Dancing for Food in the Deep Sea: Bacterial Farming by a New Species of Yeti Crab

Vent and seep animals harness chemosynthetic energy to thrive far from the sun's energy. While symbiont-derived energy fuels many taxa, vent crustaceans have remained an enigma; these shrimps, crabs, and barnacles possess a phylogenetically distinct group of chemosynthetic bacterial epibionts, yet the role of these bacteria has remained unclear. We test whether a new species of Yeti crab, which we describe as Kiwa puravida n. sp, farms the epibiotic bacteria that it grows on its chelipeds (claws), chelipeds that the crab waves in fluid escaping from a deep-sea methane seep. Lipid and isotope analyses provide evidence that epibiotic bacteria are the crab's main food source and K. puravida n. sp. has highly-modified setae (hairs) on its 3rd maxilliped (a mouth appendage) which it uses to harvest these bacteria. The ε- and γ- proteobacteria that this methane-seep species farms are closely related to hydrothermal-vent decapod epibionts. We hypothesize that this species waves its arm in reducing fluid to increase the productivity of its epibionts by removing boundary layers which may otherwise limit carbon fixation. The discovery of this new species, only the second within a family described in 2005, stresses how much remains undiscovered on our continental margins.


Introduction
From ants that use symbiotic bacteria to protect the fungi that they farm [1] to polychaetes that ''garden'' bacteria [2], animals have developed a diversity of mechanisms to increase their symbionts' productivity and health. Few places provide as many examples of these symbioses as reducing systems, including hydrothermal vents and cold seeps, where shrimp, mussels, clams, and mouthless tubeworms are fueled by their chemoautotrophic bacterial symbionts [3][4][5][6][7][8]. As we try to understand how these systems function, novel species are constantly revealed. In 2005 a new family of crab was discovered at a hydrothermal vent. This crab had chelipeds (claws) covered in dense setae and epibiotic bacteria that lead to this species, Kiwa hirsuta, to be called the ''Yeti crab'' [9]. Only a single individual of K. hirsuta was collected leaving the ecology and role of its bacterial epibionts largely unknown [9,10]. During June 2006, we discovered a second species of Yeti crab, which we formally describe here at Kiwa puravida n. sp, swinging its bacteria-laden chelipeds rhythmically at a Costa Rican methane seep (Figures 1A-C and 1F; Video S1). In this study, we describe how this new species farms its epibiotic bacteria in a unique form of symbiosis.
The symbiont-host interactions of epibiont bearing hydrothermal vent crustaceans have remained perplexing; these shrimp, barnacles, and crabs, share a distinct phylogeny of chemosynthetic epibionts [11] but if and how they harvest their symbionts has remained elusive (summarized in [8]). The best studied of this group is the vent shrimp Rimicaris exoculata that gains a large proportion of its energy from the eand c-proteobacteria symbionts that grow inside its carapace and on its enlarged mouthparts [8,[12][13][14][15]. What remains unknown is how R. exoculata harvests the bacteria that grow on these structures. Competing hypotheses suggest that R. exoculata either uses its chelipeds to transfer its epibiotic bacteria to its mouth or that R. exoculata consumes its discarded bacteria-rich exoskeleton after molting [12]. This has led some to question the nutritional role of these symbionts [8]. Vent barnacles also appear to have morphological adaptation to harvest their symbionts [16], but no observations support this hypothesis. Only the vent crab Shinkaia crosnieri, has been observed to scrape off its epibiotic bacteria and transfer them to its mouth, providing a clear mechanism for symbiont harvesting [17]. Isotopic evidence further supported S. crosnieri's consumption of its epibionts [18]. Yet the importance of S. crosnieri's symbionts to its overall nutrition remains unknown [18]. The hydrothermal vent Yeti crab, K. hirsuta, may rely on its chemosynthetic symbionts for a food source, as suggested by the presence of enzymes necessary for carbon fixation within K. hirsuta's epibiotic bacteria, but it has not been observed to consume these bacteria [8,10,11].
A key aspect of farming, a form of mutualistic symbiosis, is the direct trophic transfer of energy from a symbiont to its host, a phenomenon that can be shown through biomarker analysis. Biomarkers, specifically carbon isotopic and fatty acid (FA) analyses, track a signature from an individual's diet into its tissues providing a time-integrated view of what that species eats [19,20]. Chemosynthetic and photosynthetic derived biomass commonly differ in the ratio of carbon-13 to carbon-12 and the length, bonding and branching patterns of certain FAs in the producer's lipid membranes [21][22][23]. Consumers derive their isotopic and a portion of their FA composition from their diet [20,24]. Through analysis of an individual's tissue, that individual's main food sources can often be identified. In certain cases, rare or unique FAs can be traced from a symbiont into their host, and provide a robust measure of direct consumption of that symbiont. An example of this is R. exoculata, as this shrimps' bacteria have relatively unique 16:2(n24) and 18:2 (n24) FAs that can be traced into the shrimp's tissue [25]. Perplexingly, these same (n24) FAs were also present in S. crosnieri's epibionts but they were not found in S. crosnieri's tissue as would be expected if the crab's bacteria were its main food source [18]. Isotopic and FA analyses of K. puravida n. sp. may provide insight about its consumption of chemoautotrophic production and identify whether it is gaining energy from its symbionts or not.
Vents and seeps are both fueled by similar chemical reactions which has lead to cross-ecosystem comparisons since the first seep community was discovered [26][27][28][29]. Symbionts have been especially enlightening in demonstrating the similarity among these habitats [27,29,30]. An initial description of the epibionts found on the Costa Rican Kiwa species found provocative crossecosystem similarities among vent and seep crustacean epibionts based on a limited number of short (,500 base pairs) 16S rRNA gene sequences [11]. Here we build upon that research through description of the host's phylogeny and add additional epibiont analysis. Our comparison of host and epibiont evolutionary histories will provide a better understanding of the biogeography of these disparate ''islands'' of chemoautotrophy in the deep-sea, as we test the hypothesis that K. puravida n. sp. farms its epibiotic bacteria.

Materials and Methods
Specimens were collected with DSRV Alvin on RV Atlantis Cruises  Specimens were preserved in 8% buffered formalin or 95% ethanol. In two instances a pereopod was removed for isotopic, genetic, epibiont, and fatty acid analysis and frozen at 280uC.
Measurements of specimens are given in millimeters (mm) and indicate the postorbital carapace length unless otherwise indicated. Specimens are deposited at the Smithsonian (Holotype and Paratype 1) and University of Costa Rica (Paratype 2). Additional specimens are being deposited at NIWA Invertebrate Collection, Wellington, New Zealand, and the Scripps Institution of Oceanography Benthic Invertebrate Collection. Descriptions were prepared using DELTA (DEscriptive Language for Taxonomy [31]). Drawings were made using a WACOM Intuous3 and Intuous4 Graphics Tablets and Adobe Illustrator CS2-CS4.
Stable isotopic and fatty acid (FA) analyses were performed on two specimens that underwent molecular analysis and ''bulk plankton'' collected during cruise AT 15-5. The bulk plankton sample was collected less than 50 m above Mound 12 and sieved through a 63 mm sieve before being frozen at 280uC. Stable isotopic analysis was performed on muscle tissue of an additional 28 specimens, collected during cruises 15-44 and 15-59 at which time particulate organic carbon (POC) isotopic measurements were also made. POC isotopic analysis was performed on surface and bottom water at areas of active seepage along the Costa Rican margin. POC samples were collected by 9 CTD deployments and water was filtered through pre-combusted glass-fibre filters to estimate of the potential range of planktonic isotopic signature for this region. 0.2 mg of tissue, bulk plankton, or scraping of the GF/ Fs were placed in tin boats, dried at 60uC overnight, acidified with 1% platinum chloride and analyzed on a Eurovector elemental analyzer interfaced with a continuous flow Micromass Isoprime isotope ratio mass spectrometer at Washington State University in the lab of Dr. Raymond Lee. Isotope ratios are expressed as d 13 C in units of per mil (% -notation explained in [21]) using Pee Dee Belemnite as the standard. Fatty acids were extracted in a one step extraction-transesterification method [32]. Freeze-dried tissue was placed in 3 ml MeOH:HCl: CHCl 3 (10:1:1 v/v/v) at 60uC for 60 min, cooled and 1 ml Milli-Q H 2 O added prior to extraction in hexane:chloroform, (4:1 v/v), and dried over sodium sulfate. FAs were analyzed on a Thermo Finnegan Trace gas chromatograph/ mass spectrometer and peak integration was performed using Xcaliber software. Percentage of total FAs are given for abundant FAs defined as those that composed more than 1.0% from any of  Figure 4E for scale). (F) Dense aggregation in situ. (G) Shipboard photo of K. puravida using its 3 rd maxilliped to harvest its epibiotic bacteria. (H) Comb-row setae with bacteria filaments stuck among combs (indicated by arrow). doi:10.1371/journal.pone.0026243.g001 the samples. As peak area measurements from mass spectra are not a function of concentration alone (spectra response varies with FA analyzed), these FA data are comparable within this study only.
Molecular analyses were performed on the two specimens collected during 2006. Genomic DNA was isolated from ethanolpreserved muscle (50 mg) and treated with the QiagenDNeasy isolation kit, according to manufacturer's instructions (Qiagen Inc., Valencia, CA). Polymerase chain reaction (PCR) conditions for amplification of the gene regions were as follows: 100 ng of template DNA, 5 ml 10X buffer (supplied by manufacturer), 5 ml MgCl 2 (2.5 mM), 2 ml of each primer (10 mM final conc.), 2.5 units of Taq polymerase (Promega Inc., WI), 5 ml of a 2 mM stock solution of dNTPs, and sterile water to a final-volume of 25 ml. A <2000 basepair fragment of the 18 S rRNA gene was amplified with universal 18 S rDNA primers, 18e (59-CTGGTTGATCC-TGCCAGT-39) and 18P (59-TAATGATCCTTCCGCAGGTT-CACCT-39) [33]. PCR products were sequenced bidirectionally with an ABI 3100 DNA sequencer (Applied Biosystems Inc., Foster City, CA). GenBank 18 S rRNA sequences (Anomura: AF439381-AF439391, Z14062; Upogebia affinis: AF439392) were aligned with the yeti crab 18 S rRNA (DQ219316) using ClustalX 2 followed by manual checking. Secondary structure of rRNA (i.e. stems and loops) was inferred using the program GeneBee [34]. Bayesian inference of phylogeny was performed using MrBayes v3.0B4 [35] with data partitions using RNA secondary structure. Six chains were run simultaneously for 1,100,000 generations and trees sampled every 1000 generations. The first 1000 trees were discarded as ''burn in'' and Bayesian posterior probabilities were estimated on the 95% majority rule consensus. The portion of the mitochondrial genome including Cytochrome Oxidase I (COI) and the intervening region to Cytochrome Oxidase II (COII) was amplified using previously published primers and conditions [9] to determine tRNA composition and order. 16 S rRNA from bacterial genomic DNA isolated from hairs on pereopods were amplified using 27F and 1492 primers. 16 S rRNA PCR products were cloned and sequenced using published protocols [10]. Separate Neighbor-Joining Trees were constructed to compare the e-proteobacteria and c-proteobacteria 16 S rRNA genes found on K. puravida n. sp. to sequences available in GENBANK. Sequence were selected for inclusion in the tree if they were identified as the most similar to sequences from reducing habitat decapoda and were greater than 1000 basepairs in length. The 16 s rRNA sequences were aligned using ClustalW within MegAlign (DNASTAR) followed by manual checking and trimming. 3000 iterations provided bootstrap confidence values and the trees were visualized using FigTree v1.3.1 with the eproteobacteria tree rooted by Leucothrix mucro (X87277) and c-proteobacteria rooted with Sulfurospirillum arcachonense (Y11561).

Diagnosis
Body depressed, symmetrical. Carapace calcified, slightly convex, smooth. Rostrum well developed, triangular. Cervical grooves clearly distinct between gastric and anterior branchial regions and between anterior and posterior branchial regions; either side of mesogastric region with small sharply defined pit. Cardiac region small and depressed and separated from branchial regions by shallow grooves. Anterior branchial regions well delimited and separated by short median longitudinal groove; small W-shaped groove over this groove. Posterior branchial regions separated by median longitudinal groove. Intestinal region well circumscribed and separated from branchial regions by distinct grooves. Posterior half of pterygostomian flap with two longitudinal and subparallel carina. Abdominal segments smooth, not folded against thorax; telson folded beneath preceding abdominal somite, with a median transverse suture and a longitudinal suture in the posterior half of telson; uropods spatulate. Epistome unarmed. Mandibular cutting edge with chitinous teeth along incisor process. Sternal plate between third maxillipeds (sternite 3) well developed, strongly produced anteriorly; sternal plate between fifth pereopods (sternite 8) absent. Eyes strongly reduced to small soft tissue, not calcified, movable, without pigment, inserted near antennulae. Antennal peduncle 5segmented, without antennal scale; flagellum of moderate length. Third maxillipeds with crista dentata in proximal half to third of ischium; epipods absent. Chelipeds (pereopod 1) strong, subequal, and greatly elongate; dense corneous spinules along distal portion of occlusal margin. Walking legs (pereopods 2-4) stout, with clawlike dactyli bearing dense corneous spinules along flexor margin. Etymology. Puravida, is a conjunction of the Spanish words ''pura'' and ''vida'' meaning pure life and is a common saying within Costa Rica, in whose waters these specimens were collected in. The genus is feminine as is the species name.

Description
Carapace. 1.3 times as long as broad (including rostrum) (1.5 times without rostrum). Dorsal surface unarmed and sparsely setose with stiff, barbed setae (SBS, Figure 4E, right image). Hepatic and epigastric regions depressed; distinct pit on either side in posterior gastric region. Region between the anterior and posterior portion of the cervical groove with distinct '-shaped groove. Frontal margin oblique, relatively straight, with large tooth near rostrum. Anterolateral margin rounded, lateral margin slightly divergent posteriorly (widest at distal quarter); slightly irregular with small granule at anterior margin of posterior branchial region, immediately posterior to deep groove. Posterior margin unarmed. Rostrum broadly triangular, horizontal, 0.2 times the length of remaining carapace; dorsal surface dorsally convex on either side of median ridge, smooth and sparsely furnished with SBS (dense cluster at apex); lateral margins slightly convex, irregular with granules but straight. Pterygostomian flap lateral surface granulate, with additional short striae in median portion and with two separate longitudinal carinae under posterior branchial region; anterior margin produced into a spine.
Sternum. sternal plastron 1.6 times as wide as long (at midlength), convex lateral margins; surface smooth, sparsely furnished with fine whip-like barbed setae (WBS, Figure 4E left image), lateral margins serrated. Sternite 3, with strong spine at lateral midlength, anterior portion forming a roughly equilateral triangle, with large granules along lateral margins. Sternite 4 2.3 times as wide as sternite 3, anterior margin shallowly concave, anterior midline grooved. Anterolateral margin produced to tooth not overreaching sternite 3 and with large granules laterally. Sternite 6 widest. Posterior margin of sternite 7 deeply concave and deep median emargination.
Abdomen. tergites sparsely setose with SBS. Pleura each with anterior transverse carina and medially depressed; pleural margins of segments 2-6 strongly tapering and with concave anterior margins. Telson 1.1 times as broad as long; distal portions 2.6 times length of proximal portion, distally distinctly bi-lobed with median notch (0.4 times the length of distal portion).
Antennal peduncle. article 1 distomesial margin unarmed, distolateral margin produced to 3 small distal spines; article 2 with strong lateral projection reaching end of article 4, strongly toothed and with additional prominent ventral spine; article 3 laterally and distally toothed, nearly reaches distal end of segment 4. Penultimate article with 3 distal spines (2 mesial and 1 lateral); ultimate article armed with 1 dorsal and 1 ventral spine. Sparsely furnished with thick SBS.
Maxilliped 3. surface smooth and unarmed except for small granules at bases of plumose setae; coxa with fine serration along distal border, fully calcified (not corneous); ischium with 20 teeth on proximal third of mesial ridge. Two types of setae present including comb-row setae (CRS, Figure 1D and1H; Figure 4E center image) on ventral and distal carpus, propodus and dactylus and WBS on dorsal portion of all segments.
Pereopod 1 (cheliped). strongly spinose, 2.6 times as long as carapace (excluding rostrum) (2.1 times including rostrum). Ischium with 2 dorsal distal spines, serrated along mesial margin. Merus with scattered strong spines, most prominent along mesial margin, with 6 distal spines. Carpus with multiple longitudinal rows of spines, with 6 distal spines, length of carpus  slightly longer than palm. Propodus with palm 1.8 times as long as high, with distinct rows of spines. Length of dactylus 0.7 times as long as propodus, proximal tooth on lateral margin; occlusal margin distally hollowed, opposable margins strongly gaping proximally with prominent median tooth; fingers each distally with strong triangular corneous tooth. Ventral portion of ischium and merus densely furnished with WBS setae, remaining surfaces uniformly covered with SBS.

Remarks
Neither of the original two specimens were complete, the male holotype specimen is lacking the posterior left two walking legs (P3 and P4), the male paratype specimen is lacking the right cheliped and the two posterior right walking legs. Sixteen additional specimens were collected during 2009 which included 6 females between carapace length (including rostrum) 4.9 and 24.5 mm and 7 male specimens between 7.4 and 38.6 mm. A single specimen, 4.6 mm, had potentially developing gonopores, indicating that it was female and this specimen and those smaller were classified as juveniles. Female gonopores were clearly visible making identifi-cation of sex possible ( Figure 4F). A single gravid female was collected that had 88 eggs that were approximately 1 mm in diameter.
The overall morphology and morphometrics of the paratypes and other specimens examined agree with the holotype, except for a variance in the shape of the rostrum. The rostrum of one of the paratypes is not evenly triangular but more leaf-shaped along the left margin.
Kiwa puravida n. sp. is similar to K. hirsuta [9] but can be clearly distinguished by the following characters: -frontal margin with prominent tooth at base of rostrum in K. puravida n. sp. (K. hirsuta only has a small tooth; Figure 2E and 4A). -pteryogostomian flap covered with large granules and short striae in addition to longitudinal carinae, anteriorly produced to spine and with short, scattered setae (K. hirsuta only has a few scattered granules in the anterior portion, anteriorly rounded and without setae; Figure 3A and 4B). -the proportions of the anterior and posterior portions of the telson differ with K. puravida n. sp. having a distinctly larger posterior portion (1.6 to 2.4 times a long as anterior portion) compared to K. hirsuta (0.8). Additionally, the distal cleft of the telson is distinctly deeper in the new species with the distal portion of the telson in K. hirsuta being only shallowly emarginated ( Figure 3F and 4C). The extent of the cleft was reduced in a specimen whose carapace was 4.7 mm and should not be used as a distinguishing feature in small specimens. -anterior portion of sternite 3 in the shape of a wider triangle with length-to-width ratio 0.7 (K. hirsuta sternite 3 anterior portion is acute with approximate length-to-width ratio of 1.4; Figure 3F and 4C). -anterior portion of sternite 4 is shallowly concave (K. hirsuta anterior margin of sternite 4 is deeply concave; Figure 3F and 4C). -lateral margins of sternal plastron are distinctly serrated (smooth in K. hirsuta; Figure 3F and 4C). -all articles of the antennal peduncle are distinctly more spiny and with the prominent lateral process on article 2, reaching the distal margin of article 4 (instead of only reaching to midlength in K. hirsuta; Figure 3C). -the coxa of the third maxilliped is distally not strongly produced, only slightly serrate and without corneous spines (distal border strongly produced and denticulate with each tooth with corneous margin in K. hirsuta; Figure 3D). -each tip of the cheliped fingers bear a single corneous spine only ( Figure 4D) while in Kiwa hirsuta bears two corneous tips on the fixed finger. -propodi of walking legs with 4-8 movable corneous spines along distal half portion of flexor margin (K. hirsuta has 11-16 spines along nearly the entire margin Figure 2B-2D).
The various types of setae that cover the surfaces of the body and appendages in both species remain intriguing. Macpherson et al. [9] describe the setation as dense long plumose setae (similar to WBS) mainly on sternum and ventral surface of pereopods and rigid chitinous seta with barbules (analogous to our SBS) inserted in pairs mainly on the merus of the cheliped. Unlike the rigid chitinous setae of K. hirsuta, the SBS of K. puravida n. sp. were covered with bacteria, even though the density of bacteria was much reduced compared to the WBS. The barbules on WBS were not easily visible except under 200x power magnification. The CRS, which were not reported on K. hirsuta, were limited to the carpus, propodus and dactylus of the third maxilliped ( Figure 1D and 4E). The second maxilliped of K. puravida n. sp. had CRS present but with much reduced combs.
The diagnosis of the family Kiwaidae and genus Kiwa is adjusted; the presence of the row of corneous spines on the coxa of the third maxilliped is excluded as this character varies between the two species now known in genus Kiwa. However, the presence of a dense row of corneous spinules along distal margins of the cheliped fingers is added. Furthermore, the gill structure appears to be unique in this family compared to other anomurans, specifically involving the presence of only four pairs of arthrobranchs with gills absent from the third maxilliped. The closely related Chirostylidae have an additional pair of arthrobranchs on the third maxilliped. Albuneids, hippids and aeglids are comparably more similar in having a pair of arthrobranchs on P1-P4 and a single arthrobranch on the maxilliped [36].
Additionally, the two Kiwa species differ in their habitat. The new species was collected from a methane seep at 1000-m depth while K. hirsuta was observed and collected at greater depth (exceeding 2200 m) at a hydrothermal vent. The collection locations of K. hirsuta and K. puravida n. sp. are 6,500 km apart.

Nomenclatural Acts
The electronic version of this document does not represent a published work according to the International Code of Zoological Nomenclature (ICZN), and hence the nomenclatural acts contained in the electronic version are not available under that Code from the electronic edition. Therefore, a separate edition of this document was produced by a method that assures numerous identical and durable copies, and those copies were simultaneously obtainable (from the publication date noted on the first page of this article) for the purpose of providing a public and permanent scientific record, in accordance with Article 8.1 of the Code. The separate print-only edition is available on request from PLoS by sending a request to PLoS ONE, 1160 Battery Street, Suite 100, San Francisco, CA 94111, USA along with a check for $10 (to cover printing and postage) payable to "Public Library of Science".
In addition, this published work and the nomenclatural acts it contains have been registered in ZooBank, the proposed online registration system for the ICZN. The ZooBank LSIDs (Life Science Identifiers) can be resolved and the associated information viewed through any standard web browser by appending the LSID to the prefix "http://zoobank.org/". The LSID for this publication is: urn:lsid:zoobank.org:pub:B59E7704-9AD0-4D45-A5E2-6EC8C-00E9508 The LSID for Kiwa puravida n. sp. is: urn:lsid:zoobank.org:act:36679362-65B9-4CE9-840A-9D1AF4A-9DFD9

Phylogenetic affinities
Both morphology and molecular data support K. puravida n. sp. as a new species of Kiwaidae. The type species of the Kiwaidae, K. hirsuta that was collected from 2228 m at an eastern Pacific hydrothermal vent, is closely related to but genetically distinct from the methane-seep species described here, having a COI nucleotide similarity of 86% and 18 S rRNA gene sequence similarity of 98% (GenBank Accession Number JN383822 for COI and JN367460 for18 S rRNA; Figure 5). Furthermore, both Kiwa species have four tRNAs between COI and COII in the mitochondrial genome (Leu(CUN), Leu(UUR), Ala, Gly); this appears to be a unique feature to Kiwaidae (see [37][38][39]). Shinkaia crosnieri and K. puravida are distinct both morphologically and genetically as well, having only 77% COI nucleotide similarity, which conforms with the family-level differences between these two taxa [40]. Although COI is a poor determinant of higher level relationships, no 18 S rRNA gene sequence data are currently available for S. crosnieri leaving COI data as the only available comparison between Kiwa (Kiwaidae) and Shinkaia (Munidopsidae) families.
A total of 26 full-length 16 S rRNA bacterial gene sequences were collected from a single individual of K. puravida n. sp. that included 17 e-proteobacteria, four d-proteobacteria, two cproteobacteria, and three bacteroides phylotypes (Table 1; GenBank accession JN255989-JN256014). The c-proteobacteria recovered from K. puravida n. sp. were similar to the other epibionts collected from reducing-habitat decapods, including K. hirsuta, S. crosnieri and R. exoculata, being 97% and 98% similar among these four taxa. The sequenced e-proteobacteria fell within the Marine Group 1, Thiovulgaceae, and all of the e-proteobacteria were between 95-98% similar to epibionts of K. hirsuta and S. crosnieri and 95-96% similar to epibionts of R. exoculata ( Table 1). The dproteobacteria and bacteroides were most similar to environmental samples collected from cold seep, hydrothermal vent, and cave systems with complex sulfur cycling.
A phylogenetic analysis of eand c-proteobacteria found that there was poor support for the relationships displayed within the trees except in six host specific clades ( Figure 6). Two clades of eproteobacteria were unique to K. puravida n. sp. (Groups A and B) and were separate from the remaining branches in 82% and 98%, respectively, of the possible trees. Two of the K. puravida n. sp. eproteobacteria phylotypes were more similar to phylotypes from other species than those collected on the same host ( Figure 6; Groups C and D). Intriguingly, within the c-proteobacteria, a Kiwa clade including both species' epibionts was formed (Group E), separating these phylotypes from the remaining similar sequences. Bootstrap confidence for this clade was low at 66% yet in 81% of the possible trees a hydrothermal phylotype (EU265784) and the methane-seep phylotype (JN256007) were found adjacent to each other. The higher level relationship among the c-proteobacteria epibionts from each of the other taxa were not well defined. The poor bootstrap support for the location of the basal branches of both trees was as would be expected for sequences that were similar to each other and thus small sequence differences would cause large impacts to the location of each branch of the tree.
Both the similarity analysis (i.e. Table 1) and the phylogenetic analysis supported the hypothesis that there is an epibiotic fauna that specializes on inhabiting both vent and seep decapods. Thus our full length 16 s rRNA gene sequences corroborate the findings of Goffredi [11] who inserted comparably shorter reads into a full length constructed tree. The presence of closely related cand eproteobacteria on diverse hydrothermal-vent hosts has been suggested to be evidence of multiple horizontal transmission events by vent-species epibionts [8]. Furthermore, Peterson et al. [8] hypothesized that, as it is unlikely that the host species co-occur at any one site, these similarities indicate that there are free living bacterial stages of these symbionts that disperse among hydrothermal-vent sites. Since we have now found both lineages at a methane seep, specifically on K. puravida n. sp, we can include methane seeps as likely stepping stones among hydrothermal vent locations. Furthermore, it appears that the one K. puravida n. sp. individual studied had two distinct clades of eproteobacteria and potentially two clades of cproteobacteria co-occurring on its setae. The presence of both bacterial phyla with high similarity to epibionts at hydrothermal vents suggests that K. puravida n. sp. may possess a duel symbiosis similar to that found on R. exoculata [8]. As our bacterial phylogenetic study results deals with sequences from a single individual and has not identified the distribution of the bacterial phylotypes even within that species (such as through fluorescent in situ hybridization analysis) the results are largely preliminary. Yet it clearly shows the importance of including methane-seep fauna when trying to understand the biogeography of hydrothermal-vent habitats, highlighting the importance of studies such as Goffredi [11].

Nutrition and Farming
During submersible dives with the DSRV Alvin off of Costa Rica, K. puravida n. sp. were observed to have a patchy distribution on the tops and within crevices of carbonate outcroppings, on Lamellibrachia cf. barhami colonies, amongst bathymodiolin mussels, and in pits within carbonate rocks with alvinocarid shrimp ( Figure 1C and 1F). The presence of these other symbiont-bearing species indicates that these areas have active methane-seep fluid release. This new Kiwa species was not observed scavenging food, a strategy previously suggested for its congener, K. hirsuta [9]. In addition, K. puravida n. sp. were often seen using their chelipeds to force shrimp away that got close to the crab; the crab made no obvious attempt to capture these shrimp. However, individuals were commonly observed slowly waving their chelipeds (pereopod 1) back and forth in these areas of active seepage (Video S1).
Both isotopic and FA biomarker approaches provided multiple lines of evidence to indicate that K. puravida n. sp. used its epibiotic bacteria as a main food source. The carbon isotopic composition of this species (d 13 C muscle = 220.1 to 244.2 %, n = 30) was significantly lighter than phytoplanktonic production in this community (d 13 C plankton = 216 to 221 %, n = 18: T-test t = 28.9, df = 42.8, p,0.001), clearly indicating a chemoautotrophic food source for K. puravida n. sp. The light and wide ranging isotopic values of this species indicate that sulfide-and potentially methane-oxidation fuel this chemoautotrophic symbiosis [21]. Kiwa puravida n. sp. 's muscle FA profile was also divergent from the planktonic FA composition and mirrored that of its bacteria-laden setae (Figure 7). The FA plankton composition included an abundance of polyunsaturated fatty acids (PUFAs) with 18% of the FA composition made up by 22:6(n23) and 7% from 20:5(n23); both diagnostic for photosynthetic production ( [20] but see [41]). In contrast, 20:5(n23) was not common in the K. puravida n. sp. tissue, composing only 3 and 5% of the FAs measured in the two specimens tested and this FA was never observed in either of the specimens' setae sampled. 22:6(n23) was present between 4 and 1% in the tissue and was present in one of the specimens' setae sampled, comprising 6% of that species total FAs. 16:2 and 18:2 FAs, which are abundant in R. exoculata symbionts [25], were present on K. puravida n. sp. 's bacteria-setae and tissue sample with the majority being within the tissues; the sum of the two diene FAs comprised 33% of both specimens' tissue samples. 16:2 biomarkers have also been found in limited abundance in mussel tissues [42] and diatoms [43] yet were not found in a species of bathymodiolin mussel from Costa Rica (Thurber, Pers. Obs.) or the phytoplankton sample collected. The relative input of free living microbes to K. puravida n. sp. from carbonate rock was not tested, a food source thought to augment the diet of R. exoculata [15,44]. We did not collect the rock that Kiwa were discovered on, although active carbonate rock from the same seep did not have 16:2 FAs (Thurber, Pers. Obs.). The isotopic composition of this species does not eliminate the possibility that K. puravida n. sp. may be a scavenger or consume symbiont-bearing fauna, but we did not observe them grazing upon other fauna and the unique 16:2 FA within their tissues further makes this lifestyle less probable. The monounsaturated fatty acid, 16:1(n27), a common constituent of sulfide-oxidizing bacteria [45], was always more than four times as abundant in tissue and spines of K. puravida n. sp. compared to the plankton sample. The presence of 16:2 FA, an abundance of 16:1(n27), the similarity of bacteria-laden setae with muscle tissue, and the isotopic composition indicative of chemosynthetic nutrition, support the hypothesis that K. puravida n. sp. 's main food source is the epibiotic bacteria growing on its setae.
In addition to this biomarker evidence, K. puravida n. sp. possessed both morphological and behavioral adaptations to harvest its epibionts. Kiwa puravida n. sp. had specialized CRS setae on its 3 rd maxilliped (a mouth appendage; Figure 1D and present in all 3 individuals inspected via light or scanning electron microscopy). The crab uses its CRS to scrape bacteria off of the WBS which adorn its chelipeds, sternum, and pereopods and then transfers the harvested bacteria into its mouth ( Figure 1G, Video S2). Dissection of a single individual revealed that the bacterial filaments were commonly stuck within these CRS, indicating that the crab was effective at removing the filaments from its setae ( Figure 1H). Bacterial filaments were found in the same individual of K. puravida n. sp. 's cardiac stomach supporting the observation that these bacteria are transferred into its mouth, an observation similar to what has been recorded in R. exoculata [46]. However, a gut-specific microbiota are known to occur in R. exoculata that are similar in appearance to those that grow upon the shrimp epibiotically [47]. Thus a molecular identification is necessary to confirm that the bacteria found in the cardiac stomach of K. puravida n. sp. is the same bacteria that grow on the WBS. Yet the bacteria observed in the gut did have the same morphologic structure as those found growing epibiotically on this new crab species. In addition to bacteria, detritus was observed both attached to the bacteria-laden setae (in all specimens observed) and was present within the Kiwa mouth of the individual that was dissected. As potential photosynthetic fatty acid biomarkers were observed, albeit minimally, both within the tissue and the setae, it seems likely that this species may periodically augment its diet through sweeping up detritus, the source of this photosynthetically-derived particulate organic matter. Yet this did not appear to be the main food source for K. puravida n. sp. as shown by the stable-isotopic composition of this species.
For a species to farm bacteria it must facilitate the growth of its epibionts. In a minimalist sense, K. puravida n. sp. does this by providing an attachment substrate for its bacteria, yet through its continual cheliped movement, K. puravida n. sp. likely facilitates increased epibiont productivity as well. Chemoautotrophic symbionts require access to oxygen from the water column and reduced compounds, i.e. sulfide or methane, from the seep. During periods of carbon fixation, a boundary layer depleted in one or more of these solutes likely develops, which limits epibiont productivity. This is analogous to reef-building corals whose photosynthetic symbionts become carbon limited during periods of high productivity [48]. In coral symbioses, carbon limitation becomes ameliorated at increased current speeds or mixing of the water column which replenishes these boundary layers [49,50]. We hypothesize that K. puravida n. sp. waves its chelipeds to shear off boundary layers formed by their epibionts productivity, increasing both the epibionts and, in turn, their own access to food. As in corals, boundary layers are greatest in areas of reduced flow, such as pits and depressions [51], which is a habitat where K. puravida n. sp. commonly occurs, making this behavior even more adaptive. Thus the cheliped waving motion may increase its epibionts chemoautotrophic productivity and yield.

Ecology and Behavior
In addition to its bacteria harvesting, this species demonstrated intriguing intra-species interactions (Video S3). An individual that appeared to have recently molted due to its minimal bacterial covering, began grappling with a larger specimen that it approached. This ended in a dominance display where the challenged individual forced the challenging individual off the carbonate outcropping while both individuals had their chelipeds spread apart. As decapods commonly reproduce after molting, as has also been observed in the hydrothermal vent S. crosnieri [17], the individual that was forced off may have been inseminated Figure 6. Neighbor-Joining phylogenetic tree based of 16 S rRNA of eand c-proteobacteria of epibionts found on Kiwa puravida n. sp. Tree's were rooted using Leucothrix mucor (X87277) and Sulfurospirillum arcachonense (Y11561) for eand cproteobacteria, respectively, and the roots are not shown. Sequence from Table 1 were excluded from these trees if they were less than 1000 base pairs. Bootstrap confidence, based on 3000 iterations, are indicated in grey. Scale bar equals number of substitutions per nucleotide. doi:10.1371/journal.pone.0026243.g006 during this display or this may have been a behavior demonstrating how this species competes for space in areas of active seepage.
A remaining question is: do K. puravida n. sp. setae act as mechanoreceptors allowing it to identify areas of active seepage? A similar role has been suggested for the setae of both S. crosnieri and K. hirsuta [52]. Through analysis of the setal structure of S. crosnieri and K. hirsuta and the mitten crabs, specifically Eriocheir sinensis and E. japonica, all of which are characterized as having dense setae, Ehrlich [52] concluded that it is unlikely that the setae of these species posses a mechanoreceptor ability. As K. puravida n. sp. has three distinct setae types, morphological analysis would be a useful approach to see if any of the distinct setae forms act as these mechanoreceptors in contrast to the previously studied highlysetose anomurans.
The nutritional role of epibionts that grow on hydrothermal vent decapods has remained perplexing. Kiwa hirsuta is known from only one specimen thus many analyses have not been performed to identify its food [9]. Rimicaris exoculata has not been observed harvesting its epibionts and doesn't possess the morphological structures to do so [9]. Shinkaia crosnieri posses the behavior and morphological adaptations to consume its epibiotic bacteria and stable isotopic evidence further supports that it consumes its epibiotic bacteria, but this species does not have a FA signature indicative of a diet based on the bacteria that grows upon it [18]. The multiple lines of evidence that we present here for Kiwa puravida n. sp. together demonstrate clearly that this new species of yeti crab farms the bacteria that grow upon it in a novel form of symbiosis.

Supporting Information
Video S1 Kiwa puravida at Mound 12, Costa Rica demonstrating the rhythmic waiving of its chelipeds.

(MP4)
Video S2 Kiwa puravida harvesting its symbionts with its 3rd maxilliped. After scraping bacteria off of its pereopods and sternum K. puravida's transfers the bacteria to its mouth with the aid of its 2nd maxillipeds. (MP4) Video S3 Two individual of K. puravida performing either a courtship or competitive display. (MP4)