Asymmetric, biraphid diatoms from the Laurentian Great Lakes

This taxonomic account of light micrographs from the coastal Laurentian Great lakes contains taxa from the diatom genera Amphora, Halamphora, Cymbella, Cymbopleura, Delicatophycus, Encyonema, Encyonopsis, Reimeria, Gomphonema, Gomphosphenia, Gomphonella, Gomphosinica, and Gomphoneis. A total of 207 samples of surface sediment and periphyton collected from 106 wetland, high-energy, embayment, and deeper nearshore locales are represented. Light micrographs of 154 taxa are presented. Of these, 76 could not be fully identified as known taxa from the existing literature and so are given tentative names, numbers or conferred assignments. Lake and habitat specificity, modeled autecological optima for phosphorus and chloride, and tolerance to anthropogenic stressors are described for 39 of the more common taxa.


INTRODUCTION
Despite being the largest surface freshwater system in the world, The Laurentian Great Lakes diatom flora is poorly known. Great Lakes diatom studies before the mid-20 th century were rare (Thomas & Chase, 1886;Vorce, 1881;Skvortzow, 1937), though diatom accounts increased in the late 1960s with Eugene Stoermer and his colleagues' ecological assessments supporting water quality conservation. The earliest work focused on Lake Michigan (Stoermer & Yang, 1969;Stoermer & Ladewski, 1976) and eventually led to more detailed reports from extensive field efforts (Stoermer & Yang, 1971;Stoermer, 1978;Stevenson & Stoermer, 1978;Kreis & Stoermer, 1979;Kociolek & Stoermer, 1990Theriot & Stoermer, 1984). Eventually this work led to retrospective interpretations through paleoecological assessments of fossil diatom records (Stoermer, Wolin & Schelske, 1993). Stoermer's use of diatoms as powerful inferential tools in the Great Lakes paved the way for their use in large environmental programs, including the Great Lakes Environmental Indicators (GLEI) project (Niemi et al., 2006), and the multi-decadal Great Lakes Biological Monitoring Program led by the USEPA's Great Lakes National Program Office (Barbiero et al., 2018). In 2000, GLEI was initiated to focused on development of indicators of coastal conditions in the lakes. Diatom collections from coastal habitats were used to create diatom-based indicators of ecological stress (Danz et al., 2005;Reavie et al., 2006). Diatom evaluations using light microscopy (LM) were based on ornamentation of the siliceous cell wall. To date, applied works from these taxonomic assemblages have included inference models for phosphorus and coastal stressors, multimetric indices, and habitat assessments Sgro et al., 2007;Reavie, 2007;Brazner et al., 2007;Reavie et al., 2008;Kireta et al., 2007). So far, iterative publications of Laurentian Great Lakes diatoms have covered 'centric' and 'araphid' taxa (Reavie & Kireta, 2015), monoraphid taxa (Reavie, 2020), Navicula Bory (Reavie & Andresen, 2020) and the non-Navicula symmetric taxa (Reavie, 2022). In this article I aim to illustrate the asymmetric, biraphid taxa observed during the GLEI project, summarizing their autecology when possible.
Field sites ( Fig. 1) were sampled from June to September 2002 and May to August 2003 (Reavie & Kireta, 2015). In addition to diatom samples, a suite of environmental measurements was collected at each sample location, and a detailed account of these parameters is provided by Reavie et al. (2006). Benthic and sedimented diatoms were sampled from natural substrates from 0.5 to 3 m depth. Additional surface sediment samples were collected from nearshore locations at a 30-m depth from the USEPA's research vessel Lake Explorer. Surface sediments were sampled using a 6.5 cm diameter push corer and core tube. Sediments were extruded in the boat or on shore, and the top 1 cm of sediment was carefully removed using a spoon and/or spatula. In areas where coring was not feasible, a 'petite' PONAR sampler was used to collect unconsolidated bottom substrates, or rocks were carefully collected by hand. Approximately 1 cm of surface sediments from PONAR samples was removed using a spoon and/or spatula. The surfaces of rocks and pebbles were scrubbed clean with a small brush or plastic knife and collected in vials as epilithic samples. All samples were iced at 4-6 C until processing. Approximately 75% of sites were cored, 13% required PONAR grab samples and 12% relied on epilithic samples collected by hand. The full set of sample locations, sample types and associated environmental data are provided in Table S1.
Sample preparation and analysis (Reavie & Kireta, 2015) In the lab, subsamples were taken from homogenized sediment samples, and the diatom remains were cleaned using concentrated nitric or hydrochloric acid, or 30% hydrogen peroxide. Samples were digested in a water bath (85 C) for 1 h. Samples were allowed to cool and settle at room temperature for 24 h and then were centrifuged at 1,800 RPM for 10 min. The tubes were aspirated, refilled with deionized water and shaken to break up the pellet. This centrifugation process was repeated five times. Four microscope slides were prepared for each sample using the Battarbee (1986) method. Diatom assessments for the GLEI project relied on light microscopy for timely data collection. For each sample, 400 diatom valves were counted along random transects at 1,000× magnification using oil immersion microscopy. Counts were made continuously along transects as wide as the field of view until sufficient valves were counted. Specimen photography did not follow a regimented protocol; digital photographs of diatom valves generally occurred as new taxa, or variations thereof, were encountered. Species were identified to the lowest taxonomic level possible using numerous diatom checklists, journal articles and iconographs. The most recently published nomenclature was used as much as possible for identifications (e.g., Guiry & Guiry, 2022;Spaulding et al., 2022).
Photographic sessions involving scanning and photographing specific genera also occurred in preparation for the many taxonomic workshops held throughout the project period (Reavie & Kireta, 2015;Reavie & Andresen, 2020). Photographs were always collected at the highest magnification possible (1,000× and higher) using standard brightfield, or employing differential interference contrast or oblique light path interference. Photography occurred at three primary locations: The University of Minnesota Duluth, John Carroll University, and the University of Michigan. Because of the variation in locations and equipment, a range of gray tones are observed across images. Although this article does not provide high-resolution diagnostic images that would be acquired through scanning electron microscopy, it is hoped that it will be used by taxonomists interested in species occurrences and distributions in the Great Lakes. Preserved material and prepared microscope slides are stored in the Natural Resources Research Institute's diatom collection at the University of Minnesota Duluth.

RESULTS, DISCUSSION AND TAXONOMY
As portions of the diatom flora for the coastal Laurentian Great Lakes are iteratively published, it has become clear that at least one-third of the species have never been adequately described. Even dominant diatom species in the Great Lakes phytoplankton have been described only recently after decades of incorrect identifications (Alexson et al., 2018Van de Vijver et al., 2021); a similar lack of understanding is observed in the highly diverse benthic communities from the coastal Great Lakes. For this manuscript suitable photos were acquired of 154 of the~2,200 diatom taxa encountered in GLEI samples. Unlike most existing diatom monographs, imperfect specimens such as broken valves or those with plane-of-focus issues are sometimes presented to account for occurrence of a taxon and at least some morphological information. Taxonomic accounts herein vary based on their abundance and need for documentation. Species that sufficiently meet the parameters of previously published accounts are afforded fewer details than uncertain or unknown taxa, which require greater details on how they differ from known taxa. Diagnostic information for most taxa includes the observed ranges of valve length, width and stria density. We acknowledge that the measured specimens represent those that were observed during the environmental assessment, and taxon-focused assessments (such as collection and analysis of additional material where a taxon was present) were not performed. Hence, the diagnostic information provided for less common taxa likely does not capture the full ranges present in the Great Lakes, and instead these values were used to provide evidence for linking identifications to previously published accounts. The 'cf.' qualifier (="confer", meaning "compare with") is used frequently because many taxa did not adequately fit published accounts. I do not establish new taxa because of uncertainty about whether a given difference reflects variability in an existing species, and so additional specimen observations are recommended. Despite decades of diatom work in the Laurentian Great Lakes, the lack of viable assignment to known species is a result of past reliance on (mostly) Eurasian concepts, such as the many published accounts from European authors. Now, we are fortunate to have an accelerating publication of North American diatoms, such as through the open-access Diatoms of North America portal ; cited multiple times herein). Despite the many uncertainties in my species descriptions, this specific account of the asymmetric diatoms should provide a floral basis for future diatom-based assessments in the Great Lakes. Many of the diatom taxa described herein are not found in existing literature.
A total of 39 taxa were sufficiently abundant to generate environmental characteristics . These environmental data are called out in the following taxonomic accounts. Slides and freeze-dried materials are retained at the Natural Resources Research Institute (NRRI).

Systematic account
As for the methods, the following is again closely paraphrased from Reavie (2022). Photographs are linked to taxonomic and autecological descriptions. In cases where specimens could not be reconciled with existing literature, a few approaches were followed (Reavie, 2022). For species resembling known taxa but differing slightly (e.g., slightly lower or higher stria density, slightly smaller or larger valve), a 'cf.' qualifier is applied. Sometimes a question mark (?) is associated with ambiguous photographs with uncertain affinity but enough similarity to the described taxon for tentative consideration. Sufficiently unique taxa with poor correspondence to existing published records were assigned 'sp.' or a numeric identifier given at the time of GLEI sample assessment. In many uncertain cases I include recommendations for future scanning electron microscopy (SEM) assessment. Ranges of morphological parameters in Great Lakes specimens are provided relative to published accounts; ranges of those parameters from previously published account are provided and cited. Environmental information for taxa: Environmental characteristics were quantified for common taxa (Reavie, 2020(Reavie, , 2022Reavie & Kireta, 2015;Reavie & Andresen, 2020). Taxa were considered common if: they occurred in at least five samples with greater than 1% relative abundance in at least one of those samples; or they represented more than 5% relative abundance in at least one sample. Autecology for each of these common taxa is presented in a summary diagram . A histogram presents habitat affinity according to the relative frequency the taxon was encountered in each geomorphic habitat (high-energy (HE), embayment (EM), riverine wetland (RW), protected wetland (PW), coastal wetland (CW), open water nearshore (NS); Kireta et al., 2007), based on samples in which the species was observed. This diagram is intended to depict whether taxa have habitat specificity in our samples or if one should expect to encounter a taxon across a wide range of physical conditions. A second histogram illustrates the relative occurrence of each taxon in the five lakes (Superior (SU), Michigan (MI), Huron (HU), Erie (ER), Ontario (ON); Kireta et al., 2007) incorporating standardized weighting by the number of samples collected in each lake. Again, this relative representation of occurrence is based on samples in which the given species was observed. When a taxon was observed in at least 10 locations, water quality optima for total phosphorus (TP) and chloride (Cl) were presented along vertical bars representing the measured water quality gradient for all GLEI samples. These autecological data for common taxa were based on statistical evaluations of species-environmental relationships (covered in greater detail by Reavie et al. (2006) and Kireta et al. (2007)). Species optima for these variables were estimated using weighted averaging regression and calibration, implemented by the rioja package (Juggins, 2015) using the R statistical program. Diatom assemblages were related to water chemistry assuming unimodal species responses along log-transformed environmental data: 1-521 µg TP/L (average 20 µg/L); 0.33-120.74 mg Cl/L (average 13.39 µg/L). Two additional bars, 'stress power' and 'stress rank,' depict the relative ability of a taxon to track stress and whether that taxon reflects low or high stress. These results were achieved by evaluating relationships between each individual taxon and agricultural, industrial and urban development stressors quantified for each sample location. Briefly, the U.S. coastline of the Great Lakes was divided into 762 segments, each consisting of a shoreline reach and associated watershed (i.e., a 'segment-shed'). Each segment-shed was summarized using 207 geographic information system (GIS)-based environmental variables that included anthropogenic activities (e.g., agricultural activities, urban density, industrial polluters) (Danz et al., 2005). For example, 26 agricultural variables (including pesticide runoff and leaching, cropland area, nitrogen and phosphorus exports, percent of county treated for various pests and livestock inventories) comprised an agricultural category. Principal components analysis (PCA) within each category of environmental variation was used to reduce dimensionality and derive comprehensive gradients for agriculture, atmospheric deposition, point source pollution and urbanization.
For each common taxon, species relative abundance data were regressed against the set of comprehensive watershed-level predictors using multiple linear regression and evaluated using the coefficient of determination (R2). The 'stress power' value for a taxon represents its R2 position on the gradient of the R2 values for all common taxa. Taxa with higher values along this gradient had more acute relationships with watershed stressors, whether they reflected high or low stress, and so those taxa are assumed to be better indicators of conditions. To derive 'stress rank,' a canonical correspondence analysis (CCA) was performed including the species assemblages and the comprehensive stressor variables. A primary gradient of stress, largely driven by agricultural activities, was derived by the distillation of these data, similar to that achieved by Reavie (2007). A taxon's stress rank was taken as its score relative to that stressor gradient, standardized by the range of scores for all common taxa.

Amphora Ehrenberg
Amphora aequalis  This taxon agrees with Krammer's (1980) depiction, although it was sometimes indistinguishable from A. inariensis as described in the same article. In general, using LM the puncta should be visible in A. aequalis, especially on the dorsal valve, but this was not always certain. A. inariensis has equal conopeum development of the ventral and proximal dorsal regions while A. aequalis has a ribbed dorsal distal region, but neither of these features are easily visible using light microscopy. Krammer gives valve length range 17-33 µm; width range 3-6 µm; dorsal stria density 14-18/10 µm. Great Lakes specimens have valve length range 17-22 µm, width range 3.5-4.5 µm, dorsal stria density 18-20/10 µm. Environmental characteristics for this species ( Fig. 2A) indicate occurrence across the Great Lakes and especially in mesotrophic protected wetlands. The TP optimum is 25 µg/L while the chloride optimum is 12 µg/L. It is a weak indicator of anthropogenic stress.
Amphora cf. macedoniensis  This taxon somewhat agrees with depictions of A. macedoniensis by Nagumo (2003). During Great Lakes assessments these specimens were identified as a small form of A. libyca Ehrenberg (Krammer, 1980). Because of a lack of detailed micrographs and the large size of Great Lakes specimens, confirming this nomenclature would be premature. For A. macedoniensis the published valve length range is 13-30 µm; width range 3-6 µm; dorsal stria density 16-18/10 µm. Great Lakes specimens have valve length range 26-~38 µm, width range 6-7 µm, dorsal stria density 16-19/10 µm.  Patrick & Reimer's (1975) and Krammer & Lange-Bertalot (1986), although the dorsal striae near the central area were very faint or not visible. Published valve length range 35-85 µm; width range 9-17 µm; dorsal stria density 10-12/10 µm. Great Lakes specimens have valve length range 48-58 µm, width range 10-15 µm, dorsal stria density 11-13/10 µm, with one questionable specimen having a density as high as 14/10 µm. Environmental characteristics for this species (Fig. 2D) indicate occurrence across the Great Lakes and in all geomorphic habitats, especially embayments and riverine wetlands. The TP optimum just exceeds the eutrophic threshold at 32 µg/L while the chloride optimum is 13 µg/L. It is a relatively weak indicator of medium levels of anthropogenic stress.

Amphora neglecta
Amphora pediculus (Kützing) Grunow (Figs. 8A-8HHHHH) Despite being the most photographed diatom from the GLEI project, identifying this small, widespread Amphora by light microscope is problematic. In many cases in the Great Lakes, larger forms of A. pediculus cannot be distinguished from A. inariensis. Despite Krammer's (1980) statement that the dorsal raphe in A. pediculus is wider than the ventral raphe (whereas both sides are identical in A. inariensis), this distinction was not easily established. Most often this taxon was distinguished by its usual smaller size. Attempts were made to distinguish Amphora perpusilla (Grunow) Grunow from A. pediculus, but substantial intergrading and difficulty resolving differences (Krammer, 1980;Patrick & Reimer, 1975;Levkov, 2009) favored a single designation. Though both species are considered unique based on the latest taxonomic literature, it has long been suggested by some authors that they are identical (Schoeman & Archibald, 1978). However, I acknowledge that A. perpusilla may be an unrecognized member of this complex. Similarities were also noted with A. subcostulata Stoermer & Yang (1971), which may be present in my complex, but the usual lack of a shortened central stria favored A. pediculus.  Grunow [374,622,591,591,462,393,700,462,420,105,311,311,294,273,425,374,591,369,700,425,425,294,374,389,422,311,273,420,425,425,393,389,374,325,393,311,275,273,596,393,68,361,462,299,273,354,294,273,275,461,68,273,299,425,294,361,273,459,311,420,425,420,68,453,311,299,393,361,275,273,273,278,208,273,425,294,325,299,393,411,311,369,281,275,470,425,393,68,311,294,208,294,425,417,294,299,275,311,311,299,273,192,311,311,192,425,425,393,299,192,311 [617,617,393,213,104,393,369,339,104,417]. Lowercase letters indicate multiple images of the same specimen, and a question mark indicates a specimen with taxonomic uncertainty as described in the text. Square brackets contain sampling locales ( Fig. 1)  It is highly likely that higher resolution observations would resolve several species from this abundant taxon. Valve length range 4-24 µm; width range 2-5 µm; dorsal stria density 18-28/10 µm. Great Lakes specimens have valve length range 4.5-17 µm, width range 2-4 µm, dorsal stria density 17-28/10 µm. Certain specimens are especially questionable by lacking the dorsal space in the striae (e.g., Fig. 8FFF) or by having slight capitation in the ends (e.g., Figs. 8LLL and 8EEEE). Environmental characteristics for this species (Fig. 2E) indicate occurrence across the Great Lakes and in all geomorphic habitats. The TP optimum indicates a preference for mesotrophic conditions at 24 µg/L while the chloride optimum is 15 µg/L. It is a fair indicator of medium levels of anthropogenic stress.  Stoermer & Yang's (1971) depictions of "A. siberica" (orthographic error) from Lake Michigan collections. Great Lakes specimens have a higher dorsal striae count than described by Nagumo (2003). From the literature: valve length range 25-40 µm, width range 5-8 µm, dorsal stria density 16-24/10 µm. Great Lakes specimens have valve length range 25-31 µm, width range 5-7.5 µm, dorsal stria density 20-27/10 µm.
Amphora sp. 104 ( Fig. 7A) This rare taxon from an embayment at Caseville (Saginaw Bay, Lake Huron) could not be adequately matched with known species of Amphora. Valve shape and raphe structure was like A. libyca, although no gap in the central, dorsal striae was observed and the dorsal margin was flatly arched. Valve length range 32 µm, width 7 µm, dorsal stria density 14 µm.
Amphora sp. 106 (Fig. 7B) This rare taxon from a protected wetland in the Cheboygan River outlet (Lake Huron) did not match well with any known species of Amphora. Valves had a strong convex dorsal margin and straight ventral margin. The axial area was narrow. Striae were punctate on the dorsal side with a distinct gap in the central area. Ventral striae and raphe characteristics were not well observed. Specimens were similar to Amphora michiganensis Stoermer & Yang (1971), but in our specimens the axial area was closer to the ventral margin. Valve length range 16-17 µm; width range 5-7 µm; dorsal stria density 20/10 µm.  Stoermer & Yang's (1971) and Patrick & Reimer's (1975) depiction of the species (as Amphora). The specimen appears to have a slightly sunken dorsal central area and a too-high striae density, resulting in intergrading with H. montana. For H. bullatoides Patrick & Reimer (1975, as Amphora) give valve length range 17-30 µm, width range 4-6 µm, dorsal stria density 9-14/10 µm. The Great Lakes specimen has valve length range 26 µm, width 5.5 µm, dorsal stria density 18/10 µm, much finer near the ends.
Halamphora coffeaeformis (C.Agardh) Z. Levkov (Figs. 6R and 6S) Great Lakes specimens matched depictions of this species (as Amphora) by Archibald & Schoeman (1984) and Patrick & Reimer (1975). This taxon intergrades with A. veneta but is generally distinguished by having little or no apparent decrease in striae count at the dorsal central area.
Stepanek (2011a) gives valve length range 15-40 µm; width range 5-7 µm; dorsal stria density 17-21/10 µm at center, 22-24/10 µm near the ends. Great Lakes specimens have valve length range 18-20 µm, width 4 µm, dorsal stria density 26-32/10 µm, finer at ends. Environmental characteristics for this species are provided in Fig. 2F. As was typical for many species of Halamphora and Amphora in the Great Lakes, H. coffeaeformis had relatively high phosphorus and chloride optima, though it is a weak indicator of a medium amount of anthropogenic stress. It was observed particularly in Lake Erie protected wetlands.
Halamphora oligotraphenta (Lange-Bertalot) Z.  This taxon exhibited great variation in valve shape, and during original GLEI assessments depictions of Amphora veneta Kützing by Krammer & Lange-Bertalot (1986) were used for identification. Similar specimens to those from the Great Lakes are presented by Germain (1981), Patrick & Reimer (1975) and Krammer & Lange-Bertalot (1986) as A. veneta. Expression of the ends varied, with some being ventrally deflected and others being distinctly capitate and/or bulbous (as in A. bullatoides). Some intergrading with A. coffeaeformis was observed, but previous descriptions of A. veneta were usually distinguished by a coarser strial density near the central dorsal area. More recent taxonomic scrutiny separates this species as H. oligotraphenta (Levkov, 2009).

Cymbella affinis Kützing (Figs. 10A-10E)
Great Lakes specimens of this species match well with Potapova's (2011) depiction from Idaho, but not so much with those shown by  and Krammer (2002), which have shortened striae around the central area, dorsal side. Potapova (2011) gives valve length range 19-36 µm; width range 6.9-9 µm; stria density 9-12/10 µm in the center valve, 13-14/10 µm at the ends. Great Lakes specimens have valve length range 25-33 µm; width range 7.5-8 µm; stria density 12/10 µm. Based on the striae structure around the central area these specimens may be representatives of Cymbella affiniformis Krammer (Krammer, 2002) originally identified from Germany, but evaluating associations between European and North America examples is prudent. Environmental characteristics for this species (Fig. 2I) indicate occurrence across the Great Lakes and in all geomorphic habitats, especially nearshore and embayments. The TP optimum is 11 µg/L and the chloride optimum is 11 µg/L. It is a weak indicator of relatively low levels of anthropogenic stress.
Cymbella cf. aspera (Ehrenberg) Cleve ( Fig. 11F) The shape and size of this specimen is similar to C. aspera, among other representatives of the Cymbella cistula complex (Krammer, 2002), but the striae density is too high, and the lack of clear filiform proximal raphe ends precludes any of those species. The specimen has valve length 93 µm; width 16.5 µm; stria density 17/10 µm.
Cymbella cf. compacta Østrup (Fig. 10H) Despite being a broken valve, the specimen is a fair fit for the species as described by Krammer (2002), though I may have something that has a higher width to length ratio and less apiculate apices. The Great Lakes specimen has valve length 34 µm; width 12 µm; stria density 12/10 µm. Cymbella cf. excisa Kützing (Figs. 10F and 10G) While originally identified as Cymbella affinis, the characteristic single, ventral stria connected to the stigma is like observations of this species by Krammer (2002). However, striae densities of Great Lakes specimens exceeded those described by Krammer (2002): valve length range 17-41 µm; width range 6-10.7 µm; stria density 9-13/10 µm in the dorsal center, 12-14/10 µm at the ends. Great Lakes specimens gives valve length range 20-21 µm; width range 7-7.5 µm; stria density 14-15/10 µm in the dorsal center.  (2011) give valve length range 38-100 µm; width range 10-16 µm; dorsal stria density 8-12/10 µm. Great Lakes specimens have valve length range 60-69 µm; width range 10-14 µm; dorsal stria density 12-14/10 µm. While similar in many aspects, often including the subtle inflection of the raphe proximal ends, the slightly higher striae density in Great Lakes specimens prevents confident association with this species. Further, while the two specimens presented have similar identifying characteristics, the wider valve and slight inflection in the apices of Fig. 9L suggest additional taxonomic splitting may be warranted.
Cymbella neocistula var. islandica Krammer (Figs. 11A-11E) Despite their large, robust features, discerning the many taxa in the "Cymbella cistula" complex in the Great Lakes is challenging due to intergrading of species features such as shape and striae density, and a complicated history of taxonomic splitting, combining and renaming. Cymbella cistula (Ehrenberg) Kirchner was separated into three new taxa by Krammer (2002), one of them being Cymbella neocistula Krammer, a variety of which was later determined to be the same as Cymbella neocistula var. islandica, which was identified from Lake Michigan as Cymbella cistula var. gibbosa Brun by Patrick & Reimer (1975). Certain specimens (e.g., Fig. 11D) have the strongly arched dorsal margin like that of Cymbella perfossalis Krammer (2002), though that is only known to be a fossil species from the type locality in Oregon. Patrick & Reimer (1975) give valve length range 35-130 µm; width range 11-26 µm; dorsal stria density 6-11/10 µm. Great Lakes specimens have valve length range 50-69 µm; width range 13.5-16.5 µm; stria density 9-11/10 µm. Environmental characteristics for this species (Fig. 2J) (2011) give valve length range 15-47 µm; width range 7-10 µm; dorsal stria density 9-11/10 µm, and as high as 13/10 µm at the apices. Great Lakes specimens have valve length range 28-38 µm; width range 7-10 µm; dorsal stria density 10-12/10 µm. Unlike previous descriptions, the characteristic central stigma was sometimes an extension of the middle stria, in some cases accompanied by smaller stigmata on the two adjacent striae. Ventral structure ranged among slightly convex, slightly concave, and flat. Some specimens also had a more constricted central area than previously observed. Though additional splitting of this taxon may be warranted,  (2011) give valve length range 20-50 µm; width range 7.8-12.7 µm; dorsal stria density 8-12/10 µm. The Great Lakes specimen has valve length range 55 µm; width 9 µm; dorsal stria density 10/10 µm. Although it is slightly longer than previous European observations, this appears to be a suitable identification.
Cymbella sp. (Muskegon, Lake Michigan) (Fig. 11G) This interesting specimen has several structural similarities to Cymbella neocistula var. islandica, though the striae count is too dense, the characteristic inflection in the proximal raphe ends is not present, and the central area has a unique, circular central area containing a few short, excised striae where stigmata would normally be (as in Cymbella proxima Reimer). The specimen has valve length range 46 µm; width 11 µm; stria density 15/10 µm.
Cymbella sp. (Old Woman Creek, Lake Erie) (Fig. 9O) This partly obscured specimen may be Cymbopleura but was not adequately identifiable as a known species. The Great Lakes specimen has valve length 24.5 µm; width 6 µm; dorsal stria density 14/10 µm.
Cymbopleura incertiformis Krammer (Fig. 15NNNN) This specimen matches well with Krammer's (2003) description for the species. Krammer (2003) gives valve length range 24-60 µm; width range 6.5-8.5 µm; stria density 15-19/10 µm at the central area, up to 22/10 µm at the ends. The Great Lakes specimen has valve length 31.5 µm; width 6.5 µm; stria density 18/10 µm at the central area, 22/10 µm at the ends. (J-Q) Encyonopsis cesatii (Rabenhorst) Krammer [423,423,247,423,419,247,247,200]; (R-V, W? and X?) Cymbopleura cf. incertiformis Krammer [247,420,411,213,247,273,461]. Lowercase letters indicate multiple images of the same specimen, and a question mark indicates a specimen with taxonomic uncertainty as described in the text. Square brackets contain sampling locales ( Fig. 1)  Cymbopleura cf. incertiformis Krammer (Figs. 16R-16V) These specimens were originally identified as Cymbella incerta (Grunow) Cleve based on the broad depiction of that species by Krammer & Lange-Bertalot (1986). Further scrutiny of the valve shape and comparison with additional specimens for Cymbopleura incerta provided by Krammer (2003) indicate that no published descriptions adequately characterize these Great Lakes specimens. Comparison can be made with Cymbopleura frequens Krammer and its varieties (Krammer, 2003), and Cymbopleura hybrida (Grunow ex Cleve) Krammer (Bahls, 2015), but those species have a coarser striae density and frequens has a more rostrate shape in the valve ends. Comparison is also made with Cymbopleura incertiformis Krammer (valve length range 29.5-60.7 µm, width range 7.0-9.4 µm, stria density 14-17/10 µm at the valve center and 20-24/10 µm at the ends on the dorsal side; Bahls, 2014a), which tends to be longer, sometimes has no central area, and has a noticeably oblique raphe in approximately two-thirds of the axial area; features not observed in Great Lakes specimens. Valves of Cymbopleura cf. incertiformis are lanceolate, very slightly dorsiventral to symmetric along the long axis, subtly triundulate with blunt, apiculate apices; valve length range 19-28 µm; width range 5-7 µm; stria density 14-18/10 µm. The axial area (sternum) is narrow and linear-lanceolate, narrower at the poles, widening gradually then merging with an irregular to circular or oval central area. The central area comprises a third to half of the valve width. The raphe is very weakly lateral and is filiform at the proximal ends, which are expanded and not visibly tipped toward either side. The distal raphe ends are deflected dorsally, a feature of Cymbopleura that is confirmed in specimen Fig. 16Va. Striae are weakly radiate at the poles and central area, approaching parallel elsewhere. Similar specimens were observed from multiple locations around the Great Lakes, supporting the view that this may be an undescribed species. The specimen with relatively high striae density and a narrower axial area (Fig. 16W), and the more dorsiventral specimen (Fig. 16X) are presented as additional variations.
Cymbopleura sp. 20 (Fig. 12E) A matching description for this likely Cymbopleura could not be found in the literature. The Great Lakes specimen has valve length 40.5 µm; width 10.5 µm; stria density 13/10 µm. The ends are slightly capitate and no central area is present.
Delicatophycus M.J.Wynne The genus name Delicata Krammer was deemed invalid because it is a technical term and was recently replaced by Delicatophycus by Wynne (2019).
Delicatophycus neocaledonicus M.J.Wynne (Figs. 14B-14Q) During the GLEI assessment most of these widespread specimens were identified using older protocols as Cymbella delicatula Kützing. Subsequent coverage of further taxonomic refinement by Krammer (2003) allowed for new identifications. Based on thorough consideration of species parameters, the majority of former "Cymbella delicatula" specimens appear to be Delicata neocaledonica Krammer. Despite widespread observation of Delicata delicatula (Kützing) Krammer in North America (e.g., Bahls, 2017aBahls, , 2017b, including in the Great Lakes (Kingston, 1980), that identification is not appropriate for these specimens. This identification is justified by non-protracted ends and appropriate length in Great Lakes examples, though the size range of the species is slightly expanded. The inflection of the proximal raphe ends is variable, though the same variation is presented by Krammer (2003;their pl. 134:34-42). Krammer (2003) gives valve length range 20-31 µm; width range 4.4-5.2 µm; stria density 18-20/10 µm. The Great Lakes specimens give valve length range 19.5-30 µm; width range 4-5.5 µm; stria density 18-20/10 µm. If correct, this identification appears to be the first observation outside of the type locality of New Caledonica. Expansion of the critical dimensions of Delicata montana Bahls (2017cBahls ( , 2019 is also a possibility, but that species has a lower width limit of 5.1 µm and a slightly lower striae density. Environmental characteristics for this species (Fig. 3B) indicate occurrence in the upper Great Lakes, especially in Lake Huron, and especially in embayments. The TP optimum suggests this is an oligotrophic indicator with a low chloride optimum of 7 µg/L. It is a fair indicator of relatively low levels of anthropogenic stress.
Delicatophycus cf. neocaledonicus M.J.Wynne (Fig. 14A) I hesitate to confirm this identification as D. neocaledonicus due to a too-wide valve: The Great Lakes specimen is 6.5 µm, while the maximum for this species (Krammer, 2003) is purportedly 5.2 µm.
Delicatophycus sp. "Bell River 2" (Fig. 14Y) This specimen, also from Bell River, has many of the characters of Delicatophycus neocaledonicus (shape, size, raphe structure) but has a too-fine striae count (23/10 µm) and a wider axial area than other small, uncertain Delicatophycus from Bell River.

Encyonema auerswaldii Rabenhorst (Figs. 18A-18M)
This species was incorrectly reported as Encyonema cespitosum Kützing during original GLEI assessments. The higher puncta (areolae) density in Great Lakes specimens indicates that the Great Lakes contains E. auerswaldii, which agrees with observations that this species is common in freshwater systems across the USA (Lowe, 2015). In Great Lakes specimens the presence of slightly subrostrate, bent apices and variation in the width of the axial area appear to be transient features. For Encyonema cespitosum var. comensis Krammer (1997a) reports ventrally bent ends (Table 1), a feature that sometimes appeared in Great Lakes specimens of E. auerswaldii. That character appears inconsistent in Krammer's photographic examples, so I do not designate any varieties based on this feature. Careful observation of the puncta density is required to separate this species from similar (but larger, coarsely punctate) species (e.g., Encyonema kamtschaticum Krammer, E. cespitosum Kützing, Encyonema sinicum Krammer). Environmental characteristics for this species (Fig. 3C) indicate occurrence particularly in Lake Ontario, and in nearshore and embayment habitats. The TP optimum is 17 µg/L while the chloride optimum is 12 µg/L. It is a weak indicator of medium levels of anthropogenic stress.
Encyonema cf. auerswaldii Rabenhorst (Fig. 18N) This specimen adheres to features of the species except for a higher striae density (18/10 µm) and erratic strial lengths on the ventral portion of the valve.
Encyonema cf. elginense (Krammer) D.G.Mann (Figs. 17N-17Q) Krammer's (1997a) concept of this species is a fair fit, though Great Lakes specimens are slightly narrower. Krammer (1997a) gives valve length range 26-63 µm; width range 10-17 µm; stria density 8-13/10 µm. Great Lakes specimens have valve length range 30-36 µm; width range 8-9 µm; stria density 11-12/10 µm.   (2021)) was determined to be a close fit to these specimens during GLEI assessments, but on further scrutiny they likely represent at least one undescribed species, or greatly increase the permissible striae density for E. evergladianum. The shape of Great Lakes valves is a good match for that species, though the presence of a very small central area also differs. Torch Lake, which is connected to Grand Traverse Bay (northeastern Lake Michigan) was noted to lack E. evergladianum in favor of a similar but more slender-valved taxon , further suggesting that adjacent specimens from the Great Lakes represent a different species. Krammer (1997b) gives valve length range 13-28 µm; width range 3.7-5 µm; stria density 21-23/10 µm. Great Lakes specimens have valve length range 13.5-22.5 µm; width range 3.5-4.5 µm; stria density 22-28/10 µm. Environmental characteristics for this species (Fig. 3D) indicate occurrence particularly in lakes Superior and Michigan, and in coastal and protected wetland habitats. The TP optimum is 6 µg/L while the chloride optimum is 5 µg/L. It is a fair indicator of relatively low levels of anthropogenic stress.
Encyonema reichardtii (Krammer) D.G.Mann  Mainly found in nordic-alpine areas and northern latitudes, the Pennsylvania specimen Cymbella brehmii Hustedt (Patrick & Reimer, 1975) has been brought into synonymy with E. reichardtii by Mann (Round, Crawford & Mann, 1990). Specimens are generally small with convex dorsal and weakly convex ventral sides. Asymmetric Fig. 19Z most likely represents an abnormal valve. Raphe branches are slightly flexed ventrally. The central area is formed by the shortening of the middle striae with the ventral stria being the shortest and the dorsal stria having additional space between it and its neighboring striae. Krammer (1997b) provides valve length range 6.7-14.5 µm, width range 3.2-4 µm, stria density 18-22/10 µm and Patrick & Reimer (1975) provide valve length range 11-16 µm, width range 4-5 µm, and dorsal stria range 12-14/10 µm, ventral stria range 14-16/10 µm. Great Lakes specimens have valve length range 9.5-13.5 µm, width range 3.5-4.5 µm, stria density 16-21/10 µm. US literature suggests our specimens should have coarser striae density, while in fact they are finer like European examples. Patrick & Reimer (1975) emphasize the ventral flex of the raphe (prominent in my specimens) as a character to use to separate their Cymbella (Encyonema) brehmii from the similar Cymbella (Encyonema) hustedtii. Environmental characteristics for this species (Fig. 3F) indicate occurrence particularly in lakes Superior and Ontario, and in high-energy habitats. The TP optimum is 11 µg/L while the chloride optimum is 6 µg/L. It is a fairly weak indicator of medium levels of anthropogenic stress.
Encyonema reimeri S.A.Spaulding, J.R.Pool & S.I.Castro (Figs. 19A and 19B) My specimens fit well with features described by Spaulding (2010b), who gives valve length range 30-89 µm, valve width range 16-26.5 µm, stria density 10/10 µm. Great Lakes specimens have valve length range 50-52 µm; valve width range 21-23 µm; stria density 9-10/10 µm. Krammer (1997a) presents a highly variable concept for E. silesiacum. Distinguishing this species from the similarly highly variable Encyonema elginense (Krammer) D.G.Mann (Krammer, 1997a) is difficult, though E. elginense generally lacks a central stigma and North American examples suggest the subtle "shoulder" at the valve ends (e.g., Fig. 17P) may be characteristic (Bahls, Boynton & Johnston, 2018). The conspicuousness of the central stigma in E. silesiacum is transient in Great Lakes specimens, so its presence is not always useful for identification. This taxon was part of a confusing complex of similar species that were often difficult to distinguish in Great Lakes samples. In Table 2 I compare three taxa that were often included with E. silesiacum during sample assessments, and though challenges remain I have attempted to discern taxonomy of available photographs. Unfortunately, features described in the literature for these species were frustratingly transient, such as a subtle subrostrate character in the apices of E. silesiacum, a feature that is more characteristic of E. lange-bertalotii. Also, separation of E. ventricosum from both of these species is largely based on a higher stria density, but given otherwise similar characteristics with these species I am not convinced this separation is warranted. Further, potential varieties of E. silesiacum as presented by Krammer (1997a;distinctepunctata, altensis, lata, Encyonema persilesiacum) were not easily discerned and may be incorporated into our Great Lakes definition of this species. Environmental characteristics for E. silesiacum (Fig. 3G) indicate widespread occurrence across the Great Lakes, but especially in nearshore locations in Lake Superior. The TP optimum is 19 µg/L while the chloride optimum is 8 µg/L. It is a fair indicator of medium levels of anthropogenic stress. Environmental characteristics for E. ventricosum (Fig. 3H) indicate it is mainly found in the upper lakes and is a strong indicator of medium levels of anthropogenic stress.
Encyonema cf. silesiacum (Bleisch in Rabenhorst) D.G.Mann (Fig. 17S) Although Encyonema silesiacum has comprised a broad definition from a set of largely European examples (Krammer, 1997a), this specimen is relatively large and coarsely striate for the species, and it has a clear dorsal inflection of the raphe at the proximal ends. Other similar European concepts include Encyonema jemtlandicum var. venezolana Krammer and Encyonema vulgare Krammer, but additional specimens from the Great Lakes are needed for confirmation. Puncta are also relatively coarse. The Great Lakes specimen has valve length 40.5 µm; width 11.5 µm; stria density 10/10 µm at the central area, 12/10 µm at the ends.
Encyonema sp. "Lake Huron" (Fig. 18P) This rare, lanceolate and probable Encyonema from Lake Huron did not belong to any species I observed in the existing literature, though it may be a very small example in the asexually declining size series of a species such as E. norvegicum, which has a similar stria density.
Encyonema/Cymbella sp. (Fig. 18Q) The curved raphe separates this specimen from other diatoms in the E. silesiacum complex. It is likely a Cymbella with similarities to Cymbella laevis Naegeli (pl. 191 in Krammer (2002)), but the central stria on the ventral side possibly connected to a stigma is distinctive and unique. The specimen has valve length range 20.5 µm; width 6 µm; stria density 11/10 µm at the center and~15/10 µm at the ends.

Encyonopsis Krammer
Encyonopsis cesatii (Rabenhorst)  These original identifications during the GLEI assessment maintain this species name. While some specimens could be identified as Encyonopsis cesatiformis Krammer (Bahls, 2014b), the wider valves expected for that species were not observed, so the presence of E. cesatiformis in the Great Lakes is questionable. Valves tended to be more lanceolate than the more rhombic Encyonopsis montana.  give valve length range 18-60 µm; width range 4-6 µm; stria density 18-22/10 µm. Great Lakes specimens have valve length range 26-38 µm; width range 4.5-6 µm; stria density 17-22/10 µm. Environmental characteristics for this species (Fig. 3I) indicate it occurs in the upper Great Lakes in all coastal habitat types. The TP optimum is 10 µg/L while the chloride optimum is 3 µg/L. It is a good indicator of medium levels of anthropogenic stress.
Encyonopsis cf. cesatii (Rabenhorst) Krammer "Ford River" (Fig. 16G) This specimen from near the Ford River outlet (Green Bay, Lake Michigan) generally matched valve parameters for Encyonopsis cesatii with the exception of a very high striae density of 28/10 µm.
Encyonopsis cf. cesatii (Rabenhorst) Krammer "Peterson Creek" (Figs. 16H and 16I) These narrow valves from near the Peterson Creek outlet (south shore, Lake Superior) largely take on the character of smaller Encyonopsis cesatii, but the higher striae density of 26-28/10 µm indicates they may be an undescribed species or variety.
Encyonopsis cf. krammeri "very fine" (Figs. 15LL-15VV) These specimens largely follow depictions of this species by  and Krammer (1997b), and "cf." is applied because of striae densities that exceed the upper limit of 30/10 µm for Encyonopsis krammeri. Although Encyonopsis angusta Krammer has sufficiently dense striae (32-35/10 µm; Krammer, 1997b), the strongly capitate ends present in that species are not consistent in Great Lakes examples. Great Lakes specimens have valve length range 9-17.5 µm; width range 2.8-3.5 µm; stria density 31-36/10 µm. Clearly there remains variation in the prominence of the "shoulders" and capitation at the valve ends, which may warrant further taxonomic splitting. Environmental characteristics for this species (Fig. 3J) indicate it occurs mainly in the upper Great Lakes, especially in riverine wetlands. The TP optimum is 7 µg/L while the chloride optimum is 10 µg/L. It is a weak indicator of relatively low levels of anthropogenic stress.
Encyonopsis cf. krammeri "very fine and narrow" (Figs. 15NNN-15RRR) As for Encyonopsis cf. krammeri, these specimens have higher than previously published striae densities, but they also represent distinctly narrow valves, a character which may specify a unique species. Great Lakes specimens have valve length range 10.5-15 µm; width 2.5 µm; stria density 31-36/10 µm. Environmental characteristics for this species (Fig. 4A) indicate it occurs across the Great Lakes except in Erie, and in all coastal habitat types. The TP optimum is 15 µg/L while the chloride optimum is 11 µg/L. It is a relatively weak indicator of medium levels of anthropogenic stress.
Encyonopsis cf. montana L.L.Bahls (Fig. 16F) This uncertain specimen from a riverine wetland at Buck Pond, Lake Ontario, has a vague similarity to E. montana but it is too short (26 µm), narrow (6.5 µm) and coarsely striate (16/10 µm). Strong strial radiation at the ends also seems to be lacking and there may be a central stigma, which is not typical for the genus.  This species identification is held for specimens with a striae density of 24/10 µm; low enough to indicate proper association with the species. However, given the difficulties described for the E. subminuta/krammeri complex, many specimens in that complex could also be E. subminuta. Given overlap in measurements (size, shape, character of the ends and central area, and especially striae density) and transitions in these characters among species, LM assessments of Great Lakes assemblages are challenging. It is likely many of these specimens could be identified as Encyonopsis minuta Krammer & E.Reichardt, but transition between these two species makes this difficult. For instance, a comparison of the two species using the associated diatoms.org pages (Bahls, 2013b;Kociolek, 2011a) states that "Encyonopsis subminuta has larger, more elliptic and more symmetric (less dorsiventral) valves"; features that appear to be transient within species and transitional between the two species. Overlap of lengths and striae densities also prevents distinction. Further, despite use of the name during GLEI assessments, the purportedly ubiquitous Encyonopsis microcephala is not applied to any Great Lakes specimens due to (a) rare observation of specimens with the required lower striae density, and (b) the inability to distinguish it from Encyonopsis subminuta based on published valve characteristics and LM limitations. Given that European concepts are used for these names, it is clear that more detailed, high-resolution studies are needed to better define these species for the Great Lakes. For Encyonopsis subminuta,  give valve length range 10-25 µm; width range 3.4-4.5 µm; stria density 12-14/10 µm, up to 23-26/10 µm at the ends. The Great Lakes specimens have valve length range 8-16 µm; width range 3-4 µm; stria density 24/10 µm. Environmental characteristics for this species (Fig. 4B) indicate it occurs mainly in the upper Great Lakes in most habitat types except riverine wetlands. The TP optimum is 8 µg/L while the chloride optimum is 10 µg/L. It is a fair indicator of medium levels of anthropogenic stress.

Encyonopsis subminuta
Encyonopsis subminuta "broadly rounded form" (Figs. 15WW and 15XX) This taxonomic assignment matches that used for Encyonopsis subminuta, though these specimens were characterized by broadly rounded ends and a distinct widening of the axial area in the center of the valve. Great Lakes specimens have valve length range 8-16 µm; width range 3-4 µm; stria density 24/10 µm.
Encyonopsis subminuta/krammeri (Figs. 15A-15KK) Inspection of dozens of specimens from many locations throughout the upper (Superior, Huron, Michigan; locations 29-486; Fig. 1) coastal Great Lakes reveals that previous taxonomic accounts for these small Encyonopsis are inadequate to consistently identify species. Given frustrating transition in valve characters that are supposed to be unique to each species-rostrate to distinctly capitate ends, dorsiventrality, striae density, small to non-existent central areas-I eventually decided to create an ambiguous complex that clearly requires additional high-resolution scrutiny. For instance, a subset of these specimens (Figs. 15FF-15KK) shows a combination of greater capitation and dorsiventral difference, similar to Encyonopsis microcephala (Grunow) Krammer, which may support separation. Because it is described as a defining character for Encyonopsis krammeri, a maximum striae density of 30/10 µm was arbitrarily selected as the cutoff for assignment to this complex. Higher striae densities are not presented in previous literature on these taxa, so another category (Encyonopsis cf. krammeri "very fine") was created for these Great Lakes-specific forms. A lower cutoff of 25/10 µm was also chosen to separate this complex from likely specimens of Encyonopsis subminuta. Though largely European concepts (Krammer, 1997b) were used in naming, special consideration is clearly needed for the Great Lakes. Great Lakes specimens in this complex have valve length range 8-20.5 µm; width range 3-4 µm; stria density 25-30/10 µm. Environmental characteristics for this species (Fig. 4C) indicate it occurs mainly in the upper Great Lakes in all coastal habitat types except protected wetlands. The TP optimum is 15 µg/L while the chloride optimum is 9 µg/L. It is a weak indicator of medium levels of anthropogenic stress.

Encyonopsis thumensis Krammer (Figs. 15GGG-15MMM)
Though specimens in this taxonomic assignment graded into Encyonopsis subminuta and the ambiguous Encyonopsis subminuta/krammeri complex, they were characterized by a subtle rostrate character at the apices. Krammer (1997b) gives valve length range 9.5-18 µm; width range 3.5-4 µm; stria density 23-26/10 µm. The Great Lakes specimens have valve length range 12-17.5 µm; width range 3-3.5 µm; stria density 24-28/10 µm. Using LM, I see no adequate way to distinguish these specimens from Encyonopsis vandamii Krammer as presented by Krammer (1997b). As for most of these small Great Lakes Encyonopsis, this name should be considered tentative until SEM or genetic comparisons can be made against examples from the type locality, in this case Bavaria.
Encyonema sp. (Riley's Bay, Lake Michigan) (Fig. 19AA) This appears to be an unknown Encyonema with similarities to Reimeria. The construction of the central area is similar to E. reichardtii but has an isolated stigma in the center. The raphe is not flexed ventrally as is E. reichardtii, though the ventral side is weakly convex. Valve length 13.3 µm, width range 4.4 µm, stria density 17.5/10 µm are from a single specimen and combined with the morphological characters provide the only evidence for a new species.

Reimeria Kociolek & Stoermer
Reimeria observations in the Great Lakes largely match with previously described species, but some detailed descriptions are provided due to ambiguities in previous literature.  Valves are linear with ends that are broadly rounded. The central area usual possesses a unilateral swelling on the ventral side with the dorsal side being tumid. The striae are slightly radiate to parallel on the dorsal side and more strongly radiate to parallel on the ventral side. The ends of the ventral side have a small 'blank' area lacking striae and instead filled with the apical porefield (Schoeman & Archibald, 1979) that extends to the mantle (Kociolek & Stoermer, 1987). The raphe is straight throughout the valve, becoming flexed toward the dorsal side at the ends of the valve, although one specimen exhibited a weakly arched raphe (Fig. 19J). There is a central stigma located just to the dorsal side or between the proximal raphe ends. Figures 19L and 19M have a slightly bilateral symmetry unlike Figs. 19H-19K, so it is possible that Figs. 19L,M represent a different species of Reimeria. The critical dimensions of forms identified during this project exhibit expanded variability. Kociolek & Stoermer (1987) give length 12-20 µm, breadth 4-5 µm and striae 10-12/10 µm (10-16/10 µm presented by Potapova (2009)). These new Great Lakes specimens are length 10.5-20.5 µm, breadth 3.5-6 µm and striae 12-16/10 µm. Environmental characteristics for this species (Fig. 4D) indicate it occurs mainly in Lake Ontario, especially in high-energy environments. The TP optimum is 15 µg/L while the chloride optimum is 12 µg/L. It is a fair indicator of medium levels of anthropogenic stress.
Reimeria sinuata f. antiqua (Grunow) Kociolek & Stoermer (Figs. 19Q and 19R) Though Krammer & Lange-Bertalot (1986) do not recognize the form antiqua, stating that the circumscription of varieties by outline is not feasible (translated), Kociolek & Stoermer (1987) recognize R. sinuata f. antiqua and describe the unique, fine structure with SEM. The ends are distinctly capitate. The striae are radiate throughout the valve, appear as solid lines in the LM but are doubly punctate in the SEM. A central pore (stigma) is located off-center externally but penetrating the valve on an angle, thus when focusing through the valve the stigma appears like a dash. The raphe is straight curving toward the dorsal side at the distal end. The critical measurements for the specimens observed fall well within those given by Patrick & Reimer (1975, as Cymbella sinuata f. antiqua (Grunow) Reimer): valve length range 11-40 µm, width range 3.5-9 µm, stria density 9-14/10 µm. Great Lakes specimens are valve length range 15-15.5 µm, width 4.5 µm, stria density 12-14/10 µm.
Reimeria sp. (Lake Ontario) (Fig. 19S) This appears to be an undescribed form of Reimeria and is possibly teratogenic. The shape is reniform with the ventral side central area slightly concave. The raphe appears to be straight with the distal ends not resolved. Striae puncta are not resolved with the LM and are perpendicular to the raphe at the center, becoming radial at the ends. The central area is somewhat unilateral but similar to other species of Reimeria. The isolated stigma is characteristically placed midway between the proximal raphe ends. The critical dimensions are valve length 12.5 µm, width 4 µm, stria density 12/10µm.

Gomphonema Ehrenberg
Gomphonema and similar taxa are represented in Figs. 20-25. Though not detected in any samples analyzed as part of the GLEI project, the well-known gomphonemoid diatom Didymosphenia geminata (Lyngbye) M.Schmidt is well presented by other authors (Stoermer, Yu-Zao & Ladewski, 1986 (Lake Superior north shore)). While this mat-forming species is often abundant, its presence in tributaries well upstream from the GLEI sediment and epiphyton samples likely prevented its detection in coastal samples.
Gomphonema acuminatum Ehrenberg ( Fig. 20M) As depicted by Thomas et al. (2009), G. acuminatum has a bulge in the wider portion of the valve that is wider than that for the central area.  give valve length range 17.5--57 µm, width range 6.7-10.8 µm, stria density 9-15/10 µm. The Great Lakes specimen is valve length 35.5 µm, width 10 µm, stria density 12/10 µm. Environmental characteristics for this species (Fig. 4E) indicate it occurs across the Great Lakes except in Lake Erie and was not observed in coastal and protected wetlands. The TP optimum reflects eutrophic conditions at 34 µg/L while the chloride optimum is 33 µg/L. It is a weak indicator of medium levels of anthropogenic stress.
Gomphonema cf. innocens E.Reichardt (Fig. 23Q) Due to its small size, a lack of visible features makes identification difficult. The striae density and arrangement around the central area suggest it may be a valve of G. innocens, but it is narrower than the range published by . The Great Lakes specimen is valve length 9 µm, width 3.5 µm, stria density 18/10 µm.
Gomphonema cf. insigne Gregory (Fig. 25A) Valve features largely match that for G. insigne as depicted by Reichardt (1999), with the exception that the Great Lakes specimen has a distinctly subrostrate, triangular head-pole, which is sometimes observed in the more finely striate Gomphonema insigniforme E.Reichardt & Lange-Bertalot. For G. insigne Reichardt (1999) gives valve length range 20-84 µm, width range 8.6-12.8 µm, stria density 6.5-10/10 µm. The Great Lakes specimens has valve length 58 µm, width 11.5 µm, stria density 10/10 µm.  (2011), but striae are finer than for G. vibrio. This taxon is likely the same as G. intricatum as identified from Lake Superior by Kociolek & Stoermer (1991), who indicate valve length range 19-65 µm, width range 4-9 µm, stria density 10-15/10 µm. The Great lakes specimens have valve length range 28-37 µm, width range 5-7 µm, stria density 10-13/10 µm. As noted in the plate, there may be a wider form of this species, and the central area was not always as large as depicted in previous literature. The specimen marked with a question mark is uncertain.
Gomphonema cf. kobayasii Kociolek & Kingston (Figs. 21V and 21W) Though initially identified as Gomphonema cf. bavaricum, Great Lakes specimens are narrower than that described by Reichardt & Lange-Bertalot (1991) for that species. Our narrower valves suggested similarity to G. kobayasii as described by Kociolek & Kingston (1999) from United States rivers, but those valves have a more lanceolate shape, whereas these specimens taper more consistently from the central area to the ends. Otherwise, primary features of Great Lakes specimens (valve length 24 µm, width 4 µm, stria density 10-12/10 µm) adhere to previous descriptions of the species. Environmental characteristics for this species (combined with G. kobayasii) are provided in Fig. 4H. This taxon has one of the highest chloride optima, reflecting the high chloride concentrations at Dunes State Park (site 364, Lake Michigan) where several specimens were observed.
Gomphonema cf. micropumilum E.Reichardt (Figs. 22N-22S) Largely identified as G. pumilum during original assessments, these specimens have a striae density that is too high for that species. They match some of the valve features presented by Reichardt (1997), but the nearly parallel-sided valve shape, lack of even subtle capitation, and sometimes higher striae density suggest that this is an undescribed species. (2011) give valve length range 10-35 µm, width range 4-8 µm, stria density 8-18/10 µm. Great Lakes specimens have valve length range 15-31.5 µm, width range 3.5-5 µm, stria density 12-17/10 µm. The specimen noted with a question mark is uncertain due to its small central area and relatively high ratio of width to length. Environmental characteristics for this species (Fig. 5A) indicate it occurs in all coastal habitats across the Great Lakes, especially in lakes Michigan and Ontario. The TP optimum is 19 µg/L while the chloride optimum is 35 µg/L. It is a weak indicator of medium levels of anthropogenic stress.

Gomphonema parapygmaeum Jüttner & Kociolek (Figs. 23V-23KK)
These lanceolate valves fit the species as depicted by Jüttner et al. (2018), who give valve length range 13.5-25.0 µm, width range 3.0-4.0 µm, stria density 12-16/10 µm. Several species in the G. pumilum complex are easily confused with this. The main feature distinguishing these specimens is the lack of tapering at the ends, which turns an otherwise similar diatom into G. minusculum as depicted by Reichardt (1997) from Germany. Using LM I see no reliable way to distinguish this species from G. pumilum var. rigidum E.
Reichardt & Lange-Bertalot as depicted by Reichardt (1997), which has valve length range 12-36 µm, width range 3-5.3 µm, stria density 11.5-14/10 µm. Reichardt indicates it may be distinguishable by its rectangular central area, but this feature appears to be transient. Great Lakes specimens are valve length range 12-29 µm, width range 3.5-5 µm, stria density 12-14/10 µm. The specimen marked by a question mark has a higher striae density that stands out among other specimens, but it is still within the published range. Environmental characteristics for this species (Fig. 5E) indicate it occurs mainly in lakes Superior and Ontario in a variety of coastal habitats. The TP optimum reflects oligotrophic conditions at 3 µg/L while the chloride optimum is 5 µg/L. It is a weak indicator of medium levels of anthropogenic stress.  (2011), give valve length range 10-36 µm, width range 5-8 µm, stria density 7-20/10 µm, a wide range of striae density that captures several Great Lakes specimens. Great Lakes specimens are valve length range 10-28 µm, width range 4-6 µm, stria density 12-18/10 µm, with one specimen (Fig. 22BB) noted to be as high as 20/10 µm (marked with a question mark). Another 8-µm-long specimen (Fig. 22DD) is below published lengths and is placed next to G. calcifuga for comparison. Environmental characteristics for this species (Fig. 5G) indicate it occurs across the Great Lakes in all coastal habitats. The TP optimum reflects eutrophic conditions at 47 µg/L while the chloride optimum is 15 µg/L. It is a weak indicator of medium levels of anthropogenic stress.
Gomphonema cf. parvulum (Mona Lake) (Fig. 22CC) Such a small specimen is difficult to identify, as it may be at the end of a diminution series and is missing the diagnostic features of larger valves. It may be a very small example of G. parvulum. Valve length 9.5 µm, width 4.5 µm, stria density 16/10 µm.  Jüttner et al. (2018) give valve length range 9.5-29.0 µm, width range 5.0-7.0 µm, stria density 14-16/10 µm. Great Lakes specimens are valve length 13 µm, width 5 µm, stria density 16/10 µm. Using LM there appears to be no way to distinguish smaller valves of G. parvulum from G. himalayaense, so these specimens from the Great Lakes remain ambiguous. Environmental characteristics for this species (Fig. 5F) indicate it occurs largely in high-energy habitats from Lake Ontario. In concord with a name indicating tolerance of polluted environments , G. parvulum f. saprophilum has one of the highest phosphorus optima (114 µg/L) for the Great Lakes diatoms. It is a weak indicator of impacted conditions.
Gomphonema cf. pseudotenellum Lange-Bertalot (Figs. 21OO-21EEE) Unlike G. parapygmaeum, the valve outline tapers gradually in some specimens from the center to the apices, though this is not clearly observed in all Great Lakes specimens. The 'cf.' qualifier is further used because LM images of G. pseudotenellum as depicted by Hofmann, Werum & Lange-Bertalot (2011) have a barely visible central stria on each side of the central area, whereas in Great Lakes specimens this feature is more clearly apparent. This is not easily distinguishable from smaller specimens of Gomphonema minusculum Cleve-Euler Jüttner et al., 2018), and they may be conspecific. Many of these may be narrow, fine specimens of a greater complex. For G. pseudotenellum Krammer & Lange-Bertalot (1986) give valve length range 10-28 µm, width range 2.5-4 µm, stria density 15-22/10 µm. The Great lakes specimens have valve length range 9.5-23 µm, width range 2.5-3.5 µm, stria density 16-20/10 µm.  (2015) originally identified similar specimens as unknown forms like G. parvulum and G. pumilum. Great Lakes specimens are also more lanceolate than Gomphonema pseudotenellum Lange-Bertalot. These somewhat match specimens identified as Gomphonema pygmaeum Kociolek & Stoermer from the Great Lakes (Kociolek & Stoermer, 1991; holotype from Lake Superior presented), but they do not have the more parallel sides of G. pygmaeum as later depicted by Kociolek (2015). Overall, Great Lakes specimens have valve length range 15-25 µm, width range 3-4 µm, stria density 12-18/10 µm. These diatoms are part of the G. pumilum complex and should be subjected to greater scrutiny using SEM. Environmental characteristics for this species (Fig. 5D) indicate it occurs across the Great Lakes in embayments. The TP optimum is 16 µg/L while the chloride optimum is 18 µg/L. It is a weak indicator of medium levels of anthropogenic stress.
Gomphonema cf. rexlowei Liu & Kociolek (Fig. 21X) Despite many articles resolving similar Gomphonema species, Great Lakes specimens such as this do not fit with known species. This is similar to that shown by Liu, Kociolek & Wang (2013), but both striae at the central area are shortened, instead of just one for G. rexlowei. This was called Gomphonema minutum "v10" during assessment due to similarities in valve shape and strial density, but G. minutum has a less prominent central area formed by shortened striae. Another similar species is Gomphonema juettnerii Karthick et al. (2015), but that species has a larger axial area and more evenly distributed striae around the central area. For G. rexlowei Liu, Kociolek & Wang (2013) give valve length range 15-35 µm, width range 4.6-6.2 µm, stria density 8-14/10 µm. The Great lakes specimen has valve length 24.5 µm, width 4.5 µm, stria density 13/10 µm. Gomphonema sphaerophorum Ehrenberg (Figs. 20J and 20K) This generally wider species is similar to G. pseudosphaerophorum but has a more asymmetrically inflated "shoulder" end. Specimens were sometimes mistaken for Gomphonema augur Ehrenberg but that species does not have an inflated headpole like Great Lakes specimens, although Thomas et al. (2009) present examples of what they consider G. augur that are not distinguishable from my specimens based on LM assessment. Polaskey & Bishop (2016) give valve length range 36.7-55.9 µm, width range 12.8-13.2 µm, stria density 12-16/10 µm. The Great Lakes specimen has valve length range 38-40 µm, width 12 µm, stria density 14/10 µm.
Gomphonema cf. subclavatum Grunow (Figs. 24C and 24D) The valve shape is similar to that described for G. subclavatum by Patrick & Reimer (1975), but the size is too small and is slightly more finely striate. For G. subclavatum Patrick & Reimer give valve length range 35-70 µm, width range 8-10 µm, stria density 9-13/10 µm. The specimens have valve length range 21-27 µm, width 6 µm, stria density 14/10 µm. It also has similarities to Gomphonema mexicanum Grunow as depicted by Reichardt (1997) but lacks the clearly punctate striae of that species. Further, the two specimens presented for this taxon have different valve shapes, indicating further taxonomic division may be possible.

Gomphonema submehleri
Kociolek & Stoermer (Figs. 24A and 24B) While similar to G. leptocampum, G. submehleri has a wider axial area as in specimen Fig. 24B, a characteristic that is well illustrated by Kociolek & Stoermer (1991), who collected the holotype from a Lake Superior epilithic sample. Specimen Fig. 24A does not carry the wide axial area through the central area, which may reflect a variation within the species. Kociolek & Stoermer (1991) give valve length range 25-55 µm, width range 7-11 µm, stria density 8-14/10 µm. The specimens have valve length range 27.5-49 µm, width range 7-8.5 µm, stria density 11-13/10 µm. Environmental characteristics for this species (Fig. 5H) indicate it occurs mainly in embayments and protected wetland habitats in Lake Ontario. The TP optimum reflects eutrophic conditions at 54 µg/L while the chloride optimum is 27 µg/L. It is a weak indicator of medium levels of anthropogenic stress.

Gomphonema truncatum Ehrenberg (Figs. 20A-20E)
This species has been previously identified from The Great Lakes by Kociolek & Stoermer (1991). For the Great Lakes they give valve length range 40-75 µm, width range 9-11 µm, and for the species overall Patrick & Reimer (1975) give stria density 10-12/10 µm. My Great Lakes specimens are valve length range 25-56 µm, width range 8-10.5 µm, stria density 10-13/10 µm. My observations increase the size range for this species in the Great Lakes. Environmental characteristics for this species (Fig. 5I) indicate it occurs across the Great Lakes in all coastal habitat types.
Gomphonema sp. (Big Sable) (Fig. 21U) Superficially it would be easy to place this specimen within the broad complex of G. pumilum and its varieties as depicted by Reichardt (1997), but the slightly capitate ends are unique.
Gomphonema sp. (Dunes State Park) (Figs. 21CC and 21DD) While a likely member of the confusing G. pumilum complex, these valves from Lake Michigan have a slightly rostrate southern pole, possibly distinguishing it from other species in the complex.
Gomphonema sp. (Grand Lake outlet, Lake Huron) (Fig. 21Y) While associated with the G. pumilum complex during sample assessment, this specimen has more distinctly capitate ends as in G. angustum , but not having the same central area features as that species.
Gomphonema sp. (Green Bay) (Figs. 25C and 25D) With its distinctly convergent striae around the central area and inflation of the central region of the valve, no adequate match for these specimens could be found in the literature. The intact specimen has valve length 53 µm, width 8.5 µm, stria density 13/10 µm.