Efficient dispersal and substrate acquisition traits in a marine invasive species via transient chimerism and colony mobility

D. vexillum colony collection and maintenance

D. vexillum fragments (hereafter ‘ramets’) were harvested from submerged man-made structures (0.3–1.5 m below sea surface) in the Nelson city marina (New Zealand: 41°S, 173°E) and then maintained in the laboratory as described in Rinkevich & Fidler (2014). All collection and culturing procedures complied with the legal and ethical requirements of the New Zealand (N.Z.) government, the Cawthron Institute (Nelson, N.Z.) and University of Auckland (Auckland, N.Z.). Note that the waters of the Nelson city marina are publically owned and collection of invasive invertebrates from such waters did not require legal permits. The harvested colonies were separated by a distance of >1.0 meter in an effort to minimize the probability that colonies may be related by asexual reproduction and therefore share identical genotypes. The D. vexillum ramets were maintained in 16 L glass tanks with high flow-through (∼1 L/min.) unfiltered, recirculated seawater with 24 h florescent lighting. The recirculated seawater was continuously monitored and regulated (salinity: 34.5 ± 1.0 ppt; pH: 8.13 ± 0.08; water temperature: 17.7 ± 0.7 °C; ambient air temperature: 20.0 ± 1.0 °C, oxidation/reduction potential: 332 ± 21 mV), with continuous aeration provided by air stones. Ramets were fed with algal cultures, ∼100 ml/tank of Isochrysis galbana (∼9 × 10⁶cells/ml) provided three times/week. At the end of each experiment tanks, and associated materials, were cleaned of biofouling organisms using fresh water.

Pairing of D. vexillum ramets

Small ramets (∼1.0 cm²) were cut from the colony edges and attached with cotton thread to 5.0 × 7.5 cm glass slides, positioned vertically in slide staining racks (Rinkevich & Fidler, 2014; Rinkevich & Shapira, 1998). Ramet surfaces were gently cleaned of surface fouling organisms using small paint brushes while their vertical positioning helped to reduce on-going accumulation of debris, faecal residues and food particles. To initiate inter-ramet fusions the paired ramets were juxtaposed on the glass slides, with their natural growing edges placed at a distance of ∼1 mm. Subsequent growth and movement of the ramets were monitored and recorded using still or time-lapse photography. For subsequent genotyping small sections were dissected from the ramets on the final day of the fusion process and these sub-samples stored (100% (v/v) ethanol, −20 °C).

Photographic recording of fusion experiments

Set I—Still image photography: For the majority of the pairing experiments the glass plates were taken from the seawater tanks at intervals of 1–2 days and images recorded as rapidly as possible, using a Canon PowerShot G15 camera (Canon, Tokyo, Japan). The whole process of photographing and taking written notes took approximately 10 min with the D. vexillum immersed in shallow seawater throughout the time.

Set II—Time lapse photography: For a subset of the pairing experiments time lapse photography (1 image /5 min. over 10–12 days) was used with the camera, in a weather resistant housing, positioned on the outside of the tanks (Brinno TLC 200 Pro digital camera; Brinno, Taipei, Taiwan; infrared filtered 6 mm. CS mount lens). The time lapse camera file was exported into an image sequence using VirtualDub (http://www.virtualdub.org/), and the resulting image sequence then imported into ImageJ 1.48v (http://imagej.nih.gov/ij/) and transformed into a movie (.avi file) using JPEG compression (frame rate: 65 images/s).

Genomic DNA purification

Ramet samples of varying size (0.2–0.8 cm²), containing both tunic and zooids, were placed in ∼1.0 mL 100% (v/v) ethanol which was changed twice in the subsequent week and then stored long-term at −20 °C. Total genomic DNA (gDNA) was extracted from the samples using a commercial kit following the manufacturer’s instructions (GSpin™ Total DNA Extraction Kit; iNtRON Biotechnology, Inc., Gyeonggi-do, South Korea, animal tissue protocol with a tissue lysis step of 3 h, 56 °C). The purified gDNA was stored at −20 °C with DNA concentrations determined using a nanophotometer (Implen GmbH, Munich, Germany).

Polymorphic microsatellite loci genotyping

The five microsatellite loci used for D. vexillum ramet genotyping were from two sources: (i) three loci DVEX05, DVEX23 and DVEX32 (Table S1) were previously reported by Abbott et al. (2011); while (ii) two loci DVEX18 (GenBank accession number KU167099) and DVEX19 (acc. no. KU167100) (Table S1), were identified in this work using 454 pyrosequencing of D. vexillum genomic DNA performed by an external contractor (Ecogenics, Balgach, Switzerland). Amplification and florescent labelling of microsatellite loci derived PCR products was achieved using the three primer strategy of Schuelke (Schuelke, 2000). Thus, an 18 bp generic ‘M13-tag’ sequence (5′-TGTAAAACGACGGCCAGT-3′) was added on the 5′ end of the loci specific forward primers Schuelke (Schuelke, 2000) (Table S1). PCR mixes (10.0 µL) consisted of 1x MyTaq™ HS Mix (cat. no. BIO25045, Bioline, London, UK), 30–35 ng of template genomic DNA, M13-tagged-locus specific forward primer (0.06 µM), locus specific reverse primer (0.2 µM) and labelled M13-tag primer (0.2 µM) which as 5′ labelled with one of four alternative fluorescent dyes: 6-FAM™ (blue), VIC^® (green), NED™ (yellow), PET™ (red) (Table S1). Thermo-cycling conditions for the DVEX18 and DVEX19 loci amplifications were: 94 °C/2 min, 1 cycle ; 94 °C/30 s, 60 °C/30 s, ramping to 72 °C at +0.2 °C/s, 72 °C/30 s, 15 cycles; 94 °C/30 s, 65 °C/30 s, 72 °C/30 s, 30 cycles; 72 °C/10 min; 60 °C/30 min; 15 °C/hold. Thermo-cycling conditions for the DVEX05, DVEX23 and DVEX32 loci followed the ‘touchdown PCR’ protocol of Abbott et al. (Abbott et al., 2011): 95 °C/2 min, 1 cycle; 95 °C/30 s, 62 °C dropping 2 °C/2 cycles to 54 °C/30 s, 72 °C/30 s, 10 cycles; 95 °C/30 s, 54 °C/30 s, 72 °C/30 s, 23 cycles; 72 °C/10 min; 60 °C/30 min; 15 °C/hold. Both thermo-cycling protocols included a 60 °C/30 min step at the end, as this is thought to promote the Taq DNA polymerase catalyzed addition of 3′ As to the amplicons thereby reducing length heterogeneity arising from the PCR itself rather than from actual allele length variation. The fluorescently labelled amplicons were stored in the dark at +4 °C before being diluted 1:4 in milliQ water and then pooled so that each microsatellite amplicon in a ‘pool’ was labelled with a different dye and could be identified by their differing emission spectra (Table 1). Amplicon lengths were estimated by an external contractor (Massey Genome Service, Massey University, New Zealand) using capillary electrophoresis (ABI3730 DNA analyser; Applied Biosystems, Waltham, U.S.A.) and GeneScan™-500 LIZ™ size standards (Applied Biosystems). Electropherogram results were used to estimate amplicon sizes, using the software Peak Scanner™ v2.0 (Life Technologies, Carlsbad, U.S.A.).

To discriminate between the two genotypes that potentially contribute to a particular D. vexillum entity, we required the two genotypes to differ by a minimum of one allele at a minimum of two microsatellite loci (Tables S1, S2). The specific loci used in the pairings studied in detail are summarised in Table S2. The relative heights in the electropherogram traces of the genotyping discriminating alleles were used to provide semi-quantitative estimations for the relative amounts of the two genotypes in the gDNA preparations from sections taken from the chimeric entities and, by implication, of the relative biomass of the two genotypes in the chimeric entities. The relative genotype ratios, as deduced from the electropherogram traces, were summarised using the following symbolism where X and Y denote the two different genotypes: X∼Y, approximately equal amounts of both genotypes; X/0 or 0/Y, one of the genotypes not detected; X>Y or Y>X, one of the two genotypes clearly more abundant then the other and X≫Y or Y≫X, one genotype is much more abundant but there is a trace of the other genotype.

Estimation of ramet surface areas

Digital photographs of the slides from the day of initial fusion and the day of experimental termination were processed using the area/length analysis tool of the NCRI CPCe software (Kohler & Gill, 2006). At day of fusion, the outline of each fragment from either D. vexillum ramet was manually traced and converted by the software into surface area values (cm²), a protocol employed at the end of the experiments for each sample within a chimera. All the CPCe output files were exported to Microsoft Excel 2016. Then, we imposed the microsatellites’ semi-quantitative ratio results on the area calculations, as follow: a single genotype detected in the sample (X or Y; 100% of the fragment’s surface area to the appropriate genotype), X∼Y (both genotypes in the tissue sample revealed approximately the same DNA amounts, 50% surface area for each genotype), X>Y or Y>X (75% and 25% of the surface area, correspondently) and X≫Y or Y≫X (95% and 5% of the surface area, correspondently). Next, a cumulative surface area / genotype for each chimeric combination was computed, not including the ‘nd’ cases where the microsatellite based discrimination had not been successful. Statistical analyses were performed using the IBM SPSS Statistics software for Microsoft Windows, version 23. After determining normal distribution for all genotypes a one-way analysis of variance (ANOVA) test with internal pairwise comparisons was used to determine the differences between the mean changes per day for the D. vexillum genotypes.

Set I: single frame photographic records of D. vexillum inter-colony fusion processes

Using ramets taken from four (termed: A to D) putatively allogeneic colonies, allogeneic and isogeneic pairings were established. Six possible pairwise combinations of colony ramets (i.e., AxB, AxC, AxD, BxC, BxD, CxD), along with their isogenic pairing controls, were established in duplicate (n = 12 allogeneic pairings in total) as described in the materials and methods. Observations and still photographs of the paired ramets were taken on a daily basis. The photographic records of all 6 allogeneic pairings are shown in Fig. S1 with summary written descriptions provided in Table 1. Set I pairing experiments were terminated seven to twelve days after initial pair establishment and sub-sections of the D. vexillum distributed on the slides were genotyped using polymorphic microsatellite loci thereby allowing semi-quantitative determination of the spatial distribution of the paired ramet genotypes (Fig. S1). As an example Fig. 1 shows images from four separate days following a ramet A vs. ramet D pairing and is illustrative of the general phenomena observed in such allogeneic pairings (Fig. 1). What follows is a summary of the observations made from the ramet pairings and generalisations that emerge (Table 1, Fig. S1). In all ramet pairings (iso- and allogeneic combinations) tunic contacts occurred one to three days after initiation of the experiment and, in some cases, such tunic contacts were followed by apparent fusion of the two ramet matrices. All the isogeneic paired ramets fused to form single colonies. Initial fusions between allogeneic ramets were followed by either morphologically-mixed chimerism (i.e., where zooid containing regions from both partners appeared to be intermingled with each other) or by visibly-sectorial chimerism (i.e., where only very limited mixing of zooid containing regions was observed) (Fig. 1, Table 1, Fig. S1). Whatever the initial degree of chimera mixing, shortly after tunic fusion (∼1–2 days), one or both of the paired ramets started coordinated movement which were apparently directionally away from the fusion zone (Fig. 1C, Fig. S1). The coordinated movement of zooid containing regions was followed by the formation of zooid-depauperate areas which appear as translucent zones that clearly lack the mustard-coloured zooids (Fig. 1, Fig. S1). Such regions, lacking large numbers of zooids, appear to degenerate quite quickly (∼2–3 days after fusion), in contrast with the opaque mustard-coloured tunic regions which contain many zooids and remain apparently healthy. In some cases the gliding-movements of zooid clusters across the glass substrate have gone as far as the glass plate edges and indeed even over to the opposite side of the glass slide (Table 1, Fig. S1), resembling the retreat growth phenomenon in interacting allogeneic pairs of the colonial tunicate Botryllus schlosseri (Rinkevich & Weissman, 1988). The sliding movements of the zooid-rich regions and the associated generation of zooid-depauperate zones led in many cases to the splitting of the chimeric entities into two non-equal parts seven to twelve days after initial tunic fusions. Isogeneic D. vexillum isogeneic pairs, while fused in the same way as the allogeneic pairs and showed fast growth and movement on the substrates, did not retreat from each other, and when splitting into daughter subclones following fast growth in every direction, these events occurred haphazardly, in different parts of the growing colonies, unrelated to the former fusion sites. It is also evident that chimerism had no apparent impact on either ramet’s survivorship, as all pairs survived over the period of observation.

To assess the degree of genetic chimerism, at the end of the fusion experiments the D. vexillum ramets, including both the translucent zones and the zooid-rich regions, were sampled (typically as 12–25 fragments, Fig. S1) for genotyping at polymorphic microsatellite loci that discriminate the two genotypes of the paired ramets. The genotyping results for each pairing experiment are shown in detail in Fig. S1. What follows is a summary of generalisations that emerge from the genotyping and the associated semi-quantitative estimation of genotype ratios in the dissected regions. It appears that in most, perhaps all, of the pairings, the initial chimerism arising from tunic fusions is followed by what appears to be a genotype-based ‘depuration’ process whereby the two genotypes associated with the chimera actively spatially separate from each other. Firstly, zooid intermingling is restricted to the fusion border areas and does not involve the intermixing of genotypes throughout the fused entity (Fig. S1). Secondly, and subsequent to the tunic boundary fusions, one or both of two interacting genotypes start to grow away from the zooidal-intermingled areas to form zones with exclusively, or nearly exclusively, a single genotype. The associated splitting of the formerly chimeric entity into two or more fragments generates translucent, zooid-depauperate zones which contain a mixture of both the paired genotypes (Fig. S1). Note that throughout the observed processes there was no clear evidence of any morphological allorejection type events or of general tissue damage, an observation that should be interpreted with caution as it may simply reflect the low level of magnification used in this study. In addition, replicates of the same pair-wise combinations often have somewhat differing outcomes (Table 1, Fig. S1), suggesting a non-genetic and non-hierarchical nature to the fusion and subsequent depuration processes.

We evaluated each genotype’s total surface area (cm²) at both the onset and termination of pairing experiments (Table 2). Then, we calculated percentage changes in area/day for each genotype in each chimera (Table 2). On the days of fusion ramet sizes had a size range of 0.4–3.5 cm² with most 1.0–2.3 cm². At the termination of the experiments (i.e., after 7–13 days) the ramets, occupied 1.3–15.1 cm² of surface area, with most in the range of 3.6–7.9 cm², reflecting a very wide range of 4.5–112.1% increases in area/day with most (13/22) ramet areas increasing at rates of 34–77%/day (Table 2). The ‘nd’ samples were limited and few, representing 1.2–3.1% of the total surface areas (combinations AxC-1, BxD-2 and CxD-2) and 10.2% of the total surface area for combination AxB-2 (Table 2).

As no hierarchy in growth patterns was documented when aligning the growth rates with genotypic combination, the average genotypic growth rates/genotype was performed, averaging the 6 ramets from each genotype. Results (genotypes: A = 10.4 ±5.3 cm²; B = 51.7 ± 28.7 cm²; C = 40.8 ±20.3 cm²; D = 39.4 ± 13.7 cm²) showed that genotype A represents the slowest growth rates (∼0.25 of genotypes B, C, D) but statistically significantly different from only genotype B (ANOVA, p = 0.0165).

Set II: time-lapse photographic records of D. vexillum inter-colony fusion processes

To obtain a higher resolution description of D. vexillum inter-colony fusion and associated mobility we used time lapse photography. As is Set I, D. vexillum ramets were paired on glass slides and then filmed using a camera placed outside the seawater tanks in which the colonies were maintained. High quality time lapse records were obtained for two allogeneic pairwise combinations (made of ramets of genotypes H, I, L), denoted H x I and H x L. that are analysed here and can be viewed using the supplementary movie files: movie-S1 (H x I) and movie-S2 (H x L). Single frame images separated by 24 h/1 day increments were selected to help summarise the fusion/mobility processes: H x I, Fig. S2; H x L, Fig. S3, For the H x I pairing, where the partner genotypes could be distinguished by two microsatellite loci (Table S2), 12 dissected regions were taken 12 days after the pairing establishment (Fig. 2).

In both the H x I and H x L pairings the edges of the two paired ramets came into visible contact after ∼2–3 days (Figs. S2 and S3, images on Days 2 and 3). However events subsequent to this initial contact differed strikingly between the two pairings. In the H x I combination there was, at least initially, clear fusion of the two ramet matrices and an apparent intermingling of zooids (Fig. S2, Days 3, 4, 5; movie-S1 seconds ∼00:03–∼00:08). By ∼day 5, the H/I chimeric entity started separating into two physically distinct sections in association with retreat movement of at least one or both of the ramet types, with the result that by days 6–7 the fragmentation of the transitory H/I chimeric entity was clearly apparent, resembling an ‘indifference’ reaction (Rinkevich & Weissman, 1992) or the ‘non-fusion’ reaction in botryllid ascidians (Tanaka & Watanabe, 1973). From Days 8–12 ramet fragments were observed moving in a variety of directions, with the movement directions strongly suggesting an active separation process. The separating fragments were separated by a remnant zone largely devoid of zooids while the region corresponding to ramet ‘H’ was moving over the edge of the glass slide to the other side (Fig. S2, Movie S1).

In contrast to H/I combination, the H/L combination does not display any, even transitory, apparent intermingling of the zooid types in a single matrix (Fig. S3, Movie S2). Indeed after ∼2 days of close contact the genotype L ramets started displaying a coordinated movement away from the H ramet that resulted, within 6 days, in only remnant tunic at the previously contacting areas (Fig. S3 Days 3–6, Movie S2 seconds ∼00:01–∼00:09). As with the H x I combination the movements of the ‘H’ and ‘L’ ramets strongly suggest some sort of active separation/avoidance process (Fig. S3, Movie S2). For the H/I pairing it proved possible to discriminate the ‘H’ and ‘I’ genotypes using the microsatellite loci DVEX18 and DVEX19, both of which differed between the two genotypes by at least one allele (Table S2) and to obtain a semi-quantitative measure of the ratios of the two genotypes in the sections taken upon termination of the fusion experiment (Day 12, Fig. 2). Such genotyping results showed that five small zooid-rich regions that emerged from the transitory fused/chimeric entity colony (denoted regions a–e in Fig. 2) were largely composed of zooids of genotype ‘I’, while the others were largely of genotype ‘H’. Two of the apparently decaying regions that remained behind the highly mobile regions (denoted nos. d, e in Fig. 2, Fig. S2) were mainly of the ‘I’ genotype while a third remnant region (fragment f; Fig. 2, Fig. S2) was composed almost equally of both, the ‘H’ and ‘I’ genotypes.