A morphological analysis of 16 O. gammarellus populations from Canadian, Icelandic and European locations was conducted by Henzler and Ingólfsson (2008). For the non-sexually dimorphic characters chosen, all populations differed significantly in one or more characters. Including within larger geographic areas (e.g. Iceland), with evidence that warm spring populations within Iceland differed the most. This is evidence that environmental variations have influenced morphometry. Henzler and Ingólfsson (2008) consider that the morphometric results were consistent with a recent (in geological terms) post-glacial, European origin.

Myers and Lowry (2020) proposed a new name for the Canadian and Icelandic populations of O. gammarellus, based on four minor morphological differences. The distinguishing features of O. forchuensis from O. gammarellus (in brackets) listed by Myers and Lowry (2020) are:

• In adult male, p7 merus parallel-sided and 2× as broad proximally (versus merus triangular and 3× as broad proximally)
• In adult male, p7 carpus much longer than broad (versus almost as broad as long)
• In adult female, the basis of p7 is sub-ovoid (versus sub-quadrangular)
• In both sexes(?), p5 has a weakly developed posterodistal lobe (versus well-developed posterodistal lobe).

These characters apply to adults and cannot distinguish immatures and juvenile members of the two populations.

Morphological plasticity has been recognised within talitrids, notably those species living in a wrack primary ecotope, such as Platorchestia platensis. Platorchestia platensis can acclimate to a secondary ecotope: driftwood (Wildish and Robinson 2016) which forces a change in diet with profound effects on morphology, growth, physiology and reproduction (Wildish and Robinson 2018). Such changes can lead to false conclusions during taxonomic decision-making which is solely based on morphological characters. This can result in errors of nomenclature, as occurred during a description of a new genus and species, which was ultimately recognised to be a driftwood morphotype of O. mediterranea (Wildish 2014, 2018). Morphological plasticity needs to be accounted for during taxonomic decision-making, as there is growing evidence that morphological plasticity is a general property of wrack generalist talitrids. The driftwood morphotype of O. mediterranea also experienced a reduced growth pattern as in driftwood-acclimated P. platensis. This could also be the case for O. gammarellus which was found in driftwood, as a concomitant of the xylophagous specialist, Macarorchestia roffensis (Wildish, 1969) and described then as juveniles (Wildish 1982). Perhaps the juveniles were actually driftwood morphotypes of O. gammarellus? We believe this is worth following up experimentally (Suggestion #1).

Apart from the three amphi-Atlantic talitrids, inclusive of O. gammarellus, the talitrid biodiversity of the east and west coasts of the North Atlantic Ocean lack species in common (table 1 in Wildish and Chang (2017)). Thus, for the sand-burrowing ecotype there are five species (in one genus) in the northwest and six species (in five different genera) in the north-eastern Atlantic. For the wrack generalist ecotype, there are five species in the northwest versus five (four if montagui is re-assigned to new genus, see above) in the northeast Atlantic which all belong to the genus Orchestia. Palustral specialists appear to be limited to the northwest Atlantic, with three species identified morphologically, although genetic methods suggest up to an additional seven species or subspecies (Radulovici 2012). Driftwood specialists appear to be limited to the northeast Atlantic with five species in a single genus.

Apart from the three amphi-Atlantic species, the different talitrid fauna on both coasts we consider as evidence that long-term separation occurred after isolation of the two coasts by continental drift in the upper Cretacious/Eocene (Bousfield 1984). The single species of Orchestia on the northwest Atlantic coast (versus five, or four, Orchestia species on the northeast Atlantic coast) does not support a ‘proto-gammarellus’ presence in the Eocene here either. We would expect more Orchestia species diversity, in line with that of the northeast Atlantic Orchestia, if ‘proto-gammarellus’ had been present since the Eocene.

In post-glacial times, colonising O. gammarellus in Iceland would reach an open supralittoral wrack habitat, without competitors, whereas in Newfoundland and Maritime Canada, they would compete with P. platensis, which was spreading northwards following glacial retreat. Competition between these two species (Persson 2001) is the likely explanation for the sporadic distribution of O. gammarellus and its replacement by P. platensis at several isolated beaches during the last 50 years (Wildish, unpublished data).

Sceptics of dispersal have condemned it as immeasurable and unimportant, but evidence of a resurgence of interest in natural dispersal was identified by Nathan (2001). Both Henzler and Ingólfsson (2008) and Myers and Lowry (2020) appear to be sceptics of natural dispersal in O. gammarellus. Ingólfsson (1995, 2006) conducted intensive studies of floating wrack off the coast of Iceland without finding O. gammarellus within it. Yet, cursory studies (five sampling excursions) in the Medway Estuary, England (Wildish 1970) found one adult individual of O. gammarellus in floating wrack in one sample on one sampling excursion. Clearly, the Estuary is the right place to seek evidence of passive dispersal and more intensive studies are needed in estuaries (Suggestion #2). Convincing evidence of passive dispersal at sea was provided by studying the stomach contents in several species of inshore fish which contained supralittoral talitrids (species undetermined, but likely to be O. gammarellus) off the Scottish coast (Gibson and Robb 1996) and in a macrotidal saltwater marsh inlet in France (Laffaille et al. 1999) where the prey was predominantly O. gammarellus. These findings show that the talitrids were predated by fish whilst they were present in the sea.

The range of O. gammarellus, from the north-eastern Atlantic Islands (Azores, Madeira, Canaries) to the Lofoten Islands, demonstrates that this species has dispersed widely within the North Atlantic. In a largely theoretical study of dispersal, Wildish (2012) proposed a lineal ‘island’ theory to account for the passive dispersal in floating wrack or floating driftwood. An important implication for all wrack generalist ecotypes with the ability to acclimate to a woody diet is that, when they do so in natural conditions, they may become subjects of long-range dispersal in floating driftwood. The theory helped to determine the dispersal potential of the three amphi-Atlantic talitrids (inclusive of O. gammarellus) with the user-friendly predictive model of present-day surface currents in the North Atlantic (van Sebille 2014). Four starting points for dispersal on the northeast Atlantic coast (Lands End, England to the Canary Islands) were modelled (Wildish and Chang 2017) and the results show that O. gammarellus could not be passively dispersed in either wrack or driftwood in the east to west direction. For the reverse direction, from west to east, a talitrid capable of living within driftwood could survive long enough to make the trip as a passenger within a driftwood raft, because of the fast Gulf Stream currents streaming in the direction of Europe. Unfortunately, we did not model starting points in Scotland or from the Norwegian Atlantic coasts and so are unable to determine if passive dispersal from there to Iceland was feasible given contemporary oceanographic conditions. The Norwegian current (Anon 1979) flows in a north-easterly direction close to the Atlantic Norwegian coast carrying potential dispersers in a counter-clockwise current around the Norwegian Sea and eventually to Iceland.

Within the last ten thousand years, during which humans have travelled to most parts of the planet and established trading routes by marine transport, they have often taken with them hitchhikers which colonise the new habitats. This process of synanthropy has been identified amongst talitrids, particularly those of terrestrial origin which come in soil or attached to imported tropical plants (e.g. Shoemaker (1935); Gregory (2016)). Henzler and Ingólfsson (2008) mentioned occasional accidental transport of O. gammarellus by ships associated with rocks, sand or other nearshore debris. An obvious candidate for transport would be in the wooden ships used to transport the Vikings from Norway to Iceland beginning in 874 AD and to Greenland in 982 (Anon 2022). The ships traversing the northern Atlantic used by the colonisers were stout ‘knorr’ or cargo ships, some 16–22 m in length and broad in beam and capable of carrying up to 25 tonnes of cargo, including live horses and cattle (Anon 2022). It is not clear whether they used ballast (stones, gravel or sand) in the keel, but if so, it would have been obtained by beaching the boat and loading directly from the beach, with the possibility that wrack and talitrids would have been inadvertently included.

The alternative mechanism of natural dispersal considered by Henzler and Ingólfsson (2008) to Iceland involved floating wrack (= floating seaweed). However, extensive field studies examining contemporary samples of this material showed that O. gammarellus was not present in coastal seas off Iceland (Ingólfsson 2006). This caused Henzler and Ingólfsson (2008) to discard natural dispersal in favour of synanthropic dispersal to Iceland.

Genetically separate lineages of multicellular organisms accumulate increasing amounts of DNA sequences which diverge during the passage of time. As a result, it is possible to estimate the geological time at which divergence first occurred using DNA sequence data. The mitochondrial cytochrome oxidase I (COI) gene is widely used for species identification and can also be used to estimate divergence times of two genetically separate lineages. Loeza-Quintana et al. (2019) used COI data from multiple pairs of arthropod species and estimated the rate of divergence at 5 to 5.2% per million years (Kimura 2-parameter corrected distance); they review earlier studies with estimates for arthropods ranging from 1.4 to 6.6% per million years.

Henzler and Ingólfsson (2008) sequenced a 532-bp fragment of COI in O. gammarellus from north-western Europe, Iceland and Canada. All three hypotheses for the origin of O. gammarellus (synanthropic dispersal, post-glacial natural dispersal and Eocene vicariance) make different predictions about the amount of DNA divergence to be expected. Considering an estimate of 5% divergence per million years, if the Icelandic/Canadian populations of O. gammarellus were completely isolated since the first Viking ships sailed to Iceland ~ 1100 years ago, the expected divergence would be only 0.007%, which would likely be undetectable with a 532-bp fragment. Postglacial dispersal 20,000 years ago would produce an estimated divergence of 0.1% or one site changed in 1000, so the 532-bp COI fragment might have some haplotypes that were found in Canada or Iceland, but were not present in Europe. However, if Canadian and Icelandic populations of O. gammarellus had been genetically isolated since the Eocene, their COI sequences would be saturated with divergence at every site that was not constrained by selection.

Henzler and Ingólfsson (2008) found a single, major haplotype in most of their European and Icelandic samples, which was also present in a minority of Canadian O. gammarellus. All of the other haplotypes in their sample differed from the most common one by at most two nucleotides. This is inconsistent with the genetic separation of north-eastern and north-western Atlantic populations of O. gammarellus since the Eocene. Henzler and Ingólfsson (2008) also estimated the divergence time between Icelandic/Canadian versus European populations to be about 14,000 to 18,000 years, which fits better with the post-glacial natural dispersal hypothesis rather than with synanthropic Viking colonisation, which occurred at the earliest 1,148 years BP.

Henzler and Ingólfsson (2008) showed that Iceland had a common minority haplotype that was not found in their European or Canadian samples and the most common haplotype in Canada was not found in any other sample. The presence of a few local haplotypes is consistent with an early postglacial dispersal. However, Henzler and Ingólfsson (2008) also found considerable divergence, including some local haplotypes, amongst sampling sites within Europe. One other possible way of explaining this result is that the Icelandic/Canadian populations originated by recent synanthropic dispersal from a source population that Henzler and Ingólfsson (2008) did not sample.

Myers and Lowry (2020) described O. gammarellus populations in Iceland and Atlantic Canada as a new species, O. forchuensis, based solely on morphological differences and did not provide an adequate discussion of the COI sequence data of Henzler and Ingólfsson (2008).

Other published molecular data of pairs of closely related species generally differ in their COI sequences by several percent; Hebert et al. (2003) found that 99.9% of pairs of congeneric crustacean species differed by 1–20%. Costa et al. (2009) considered 16 species of Gammarus and found that the closest pair of species, G. insensibilis and G. aequicauda, differed by 5.2%. The closest pair of Orchestia species in Pavesi et al. (2015), O. mediterranea and O. aestuarensis differed by 11%. Davalos et al. (2018) sequenced COI and histone H3 in European populations of O. gammarellus, O. montagui, O. stephenseni, O. mediterranea and O. aestuarensis; the most similar pair of species, O. mediterranea and O. aestuarensis, differed by over 5%. The reason that the Kimura 2-parameter corrected distance was different in the latter species pair was that Pavesi et al. used a longer fragment of COI (497 bp) than Davalos et al. (363 bp) and because the distances in the latter combine COI data with the more conservative histone H3 gene.

We used MEGA version 11 (Tamura et al. 2021) and the COI data of Henzler and Ingólfsson (2008) to calculate the average Kimura 2-parameter corrected distance between “O. forchuensis” (Canada plus Iceland) and O. gammarellus from Europe. The distance was only 0.1%, with many individuals in the two groups sharing identical haplotypes. This is much less than the distances between closely-related pairs of well-established other Orchestia species. The COI data are, therefore, in conflict if the Icelandic/Canadian populations are considered as a separate species.

Molecular genetic methods applied to amphipod taxonomy are an important addition to morphology, rather than a replacement of it, resulting in a more accurate and useful combined discipline. The particular molecular genetic methods needed to achieve this include:

• Determining genetic distances between populations of closely-related species.
• Determining the geological age of separation between isolated populations of the same or closely-related species.
• Providing a convergence-free talitrid phylogeny which is central to an accurate phylogenetic tree, as the proper basis for defining the higher levels of taxonomy within Talitroidea.

The COI phylogeny of Henzler and Ingólfsson (2008) suggested a population expansion from a single, bottlenecked source population which arrived in Iceland at some point during the last post-glacial period.