INTRODUCTION

Accurate knowledge of the status and distribution of biota is fundamental for proper conservation of natural resources. Diversity is significant within unionid mollusks in the Mississippi basin (van der Schalie and van der Schalie 1950; Johnson 1980; Turner et al. 2000), yet an incomplete understanding of the genetic structure of many taxa (e.g., Campbell et al. 2005, Graf and Cummings 2007) leads to uncertainty regarding species distributions. Illinois has a diverse, well-documented freshwater mussel fauna that historically consisted of more than 80 species of Unionidae and one species of Margaritiferidae (Baker 1906, 1912; Parmalee 1967; Cummings and Mayer 1997; Tiemann et al. 2007). Range updates, such as discovering Bankclimber Plectomerus dombeyanus (Valenciennes, 1827) in Illinois in 2012, have been documented through sporadic or systematic surveys (Tiemann et al. 2007, 2013). Publication of such findings is valuable to regional conservation efforts, because federal and state agency conservation plans can apply only to species that are known to be present.

Figure 1.

Example of variation in morphology of Lampsilis species included in our analyses: INHS Mollusk Collection Catalog Number, locality details, state. Sex noted by ♀ (female) or ♂ (male) and was determined by external shell morphology. (a) Lampsilis abrupta INHS 21521, Ohio; (b) Lampsilis higginsii INHS 30606, Mississippi River, Dubuque County, Iowa; (c) Lampsilis hydiana INHS 87783 Boeuf River, Richland Parish, Louisiana; (d) Lampsilis radiata INHS 38141, Yates County, New York; (e) Lampsilis sietmani INHS 32502, Illinois River, Pike County, Illinois; (f) Lampsilis siliquoidea INHS 41996, Mackinaw River, McLean County, Illinois; (g) Lampsilis straminea INHS 22926, Black Warrior River, Jefferson County, Alabama; (h) Lampsilis virescens INHS 21586, Paint Rock River, Alabama.

More than 20 species of Lampsilis are currently recognized in North America (Williams et al. 2017; FMCS 2019), and seven of those have been documented in Illinois by live material or shell (Tiemann et al. 2007). This diverse genus ranges across eastern and central North America and has shell morphology that varies from ovate—like Pink Mucket Lampsilis abrupta (Say, 1831)—to elongate and terete—like the newly described Canary Kingshell Lampsilis sietmani Keogh and Simons 2019 (Keogh and Simons 2019; Fig. 1). Fatmucket Lampsilis siliquoidea (Barnes, 1823) is one of the most widespread unionids in the world and has stable populations across most of its range. It occurs widely in the Mississippi and Great Lakes basins and is commonly encountered in Illinois rivers (Tiemann et al. 2007; Watters et al. 2009). Louisiana Fatmucket Lampsilis hydiana (Lea, 1838) (Fig. 1c—a) species previously reported from eastern Texas, Oklahoma, and Arkansas and east to Alabama (Burch 1975; Howells et al. 1996)—has a similar morphology to L. siliquoidea (Fig. 1f), but L. hydiana has never been genetically confirmed to exist in Illinois. Neither L. hydiana nor L. siliquoidea is of conservation concern in Illinois or at the federal level.

Lampsilis hydiana is described as having an elliptical, rayed, somewhat inflated shell and is distinguished from L. siliquoidea by a pearlier nacre, an umbo that is anterior, and an overall smaller average total length (Lea 1838). However, these two species have been considered indistinguishable at times (Vaughn et al. 1996) or as synonyms (Call 1895), which has led to uncertainty regarding their distributions. Based on literature reports and museum shell records, these species presumably co-occur in several drainages, such as the Big Black and Yazoo rivers in Mississippi (Jones et al. 2005). Additionally, specimens from Arkansas initially identified as L. hydiana included three genetically distinct groups that represented L. hydiana and two additional undescribed species (Harris et al. 2009). These divisions were supported by a shape analysis, though there was some overlap in morphology (Harris et al. 2004; Harris et al. 2009). Thus, the range extent of L. hydiana remains unknown, and morphological characteristics to distinguish among L. siliquoidea, L. hydiana, and other similarly shaped Lampsilis species are lacking.

Figure 2.

Representative images of some individuals of Illinois-collected Lampsilis included in our analyses (other images at  https://doi.org/10.13012/B2IDB-5609050_V1): INHS Mollusk Collection Catalog Number (lower specimen number is arranged on top of each pair of images), locality details, and predetermined phenotype and confirmed genotype. Sex noted by ♀ (female) or ♂ (male) and was determined by external shell morphology. (a) INHS 45463-2 and 45463-3, Skillet Fork, Wayne County Lampsilis hydiana phenotype and L. hydiana genotype; (b) INHS 41996-1 and 41996-2, Mackinaw River, McLean County, Illinois, Lampsilis siliquoidea phenotype and L. siliquoidea genotype; (c) INHS 86787-5 and 86787-6 Lusk Creek, Pope County, L. siliquoidea phenotype and L. hydiana genotype; (d) INHS 45615-2 and 45615-9 Lusk Creek, Pope County, L. siliquoidea phenotype and L. siliquoidea genotype.

Certain specimens in several southern Illinois watersheds morphologically resemble L. hydiana (Fig. 2), though collection localities are well outside the published range of this species (Fig. 3). These specimens were typically identified as L. siliquoidea, despite morphologic resemblance to L. hydiana. The objective of our study was to determine taxonomic placement of the specimens that morphologically resemble L. hydiana to gain a better understanding of the distribution of L. siliquoidea and related species in Illinois.

METHODS

Mantle tissues of putative L. hydiana and L. siliquoidea from Illinois (n = 83 specimens from 25 sites) were collected from fresh, frozen, or ethanol-preserved individuals, used for DNA extraction, and catalogued in the Illinois Natural History Survey (INHS) Mollusk Collection, Champaign, Illinois (Appendix 1). Specimens came from the Big Muddy, Cache, Embarras, Kaskaskia, Little Wabash, Little Vermilion, Mackinaw, Sangamon, and Skillet Fork drainages and direct tributaries to the Ohio River (Big Grande Pierre, Lusk, and Rose creeks); images of external and internal valves of each specimen were made available via the Illinois Data Bank ( https://doi.org/10.13012/B2IDB-5609050_V1). Initial species identifications were made from external shell morphology of each specimen prior to genetic analysis. Those that were more inflated, had a pearlier nacre, and had a shorter average total length in mature individuals were identified as putative L. hydiana (n = 46; Fig. 2a), while specimens that were more compressed, had a duller nacre, and had a longer average total length in mature individuals were identified as L. siliquoidea (n = 37; Fig. 2b). Most of the putative L. hydiana were from specimens collected from the southern half of Illinois. Specimens used in this study were collected as part of other research projects, primarily during a statewide mussel survey for Illinois from 2009 to 2012. Funding constraints or curated tissue quality prevented us from using all available tissue samples from putative L. hydiana or L. siliquoidea in Illinois. Four L. hydiana specimens were collected from the Boeuf River, Louisiana, to provide comparative material (INHS 87783). In addition, comparative sequences were obtained from GenBank (Appendix 2).

Figure 3.

Approximate locations of reference materials used for this study. The previously published range for Lampsilis hydiana was adapted from Burch (1975) and Howells et al. (1996).

DNA was extracted from approximately 2 mm × 2 mm mantle and muscle biopsies using the MagMAX–96 DNA Multi-Sample Kit (ThermoFisher Scientific, Waltham, MA, USA) according to the manufacturer's instructions, except samples were eluted in 40 µl of elution buffer 1 and 2 instead of 100 µl. Polymerase chain reactions (PCR) and primers for cox1 and nad1 DNA amplification followed Campbell and Lydeard (2012). PCR products were sequenced on a Life Technologies 3730xl DNA Analyzer (Applied Biosystems, University of Illinois Chicago Genome Research Core). The cox1 region was 660 base pairs long, and the nad1 region was 834 bases long (including 30 bases of tRNA-Leu). Not all reads clearly resolved all bases, however, and unreadable bases were entered as unknowns. Sequences were aligned using BioEdit (Hall 1999). The sequence alignments are available at the Illinois Data Bank ( https://doi.org/10.13012/B2IDB-5609050_V1).

The relationships between species currently assigned to Lampsilis are not well resolved (Keogh and Simons 2019). To determine appropriate comparison taxa for our specimens, we performed preliminary phylogenetic analyses (details below) of all available cox1 and nad1 sequences for species currently assigned to Lampsilis (based on Williams et al. 2017), along with representatives of other genera in the tribe Lampsilini. These supported a clade of morphologically similar taxa that included L. siliquoidea and L. hydiana, along with Guadalupe Fatmucket Lampsilis bergmanni Inoue & Randklev, 2020, Arkansas Fatmucket Lampsilis powellii (Lea, 1852), Eastern Lampmussel Lampsilis radiata (Gmelin, 1791), L. sietmani, Rough Fatmucket Lampsilis straminea (Conrad, 1834), and Alabama Lampmussel Lampsilis virescens (Lea, 1858). In turn, this siliquoidea clade was most closely related to a clade that included Mucket Ortmanniana ligamentina (Lamarck, 1819), L. abrupta, and Higgins Eye Lampsilis higginsii (Lea, 1857), consistent with previous findings (Porto-Hannes et al. 2019; Inoue et al. 2020). Nomenclature follows Williams et al. (2017), with updates from recent works for O. ligamentina (Pfeiffer et al. 2019; Graf and Cummings 2021). As noted by Keogh and Simons (2019), confident assessment of the phylogenetic relationships of Lampsilis species within Lampsilini will require extensive sampling. Our goal was to find appropriate taxa for comparison with our L. siliquoidea–like and L. hydiana–like populations from Illinois, and we did not pursue the general phylogeny further. Based on these preliminary results, we included all available cox1 and nad1 sequences from the siliquoidea clade in our detailed analyses and used the ligamentina clade as the outgroup. The sequence identified as L. powellii in GenBank was treated as L. hydiana in our analyses (MF326971). Walters et al. (2021) also found this sequence to be L. hydiana, whereas true L. powellii was nearest to L. siliquoidea. A few sequences currently listed as L. radiata in GenBank (cox1: HQ153601, HQ153602, HQ153605; nad1: HQ153683, HQ153684, HQ153687, and HQ153691) were found to represent the “Cryptic Lampsilis sp.” of McCartney et al. (2016). Those sequences did not place in the siliquoidea clade based on McCartney et al. (2016) and our preliminary analyses, thus we excluded them from the present analyses. Percent differences and number of base-pair differences were calculated for all sequences from the siliquoidea clade using PAUP*4.0a167 (Swofford 2002). Because many individuals had only one gene or the other sequenced, cox1 and nad1 were compared separately in these analyses. These calculations omit bases with uncertainty (e.g., A versus N is not counted as a difference, nor is that position counted in the total number of bases for calculating percentage). We used the program ABGD (Puillandre et al. 2012) to test the differentiation between species in the siliquoidea clade. To test the cutoff for different divisions, the number of steps was increased to 20 and relative gap width decreased to one; other settings used the default values.

For phylogenetic analyses, we used both parsimony and Bayesian approaches and included all individuals with data for both nad1 and cox1. We concatenated the two genes, omitting the tRNA-Leu region. Lampsilis sietmani and L. abrupta had no nad1 data available, but we included representative cox1 sequences. In the ABGD analysis, one published cox1 sequence identified as L. hydiana (EF033270, from the Cossatot River in Arkansas), the Escambia River L. straminea (four sequences), and the Neches River sequence of L. sietmani (two individuals with identical sequences) were somewhat divergent from the other sampled individuals, so they were also included despite having only cox1 data available. If two individuals had the same haplotype for both cox1 and nad1, that combined haplotype was included only once in the phylogenetic analyses. Maximum parsimony and “Group present/Contradicted” (GC) bootstrap analysis (Goloboff et al. 2003) in the computer program TNT 1.5 (Goloboff and Catalano 2016) used all the “new technology” search options. Parsimony analysis used 500 random addition replicates, and the bootstrap analysis used 500 bootstrap replicates, each with 10 random addition replicates. Bayesian analyses used 10,000,000 generations with 10 runs, each with eight chains. We used PAUP* to test data partitions, setting the codon positions as data blocks. Using likelihood criteria and the “greedy” heuristic, the AICc criterion supported a GTR+1 model for cox1 positions 1 and 3 and nad1 position 2, GTR for cox1 position 2 and nad1 position 3, and GTR+G for nad1 position 1. MrBayes 3.2.7 was used for Bayesian analyses (Ronquist et al. 2012). Each codon position was treated as a separate partition. The parameters revmat, shape, pinvar, and statefreq were all unlinked. Convergence was determined by examining the standard deviation of split frequencies and confirming that they were under 0.01 (Ronquist et al. 2011), as well as by examination of the ESS values and trace plot in Tracer 1.7.1 (Rambaut et al. 2018). Tracer showed all ESS values well over 200, and the trace plot did not show any anomalies, so the standard 25% burn-in was used. We used PAUP* to calculate a majority-rule consensus of the Bayesian trees to obtain posterior probabilities, which facilitated outputting the tree as a graphic. Additionally, haplotype networks were constructed for L. siliquoidea and L. hydiana using median joining in PopART (Leigh and Bryant 2015).

RESULTS

The genetic results indicate that the L. siliquoidea and putative L. hydiana specimens from Illinois represent two distinct but closely related Lampsilis species. Sequences obtained for this study are available in GenBank (accession numbers MH560712-MH560762, MH560764-MH560777, MH588322-MH588394, MT537705-MT537725; Appendices 1, 2). Parsimony and Bayesian methods produced nearly identical results, with no differences in the affinities of the Illinois specimens. Parsimony analyses produced 319 trees of length 545 (as counted by TNT, which collapses polytomies, making a much smaller number of trees than PAUP*). In the Bayesian analysis, the standard deviation of split frequencies reached 0.01 after 1,735,000 generations. Lampsilis hydiana and L. siliquoidea were not sister taxa, but instead placed on different branches within the larger siliquoidea clade (Fig. 4). Lampsilis siliquoidea and L. radiata are sister taxa with relatively low genetic divergence, while L. hydiana is sister to L. bergmanni.

Sequences for L. hydiana versus L. siliquoidea had an average of 5.67% difference between them in cox1 and 7.43% in nad1, similar to most other interspecies differences within the clade (Table 1). In contrast, the average differences within L. hydiana and within L. siliquoidea for both genes were under 0.5%, with some Illinois specimens sharing haplotypes with specimens from elsewhere. In particular, identical haplotypes were found in some Illinois specimens and some of the topotypic L. hydiana specimens sampled in this study (Appendix 2, Figs. 5, 6). Likewise, the haplotype networks show much larger differences between L. hydiana and L. siliquoidea than within them. In ABGD for cox1, all partitions with gap priors between 0.0183 and 0.00162 separated the Illinois specimens (along with many from other states) into two groups, corresponding to L. hydiana and L. siliquoidea. No partitions supported any further division of L. hydiana or L. siliquoidea, except for recognizing the Cossatot River, Arkansas “L. hydiana” as distinct for priors of 0.00886 or less in the initial partition and 0.0144 or less in the recursive partition. The species most difficult to distinguish from L. hydiana were L. bergmanni and Mobile Basin L. straminea, which separated only at gap priors of 0.00428 or less, whereas L. siliquoidea and L. radiata were separated at gap priors of 0.0546 or less. Gap priors of 0.00127 or less split up individual variation, which produced 116 groups. For nad1, partitions with gap priors between 0.0183 and 0.00264 separated L. hydiana and L. siliquoidea without dividing either one. Again, separation between L. hydiana and L. bergmanni or L. straminea was less clear, requiring gap priors of 0.00207 or less, which also began to split off divergent sequences within L. hydiana. Separation of L. siliquoidea from L. radiata was supported with the recursive partition at a gap prior of 0.0183 or less and the initial partition at a gap prior of 0.0144 or less. Intermediate gap priors generally agreed with currently recognized species, though some currently recognized species were divided into more than one group, especially if there was a geographic gap in the sampling (such as L. sietmani from Texas versus the upper Mississippi drainage).

Our results support the presence of L. hydiana in the Big Muddy, Cache, Embarras, Kaskaskia, Sangamon, Ohio, and Little Wabash drainages of Illinois (Fig. 7). Most drainages that we examined contained only L. siliquoidea or L. hydiana; however, both L. siliquoidea and L. hydiana were confirmed in the Sangamon River basin and in Horse, Big Grande Pierre, and Lusk creeks. Our morphological identifications matched the genetic confirmation in most cases (72 of 83 individuals were identified correctly; Appendix 1). Ten specimens that were determined morphologically to be L. siliquoidea were genetically confirmed as L. hydiana, and one specimen that was determined morphologically to be L. hydiana was genetically confirmed as L. siliquoidea. Three of four sites where these mismatches occurred had both L. hydiana and L. siliquoidea genotypes present (Big Grande Pierre Creek, Lusk Creek, and Horse Creek; Fig. 2c, 2d). The only individual sequenced from Salt Creek (of two total specimens from the Sangamon River drainage) was genetically confirmed as L. hydiana but was determined morphologically to be L. siliquoidea.

DISCUSSION

We used genetic analyses to confirm the presence of L. hydiana in Illinois. This genetic confirmation supports the species determinations by Anson A. Hinkley and Frank C. Baker more than a century ago (Illinois Natural History Survey, Prairie Research Institute 2021 [INHS Collections Data], referenced via previous identification field), that were made prior to the availability of genetic tools. It is unclear why L. hydiana was never included on Illinois species lists even though shells were deposited in the INHS Mollusk Collection bearing this identification. Regardless, we now have genetic support that the range of L. hydiana extends to latitude 40.1° N in the Sangamon River drainage, which is well north of the previously published range limit of latitude 34.6° N (Burch 1975; Howells et al. 1996; Inoue et al. 2020). While historical literature proclaimed the morphological differences between L. hydiana and L. siliquoidea to be “very clear cut” (Isley 1924), we obviously did not find that to be the case for all the individuals analyzed. At sites where both L. hydiana and L. siliquoidea genotypes were present, we were unable to separate these individuals using only shell morphology (Fig. 2c, 2d). A more detailed morphological analysis may reveal additional characters that we did not consider, such as quantifying height to length ratio or measuring shell thickness (Keogh and Simons 2019). We recognize that our study's small sample size limits our understanding of the overall extent of L. hydiana in Illinois. Likewise, mitochondrial introgression, selective pressures, or incomplete lineage sorting (Doucet-Beaupré et al. 2012; Chong et al. 2016) could have produced anomalous genetic patterns. Additional nuclear molecular markers and a more detailed morphometric analysis of these populations may provide a clearer picture of relationships of Lampsilis populations in Illinois (Graf and Cummings 2006; Bogan and Roe 2008; Chong et al. 2016).

Figure 4.

Phylogram of the Bayesian majority-rule consensus tree. Numbers on branches are Bayesian posterior probability/bootstrap GC percentage, - indicates under 50% bootstrap support; * denotes branches that did not have room for labeling the probabilities directly on the branch. Numbers after a name indicate cox1 haplotype, followed by nad1 haplotype (see Appendix 2). Letters after a name indicate a new sequence from Illinois, topotype Lampsilis hydiana (t), the sequence identified as Lampsilis powellii (p), or haplotypes found in other published sequences and in new ones from Illinois (o). L. “hydiana” indicates the divergent Cossatot River sequence, and L. “straminea” indicates the Escambia River population.

Table 1.

Average, minimum, and maximum percent difference and number of base-pair differences in cox1 and nad1.

Figure 5.

Haplotype network for cox1 data. Numbers are the haplotype number (Appendices 1, 2). Bars on connecting lines indicate the number of base-pair differences between specimens; size of circles indicates the number of individuals with that haplotype.

Figure 6.

Haplotype network for nad1 data. Numbers are the haplotype number (Appendices 1, 2). Bars on connecting lines indicate the number of base-pair differences between specimens; size of circles indicates the number of individuals with that haplotype.

Figure 7.

Locations of all shell records observed for Lampsilis hydiana and Lampsilis siliquoidea from Illinois in the INHS Mollusk Collection, Champaign, Illinois (gray closed circles), with genetic confirmation of L. hydiana (green triangles) and L. siliquoidea (black squares) plotted within each watershed, with pertinent rivers labeled. Watershed shading indicates species assignments based on observed external shell morphology prior to genetic analysis; green shading = shell characters match L. hydiana and yellow shading = shell characters match L. siliquoidea.

Our results suggest that L. siliquoidea and L. hydiana are closely related to each other but are not sister taxa. The sister taxon relationship between L. siliquoidea and L. radiata fits with previous classifications, as L. siliquoidea has been treated as a subspecies of L. radiata (Watters et al. 2009). Relationships between other members of the siliquoidea clade have not been discussed in detail, particularly as L. sietmani and L. bergmanni were described very recently. However, a relationship between L. straminea, L. bergmanni, and L. hydiana would not be surprising on biogeographic grounds, as their ranges adjoin each other.

Our discovery of both L. hydiana and L. siliquoidea in Illinois highlights the possibility of overlooked diversity elsewhere. Previous studies found some specimens identified as L. hydiana from the Arkansas and Red River systems in Arkansas were genetically distinct from topotypic L. hydiana (Turner et al. 2000; Lewter et al. 2003; Harris et al. 2004). A cox1 sequence from one of those populations (Chapman et al. 2008; GenBank accession number EF033270) was divergent from true L. hydiana (Keogh and Simons 2019 and present analyses). Similarly, sequences in GenBank identified as L. powellii (from Breton et al. 2011 and Robicheau et al. 2018; GenBank accession numbers HM849075 and HM849218) matched topotypic L. hydiana (Walters et al. 2021 and present analyses). However, Harris et al. (2004) and Walters et al. (2021) found their sequences for L. powellii were closest to L. siliquoidea. Lampsilis straminea is reported to range from eastern Louisiana to central Florida, but data for cox1 separated specimens from the Escambia drainage versus those from the Mobile basin; no other populations have been analyzed genetically. Thus, further analyses of the siliquoidea clade are likely to reveal additional new records. The recent descriptions of L. sietmani and L. bergmanni highlight the possibility of additional undescribed or incorrectly synonymized species in this group (Inoue et al. 2020; Keogh and Simons 2020).

Our analysis provides additional support showing that the siliquoidea clade is one of several distinct groups currently assigned to the genus Lampsilis, even though species in this clade are morphologically and genetically distinct from the type of the genus, Pocketbook Lampsilis ovata (Say, 1817). Other species seem to be genetically divergent from both the siliquoidea clade and from type Lampsilis, including Texas Fatmucket Lampsilis bracteata (Gould, 1855) (Harris et al. 2004; Porto-Hannes et al. 2019; Inoue et al. 2020), the cryptic Lampsilis sp. of McCartney et al. (2016), and the clade of Northern Brokenray Lampsilis brittsi Simpson, 1900, Arkansas Brokenray Lampsilis reeveiana (Lea, 1852), and Speckled Pocketbook Lampsilis streckeri Frierson, 1927 (Harris et al. 2004). One other species recognized in Lampsilis, Neosho Mucket Lampsilis rafinesqueana Frierson, 1927, has not yet been analyzed genetically but has an unusual combination of anatomical and shell features (Harris et al. 2004). As Keogh and Simons (2019) pointed out, a thorough analysis of Lampsilini will be necessary to determine the correct placement of these taxa.

Accurate species delineation is critical to developing sound conservation strategies for freshwater mussels, particularly because many species of conservation concern are managed or closely monitored at the state level. At press time, three Lampsilis species are endangered in Illinois: L. abrupta and L. higginsii are federally protected, while Wavyrayed Lampmussel Lampsilis fasciola Rafinesque, 1820 is listed only at the state level. Other common, widespread Lampsilis species, such as Plain Pocketbook Lampsilis cardium (Rafinesque, 1820) and L. siliquoidea, are often used by local and state authorities for propagation and augmentation following habitat restoration efforts. Our analysis emphasizes the need for managers to follow best practices during augmentation and reintroduction activities to avoid cross-basin contamination, as hidden diversity may be present even in common, presumably well-understood species (McMurray and Roe 2017; Inoue et al. 2020).