Macrostylis sabinae, Bober & Riehl & Henne & Brandt, 2018

REMARKS FOR MACROSTYLIS SABINAE View in CoL SP. NOV. AND MACROSTYLIS AMALIAE SP. NOV.

The females of Macrostylis sabinae sp. nov. and M. amaliae sp. nov. are remarkably similar to each other. The coxal seta only on pereopod VII seems to be a synapomorphic character for these sister species. Coxal setae on pereopods V–VII are found in M. wolffi Mezhov, 1988 from the Pacific Ocean. Apart from the shared coxal seta on pereopod VII the species have no further similarities.

The distributions of the morphologically indistinguishable (in the case of females, mancas and subadult male stages) M. sabinae sp. nov. and M. amaliae sp. nov. are sympatric but they are genetically distinct ( Fig. 35). However, the adult males of both species are morphologically distinct. While subadult males and the females have ventral projections on pereonites 5 and 6 ( Figs 13B; 25B, 29A; 30B, C; Supporting Information S2B, D, S3B, S5), they are absent in the adult male of M. sabinae sp. nov. ( Figs 20B, 21B, C). Furthermore, in M. sabinae sp. nov. the lateral lobes of pleopod I project beyond the medial lobes distally ( Fig. 24A, B), which is not the case in M. amaliae sp. nov. ( Fig. 33A, B). Another difference between the terminal males of both species is featured in the aesthetascs, which are different in length and structure ( Fig. 22). The aesthetascs of M. amaliae sp. nov. extend up to the distal margin of the fourth segment of the antenna (merus) ( Figs 22B, 29A, 30B, Supporting Information S5) while the aesthetascs of M. sabinae sp. nov. extend further until the distal margin of the fifth segment (carpus) ( Figs 20A, B, 21A, B, 22A). Furthermore the aesthetascs of M. sabinae sp. nov. have a conspicuous constriction distally to the macrostylids’ common belt of constriction, which is situated medially along the aesthetasc proximo-distal axis ( Figs 20D, E, G, 22A). This additional constriction was present on the majority of aesthetascs. There was no interspecific difference found in the aesthetascs of juvenile males ( Fig. 20F).

IDENTIFICATION KEY TO THE SPECIES OF MACROSTYLIDAE FROM THE NORTHWEST PACIFIC

Remarks: Except where mentioned otherwise, the key is based on females. For the identification key all adequately described species known for the KKT region were included. However, four species were excluded. The descriptions of Macrostylis profundissima Birstein, 1970 , M. sensitiva Birstein, 1970 and M. quadratura Birstein, 1970 are based on male specimens only and the females remain unknown. Macrostylis ovata Birstein, 1970 was excluded because we assume, based on the development of the seventh pereonite and the similarly weakly developed setation on the anterior pereopods, that the individual upon which the species was described may be a manca, possibly of M. grandis Birstein, 1970 , which occurs sympatrically.

KEY TO THE NORTHWEST PACIFIC MACROSTYLIDAE 1. Pereonite 1 ventral projection directed ventrally, rounded or acute, spine-like .................................................... 2 Pereonite 1 ventral projection spine-like, orientated anteriorly.............................................................................. 5

2(1). Pereonite 6 posterolateral margin produced posteriorly; pereonite 7 with posterolateral protrusions, similar to pereonites 5 and 6; operculum ventrally roundedly keeled; pleotelson ventrolateral setal ridges present..................................................................................................................................................................... 3 Pereonite 6 posterolateral margin not produced posteriorly; pereronite 7 without or with weakly developed posterolateral protrusions; operculum without keel; pleotelson ventrolateral setal ridges absent. 4

3(2). Pleotelson waist absent, without lateral constriction anteriorly to uropod insertions; operculum elongate (length 1.8 width); pleotelson as wide as pereonite 7; pereopod VII dorsal (posterior) margin row of elongate setae present; pleotelson setal ridges not visible in dorsal view; pleotelson posterior apex acutely tapering; antennula (antenna 1) comprising one segment .............. Macrostylis curticornis Birstein, 1963 Pleotelson waist present, constricted anteriorly to uropod articulation; operculum elongate (length clearly more than 1.5 width); pleotelson narrower than pereonite 7; pereopod VII dorsal (posterior) margin row of elongate setae absent; pleotelson setal ridges visible in dorsal view; pleotelson posterior apex smoothly rounded; antennula (antenna 1) five segments.................................................................................... Macrostylis daniae sp. nov.

4(2). Pereronite 4 posterolateral setae present and segment widened in the middle; pleotelson waist present, constriction anteriorly to uropod articulation; operculum elongate (length clearly more than 1.5 width), acutely tapering posteriorly; pleotelson shape ovoid, lateral margins convex (outlines of anterior part in dorsal view); fossosome ventral surface with sharp keel; antennula (antenna 1) of three segments (unlike in the original description); distinctly elongate and slender body; uropod protopod 4.5 times the length of endopod...................................................................................................... Macrostylis longula Birstein, 1970 Pereronite 4 posterolateral setae absent; waist absent; operculum stout (length 1.5 width or less), smoothly rounded posteriorly; pleotelson shape narrowing evenly towards uropodal insertions, lateral margins straight (outlines of anterior part in dorsal view); fossosome ventral surface without keel; antennula (antenna 1) of five segments; distinct by heavy imbricate ornamentation on all segments........................................................................................................ Macrostylis reticulata Birstein, 1963

5(1). Pereonite 4 posterolateral margins produced posteriorly and posterolateral setae present; pleotelson waist absent, constriction anteriorly to uropod articulation; operculum stout (length 1.5 width or less); fossosome ventral surface without keel .................................................................................................................................. 6 Pereonite 4 posterolateral margins not produced posteriorly and posterolateral setae absent; pleotelson waist present, constricted anteriorly to uropod articulation; operculum elongate (length clearly more than 1.5 width); fossosome ventral surface with sharp keel ............................................................................................... 8

6(5). Distinct body shape: stout (L/ W ratio <3.0) and rather large (7.8 mm); pleotelson anteriorly much wider than posteriorly, convex, progressively narrowing towards uropod insertions (outlines of anterior part in dorsal view); pereonite 3 posterolateral margin with tapering posterior projection; pereonite 7 without or with weakly developed posterolateral protrusions (as in manca); pleotelson as wide as pereonite 7 .............................................................................................. Macrostylis grandis Birstein, 1970 Body shape elongate (L/ W ratio> 3.0); pleotelson ovoid, lateral margins convex (outlines of anterior part in dorsal view); pereonite 3 posterolateral margin not produced posteriorly, pereonite 7 with posterolateral protrusions, similar to pereonites 5 and 6; pleotelson narrower than pereonite 7 ........................................................ 7

7(6). Pereonite 6 length clearly larger pereonite 5 length; pereonite 5 length smaller or subequal pereonite 4 length; pereonite 6 posterolateral margin rounded; pereonite 4 pereonal collum laterally expressed (segment anteriorly constricted); pereonite 4 shape generally resembling more posterior pereonites; posterolateral spine like setae and ventral projection present on pereonite 4; ventral and dorsal row of elongate setae present on pereopod VII............................................................. Macrostylis zenkevitchi Birstein, 1963 Pereonite 6 length smaller or subequal pereonite 5 length; pereonite 5 length clearly greater pereonite 4 length; pereonite 6 posterolateral margin tapering; pereonite 4 pereonal collum laterally not expressed (segment anteriorly not constricted); pereonite 4 shape clearly distinct from both anterior and posterior pereonites ........................................................................................................... Macrostylis affinis Birstein, 1963

8(5). In males ventral projections similar to female on all pereonites; in males aesthetascs all of same type; pleotelson L/ W ratio in male subequal to female........................................................................................... Macrostylis amaliae sp. nov. In males ventral projections differ from females, ventral projections on pereonites 5 and 6 absent; in males aesthetascs of multiple types; pleotelson L/ W ratio in male greater than in female............................. Macrostylis sabinae sp. nov.

SPECIES DELIMITATION OF KNOWN SPECIES FROM THE NORTHWEST PACIFIC WHICH WERE EXCLUDED FROM THE KEY

Macrostylis sensitiva Birstein, 1970 : The adult males are distinguishable from Macrostylis daniae sp. nov. by the shape of the first antenna (antennula). Macrostylis daniae sp. nov. has a squat and not elongate terminal segment. In contast to Macrostylis sabinae sp. nov. and Macrostylis amaliae sp. nov., this species has a straight apex seta on the ischium of pereopod III. Furthermore, the shape of the pleotelson narrows continuously to the uropod insertion. The pleotelson is clearly wider anteriorly than posteriorly. Macrostylis profundissima Birstein, 1970 : In contrast to the three species described here the first antennula is composed of a single segment and the pleotelson waist is absent.

Macrostylis quadratura Birstein, 1970 : The pleotelson is rectangular in form. The antennula is short, thick and composed of three segments only.

GENETIC RESULTS, PHYLOGENETIC INFERENCE AND MOLECULAR SPECIES DELIMITATION

We were able to successfully amplify 50 sequences for the 16S gene fragment and nine sequences for the 18S gene fragment. Unfortunately, we were not able to amplify the barcoding marker COI. The 16S gene fragment varied between 384 and 493 base pairs (bp) in length and had a high AT content (66.4%) typical of this gene ( Simon et al., 1994). The MAFFT alignment for the 16S gene was made from 109 sequences including the outgroup and had a length of 519 bp of which 200 bp was conserved, 242 bp was variable and 210 bp was parsimony informative. Following the AIC and hLRT, the best substitution model was GTR+G with no invariable sites and a gamma distribution shape parameter of 0.3188. The nucleotide frequencies of the alignment were A = 0.3438, C = 0.1354, G = 0.1811 and T = 0.3397. The substitution rates were R[AC] = 0.9425, R[AG] = 7.3100, R[AT] = 3.0851, R[CG] = 0.3979, R[CT] = 9.4418 and R[GT] = 1.000. The 18S gene fragment amplified varied between 1766 and 2221 bp in length and had a balanced AT to GC content (AT = 50.5%). The MUSCLE alignment for the 18S gene was made from 16 sequences including the outgroup and had a length of 2369 bp of which 1931 bp was conserved, 402 bp was variable and 263 bp was parsimony informative. The AIC and the hLRTs suggested the same substitution model, which was GTR+I+G with a proportion of invariable sites of 0.6605 and a gamma distribution shape parameter of 0.4811. The nucleotide frequencies of the alignment were: A = 0.2387, C = 0.2229, G = 0.2706 and T = 0.2678. The substitution rates were: R[AC] = 0.4540, R[AG] = 1.3860, R[AT] = 0.7423, R[CG] = 0.3988, R[CT] = 2.5437 and R[GT] = 1.000. Considerably more sequences were amplified for the 16S genetic marker. However, the phylogenetic reconstructions of the two markers separately resulted in similar topologies.

For both M. sabinae sp. nov. and M. amaliae sp. nov., the 16S gene showed a low maximum within-group divergence of 0.8% uncorrected p -distance ( Fig. 35). The clades formed by the individuals of these two respective species divided into two monophyletic groups ( Fig. 36, posterior probability = 1), representing the species proposed here. They are genetically distinct by 7.7–8.0 % without intermediate distances ( Fig. 35). The most closely related species to M. sabinae sp. nov. and M. amaliae sp. nov. in the 16S dataset is M. scotti . This relatedness is statistically well supported (post. prob. = 1). The rest of the cladogram is not well supported and not sufficiently resolved; for a better resolution more sequences from more species would be necessary. In the 16S cladogram, the species M. roaldi represents the well-supported sister taxon (post. prob. = 0.97) to all other tested Macrostylidae . Macrostylis daniae sp. nov. occupies one distinct clade ( Fig. 36), but its position is not well supported (post. prob. = 0.51).

The 18S cladogram differs slightly from the 16S cladogram. Since 18S is a more slowly evolving gene than 16S and the species composition differed between the alignments, the 18S cladogram is better resolved and better supported. Macrostylis sabinae sp. nov. and M. amaliae sp. nov. remain in two monophyletic groups (Supporting Information S6, post. prob. = 1). However, M. roaldi does not sit opposite to all other Macrostylidae ; here it is a rather ‘recent’ species forming a monophyletic group with Macrostylis sp. ( EU414442 View Materials ) (post. prob. = 1). Macrostylis daniae sp. nov. forms a monophyletic group with Macrostylis sp. ( AY461477 View Materials ) (post. prob. = 0.88). The species Macrostylis sp. ( AY461476 View Materials ) is placed opposite to all other Macrostylidae for the 18S marker (post. prob. = 0.72). The genetic distinction of the three newly described species was apparent in the haplotype network as well ( Fig. 37). In the haplotype network 13 haplotypes are represented. Haplotypes 1–4 represent M. amaliae sp. nov., haplotypes 5–8 represent M. sabinae sp. nov. and haplotypes 9–13 represent M. daniae sp. nov.

Macrostylis sabinae sp. nov. is separated from M. amaliae sp. nov. by 27 mutation steps (7.7% divergence), while there is a maximum of six mutations (0.8%) within these species. Macrostylis daniae sp. nov. is separated from M. sabinae sp. nov. by 97 mutations (28.6%) and from M. amaliae sp. nov. by 124 mutations (29.4%) and has a higher intraspecific variation than M. sabinae sp. nov. or M. amaliae sp. nov. with a maximum of 15 mutations (1.9%) within its clade. Those distances, however, are mainly caused by one individual which was the only specimen of this species sampled at station 6–11. KBMa120 is separated from the closest other specimens of M. daniae sp. nov. by nine mutations (1.6%).

BODY-SIZE VARIATION BETWEEN MALES AND FEMALES OF M. SABINAE SP. NOV. AND M. AMALIAE SP. NOV.

Variations in body length were found between adult conspecific males and females ( Fig. 38). We were interested in the size difference between males and females of comparable stages. To compare these, only ovigerous females and adult males were considered for analyses. The genetic dataset was unbalanced due to the low numbers of terminal males. As a result, it was not suitable for statistics, but a boxplot ( Fig. 39A) provided an overview of the available data. Furthermore, this dataset confirms similar variations in both species.

To test whether the observed size variability between males and females among these two species was statistically significant, the formalin-fixed material was included in the analysis. The formalin-fixed animals of M. sabinae sp. nov. and M. amaliae sp. nov. were analysed together with the ethanol-fixed material, but all individuals were treated as one species ( Macrostylis sabinae-amaliae complex). A Wilcoxon– Mann–Whitney U test was conducted to compare the body length of ovigerous females and adult males ( Fig. 39B). The ovigerous females are significantly larger than the adult males (W = 129.5, P <0.0001). While equally significant (Welch two-sample t -test, M females = 1.961, M males = 1.632, t (35.66) = 5.707, P <0.0001) the non-ovigerous but seemingly adult females may be of interest, but they represent a rather roughly defined group possibly comprising multiple developmental stages and are not a sufficient group for a size comparison.

A significant size difference was further found between ovigerous females among stations (Kruskal– Wallis test, χ 2 (11) = 20.985, P <0.05) ( Fig. 39C). The effect of this incident on the present data was analysed in Fig. 39D. Based on the results ( Fig. 39D) it is clear that the size difference between males and females was similarly distributed among all stations.

DISPERSIBILITY OF M. SABINAE SP. NOV. IN THE ABYSS

Conspecific specimens were collected at abyssal depths from both sides of the hadal KKT. With its maximum depth of over 9700 m, the KKT may well represent a dispersal barrier for abyssal benthos. Station 3–9 was located north of the KKT, while all other stations were located south of the trench ( Fig. 1). It was hence possible to test for connectivity of abyssal species across the KKT. Three individuals of M. sabinae sp. nov. of the same 16S haplotype were found north of the KKT (station 3–9) (ZMH K-45929, ZMH K-45933, ZMH K-45926) ( Fig. 37: haplotype 5). The closest station across the KKT was station 2–9 ( Fig. 1), where 12 individuals of M. sabinae sp. nov., also sharing one haplotype, were found ( Fig. 37: haplotype 6). Both haplotypes, geographically isolated by the KKT, were separated by four mutation steps equalling 0.5% uncorrected p -distance. All three haplotypes south of the KKT were separated by only one mutation step (0.3%) ( Fig. 37: HT6 vs. HT7, HT8). No correlation was found between genetic and geographical distance (Mantel test, r = 0.191, 9999 replicates, P> 0.30). A possible genetic barrier was analysed using a ‘genetic landscape shapes’ interpolation ( Miller, 2005) ( Fig. 40). The three-dimensional genetic landscape presented high genetic p -distances across the KKT. The interpolation was based on a Delauney triangulation network ( Watson, 1992; Brouns, Wulf & Constales, 2003) ( Fig. 40: black lines). One high peak was found from station 3–9 to 7–9/5–9 and a further peak was found between 3–9 and 2–9, indicting high genetic distances. Among the stations south of the KKT, three low peaks for low genetic distances were found. Monmonier’s algorithm implemented in Alleles in Space detected a barrier in the tested dataset ( Fig. 40: red line). This suggested barrier overlapped with the extent of the KKT, indicating that hadal trenches represent a physical barrier to deep-sea benthic organisms.