Polymorphism in the aggressive mimicry lure of the parasitic freshwater mussel Lampsilis fasciola

Tissue sample collection

L. fasciola mantle tissue samples were collected for genotyping purposes by taking non-lethal mantle clip biopsies (Berg et al., 1995) from wild population lure-displaying female mussels during the summers of 2017, 2018, and 2021 in three rivers (Fig. 2). Maps were made in ArcGIS (ESRI, 2022) using U.S. Geological Survey (2022) as a basemap layer. Two of the sampling locations were in southeastern Michigan: the River Raisin at Sharon Mills County Park (42.176723, −84.092453; N = 30; 24 “darter-like”, six “worm-like”, collectively sampled in 2017, 2018 & 2020), and the Huron River at Hudson Mills Metropark, MI (42.37552, −83.91650; N = 13; 7 “darter-like”, six “worm-like”, collectively sampled in 2017, 2018, and 2020 under the MI Threatened and endangered species collection permit TE149). Both rivers flow into Lake Erie and are part of the Saint Lawrence drainage. The third location was in North Carolina: the Little Tennessee River (N = 10; 35.32324, −83.52275; N = 10, all were “darter-like” and sampled in 2017); this river is a tributary of the Tennessee River and part of the Mississippi drainage. Prior to each biopsy, photographs of the intact, undisturbed, lure display were taken with an Olympus Tough TG-6 underwater camera (Fig. S1).

Captive brood tissue samples

We also obtained tissue samples from 50 captive-raised individuals of a single brood that had been ethanol-preserved. In 2009, the Alabama Aquatic Biodiversity Center (AABC) established a culture facility for endangered freshwater mussels. The Center’s inaugural culture attempt, by co-authors Paul Johnson and Michael Buntin, was a proof-of-concept trial involving a single gravid female L. fasciola sourced from the Paint Rock River (another Tennessee River tributary; N 34˚ 47.733′,W 86˚ 14.396′) in Jackson County, AL (Fig. 2) on June 11, 2009. This female L. fasciola had a “worm-like” lure: the AABC data sheet for the trial 2009 host infection (Fig. S2) records that it was “bright orange and black” and lacked the “eyespots”, mottled body coloration, marginal extensions, and “tail” of the “darter-like” lure phenotype (Buntin & Johnson, 2009, personal observations). On July 13 2009, about 31,000 glochidia larvae were extracted from the female’s marsupia and used to infect Micropterus coosae (Redeye Bass) hosts sourced from the Eastaboga Fish Hatchery (Calhoun County, AL, USA) using standard protocols (Barnhart, Haag & Roston, 2008). The female mussel was then returned live to the Paint Rock River. Following completion of larval development on the fish hosts, about 9,300 metamorphosed juvenile mussels were recovered and reared, initially for the first few weeks in mucket bucket systems (Barnhart, 2006), then in a suspended upwelling system (SUPSYS) for 2 years with about 2,200 surviving. In 2011, this proof-of-concept culture experiment was terminated, and the survivors were donated to several research groups, with the majority used for toxicology experiments (Leonard et al., 2014a, 2014b).

Prior to the brood’s termination, Johnson noticed that a few females had attained sexual maturity and were displaying polymorphic lures (Figs. 3B, 3C). To substantiate that 2011 observation, we examined 50 individuals that had been preserved in 95% ethanol and shipped to Nathan Johnson (USGS) in Gainsville, FL in 2011. Because Lampsilis spp. juveniles and males produce a rudimentary mantle lure (Ortmann, 1921; Kramer, 1970), we were able to determine the primary lure phenotype (darter-like” or “worm-like”) of all 50 preserved brood members. Using a Leica MZ16 dissecting microscope, individual photomicrographs were taken of the preserved rudimentary lure structures (Figs. 3D, 3E and S3), and their respective lure phenotypes were identified independently by both T. Hewitt and by D. Ó Foighil. Additionally, tissue samples were acquired from all 50 individuals and included for phylogenomic analyses.

Phylogenomic analyses

DNA sequencing and raw data processing were performed using the protocol outlined in Hewitt, Haponski & Ó Foighil (2021a, 2021b). Genomic DNA was extracted from tissue samples using E.Z.N.A. Mollusk DNA kit (Omega Bio-Tek, Norcross, GA, USA) according to manufacturer’s instructions and then stored at −80 °C. The quality and quantity of DNA extractions were assessed using a Qubit 2.0 Fluorometer (Life Technologies, Carlsbad, CA, USA) and ddRADseq libraries were prepared following the protocols of Peterson et al. (2012). We then used 200 ng of DNA for each library prep. This involved digestion with Eco-RI-HF and MseI (New England Biolabs, Ipswich, MA, USA) restriction enzymes, followed by isolating 294–394 bp fragments using a Pippen Prep (Sage Science, Beverly, MA, USA) following the manufacturer’s instructions. Prepared ddRADseq libraries then were submitted to the University of Michigan’s DNA sequencing core and run in four different lanes using 150 bp paired-end sequencing on either an Illumina HiSeq 2500 or Illumina novaseq shared flow cell. Two control individuals of L. fasciola were run in each lane and reads for both individuals clustered together in every analysis with 100% bootstrap support, indicating no lane effects on clustering among individuals. Raw demultiplexed data were deposited at GenBank under the bioproject ID PRJNA985631 with accession numbers SAMN35800743–SAMN35800847. Individuals included in phylogenomic analyses can be found in Table 1, and museum ID numbers can be found in Table S1.

The alignment-clustering algorithm in ipyrad v.0.7.17 (Eaton, 2014; Eaton & Overcast, 2020) was used to identify homologous ddRADseq tags. Ipyrad is capable of detecting insertions and deletions among homologous loci, which increases the number of loci recovered at deeper evolutionary scales compared to alternative methods of genomic clustering (Eaton, 2014). Demultiplexing was performed by sorting sequences by barcode, allowing for zero barcode mismatches (parameter 15 setting 0) and a maximum of five low-quality bases (parameter 9). Restriction sites, barcodes, and Illumina adapters were trimmed from the raw sequence reads (parameter 16 setting 2), and bases with low-quality scores (Phred-score < 20, parameter 10 setting 33) were replaced with an N designation. Sequences were discarded if they contained more than 5 N’s (parameter 19). Reads were clustered and aligned within each sample at an 85% similarity threshold, and clusters with a depth <6 were discarded (parameters 11 and 12). We also varied the number of individuals required to share a locus from ~50% to ~75%.

We analyzed the two concatenated ddRAD-seq alignment files (50% and 75% minimum samples per locus) using maximum likelihood in RAxML v8.2.8 (Stamatakis, 2014). A general time-reversible model (Lanave et al., 1984) was used for these analyses that included invariable sites and assumed a gamma distribution. Support was determined for each node using 100 fast parametric bootstrap replications. Lure phenotype information was recorded and mapped on to the phylogenetic tree. Phylogenetic signal of lure phenotype was tested using Pagel’s (1999) λ in R (R Core Team, 2018) with the ‘phylobase’ package (Hackathon et al., 2013).

River raisin mantle lure phenotype ratios over time

Mid-20^th century L. fasciola specimens collected at the Sharon Mills County Park site (Raisin River, MI, USA; Fig. 2A) are preserved as part of the University of Michigan’s Museum of Zoology wet mollusk collection. They stem from eight different collecting events between 1954 and 1962 (Table S2), and their presence afforded an opportunity to assess the stability of the L. fasciola “darter/worm” mantle lure polymorphism in that population over a six-decade time interval. All of the museum specimens, males as well as females, were examined to determine whether their fully-formed (female) or rudimentary (male) mantle lures were “darter-like” or “worm-like”. For females, this could be achieved by simple visual examination, but male lure classification required a dissecting microscope. The percentages of mantle lure phenotypes observed in the Sharon Mills County Park population was compared among mid-20th century (UMMZ preserved females and males) and 2017 (field photographs and videos of displaying females) samples using a Fisher’s exact test, implemented in R.

Putative lure mimicry models

Population-specific putative model species for the L. fasciola mantle lure mimicry system were investigated at the River Raisin Sharon Mills County Park study site (Fig. 2), in part because of the availability of a comprehensive ecological survey of Raisin River fishes (Smith, Taylor & Grimshaw, 1981). “Darters”—members of the speciose North American subfamily Etheosomatinae— have been implictly identified as models for the predominant “darter-like” mantle lure phenotype (Zanatta, Fraley & Murphy, 2007), and they are preyed upon by Micropterus dolomieu (Surber, 1941; Robertson & Winemiller, 2001; Murphy et al., 2005), L. fasciola’s primary fish host (Zale & Neves, 1982; McNichols, 2007; Morris et al., 2008; McNichols, Mackie & Ackerman, 2011; VanTassel et al., 2021). Ten species of Etheosomatinae occur in the River Raisin, as does M. dolomieu (Smith, Taylor & Grimshaw, 1981).

River Raisin gravid female L. fasciola engage in mantle lure displays from May-August. During the summer of 2017, a total of 27 different displaying females were photographed along a 150-m stretch downstream of the dam at Sharon Mills County Park using an Olympus Tough TG-6 underwater camera. Individuals were located by carefully scanning the river bed with mask and snorkel to try and approximate the real ratios of phenotypes at this site. Additional lure photos were taken by coauthor Paul Johnson at the AABC of individuals from the Paint Rock River (AL). The lures were first categorized into broad groupings based on visual similarity, in terms of morphology and coloration. These groupings were then used to identify putative host prey fish model species from those present in the River Raisin drainage (Smith, Taylor & Grimshaw, 1981), based on similarities in size, shape, and coloration. Putative model species were further assessed based on their relative local abundance (Smith, Taylor & Grimshaw, 1981) and on their range overlap with both mimic and receiver. We also photograph and document the male rudimentary lures for both L. fasciola and L. cardium, taken from the River Raisin (Fig. S4). Geographic ranges of L. fasciola, the primary host M. dolomieu, and each prospective model species were produced by hand in Arcgis software (ESRI, 2022), and the overlap between L. fasciola, M. dolomieu, and each putative model species was assessed using Arcgis software.

Behavioral analyses

Standardized video recordings of 27 mantle lure-displaying female L. fasciola (15 “darter-like” and 12 “leech-like”) were recorded using a Go Pro Hero 6 camera in the summer of 2018 at the two different southeastern Michigan study sites: Sharon Mills County Park (River Raisin) and Hudson Mills Metropark (Huron River). All “darter-like” individuals were grouped together. An additional four video recordings of the lure behavior of sympatric Lampsilis cardium, a well studied congener lacking pronounced mantle lure polymorphisms (Kramer, 1970; Haag & Warren, 1999) were collected from the Sharon Mills site to assess interspecific variability in lure behavior. Recordings were captured from a top-down perspective during daylight hours using a standardized frame that included a metric ruler and a Casio TX watch to record date, time, and water temperature data within the video frame. For each displaying female, videos of the lure movements were recorded for 10 min at 120 frames-per-second. Setting up the camera occasionally disturbed the mussels, and video recordings began after waiting some time (usually 2–15 min) until the behavior qualitatively returned to its prior state. Analysis of the videos involved manually recording mantle lure movements for 20,000 frames (2.8 min), starting at 5,000 frames (42 s) to to avoid any camera shaking or hands accidentally blocking the view. The frame numbers when an individual movement began, defined as the first frame where contraction of mantle tissue was observed, and ended, defined as the time that mantle lure returns to it resting state, were noted. Movements of the left and the right mantle lure flaps were recorded seperately.

To quantitatively assess behavioral differences among samples, gait analysis diagrams were created in R for each displaying mussel. Because the lure is mimicking the swimming locomotion of fish, and fish locomotion has been characterized using gait analysis (Liao et al., 2003), we used gait analysis methods to characterize the non-locomotory motions that generate the luring behavior. Averages and standard deviations for the time intervals between lure undulations (the time between the start of one movement and the start of the next) were calculated for each side of each individual, as well as duration undulation (the time between the start of one movement and the end of that movement) and proportion of movements synchronized. Movements were defined as synchronized if the start of a movement on one side was within four frames of the start of a movement on the corresponding side. Proportion of movements synchronized were calculated by dividing the number of synchronized movements by the sum of left movements only, right movements only, and synchronized movements. A Kruskal-Wallis test was used to test for overall differences among lure groups (L. fasciola “darter-like”, L. fasciola “worm-like”, and L. cardium), and pairwise Wilcoxon Signed rank tests were used to compare groups directly with a Bonferroni p value adjustment to correct for multiple tests. A Spearman correlation was used to test for an effect of water temperature on time interval between lure undulations.

To further explore differences in lure behavior among groups, we used a general linear mixed model (GLMM), with sample ID as a random factor, to test for differences in lure movement intervals. The GLMM approach, unlike simple mean comparisons, allows the inclusion of all lure movements for all individuals in the model. Because displaying mussels all varied in the number of lure movements recorded over 20,000 frames analyzed, a dataset of 1,000 random bootstrap values was constructed for each individual by randomly sampling values, with replacement. Models were fitted using the ‘lmerTest’ package in R, and Satterthwaite’s (1946) Method (Kuznetsova, Brockhoff & Christensen, 2017) was used to test for significance of fixed effects of lure phenotype on the interval between lure undulations.

Captive brood

Two independent classifiers concurred that the 50 preserved specimens from the same maternal brood included 33 “darter-like” (66%) and 17 “worm-like” (34%) individuals (Figs. 3D, 3E and S4).

ddRAD-seq and phylogenomic analyses

Genomic sequencing returned raw reads ranging from 258,664 to 13,366,692 per individual across the 108 unionid specimens included in the analyses comprising samples of the ingroup L. fasciola, sourced from four different populations, along with outgroups L. cardium and Sagittunio nasuta. Mean coverage depth for the 85% clustering threshold ranged from 2.03 (L_fasciola_AL_mom_2) to 14.25 (L_fasciola_Raisin_16; Table 1). Between 28,725 and 16,161 homologous loci were identified across the two best ddrad datasets (85–50% and 85–75% respectively) and the number of loci recovered was generally consistent among all samples.

The maximum likelihood tree produced by RAxML (Fig. S4) recovered the following ingroup/outgroup topology: (S. nasuta (L. cardium, L. fasciola)) with outgroup branch lengths greatly exceeding those of the ingroup. To optimize the legibility of ingroup relationships, a compressed, color-coded graphic excluding S. nasuta was constructed (Fig. 4). A nested series of phylogenetic relationships was recovered for the four L. fasciola fluvial populations with the two Michigan drainages being paraphyletic: (Little Tennessee River (Paint Rock River (River Raisin (River Raisin, Huron River))). The ingroup topology also showed evidence of within-population genealogical relationships with all Paint Rock River brood members forming an exclusive clade (Fig. 4).

The respective primary mantle lure phenotypes—“darter-like” or “worm-like”—of all 92 L. fasciola ingroup individuals are indicated in Fig. 4. Note that three of the four population samples—Little Tennessee River, River Raisin and Huron River—were exclusively composed of mantle-lure displaying wild females, and the latter two samples were polymorphic in mantle lure composition. Regarding the Paint Rock River sample, polymorphic lures were restricted to the 50 captive-raised AABC brood members sourced from a gravid, wild female in 2009 (not included in the analyses). The ingroup phylogeny (Fig. 4) contained two polymorphic mantle lure clades, one composed of both Michigan populations (River Raisin and Huron River), the other consisting only of the AABC brood, and both clades had individuals of either lure phenotype interspersed across their respective topologies. Little phylogenetic signal associated with either primary mantle lure phenotype (λ = 0.21; P = 0.13).

Phenotypic ratios over time

Table S2 summarizes the sex and primary lure phenotypes of 57 L. fasciola specimens collected from 1954–1962 at the River Raisin Sharon Mills County Park study site (Fig. 2A) and preserved in the University of Michigan Museum of Zoology’s wet mollusk collection (Figs. 5B and 5C). These historical samples had a collective “darter-like” to “worm-like” ratio of 48:9, with 84.2% of individuals having the more common “darter-like” mantle lure phenotype and 15.8% having the “leech-like” phenotype. Figure 5A contrasts the mid-20^th century lure phenotype ratios with a contemporary (2017) estimate in that same population, based on photographic recordings of 27 displaying females. The contemporary ratio was 23:4, with 85.2% of individuals having the more common “darter-like” mantle lure phenotype and 14.8% having the “leech-like” phenotype. The contemporary ratio was not significantly different from the historical ratio (Fisher Exact Test, Χ² = 0.01, P = 0.91).

Putative raisin river lure mimicry models

The field photographs of 27 displaying female L. fasciola mantle lures in the Raisin River Sharon Mills County Park population in 2017 (Fig. S1) were categorized into either “darter-like” (Zanatta, Fraley & Murphy, 2007) or “worm-like” (McNichols, 2007), as summarized in the Materials & Methods section. In addition to the specific features that separate these two primary mantle lure phenotypes (presence/absence of “eyespots” mottled pigmentation, marginal extensions and a “tail”), “darter-like” lures exhibited a much higher degree of variation than did “worm-like” lures, both within populations and across the species range. The latter lure phenotype exhibited a relatively simple, uniform morphology combined with a bright orange coloration underlain with a black basal stripe phenotype in Michigan (Figs. 6F–6H), in Alabama (Figs. 6I and 6J), and in North Carolina populations (Fig. 2A in Zanatta, Fraley & Murphy, 2007). In contrast, Raisin River “darter-like” mantle lures exhibited individual-level variation that was sometime quite marked, especially in details of their pigmentation, and to a more limited degree in their marginal extensions (Figs. 6A–6D and S1). Among individual variation was most pronounced for inter-population camparisons, e.g., see the much larger “tail” in the lure displaying Paint Rock River, Alabama specimen shown in Fig. 6E, and also the wider range of phenotypes present in North Carolina populations (Figs. 2B–2D in Zanatta, Fraley & Murphy’s (2007). Male mantle lure rudiment photos are found in Fig. S5.

Despite the considerable individual variation among the 24 photographed Raisin River “darter-like” mantle lures (Fig. S1), it was possible to identify some shared phenotypic motifs, especially in pigmentation pattern, and to informally categorize 23/24 mantle lures with those shared motifs into four general groupings. Group 1 “darter-like” mantle lures were characterized by prominent, chevron-like, darker pigmented blotches, spaced regularily along the flanks of the lure, over a lighter background coloration (Fig. 6A). This general pattern occurred in 7/24 Raisin River “darter-like” lures examined. Group 2 was rarer (3/24 individuals) and consisted of a darker background coloration with large orange blotches spaced regularily along the lure flanks, some divided into “dorsal” and “ventral” elements (Fig. 6B). Group 3 (9/24 individuals) lures were characterized by prominent dark lateral maculation spatially divided into a “ventral” pattern of larger, regularly spaced blotches and a “dorsal” pattern of more numerous, irregular blotches of different sizes (Fig. 6C). Finally, Group 4 (3/24 individuals) lures were characterized by an evenly-dispersed, fine grained freckling of numerous pigmented spots over a lighter background (Fig. 6F).

To explore putative model species for the four L. fasciola Raisin River “darter-like” mantle lure groupings (Figs. 6A–6D and 6F), potential matches (in terms of size, shape and coloration) were sought among the 10 species of Etheosomatidae that occur in the River Raisin (Smith, Taylor & Grimshaw, 1981), many of which display pronounced sexual dimorphism in body coloration (Kuehne & Barbour, 2014). The best apparent matches, depicted in Fig. 7, are as follows: Group 1 (Fig. 6A)-Etheostoma blennioides (female coloration), Group 2 (Fig. 6B)-Etheostoma exile (male coloration), Group 3 (Fig. 6C)-Percina maculata (male and female coloration) and Group 4 (Fig. 6D)-Etheostoma microperca (female coloration).

The distinctive color combination of the L. fasciola “worm-like” lure-solid orange with a black underlay (Figs. 6F–6J) does not match that of any Raisin River darter, or other Raisin River fishes (Smith, Taylor & Grimshaw, 1981). It does, however, match the coloration and size/shape, of the common North American leech, Macrobdella decora, which is widespread and abundant in eastern North America watersheds and typically feeds on aquatic vertebrates (Klemm, 1982; Munro et al., 1992). M. dolomieu, L. fasciola’s primary host fish, is a generalist predator with a diet of aquatic invertebrates, including leeches, in addition to small fishes (Clady, 1974), and recreational fishers frequently use live and/or artifical leeches as bait to catch this species (Cooke et al., 2022). Based on the available data, it seems that Macrobdella decora may be the best model species candidate for the “worm-like” (McNichols, 2007) L. fasciola mantle lure phenotype, and will hereafter be referred to as the leech phenotype.

The geographic range of the mimic, L. fasciola, is a subset of that of its receiver/host M. dolomieu (Fig. 2), and the extent of range overlap with all five putative River Raisin mantle lure models were calculated using Arcgis (Table 2) and are shown in Fig. 8. Three of the five putative models-Etheostoma blenniodes, Percina maculata and Macrobdella decorata have extensive overlap with L. fasciola’s range, but E. exile and E. microperca are restricted to northern portions.

Behavioral analyses

Lure movements for both species consist of small undulations along the length of the mantle lure, beginning about two thirds of the way towards the “tail” side of the lure, and travelling towards the “head” of the lure. The L. cardium lure movements always occur on both left and right sides of the mantle lure simultaneosly, while both L. fasciola lure phenotyopes exhibit independent movement of the left and right sides of the lure. Qualitatively, L. fasciola and L. cardium have very different mantle lure display behaviors. Gait diagrams show a clear distinction between L. cardium and both primary L. fasciola lure phenotypes (“darter” and “leech”). L. cardium consistently exhibited a synchronized lure undulation of both mantle lure flaps, whereas L. fasciola samples frequently moved left and right mantle flaps independently (Fig. 9 and S6). Gait diagrams also qualitatively showed that L. fasciola is charicterized by a high level of variability in undulation interval, L. cardium is much more regular in undulation interval with a steady beat frequency.

Intervals between movements in L. cardium were shorter (Wilcoxon test, W = 0, N = 4 L. cardium, 15 darter lure L. fasciola, 13 leech lure L. fasciola, p < 0.01 for both comparisons), less variable (Wilcoxon test, W = 0, p < 0.01 for both comparisons) and more synchronized (Wilcoxon test, W = 60, 48, p < 0.01 for comparisons with darter and leech lures, respectively) than in L. fasciola (Fig. 10). There was no difference in duration of lure undulations between L. cardium and both L. fasciola lure phenotypes (Wilcoxon test, W = 42,40, p = 0.26,0.06 for comparisons with darter and leech lures, respectively). Differences between the lure types of L. fasciola were smaller, with inter-movement intervals in the darter phenotype that were longer (Wilcoxon test, W = 142, p = 0.01) and marginally non-significantly more variable (W = 128, p = 0.07) but similar in duration (Wilcoxon test, W = 97, p = 0.76) and degree of synchronization (Wilcoxon test, W = 64, p = 0.22, Fig. 10). Table S3 details the time, date, location, temperature and summary statistics of all 34 lure display field recordings.

GLMM were used as an alternative analytical approach that included a large, bootstrapped dataset of lure movements. GLMM results were similar to those of the mean comparisons, with L. cardium individuals having shorter movement intervals than either L. fasciola lure morphs (an estimated 0.21 s for L. cardium vs. 3.2 and 1.0 s, respectively for L. fasciola “darter” and “leech” lures). However, these fixed effects are not statistically significant.