CYPERACEAE

CYPERACEAE View in CoL View at ENA

Carex : A morphometric analysis of the Carex intumescens complex supporting the recognition of two varieties. Primary authors:H.C.Medford,Derick B. Poindexter,and Alan S.Weakley

Carex intumescens Rudge ( Cyperaceae sect. Lupulinae Tuck. ex J. Carey) has an extensive range across eastern North America, from Newfoundland west to Manitoba, Minnesota, western South Dakota and northeastern Colorado, and thence south to central peninsular Florida and eastern Texas ( Reznicek & Ball 1974; Reznicek 2002; Kartesz 2015).

Within C. intumescens View in CoL , one variety, C. intumescens var. fernaldii L.H. Bailey View in CoL , has been described based on populations from Maine ( Bailey 1893). Compared with typical var. intumescens View in CoL , this taxon purportedly has slimmer and fewer perigynia per pistillate spike, as well as narrower leaves ( Bailey 1893). Fernald (1942) expanded upon this morphological concept and used achene shape as a significant distinguishing feature, attributing elliptic achenes to var. intumescens View in CoL and obovoid achenes to var. fernaldii View in CoL . Confusingly, Fernald (1942) described a form of C. intumescens var. fernaldii View in CoL ( =f. ventriosa ), which resembles typical C. intumescens View in CoL in all characters except for the obovoid achenes he associated with var. fernaldii ( Fernald 1942) View in CoL . Uttal (1971) found an additional feature, the curvature of achene styles, to support recognizing var. fernaldii View in CoL . Biogeographically, all authors interpreted var. intumescens View in CoL as occurring predominantly at lower latitudes and var. fernaldii View in CoL as distributed at upper latitudes.

In contrast to these observations, Reznicek and Ball (1974) found no difference in achene shape between var. fernaldii View in CoL and var. intumescens View in CoL and therefore rejected var. fernaldii View in CoL as a distinct taxonomic entity within C. intumescens View in CoL . Likewise, in our preliminary assessments we found that style morphology (as emphasized by Uttal [1971]) was unrelated to perigynium shape, achene shape, geography, or habitat, while also being difficult to quantify in a replicable way.

Since the description of var. fernaldii , a majority of floristic treatments or studies have treated C. intumescens as an indivisible species, with var. fernaldii synonymized ( Mackenzie 1931 –1935; Small 1933; Gleason 1952; Radford et al. 1968; Godfrey & Wooten 1979; Gleason & Cronquist 1991; Ball & Reznicek 2002; Rhoads & Block 2007; Haines 2011; Weakley et al. 2012; Voss & Reznicek 2012), likely owing to the lack of clarity surrounding its description.A minority of floristic treatments list C. intumescens as having two varieties (Robinson & Fernald 1908; Fernald 1950; Tennessee Flora Committee 2015; Weakley 2020; Weakley et al. 2020a).

It is evident that considerable morphological variation occurs within the species, likely driving the original description of infraspecific taxa. However, distinguishing features and correlated biogeography have not been investigated rigorously enough to confidently circumscribe meaningful taxonomic units.

After years of qualitative field observations throughout the southeast by the second and third authors, coupled with cursory examinations of herbarium specimens, we chose to embark on a morphometric study to potentially clarify taxonomic units within C. intumescens . This need was particularly evident when juxtaposing populations of plants across multiple elevations and physiographic provinces.In the Southern Appalachians specifically, we have noticed that there seem to be two entities with different morphologies correlated with different habitats. Plants with slimmer perigynia (possibly var. fernaldii ) are found in high-elevation non-wetland sites, notably in spruce-fir forests, northern hardwood forests, and grassy balds, compared to plants with more robust perigynia (typical of var. intumescens ), which are found in high- to low-elevation wetland sites ( Uttal 1971; Weakley 2020). Here, we assess the morphological variation within C. intumescens and assign discrete morphological groups appropriate taxonomic ranks.

We analyzed 263 herbarium specimens, which encompassed the full geographic distribution of Carex intumescens , from the following herbaria: MICH, NCU, TEX, and USF. We also examined type material of C. intumescens var. fernaldii (GH) and gathered approximate measurements digitally from the putative holotype of C. intumescens var. intumescens (BM) .To analyze morphological variation, we made 1 discrete, 2 ratio, and 10 continuous morphological measurements ( Table 1 View TABLE ).

We performed a Principal Component Analysis (PCA) in PC-ORD (McCune & Mefford 2011) to elucidate different groups present evident via separation on ordination space. The PCA was based on a correlation distance-based biplot and only used continuous or ratio variables. We excluded from the final PCA characteristics that were not present on all specimens (e.g. subtending bract length, culm height) and other non-informative variables ( Table 1 View TABLE ). We used preliminary PCAs to ascertain groupings and compiled ranges and standard deviations for each entity present ( Table 1 View TABLE ). We used these ranges to refine the groups on a specimento-specimen basis and ran the final PCA.

Two imperfectly separated groups are depicted in our final PCA ( Fig. 1 View FIG ), one of which includes the isotype specimens of C. intumescens var. fernaldii and the other containing the putative holotype of C. intumescens var. intumescens . We found the strongest determining morphological characteristics between these two varieties to be perigynium length-to-width ratio, perigynium width, and perigynium scale length. Two of these relate to perigynium dimensions: var. fernaldii has slimmer perigynium compared to var. intumescens , which is reflected in a higher perigynium length/width ratio and narrower perigynium widths ( Figs. 2 View FIG & 3 View FIG ). This, along with the lower number of pergynia per pistillate spike (measured as max number of perigynia per pistillate spike here, but excluded from PCA as a noncontinuous variable), fits Bailey’s original description of C. intumescens var. fernaldii ( Bailey 1893) . We found an additional characteristic (perigynium scale length) to be reliably useful in making determinations between var. fernaldii and var. intumescens ( Figs. 2 View FIG & 3 View FIG ). However, we did not find any significant trend in achene characteristics, in agreement with Reznicek and Ball (1974).

We also documented a broad range of morphological variation within C. intumescens var. intumescens . Most noticeably, previous literature described C. intumescens var. intumescens with fewer perigynia per pistillate spike and smaller perigynium scales (Rezineck & Ball 1974; Ball & Reznicek 2002) than we have documented. This is important, because some C. intumescens var. intumescens specimens with many perigynium per pistillate spike may be confused with the closely related C. grayi . However, C. intumescens can be distinguished by the glabrous surface of perigynia, whereas C. grayi has a puberulent vestiture visible under magnification (pers. obs.).

We created a map in ArcMap (ESRI 2017) to reflect the geographic range of these entities. Carex intumescens var. intumescens has an extensive range across all of eastern North America. Contrastingly, C. intumescens var. fernaldii is restricted to the northern regions of the range of the typical variety, only extending southward in the high elevations of the Appalachian Mountains ( Fig. 4 View FIG ). This geographic pattern of var. fernaldii is supported by the habitat information provided on herbarium labels. Carex intumescens var. fernaldii is increasingly restricted to high-elevation spruce-fir forests at lower latitudes, while C. intumescens var. intumescens is widespread across a variety of habitats.

We did not find evidence for a clear northern/southern latitudinal separation, like that which Fernald (1942) proposed. This lack of the latitudinal separation is in agreement with Reznicek and Ball (1974). Carex intumescens var. intumescens is widespread throughout ( Fig. 4 View FIG ) and exhibits a broad range of morphological variation. Similarly, C. intumescens var. fernaldii is not restricted to northern latitudes ( Fig. 4 View FIG ). However, it may appropriately be considered a “northern ” variety since it extends south in only high-elevation habitats typically mimicking northern latitudes, such as spruce-fir forests (Schafale 2012). Reznicek and Ball (1974) discredited a geographic pattern present in either variety, but they primarily looked at northern material, and this pattern was not apparent until southern material was examined.

The imperfect morphological separation of the two entities ( Fig. 1 View FIG ) suggests that either no taxonomic recognition should be given to them, or that a lower rank (e.g., variety) is warranted rather than a higher rank (i.e., species). However, the two entities do exhibit a strong geographic and habitat separation, especially in the southeastern United States. Var. intumescens is widely and generally distributed, and seemingly found only in wetlands.In contrast, var. fernaldii appears to be restricted to high-elevation sites in the Southern Appalachians that are moist but not wetlands ( Fig. 4 View FIG ). By not recognizing these morphologically and ecologically differentiated entities, we lose valuable understanding of the biological diversity in this complex.

Our awareness of cryptic and semicryptic species is increasing with modern taxonomic studies, which employ tools like high-throughput genetic analysis. “Cryptic” refers to taxa that warrant specific rank but differ (only or primarily) on genetic or micromorphological levels ( Bickford et al. 2007). It is completely possible that the two entities discussed here represent semicryptic species. To address this, data from genetic analyses could be compared with an expanded set of morphological characters. This would help conclusively ascribe biologically accurate ranks for these two entities. However, until such study is done, it seems best to conservatively represent these entities at varietal rank. This maintains their names and taxonomic concepts but avoids overstating their taxonomic status, pending further study. In NatureServe conservation ranks, we consider both taxa to warrant G5T5 rangewide ranks: The species as a whole is G5, and each variety is T5: Globally Secure.