Further morphological and molecular studies of driftwood hoppers (Crustacea: Amphipoda: Talitridae) from Mediterranean/north-east Atlantic coastlines

An unknown talitrid was discovered in 2011 in the Swale, UK, living in driftwood. Sequencing of multiple mitochondrial and nuclear genes (cytochrome oxidase I, 18S rDNA and 28S rDNA) confirms that the unknown taxon was close to Orchestia mediterranea A. Costa 1857. The driftwood in which it was found was of Douglas fir (Pseudotsuga menziesii), probably imported from North America. Relative growth methods allow us to reject the null hypothesis that the unknown taxon was a juvenile Orchestia mediterranea. This is because the unknown taxon initiates sexual maturation at an earlier moult number (neoteny) and is predicted to reach terminal moult stages at a much smaller size (dwarfism) than its closest relative O. mediterranea. Both molecular and morphological data confirm that the unknown taxon from the Swale, UK, is a new driftwood specialist taxon, distantly related to the driftwood specialist genus Macarorchestia. Additional data for body length and sex ratios are presented for Macarorchestia. In addition, relative growth methods have shown that the driftwood species Orchestia microphtalma Amanieu and Salvat 1964 is closer to the genus Macarorchestia rather than to the wrack generalist one, Orchestia. This evidence is in line with the molecular results that do not cluster morphologically defined species of Orchestia monophyletically.


Introduction
In an earlier publication ) an initial study of some of the known species of driftwood specialist talitrids found in the north-east Atlantic/ Mediterranean region (inclusive of the Azores, Madeiras and Canaries archipelagos) was presented. In that paper, morphological-based, relative growth methods in combination with modern molecular analyses were employed to determine: (1) a close morphological and genetic similarity between Macarorchestia remyi from Italy, M. roffensis from the UK, and M. martini from the island of Terceira, the Azores; (2) that Macarorchestia was a monophyletic genus; and (3) that a putative new driftwood specialist taxon from the Swale, UK, was closely related to Orchestia mediterranea, rather than to Macarorchestia.
In a continuation of this work using similar methods and analyses with the addition of nuclear genes and additional taxa, we tested the following specific null hypotheses: Hypothesis 1 that Orchestia microphtalma belonged to Orchestia rather than Macarorchestia. Hypothesis 2 that the putative new taxon from the Swale, UK, was a juvenile of Orchestia mediterranea A. Costa 1857.
The results of this study, as well as the earlier one , will be used in preparing formal taxonomic descriptions of two new taxa of driftwood talitrid specialists (Wildish 2014).

Archived material
The Natural History Museum in London, UK (NHM) made available 71 paratype individuals of Orchestia microphtalma for non-destructive morphological analysis, plus an additional five individuals for destructive sampling for molecular analyses. All specimens were collected from the type locality at Cap Ferret, near Arcachon, France (Amanieu and Salvat 1963). They were collected on 11 September 1967 by D.J. Wildish and deposited at the NHM with the following registration number: NHM 1967.10.6.1-75. Data matrices containing morphological measurements and counts against age and moult number in cultured juveniles of O. mediterranea from the Medway estuary, UK, originally made in 1968 and archived at the NHM, were also used in this study. The data for the juvenile phase of the life history was part of a comparative growth study of talitrids from the Medway estuary, UK (Wildish 1972). A sample (n = 30) of O. mediterranea from the Ria Formosa, Faro, Portugal collected by Ms Natalia Dias in 1998-1999 was also analysed here.
The sample from the Swale, Kent, UK ) consisted of 16 individuals (one immature male, two immature females and 13 juveniles) collected during the summer of 2011 by K.J. Wildish from a single driftwood log. A total of eight individuals, including all immature stages, were used in the relative growth study.

Relative growth
As far as possible we used a non-destructive method to make measurements and counts on each individual. Observational methods were essentially the same as used previously ). The following antennal flagellum counts/measurements in mm were made for each individual examined: • total body length from anterior cephalon to tip of telson (TBL) measured from photographs of each animal made with a Canon 990 digital camera attached to a binocular microscope. Using suitable calibration photographs TBL was measured with Image Pro Plus software, which allows curved bodies to be measured accurately. • second antenna article count (A2 FA) • A2 peduncle length, measured dorsally on the two most distal segments (A2 PL) • eye diameter (ED) • third peraeopod propodus length measured on the anterior side (P3 PL) • seventh peraeopod propodus length measured on the anterior side (P7 PL) • third pleopod exopod length (pl3 ExL) • third pleopod basis length (pl3 BL) • first uropod exopod length (Up1 ExL) All four population samples were treated separately and measurements/counts made on them as previously . Data matrices were prepared for each population sample on separate worksheets in Microsoft Excel (2003). The model I linear regression equation used was: y = ax + b, to analyse each population sample, with × = TBL and y = each of the counts/measurements in turn.

Sex ratio
The sex ratio was calculated as the percentage of males within discrete population samples for all immature and adult specimens as follows: '%♂ = total of immature and mature males × 100/total of immature and mature females plus intersexes'.
Each animal was examined with the aid of a binocular microscope to determine the presence of penes at the base of P7 (= male) and oostegites at bases of P2 to P5 (= female). Where secondary sexual characters were mixed we have designated them as intersex individuals.
Determining the provenance of driftwood occupied by the unknown taxon Because it might indicate the origin of the driftwood and thus help identify the place of origin of the unknown taxon, we sought to identify the driftwood tree species. This was achieved by boiling a piece of driftwood from the Swale, UK, in which the unknown taxon was found; it was then sectioned transversely, tangential longitudinally and radial longitudinally at 20 µm thickness. Slide preparations mounted in glycerol were examined with a light microscope.

Statistics of morphology based measures/counts
Mean total body lengths (TBL) was calculated for each population sampled and compared by analysis of variance.
Discrimination of relative growth difference between counts and measurements was made by model I regression analysis with TBL as the independent variable. The fitted regression lines were compared by Reeve's (1940) method, from the goodness of fit with single, separate and parallel lines. Mean square differences, or residual variance, from all fittings were compared with the F statistic: F = larger mean square/smaller mean square.
We used Statgraphics XV software, which uses Reeve's method, to compare population sample slopes and intercepts, and determined the probabilities of significant difference by reference to F-ratios. The same probabilities of significant difference of p-values as indicated for the sex ratio analyses were adopted in this analysis. Significance levels test the null hypothesis that population #1 ≠ #2 for slope coefficient. If the slopes were not significantly different (ND) a further test of the intercept coefficient was performed.

Additional specimens for molecular study
The dataset of Wildish et al. (2012) was re-analysed and additional talitrid species and molecular markers (see next paragraph) were included in this study. In total, we analysed nine genera and 19 species (inclusive of two unrelated taxa as outgroups). For the genus Orchestia, we analysed seven species from 18 locations (Table 1), including an attempt to retrieve DNA from the historical, formalin preserved specimens of O. microphtalma. A complete list of the populations, species and molecular markers used is shown in Table 1.

Molecular methods
Laboratory procedure DNA extractions and polymerase chain reaction (PCR) conditions for the historical formalin preserved specimens of O. microphtalma were performed as described in Wildish et al. (2012).
For contemporary samples DNA extractions, PCR reactions and sequencing were carried out according to Ketmaier et al. (2008), Pavesi et al. (2011Pavesi et al. ( , 2013 for the COI gene; primers and conditions detailed in Spears et al. (2005) and Jarman et al. (2000) were used for the 18S gene and two fragments of the 28S gene (primer pairs 28Sdd/ 28Sff and 28Sv/28Sx), respectively.
Sequencing was outsourced to LGC Genomics (http://www.lgcgenomics.com) and performed in both directions for each gene fragment. Sequences were edited with SEQUENCER v. 4.1 (Gene Code Corporation, Ann Arbor, MI, USA). Aligning the COI sequences was straightforward and was done in SEQUENCER. For the 18S and 28S fragments, we used CLUSTALX (Thompson et al. 1997) using the default settings (pair-wise alignment parameters: slow-accurate pairwise alignment method, gap opening penalty = 15.00, gap extension penalty = 6.66, IUB DNA weight matrix; and multiple alignment parameters: gap opening penalty = 15.00, gap    extension penalty = 6.66, delay divergent sequences = 30%, DNA transition weight = 0.50), followed by limited manual editing to improve inferences of positional homology. No large gaps or poorly alignable regions were present in the 18S and 28S alignments.

Statistical analyses
For the COI dataset, Kimura two-parameter (K2P COI ) genetic distances were calculated with PAUP * Portable version 4.0b10 (Swofford 2002). In order to assess levels of genetic divergence at the 'among' and 'within' genus levels, minimum, maximum and mean K2P distances were calculated within Macarorchestia and Orchestia spp. For comparative purposes, K2P all_genes distances were also calculated on the 18S and 28S gene partitions with the two 28S gene fragments combined together. The automatic barcode gap detection (ABGD) method (Puillandre, Lambert et al. 2012), which implements an automatic procedure of barcode gap discovery, was used to define species assignment based on COI haplotype groups within the genus Orchestia. The analysis was performed online (http://wwwabi.snv.jussieu.fr/public/abgd/abgdweb. html) with both Kimura 2P (K89) and Jukes-Cantor (JC69) distances and with default parameters. Given the limited number of sequences per species at our disposal, we additionally tested model combinations with P (prior limit to intraspecific diversity) ranging between 0.001 and 0.1, and X (proxy for minimum gap width) from 1.5 to 10. This was done to both evaluate the effect on our dataset, and to avoid the capture of smaller gaps as a result of weak sampling (Castelin et al. 2012;Puillandre, Modica, et al. 2012;Marin et al. 2013).
Phylogenetic searches were conducted on each gene separately, on COI plus the 28S fragment obtained with the 28Sv/x primer pair and all genes combined together. Phylogenetic trees were rooted using the homologous sequences of the hyperiid amphipod Themisto gaudichaudii retrieved from GenBank as outgroup (accession nos. HM053513, AY743940, AF244091 and AF244092 for the COI, 18S and 28S fragments, respectively). We used JMODELTEST 2 (Darriba et al. 2012) to determine the best model of sequence evolution for each data partition analysed. JMODELTEST 2 returned the GTR + G model for the 18S and 28Sdd/ff partitions and the GTR + G + I for all the others. These models were then implemented in the Bayesian searches carried out in MrBAYES (Ronquist et al. 2012). We allowed sitespecific rate variation partitioned by gene and, for COI, by codon position in the Bayesian analysis. MrBAYES was run for 2,000,000 generations with a sampling frequency of 100 generations. We ran one cold and three heated Markov chains and two independent runs. To establish if the Markov chains had become stationary, we plotted the likelihood scores of the sampled trees against generation time. Trees generated before stationarity were discarded as burn-in (first 10% of the sampled trees) and posterior probability values for each node were calculated on the basis of the remaining 90% of sampled trees. The output of each independent Bayesian run was visualized using TRACER v1.4 (Drummond and Rambaut 2007) to check whether a reasonable effective sampling size (ESS, number of independent samples of the posterior distribution) had been reached during each search. Convergence of the Markov chains was further evaluated with AWTY (Wilgenbusch et al. 2004). TRACER and AWTY both consistently suggested that the Markov chains in both independent runs have converged and reached stationarity after the 10% burn-in. We tentatively dated the main splits in the COI phylogeny using K2P COI values and the mutation rate adopted for O. montagui in Pavesi et al. (2013).

Relative growth
Hypothesis 1 A scatterplot comparing O. microphtalma and O. mediterranea from Faro, Portugal for pl3 ExL on TBL is shown in Figure 1A and suggests that the regression lines are far apart. All dependant variables tested are shown in Table 2. Because scatterplots of ED, A2 FA and A2 PL for O. microphtalma and M. remyi showed an inflection between juvenile and immature data, the juveniles were excluded in calculating the regression constants (indicated by brackets in Table 2). Of the eight dependent variables of Table 2 significant differences in slope values between O. microphtalma and O. mediterranea were found in: ED*, A2 FA***, A2 PL****, and Pl3 ExL****. For bivariates lacking significant difference in slope (Table 2) we tested the intercept differences with the following results: P3 PL (ND), P7 PL**, Up1 ExL**** and Pl3 BL****. The preponderance of allometric slope differences (four of eight) between the two taxa allow us to reject the null hypothesis that O. microphtalma belongs to Orchestia.
We examined an alternative hypothesis that O. microphtalma belonged to Macarorchestia by comparing the allometric slope differences between O. microphtalma and M. remyi from Principina a Mare, Italy (data from Wildish et al. 2012). We found no slope coefficient differences ( Figure 1B) between the two species, consistent with O. microphtalma belonging to Macarorchestia. Only intercept differences were present between the two sampled populations as follows: A2 Fa*** ( Figure 1B), A2 PL**, Pl3 ExL** and Up1 ExL**** (not shown).

Hypothesis 2
A scatterplot comparing the unknown taxon and O. mediterranea juveniles from the Medway estuary, UK, suggested that the fitted regression lines for A2 PL on TBL ( Figure 2A) and ED on TBL ( Figure 2B) was linear and parallel for both taxa, with no evidence of an inflection (although this does occur in O. mediterranea between the juvenile stage and at the onset of sexual maturity). There appears to be no slope difference, but a significant intercept difference in the regression lines compared in Figure 2. Other points to note from Figure 2 are the few data points for the unknown taxon (maximum of eight) which renders their use in relative growth as questionable, although R 2 = 0.89 in Figure 2A and R 2 = 0.88 in Figure 2B for this dataset. The three largest individuals for the unknown taxon shown in Figure 2 consist of immature male and two immature females, as recognized independently by observing the presence of penes and oostegite buds at the base of P7 and P2 to P5, respectively. The regression data are shown in Table 3 and confirm the impression of no slope, but large intercept difference between the unknown taxon and juvenile O. mediterranea. Thus for all six measures/counts (Pl3 measures are excluded because only two measurements were available for the unknown taxon) none of the slope coefficients (except one), but all of the intercept differences were significantly different. Thus: ED***, A2 FA****, A2 Pl****, P7 PL****, and Up1 ExL****. The one exception was for P7 PL where the slope coefficient was significantly different at the 95% level (Table 3). In general the regression lines for this pair of populations were similar in slope but far apart positionally, suggestive of a close relationship, but differing in relative growth characteristics which include the initial and final size reached by each population. We examined one possible alternative hypothesis, that the unknown taxon belonged to Macarorchestia, by comparing the slope differences between it and M. roffensis (data from Wildish et al. 2012). For the latter species an inflection was present for ED, A2 FA and A2 PL at the onset of sexual maturity and so the juvenile data were excluded in calculating the regression coefficients for these characters. Of the six available dependent variables, significant allometric slope differences were found in one of them (ED****), and intercept differences in A2 FA***. Of two measurements we were able to make of the Pl3 characters in an immature female and immature male of the unknown taxon, the Pl3 ExL was 0.500 and 0.600 mm. For M. roffensis at the same TBL as these specimens the Pl3 ExL as predicted by the regression line would be 0.250 and 0.300 mm respectively. These large differences suggest that the unknown taxon was not directly related to Macarorchestia.

Total body length (TBL)
If we combined body size data from our earlier study  with the additional taxa of this study, there was a considerable range in body size among species, as shown in Table 4. Whereas the smaller species, such as M. roffensis, had females > males, the larger species, such as M. remyi, had males > females, as measured by TBL. Insufficient population sample numbers were available for M. martini from the Azores and M. remyi from Lesina, Italy to decide if they too followed this trend.

Sex ratio
Although insufficient sample numbers were available to determine sex ratios for M. martini from the Azores and M. remyi from Lesina, Italy, the general trend was for the sex ratio (Table 5) to be positively related to size, such that the smaller the final body size, the more biased was the sex ratio to an excess of females.

Driftwood provenance
Growth rings in the Swale driftwood sample were narrow with 18 complete growth rings observable, which were 586-970 µm in width (mean = 762 µm). Axial resin canals were present mainly in the latewood. In tangential section some of the rays had radial canals. In both tangential and radial view the tracheids possessed shallow helical thickenings. This combination of axial and radial resin canals plus helical thickening is diagnostic of Pseudotsuga sp. (Douglas fir). Of the two known North American species of Pseudotsuga sp (Farjon 2001) the likely one is P. menziesii as it is the only one used in commercial lumbering. Pseudotsuga menziesii was introduced into the UK in 1827 (Edlin 1970) and today is widely grown, e.g. by the UK Forestry Commission. Since the early twentieth century large quantities of Douglas fir have been imported to the UK from North America. Thus the provenance of the Swale driftwood is uncertain (imported or locally grown Douglas fir?). The limited evidence suggests that the driftwood originated as imported Douglas fir, because growth rings indicate slow growth. Typically, North American growth conditions result in slow growth, versus faster growth in the UK. The large dimensions of the sawn driftwood sample found (~3 × 0.5 × 0.25m) in the Swale driftline suggests that its original human use was a wharf marine piling in the Thames or Medway estuaries.

Hypothesis 1
In spite of the considerable effort and use of the specific protocols adopted successfully for the recovery of short fragments of the historical, formalin preserved specimens of M. roffensis and M. martini in Wildish et al. (2012), it was not possible to obtain any PCR product for O. microphtalma. Formalin preservation for more than 40 years had probably caused high degradation of the DNA of these specimens. Similarly, we were not able to amplify fragments of the 18S and 28S genes for M. roffensis and M. martini. All markers worked well for most species if contemporary samples were available, with only a few exceptions (Table 1). All sequences obtained

Hypothesis 2
Pertinent genetic distances based on K2P of the COI gene are shown in Table 6 along with those based on the two nuclear genes. For the unknown taxon from the Swale, UK, and an O. mediterranea population from the euhaline part of the Medway estuary, UK, K2P COI = 2%, which shows the closeness between the two species. 18S and 28S genes showed no differences between the unknown taxon and O. mediterranea (Table 6). Our results are consistent with the null hypothesis that the unknown taxon is an intraspecific form within O. mediterranea.

Other K2P differences
Other differences of interest (Table 6) show that between O. mediterranea and O. aestuarensis K2P COI = 11%, which is consistent with this pair being closely related sister species, although able to interbreed both in the laboratory and the field (Wildish 1970). The possibility that a mesohaline population of O. gammarellus from the Medway estuary was a new species was discounted because of an insufficient K2P COI difference of 4.3%. The genetic structuring    Lowry and Fanini (2013). Villacorta et al. (2008) found both O. gammarellus and O. guancha close to O. gomeri from the island of La Gomera in the Canaries, a species that unfortunately we were not able to include in the study. The genetic difference between Macarochestia remyi from Italy and Palmorchestia epigaea from the Canaries is K2P COI = 25%. Both are basal to the large clade which includes many Orchestia spp. (Figure 3; all genes). Multiple comparisons within the genus Orchestia returned K2P COI values comparable to that between the genera Macarorchestia and Palmorchestia, supporting the lack of monophyly for the genus Orchestia as apparent in the phylogenetic trees (Figures 3 and 4). Divergence times of supported cladogenetic events in the COI phylogeny ranges from about 0.83 million years (myr) for the split involving the putative new taxon and O. mediterranea to 9.2 myr between O. gammarellus and O. guancha (Table 6). We emphasize here that these time estimates should be considered with extreme caution given that (i) they are based on the assumption of a linear relationship between genetic divergence and evolutionary times; and (ii) our phylogeny is not exhaustive taxon-wise and this prevented us adopting more complex approaches (i.e. Bayesian approaches) that, being tree-based, require taxon-sampling denser than ours to avoid inflating divergence time estimates due to the presence of large gaps (unsampled taxa) in the branching pattern.
Talitrid phylogeny of the north-east Atlantic/Mediterranean region Included in this region are the north-east Atlantic islands of the Azores, Madeira and Canary archipelagos. All of the known talitrid species found there by conventional, morphological methods are shown in Table 7. Species are categorized by the primary habitat (where this is known) that they normally occur in. The list in Table 7 includes nine genera and 26 morphologically defined species, for which we provide molecular data for seven of the nine genera, but only 13 of the 26 species. This limitation should be borne in mind when considering the phylogenetic analysis (Figures 3 and 4).
Based on the COI alone ( Figure 4) the evidence suggests that there are three groups, consisting of C. cavimana only and two subgroups of O. montagui + O. stephenseni and the rest. The latter consists of two subgroups made up of Platorchestia + Orchestia and Deshayesorchestia + Talitrus and a further subgroup consisting of Macaror chestia + Africorchestia + Palmorchestia. All genes combined yielded similar relationships ( Figure 3) with the main discrepancies being the placement of A. spinifera and P. epigea; the two species cluster together in the COI phylogeny and separately in the tree based on all genes. Both phylogenies consistently rejected the genus Orchestia as a monophyletic.

Growth data
Four of the eight dependent variables in the regression analysis shown in Table 2 are significantly different in slope between an O. mediterranea population from Portugal and the O. microphtalma one from France. These variables included ED and Pl3 Exl, which are reduced in O. microphtalma. Both these features are included as diagnostic for Macarorchestia by Stock (1989) and allow us to reject the null hypothesis that O. microphtalma belongs to the genus Orchestia, where both characters are not reduced. Our results suggest that O. microphtalma is more closely related to Macarorchestia and we examined this further by comparing O. microphtalma and a population of M. remyi from Principina a Mare, Italy. We found that none of the characters had a significant slope difference between O. microphtalma and M. remyi, although there were four intercept significant differences which involved extensive overlapping, rendering them of little use in distinguishing the two species (Figure 1 B). The intercept differences are consistent with the hypothesis that O. remyi is neotenously derived from O. microphtalma and that both species belong to the genus Macarorchestia.
The finding that the Swale unknown taxon included immatures at TBL sizes that in O. mediterranea would still be in the juvenile stage, supports a rejection of the null hypothesis that the unknown species is a juvenile O. mediterranea. The appearance of secondary sexual characters in the unknown taxon at a 'juvenile' TBL size for O. mediterranea suggests a neotenous origin for it. The explanation as to why in Figure 2 the parallel regression lines for the unknown species occurs above that of juvenile O. mediterranea is because the former first moult juvenile is hatched from the ova at a smaller TBL (~2.25 mm) size than the latter (3.7-4.2 mm). This is in line with the observed intercept differences in most of our data between the unknown Figure 4. Bayesian estimate of the evolutionary relationships among the talitrid species included in the study based on the COI partition. Numbers at nodes are posterior probabilities; divergence time (myr) ± standard deviation (SD) is reported at each node. Taxa are labelled following the codes in Table 1. Species delimitations obtained with the ABGD method are shown next to the tree. taxon and O. mediterranea juveniles. One would normally only accept slope (allometric), and not intercept (isometric), differences between pairs of taxa as evidence of taxonomic difference. In the case between the unknown taxon and juvenile O. mediterranea from the Medway estuary the intercept differences occur for most of the dependent variables tested and demonstrate isometry, which supports the hypothesis of neotenous origin for the unknown species. On the other hand, the slope coefficient similarities also found support the view that the unknown taxon is closely related to O. mediterranea, perhaps only differing in that it is a neotenic form of it.
Having accepted the alternative hypothesis that the unknown taxon was a new form of driftwood hopper we tested the possibility that it belonged taxonomically to Macarorchestia by comparing it with M. roffensis, also from the Medway estuary. We found two slope differences (ED and PL3 ExL), which according to Stock's (1989) diagnosis of Macarorhestia discount it as belonging to that genus. The large, unreduced eyes and pleopod rami of the unknown taxon suggests that it did not belong to Macarorchestia.
Considerations based on relative growth measures, as above, hide the typically slow growth rates which are shown by all driftwood talitrids so far examined. Limited talitrid growth data are available (Wildish 1972) and we have used this to help construct hypothetical growth curves based on archived juvenile growth data available at the NHM for O. mediterranea and M. roffensis, from the Medway estuary, UK. Maximum adult TBLs for the former are 21 mm and 8 mm for the latter and these have been used to construct the growth curves of Figure 5. Shown for M. roffensis is a 12-month and for O. mediterranea an 18-month lifespan. The linear fitting for the unknown taxon ( Figure 5A) is based on the assumptions of a first moult TBL = 2.5 mm and similar growth slope coefficient as in O. mediterranea. For a 12-month lifespan the maximum TBL reached by the unknown taxon would be 12 mm and for an 18-month lifespan it would be~16 mm. A lognormal fitting for the same data is shown in Figure 5B.
The findings presented here suggest that neoteny, and the consequent small body size finally achieved (dwarfism), may be fundamentally important in the origin of all driftwood talitrids. If this is so the size, as determined by the TBL of driftwood hoppers, is important and so the accurate TBL of the data generated here and in Wildish et al. (2012) is included in Table 4. The range found is from~14 mm to as low as 6 mm for M. martini from the Azores. A concomitant effect of reductions in body size, due to slower growth, is that the number of ova per brood is reduced (Wildish 1988). One way of offsetting small body size and consequent reductions in number of ova/brood is to increase the proportion of functional females in the brood. This is known to occur in talitrids commonly as a result of co-evolution of an 'endoparasite' and its talitrid host (Ginsburger-Vogel 1991) and the adaptive advantages are outlined (Wildish 1971). In our own data (Table 4) the tendency to female bias in the immature and adult populations does seem to increase as TBL decreases. However, our data are incomplete, particularly for M. remyi from Lesina, Italy and M. martini from the Azores and further sampling is required to firmly establish this point.
Our finding that the unknown driftwood talitrid found in the UK in 2011 was present in driftwood made up of Douglas fir confirms that driftwood talitrids can utilize both coniferous (Gymnospermae) and deciduous (Angiospermae) driftwood. This is consistent with earlier observations in the UK, of M. roffensis found in deciduous driftwood by Wildish (unpublished) and in the Mediterranean Sea region of other species of Macarorchestia in both deciduous (Alnus sp., Betula sp.) and coniferous (Pinus sp.) driftwood by Pavesi (2009). These findings are important because they establish that suitable driftwood (gymnosperms originated 300 mya and angiosperms 140 mya Atlantic Geoscience Society 2001) was available for driftwood colonization by talitrids prior to the supposed origin of talitrids from hyalidlike ancestors some 110 mya (Bousfield 1984).

Molecular data
The K2P COI distances found in the current study confirmed the findings of Wildish et al. (2012) and are in line with the available literature on Crustacea (Adamowicz  Costa et al. 2009). The driftwood hoppers (Macarorchestia spp.) are distinct from the non-driftwood talitrids, and the new taxon appears to be closer to O. mediterranea than to M. remyi.
Populations' ABGD was congruent with the current taxonomy for almost all species. In each analysis, when the prior maximal distance (P max ) is small, the partition tends to find a higher number of groups, probably due to intra-rather than inter-specific genetic divergence. Indeed, many talitrid species show a high level of genetic structuring (see Ketmaier and Pavesi 2013 for a review on the issue). A threshold of 4% was the best to describe our data. The putative new taxon clustered, in each dataset and each parameter set-up, always together with O. mediterranea. For P max >4% O. aestuarensis (H) would be grouped with O. mediterranea, and all the O. gammarellus populations would cluster together, while for P max <4% the analysis would assign population B and all the other O. gammarellus populations to two distinct groups. The occurrence of probable divergent genetic lineages within amphipod species has sometimes been reported (Witt et al. 2006;Cabezas et al. 2013), including in some Mediterranean talitrids .
Although our phylogenetic tree includes only one half of the known species found in the Mediterranean/north-east Atlantic region we believe that it provides a preliminary view of what a more complete phylogeny might look like. The following are the salient findings: -The divergence time for the splitting between the unknown taxon and O. mediterranea could be very recent (0.83 million years). -O. guancha is phylogenetically close to O. gammarellus, supporting the earlier findings of Villacorta et al. (2008); this is circumstantial evidence in support of the hypothesis that the former species arose from a wrack generalist (Wildish 2012). -Palmorchestia epigea is either close to A. spinifera (COI only) and both species form a clade sister to M. remyi; or close to P. epigea (all genes) and is basal to a large clade including various Orchestia species + the putative new taxon + M. remyi. This is supportive of the proposal of Lowry and Fanini (2013) that the former Orchestia cavimana belongs to a new genus Cryptorchestia, but not that the northeast Atlantic island terrestrial talitrids referable to Orchestia belong to the same new genus, represented in our sampling by the former O. guancha. The latter species is phylogenetically close to O. gammarellus.

Conclusions
The main findings of this study are that on morphological grounds O. microphtalma belongs to the genus Macarorchestia. Regarding the putative new taxon, morphological results indicate that the unknown taxon is not a juvenile O. mediterranea. Molecular analyses confirms that it does not belong to the well-established driftwood genus Macarorchestia and is very close genetically to the wrack generalist O. mediterranea, having diverged from it c.0.83 myr. Based on allometric growth observations the unknown taxon is a neotenic dwarf form of O. mediterranea, differing only in that it is slower growing, begins sexualization at an earlier moult number and lacks dorsal pigment patterns. Microscopic examination of the driftwood sample in which the unknown taxon was found established that it was from a Douglas fir tree.
The molecular results confirm the close relationship between the unknown taxon and O. mediterranea, with two additional nuclear genes supporting this. During this study approximately one half of the morphologically defined species present in the north-eastern Atlantic/Mediterranean region were analysed (for COI and 18S and 28S rDNA). The resulting phylogenetic tree provides a preliminary glimpse of the phylogenetic relationships of species from seven, of a total of nine, regional talitrid genera.