Study area

The Pacific margin of Costa Rica is an offshore convergent margin, where fluid venting has long been documented in the area with evidence for more than 100 seeps [37, 38]. These seeps are associated with landslide scars, seamount-subduction related fractures, faults, and many mounds. Over 60 mounds 50–100 m high and up to 1 km wide at the base were identified in the region [38]. Among these, Mound 12 (8°55.8’N, 84°18.7’W) is located at around 1,000 m water depth [39], below the oxygen minimum zone [11]. The benthic fauna at active seepage sites is characterized by carbonate rocks hosting the yeti crabs Kiwa puravida and bathymodiolin mussel beds [11, 38, 40], as well as smaller animals such as neolepetopsid limpets and provannid snails [13]. Transition sites are located at the intersection of seep and background environments (described above) and are identified by the presence of scattered seep-specialist megafauna (e.g., Bathymodiolus mussels and yeti crabs), shell remains of seep-associated bivalves (e.g., Bathymodiolus mussels and vesicomyid clams), and heterogeneous soft and hard bottom with partially buried but visible carbonate rocks [17] hosting seep species (e.g., various gastropod molluscs and annelid worms) and background species (e.g., amphipods, ophiuroids), which are more commonly found at seep and background sites, respectively [13].

Experimental design

A colonization experiment was performed at Mound 12 on the Pacific margin of Costa Rica to test the interaction between methane seeps and different CBEs (organic falls). Bone (cow bone) and wood blocks (untreated Douglas Fir blocks 10 cm x 10 cm x 25 cm) were deployed in two active (2 bones, 4 wood blocks) and two transition (2 bones, 4 wood blocks) sites in January 2010 on R/V Atlantis cruise AT15-59 with the ROV Jason at Mound 12 (Fig 1). Active sites had extensive carbonates colonized by two species of Bathymodiolus mussels [41] and other macrofaunal seep species, and transition sites had partly buried carbonates with soft sediment and scattered dead mussels (Fig 1). Within each site, one bone and two wood blocks were deployed within 25–50 cm of each other. Substrates were recovered after 7.4 years with the submersible ALVIN in May 2017 (AT37-13; Fig 2). Substrates were placed into individual containers within Delrin bioboxes on the Alvin basket to avoid cross contamination during recovery. Defaunated carbonate rocks were also deployed for 7.4 years at the same active (n = 4) and transition (n = 4) sites (from rock experiment, see [13]) for comparison of colonization rates among different substrates. These defaunated carbonate rocks had been collected from Mound 12 in 2009. At that time, the fauna were removed upon collection, and the rocks were left out to dry (see [13]). In situ, unmanipulated carbonate rocks that were collected in 2017 at active and transition sites were used as control samples for the experiment (see [13]). Here we define control as what we expected the colonizing community to look like at the end of the experiment if full recovery to baseline state was achieved. Here we present the results for the wood and bone experiments, comparing to the results of the rock experiment in Pereira et al. [13].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Sampling locations of the experimental substrates and photos in situ.

Experimental bones, wood, and carbonate rocks were deployed for 7.4 years (2010–2017) at (B) active and (C) transition sites at Mound 12. The map was plotted in R Software using marmap 1.0.6 package ‘getNOAA.bathy()’ function, which queries the ETOPO1 database hosted on the NOAA website in the public domain.

https://doi.org/10.1371/journal.pone.0271635.g001

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Photos of the experimental substrates upon recovery.

Experimental bones, wood, and carbonate rocks were deployed for 7.4 years (2010–2017) at (A-C) active and (D-F) transition sites at Mound 12.

https://doi.org/10.1371/journal.pone.0271635.g002

In addition, we have observed natural organic falls in the vicinity of methane seep sites off the Costa Rican margin; these included natural wood pieces at Mound 12 (~1,000 m) and Jaco Scar (~2,000 m), and a swordfish skeleton at Quepos Slide (~400 m) (Fig 3). We collected macrofaunal samples from the natural wood at Mound 12 in 2017 and 2018 and at Jaco Scar in 2017 for taxonomy and isotope analyses following the same methods described below for the experimentally deployed substrates.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Natural organic falls observed in the vicinities of methane seeps off the Costa Rican margin.

(A) Wood found at Jaco Scar (~2,000 m) and a (B) closeup. (C) Wood found at Mound 12 (~1,000 m). (D) Swordfish skeleton found at Quepos Slide (~400 m). Although we only collected samples from the woods, the skeleton observation illustrated here further highlights the common occurrence of natural organic falls in the vicinities of methane seep sites off the Costa Rican margin.

https://doi.org/10.1371/journal.pone.0271635.g003

Sample processing

Shipboard and laboratory processing followed Pereira et al. [13]. In summary, substrates were photographed intact upon recovery, the associated fauna was removed from each substrate and its recovery container and sorted and identified to the lowest possible taxonomic level. Tissue subsamples were collected for stable isotope analyses to examine trophic diversity and reliance on chemosynthetic production. Bones and rocks were wrapped in aluminum foil to determine the approximate substrate surface area (in cm²) later in the laboratory by dividing the total weight of the foil used to wrap the substrate by the average weight of a 1 cm² piece of foil. Foil wrapping was the most accurate way to estimate surface area due to the rugose, pitted and lumpy nature of the substrates. Initial wood block surfaces were smooth, and area (1047.06 cm²) was readily calculated from wood dimensions (10 cm x 10 cm x 25 cm). Substrates were left at room temperature overnight in filtered seawater for additional fauna to crawl out; the fauna were then preserved in 95% ethanol.

For constraining food webs, we collected samples for stable isotopic analyses (δ¹³C and δ¹⁵N) of the potential food sources: Water samples were collected from Mound 12 at the surface (< 5 m deep) and near-bottom (<15 m above bottom) via Niskin bottles on a CTD Rosette and filtered on glass microfiber filters to characterize particulate organic carbon (POC). Bacteria (mainly Thioploca) were scraped off the rocks, and subsamples of each of the rock, wood and bone substrates were powdered using a ceramic mortar and pestle.

At Scripps Institution of Oceanography, the preserved samples were re-sieved and sorted, following Pereira et al. [13]. Species-level IDs based on morphology were considered unreliable since subsequent molecular sequencing showed that some taxa were cryptic species (G.W. Rouse, personal observation), and, thus, Annelida and Mollusca were identified at the family level, Arthropoda (crustaceans) at the order or infraorder level, Cnidaria at the order level, and Echinodermata at the class level. Other macrofauna that occurred in low abundance included Nemertea and Platyhelminthes identified to phylum, and other mollusks (aplacophorans and polyplacophorans) and crustaceans (ostracods and mites). Animals were counted, adding the counts of animals that were removed at sea upon recovery for genetic or isotopic analysis to obtain totals for each substrate. Tissue subsample preparation for isotope analyses followed Pereira et al. [13] and δ¹³C and δ¹⁵N measurements of tissues, food sources and substrates were carried out at Washington State University (WSU).

Data synthesis and statistical analyses

Densities of the total macrofaunal community on the different substrates were calculated for a standard surface area of 200 cm² [11, 13], which represents the surface area of an average-sized carbonate rock at Mound 12. As the data were not normally distributed nor showed a homogeneous variance, Kruskal-Wallis tests followed by Dunn’s test using the Benjamini-Hochberg adjustment [42] were performed to check for variability in densities among substrates, and Wilcox tests were performed to check for variability between seepage activity within substrates. Community composition was analyzed by percent composition and density per 200 cm² by taxonomic group. Multi-dimensional scaling of Bray-Curtis dissimilarities, two-way ANOSIM and two-way SIMPER analyses were conducted using Primer v7 [43] after standardizing the data by the total number of individuals and fourth-root transformation.

The stable isotope data were also not normally distributed and did not show homogeneity of variance, thus Kruskal-Wallis tests followed by Dunn’s test using the Benjamini-Hochberg adjustment [42] were performed to check for variability among substrates, and Wilcox tests were performed to check for variability between seepage activity within substrates. Community-level isotope metrics [36] were generated using Stable Isotope Bayesian Ellipses in R (SIBER) package [44], following Pereira et al. [13], to evaluate trophic diversity, trophic redundancy, and niche breadth. Similarities in trophic structure among substrates and seepage activity were determined by computing pairwise ellipse overlaps based on Bayesian posterior estimates using 2 chains of 20,000 iterations with a burn-in of 1,000 and thinning of 10 [45].

Organic-fall colonizers at seep transition zones

At transition sites, the community composition on wood and bone was similar to natural wood- and whale-fall community composition, which rely on decay of the organic substrates, with high abundance of mainly annelids, and a few peracarid crustaceans and echinoderms (Fig 5A) (see [20, 46]). Unfortunately, some basal food sources may be missing in our isotopic analyses, given that some macrofauna isotopic values do not fall within the range expected for the food source isotopic values we have, and, thus, we could not perform mixing models analyses. However, the higher carbon and nitrogen values at transition sites (Fig 8) reflect less use of methane-derived production and locally fixed nitrogen. Interestingly, macrofauna on wood, carbonate and bone at transition sites seem to be feeding on identical food sources (most likely POC and chemosynthetic bacteria utilizing sulfide from organic-matter decay) (Fig 8) as suggested by the high niche overlap among substrates relative to active sites (Fig 9). This is notable, given that the different substrates shared fewer taxa at transition than at active sites (i.e., lower overlap of taxa) (Table 1).

Bones at transition sites were dark-colored indicating reducing conditions [23] and covered in white bacterial mat (Fig 2D). This reducing state has been seen before on an implanted gray-whale carcass on the California margin after 6.8 years at the seafloor in a sulfophilic stage indicated by the presence of thick white and yellow bacterial mats [18], and on a whale vertebrae from a natural whale fall in the Southwest Atlantic, where the bone was not colonized by Osedax and was intact with chemosynthetic bacterial mats [47], considered by the authors to be an intermediate reducing state [20]. The intact state of our colonization bones after 7 years is not a surprise since bone preservation is increased in the absence of Osedax [12].

Hesionid and dorvilleid polychaetes were the dominant families on our bones at transition sites after 7.4 years. Previous studies reported that whale-fall macrofaunal communities are dominated by these families of polychaetes, as well as capitellids, including many species of dorvilleids especially from the genus Ophryotrocha and Parougia [23, 47–49]. Our observation suggests that these annelids are not only successful colonizers of whale falls in the short-term but also in the long-term. Animals on natural bones in an intermediate reducing state relied on carbon derived from free-living chemosynthetic bacteria on the bone [20], supporting our hypotheses that animals were feeding on chemosynthetic bacteria on our bones.

Woods blocks at transition sites were lightly bored by wood-boring bivalves (Fig 2E), although we found only one live Xylophaga (see discussion below). The bivalve tunnels provide structural heterogeneity for the colonizing macrofauna [50]. Provannid snails were the dominant taxon on wood at transition sites (38.48%) in our 7.4-year long experiment; they accounted for 20–30% of the gastropods on wood from a previous colonization 10.4-month experiment at Mound 12 [21]. Lepetodrilid limpets showed the opposite pattern; they accounted for ~20% of the wood community after 10.4 months [21] but only 4.2% after 7.4 years. Previous studies reported provannid snails to be less resilient under low seepage activity in the short term [8, 13], and lepetodrilids were more abundant than provannids on the carbonate rocks deployed for 7.4 years (see [13]). Both provannids and lepetodrilids are bacterial grazers or get their nutrition from chemosymbionts [51–55]; provannids appeared to be favored by the bacterial community that established on the wood.

Our woods also showed high abundance of dorvilleid polychaetes and some skeneid snails, as observed in a 15-month colonization experiment [23], sharing the substrate with seep-associated species (e.g., provannid snails and neolepetopsid limpets). Dorvilleids and skeneids most likely feed on bacterial mats on the wood [23], most provannid snails graze on bacteria as mentioned above [55], and neolepetopsid limpets also show flexible diets feeding on bacteria and POC [56, 57]. Our isotope data agrees with these, where animals on wood at transition seem to be feeding on POC and/or bacteria (Fig 8). Nevertheless, the dark coloration on the bottom half of the wood, which was likely buried in sediment, indicate a reducing state. A previous study showed that in the absence of Xylophaga, the microbial community can produce hydrogen sulfide and, thus, support a chemosynthesis-based community while degradation is slow [58].

Active seepage influence on organic-fall colonizers

At active sites, fauna seems to depend less on the organic matter provided by the organic substrates for the following reasons: Although we are missing some food sources, we can infer that there was limited direct use of the bone, wood, or rock substrate organic matter as a food source by the macrofauna at active sites (Fig 8A). If animals were using the substrates as food source, we would expect to see higher macrofaunal isotope values on the organic substrate, especially nitrogen, to reflect the higher values of the substrates compared to chemosynthetic production values (Fig 8). In addition, there was a clear shift in trophic structure from transition to active sites, with lower carbon and nitrogen stable isotope values at active sites, indicating more use of chemosynthetic production (Fig 6).

Lepetodrilids, provannids, anomurans (mainly yeti crabs), and neolepetopsids showed high abundance on wood and bone. These taxa are commonly found associated with hard substrates at active seepage, where seepage activity is a strong determinant of the macrofaunal community composition [11, 13]. Thus, at active sites, it seems that seepage activity overrides the substrate influence, i.e., seep species can colonize wood and bone but not necessarily to feed directly on their organic-rich content. This conclusion is also supported by the similar trophic structure across substrates at active sites, especially for wood and rock (Figs 7–9). In addition, communities on bone and wood attained the same level of dissimilarities between active and transition sites as observed on experimentally deployed carbonate rocks [13], and higher abundance of seep species at active than at transition sites (Table 1), highlighting the stronger influence of seepage activity on community composition across substrates. It seems that the communities associated with organic substrate at active seepage are mainly influenced by the energy available from the seep chemosynthetic production (rather than the organic-rich substrate itself) and the niche dynamics shaped by seepage activity [46, 59, 60].

The similarity between wood and rock could be explained by a delayed wood degradation at active sites. Initial decay of wood in the deep sea usually is mediated by Xylophaga, and microbial succession and wood decomposition depend on environmental conditions, especially oxygen [2]. As mentioned above, we did not find Xylophaga on the wood at active sites, and seep communities use 2x more oxygen than non-seep communities [61].

Although we did not measure the substrates before the experiment to account for decay, a more rapid decay of the bone could be a possible explanation for the lower niche overlap between macrofauna of the bone and the other two substrates at active sites. However, our bones at active sites did not show signs of degradation, such as black discoloration that indicates reducing conditions [23]. It is possible that this niche difference between bone and wood/rock resulted from the low sample size, as the few species that we collected from the bones for isotope analyses did not reflect use of bone or differ in isotopic composition from the same species on wood and rock (S1 File).

Lack of organic-fall specialists

Previous experiments with wood deployed for 2 weeks to 2 years suggest that the wood-boring xylophagaid bivalves are early colonizers [23, 46, 62, 63]. The absence of xylophagaids in our 7.4-year experiment, but presence of a single Xylophaga individual and burrows in the wood blocks at transition sites reflect past Xylophaga activity at earlier successional stages.

This lack of wood specialist species could be explained by a lack of larval supply. Young et al. [23] suggested that local abundance of wood falls is more important to wood specialists than larval transport from distant sources. However, xylophagaid bivalves have been reported from natural wood off Costa Rica, including the ones we collected at Mound 12 and Jaco Scar in 2017 and 2018 and near Cocos Island in 2019 (G.W. Rouse, personal observation, e.g., https://sioapps.ucsd.edu/collections/bi/catalog/M16099/?q=xylophagaidae&image=Any&idx=305, https://sioapps.ucsd.edu/collections/bi/catalog/M16111/?q=xylophagaidae&image=Any&idx=306). Predation also seems unlikely to explain the absence of Xylophaga since we did not find common xylophagaid predators, e.g., chrysopetalid, and polynoid polychaetes, gastropods, and flatworms [64, 65] on our wood blocks.

There are then two other possible scenarios that could explain the lack of organic-fall specialists at actively seeping sites: (1) Our wood blocks were at the end of the sulfophilic stage upon recovery [52, 63], and/or (2) seepage activity creates a toxic environment for organic-fall specialists.

Scenario 1: In experiments carried out in other regions, wood blocks from shorter experiments were much more degraded than ours. Wood blocks deployed off the coast of Washington for 15 months were crumbling upon recovery due to heavy boring by mostly Xylophaga oregona [23], and others deployed in the Southwest Atlantic hosting multiple xylophagaid species lost up to 75% of their volume after almost 2 years [46]. Our wood blocks were not highly degraded, whereas the previously described experiments yielded wood degradation described as ‘crushable by hand’ [23, P.Y.G. Sumida personal communication for Saeedi et al. 46], and ours were not. As lightly-bored wood blocks may represent an early colonization stage where space and organic matter are not limited yet [23], it is then unlikely that our wood blocks at either active or transition sites, were at the end of the sulfophilic stage.

Scenario 2: Considering this low degradation state, it is striking that we only found a single Xylophaga sp. individual and empty burrows on our wood blocks at transition sites, and none were seen at active sites. As wood-borers are excluded in anoxic conditions [66], seepage activity could create a low-oxygen environment that is toxic for xylophagaid bivalves that were not able to colonize the wood at active sites. At transition sites, where there is less seepage activity, xylophagaid larva were able to settle on the wood blocks but it is unclear why they did not survive. However, in the absence of Xylophaga, the wood becomes anoxic due to microbial fermentation [58], possibly impeding new xylophagaid settlement within the 7.4 years of our experiment.

Grupe [21] also hypothesized that nearby seepage could have caused the lack of wood-boring taxa on their wood deployed for 10.5 months at active and transition sites at Mound 12. Surprisingly, other experiments with wood in the Mediterranean did recover high densities of wood-boring bivalves Xylophaga atlantica and Xyloredo ingolfa on wood pieces deployed for 1 year at a seep [63] and Xylophaga dorsalis on ~1–2 years experiments at mud volcanoes [67]. However, on both experiments the deployment sites seem to be heterogenous with soft sediment and some carbonates, or mostly covered in soft sediment with some dead mussels, which we would consider a transition site (see photos in [63, 67]).

In the case of the bones, Osedax is considered an early but short-term colonizer [68]. We did not find Osedax on our bones nor signs that they were previously there (there were no burrows). Parallel to the wood, a lack of larval supply or bone sources seems unlikely since Osedax has also been collected from natural bones and other experimental bones deployed off Costa Rica, including the swordfish skeleton shown in Fig 3D from a low oxygen site (G.W. Rouse, personal observation). It is possible then that seepage is also toxic for Osedax. Grupe [21] deployed bones at active and inactive sites in Hydrate Ridge, Oregon, for 10.5 months and found that the bones at inactive sites were colonized by three species of Osedax, but not the bones at active sites. However, it is also important to point out that Osedax might show bone preference; vertebrae colonized and not colonized by Osedax have been collected from the same whale fall in the Southwest Atlantic [47], and Osedax were denser on mandibles and thoracic vertebrae and less prevalent on the ribs, sternum, humerus, lumbar vertebrae, and caudal vertebrae on a natural whale fall in Antarctica [69]. Thus, the lack of Osedax in our bones could also have been the result of bone characteristics.

Another location where toxic seepage could explain the lack of organic-fall specialists is the Southwest Atlantic, where six landers with bone and wood were deployed for almost 2 years. At the end of the experiment, wood and bones from one of the landers were not colonized by wood-boring bivalves [46] or Osedax [70], respectively. Interestingly, as previously mentioned, individuals of the shrimp Alvinocaris muricola [24] and the snail Cordesia sp. [25], two species previously know only from seeps, were collected from the same wood and bones.

Organic falls in Costa Rica and linkages to methane seeps

Wood elements can enter the ocean via river systems that carry wood after storms, landslides, hurricanes, and floods. In 2005, hurricane Rita caused major flooding and landslides in Costa Rica, supplying river channels with large quantities of trees [71]. The high wood decay rates in tropical regions (due to higher rate of biological activity, microbial diversity, and year-round warm and moist conditions) allow for breakdown of wood into smaller pieces that can be more easily transported [72]. Tropical storms and associated rapid streamflow also lead to higher wood mobility [72]. The increase in frequency and intensity of extreme climate events will likely increase the transfer rate of wood debris from the land to the marine environments [73]. Moreover, tropical forests in Latin America are threatened by deforestation due to agricultural expansion [74]. In Costa Rica, once the transported wood reaches the ocean, it can be deposited in proximity to the many methane seeps on the continental slope [75].

Similarly, whale falls are commonly found on continental margins along cetacean migration routes [2]. In Costa Rica, humpback whales (Megaptera novaeangliae) are the most abundant species of whales, and both humpback and blue whales (Balaenoptera musculus) occupy breeding grounds, where they reproduce and give birth during the winter [76, 77]. Dolphins (family Delphinidae) are also an extremely diverse group, with some year-round resident populations [76]. Despite the economic value of these cetaceans to local communities via tourism (whale watching), they remain affected by incidental mortality in fisheries, direct killing for shark bait, and noise pollution [76]. Other carcasses such as large pelagic fishes can also create food-rich islands in the deep sea [78, 79]. In Costa Rica, important fisheries include yellow fin, big-eyed, and skipjack tuna, swordfish, marlin, dolphin and shark [80]. Thus, there is great potential for organic falls in the Costa Rican continental margin, in proximity to methane seeps, supported by our personal observations (Fig 3) Although we used cow bones for this experiment, which have lower lipid contents than whale bones [81], the bone-eating worm, Osedax sp., previously thought to be a whale-fall specialist, has been observed colonizing experimentally deployed cow [82], fish [83], alligator [84], bird, and turtle [85] bones, showing that many kinds of bone may support chemosynthetic communities.

The presence of seep fauna such as neolepetopsid limpets on the natural wood, and other seep fauna (e.g., yeti crabs, lepetodrilid limpets, and provannid snails) coexisting with fauna commonly found in organic falls (e.g., dorvilleid and capitellid polychaetes) (see [20]) on the experimentally deployed wood blocks and cow bones supports the hypothesis of organic falls functioning as stepping-stones for dispersal of seep fauna. Transition zones expand the influence of the seep and, as seepage activity might be toxic for organic-fall specialists, it delays the degradation of organic substrates, and seep animals colonizing these at transition zones are not overwhelmed and limited by organic fall specialists [67]. Although these seep species do not seem to be feeding on the organic content of the substrates as do Osedax and Xylophaga, they are able to colonize them and feed on bacterial mat that establishes on the substrates during the sulfophilic stage. The ability to colonize and reproduce on different substrates increases the likelihood that larva of seep species will find a settlement site [86], thus wood-fall and whale-fall substrates can create ephemeral chemosynthetic habitats that attract a species-rich assemblage with a range of trophic niches, functioning as stepping stones for vent and seep species [18, 20, 87–89]. This habitat heterogeneity can be exploited by both opportunists and endemic fauna (including bone or wood trophic specialists), acting to sustain the persistence and evolution of chemosynthesis-dependent species [90].