A Review of Gammaridae (Crustacea: Amphipoda): the Family Extent, Its Evolutionary History, and Taxonomic Redeﬁnition of Genera

Home , Acanthogammaridae, Brandtia, Gammaracanthus, Gammaridae, Gammaridea, Gammaroidea

Zoological Journal of the Linnean Society, 2016, 176, 323–348.

A review of Gammaridae (Crustacea: Amphipoda): the family extent, its evolutionary history, and taxonomic redeﬁnition of genera

ZHONGE HOU1 and BORIS SKET2*

1Key Laboratory of Zoological Systematics and Evolution, Institute of Zoology, Chinese Academy of Sciences, Beijing 100101, China 2Department of Biology, Biotechnical Faculty, University of Ljubljana, PO Box 2995, Ljubljana SI-1001, Slovenia

Received 30 March 2015; revised 22 June 2015; accepted for publication 24 June 2015

By molecular analysis of a high number of gammarids, including 29 out-group genera, we could assure the monophyly of Gammaridae. To avoid the paraphyly of the family, we propose the omission of Pontogammaridae, Typhlogammaridae, and all Baikalian families. Similarly, the genera Fontogammarus, Sinogammarus, Lagunogammarus, Pephredo, Neogammarus, and Laurogammarus may be cancelled. But, tens of Baikal genera, nested within Gammarus, are so diverse that they must be retained, although rendering Gammarus paraphyletic. Besides we propose the polyphyletic Echinogammarus–Chaetogammarus group to be divided into monophyletic genera Echinogammarus s. str., Homoeogammarus, Parhomoeogammarus, Marinogammarus, Relictogammarus gen. nov., Chaetogammarus, and Trichogammarus gen. nov. These solutions made it possible to complete the ﬁrst analysis of the family evolution in light of its phylogeny. Perimarine clades are mainly basally split clades, whereas in some ancient lakes extremely rich endemic faunas had developed polyphyletically. The troglobiotic Typhlogammarus group from Dinarides and Caucasus formed a monophylum, whereas the troglobiotic assemblage of Gammarus species is highly polyphyletic. Reduction of the uropod III endopodite, which classically distinguishes between the genera Gammarus and Echinogammarus, appeared to be highly polyphyletic. Protective dorsal pleonal projections occur scattered across the family and beyond, whereas lateral projections were limited to species of ancient lakes, so both structures were polyphyletic. The evolutionary history of Gammaridae was investigated with ten different calibration schemes, which produced incompatible results; however, the most probable scenario is a late rise of the family, which can only explain the absence of Gammaridae species around the Indo-Paciﬁc.

ADDITIONAL KEYWORDS: ancient lakes – molecular – phylogeny – radiation – subterranean – taxonomy – time estimation.

INTRODUCTION They may occur in dense masses of over 4000 indi- viduals (Davey et al., 2006), or 1.5 g dry weight per The mainly centimetre-sized crustacean species ha- m2 (Mortensen, 1982), below stones and fallen leaves. bitually labelled as ‘gammarids’ or ‘Gammaridae’ are Several hundreds can be taken with one pull of the very common inhabitants of nearly the whole Holarctic hand net. Gammarids are one of the most conspicu- area (except the south-west of North America), where ous animal groups (with hundreds of species) in the they inhabit mainly springs, but also rivers, lakes, ancient lake Baikal (Bajkal, Bajkalskoe ozero). On the coastal lagoons, and the marine littoral zone. Spring other hand, they are only locally present and not very stretches of streams may be literally crowded by them. abundant in subterranean waters. Gammarids are very important constituents of food chains in running waters, and are even being dried and commercially offered as *Corresponding author. E-mail: [email protected] ﬁsh food for aquaria or ﬁsheries. A journal issue was

© 2015 The Linnean Society of London, Zoological Journal of the Linnean Society, 2016, 176, 323–348 323 324 Z. HOU AND B. SKET devoted to the biology of Gammarus recently (Gerhardt, Crangonyctidae, and some other ‘crangonyctoid’ groups, Bloor & Mills, 2011). Anisogammaridae, Phreatogammaridae, and some other Throughout the history, the concept of gammarids southern groups, the Ponto-Caspian Pontogammaridae has varied remarkably. At the beginning of the 20th and some Baikalian families, and the Bogidiellidae ob- century, Stebbing (1906) divided Amphipoda into three tained family status. Moreover, from the morphologi- legiones (Gammaridea, Hyperiidea, and Caprellidea), cally exceedingly diverse Baikalian group of gammarids with Gammaridea comprising all amphipods with head the Russian authors Kamaltynov (1999a) and Tahteev not fused with the first pereonite and with a developed (2000) defined some additional families. maxilliped palp. There were 41 families included in In their most recent morphologically based and Gammaridea, whereas the family Gammaridae includ- cladistically elaborated study, Lowry & Myers (2013) ed 52 genera. Besides the obligatory Gammarus and erected the suborder Senticaudata, including the Echinogammarus, among them were also Crangonyx, infraorder Gammarida Latreille, the parvorder Niphargus, Melita, Maera, Phreatogammarus, prob- Gammaridira Latreille, and the only superfamily ably all Baikalian genera and Ponto-Caspian genera. Gammaroidea Latreille. Beside the family Gammaridae Evidently, virtually the same conception has been ac- Latreille, this superfamily contains some other cepted by Chevreux & Fage (1925) with their ‘S. O. groups of interest. These are the Baikalian endemic GAMMARIENS’ and by Schellenberg (1942) with his families: Acanthogammaridae Garjajev, Baikalogam- ‘Unterordnung Gammaridea’. The splitting of these com- maridae Kamaltynov, Iphigenellidae Kamaltynov, prehensive higher taxa only really began after the Second Macrohectopidae Sowinsky, Micruropodidae Kamaltynov, World War. A review of marine gammaridean amphipods Pachyschesidae Kamaltynov, Pallaseidae Takhteev, the (Barnard, 1969) divided the suborder into informal sec- Ponto-Caspian Pontogammaridae Bousfield, and the tions A–H, but stopped short of naming formal taxa Dinaric-Caucasian troglobiotic Typhlogammaridae between the suborder and family level. In their fol- Bousfield. This is the assemblage that we will call here lowing review of freshwater amphipods Barnard & ‘gammarids’ or ‘Gammaridae’, following the tree in Hou, Barnard, (1983) even went a step back, dissolving es- Sket and Li (2014: Fig. S1). tablished families and treating the gammaroid amphipods A molecular study by MacDonald, Yampolsky & Duffy only by groups appropriate for identification, either mor- (2005), devoted mainly to amphipods from the lake Baikal, phologically or by distribution areas. was very elucidating. Besides the in-group Baikalians, Bousfield was very active in building a phyletic this study also included quite a number of gammaroids system. First, he divided Gammaridea into six from the out-group. They show that all Baikalian and superfamilies and 24 families (Bousfield, 1977). Later, Ponto-Caspian taxa considered are nested within the he divided the suborder Gammaridea into 19 Gammarus–Chaetogammarus group, rendering it superfamilies (Bousfield, 1982, 1983). Within the su- paraphyletic, probably even polyphyletic. On the other perfamily Gammaroidea he formulated and analysed hand, the few representatives of Anisogammaridae split eight polytypic and two unnamed monotypic ‘family basally, and behave as an out-group to the gammarids. groups’: Gammaridae included a number of The study by Englisch (2001), which includes a higher ‘baikalianized’ species; Metohia group (hypogean large number of amphipod families (but with each family gammarids) should be dismembered and mainly placed represented by very few genera) gave a stepwise tree into Gammaridae; Pontogammaridae; the parasitic with the Gammarus–Chaetogammarus as one of the Cardiophilus group and Iphigenella group should distal branches, including also a representative of the be retired, and their species placed into Baikalian fauna (Parapallasea). Pontogammaridae and Acanthogammaridae, respec- Englisch (2001) was also a base for the attempt to tively; Acanthogammaridae (Baikal group) are a hetero- define the family by Hou et al. (2014), where the afore- geneous group of Baikal inhabitants, not included into mentioned Baikalian, Dinaric-Caucasian, and Ponto- Gammaridae; Anisogammaridae are ‘an excellent cluster’ Caspian families are clearly nested within Gammaridae. but were rejected as a taxon, mainly for their (at that Unfortunately, a paper by Browne, Haddock & Martindale time!) paucity. As a result of these controversies, Martin (2007) noticed later, although dealing with hyperiids, & Davis (2001) wrote: ‘The Amphipoda, despite a large has cast some doubts on our decision in delimiting the number of dedicated workers and numerous pro- family. This resulted in the present attempt to verify posed phylogenies and classificatory schemes, remain the monophyly of the family by the inclusion of a richer to a large extent an unresolved mess’. So, they only out-group set. Moreover, this is an appropriate occa- shelled one additional, small suborder (Ingolfiellidea), sion for rethinking the evolutionary history of the family. whereas the suborder Gammaridea remained other- Thus, the monophyly of the group will be again proven wise the same. The family Gammaridae suffered a great in this study, and some hypotheses about its morpho- transformation, as the suborder is divided here into logical development, and geographical and phylogenetic ∼125 families. Of the freshwater groups, Gammaridae, splitting, will be formulated. The taxonomy of the family

© 2015 The Linnean Society of London, Zoological Journal of the Linnean Society, 2016, 176, 323–348 PHYLOGENETICS OF GAMMARIDAE 325 at the genus level will be cleared of polyphyletic genera samples of any previously abandoned taxa, mainly that had been mainly treated as Echinogammarus or Karaman’s and Martynov’s species synonymized with Chaetogammarus. Thus, the most urgent proposals and G. balcanicus by later authors (Birštejn, 1961; Karaman decisions on the family and genus level of the taxono- & Pinkster, 1987). We must accept all other identifi- my will be presented. cations as potentially unreliable, including identifica- tions accepted here from previously published studies (‘Gammarus balcanicus’ from Kazakhstan, MacDonald et al., 2005). An attempt to exactly identify these MATERIAL AND METHODS amphipods will follow our molecular analyses, rather TAXA SAMPLING AND IDENTIFICATION than precede them. Our data set is composed of 467 Gammaridae samples The conceptions of different genera are also chang- from Hou et al. (2014), representing the whole mor- ing fluently. Therefore, only the type species may be phological and geographical array of the family and used as the genus standard. To avoid such problems, 39 non-Gammaridae out-group taxa belonging to 29 we also tried to include type species (again possibly genera from 19 families. Our sampling covered the whole from their type localities) for most of the named genera. distribution area of freshwater Gammaridae and their Animals were mainly caught by a hand net from attested marine relatives. For the moment the Atlan- below stones or between submerged plants or fallen tic islands (Canarias, Azores, and Madeira) remain out leaves in springs, rivers and lakes, some also in cave of reach, but will be sampled for the next study. If we waters. From the deeper bottoms in lakes such as Ohrid exclude the endemic Baikal and Ponto-Caspian faunas, and Caspian, they were obtained with a small dredge Barnard & Barnard (1983) mentioned ∼20 valid genera (triangular frame with 30-cm frame margins) and with and close to 10 synonyms, 19 of them represented in baited traps made of plastic bottles. The samples were our samples, 14 with type species. For Baikal and Ponto- fixed with 96% ethanol, with a small quantity of glyc- Caspian lacustrine faunas we took some representa- erol added. Samples have been kept in a freezer at tive genera. Moreover, we succeeded to include 63 species roughly −20 °C. Single specimens from samples were with samples from their topo-type populations. Despite analysed molecularly. Every specimen for DNA analy- the clear phylogenetic relationships within the family sis was first provisionally identified, then was par- Gammaridae revealed in the previous study (Hou et al., tially dismembered, and pieces of the body or of the 2014), the primary focus of this study is the verifica- musculature were used. All the rest of every such speci- tion of the monophyly of the family Gammaridae, testing men was stored for subsequent verification of taxo- for additional dating attempts and making a taxonomic identity. If at all possible, the head, pleonite III, nomic revision. Thus, we reduced the number of samples and urosomites, as well as most pereopods from one to 91 supposed Gammaridae, representing the whole side were saved for this purpose. morphological and geographical array of the family, and 39 out-group taxa (Table S1). A number of factors make the taxonomic identifi- PHYLOGENETIC ANALYSES cation of gammarids very uncertain. Some species may Genomic DNA was extracted from pieces of speci- morphologically vary beyond the accepted species cri- mens using a standard Phenol Chloroform Isoamyl teria, either locally (some biologically congruent protocol (Hillis et al., 1996). Four different gene regions Gammarus species; Pinkster, 1983) or seasonally, as were amplified using the polymerase chain reaction, is the case for at least some of the ‘Echinogammarus’ including a portion of the cytochrome c oxidase I gene species (Pinkster, 1988). On the other hand, named (COI), a fragment of the gene coding for elongation species are sometimes in fact aggregates of morpho- factor 1α protein (EF-1α), as well as nuclear genes logically yet indistinguishable species (cryptic species, coding for 18S and 28S rRNA. Sequence chromatograms sibling species; note the case of Gammarus fossarum were proofed and edited using SEQUENCHER 4.2 Koch, 1836 in Panzer, 1836; Weiss et al., 2014). A DEMO (Gene Codes Corporation, Inc.). Sequences were number of species had been named, then retracted (as aligned using Clustal X (Thompson et al., 1997) and in the case of the Gammarus balcanicus aggregate; adjusted by eye using MacClade 4.06 (Maddison & Karaman & Pinkster, 1987), but might be molecu- Maddison, 2000). The COI and EF-1α fragments were larly re-established (own experiences; Hou et al., 2014). translated using Drosophila mitochondrial DNA Therefore, we insisted on obtaining as many topotype (mtDNA) or universal genetic code on MacClade to check samples as possible. We also attributed the exact taxo- for the presence of nuclear mitochondrial pseudogenes. nomic names only to morphologically checked speci- The best-fitting partitioning schemes and nucleotide mens from topotype samples; there are some deliberate substitution models were selected using PARTITION exceptions, mainly for marine or morphologically par- FINDER 1.1.0 with the Bayesian information criteri- ticularly aberrant species. We also included topotype on (Lanfear et al., 2012). The four-partition scheme

© 2015 The Linnean Society of London, Zoological Journal of the Linnean Society, 2016, 176, 323–348 326 Z. HOU AND B. SKET defined by gene region was selected, and the general were combined using LogCombiner 1.7.5 and exclud- time-reversible (GTR) model with a proportion of in- ed the non-stationary burn-in phase, resampling every variable sites (I) and gamma-distributed rate hetero- 2000 generations. The maximum clade credibility tree geneity (G) was seleceted as the substitution model was produced with TreeAnnotator 1.7.5, viewed in for all genes. The phylogeny was reconstructed under FigTree 1.3.1. maximum parsimony (MP), maximum likelihood (ML), The fossil records of amphipod crustaceans are very and Bayesian inference (BI). MP analyses were per- few, and no earlier than Upper Eocene. We used three formed using PAUP* 4.0b10 (Swofford, 2002). Align- geological events for calibrations for the first estima- ment gaps were treated as missing data. All characters tion (Hou et al., 2014). First, the ancient Lake Baikal were equally weighted and unordered. All originated around 28 Mya (MacDonald et al., 2005), and phylogenetically uninformative characters were ex- the Baikalian main group of acanthogammarid cluded from the analysis. Heuristic searches were con- amphipods was estimated to be present no later than ducted using tree bisection reconnection (TBR) branch 30 Mya (Ogarkov et al., 1997). Therefore, the time of swapping, with a limit of 10 million rearrangements the Baikalian clade was set to 28 ± 2 Mya. Second, the for each replicate. Bootstrap support indices were gen- retreat of the Tethys Ocean from the Pamir region re- erated for each node using 1000 bootstrap replicates sulted in the split between Sarothrogammarus and executed over ten random-addition sequences. ML analy- Rhipidogammarus at 38 ± 2 Mya. Third, the separa- sis was performed using GARLI 2.01 (Genetic Algo- tion of marine and freshwater Gammaracanthus was rithm for Rapid Likelihood Inference; Zwickl, 2006), set to 6 ± 1 Mya (Väinölä, Vainio & Palo, 2001). Now, with four gene partitions under the GTR+I+Gmodel. we have found a number of additional palaeogeographic Rapid bootstrap analysis with 1000 replicates was con- events that could serve as clock calibrations. Besides ducted with RAxML 7.0.3 (Stamatakis, Hoover & these, we could add rich palaeontological data on the Rougemont, 2008). Initially, ML analyses were run on molluscs that are supposed to develop in parallel with individual genes and then compared with one another geographically close (sympatric) amphipods. A list of to assess compatibility for combined analysis. No well- 15 timing data points is presented in Table 1. Each supported branches (80% bootstrap support) among dif- calibration point was set with a normal distribution ferent genes were in conflict; therefore, the four genes prior. The calibration points were combined based on were combined and analysed in the following. BI was the following rules: first, without conflict among points, conducted using MrBayes 3.2.1 (Ronquist et al., 2012) such as ancient node with older time and tip node with with a partitioned four-gene model. We ran the Monte younger age; second, calibration points evenly distrib- Carlo Markov chain for 10 million generations, sam- uted in the tree, preventing undue emphasis on any pling every 100 generations. The first 50 thousand trees one clade; and third, different schemes confirming the were discarded as burn-in, after checking for stationarity compatibility among calibration points. Unfortunate- and convergence of the chains in TRACER 1.5 (Rambaut ly, groups of these events are producing very differ- & Drummond, 2009). The majority consensus tree with ent timing schedules for the gammarid phylogeny. posterior probabilities was constructed from the last 50 thousand trees. RESULTS The phylogenetic tree and the sample list from Hou et al. (2014: figs 1, S1, table S1) were widely used in PHYLOGENETIC ANALYSIS this study. The alignment of the combined data set contained 130 taxa with 5693 base pairs (bp), including 1514 bp for 28S, 677 bp for COI, 2901 bp for 18S, and 601 bp for ESTIMATION OF PHYLOGENY DIVERGENCE TIMES EF-1α. All new sequences were deposited in GenBank The hypothesis that our in-group data evolved accord- (accession numbers KT180183–KT180192). The ma- ing to a strict molecular clock was rejected by a like- trices are available in TreeBase (study accession number lihood ratio test using PAUP* (lnLclock = −56021.7, S17777). The MP, ML, and BI analyses produced con- lnLnoclock = −55608.8, df = 89, P = 0). Consequently, gruent phylogenetic trees, except for a few disagree- time estimation was performed using a relaxed model. ments existing in nodes with lower support values. The The analysis was carried out in BEAST 1.7.5 main discrepancy was the position of Baikalian (Drummond et al., 2012), under an uncorrelated log- amphipods. They are divided into two groups by ML normal relaxed molecular model and Yule speciation and BI analyses: one as a sister group to the Eura- prior. Two parallel runs of 50 million generations were sian Gammarus clade, and the other as a sister group conducted, sampling every 1000 generations. The to the Oriental Gammarus clade. Whereas the two stationarity of each run was examined using TRACER groups are nested in the Eurasian Gammarus clade with an effective sampling size of each parameter, in the MP tree. MP topology had comparatively lower ESS > 200. The resulting files from independent runs support values for major clades (Figs 1, 2).

Table 1. Molecular clock calibration points

Calibration point Age (Myr) Geographical timing event Possible phylogenetic consequence a 0.07 ± 0.01 70 000 years ago (Väinölä et al., 2001) Split between Gammaracanthus caspius and Gammaracanthus aestuariorum b 3 ± 1 3 Mya cooling and oxygenation of Baikal Radiation of both Baikalian groups depths c 6 ± 1 6 Mya Salinity Crisis – isolated habitats Main radiation era of Mediterranean ‘Echinogammarus’ d 8 ± 1 8 or 12 Mya, distinct Pannon Basin Radiation of Ponto-Caspian flock e 12 ± 1 12 or 8 Mya, radiation of Mollusca in Radiation of Ponto-Caspian flock Paratethys, fossil transformed Amphipoda g 17 ± 1 17–15 Mya, Dinaric Lakes system Separation of Dinaric from mediterranean ‘Echinogammarus’ h 19 ± 2 19 Mya, disruption of Dinaric–Caucasian Split between Typhlogammarus–Metohia and connections Echinogammarus s. str. j 29 ± 2 29 Mya, separation of Turan (i.e. Paratethys) Separation of the Dinaric + Ponto-Caspian Sea clade k 29 ± 2 29 Mya, enlargment of Dinaric land Radiation of Dinaric ‘Echinogammarus’ l 30 ± 2 30 Mya, final break of corridor between Separation of Atlantic marine lineages: Turan Sea (= eastern Paratethys) and Gammarus fasciatus, Gammarus locusta, North Sea and Gammarus zaddachi m 32 ± 2 32 Mya, closure of the Turgaj Strait Separation of Gammarus o 34 ± 2 34 Mya, Central Asia becomes dry land Split between Rhipidogammarus and Sarothrogammarus p 35 ± 2 40–30 or 20–15 Mya, faunal exchanges Split between between east and west Atlantic Gammarus locusta–zaddachi and Gammarus fasciatus groups r 18 ± 2 40–30 or 20–15 Mya, faunal exchanges Split between between east and west Atlantic Gammarus locusta–zaddachi and Gammarus fasciatus groups x 6 ± 1 6 Mya, secondary calibration from Väinölä Separation of Arctic marine and freshwater et al., 2001 Gammaracanthus y 28 ± 2 28 Mya, formation of Lake Baikal The most recent common ancestor of Baikal amphipods z 38 ± 2 38 Mya, Late Eocene retreat of the Tethys Split between Sarothrogammarus and from the eastern Pamir Rhipidogammarus

The phylogenetic relationship within Gammaridae cally, but with each branch being morphologically very is in accord with that in the complete tree published heterogeneous. The clade of Gammarus is a separate by Hou et al. (2014). The main novelty of this study lineage, including the genera Gammarus, Sinogam- is a richer set of the out-groups, representing 29 genera marus, and Fontogammarus, distributed across Eurasia of 20 families (Fig. 1). A group of 19 genera form a and North America, as well as two clades of endemic monophylum with Gammaridae, mainly correspond- Baikalian genera. Gammarus further splits into ing to Senticaudata (Lowry & Myers, 2013), but in- perimarine Gammarus species endemic to the cluding Haustorius and Monoporeia; the sister group Mediterranean–Atlantic–Arctic area, and including North to this Senticaudata clade contains six genera of five American freshwater species, and the monophyletic group families. The hyperiids is placed at the base, in clear of freshwater gammarids widely distributed in the contradiction to the mixing scenario between gammarids Palaearctic. Within freshwater Gammarus, five clades and hyperiids (Browne et al., 2007). were recovered. The Eurasian (or Palaearctic) Gammarus Gammaridae forms three monophyletic groups: the clade is composed of Gammarus lacustris and northern branch (genus Gammarus), the southern branch G. balcanicus aggregates. Gammarus lacustris aggre- (pontogammarids in Hou et al., 2014), and gate is widely spread in Holarctic fresh waters, whereas sarothrogammarids, which are well defined geographi- G. balcanicus aggregate is mainly located in eastern

Figure 1. Phylogenetic tree of Gammaridea derived from maximum-likelihood (ML) analyses based on a combined mitochondrial and nuclear data set, indicating the monophyly of Gammaridae. Numbers above and below branches are ML bootstrap values and Bayesian posterior probabilities. The reputed families are labelled.

and south-eastern Europe and Central Asia, and reach pulex)+Gammarus roeselii) distributed in Europe and the extreme south-east of Kazakhstan and neighboring in south-west Asia. area of Xinjiang in China; the genus Fontogammarus The southern branch, including sarothrogammarids, is nested in the G. balcanicus group. The oriental is originally limited to fresh and brackish waters of clade unites all the Eastern Asian Gammarus and southern Europe and to the north Atlantic littoral zone. includes Sinogammarus species. The Baikalian It again comprises a highly morphologically diversi- amphipods Micruropodidae with the nested ﬁed group, monophyletic and originating in Ponto- Macrohectopidae are one monophylum nested Caspian lakes. within Gammarus; a separate clade nested within Gammarus is the set of four families (Acantho- gammaridae + Eulimnogammaridae + Pallaseidae + TAXONOMIC SOLUTIONS Odontogammaridae), including the bulk of Baikalian In the following diagnoses we deﬁne clades listed in gammarid diversity. The European clade consists of the phylogenetic tree presented in Figure 2. For ad- species of aggregates ((Gammarus fossarum + Gammarus ditional data see the Discussion.

Figure 2. Maximum-likelihood (ML) phylogeny of Gammaridae with out-groups removed, based on a combined mitochondrial and nuclear data set. Numbers above and below branches are ML bootstrap values and Bayesian posterior probabilities. The major groups are labelled. For details, see Hou et al. (2014: ﬁgure S1).

GENUS HOMOEOGAMMARUS SCHELLENBERG, 1937, (1973, 1993) and Atlantic Echinogammarus by Hou et al. WITH THE TYPE SPECIES GAMMARUS SIMONI (2014). Its type species is Gammarus berilloni Catta, CHEVREUX, 1894 1878. Diagnosis Peri-Mediterranean, gammariform amphipods. Pereon Diagnosis and pleon dorsally smooth, seldom with dorsal keels South-west European gammariform amphipods similar or teeth. Each telson lobe less than twice as long as to Homoeogammarus, but pereon smooth, pleonites with broad. Antennae I and II normal, with slender pedun- prominent, numerous, and irregularly disposed stout cle articles, anntena II shorter or (seldomly) subequal spines (spiniform setae), and/or tufts of long setae dor- to antenna I. Antena II with short or long, dense or sally or/and laterally; no keels or teeths. Gnathopod sparse, straight setae. Excretory cone generally short. propodus II longer than propodus I. Uropod III exopodite Mouth parts as in Gammarus; outer lobe of maxilla I well developed, biarticulate, with marginal spines and with ∼11 spines, palps asymmetric. Gnathopods I and long, straight setae. II subchelate, propodus II longer than, seldom equal to, propodus I. Coxal plate IV distoposteriorly lobate. GENUS MARINOGAMMARUS SCHELLENBERG, 1937B, Pereopod VII basis without a distoposterior lobe. WITH THE TYPE SPECIES GAMMARUS MARINUS Uropods I and II usually normal, with distal and lateral LEACH, 1815 spines. Uropod III exopodite with marginal groups of Diagnosis spines, usually accompanied by long setae that are Peri-Atlantic gammariform amphipods similar to always straight; endopodite reduced and scale-like, with Homoeogammarus, but each telson lobe at least twice terminal spine(s) only (without facial or marginal spines). as long as broad, with a distal and a lateral cluster of spines. Gnathopod propodus II longer or shorter than propodus I. Uropod III exopodite well developed, with GENUS CHAETOGAMMARUS MARTYNOV, 1924 WITH long and straight setae joined to marginal groups of THE TYPE SPECIES GAMMARUS ISCHNUS spines (rarely on one side only); endopodite slightly STEBBING, 1899 elongated (not scale-like) with (beside terminal spines) Diagnosis at least one facial or marginal spine. Ponto-Caspian gammariform amphipods similar to Homoeogammarus, but antenna II densely set with long GENUS PARHOMOEOGAMMARUS SCHELLENBERG, setae ventrally, those on article 4 more than twice as long as the diameter of the article. Gnathopod 1943, WITH THE TYPE SPECIES GAMMARUS propodus II longer than propodus I. Uropod III exopodite (PARHOMOEOGAMMARUS) LUSITANUS well developed, biarticulate, with marginal spines and SCHELLENBERG, 1943 just a few, short setae; endopodite rudimentary, scale- Diagnosis like, with rudimentary spination/setation. Gammariform amphipods similar to Homoeogammarus, but gnathopod propodus II shorter than propodus I. Uropod III endopodite less than 50% exopodite length, GENUS TRICHOGAMMARUS GEN. NOV., WITH THE but linear, with terminal and marginal spines (only TYPE AND ONLY KNOWN SPECIES CHAETOGAMMARUS exceptionally scale-like). TRICHIATUS MARTYNOV, 1932

Diagnosis GENUS RELICTOGAMMARUS GEN. NOV., WITH THE Peri-Pontic gammariform amphipods similar to TYPE AND THE ONLY KNOWN SPECIES GAMMARUS Homoeogammarus, but antenna II densely set with long STOERENSIS REID, 1938 setae ventrally, on article 4 more than twice as long Diagnosis as the diameter of the article. Gnathopod propodus II Peri-Atlantic gammariform amphipods similar to longer than propodus I. Uropod III exopodite well de- Homoeogammarus, but each telson lobe at least twice veloped, biarticulate, with marginal spines and densely as long as broad, with a distal and a lateral cluster set long and curled setae; endopodite rudimentary, scale- of spines. Gnathopod propodus II longer than propodus I. like, with apical spines only. Uropod III exopodite without setae; endopodite scale- like, with only apical spine(s).

GENUS ECHINOGAMMARUS STEBBING, 1899 (ECHINOGAMMARUS S. STR.) DIVERGENCE TIME ESTIMATIONS Echinogammarus in a revised sense was previously We succeeded in identifying 15 palaeogeographic events named Echinogammarus berilloni group by Pinkster and some series of palaeontological data (Table 1) that

Figure 3. Bayesian inference chronogram of Gammaridae based on a combined data set. Black dots are calibration points: e, h, and r. Blank dots indicate the nodes listed in Table 1, with the basal split within Gammaridae marked by a black triangle. The grey vertical bar indicates the boundary of the Cretaceous and the Palaeogene for the fall in sea level. For the complete set of ten possible chronograms, see Figure S1.

could be decisive or supporting for some phylogenetic DISCUSSION splits in Gammaridae. In ten chronograms, starting MOST FUNDAMENTAL TAXONOMIC PROBLEMS with two or three calibration points each, the origin of basal Gammaridae range across very different time What may be Gammaridae? points, from 160 Mya to less than 6 Mya (Fig. S1E, Our present study is devoted to the group estab- H). Four analyses produced similar divergence esti- lished in Hou et al. (2014) as Gammaridae. We decided mates for the origin of Gammaridae at the boundary to unite the family as a molecularly compact clade of of the Oligocene and the Eocene (Fig. S1A–D). We amphipods, separated from its sister clade, the small consider the most credible chronogram (Figs 3, S1A) family Gammaracanthidae, by a longer branch (Fig. 1). based on the final disruption of the marine connec- The positions of the included taxa are the same as in tion between the Dinaro-Caucasian and the western previously published trees based on more conserved European areas, the time of the last palaeontologically genes (Englisch, 2001; MacDonald et al., 2005). In both, documented faunal exchanges between the east and Gammaridae are represented by only small numbers west Atlantic, supported by palaeontological data in of taxa and both trees are without representatives of Bivalvia and Gastropoda. This puts the origin of Gammaracanthidae. In both papers Haustoriidae and Gammaridae at the time point when gammarids were Pontoporeiidae are close to the root of Gammaridae, most probably no longer able to pass the Mediterranean– as in our tree. Pacific connections. Then Gammaridae diversified as In the study by Browne et al. (2007), the gammarid the lineages of Gammarus around 19 Mya, clade was split into three branches, interspersed by pontogammarids at 25 Mya, and sarothrogammarids some non-gammarid clades. The aim of that study was at 16 Mya. As the divergence time estimates heavily different and only COI was used as the marker, which depend on the calibration schemes, we should discuss is very variable and not appropriate for phylogenetic time ranges rather than absolute times. conclusions on higher taxonomic (phylogenetic) levels

(Costa et al., 2009). In our recent analysis, we includ- Caspicolidae, Birstein 1945, not represented in our ed genes of a higher relevance for all non-gammarid samples. taxa of concern, and they were all revealed as out- group to Gammaridae (Fig. 1). What is Gammarus? With the inclusion of the nested pontogammarids and Gammarus is an extensive monophylum consisting of Baikalian genera, our Gammaridae (Fig. 2) are a a series of subclades (Hou et al., 2011, 2014), i.e. of monophyletic and holophyletic, biogeographically con- taxonomically informal aggregates. Although mainly sistent group inhabiting mainly fresh waters of well supported molecularly, and geographically well the Palaearctic and eastern Nearctic land, and the At- defined, the structure of most of them does not allow lantic (and Mediterranean) coastal seas in between. us to define and distinguish these clades morphologi- In our conception, Gammaridae comprises most cally. This is annoying, as two subclades, quite distal genera traditionally called gammarids, but also in- within the tree, are extraordinarily modified and diverse. cludes Baikalian and Ponto-Caspian assemblages of These are two groups of the endemic Baikalian species, morphologically very aberrant taxa, which have not diverse to such an extent that several hundreds of mor- been accepted in any of the morphology based gammarid phologically distinguishable species were described, and systems. This solution has a high degree of support, as many as tens of genera have been erected 83% for MP and 95% for ML topologies (Hou (Kamaltynov, 2001). They were even grouped into three et al., 2014), whereas in the current study the support independent families (Kamaltynov, 1999a; Tahteev, 2000) value decreased a little because of the large mentioned above, with the addition of Pallaseidae out-groups. Tachteew, 2000, Carinogammaridae Tachteew, 2000, We are leaving Gammaracanthus (and the potential Eulimnogammaridae Kamaltynov, 2001, and prob- Gammaracanthidae) outside of Gammaridae as a sister ably Pachyschesiidae Tachteew, 1998 (Tahteev, 1998, taxon because of a very high patristic distance, of the 2000). Considering the most recent system of Baikal same order as that for Haustoriidae, which according amphipods (Kamaltynov, 2001), the Baikal group 1 in to Englisch (2001) should be the closest relative of our tree (Fig. 2; Hou et al., 2014: fig. S1) includes diverse Gammaridae. Some other easily recognizable Gammarus- members of the families Acanthogammaridae, shaped groups are also excluded from Gammaridae: Eulimnogammaridae, and Pallaseidae. Baikal group 2 Anisogammaridae (probably sister to Gammaridae), includes members of Micruropodidae and (the single Crangonyctidae (and possibly Crangonyctoidea), and species of) Macrohectopidae. Within both monophyla, Niphargidae. On the other hand, a number of occa- branchlets with members of the named families sionally recognized families remain within Gammaridae mentioned are intermingled, thus none of the mor- according to molecular analysis, including some morphologically defined families Micruropodidae, phologically aberrant groups (Fig. 2; Hou et al., 2014: Acanthogammaridae, and Pallaseidae is a monophylum. fig. S1). The Ponto-Caspian ‘Pontogammaridae’ are sister We may discard the nested (Baikalian) families here to peri-Mediterranean ‘Echinogammarus’ (here dis- to reduce the paraphyly of the genus Gammarus to placed to Homoeogammarus). The consistently troglobiotic the minimum possible, but it is not possible to cancel and troglomorphic metohiids or ‘Typhlogammaridae’ are all Baikalian genera and fuse their species with sister to Echinogammarus s. str. At least the ‘families’ Gammarus as the cladistic praxis would demand Macrohectopidae, Micruropodidae, Acanthogammaridae, (Morrone, 2006). The utilitarian aspect of taxonomy and Pallaseidae, endemic in lake Baikal, are nested definitely deserves the retention of these extremely within the genus Gammarus and, excepting diverse taxa. The morphological diversity of Baikalian Macrohectopidae, are polyphyletic. They will be treated amphipods is not far off the entire extent of diversity below. Excluding the aforementioned families from in Amphipoda! Gammaridae renders this family paraphyletic; in this The genus Gammarus, as generally conceived today, particular case, we are avoiding the paraphyly of is morphologically practically indivisible, in accord with the taxon, as it would bring no practical benefits. Thus, the molecular relationships held within it. Any at- of the families listed by Martin & Davis (2001) and by tempts to systematize Gammarus species could merely Lowry & Myers (2013), our Gammaridae would include: result in some explicitly morphologically defined groups Gammaridae Latreille, 1802, Acanthogammaridae (Stock, 1967; Karaman & Pinkster, 1977a,b, 1987). Most Garjajev, 1901, Macrohectopidae Sowinsky, 1915, of them actually turn out to be polyphyletic in this study. Micruropodidae Kamaltynov, 1999, Pontogammaridae On the other hand, the established monophyletic ag- Bousfield, 1977, and Typhlogammaridae Bousfield, gregates generally overlap in morphology, as well as 1977. in geographical distributions; however, we may cancel What remains unclear is the status of Meso- some genera that are nested in other aggregates gammaridae Bousfield, 1977 and Pachyschesiidae of Gammarus (compare Hou et al., 2014: fig. S1). Tachteew, 1998, as well as the monotypic Caspian The Dinaric Fontogammarus S. Karaman, 1931 is

© 2015 The Linnean Society of London, Zoological Journal of the Linnean Society, 2016, 176, 323–348 PHYLOGENETICS OF GAMMARIDAE 333 characterized by small differences in the mouth ap- unrecognizable. We will try to clarify the relationships pendages alone, and is phylogenetically situated within here. the G. balcanicus aggregate, very close to the topotype The monophylum containing all Echinogammarus– population of G. balcanicus itself. Omission already sug- Chaetogammarus types also includes three clades with gested by S. Karaman (1935). Sinogammarus Karaman groups(!) of unanimously accepted genera. Any & Ruffo, 1995 with type species Sinogammarus trog- association (fusion) of all Echinogammarus– lodytes Karaman & Ruffo, 1995 is characterized by some Chaetogammarus-like genera appears a highly troglomorphic traits, and is nested within the polyphyletic assemblage after our molecular analy- Oriental aggregate. The Caucasian troglobiotic sis. Unfortunately, most monophyla are only weakly Anopogammarus Deržavin, 1945 is not represented here morphologically different, sometimes even not at all, with the type species; if Anopogammarus revazi (Birštejn but at least geography/ecology supports their molecu- & Levuškin, 1970) is a good representative of the genus, larly established distinction. it can also be abandoned. The proposed Relictogammarus gen. nov., which is Some other genera or subgenera have been a sister clade to all other Gammaridae (including synonymized with Gammarus earlier (Barnard & Gammarus), is morphologically distinctly although very Barnard, 1983): Lagunogammarus Sket, 1971 (with weakly different from Marinogammarus. The sup- Gammarus zaddachi Sexton, 1912), Mucrogammarus posed Parhomoeogammarus is morphologically and Barnard & Grey, 1968 (with Gammarus mucronatus geographically well defined. The proposed Echino- Say, 1818), Pephredo Rafinesque, 1817 (with Pephredo gammarus s. str. is a morphologically well-defined taxon, potamogeti Rafinesque, 1820), and Lepleurus Rafinesque, different from any non-lacustrine gammarid species, 1820 (with Lepleurus rivularis Rafinesque, 1820 = and is geographically and ecologically separated from Gammarus minus Say, 1818) are subclades of the its sister taxa (of the Typhlogammarus group) in Dinaric perimarine and American Gammarus clades; and Caucasian subterranean waters. Marinogammarus Fluviogammarus S. & G. Karaman, 1959 (with is morphologically close to Homoeogammarus, but seems Carinogammarus triacanthus Schäferna, 1922; nec to be geographically/ecologically well defined. Fluviogammarus Dorogostaisky, 1917) and Homoeogammarus is a large and modestly diverse Carinogammarus sensu Schäferna, 1922 (with genus, but is geographically well defined and sister to Gammarus triacanthus Schäferna, 1922; nec the Pontocaspian groups. The inclusion of the mor- Carinogammarus Stebbing, 1899) are subclades of phologically indistinguishable Dinaric continental G. roeselii aggregate; Rivulogammarus S. Karaman, 1931 forms into this genus makes Homoeogammarus [with Gammarus pulex (Linnaeus, 1758)] is a synonym paraphyletic, but applicable. Chaetogammarus and of Gammarus. Trichogammarus gen. nov. are morphologically similar, but are nevertheless distinct and molecularly nest sepa- What can be Echinogammarus? rately within the morphologically very diversified group Besides Gammarus,‘Echinogammarus’ has been a steady of Pontocaspian genera. We do not discuss the other member of Gammaridae since the beginning of the 20th genera of the Pontocaspian group mentioned here (except century. Characterized by the dorsal spinosity of the Chaetogammarus and Trichogammarus), for that would pleonites, it had contained mainly Baikalian species require a more extensive sampling within its entire recently regarded as Eulimnogammarus. It was later endemic distribution area. The phylogenetic position considered a close counterpart of Gammarus, charac- of the aforementioned genera is clearly presented in terized consistently, and ultimately only(!), by a short- Figure 2, but for more information compare with Hou ened endopodite of uropod III (representing a parviramus et al. (2014: fig. S1). All of these clades are molecu- clade by Schellenberg, 1937b; Barnard & Barnard, 1983; larly well supported, as are the phylogenetic relation- Karaman, 1993). In fact, this genus has been a tem- ships in the main. porary bin for members of a number of other named genera, such as Chaetogammarus, Marinogammarus, Homoeogammarus, Parhomoeogammarus, Neogammarus, GENUS HOMOEOGAMMARUS SCHELLENBERG, 1937B and exceptionally even Gammarus and Eulimno- Exceptions in morphology gammarus. Our tree (Fig. 2; Hou et al., 2014: fig. S1) There are a few details that hamper a clear picture shows that most of these taxa are branches of a of Homoeogammarus. Morphologically, five supposed sister clade to Gammarus, whereas some of them split member species are without (or nearly so) long setae from the tree even more basally. Most of the men- on uropod III. In Homoeogammarus tibaldii (Pinkster tioned genera are molecularly objective, definable, & Stock, 1970), antenna II setae are ‘sometimes curled’ biogeographically sound, and some even have existing (Pinkster, 1993), on article 4 densely set. In (available) potential type species. On the other hand, ‘Laurogammarus’ scutarensis (Schäferna, 1922), the ex- they may be morphologically indistinct or even cretory cone is elongated. In ‘Neogammarus’ species,

© 2015 The Linnean Society of London, Zoological Journal of the Linnean Society, 2016, 176, 323–348 334 Z. HOU AND B. SKET uropods I and II are slightly reduced and without lateral the Black Sea and the area of the former Dinaric Lakes spines. System (sensu Harzhauser & Mandic, 2008). Homoeogammarus is a sister clade to the diverse Ponto- Type species Caspian group of genera. Gammarus simoni Chevreux, 1894, described from Karaman (1977b) already substantiated the Tunisia and Algeria. synonymizing of Neogammarus Ruffo, 1937 with Echinogammarus. In our tree and amended system, Additional species Neogammarus nudus Stock, 1971 is clearly a member Molecularly (conditionally) proven (Hou et al., 2014, as of Homoeogammarus. Similarly, the genus Mediterranean Echinogammarus and Dinaric Laurogammarus (for Carinogammarus scutarensis Echinogammarus) taxa are: (Mediterranean) Schäferna, 1922), characterized only by an elongated Carinogammarus thoni Schäferna, 1922, Gammarus and curved conus excretorius (Pinkster, 1993), may be pungens H. Milne Edwards, 1840, Gammarus veneris abandoned and its only species retained as Heller, 1865, Ostiogammarus stammeri S. Karaman, Homoeogammarus scutarensis (Schäferna, 1922). 1931, Carinogammarus scutarensis Schäferna, 1922, Homoeogammarus as conceived here is inevitably Echinogammarus cyrtus Pinkster and Platvoet, 1986, paraphyletic, as the mainly continental Dinaric group Echinogammarus platvoeti Pinkster, 1993, Gammarus of species (dinarogammarids) fall morphologically within foxi Schellenberg, 1928, Echinogammarus monomerus the Homoeogammarus and could not be morphologi- Stock, 1977, Echinogammarus tibaldii Pinkster and cally deﬁned as a practicable genus. On the other hand Stock, 1970, Neogammarus nudus Stock, 1971; (Dinaric) the very diverse Ponto-Caspian group, by taxonomi- Gammarus cari S. Karaman, 1931, Gammarus cal practice split into 33 genera (Grabowski, 2014), acarinatus S. Karaman, 1931, and Ostiogammarus cannot be merged with Homoeogammarus. bosnensis S. Karaman, 1934. The majority of the Echinogammarus pungens group, sensu Stock, 1968, GENUS CHAETOGAMMARUS MARTYNOV, 1924 are probable members. Type species Distribution Gammarus ischnus Stebbing, 1899 (synonym Gammarus Homoeogammarus spp. mainly inhabit fresh and brack- tenellus Sars, 1896, a homonym). ish waters draining towards the Mediterranean sea(s). The genus is represented around the Black Sea only Aditional species by the very euryecious Echinogammarus foxi. Its pres- Molecularly proven (Hou et al., 2014) members are: ence there and in the Euphrates may be a more recent Gammarus warpachowskii G.O. Sars, 1894, Gammarus invasion, similar to E. berilloni in the West. The placidus G.O. Sars, 1895, and probably Gammarus placement of Gammarus olivii Milne Edwards, 1830 behningi Martynov, 1919, Chaetogammarus pauxillus is debatable – it might also be present in the Atlan- G.O. Sars, 1896, Sovinskya macrocerus Deržavin, 1948, tic, as supposed by Stock (1993), or belong to Chaetogammarus hyrcanus Pjatakova, 1962, and Marinogammarus, as supposed by Sexton & Spooner Gammarus sowinskyi Behning, 1914. (1940). Distribution Remarks The natural distribution area of Chaetogammarus are Homoeogammarus is an available name for the ma- the Black and Caspian seas, and their affluents. All jority of the ‘Echinogammarus’ species designated known species are present in the Caspian, and most ‘Echinogammarus pungens group’ by Stock (1968) or are found in both the Black Sea and the Caspian drain- ‘Mediterranean Echinogammarus’ by Hou et al. (2014). ages. Some species are recent strong invaders all over Ostiogammarus S. Karaman, 1931 is unavailable Europe, and even in North America. because its author did not state the type species, whereas Stock (1968) and Barnard & Barnard (1983) Remarks awkwardly foisted upon it the same type species as These species are morphologically very similar to for Echinogammarus. Schellenberg’s name Homoeogammarus (although distinguishable by Homoeogammarus was made available by Barnard & uropod III setation). As it is deeply nested within the Barnard (1983), who selected the North African (and group of Ponto-Caspian genera, we cannot fuse it with probably Lusitanian) G. (H.) simoni for the type species. the peri-Mediterranean relatives. This species appeared to be a member of the Otherwise, the monophyletic Ponto-Caspian group monophylum composed of all peri-Mediterranean of genera exhibits great diversity, slightly reminis- ‘Echinogammarus’ species treated here, inhabiting brack- cent of the Baikalian diversity; Chaetogammarus is a ish or fresh waters of the area, with the exclusion of branch between other gammaroid and pontogammaroid

© 2015 The Linnean Society of London, Zoological Journal of the Linnean Society, 2016, 176, 323–348 PHYLOGENETICS OF GAMMARIDAE 335 branches of the Ponto-Caspian group genera, and along vading Germany (Thienemann, 1950), and was found with Trichogammarus and the aberrant Derzhavinella to be invading again recently in Germany and France the only Ponto-Caspian gammarids without a (Pinkster, 1993). distoposterior lobe on bases of pereopod VII. Remarks Echinogammarus s. str. is a distinct clade including GENUS TRICHOGAMMARUS GEN. NOV. only south-west European freshwater species, as defined above. According to Stock (1993), they are morpho- Type and the only known species logically characterized by diverse and extraordinary Chaetogammarus trichiatus Martynov, 1932 (syn. setosity and/or spinosity of pleonites and urosomites, Chaetogammarus tenellus major Caraus¸u, 1943). and peduncle articles of antennae I and II very slender. Surprisingly, in our tree, this clade includes the Cau- casian troglobiotic and morphologically very different Etymology Zenkevitchia admirabilis Birstein, 1940, whereas it is Named after its hairy uropod III; in ancient Greek in sister relationship with the clade of the Dinaric thríks, trikhós = hair. troglobiotic Typhlogammarus group. The need for re- definition of Echinogammarus s. str. was already expressed by Stock (1993); he left all other Mediterranean, Distribution gammariform Ponto-Caspian and marine Atlantic Chaetogammarus trichiatus inhabits affluents of the parviramus ‘echinogammarids’ in the genus Black Sea only. Chaetogammarus. In our tree (Fig. 2) this clade is very distant from all other would-be ‘Echinogammarus’ clades, Remarks which appeared to be scattered throughout the Although very similar to both Chaetogammarus and phylogenetic tree. Homoeogammarus, it can be distinguished by the char- acteristic setosity of uropod III. Phylogenetically, this GENUS MARINOGAMMARUS SCHELLENBERG, 1937B clade is also nested within the Ponto-Caspian group, Type species distant from Homoeogammarus, but very close to the Gammarus marinus Leach, 1815. Dinaric clade of Homoeogammarus. Aditional species Molecularly (conditionally) proven (Hou et al., 2014) GENUS ECHINOGAMMARUS STEBBING, 1899 taxa are: Gammarus obtusatus Dahl, 1938, (ECHINOGAMMARUS S. STR.) Marinogammarus pirloti Sexton and Spooner, 1940; Type species probably also Gammarus finmarchicus Dahl, 1938, and Gammarus berilloni Catta, 1878. Marinogammarus atlanticus Dahl, 1958.

Distribution Additional species Marinogammarus spp. inhabit shallow eastern and Molecularly proven (Hou et al., 2014) included taxa are: north-western North Atlantic waters, particularly where Gammarus calvus Margalef, 1956, Echinogammarus inﬂuenced by freshwater inﬂow. If G. olivii belongs here, tarragonensis Pinkster, 1973, Echinogammarus as supposed by Sexton & Spooner (1940), the genus longisetosus Pinkster, 1973, Echinogammarus feminatus is also present in the Mediterranean. Pinkster, 1973, Echinogammarus aquilifer Pinkster, 1969, Echinogammarus zebrinus Pinkster & Stock, 1971, Remarks and probably all other Echinogammarus berilloni group Marinogammarus is a sister clade to species sensu Pinkster, 1993, Stock, 1993. Homoeogammarus + Ponto-Caspian group of genera, and molecularly very distant from that sister group. It also differs in biogeography. Distribution These are freshwater species, inhabiting mainly At- lantic affluents in France and Spain, and with a number GENUS PARHOMOEOGAMMARUS SCHELLENBERG, 1943 of species (also) in the Mediterranean drainages of Spain. Type species Note that large river systems are traversing the Gammarus (Parhomoeogammarus) lusitanus Pyreneean peninsula in an east–west direction. In 1920 Schellenberg, 1943 (from northern Portugal and Echinogammarus berilloni was already found to be in- north-west Spain), but tentatively represented in our

© 2015 The Linnean Society of London, Zoological Journal of the Linnean Society, 2016, 176, 323–348 336 Z. HOU AND B. SKET analysis with Gammarus anisocheirus Ruffo, 1959 from upper reaches of streams. Other bodies of water are the western French Pyrenees. inhabited by fewer species or by phylogenetically very narrow groups of species. Gammaridae inhabit all sa- Remarks linities of euhaline (exceptionally also hypersaline), The appurtenance of G. anisocheirus Ruffo, 1959 to this brackish (mixohaline), and hard or soft limnic waters. group is only putative, but is plausible morphologi- The very basally split branches of Gammaridae cally as well as geographically; otherwise, it has been (Relictogammarus and Marinogammarus), many considered by Barnard & Barnard (1983) as an Homoeogammarus spp., as well as the basally split Echinogammarus species, and by Pinkster & Stock branches of the genus Gammarus are characteristi- (1970) as a ‘western European species of the pre- cally halophilic. The ‘basal position’ of halophiles can sumed Baikal-genus Eulimnogammarus’. At the only mean that the strict freshwater clades accom- moment, the type species is the only reliable member plished much richer differentiation and speciation (Crisp of Parhomoeogammarus; G. anisocheirus is a very prob- & Cook, 2005) than perimarine clades. Even the ‘purely able member, along with the other freshwater Euro- marine forms’ such as G. locusta (Stock, 1967) can be pean ‘Eulimnogammarus’ species, Eulimnogammarus bred at low salinities of 33‰ (Neuparth, Costa & Costa, macrocarpus Stock, 1969 and Eulimnogammarus 2002). Gammarus wilkitzkii Birula, 1897 is present in toletanus Pinkster and Stock, 1970. According to recent estuaries or at depths of up to 3900 m in the sea; it descriptions (Pinkster & Stock, 1970; Pinkster, 1993), is often associated with ice and living at tempera- uropod III endopodite is less than 50% of the exopodite tures close to 0 °C, with a salinity range of 0.5– length, but only exceptionally scale-like. An addition- 34.0‰ (Cvetkova, 1975). al member might be Echinogammarus spinulicornis Pinkster & Stock, 1971 from Atlantic France. The com- Troglobiotic species bination of enlarged gnathopod I and the linear Gammaridae immigrated sporadically, at different times uropod III endopodite is very peculiar in gammarids. and places independently into subterranean waters. A number of Gammarus species from all freshwater ag- Distribution gregates and areas in Eurasia and America inhabit cave The type species inhabits streams in northern Portu- waters, and developed a lower or higher degree of gal and north-western Spain, but the genus might have troglomorphy (Karaman & Pinkster, 1977a, 1987; Stock, a wider distribution across the Pyrenaean Peninsula, 1986; Fong, 1989; Culver, Kane & Fong, 1995; Carlini, and probably also in south-western France. Manning & Sullivan, 2009). The most known and well- studied eutroglophilous (sensu Sket, 2008) species G. minus, shows that even a single species could have GENUS RELICTOGAMMARUS GEN. NOV. invaded subterranean habitats repeatedly (Culver et al., Type and the only known species 1995), as is the case in other animal groups (Sket, 1997). Gammarus stoerensis Reid, 1938. The Typhlogammarus–Metohia group of troglobiotic and highly troglomorphic species inhabits the Dinaric Etymology and Caucasian karsts. They are molecularly closely In Latin relictus = retarded, left behind. related only to the geographically and morphologically distant Echinogammarus s. str., which are also Distribution somehow specialized freshwater animals (Pinkster, The species is present intertidally along the eastern 1993). and north-western North Atlantic waters. The phylogenetically even more isolated Sarothrogammarus group is ecologically peculiar. Several Remarks species are transitory regarding salinity, as well as re- This Atlantic marine or brackish water species splits garding hypogean habitats. Several purely fresh- from the Gammaridae tree basally, and is the sister water species from Central Asian mountains are taxon to the rest of the family. The same phylogenetic regarded as ‘half subterranean’ (Stock, 1986). The Medi- position of this species was shown by MacDonald et al., terranean Rhipidogammarus and related species are 2005. slightly troglomorphic, however, but only exceptionally eyeless; they occur mainly in marginal subterranean habitats, such as interstitially in the coarse sand MOST GENERAL ECOLOGICAL DIFFERENTIATION and shingle of beaches (Stock, 1973) or in (brackish) WITHIN THE FAMILY GAMMARIDAE springs arising from coastal gravels (Sket, 1986). More Marine (halophile) species troglomorph are members of some supposedly related Most Gammarus species, Echinogammarus s. str. species, genera, not included in our analyses (Tyrrhenogammarus and sarothrogammarids are limited to springs and the and Longigammarus).

Table 2. Mean p-distance within different groups of Gammaridae calculated with MEGA 5.2

COI 28S 18S

Standard Standard Standard Groups/aggregates Estimate error Estimate error Estimate error

Gammarus lacustris 0.155 0.010 0.008 0.002 0.004 0.001 Gammarus balcanicus 0.193 0.010 0.030 0.003 0.007 0.001 European Gammarus 0.225 0.011 0.032 0.003 0.013 0.002 Oriental Gammarus 0.219 0.011 0.060 0.004 0.023 0.002 Baikal group 1 0.267 0.014 – – 0.005 0.002 Baikal group 2 0.200 0.018 – – 0.006 0.002 Perimarine Gammarus 0.224 0.011 0.077 0.004 0.046 0.003 Dinaric–continental Homoeogammarus – – 0.011 0.003 0.012 0.003 Ponto-Caspian group 0.187 0.010 0.047 0.003 0.015 0.002 Non-Dinaric Homoeogammarus 0.194 0.012 0.054 0.004 0.036 0.003 Echinogam s. str. & Typhlogammarus 0.236 0.013 0.043 0.004 0.043 0.003 Sarothrogammarus group 0.222 0.017 0.109 0.009 0.060 0.005 Out-group 0.272 0.011 0.307 0.007 0.209 0.006

Lacustrine gammarids ecologically and morphologically. It is astonishing that If we are discussing the natural situation, before the none of the endemic and diverse lacustrine faunas seems human-supported spreading of Ponto-Caspian species to originate from the aggregate G. lacustris. Both in the 20th century, Gammarus lacustris G.O. Sars, 1863 Baikalian mega-ﬂocks are comparatively distal inde- is one of very few non-endemic species inhabiting pendent branches of the freshwater Gammarus group. ephemeral (i.e. not ancient or long-lived) lakes. It has The Ponto-Caspian mega ﬂock is only weakly related a wider distribution than the rest of the entire genus, to Gammarus. The gammarid faunas in the compara- and even of the family, and is also present around the tively young and small lakes Ysyk Köl in Kyrgyzstan Himalayas and in north-western parts of North America, and Ohrid in south-east Europe are comparatively very which are both without other Gammaridae. modest. If we include related species in our consideration, The gammarid fauna of Ysyk Köl (Issyk Kul), a morphological and molecular differences are only slight- Kyrgyzian (central Asian) brackish lake at high el- ly augmented within this clade (aggregate G. lacustris), evation (1606 m a.s.l., depth 668 m; ion sum only close but p-distances within the whole aggregate are still to6gL−1; Baetov, 2006) has been poorly studied. much lower (Table 2) than within any other aggre- Endemic species include Issykogammarus hamatus gate, in spite of its exceptional distribution area and Chevreux, 1908, and probably also Gammarus bergi ecological diversity. Our samples have been taken from Martynov, 1930, Gammarus inberbus Karaman & between latitudes 35° and 50° N, and between longi- Pinkster, 1977, and Gammarus ocellatus Martynov, 1930 tudes 125°W and 120°E, from elevations of 70 to (Chevreux, 1908; Martynov, 1930; Karaman & Pinkster, 1700 m a.s.l., from lakes, but also from springs and 1977a), none of which could be studied molecularly at streams. The exceedingly wide distribution area of this time. Although I. hamatus shows some other pe- G. lacustris could be thanks to some ethological char- culiarities (Barnard & Barnard, 1983), its most strik- acteristics that enabled it to be transported by water ing character is the protective pointed coxal elongations. fowl (Segerstråle, 1954). These taxa evidently do not belong to the morphologi- Another mainly lacustrine species is Pallaseopsis cally easily recognizable G. balcanicus aggregate. quadrispinosa (G.O. Sars, 1867) (synonyma Pallasea q., Pallasiola q.), limited to Palaearctic glacial lakes in Ohrid northern Europe and north-west Asia, including Novaâ The endemic fauna of the Balkan lake Ohridsko ezero zemlâ (Novaya Zemlya; Segerstråle, 1958). The other (Lake Ohrid), as yet with no well-established age gammarid species occur in ephemeral lakes only ex- (Albrecht & Wilke, 2008), is richer. The Ohrid lacustrine ceptionally, although G. fossarum even developed an species together with those from the surrounding springs abyssal race in the Lake Bodensee (Lake Constance; constitute a very well-supported monophylum within G. Karaman, 1976). the aggregate G. balcanicus. Seven species have been Three or four long-lived lakes have been success- described (Karaman, 1977a; Karaman & Pinkster, 1987), fully invaded by Gammaridae, radiating strongly both and there are a few as yet undescribed but reliable

© 2015 The Linnean Society of London, Zoological Journal of the Linnean Society, 2016, 176, 323–348 338 Z. HOU AND B. SKET species (Hou et al., 2011, 2014; Wysocka et al., 2013). The diversity of evolutionary directions within Baikal None of them has the ‘Baikalian’ habitus, but as many amphipods has been discussed both morphologically as five of them are armed by spines (stout spiniform (Barnard & Barnard, 1983; Tahteev, 2000) and mo- setae) at the posterior borders of the pleonites (Karaman lecularly (Väinölä & Kamaltynov, 1999; MacDonald et al., & Pinkster, 1987). 2005). Therefore, we will only briefly review the whole These species are feebly specialized regarding depths morphological and ecological diversity to compare this of habitat: some are limited to the shallows and coastal with the relatively modest molecular diversity (Table 2). springs, whereas others inhabit layers between 20 and The Baikalian gammarids as a whole diverged to such >100 m in depth. No specializations have been ob- an extent that scholars are quite unanimous in served, and the species may occur in mixed popula- attributing/dividing them among several groups or fami- tions (Karaman & Pinkster, 1987; B. Sket, pers. observ.). lies (Kamaltynov, 1999a, 2001; Tahteev, 2000; Martin Some of the speciation proceeded on account of genome & Davis, 2001; Lowry & Myers, 2013). In our trees, mutations (Salemaa & Kamaltynov, 1994; Schön & the endemic Baikalian amphipods only split into two Martens, 2004). Although they are clearly discernible monophyletic clades, as is the case with MacDonald morphologically, their p-distances are extremely small, et al. (2005); these two clades are not fully in accord and represent just a small portion of diverging popu- with the morphologically based taxonomy. Both clades lations of G. balcanicus (as supposed by Karaman & are morphologically and ecologically highly different Pinkster, 1987; Birštejn, 1961). There is a similar situa- and diverse. tion with some gastropod genera in the same lake Both clades evidently (Hou et al., 2011, 2014) derive (Albrecht et al., 2006; Hauswald, Albrecht & Wilke, from ‘normal’ Gammarus species, which are of the pre- 2008). In the hirudinean genera Glossiphonia and Dina dictable, unified habitus, structure, and even colora- the Ohridan diversity in external morphology is hardly tion of brown (seldom with reddish patches). Regarding reached outside this lake (Sket, 1981, 1989), whereas ornamentation, members of the Baikal-2 clade are phylogenetic relationships are hardly resolved because mainly smooth (Micruropus, Crypturopus, and of the small p-distances (unpublished). In the isopod Gmelinoides) or weakly warty (Carinogammarus and genus Proasellus (Kilikowska et al., 2006; F. Malard, Pseudomicruropus); Kamaltynov (1999a) designates the pers. comm.) the situation is similar. For Ancylus, the entire family Micruropodidae as an ‘analogue of the split of the lacustrine flock from the closest relative Ponto-Caspian Pontogammaridae’. Evidently the only has been calculated to 1.4 ± 0.6 Mya. Wysocka et al. exception in this clade is the planktonic, mysidiform (2013) calculated the age of the Gammarus specia- Macrohectopus. Members of the Baikal-1 clade may be tion in Ohrid to 6±1Mya;according to our crude cal- smooth to most diversely armed, either with dorsal culation, that time point would be less than 4 Mya. spiniform setae or by points and bulges (like Acanthogammarus or Brandtia) or projections on virtually all parts of the trunk. In their basic trunk shape, Baikal the Baikal gammarids may be Gammarus-like (like The amphipod fauna of the ‘long-lived’ lake Baikal Eulimnogammarus), or with very elongate append- (Bajkal, Ozero Bajkal, Bajkalskoe ozero; Kamaltynov, ages (Abyssogammarus), or stout and probably able to 1999a, 2001; Tahteev, 2000) evidently contains just volvate (Gammarosphaera); they may be dwarfed (like Gammaridae and its derivatives. This is by far the Micruropus pusillus Bazikalova, 1962, reaching up to largest local quota of 344 morphologically defined 2 mm in length) or gigantic [like Acanthogammarus species, for which 70 genera have been erected grewingkii (Dybowski, 1974), reaching up to 90 mm in (Kamaltynov, 1999a, 2001). Considering high molecu- length]. lar differences within some supposed species (Väinölä, Regarding coloration, members of the Baikal-2 group Kontula & Kamaltynov, 2000), Kamaltynov (2001) even are, with just a few exceptions [such as the striped calculates that the real number of species could be Gmelinoides fasciatus (Stebbing, 1899)], white or ∼2000. According to our analysis these species stem colorless (V. Tahteev, pers. comm.) fossorial or from at least two immigration waves, as evidenced by benthopelagic to planktic animals. On the other hand, two well-supported clades (Fig. 2). This is already dis- the Baikal-1 members – if not inhabitants of great cernible in the phylogenetic trees of MacDonald et al. depths – may be very dark [like Eulimnogammarus (2005), and was confirmed by Hou et al. (2011, 2014: cruentus (Dorogostaisky, 1930)], parti-coloured figs 2, S1) and Sherbakov et al. (1999: fig. 2). The set [Eulimnogammarus maacki (Gerstfeldt, 1858)], even of species taken in these analyses is, however, very with diversely coloured reflecting structures small, but represents the whole morphological diver- [Dorogostaiskia parasitica (Dybowsky, 1874) sity of the fauna. Thus, it is not very likely that the (= Spinacanthus sp.); own data, otherwise just noted number of clades is higher. Both immigration waves as ‘golden’ or ‘blue spots’; Daneliya & Väinölä, faced very different destinies. 2013].

Similarly diverse is the ecology of Baikalian the Paratethys Sea) and its confluents, which went gammarids; compared with the morphology, the eco- through a number of permutations regarding the shape logical specialization has been presented in more in (hydrography) as well as the physics and chemistry detail by Kamaltynov (1999b) and Tahteev (2000). They of the water (hydrology) (Cristescu, Hebert & Onciu, may be specialized to different depths from the litto- 2003). Its extant, isolated remains are the Caspian ‘Sea’, ral to the abyssal, and may be burrowing, epibenthic, Aral Lake, Black and Azov seas, and their coastal bentho-pelagic, and even fully planktic, which result- waters, limans (lagoons), and estuaries. A number of ed in the mysid-like habitus in Macrohectopus branickii these species are rheophiles, penetrating upstream along (Dybowsky, 1874). Another subject of specialization may mighty rivers. be feeding, which resulted in specialists such as the Besides a few genera of obviously different ample genus Pachyschesis (Takhteyev & Mekhanikova, phylogenetic position (such as Corophium, Onisimus, 1993; Tahteev, 2000) feeding on eggs in marsupia of Monoporeia, and Gammaracanthus), the gammarid larger species, or living mainly as ‘raumparasites’ on nature is only questionable for a few others. Our set large sponge colonies [such as Dorogostaiskia parasitica of species covers the morphological diversity of the (Dybowsky, 1874) and Eulimnogammarus violaceus Ponto-Caspian gammarid group to such an extent that (Dybowsky, 1874); Mehanikova, 2001]. we can confirm its monophyly as well as its deep in- The genetic distances in genes not morphologically clusion in the tree of Gammaridae. Phylogenetically, expressed but used for phylogenetic estimations (here Ponto-Caspian gammarids are members of the south- COI, 28S, and 18S) between species, genera, and ‘fami- ern gammarid branch. According to Barnard & Barnard lies’ of both clades are of the same order of magni- (1983) the ‘Ponto-Caspian Gammaroids comprise 33 tude as the distances between the species of the oriental genera and 77 species’, which already shows the high Gammarus aggregate, which is in a sister relation- morphological diversity of this fauna. ship with them, but which is morphologically explic- For the entire Ponto-Caspian, Grabowski (2014) itly uniform. On the other hand, their p-distances are already listed 84 species and 33 genera, and distrib- slightly longer than between the Ponto-Caspian taxa. uted them into four families: Behningiellidae Just a few Baikalian species succeeded in settling Kamaltynov, 2002; Caspicolidae Birštejn, 1945; in the rivers and small lakes downstream of Baikal, Gammaridae Latreille, 1802; and Pontogammaridae as far as the surroundings of the Jenisej (Yenisey) Bousfield, 1977. Barnard & Barnard (1983) distribut- estuary, and exceptionally upstream in the river Selenga ed Ponto-Caspian species into six ‘groups’: (Kamaltynov, 2001; Takhteev, Berezina & Sidorov, 2014). Echinogammarus, Gmelina, Dikerogammarus, Gmelinoides fasciatus seems to be one of the few Cardiophilus, Pontogammarus, and Compactogammarus Baikalian amphipods that also periodically conquers group. In the hydrographically isolated Caspian, 70 the warmer waters outside Baikal. It has been arti- species are members of Gammaridae (including ficially introduced outside Baikal, but is also spread- Pontogammaridae), 15 are non-Gammaridae, and four ing actively. This was explained by a different quality belong to families of unknown affiliation (Behningiellidae of its heat-shock proteins (Protopopova et al., 2011). and Caspicolidae) (Grabowski, 2014). In the Black Sea, Less extensive is the spreading of Micruropus possolskii which is now open to the Mediterranean, the number Sowinsky, 1915 and Micruropus wohlii (Dybovski, 1874) of non-endemic species is much higher (Barbashova, Malavin & Kurashov, 2013). All emi- (Morduhaj-Boltovskoj, Greze & Vasilenko, 1996). The grants are smooth-bodied members of the Baikal-2 clade. same system was also accepted by Lowry & Myers A very different successful earlier emigrant/invader (2013). The separation of ‘Pontogammaridae’ from the might have been Pallaseopsis quadrispinosa, which rest (‘gammarid’) of the Ponto-Caspian genera results spread outside Baikal in colder Pleistocene periods and in the paraphyly of Gammaridae, which is evident from remained more or less limited to relictual glacial lakes both MacDonald et al. (2005) and our tree (Hou et al., between the Sibirian Lena estuary and northern Europe 2014; this study). Neither Behningiellidae nor (see above). Its Baikalian origin has been proven mor- Caspicolidae was included in any molecular study. phologically as well as molecularly by Väinölä & According to Barnard & Barnard (1983), most species Kamaltynov (1995) and Väinölä et al. (2000), al- are primarily bound to benthic habitats and niches, though Thienemann (1950: 495–498) and some other burrowing into the substratum, but easily leaving it. early authors (see Tahteev, 2000: 53) even discussed With the exception of Derzhavinella, Trichogammarus, the possibility of its marine origin. and the richer genus Chaetogammarus, all Ponto- Caspian gammarids exhibit distoposteriorly lobate basis Ponto-Caspian basin (article 2) of pereopod VII. It is worth noting that this Gammarids of the mainly brackish Ponto-Caspian basin otherwise rare character in gammarids is present also are evidently followers of a previously more unified in the closely related Jugogammarus kusceri fauna of an ancient relictual lake (the remainder of (S. Karaman, 1931), inhabiting mosses (Cinclidotus)in

© 2015 The Linnean Society of London, Zoological Journal of the Linnean Society, 2016, 176, 323–348 340 Z. HOU AND B. SKET rheocrene springs in Slovenia; thus, this character is (Takhteev et al., 2014). Chaetogammarus ischnus must not linked with fossorial habits, as supposed by some have invaded the Pontic rivers already in Pleistocene authors (Barnard & Barnard, 1983). It is remarkable (Cristescu et al., 2004). The presence of some species that no Ponto-Caspian species is specialized to be in Aegean tributaries [Turcogammarus spandli pelagic, and none are specialized for the depths of the (S. Karaman, 1931); Hou et al., 2014] can be ex- lake (Zenkevicˇ, 1947), even though the Caspian lake plained by ‘influxes to the Mediterranean Sea’ during is up to 1025 m in depth. the Messinian Salinity Crisis, which were document- Otherwise, the morphology of these animals is very ed by fossil dinoflagellates (Popescu et al., 2009). diverse, either Gammarus-like (like Chaetogammarus The size and history of this lacustrine system, par- and Dikerogammarus) or very stout, supposedly(!) ticularly the great changes in its shape, extent, and fossorial (Barnard & Barnard, 1983), with a broad trunk salinity (Zenkevicˇ, 1947; Starobogatov, 1994), trig- and shortened antennae I (Niphargoides spp.). Some gered a rich ecological diversification of the primary species are armed with pointed projections, similar to inhabitants, followed by their morphological diversi- Baikal gammarids. Unfortunately, there are only scarce fication and speciation. The changes in salinity were precise data about the ecology and behaviour of Ponto- probably also responsible for a remarkable euryhalinity Caspian amphipods. We know that Dikerogammarus of Ponto-Caspian taxa, which allowed their recent pen- may be a strong predator (Dick, Platvoet & Kelly, 2002; etration into rivers and, by the human intervention, Kley & Maier, 2003), whereas some others, like invasion into new regions. An additional factor seems Pontogammarus species, can be morphologically iden- to be the increased ion content in rivers caused by pol- tified as filterers (Barnard & Barnard, 1983). None of lution (Jazdzewski & Konopacka, 2002). In recent the existing parasitic species (e.g. Cardiophilus spp.) decades Ponto-Caspian species have aggressively re- could have been studied molecularly. placed native Gammarus species in streams and lakes In spite of the great morphological diversity, in Central Europe around the Baltic Sea, as well as p-distances within the clade of Ponto-Caspian in North America. Along with euryhalinity and a lower species (and genera!) are moderate, and no higher than sensitivity to some pollutants, these advances are also between species of the related genus Homoeogammarus supported by high fecundity and predatory behav- (Table 2). iour, and/or by the different feeding habits of these In contrast to the Baikal amphipods, a number of species not exploited by native species. Note that many Ponto-Caspian animals succeeded in conquering lakes other members of the Ponto-Caspian fauna (Amphipoda and non-lacustrine waters outside their original area Corophiidae, Mysida, Bivalvia, Pisces Gobiidae) are simi- (Jaz˙dz˙ewski, Konopacka & Grabowski, 2005; Gallardo larly successful, some even in America (Stepien & & Aldridge, 2013). They are even successful in Tumeo, 2006). outcompeting native amphipods over much of Europe and in North America (Cristescu et al., 2004). This can only be explained by past adaptations to highly fluc- MOST OBVIOUS MORPHOLOGICAL DIFFERENTIATION tuating salinities in their palaeo lakes. Moreover, the Gammarus is generally presented as a ‘basic ‘killer shrimp’ Dikerogammarus villosus (Sowinsky, 1894) gammaridean’ (Barnard, 1969; Dahl, 1977). As all non- exhibits particular ethological and physiological prop- gammaridean amphipod suborders (Caprellidea, erties that facilitates its spread by manmade means Hyperiidea, and Ingolfiellidea) are highly specialized (Bacela-Spychalska et al., 2013), somehow analogous ecologically, ethologically, and morphologically, with the natural spread of Gammarus lacustris via birds Gammarus can also be accepted as a standardized and (Segerstråle, 1954). the most plesiomorphic amphipod type. With only small In the middle of the 20th century, Thienemann (1950) aberrations in the body shape, such animals occur across could only report the invasion of Chaetogammarus all of the Gammaridae area, differing in details of ischnus [and the non-gammarid amphipod chaetotaxy and slightly differing in some body pro- Helicorophium curvispinum (Sars, 1895)] via Black Sea portions. All species inhabiting springs, streams, young affluents and artificial channels up to the Ostsee drain- lakes, and the sea, are virtually of the same habitus. age system and 1752 km upstream in the Danube (at The only remarkable deviations of the ‘basic Budapest). Chaetogammarus ischnus reached the Polish gammaridean’ type Gammaridae may be seen in river Wisła (Vistula) at the latest by 1928 (Jaz˙dz˙ewski, lacustrine mega-flocks in both the Baikal and the Ponto- 1980); in 1995 the same species was already estab- Caspian. Only the following taxonomically important lished in the Laurentian Great Lakes (Witt, Hebert characters will be elaborated upon here: develop- & Morton, 1997), whereas presently at least six Ponto- ment of uropod III and the protective armament of Caspian gammarid species are spread all over Europe the trunk (‘ornamentation’). The reduction of uropod (Gallardo & Aldridge, 2013), and an even higher number I was already discussed by Karaman (1977b), and of species inhabit the rivers of southern European Russia we can only corroborate molecularly the polyphyletic

© 2015 The Linnean Society of London, Zoological Journal of the Linnean Society, 2016, 176, 323–348 PHYLOGENETICS OF GAMMARIDAE 341 occurrence of its partial reduction (which excludes (Kutschera et al., 2012) clearly shows its apomorphic, Neogammarus from sarothrogammarids; Hou et al., derived origin. The magniramus type of the large genus 2014: fig. S1). Gammarus is thus a morphological relict. The details of the magniramus type uropod in some others, e.g. Uropod III Euxinia, even suggests an occasional secondary elon- The development of uropod III is, or used to be, one gation of the (previously scale-like) endopodite. Thus of the taxonomically most remarkable characters in we may agree with Barnard & Barnard (1983: 118) Gammaridae; the two states of this character were that ‘one must suppose it [the endopodite reduction] crucial for distinguishing between the major and con- has a selective advantage but the function is unknown’. stant members of ‘Gammaridae’: Gammarus and This is also supported by well-developed endopodites Echinogammarus (e.g. Barnard & Barnard, 1983; in the Sarmatian fossil amphipods (Deržavin, 1927, Karaman, 1993). Functionally, both uropods III form 1941). On the other hand, the definite structure of the tail fan at the posterior end of the trunk. Kutschera, uropod III, namely its length and setosity, is in most Maas & Waloszek (2012) characterized the amphipod species only finalized in adult males, not even in (gammarid) uropod III as ‘rod-shaped’, which is incor- females, let alone in juvenile, non-mature specimens. rect. Both rami of this appendage in gammarids are This may raise doubts about the biological impor- flat and narrowly leaf-shaped, whereas in many species tance of the appendage (but not about its endopodite a border of long and densely set marginal setae may reduction). even functionally transform both uropods into a broad fan. Some authors (Schellenberg, 1942; G. Mayer, Protective projections pers. comm.) characterize the appendage as a ‘rudder’ Dorsal spines (stout setae) on urosomites and on when swimming or to ‘enable the animals lying sta- pleonites seem at first sight to be a protective struc- tionary on the ground [at a] permanent water current ture. But the distribution of this character (state) among produced by the pleopods’ (G. Mayer, pers. comm.) Ac- gammarids gives no support for such an idea. It is dif- cording to Platvoet et al. (2009) it is also included in ficult to estimate the biological meaning of dorsal spines the filtration process for feeding. (spiniform setae) occurring on the urosomites of the The ‘typical’ uropod III of most Gammarus spp. is majority of Gammarus and Homoeogammarus spp. probably plesiomorphic for Gammaridae (or even for Much less frequent is the presence of similar struc- Amphipoda). Its biarticulate exopodite is several times tures on the pleonites. They only occur in most longer than sympodite, with a rudimentary apical article; Echinogammarus s. str. species, and in some lacustrine proximal article flat, with groups of short spines and Gammarus species (in Ohrid) and Eulimnogammarus longer setae along both edges; endopodite similarly species (in Baikal). Pleonital spines are only excep- shaped and setose as exopodite, but moderately shorter tionally present in a non-lacustrine Gammarus (e.g. and uniarticulated. This uropod type was named Gammarus accolae G. Karaman, 1973), and are evi- ‘magniramus’ type by Schellenberg (1937a). The dently absent in Homoeogammarus. Such a distribu- interspecies (and to some extent also intraspecies) vari- tion of the character could hardly testify for its protective ability of the ‘magniramus’ type consisted of the vari- function. able shortening of the endopodite and the lengths and Two basic types of protective projections can be rec- density of the marginal setae. Although transitional ognized in Gammaridae: dorsal pointed projections oc- shapes do exist, the so-called ‘parviramus’ type can curring occasionally in all habitat types, and lateral usually be clearly distinguished. Its main character- pointed or blunt projections occurring exclusively in istic is the diminution of the endopodite into a scale- ancient lakes. The most generally developed arma- like rudiment with only vestigial, mainly apical spinosity ment are three or four posteriorly directed median dorsal and setosity. Schellenberg designated the phyletic projections (points or protuberances) formed at the pos- branches in which both types and transitory forms of terior edges of tergites. They most often appear on uropod III occur as ‘variiramus’ branches (‘variiramus- pleonites I–III, less commonly on some posterior Zweig’), such as described in Marinogammarus species. pereonites. Dorsal median points occur in different Unfortunately, Schellenberg was not clear in his phyletic lines and different types of habitat: within fresh- ‘variiramus-Zweig’ definition: should it mean individ- water Gammarus in Gammarus roeselii Gervais, 1835 ual variability or differently developed species within and its relatives; in Gammarus stojicevici (S. Karaman, a phylogenetic branch? 1929) (Gammarus balcanicus aggregate); in marine A glimpse of the phylogenetic tree of Gammaridae Gammarus mucronatus (Gammarus fasciatus aggre- may give the impression that the parviramus type of gate); in some unrelated Ponto-Caspian groups uropod III is the plesiomorphic, ancestral character state, Amathillina spp., Dikerogammarus caspius (Pallas, as it is common to all basally split clades of the family. 1771), Turcogammarus spandli,andGmelina spp.; and But an out-group analysis within Malacostraca in the Mediterranean ‘Echinogammarus’ apfelbecki

(S. Karaman, 1931), ‘Echinogammarus’ annandalei following the regional rifting, by the retreat of Tethys (Monod, 1924). The same type of armament is also from the eastern Pamir area, causing the split within present in species of other families (see Sars, 1895). the sarothrogammarid subclade, and by the The homoplastic nature of this character, occurring in hydrographically supported split within the such diverse lineages, is out of question. The number Gammaracanthus subclade. Unfortunately, this sce- of median or paired protuberances along the pereon nario does not accord with the evident total absence may be higher in lacustrine gammarids in the Caspian of any Gammaridae at the coasts of the Indian Ocean. [Gmelina kusnezowi (Sowinsky, 1904)] and in Baikal Therefore, we searched for another solution (Figs 3, S1; [Axelboeckia, Acanthogammarus spp., Eucarino- Table 1). gammarus wagii (Dybowski, 1874)], as well as in nu- For calibrating the gammarid radiations we consid- merous marine amphipods (Gammaracanthus and er the most reliable scenario (Figs 3, S1A), based on Amathilla). When compared with protective struc- three calibration points. The last faunal exchanges tures in other animal groups (Cladocera, Rotatoria; between the eastern and western Atlantic 20–15 Mya Caridea, Jugovic et al., 2010) it is quite certain that (Harzhauser, Piller & Steininger, 2002) should ap- such points serve for protection against predatory ver- proximately coincide with the split between the locusta– tebrates. A support to such a hypothesis is the situa- zaddachi and fasciatus groups (point ‘r’) of Gammarus. tion in Metohia carinata Absolon, 1927 found in the The final disruption of aquatic connections of the cave Obodska pecína, where the predator (Proteus)is Dinaric–Caucasian lands towards the West c.19Mya absent and the points in the local population of the could have caused the split between the Dinaric– gammarid Metohia are reduced. We observed a similar Caucasian Typhlogammarus group and the peri- situation in the rostrum of the cave shrimp Troglocaris, Atlantic Echinogammarus s. str. (point ‘h’). The start the length of which is distinctly smaller in the absence of the palaeontologically stated bivalvian radiation in of Proteus (Jugovic et al., 2010). In neither case could the Paratethys was certainly in accord with the start we establish whether the armament prolongation is of radiation of the Ponto-Caspian amphipods (point ‘e’); phenotypically induced (by the predator’s exudates) or this is supported by the existence of a few fossil genotypically fixed by selection. It is only obvious that gammarids of the ‘Ponto-Caspian type’ (Deržavin, 1927, some gammarid taxa do not possess alleles for the de- 1941). In addition to the events already used, some velopment of such points. In a large limnocrene spring others are approximately congruent with this sce- in eastern Serbia, the well-armed Gammarus stojicevici nario. Such a scenario brings the progressing radia- (S. Karaman, 1929) occurs in a mixed population with tion of the peri-Mediterranean Homoeogammarus close the related, totally smooth-necked Gammarus cf. to the estimated time of the Salinity Crisis (c. 6.0– balcanicus (B. Sket, pers. observ.); both are present in 5.5 Mya), when extensive water level oscillations enormous density. (Gargani & Rigollet, 2007) could have caused numer- Only in some old lacustrine lineages do differently ous isolation events along the coastal belt. The cooling positioned, mainly lateral projections occur. Evident- and oxygenation of the Baikal depths at 3 Mya (Mats, ly the most common are paired lateral points. They 2013) and the progressive differentiation of both Baikal may occur in one pair or more, in a great diversity in groups approximately coincide. Baikal gammarids. These may be lateral protuber- We are placing great weight on the material palae- ances of tergites, as in Acanthogammarus godlewskii ontological data, which also reliably support the pos- (Dybowsky, 1874) from Baikal or the Caspian sibility of extensive evolutionary changes in ancient Axelboeckia spinosa (G.O. Sars, 1894), or elongations lake faunas. The scenario leaves seemingly little time of coxal plates, as in Issykogammarus hamatus for the differentiation of exceedingly diverse gammarids Chevreux, 1908 from Ysyk Köl (Barnard & Barnard, in ancient lakes (Baikal and Ponto-Caspian). But the 1983). The homoplastic nature of this character is em- relative endemicity of some well-dated gastropod faunas phasized by similar structures in some Hyalella spp. suggests that such a development of a certain fauna (Hyalellidae) in Titicaca (Stebbing, 1906) and in Fuxiana was definitely possible (Harzhauser et al., 2002). The yangi Sket, 2000 (probably Anisogammaridae) in Fuxian few fossil Sarmatian amphipods and the rich assem- Hu (Sket, 2000), which are again ancient lakes. blages of fossil bivalvians both show the possibility of such rapid radiations and morphological elaboration. Fossil amphipods have been found within the Ponto- TIMING AND GEOGRAPHY OF THE GAMMARID Caspian area, and have been morphologically de- EVOLUTION scribed in some detail (Deržavin, 1927, 1941). Although According to Hou et al., 2014, the first splits in the with differing (magniramous) uropod III and non- already existing gammarid clade should have hap- lobate pereopod bases, they already show some char- pened more than 70 Mya. Such a scenario is support- acteristics of the recent Ponto-Caspian amphipods. Three ed by the supposed age of the Baikalian amphipod fauna Andrussovia species are characterized mainly by body

(appendages) proportions, and together with two But, the biogeographically related cardiids started one Praegmelina species also by diverse lateral tergal pro- of their radiations in the same body of water much tuberances (never present in non-lacustrine species, see earlier, in the Sarmatian, at 12 Mya or earlier. The above). They are of an Upper Sarmatian age, which cardiid fauna should have been extinguished there tem- means that at least 12 Mya the Paratethyan amphipods porarily or rarefied by the freshening of the lake were already diversified. The most promising data for (Magyar, 2006), which was not necessarily the destiny timing the amphipod differentiation are from endemic, of euryhaline amphipods. The fossil Ponto-Caspian type broadly Paratethyan Bivalvia, a group with a trace- amphipods in Sarmatian (c. 10 Mya) sediments able and fairly well-studied history. Dreissenidae and (Deržavin, 1927, 1941) may be direct ancestors of the Cardiidae: Lymnocardiinae are similarly richly differ- extant Ponto-Caspians. In such a case, their first ra- entiated in the Paratethyan area from the Neogene diation could be synchronous with one of the first ra- to recent times. Evolutionarily the most similar to our diations of lymnocardiines. This occurred in the great amphipods is the group Adacnini. Both the Ponto- Sarmatian Sea, continuous from the Aral to the extant Caspian type gammarids and the cardiid Adacnini are north-western Dinaric area. still present in the Caspian Lake, and in estuaries and Besides the fossil data, the most convincing timer limans around the Black Sea (Nevesskaja, Paramonova might be the relationship between the proto- & Popov, 2001; Popa et al., 2009). Moreover, both groups Mediterranean and the Indo-Pacific areas. This would have (or had, before its artificial, anthropogene, de- position the origin of Gammaridae to 40 Mya or earlier struction) several species in the Aral ‘Sea’. These are in the Atlantic, their spread into the proto- the fossil lymnocardiines Monodacna sp. and recent Mediterranean and Asian continental seas, including Adacna sp. (according to Nevesskaja et al., 2001; Turgaj and their affluents, between 40 and 20 Mya; Zenkevicˇ, 1947), and the amphipod Dikerogammarus all these waters were contiguous in those times aralensis (Ul’janin, 1875) (Birštejn & Romanova, 1968). (Meulenkamp & Sissingh, 2003). Since that time If the time since Apscheronian (i.e. ∼2 Myr) might seem (Scotese, 1998), and more pronounced since the Early much too short for such a rich speciation of amphipods, Miocene (c. 20 Mya), the proto-Mediterranean was con- the radiation of Adacnini during that stretch of time nected with the Indo-Pacific by south-east directed has been materially proven. More than 40 recognized shallow straits only (Popov et al., 2004). They could species and ten genera developed in that time period, have been easily overheated and with raised salini- which is a remarkable morphological diversity for such ties, which is both adverse for members of this mainly a morphologically poor group. One has to consider that oligotherm and oligohaline group. Note that the tem- the morphological diversity of Cardiidae is a priori very peratures (Wolfe & Poore, 1982) were much higher than limited by a low number of morphological characters, today. compared with gammarids. We must wait for some Thus, it is possible that Gammaridae are an Atlantic– molecular analyses to be able to compare the more Mediterranean derivative of a common ancestor with equivalent molecular diversities within the extant Ponto- the Atlantic Haustoriidae and Gammaracanthidae, Caspian gammarids and Adacnini. The family which was not yet morphologically specialized. Dreissenidae is comparable: within Lake Pannon alone Some conservative high-level lineages, such as this family developed into 65 species belonging to five Relictogammarus stoerensis and Marinogammarus, re- generally accepted genera of particularly high (for mained restricted to the eastern Atlantic. The others malacological criteria) morphological and ecological di- conquered its freshwater confluents and remained cap- versity. This happened over some 5 Mya. Within the tured there, like Parhomoeogammarus and initial phase (11.5–10.0 Mya) alone the first few Echinogammarus s. str. (‘berilloni group’). Finally, some Mytilopsis species radiated into 26 species attributed penetrated through the proto-Mediterranean to the lands to three genera (Harzhauser & Mandic, 2010). At the at its eastern edges, and remained either captured in same time (1.5 Mya) an unknown number of gastro- its freshwater confluents, like the Typhlogammarus pod species (the fauna is not monophyletic) radiated group and a part of the Sarothrogammarus group, or into 150 species attributed to 34 families (Harzhauser also spread along the future Mediterranean coasts, like & Mandic, 2008); even if the starting number of species Rhipidogammarus. was remarkable, these radiations along with their speed Another early scenario, more congruent with what are comparable with the here supposed history of the follows, would be the development of the ancestral Ponto-Caspian Gammaridae. A number of cyprinid gammarid stock in the late Eocene, 37–34 Mya, in the genera in southern Europe (Perea et al., 2010) also suc- North European Region (sensu Harzhauser et al., 2002), ceeded to radiate into flocks of species in such a time. leaving some conservative relatives in the Atlantic. Any So, an alternative hypothesis would be that the time earlier presence of Gammaridae in the Mediterra- of the main radiation of ‘Ponto-Caspians’ could have nean would allow and also force the spread into the started even as late as in the Apscheronian, 2 Mya. Indo-Pacific.

It would be highly speculative to discuss the order Albrecht C, Wilke T. 2008. Lake Ohrid: biodiversity and evo- and timing of the splits between the Palaearctic lution. Hydrobiologia 615: 103–140. subclades of Gammarus, both for a weak support of Bacela-Spychalska K, Grabowski M, Rewicz T, Konopacka their mutual relationsships as well as for a poor knowl- A, Wattier R. 2013. The ‘killer shrimp’ Dikerogammarus edge of the appropriate palaeogeographic develop- villosus (Crustacea, Amphipoda) invading Alpine lakes: over- ment. We also do not have an exact idea about the land transport by recreational boats and scuba-diving gear age of the endemic Baikalian fauna. According to recent as potential entry vectors? Aquatic Conservation: Marine and data (Mats, Shcherbakov & Efimova, 2011), the shallow Freshwater Ecosystems 23: 606–618. Baetov R. 2006. Lake Issyk-Kul: experiences and lessons learned or modestly deep lake basins have been available since brief. Managing lakes and their basins for sustainable use: 70 Mya, and after 30 Mya they already reached some a report for lake basin managers and stakeholders. Kuatsu: hundreds of metres. Between 74 and 34 Mya the International Lake Enviroment Committee Foundation, 193– amphipods (and gastropods) of that area were sup- 204. ISBN 4-9901546-2-2. posed to radiate and colonize the lake at 30–28 Mya. Barbashova MA, Malavin SA, Kurashov EA. 2013. First Unfortunately, the fossil-supported history of Mollusca finding of Baikalian amphipod Micruropus possolskii Sowinsky, (Gastropoda) in lake Baikal is very scant (Mats, 2013; 1915 (Amphipoda, Crustacea) in Lake Ladoga. Russian Journal V. Mats, pers. comm.). of Biological Invasions 4: 219–224. We can explain the wide distribution of Gammarus Barnard JL. 1969. The families and genera of marine lacustris by the fact that this species is mainly lacustrine gammaridean Amphipoda. United States National Museum, and able to disperse carried by birds in very recent Bulletin 271: 1–535. times. It is easily possible that the same ability is, or Barnard JL, Barnard CM. 1983. Freshwater Amphipoda of used to be, familiar to its relatives, whereas it is defi- the world. Mt. Vernon, VA: Hayfield Associates. nitely not familiar to Gammarus pulex (Segerstråle, Birštejn (Birstein) JA. 1961. Pešcˇernye bokoplavy Kryma. 1954). The only possible timing events are the highest Bulleten Moskovskogo obšcˇestva ispitatelej prirody, Biologiâ glaciations, which left the Scandinavian localities of 66: 126–144. Gammarus lacustris and Gammarus pulex in our study Birštejn JA, Romanova NN. 1968. Otrâd bokoplavy. inhabitable only ∼15 000 years ago, as already noted Amphipoda. Atlas bezpozvonocˇnyh Kaspijskogo morâ. Moskva: by Vainio & Väinölä (2003) for Gammarus lacustris. Izdatel’stvo ‘Pišcˇevaâ promyšlennost, 1–416. This is the highest possible time difference between Bousfield EL. 1977. A new look at the systematics of the Scandinavian and central European populations gammaroidean amphipods of the world. Crustaceana Suppl. of both species. 4: 282–316. But, the above described scenarios can hardly explain Bousfield EL. 1982. Malacostraca, Peracarida, Amphipoda. In: Parker SP, ed. Synopsis and classification of living organ- the split between the peri-Mediterranean and central isms. New York: McGraw-Hill Book Company, 232–293. Asian sarothrogammarids, which falls to c. 17 Mya, Bousfield EL. 1983. An updated phyletic classification and whereas the sea retreated from Central Asia already palaeohistory of the Amphipoda. In: Schram FR, ed. Crus- at 34 Mya or earlier (Popov et al., 2004). With this case tacean issues 1. Crustacean phylogeny. Rotterdam: A. A. we wanted to show how difficult and speculative the Balkema Press, 257–277. estimation of even the sequence of phylogenetic events Browne WE, Haddock SHD, Martindale MQ. 2007. is, let alone the absolute timing of the process. Phylogenetic analysis of lineage relationships among hyperiid amphipods as revealed by examination of the mitochondrial gene, cytochrome oxidase I (COI). Integrative and Compara- ACKNOWLEDGEMENTS tive Biology 47: 815–830. We are indebted to numerous colleagues who contrib- Carlini DB, Manning J, Sullivan PG. 2009. Molecular genetic uted to this work by providing samples of gammarids; variation and population structure in morphologically dif- they are listed in Hou et al., 2014: table S1. Some of ferentiated cave and surface populations of the freshwater the study was financed by the programme P1-0184 of amphipod Gammarus minus. Molecular Ecology 18: 1932– the Slovenian Research Agency, and some by the Na- 1945. Chevreux E. 1908. II. Crustaces Amphipodes. In: Pedaschenko tional Natural Science Foundation of China (NSFC- DD, ed. Études sur la faune du Turkestan. 91–110, tab 5–6. 31172049, 31372156, 31422048). Chevreux E, Fage L. 1925. 9. Amphipodes. In: De Beauchamp P, ed. Faune de France. Paris: Paul Chevalier, 1–488. REFERENCES Costa FO, Henzler CM, Lunt DH, Whiteley NM, Rock J. 2009. Probing marine Gammarus (Amphipoda) taxonomy with Albrecht C, Trajanovski S, Kuhn K, Streit B, Wilke T. 2006. DNA barcodes. Systematics and Biodiversity 7: 365–379. Rapid evolution of an ancient species flock: freshwater limpets Crisp MD, Cook LG. 2005. Do early branching lineages signify (Gastropoda: Ancylidae) in nthe Balkan Lake Ohrid. Organ- ancestral traits? Trends in Ecology & Evolution 20: 122– isms, Diversity & Evolution 6: 294–307. 128.

Cristescu MEA, Hebert PDN, Onciu TM. 2003. umbrella of Caspian Sea Environment Program. Available Phylogeography of Ponto-Caspian crustaceans: a benthic– at: http://www.zin.ru/projects/caspdiv/caspian_amphipods.html planktonic comparison. Molecular Ecology 12: 985–996. Harzhauser M, Mandic O. 2008. Neogene lake systems of Cristescu MEA, Witt JDS, Grigorovich IA, Hebert PDN, Central and South-Eastern Europe, Faunal diversity, gra- MacIsaac HJ. 2004. Dispersal of the Ponto-Caspian am- dients and interrelations. Palaeogeography, Palaeoclimatology, phipod Echinogammarus ischnus: invasion waves from the Palaeoecology 260: 417–434. Pleistocene to the present. Heredity 92: 197–203. Harzhauser M, Mandic O. 2010. Neogene dreissenids in Culver DC, Kane TC, Fong DW. 1995. Adaptation and natural Central Europe, evolutionary shifts and diversity changes. selection in caves: the evolution of Gammarus minus. Cam- In: van der Velde G, Rajagopal S, bij de Vaate A, eds. The bridge: Harvard University Press. zebra mussel in Europe. Leiden: Backhuys Publishers, and Cvetkova NL. 1975. Pribrežnye gammaridy severnyh i Weikersheim: Margraf Publishers, 11–28. dalnevostocˇnyh morej SSSR i sopredel’nyh vod. Leningrad: Harzhauser M, Piller WE, Steininger FF. 2002. Circum- Izdatel’stvo ‘Nauka’. Mediterranean Oligo-Miocene biogeographic evolution – the Dahl E. 1977. The amphipod functional model and its bearing gastropods’ point of view. Palaeogeography, Palaeoclimatology, upon systematics and phylogeny. Zoologica Scripta 6: 221– Palaeoecology 183: 103–133. 228. Hauswald AK, Albrecht C, Wilke T. 2008. Testing two con- Daneliya ME, Väinölä R. 2013. Five subspecies of the trasting evolutionary patterns in ancient lakes, species flock Dorogostaiskia parasitica complex (Dybowsky) (Crustacea: versus species scatter in valvatid gastropods of Lake Ohrid. Amphipoda: Acanthogammaridae), epibionts of sponges in Lake Hydrobiologia 615: 169–179. Baikal. Hydrobiologia 739: 95–117. Hillis DM, Mable BK, Larson A, Davis SK, Zimmer EA. Davey AJH, Turner GF, Hawkins SJ, Doncaster CP. 2006. 1996. Nucleic Acids IV: sequencing and cloing. In: Hillis DM, Mechanisms of density dependence in stream fish: exploi- Mortz C, Mable BC, eds. Molecular systematics. Sunder- tation competition for food reduces growth of adult Euro- land, MA: Sinauer Associates, 321–381. pean bullheads (Cottus gobio). Canadian Journal of Fisheries Hou Z, Sket B, Fišer C, Li S. 2011. Eocene habitat shift from and Aquatic Sciences 63: 597–606. saline to freshwater promoted Tethyan amphipod diversifi- Deržavin (Derjavin) AN. 1927. K poznaniû verhnesarmatskih cation. Proceedings of the National Academy of Sciences of amfipod Pontokaspijskoj oblasti. Bûlleten’ Moskovskogo the United States of America 108: 14533–14538. Obšcˇestva ispytatelej prirody 35: 183–196. Hou Z, Sket B, Li S. 2014. Phylogenetic analyses of Gammaridae Deržavin AN. 1941. Iskopaemye bokoplavy El’dara. crustacean reveal different diversification patterns among sister Izvestiâ Azerbajdžanskogo filiala Akademii nauk SSSR 2: lineages in the Tethyan region. Cladistics: The International 65–69. Journal of the Willi Hennig Society 30: 352–365. Dick JTA, Platvoet D, Kelly DW. 2002. Predatory impact Jazdzewski K, Konopacka A. 2002. Invasive Ponto-Caspian of the freshwater invader Dikerogammarus villosus (Crustacea: species in waters of the Vistula and Oder basins and the south- Amphipoda). Canadian Journal of Fisheries and Aquatic Sci- ern Baltic sea. In: Leppäkoski E, Gollasch S, Olenin S, eds. ences 59: 1078–1084. Invasive aquatic species of Europe. Distribution, impacts and Drummond AJ, Suchard MA, Xie D, Rambaut A. 2012. management. Dordrecht: Kluwer Academic Publishers, 384– Bayesian phylogenetics with BEAUti and the BEAST 1.7. 398. Molecular Biology and Evolution 29: 1969–1973. Jaz˙dz˙ewski K. 1980. Range extensions of some gammaridean Englisch U. 2001. Analyse der Phylogenie der Amphipoda species in European inland waters caused by human activ- (Crustacea, Malacostraca) mit Hilfe von Sequenzen des Gens ity. Crustaceana (Leiden) 6(Suppl.): 84–107. der RNA der kleinen ribosomalen Untereinheit. ub.ruhr-uni- Jaz˙dz˙ewski K, Konopacka A, Grabowski M. 2005. Native bochum. and alien Malacostracan Crustacea along the Polish Baltic Fong DW. 1989. Morphological evolution of the amphipod Sea coast in the twentieth century. Oceanological and Gammarus minus in caves: quantitative genetic analysis. Hydrobiological, University of Gdan´ sk Studies 34 (Suppl. 1): American Midland Naturalist 121: 361–378. 175–193. Gallardo B, Aldridge DC. 2013. Priority setting for inva- Jugovic J, Prevorcˇnik S, Aljancˇicˇ G, Sket B. 2010. The sive species management: risk asessment of Ponto-Caspian atyid shrimp (Crustacea, Decapoda, Atyidae) rostrum, phy- invasive species into Great Britain. Ecological applications logeny versus adaptation, taxonomy versus trophic ecology. 23: 352–364. Journal of Natural History 44: 2509–2533. Gargani J, Rigollet C. 2007. Mediterranean Sea level vari- Kamaltynov RM. 1999a. On the higher classification of Lake ations during the Messinian salinity crisis. Geophysical Re- Baikal amphipods. Crustaceana 72: 933–944. http://dx.doi.org/ search Letters 34: 1–5. 10.1163/156854099503834 Gerhardt A, Bloor M, Mills CL. 2011. Gammarus, Impor- Kamaltynov RM. 1999b. On the evolution of Lake Baikal tant taxon in freshwater and marine changing environ- amphipods. Crustaceana 72: 921–931. ments. International Journal of Zoology 2011: Art. ID 524276, Kamaltynov RM. 2001. Amfipody (Amphipoda, Gammaroidea). 2 pages. In: Timoškin OA., ed. Annotirovannyj spisok fauny ozera Bajkal Grabowski M. 2014. Check-list for Ponto-Caspian amphipods i ego vodosbornogo bassejna. Novosibirsk: ‘Nauka’, 572– (brackish & freshwater) Caspian Sea Biodiversity Project under 831.

Karaman G. 1993. Crustacea Amphipoda di acqua dolce. Fauna Maddison DR, Maddison WP. 2000. MacClade 4: analysis d’Italia 31: 1–337. Edizione Calderini Bologna, Italia. of phylogeny and character evolution. Sunderland, MA: Sinauer Karaman G, Pinkster S. 1977a. Freshwater Gammarus species Associates. from Europe, North Africa and adjacent regions of Asia Magyar I. 2006. Radiations of lacustrine cockles (Crustacea-Amphipoda). Part I. Gammarus pulex-group and (Lymnocardiinae; Bivalvia) in the Paratethys. In: Hum L, related species. Bijdragen tot de Dierkunde 47: 1–97. Gulyás S, Sümegi P, eds. Environmental historical studies Karaman G, Pinkster S. 1977b. Freshwater Gammarus species from the late Tertiary and Quaternary of Hungary. Szeged: from Europe, North Africa and adjacent regions of Asia Department of Geology and Paleontology, University of Szeged, (Crustacea-Amphipoda). Part II. Gammarus roeseli-group and 13–23. related species. Bijdragen tot de Dierkunde 47: 165–196. Martin JW, Davis GE. 2001. An updated classification of recent Karaman G, Pinkster S. 1987. Freshwater Gammarus species Crustacea. Natural History Museum of Los Angeles County from Europe, North Africa and adjacent regions of Asia Science Series 39: i–vii, 1–124. (Crustacea-Amphipoda) Part III. Gammarus balcanicus- Martynov AV. 1930. K poznaniju fauny Amphipoda ozera Issyk- group and related species. Bijdragen tot de Dierkunde 57: kul’. Results Lake Issyk-Kul expedition in Kirghiz SSR 1: 51– 207–260. 70. Karaman G, Ruffo S. 1995. Sinogammarus troglodytes n. gen. Mats VD. 2013. Late Cretaceous and Cenozoic Stratigraphy n. sp. A new troglobiont from China (Crustacea Amphipoda). of the Baikal Rift Sediments. Stratigraphy and Geological International Journal Speleology 23: 157–171. Correlation 21: 637–651. Karaman GS. 1976. Contribution to the knowledge of the Mats VD, Shcherbakov D, Efimova IM. 2011. Late Amphipoda 79. The redescription of Gammarus fossarum Cretaceous–Cenozoic history of the Lake Baikal depression bodanicus Schell. 1934, one forgotten subspecies from Boden and formation of its unique biodiversity. Stratigraphy and Lake (fam. Gammaridae). Poljoprivreda i šumarstvo 22: 13– Geological Correlation 19: 404–423. 22. Mehanikova IV 2001. Sostav i sezonnaâ dinamika pitaniâ Karaman GS. 1977a. Contribution to the knowledge of the Brandtia parasitica (Dyb.) (Crustacea, Amphipoda) iz ozera Amphipoda 77. Gammarus ochridensis Schäf. species complex Bajkal. Issledovaniâ fauny vodoemov Vostocˇnoj Sibiri. Irkutsk. of Ohrid Lake. Glasnik Odjelenja prirodnih nauka CANU 62–70. 2: 49–89. Meulenkamp JE, Sissingh W. 2003. Tertiary palaeogeography Karaman GS. 1977b. The value of the genus Neogammarus and tectonostratigraphic evolution of the Northern and South- (Ruffo 1937) and its relation to the genus Echinogammarus ern Peri-Tethys platforms and the intermediate domains of Stebb. 1889 (fam. Gammaridae). (Contribution to the knowl- the African-Eurasian convergent plate boundary zone. edge of the Amphipoda 88). Animalia (Catania) 4: 109– Palaeogeography, Palaeoclimatology, Palaeoecology 196: 209– 121. 228. Karaman S. 1935. VII. Beitrag zur Kenntnis de Morduhaj-Boltovskoj FD, Greze II, Vasilenko SV. Süsswasseramphipoden. Zoologischer Anzeiger 110: 125– 1996. Otrâd amfipody, ili raznonogie – Amphipoda. 130. In: Morduhaj-Boltovskoj FD, ed. Opredelitel’ fauny Cˇ ernogo Kilikowska A, Wysocka A, Sell J, Zie˛tara M, Sywula T. i Azovskogo Morej. Kiev: Izdatel’stvo Naukova dumka, 2006. Notes on genetic differentiation within the endemic 440–524. Proasellus species group from Lake Ohrid (Balkan Penin- Morrone JJ. 2006. Paraphyly is bad taxonomy. Taxon 55: 831– sula). Biological Letters 43: 21–32. 832. Kley A, Maier G. 2003. Life history characteristics of the in- Mortensen E. 1982. Production of Gammarus pulex L. vasive freshwater gammarids Dikerogammarus villosus and (Amphipoda) in a small Danish stream. Hydrobiologia 87: Echinogammarus ischnus in the river Main and the Main- 77–97. Donau canal. Archiv für Hydrobiologie 156: 457–469. Neuparth T, Costa FO, Costa MH. 2002. Effects of tem- Kutschera V, Maas A, Waloszek D. 2012. Uropods of perature and salinity on life history of the marine amphi- Eumalacostraca (Crustacea s.l., Malacostraca and their pod Gammarus locusta. Implications for ecotoxicological testing. phylogenetic significance. Arthropod Systematics & Phylog- Ecotoxicology (London, England) 11: 61–73. eny 70: 181–206. Nevesskaja LA, Paramonova NP, Popov SV. 2001. History Lanfear R, Calcott B, Ho SYW, Guindon S. 2012. of Lymnocardiinae (Bivalvia, Cardiidae). Paleontological PartitionFinder: combined selection of partitioning schemes Journal 35 (Suppl. 3): S147–S217. and substitution models for phylogenetic analyses. Molecu- Ogarkov OB, Kamaltynov KM, Belikov SI, Shcherbakov lar Biology and Evolution 29: 1695–1701. DY. 1997. Phylogenetic relatedness of the Baikal lake endemial Lowry JK, Myers AA. 2013. A Phylogeny and Classification amphipodes (Crustacea, Amphipoda) deduced from partial of the Senticaudata subord. nov. (Crustacea, Amphipoda). nucleotide sequences of the cytochrome oxidase subunit III Zootaxa 3610: 1–80. genes. Molecular Biology 31: 24–29. MacDonald KS, Yampolsky L, Duffy JE. 2005. Molecular Perea S, Böhme M, Zupancˇicˇ P, Freyhof J, Šanda R, and morphological evolution of the amphipod radiation of Lake Özulug˘ M, Abdoli A, Doadrio I. 2010. Phylogenetic Baikal. Molecular Phylogenetics and Evolution 35: 323– relationships and biogeographical patterns in Circum- 343. Mediterranean subfamily Leuciscinae (Teleostei, Cyprinidae)

inferred from both mitochondrial and nuclear data. Evolu- Schellenberg A. 1937a. Kritische Bemerkungen zur Systematik tionary Biology 2010: 1–27. der Süsswassergammariden. Zoologische Jahrbücher, Abteilung Pinkster S. 1973. The Echinogammarus berilloni-group, a Systematik 69: 469–516. number of predominantly Iberian amphipod species crustacea. Schellenberg A. 1937b. Schlüssel und Diagnosen der dem Bijdragen tot de Dierkunde 43: 1–38. Süsswasser-Gammarus nahestehenden Einheiten Pinkster S. 1983. The value of morphological characters in ausschliesslich der Arten des Baikalsees und Australiens. the taxonomy of Gammarus. Beaufortia 33: 15–28. Zoologischer Anzeiger 117: 267–280. Pinkster S. 1988. Problems in the taxonomy of the freshwater Schellenberg A. 1942. Krebstiere oder Crustacea IV, Flohkrebse gammarids with special emphasis on the genus oder Amphipoda. Die Tierwelt Deutschlands 40, Jena. Echinogammarus in Italy. Crustaceana Suppl. 13: 245–255. Schellenberg A. 1943. Portugiesische Süsswasser-Amphipoden. Pinkster S. 1993. A revision of the genus Echinogammarus Memorias e estudos do Museu zoologico da Universidade de Stebbing, 1899 with some notes on related genera (Crustacea, Coimbra 139: 1–5. Amphipoda). Memorie del Museo Civico di Storia Naturale Schön I, Martens K. 2004. Adaptive, pre-adaptive and non- (serie 2a) Sezione Scienze della Vita, Verona 10: 1–185. adaptive components of radiations in ancient lakes, a review. Pinkster S, Stock J. 1970. Western European species of the Organisms, Diversity & Evolution 4: 137–156. presumed Baikal-genus Eulimnogammarus (Crustacea – Segerstråle SG. 1958. On an isolated Finnish population of Amphipoda), with description of a new species from Spain. the relict amphipod Pallasea quadrispinosa G.O. Sars ex- Bulletin Zoologisch Museum 1: 205–219. hibiting striking morphological reduction, with remarks on Platvoet D, van der Velde G, Dick JTA, Li S. 2009. Flex- other cases of morphological reduction in the species. ible omnivory in Dikerogammarus villosus. Crustaceana 82: Commentationes biologicae 17: 1–33. 703–720. Segerstråle SG. 1954. The freshwater amphipods, Gammarus Popa OA, Sarkany-Kiss A, Kelemen BS, Ei I, Murariu D, pulex (L.) and Gammarus lacustris G. O. Sars, in Popa LO. 2009. Contributions to the knowledge of the present Denmark and Fennoscandia – a contribution to the late- Limnocardiidae fauna (Mollusca, Bivalvia) from Romania. and post-glacial immigration history of the aquatic Travaux du Muséum National d’Histoire Naturelle «Grigore fauna of northern Europe. Commentationes Biologicae 15: 1– Antipa» 52: 7–15. 91. Popescu SM, Dalesme F, Jouannic G, Escarguel G, Head Sexton EW, Spooner GM. 1940. An account of MJ, Melinte-Dobrinescu MC, Suto-Szentai M, Bakrac Marinogammarus (Schellenberg) gen. nov. (Amphipoda), with K, Clauzon G, Suc JP. 2009. Galeacysta etrusca complex, a description of a new species, M. pirloti. Journal of the Marine Dinoflagellate cyst marker of Paratethyan influxes to the Medi- Biological Association of the United Kingdom 24: 634– terranean Sea before and after the peak of the Messinian 682. Salinity Crisis. Palynology 33: 105–134. Sherbakov D, Kamaltynov RM, Ogarkov OB, Väinölä R, Popov SV, Rögl F, Rozanov AY, Steininger FF, Shcherba Vainio JK, Verheyen E. 1999. On the phylogeny of lake IG, Kovac M (Eds). 2004. Lithological-Paleogeographic maps Baikal amphipods in the light of mitochondrial and nuclear of Paratethys 10 maps Late Eocene to Pliocene. Courier DNA sequence data. Crustaceana 72: 911–919. Forschungsinstitut Senckenberg (CFS) 250: 1–46. Sket B. 1981. Rhynchobdellid leeches (Hirudinea, Protopopova MV, Takhteev VV, Shatilina Z, Pavlichenko Rhynchobdellae) in the relic Ohrid Lake region. Biološki vestnik VV, Axenov-Gribanov DV, Bedulina DS, Timofeyev MA. 29: 67–90. 2011. Small HSPs molecular weights as new indication to Sket B. 1986. Ecology of the mixohaline hypogean fauna along the hypothesis of segregated status of thermophilic relict the Yugoslav coast. (Stygologia, 2(4): 317-338); In: White WB, Gmelinoides fasciatus among Baikal and Palearctic amphipods. Culver DC, eds. Benchmark Papers in Karst Science, Karst Journal of Stress Physiology & Biochemistry 7: 175– Waters Institute, Special Publication (2007) 11: 395–418. 182. Sket B. 1989. Intralacustrine speciation in the genus Dina Rambaut A, Drummond AJ. 2009. Tracer. v. 1.5. Available (Hirudinea, Erpobdellidae) in Lake Ohrid (Yugoslavia). at: http://tree.bio.ed.ac.uk/software/tracer/ Hydrobiologia 182: 49–58. Ronquist F, Teslenko M, van der Mark P, Ayres DL, Sket B. 1997. Distribution of Proteus (Amphibia, Urodela, Darling A, Höhna S, Larget B, Liu L, Suchard MA, Proteidae) and its possible explanation. Journal of Biogeog- Huelsenbeck JP. 2012. MrBayes 3.2: efficient Bayesian raphy 24: 263–280. phylogenetic inference and model choice across a large model Sket B. 2000. Fuxiana yangi g.n., sp.n. (Crustacea, Amphipoda), space. Systematic Biology 61: 539–542. a ‘baikaloid’ amphipod from the depths of Fuxian Hu, an Salemaa H, Kamaltynov R. 1994. The chromosome numbers ancient lake in the karst of Yunnan, China. Archiv für of endemic Amphipoda and Isopoda – an evolutionary paradox Hydrobiologie 147: 241–255. in the ancient lakes Ohrid and Baikal. In: Martens K, Sket B. 2008. Can we agree on an ecological classification of Goddeeris B, Coulter G, eds. Speciation in ancient lakes. Archiv subterranean animals? Journal of Natural History 42: 1549– für Hydrobiologie, 44: 247–256. 1563. Sars GO. 1895. Amphipoda. An account of the Crustacea of Stamatakis A, Hoover P, Rougemont J. 2008. A rapid boot- Norway with short descriptions and figures of all the species, strap algorithm for the RAxML web servers. Systematic Biology l, viii + 711 pp., 240 pls, 8 supplementary pls. 57: 758–771.

Starobogatov Y. 1994. Taxonomy and paleontology. In: Vainio JK, Väinölä R. 2003. Refugial races and postglacial Starobogatov YI, ed. Freshwater Zebra Mussel: Dreissena colonization history of the freshwater amphipod Gammarus Polymorpha (Pall.) (Bivalvia, Dreissenidae): Systematics, lacustris in Northern Europe. Biological Journal of the Linnean Ecology, Practical Meaning. Russian Academy of Sciences Society 79: 523–542. (Nauka Press), Moscow, 18–46. Väinölä R, Kamaltynov RM. 1995. Allozyme studies on the Stebbing TRR. 1906. Amphipoda 1. Gammaridea. Das Tierreich evolutionary diversity of Baikalian amphipod crustaceans, 21: 1–806. and their relationships to the European escapee species Stepien CA, Tumeo MA. 2006. Invasion genetics of Ponto- Pallasea quadrispinosa. The Second Vereshchagin Baikal con- Caspian gobies in the Great Lakes, a ‘cryptic’ species, absence ference, Abstracts International meeting, Irkutsk: SB RAS of founder effects, and comparative risk analysis. Biologi- Press, 40–41. cal Invasions 8: 61–78. Väinölä R, Kamaltynov RM. 1999. Species diversity and spe- Stock JH. 1967. A revision of the European species of the ciation in the endemic amphipods of Lake Baikal, molecu- Gammarus locusta-group (Crustacea, Amphipoda). Zoologische lar evidence. Crustaceana 72: 945–955. verhandelingen 90: 1–56. Väinölä R, Kontula T, Kamaltynov RM. 2000. Use of Stock JH. 1968. A revision of the European species of the mitochondrial DNA and allozyme characters to explore the Echinogammarus pungens-group (Crustacea, Amphipoda). systematic diversity of Lake Baikal amphipods (genus Beaufortia 16: 13–78. Pallasea). Biodiversity and dynamics of ecosistems in north Stock JH. 1973. The existence of interstitial members of Eurasia, Abstracts International meeting, 5: 204–206. the Gammarus-group (Amphipoda). Crustaceana 24: 339– Väinölä R, Vainio JK, Palo JU. 2001. Phylogeography of 341. ‘glacial relict’ Gammaracanthus (Crustacea, Amphipoda) from Stock JH. 1986. Amphipoda, Gammarid grouping. In: boreal lakes and the Caspian and White seas. Canadian Botosaneanu L, ed. Stygofauna Mundi. Leiden: E.J. Brill/ Journal of Fisheries and Aquatic Sciences 58: 2247– Dr. W. Backhuys, 497–503. 2257. Stock JH. 1993. Gammarus and Chaetogammarus (Crustacea, Weiss M, Macher JN, Seefeldt MA, Leese F. 2014. Mo- Amphipoda) from Macaronesia. Boletim do Museu munici- lecular evidence for further overlooked species within the pal do Funchal 45: 41–52. Gammarus fossarum complex (Crustacea, Amphipoda). Swofford DL. 2002. PAUP*: Phylogenetic analysis using par- Hydrobiologia 721: 165–184. simony (and other methods). Sunderland, MA: Sinauer As- Witt JDS, Hebert PDN, Morton WB. 1997. Echinogammarus sociates. ischnus, another crustacean invader in the Laurentian Great Tahteev (Takhteev) VV. 2000. Ocˇerki ob amfipodah ozera Lakes basin. Canadian Journal of Fisheries and Aquatic Sci- Bajkal (sistematika, sravnitelnaâ ekologiâ, evoluciâ). ences 54: 264–268. Izdatel’stvo Irkutskogo Universiteta, Irkutsk, 335 pp. Wolfe JA, Poore RZ 1982 Tertiary marine and nonmarine Takhteev VV, Berezina NA, Sidorov D. 2014. Amphipods climatic trends. Climate in Earth History, Studies in Geo- of continental waters of Russia, taxonomy, biogeography and physics. U.S. Geological Survey. Available at: http:// alien species. 8th International Crustacean Congress, Frank- www.nap.edu/catalog/11798.html (accessed 2013). furt am Main, Abstract volume, p. 204. Wysocka A, Grabowski M, Sworobowicz L, Burzynski A, Takhteyev VV, Mekhanikova IV. 1993. Some data on the Kilikowska A, Kostoski G, Sell J. 2013. A tale of time biology of Baikal parasitic amphipods of the genus Pachyschesis and depth, intralacustrine radiation in endemic Gammarus (Amphipoda, Gammaridae). Hydrobiological Journal 29: 93– species flock from the ancient Lake Ohrid. Zoological Journal 103. of the Linnean Society 167: 345–359. Thienemann A. 1950. Verbreitungsgeschichte der Zenkevicˇ LA. 1947. Fauna i biologicˇeskaâ produktivnost’ morâ. Süsswassertierwelt Europas. Stuttgart: Die Binnengewässer. Morâ SSSR, T. II. Sovetskaâ Nauka. E. Schweizerbart’sche Verlagsbuchhandlung. Zwickl DJ. 2006. Genetic algorithm approaches for the Thompson JD, Gibason TJ, Plewniak F, Jeanmougin F, phylogenetic analysis of large biological sequence datasets under Higgins DG. 1997. The Clustal_X windows interface: flex- the maximum likelihood criterion. PhD thesis, University of ible strategies for multiple sequence alignment aided by quality Texas at Austin, Austin, TX. GARLI, version 1.0. Published analysis tools. Nucleic Acids Research 25: 4876–4882. by the author. Available at: http://www.nescent.org/wg_garli

SUPPORTING INFORMATION Additional supporting information may be found in the online version of this article at the publisher’s web-site: Figure S1. Bayesian inference chronograms of Gammaridae under different calibration schemes based on a combined data set. Table S1. Taxon information and GenBank numbers.