A Review of North American Isoetes engelmannii (Isoetaceae) complex hybrids, including the description of I. × fernaldii, hyb. nov. and I. × karenae, hyb. nov

The Isoetes engelmannii complex of eastern North America consists of 30 taxa including 13 named species. Nine of the 17 hybrids within the complex (the largest group of Isoetes hybrids in the world) have been formally described. Those named hybrids are reviewed here in light of recent additions to and enhancements of the morphological and cytological evidence employed in their original description. The pedigree of three of these, I. ×brittonii, I. ×bruntonii and I. ×carltaylorii, is updated and clarified. Formal descriptions are proposed for two additional taxa: I. ×fernaldii, hyb. nov. (I. engelmannii × I. hyemalis) and I. ×karenae, hyb. nov. (I. appalachiana × engelmannii). The potential for a further eight hybrid combinations to occur in the wild is also addressed.

The identification of sterile hybrids has been important in determining the distinctiveness of these and other Isoetes species by confirming the existence of interspecific genetic barriers (Taylor et al. 1985;Brunton 2015;Taylor et al. 2016). Indeed, hybrids are recognized as an important element of the evolution and taxonomy of vascular plants in general (Plume et al 2015;Gianguzzi et al. 2017;Goulet et al. 2017). Within the Isoetaceae, hybrids also commonly present opportunities for the development of polyploid species through alloploidy (Hickey et al. 1989;Troia et al. 2016;Brunton and Troia 2018).
The existence of hybrid diversity within the Isoetes engelmannii complex has been confirmed through recognition of a suite of often subtle but consistent and usually distinctive morphological features, often supported by cytology. Features and characteristics of the megaspores and microspores, found within sporangia at the base of fertile leaves, are critical to the identification of such plants. These include ornamentation patterns reflecting features of both putative parents.
Hybrids also have polymorphic, aborted megaspores and microspores, usually in high proportion, and a wide range of spore sizes within individual sporangia (Taylor et al. 1985;Hickey et al. 1987;Musselman et al. 1996;Britton and Brunton 1996;Troia & Greuter 2014;Brunton 2015). Megaspores of hybrids often present uniquely congested ('brain coral') ornamentation and/or are fused together as 'dumbbell-shaped' megaspores (Jeffrey 1937;Britton 1991;Taylor et al. 2016). Hybrid plants often display hybrid vigour and are larger than putative parents at the same site (Britton 1991;Musselman et al 1995). Cytology is an important determinant if the putative parents have different ploidy levels (Hickey et al. 1987;Taylor and Luebke 1988;Musselman et al. 1997).
Isoetes hybrids in the wild grow with at least one and almost always with both putative parents. They also are usually found in sites especially well suited to the mixing of spores, such as downstream of intrusions into river currents (e.g. below rapids) or in seasonally flooded creek-side swales (Britton and Brunton 1989;Britton and Brunton 1996;Musselman et al. 1996).
Isoetes species of environmentally stressed habitats often exhibit polymorphic spores reminiscent of sterile hybrids but still reflect only the ornamentation features  of a single species. This is most commonly encountered in species of ephemeral wetlands such as temporary pools on bedrock outcrops (Haefner and Bray 2005;Brunton 2015). Environmentally induced spore polymorphism is rare in true aquatic species, however, and is almost always expressed in such plants by only a small number of individual spores. Dumbbell-shaped megaspores are apparently not evident in plants experiencing environmentally induced polymorphism (pers. obs.).
Recognition and clarification of many of the hybrid taxa in the Isoetes engelmannii complex occurred in a relatively short period of time in the 1990s -early 2000s. Subsequent cytological, taxonomic and morphological data have substantially altered initial conclusions for some of these taxa. Recent molecular studies (e.g. Shafran et al. 2018) indicate multiple origins in a number (most?) of the recognized polyploid Isoetes of eastern North America. This is perhaps to be expected within one of the most ancient and widespread vascular plant groups on Earth (Larsen and Rydin 2016;Wood et al. 2019). Especially so when many species within it, particularly amphibious and semi-terrestrial taxa such as some of those addressed here, consist substantially or entirely of widely dispersed populations in highly stressed environments, these habitat patches having persisted unchanged for thousands or in some cases, even millions of years (Pfeiffer 1922;McVaugh 1943;Matthews and Murdy 1969;Taylor et al. 1993;Heafner and Bray 2005;Brunton and Troia 2018).
Shafran (2019) states that "using a lineage-based species concept may require the recognition of ca. 50 new species of auto-and allopolyploid Isoëtes in eastern North America". There is a need for the systematic investigation of morphological, ecological and phytogeographic lines of evidence to provide consistent and repeatable collaboration with these molecular findings. This would demonstrate that genetic barriers exist between some of these newly distinguished 'species', and also would indicate how such taxa can be reliably distinguished in the field. In the absence of such collaboration, however, in the present study we hold to a conservative and inclusive (more traditional) taxonomic interpretation of Isoetes speciation. To do otherwise would preclude the identification of any sterile hybrids -indeed, the identification of many if not most populations of a particular polyploid Isoetes species -outside of the type location.
Clarifying the existence and status of sterile hybrids within already recognized Isoetes species will assist in future comprehensive, multifaceted investigations of relationships within the regionally important Isoetes engelmannii complex in particular and Isoetes in general.

METHODS
This study is based on extensive field experience of the first author undertaken since the early 1990s throughout the range all the species addressed. Over 2,250 voucher specimens of the taxa discussed here have been examined. The largest number of vouchers studied in over 30 herbaria are in CAN, DAO, DFB, FLAS, MICH, NYS, OAC and PH (herbarium acronyms of Theirs, continuously updated).
An extensive library of Scanning Electron Microscope (SEM) images prepared by D. M. Britton before 2007 (prepared with the methods described in Brunton and Britton 2006) of microspores and megaspores of virtually all subject taxa, was reviewed. Additional SEM images have been produced since 2017 by the authors. For these new images, air dried spores were attached to SEM stubs by means of adhesive carbon discs. These were sputter coated with a gold / palladium alloy (Au/Pd) and examined with a 2017 model FEI Apreo SEM (at 15 kV and 25 pA, with a working distance of 10 mm and a spot size of 6).
The physical characteristics of specimens, especially their megaspore form, ornamentation pattern and size, were examined through a Leica Wild M3B [light] dissecting microscope at 40x magnification, with the aid of an in-mount graticule (ocular micrometer) for measurements. Megaspore sizes reported for individual specimens represent the average width (across the equatorial region) of at least 10 spores. Comparable microspore measurements are based on the average of 20 longitudinal measurements taken from SEM images of clusters of spores.
Cytological determinations of hybrid specimens reported here were made by D. M. Britton, following the techniques described in Britton (1999, 2006). Plants from sampled populations were grown in distilled water in a growth chamber. The developing root tips were excised and pre-treated in aqueous paradichlorobenzene (PDB) at room temperature for four hours. They then were washed in distilled water, fixed in acetic alcohol (3:1 absolute ethyl alcohol to glacial acetic acid) for 30 minutes or more, hydrolysed in Warmke's solution (1:1 concentrated HCl to absolute ethyl alcohol) for 7-10 minutes at room temperature, and stained in leucobasic fuchsin (Feulgen) for two hours. The meristems were squashed under a cover glass in 45% acetocarmine stain and examined. With both putative parents having predominantly reticulate megaspore ornamentation patterns, the range of distinctive features available for detecting this hybrid is reduced from that often available for the detection of other combinations. Nonetheless, there are several features of the Dinwiddie County material that confirm its distinctiveness.

Nine sterile hybrid combinations in the
The irregular shape, ornamentation and size of megaspores of Rowanty Creek hybrid plants are conspicuous ( Figures 4A-4D). Megaspore size (514.9 µm, 1 SD 52.55 µm, N=28) was determined to be intermediate between that of diploid parent I. engelmannii (460-500 µm) and the tetraploid parent I. appalachiana (520-600 µm) (Brunton and Britton 1997;Brunton 2015). Most megaspores are distorted in shape ( Figures 4A-4B), with some of the fused, dumb-bell shaped megaspores that are diagnostic of hybrids, also being evident ( Figure 4B). Relatively few regularly globose megaspores that are typical of non-hybrid Isoetes plants were evident in the suspected hybrid plants.
Microspore ornamentation is most similar to that of the smooth-spored diploid parent Isoetes engelmannii ( Figures 4E-F), with only a subtle indication of the tuberculate ornamentation pattern of the tetraploid parent I. appalachiana. In keeping with this strong I. engelmannii expression in the hybrid, microspore size (25.4 µm, 1 SD 1.30 µm, N=40) was determined to be comparable to that diploid parent (25.5 µm) and substantially smaller than 30.4 µm length of tetraploid I. appalachiana microspores (Brunton and Britton 1997). A length of approximately 27-28 µm would have been more consistent with the size recorded with other triploid hybrids. This inconsistency, however, is believed to reflect the limited of microspore data (one plant) available at the Rowanty a mount Creek location.
No fresh or live plants were available for genomic investigation or genomic analysis. That is unfortunate since the reflection of the different ploidy levels of the putative parents would have offered stronger support of the taxonomic hypothesis presented here. This necessitates the less desirable option of inferring ploidy level from spore size.
Other occurrences of this hybrid were detected in herbarium specimens showing comparable morphological characteristics (including aborted, polymorphic spores) from elsewhere in the sympatric range of the putative parents. In each case, the suspected hybrid was either a single huge plant (suggesting selective collection of individuals demonstrating such hybrid vigour) or mixed collections with one or both of the putative parents. The following binomial is proposed for this previously undescribed taxon:   appalachiana and I. engelmannii; Megaspores -variable in size, most aborted (misshaped) with well-formed spores of intermediate size 514.9 µm (1 SD = 52.55 µm) in diameter and with densely, irregularly-reticulate ornamentation pattern formed by variable and raggedcrested muri; fused, dumb-bell shaped spores also present; Microspores -plain perispore surface obscurely verrucose, 25.4 µm (1 SD = 1.30 µm) in length; Habitat -in periodically flood-scoured bank swales along deciduous swamp forest creeks; Cytology -unconfirmed (inferred 2n=3x=33 from megaspore size).

Etymology
The taxon in named in honour of Canadian field botanist Karen L. McIntosh of Ottawa, Ontario, who not only discovered the type population but has provided valuable contributions to our understanding of numerous Isoetes taxa during 30+ years of field investigations with the first author across North America.  Figure 8D). It also has microspores with obscurely verrucose surfaces ( Figure 4F) rather than apparently obscurely broad-based spiny perispore of I. engelmannii ×hyemalis ( Figure 8F).

Paratypes
Given the extent of overlapping distributions of its putative parents, Isoetes ×karenae can be expected to occur in mixed Isoetes populations over a large area in the eastern United States. Additional occurrences may be discovered at sites from central Pennsylvania southward to at least South Carolina and northern Georgia.
Hybrid 2) Isoetes appalachiana × I. hyemalis (I. ×bruntonii D. Knepper & L.J. Musselman) This taxon was first described as a cytologically supported triploid hybrid, thought to represent Isoetes engelmannii × I. hyemalis (Musselman et al. 1996), prior to recognition that I. engelmannii (s.l.) included tetraploid I. appalachiana (Brunton and Britton 1997). Isoetes ×karenae (above) and I. ×bruntonii type locations are in adjacent creek systems in Dinwiddie County, Virginia. It is at a site well suited to the mixing of Isoetes spores and the generation of hybrids, where trees fallen into a woodland creek created back eddies where the spores of the locally abundant I. hyemalis could mix with those of other taxa ( Figure 5). Isoetes appalachiana (e.g. D. F. Brunton Taylor et al. (1985), and I. ×foveolata A.A. Eaton, pro sp. sensu Taylor et al. (1985) all exhibit more congested megaspore ornamentation. Megaspores from the holotype (ODU), an Isotype (DFB) and later cytologically determined tetraploid topotypes (D.F. ). This is a larger than the megaspores size found in triploid hybrids such as I. ×karenae (above) or I. engelmannii × hyemalis (below), which average approximately 500 µm. No microspores from type material were available for examination in the present study.
No diploid plants, nor any triploid plants since the original cytotype report (Musselman et al. 1996), have been reported at the Isoetes ×bruntonii type location. The isotype in DFB consists of two plants, one hybrid and one I. appalachiana. It is not possible to compare morphological data from the reported triploid cytotype against that of tetraploid hybrids later found at the type location as no voucher was retained of that cytotype (R. Bray, pers. comm.).
While it is technically possible that a triploid plant was originally present and has not reoccurred at the site, that 'one-off' occurrence would be contrary to our experience in all other regularly-producing hybrid sites. It seems most likely that either an error was made at the time of the original chromosome count for the cytotype or that the specimens did not originate from the Isoetes ×bruntonii type location. Regardless, existing cytological vouchers from the type location are exclusively tetraploid and morphologically consistent with the type material.
While Isoetes ×bruntonii is currently known only from eastern Virginia, it is reported (as I. ×brittonii D.F. Brunton & W.C. Taylor) to be frequent there (Musselman & Knepper 1994). Given the relative wide area of overlap between its putative parents, however, this hybrid could also be expected to occur in mixed Isoetes populations in North Carolina, South Carolina or even southern Georgia/ northern Florida.
Isoetes × brittonii remains a rare taxon, perhaps not surprisingly considering the relatively small area of overlap in the distribution of its putative parents (Brunton and Britton 1997;Brunton and McNeill 2015). It can reasonably be expected to occur elsewhere, however, in at least central and eastern Pennsylvania and adjacent New Jersey. It is presently known only in Pennsylvania from Lycoming County: West Branch Susquehanna River opposite Jersey Shore (W. The cytological criteria for hybridization in Isoetes were first described by Jeffrey (1937) who stated that I. eatonii [sic] was 'under strong suspicion of hybrid deriva-tion'. This taxon was one of the first hybrid Isoetes taxa to be formally recognized (Taylor et al. 1985;Hickey et al. 1987). Diploid I. ×eatonii plants can be exceptionally large with leaves commonly over 30 cm long and some exceeding 60 cm in length (Dodge 1897). It is a true aquatic of freshwater ponds and streams, occasionally occurring in such large numbers as to hide the presence of one or both of its putative parents (Eaton 1900;Kott and Bobbette 1980). That abundance delayed recognition of its hybrid nature (Taylor et al. 1985). Its diploid hybrid status has been confirmed, however, by the morphological, ecological or cytological evidence coming forward in recent decades.
Isoetes ×eatonii occurs in the New England region of the United States and is disjunct to the west in southern Michigan (Palmer 2018) and southern Ontario, Canada . The population decline of parent I. engelmannii (s.str.) in this region ) has resulted in a comparable decline in hybrid occurrences as well.
Hybrid 5) Isoetes engelmannii (s.str.) × I. hyemalis (hyb. nov.) This hybrid combination is known from a single site in eastern Virginia where it persisted, presumably through a series of independent propagation events, for over 30 years. A large (40 cm long), mature plant of at least several years of age was collected in 1965 and  The leaves of the hybrid plants have the gross appearance of their putative parents and a velum coverage of their sporangium of approximately 17-20% (N=4). This is centred within the 10-25% range of velum coverage of the putative parents (Brunton et al. 1994;Brunton and Britton 1997). The hybrid also exhibits conspicuously polymorphic, aborted megaspores with congested, irregular ornamentation patterns ( Figure 8B). The reticulate pattern of Isoetes engelmannii is the dominant expression on megaspores of these hybrid plants ( Figure 8C), but appear to be of a less regularly 'honeycomb' pattern than is evident with that parent ( Figure 1B). Megaspores of the hybrid also have a more congested, irregularly-reticular ornamentation on the distal side than is expressed in I. engelmannii and individual ornamentation 'cells' are conspicuously smaller ( Figure 8C, 8E). The variation amongst megaspores of the hybrid includes some having more broken-reticulate pattern of muri (walls) interspersed with very short almost tuberculate muri on the proximal side ( Figure 8C). Others exhibit an ornamentation pattern of elongated, non-reticulate cells (Figures 8D) more suggestive of tetraploid putative parent I. hyemalis ( Figure 9A, 9B).
The following binomial is proposed for this undescribed taxon:

Etymology
The epithet acknowledges the contribution made by Merritt L. Fernald (1873-1950 to our knowledge of the distribution of this genus in the southeastern United States in general and Virginia in particular. His comprehensive field investigations in the 1930s and 1940s, especially on the species-rich coastal plain (Hugo and Ware 2012), laid the foundation upon which subsequent regional floristic and taxonomic investigations such as Pease (1951) and Musselman and Knepper (1994)  Isoetes ×fernaldii is similar in appearance to tetraploid I. ×bruntonii. Its smaller spore size, however, reflects its triploid status. It is even more similar in appearance to fellow triploid I. ×karenae. Compared with the latter, however, I. ×fernaldii exhibits megaspores with thinner, higher muri ( Figures 8C, 8D vs. Figures 4C, 4D) and more pronounced suture ridges with a greater abundance of tiny spines along their flanks ( Figures 8D vs. Figures 4C), and more evidently spiny microspores.
There is considerable overlap in the distributions of Isoetes hyemalis and I. engelmannii in the eastern United States. Accordingly, additional localities for this sterile hybrid are expected to be found, particularly in Virginia and the Carolinas.
The hybrid is known only from a single freshwater lake in New Jersey where individuals were found infrequently in mixed populations with its putative parents, most frequently with Isoetes macrospora. With few areas of overlapping distribution by the putative parents, the possibility of additional populations of I. ×fairbrothersii being found are limited to northern New Jersey and adjacent eastern New York. Occurrences may also be possible, however, within the disjunct western populations of I. engelmannii in Michigan or Ontario (Cody and Britton 1989;Taylor et al. 1993, Tryon andMoran 1997;Palmer 2018;Weldy et al. 2019). Isoetes ×carltaylorii was originally described as the cytologically confirmed triploid hybrid between diploid I. engelmannii and tetraploid I. acadiensis L. Kott (Musselman et al. 1997). The latter is now considered to represent I. tuckermanii A. Braun ssp. acadiensis (L. Kott) D.F. Brunton (Brunton 2018)). The hybrid is known only from the tidal reaches of rivers flowing into the western side of the Chesapeake Bay in northeastern Virginia. It is unique amongst I. engelmannii complex hybrids in having a conspicuously subdued, reticulate megaspore ornamentation pattern of low, broad (almost vermiform) muri.
The tetraploid parent appears not to be Isoetes tuckermanii ssp. acadiensis but I. riparia var. reticulata (A.A. Eaton) G. Proctor, an uncommon (now rare?) endemic of emergent tidal shores in freshwater and brackish marsh habitat on the Chesapeake Bay (Proctor 1949). Isoetes riparia var. reticulata shares many morphological similarities with I. tuckermanii ssp. acadiensis (Brunton 2015), particularly in having a reticulate megaspore ornamentation pattern of low, broad (almost vermiform) muri. Isoetes riparia var. reticulata has consistently smaller megaspores with thinner, usually more densely arranged muri, however, and this intricacy is reflected in the megaspore ornamentation of the hybrid. In addition, I. tuckermanii ssp. acadiensis is a non-tidal freshwater species that extends no further south than Massachusetts, 650 km from the nearest I. ×carltaylorii site (Taylor et al. 1993;Musselman et al. 1997;Brunton 2018).
Isoetes ×carltaylorii is not likely to be found outside of the Chesapeake Bay of Virginia and the adjacent District of Columbia, the only area where the putative parents are sympatric.
Hybrid 8) Isoetes engelmannii (s.str.) × I. tuckermanii (I. ×foveolata, pro sp.) Along with Isoetes × eatonii, this was thought to be a distinct species when described in the 19 th Century (Eaton 1898) and is one of the first Isoetes taxa to be recognized as a hybrid (Taylor et al. 1985;Hickey et al. 1989). It is a true aquatic, found in ponds and rivers in southern New Hampshire and apparently has not been found in the wild for over a century. Megaspore ornamentation in this triploid is congested, presenting a particularly bold example of the 'brain-coral' pattern typical of Isoetes hybrids (Britton 1991) (Figures 10A, 10B).
Given the relatively large area of southern New England (Massachusetts, Connecticut, Rhode Island, New Hampshire) where its (at least formerly) common putative parents are sympatric, the rarity of this taxon is unexpected. This may reflect an ecological separation of the parents, Isoetes tuckermanii apparently occurring in more acidic substrates and oligotrophic water than I. engelmannii (pers. obs.).
Hybrid 9) Isoetes engelmannii (s.str.) × I. valida (G. Engelmann) W.N. Clute (I. ×altonharvillii) This cytologically confirmed diploid typically exhibits the classic features of Isoetes hybrids including hybrid vigour, brain coral megaspore ornamentation patterns, fragmented and aborted spores and occurrence in ponds, seepages or along river shores with one or (usually) both of its putative parents (Musselman et al. 1995). The ca. 50% velum coverage of the sporangia reported for the hybrid is the same as that determined for I. valida . This is significantly greater than would be expected of a hybrid also involving I. engelmannii which has a much narrower (10-15%) velum coverage Britton 1996, 1997). Subsequently examined specimens of I. ×altonharvillii from Tennessee, North Carolina and South Carolina (DFB), however, average 26.6% velum coverage (N=28, six plants), which is consistent with expectations for an I. engelmannii × valida hybrid.
Isoetes ×altonharvillii has been found in mixed Isoetes populations in ponds and streams adjacent to and within the Appalachian Mountains from Delaware to northeastern Alabama, occasionally growing in relatively high elevation sites.

Potential Hybrid Combinations
In addition to the nine formally named hybrid taxa reviewed above, there are eight hybrid combinations which remain to be confirmed in the Isoetes engelmannii complex. Most of these involve species with restricted range and/ habitat requirements and thus have limited potential for interaction with other Isoetes.  Table 1 lists and identifies the expected habitat, areas of occurrence (reflecting the sympatric range of the putative parents) and ploidy level of these presently unconfirmed hybrid combinations. The likelihood of discovery for each hybrid combination is subjectively appraised based on the apparent availability of suitable habitat and condi-tions for hybrid generation within the sympatric range of putative parents.

DISCUSSION
The complicated relationship within the diverse Isoetes engelmannii complex is depicted in Figure 11. Thirty taxa are represented, 13 of which are species, nine representing named sterile hybrids (including the two described here) and eight (listed in Table 1) being theoretically possible hybrids.
Specimens apparently representing some of the unconfirmed hybrid combinations noted in Table 1 are known. Herbarium specimens that appear to represent triploid Isoetes engelmannii × septentrionalis from Connecticut, New Jersey, Pennsylvania, New York and Maryland, for example, are deposited in MICH, MIL, PH and PSU. Most of these are fragmentary, pre-1950s specimens with limited site data, however. Confirmation of the natural occurrence of that and other possible hybrid combination requires stronger supporting evidence. Other vouchers suspected to represent unnamed I. engelmannii complex hybrids are old, fragmentary and/or ambiguous. They also are deemed insufficient to independently support formal designations.
In addition to enumerating the diversity and status of hybrids within the Isoetes engelmannii complex, the present study illustrates that the identification of such plants by morphological features alone is relatively straightforward. A suite of reliable morphological indi-cators of hybrid status has emerged from the investigation and description of over two dozen Isoetes hybrids in North America (Taylor et al. 1993;Musselman et al. 1994;Brunton and Britton 1999;Brunton 2015;Taylor et al. 2016), Japan (Takamiya 1999), eastern Russia (Mocholova et al. 2015, western Europe (Brunton et al. 2020) and India (Singh at al. 2018) over the last 25 years.
The determination of the parentage of hybrid plants in herbarium specimens on the basis of morphological features alone can be difficult. Indeed, without fully mature specimens and information on the associated Isoetes growing with a suspected hybrid, or in cases where voucher material is fragmentary or otherwise poorly preserved, identification may be impossible. This is particularly the case in differentiating hybrids between taxa of equal ploidy levels (and thus with spores of approximately the same size) and/ or with hybrid taxa sharing one putative parent, such as I. ×karenae and I. ×fernaldii. In these and most situations, the most compelling direct evidence of paternity can be the identity of the Isoetes species with which hybrid plants are growing.
Alloploidy involving the doubling of chromosome of interspecific hybrids is a common route of speciation in pteridophytes (Sigel et al. 2019) and is seen as an important (the most significant?) means for generating the current diversity of polyploid Isoetes species (Hoot 2004;Troia et al. 2016;Schafran et al. 2018). Accordingly, aligning important molecular indications of diversity with consistent phytogeographic, physiological, ecological and morphological lines of evidence is critical to the development of a coherent taxonomic understanding of