anthracoidea associations (Marquis, 1992 ; Ingvarsson and Ericson, 1998 ). Although the hybrids studied here are nearly sterile, they still expose their stigmatic surfaces at flowering time as the parental species do, but the ovules are rarely fertilized. When fertilization does not occur, the stigmas remain extended much longer than the parental species (Kukkonen and Vatanen, 1968 ; M. J. Waterway, personal observation), resulting in a higher probability that smut basidiospores will land on a receptive surface. An experimental study demonstrating that Anthracoidea smut spores infect Carex only at the early stages of flowering, during a relatively short period in early spring (Kukkonen and Vatanen, 1968 ), emphasizes the importance of the timing and duration of stigmatic receptivity for infection. Furthermore, the phenology of the three taxa is such that C. rariflora flowers earlier than C. limosa or the hybrid (M. J. Waterway, personal observation) and may be earlier than the release of smut basidiospores in some years. This could explain the particularly low levels of smut infection at both scales and at all sites for C. rariflora during the year of our study. Both C. limosa and the hybrids may flower during the time that the smut basidiospores are infective, but since the hybrid stigmas remain receptive longer, basidiospores are more likely to land on them, resulting in greater susceptibility in the hybrids.

Within hybrid zones, four hypotheses concerning responses to herbivores, pests, and pathogens have been described (Fritz, Nichols-Orians, and Brunsfeld, 1994 ; Strauss, 1994 ). Our results provide support for the hybrid susceptibility hypothesis, because hybrids had a significantly higher level of disease incidence than did either parent, with C. rariflora demonstrating almost no infection. Contrasting arguments have been made about whether hybrid susceptibility has an effect on the demography of the pathogen in populations of the parental species, and, if so, whether susceptible hybrids serve as a pathogen sink (Whitham, 1989 ; Ericson, Burdon, and Wennström, 1993 ; Fritz, Nichols-Orians, and Brunsfeld, 1994 ). Whitham's "hybrids-as-sinks" model (1989) has the underlying implication that the pathogen has evolved to adapt to the hybrid host, effectively removing the disease from the parental species. The increased susceptibility of the hybrids demonstrated in our study is not inconsistent with the hybrids-as-sinks model, but we were not able to find a strong link between the susceptibility to the pathogen and hybrid genotype, nor did we investigate genetic changes in the pathogen. Instead, our study provides support for a frequency-dependent model in which the occurrence of the smut spores on the hybrids allows the hybrids to act as a source for reinfection of parental species when they grow in proximity. Areas within fens with high hybrid frequency (such as our intensive scale sites in which almost 90% of fertile ramets were hybrids as compared to approximately 24% occurrence at the extensive scale sites) will necessarily provide an abundant source of smut inoculum available to reinfect year after year (Alexander, 1990 ). Because the spores are predominantly wind- or insect-dispersed (Steiner, 1984 ; Ericson, Burdon, and Wennström, 1993 ), infection will be spatially defined (see spatial autocorrelation above; Real and McElhany, 1996 ) and density or frequency dependent (Antonovics, 1992 ), resulting in a patchy distribution of disease incidence throughout the fen (Ingvarsson and Ericson, 1998 ). A hybrids-as-source model would predict that susceptible C. limosa or C. rariflora plants growing in or near areas of high hybrid density should show higher incidence of smut infection than expected on average across the fens. Our results showing that C. limosa at the intensive scale (i.e., when in proximity to hybrids) was more infected than at the extensive scale (when it is more often not in proximity to hybrids) are consistent with this model. Because proximity of host equally applies to hybrids, the observation that hybrids are also more infected in areas of high hybrid density also follows this model. Thus, the increased incidence of smut on C. limosa and the hybrids at the intensive scale provides support for a hybrids-as-source model and would not be predicted by a hybrids-as-sink model. Hybrids appear to be at least partially responsible for maintaining A. limosa populations that can infect C. limosa and C. rariflora.

Implications of host–pathogen interactions
Gene flow between sibling plant species is commonly known to occur (Rieseberg and Wendel, 1993 ) and may, in some cases, result in the breakdown of isolating barriers between two partially isolated species (Grant, 1981 ). The mechanisms maintaining species barriers in some cases are poorly understood and may be diverse (Johannesson, 2001 ). The high incidence of smut infection on hybrids in our sites has the effect of reinforcing the almost complete reproductive barrier between C. rariflora and C. limosa by reducing the already scant F1 seed production. If hybrid susceptibility occurs as generally as studies indicate (Ericson, Burdon, and Wennström, 1993 ; Strauss, 1994 ; Christensen, Whitham, and Keim, 1995 ), pathogens, pests, and herbivores may play an important role in maintaining species boundaries by acting as ecological forces contributing to reproductive isolation among sympatric species.

In this study, unlike a previous example (Ericson, Burdon, and Wennström, 1993 ), survival of A. limosa is not absolutely dependent on the existence of hybrid plants. Although incidence was much higher on the hybrids, substantial infections were also found on C. limosa. Anthracoidea limosa has also been reported to infect C. magellanica, another species in Carex section Limosae (Vanky, 1994 ). Hybrids may play a role in stabilizing the populations of the smut fungi. Due to the short window of opportunity each spring to infect flowers of fertile species through their stigmas, smut populations would be expected to fluctuate from year to year if timing of basidiospore release and timing of stigma exsertion in the host plant were dependent on slightly different meteorological conditions. In some years, the two events might be coincident, but due to stochastic factors, in other years they may overlap only slightly or not at all. Hybrids, with a much longer period of stigma receptivity, would be more likely to be infected every year, thus stabilizing the fungal population and serving as a source for infection for parental species. In this metapopulation context, it is likely that the fungus may be at least partially dependent on the hybrids in these populations for its long-term survival and success.



View larger version (71K):
[in this window]
[in a new window]
  		Fig. 5. Maps of pathogen incidence at the six extensive-scale sites. Each square represents a single ramet sampled at the intersection of a 10 x 10 m or 20 x 20 m grid, depending on the site. Blanks indicate that no ramet was present, due to open water or edge of the fen. See Fig. 1B for key to numbering and shading codes

 

   FOOTNOTES
 
1 The authors thank A. Bond, S. Burelle, G. Drake, K. Einhorn, C. Mercier, L. Sack, J. Stabile, and M. Trudeau for help with the enzyme extractions; M. Vellend and C. Novo for field assistance; K. Vanky for identifying the smut pathogen; and M.-J. Fortin for advice on the spatial analysis. We also thank the staff at the McGill Subarctic Research Station in Schefferville for logistic support. Financial support was provided by the Northern Scientific Training Program through the Department of Indian Affairs and Northern Development, Canada and by a research grant from the Natural Sciences and Engineering Research Council, Canada to MJW. Back

2 Current address: Department of Forest Sciences, University of British Columbia, 3041-2424 Main Mall, Vancouver, British Columbia, Canada V6T 1Z4. mcintire{at}interchange.ubc.ca . Back

3 Author for reprint requests (tel: 514-398-7851 ext. 7864, FAX: 514-398-7897, marcia.waterway{at}mcgill.ca ) Back


   LITERATURE CITED
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 LITERATURE CITED
 
Aguilar J. M. W. J. Boecklen 1992 Patterns of herbivory in the Quercus grisea x Quercus gambelii species complex. Oikos 64: 498-504[CrossRef][ISI]

Alexander H. M. 1990 Epidemiology of anther-smut infection of Silene alba caused by Ustilago violacea: patterns of spore deposition and disease incidence. Journal of Ecology 78: 166-179[CrossRef]

Alexander H. M. P. H. Thrall J. Antonovics A. M. Jarosz P. V. Oudemans 1996 Population dynamics and genetics of plant disease: a case study of anther-smut disease. Ecology 77: 990-996[CrossRef][ISI]

Antonovics J. 1992 Toward community genetics. In R. S. Fritz and E. L. Simms [eds.], Plant resistance to herbivores and pathogens: ecology, evolution, and genetics, 426–449. University of Chicago Press, Chicago, Illinois, USA

Barton N. H. G. M. Hewitt 1985 Analysis of hybrid zones. Annual Review of Ecology and Systematics 16: 113-148[CrossRef][ISI]

Bernard J. M. 1990 Life history and vegetative reproduction in Carex. Canadian Journal of Botany 68: 1441-1448

Boecklen W. J. R. Spellenberg 1990 Structure of herbivore communities in two oak (Quercus spp.) hybrid zones. Oecologia 85: 92-100[CrossRef][ISI]

Brown S. G. S. Kwan S. Shero 1995 The parasitic theory of sexual reproduction: parasitism in unisexual and bisexual geckos. Proceedings of the Royal Society of London, Series B 260: 317-320[CrossRef]

Burdon J. J. 1987 Diseases and plant population biology. Cambridge University Press, Cambridge, UK

Christensen K. M. T. F. Whitham P. Keim 1995 Herbivory and tree mortality across a pinyon pine hybrid zone. Oecologia 101: 29-36

Cliff A. D. J. K. Ord 1981 Spatial processes: models and applications. Pion, London, UK

Clifford P. S. Richardson D. Hémon 1989 Assessing the significance of the correlation between two spatial processes. Biometrics 45: 123-134[CrossRef][ISI][Medline]

Ericson L. J. J. Burdon A. Wennström 1993 Inter-specific host hybrids and phalacrid beetles implicated in the local survival of smut pathogens. Oikos 68: 393-400

Floate K. D. M. J. C. Kearsley T. G. Whitham 1993 Elevated herbivory in plant hybrid zones: Chrysomela confluens, Populus and phenological sinks. Ecology 74: 2056-2065[CrossRef][ISI]

Foster D. R. G. A. King M. V. Santelmann 1988 Patterned fens of western Labrador and adjacent Quebec: phytosociology, water chemistry, landform features, and dynamics of surface patterns. Canadian Journal of Botany 66: 2402-2418

Fritz R. S. L. Johansson B. Aastrom M. Ramstedt H. Haeggstroem 1996 Susceptibility of pure and hybrid willows to isolates of Melampsora epitea rust. European Journal of Plant Pathology 102: 875-881[CrossRef][ISI]

Fritz R. S. C. M. Nichols-Orians S. J. Brunsfeld 1994 Interspecific hybridization of plants and resistance to herbivores: hypotheses, genetics, and variable responses in a diverse herbivore community. Oecologia 97: 106-117[CrossRef][ISI]

Ginzberg L. R. 1979 Why are heterozygotes often superior in fitness?. Theoretical Population Biology 15: 264-267[CrossRef][ISI]

Gottlieb L. D. 1981 Gene number in species of Asteraceae that have different chromosome numbers. Proceedings of the National Academy of Science, USA 78: 3726-3729[Abstract/Free Full Text]

Graham J. H. D. C. Freeman E. D. McArthur 1995 Narrow hybrid zone between two subspecies of big sagebrush (Artemesia tridentata, Asteraceae). 2. Selection gradients and hybrid fitness. American Journal of Botany 82: 709-716[CrossRef][ISI]

Grant V. 1981 Plant speciation, 2nd ed. Columbia University Press, New York, New York, USA

Gray A. J. R. J. Parsell R. Scott 1979 The genetic structure of plant populations in relation to the development of salt marshes. In R. L. Jefferies and A. J. Davy [eds.], Ecological processes in coastal environments, 43–64. Blackwell Scientific, London, UK

Hartnett D. C. F. A. Bazzaz 1985 The genet and ramet population dynamics of Solidago canadensis in an abandoned field. Journal of Ecology 73: 407-413[CrossRef][ISI]

Hjältén J. 1998 An experimental test of hybrid resistance to insects and pathogens using Salix caprea, S. repens and their F1 hybrids. Oecologia 117: 127-132[CrossRef][ISI]

Holmberg O. R. 1929 De skandinaviska formerna av Carex-gruppen Limosae. Botaniska Notiser 1929: 209-228

Ingvarsson P. K. L. Ericson 1998 Spatial and temporal variation in disease levels of a floral smut (Anthracoidea heterospora) on Carex nigra. Journal of Ecology 86: 53-61

Jarosz A. M. J. J. Burdon 1988 The effect of small-scale environmental changes on disease incidence and severity in a natural plant–pathogen interaction. Oecologia 75: 278-281[CrossRef][ISI]

Jean M. A. Bouchard 1993 Riverine wetland vegetation: importance of small-scale and large-scale environmental variation. Journal of Vegetation Science 4: 609-620

Jermy A. C. A. O. Chater R. W. David 1982 Sedges of the British Isles. Botanical Society of the British Isles, London, UK

Johannesson K. 2001 Parallel speciation: a key to sympatric divergence. Trends in Ecology and Evolution 16: 148-153

Johnson J. B. 1996 Phytosociology and gradient analysis of a subalpine treed fen in Rocky Mountain National Park, Colorado. Canadian Journal of Botany 74: 1203-1218

Jonsson B. O. I. S. Jónsdóttir N. Cronberg 1996 Clonal diversity and allozyme variation in populations of the arctic sedge Carex bigelowii (Cyperaceae). Journal of Ecology 84: 449-459[CrossRef]

Kazuo Y. O. Kazuyuki 1997 Key variables controlling the vegetation of a cool-temperate mire in northern Japan. Journal of Vegetation Science 8: 29-36

Kükenthal G. 1909 Cyperaceae—Caricoideae. Das Pflanzenreich 4: (20)1-824. Engelmann, Leipzig

Kukkonen I. E. Vatanen 1968 Studies on the mechanism of infection and the imperfect stage of Anthracoidea (Ustilaginales). Annales Botanici Fennici 5: 10-16

Lebrun N. F. Renaud P. Berrebi A. Lambert 1992 Hybrid zones and host–parasite relationships: effect on the evolution of parasitic specificity. Evolution 46: 56-61[CrossRef][ISI]

MacDonald S. E. V. J. Lieffers 1993 Rhizome plasticity and clonal foraging of Calamagrostis canadensis in response to habitat heterogeneity. Journal of Ecology 81: 769-776[CrossRef]

Marquis R. J. 1992 Selective impact of herbivores. In R. S. Fritz and E. L. Simms [eds.], Plant resistance to herbivores and pathogens: ecology, evolution, and genetics, 426–449. University of Chicago Press, Chicago, Illinois, USA

Math Soft. 1995 S-Plus for Windows user's manual, version 3.3. Math Soft, Seattle, Washington, USA

McClintock K. A. M. J. Waterway 1993 Patterns of allozyme variation and clonal diversity in Carex lasiocarpa and C. pellita. American Journal of Botany 80: 1251-1263[CrossRef][ISI]

Mitton J. B. 1994 Molecular approaches to population biology. Annual Review of Ecology and Systematics 25: 45-69

Mitton J. B. M. C. Grant 1980 Observations on the ecology and evolution of quaking aspen, Populus tremuloides, in the Colorado front range. American Journal of Botany 67: 202-209[CrossRef]