ABSTRACT Seamounts are topographic features that extend upwards a thousand meters or more from the ocean floor, but peak below the water's surface. Initial studies of seamount biota proposed a high level of endemicity on seamounts. However, further work has called this assumption into question. For the first time, global biogeographic ranges have been gathered for a marine assemblage, namely that of 107 the Davidson Seamount (-35.7° N, 1ZZ. W). Only five percent of Davidson fauna are localized to this seamount, with the majority observed over a continuum of geographic ranges. As further evidence against the endemicity hypothesis, eighty percent of the megafauna have been found over a thousand kilometers away from Davidson Seamount. Moreover, taxonomic and sampling improvements have the potential to increase this figure, which suggests that most deep-sea assemblages are composed of widely distributed species. INTRODUCTION There are estimated to be over 14,000 (Wessel 2001, Kitchingman and Lai 2004 in McClain 2007) seamounts in the ocean, but only around two hundred sites have been sampled. The overall biology on seamounts is not well characterized because the technology required to sample the more remote and deep summits is relatively complex and expensive. Within the seamount literature, a landmark Nature publication by Richer de Forge et al. (2002) heralded the discovery of more than 850 macro-and megafaunal species from seamounts in the Tasman Sea and southeast Coral Sea. The statement that 29-34% of these were new to science and potential seamount endemics emphasized the novelty of the find. This went above the earlier estimation of maximum seamount endemism at 15.4% among invertebrates and 11.6% among fish (Wilson and Kauffman 1987). The possible existence of endemic fauna, which by definition is found at a particular location and nowhere else, has been explained in numerous ways. The term is used frequently in island biogeography literature where unusually high rates of endemic species are reported. For example, the proportion of indigenous flora that is endemic to the well-studied Galapagos Islands is estimated at 43%. If Hawaii's native fauna is broken down into taxonomic groups, 81% of birds are endemic, as are 99% of mollusks and insects (Whittaker and Fernandez- Palacios 2007). Many of the extinction or near extinctions that have been reported in the terrestrial and marine environment share one thing in common: a small range size (Roberts and Hawkins 1999). Therefore, endemic species may provide an impetus for research and conservation since they are often considered vulnerable. The mechanisms put forth to describe the process by which this condition could arise on seamounts describe processes analogous to islands. Seamounts are portrayed as an isolated habitat that is surrounded by strongly contrasting mediums. For example, hard seamount substrate in the open ocean provides a sharp contrast to the surrounding abyssal plains of soft sediment and the deep-water column. The patchy nature of hard substrate may create isolating conditions favorable for the creation and maintenance of endemic species in taxa that need this medium for settlement. A second, similar idea emphasizes the shallower depths found on seamounts in relation to the surrounding abyssal plain. Depth zonation is one of the most established trends within benthic deep-sea biology (Gage and Tyler 1991). The third hypothesis emphasizes current regimes and other hydrographic conditions as the mechanism to prohibit dispersal and retain larvae on seamounts. While these hypotheses remain largely untested (McClain 2007), high numbers of species endemic to a single seamount have not materialized. The publication of more specific studies from the same region as Richer de Forges' original study have provided evidence contrary to endemicism. Samedi et al. (2006) addressed the above suggestions by testing several species of well-known deep- sea crustaceans for fragmentation of distribution that would result in phylogenetic structure. Contrary to the isolation model, the haplotype frequency did not correlate to location, which would indicate that populations remain genetically connected. Using a different prototypical organism, O'Hara (2007) compared the species richness and endemicity rates of ophiuroids between seamounts and continental shelves and found no difference between the two habitats. Both of these studies discredit the single seamount endemicity hypothesis, but only in a limited context: a specific taxonomic group in a specific place. Lundsten et al. (In review) had a similar result for three seamounts along the California coast, another specific place, but did look at all fauna found within the study system. 1 hope to expand these previous studies to the distribution of all megafauna found on Davidson Seamount at a larger spatial scale. Davidson Seamount will act as a model system to test the endemicity hypothesis at the community level through analysis of global biogeography data. The greatest linear distance from Davidson Seamount at which a particular taxon was seen will be used as a tool to compile a frequency distribution of species range size data for the megafaunal composition of Davidson Seamount. This seamount is located around ninety kilometers off the central California coast, rising 2400m off the ocean floor to a series of peaked summits separated by sediment filled troughs nearly 1200m below the ocean's surface (Lundsten et al. In review; Davis et al. 2002— see fig. 1 for a map of the California coast that shows the relative position on the coastline, the northeast-southwest alignment and the general shape of the seamount). Megafauna were previously identified from deep sea Remotely Operated Vehicle (ROV) dives to Davidson Seamount in 2000, 2002, 2006, and 2007 as part of a partnership of several organizations: the Monterey Bay Aquarium Research Institute (MBARI), the Monterey Bay National Marine Sanctuary (MBNMS), and the National Oceanic and Atmospheric Administration (NOAA). The extent of sampling has also been included in fig. 1, represented by colored dots. MATERIALS AND METHODS The megafaunal community on Davidson Seamount was identified to the lowest possible taxonomic level using in situ video frame grabs and digital still images of specimens that were identified by taxonomists (Lundsten et al. In review). This process relies on morphological differences, which can be more helpful for some phyla than others. As a result, the phylogeny has not always been worked out to a species level for many of the organisms on Davidson. Broad identifications that lump many potential species together have been disregarded for this analysis, as the distributional patterns are scrambled together. Where a difference can be ascertained, the labels reflect only the highest confirmed taxonomic level. Examples include Family Hormathiidae sp., Porifera sp. 15, Class Holothuroidea sp. 6 or Order Poecilosclerida sp. The resulting 156 identifications were then used as a reference to gather the coordinates for all previous sightings within the extensive Monterey Bay Aquarium Research Institute video database (VARS). VARS is a compilation from over twenty-six years of Northeast Pacific deep- sea RÖV observations. Coordinates were also collected from two other databases: FishBase.org Point Maps and SeamountsOnline, managed by the University of California San Diego. Published species ranges in the literature were transformed into coordinates using GoogleEarth and the National Geospatial Intelligence Agency's Undersea Formation Coordinates (http://earth-info.nga. mil/gns/html/uf gaz jun07.pdf). Utilizing the assumption that the earth is a perfect sphere with a radius of 6378.0 km, all coordinates were transformed into a linear distance off Davidson with an applet hosted by Northern Arizona University at http://jan.ucc. nau.edu/-cym/. The maximum linear distance (in km) off Davidson published for the most conservative taxonomic level was used to create a frequency distribution of the geographic range sizes and the corresponding probability distribution using JMP Statistical Software version 5. The median value was used in this study instead of the mean value as a descriptive statistic for the average of a dataset because it is not as influenced by outlining points. The frequency distribution was subdivided into phyla, and shown in a single figure as a comparison of boxplots. RESULTS The frequency distribution of published maximal linear distances is presented in Fig. 2a. It is a right-skewed graph with a large spread. There is a high level of variation within the community, with some species only recorded on Davidson (0 km) whereas others have been recorded at the farthest possible linear distance from Davidson in the Indian Ocean 19888.6 km away. Fifty percent of species on Davidson Seamount have been observed over 1879 kilometers away (median of dataset). These types of patterns become more obvious when the data are transformed into the probability distribution seen in Fig. 2b. A near vertical slope denotes a cluster of data points, indicating furthest observations of multiple species were seen within close vicinity of each other. The most noticeable case of this is around 1500 kilometers. These patterns will be explained further in the discussion section, as they are a logical extension of the original frequency distribution of linear distance off Davidson. This frequency distribution was also broken down into phyla in Fig. 3. Only one Ctenophora was seen, so no further analysis of this group will be made. Of the remaining phyla, all had at least one specimen with a geographic range limited to Davidson or nearby. The boxplot of different Chordates on Davidson is characterized by roughly equivalent distances between each quartile and the median, indicating a uniform continuum of observations. This is also reflected in a significantly higher median, greater than the third quartile of all other groups except Arthropoda. Arthropoda has a median of 17, higher than other phyla. The Porifera, Cnidaria, Echinodermata and Arthropoda boxplots have a shorter distance between the first quartile and the median, which indicates a right skewed distribution similar to the frequency distribution as a whole. The Mollusca are the only phylum in which the distance between the first quartile and the median is greater than the distance between the third quartile and the median. The maximum linear observation for this group was 3326.1 kilometers, and does not have the tail' characteristic of the frequency distribution for all megafauna on Davidson seamount. Interestingly, the median value for this group is similar to the Porifera and Echinodermata, which have different shaped distributions (1369.9, 1202, and 1480, kilometers respectively). This also confirms the robustness of the median as an average descriptor for this dataset, since it is not skewed towards outlying points. DISCUSSION The dominance of widespread species at this location is best illustrated by the fact that eighty percent of the megafauna groups on Davidson Seamount have been observed over a thousand kilometers away. Fig. 2b allows us to examine the continuum of ranges more closely. The percentage of megafauna with geographic ranges greater than 1500 kilometers falls off sharply as this excludes the species seen as far away as the Canadian border or the Gulf of California. In contrast, the percentage of megafauna with ranges between 3-5000 km slowly declines. Two localities underlie this trend: linear observations from Alaska fall between 2300-3000 km and Hawaii has a range between 3500-4500 km. Only twenty-five percent of the megafauna on Davidson has been seen farther away than the western coast of North America or Hawaii. This may be an artifact of relying heavily on the MBARI ROV dataset, which has only sampled as far as Hawaii. The steepness of the slope around 8500 kilometers is a result of another group of observations. This is past the farthest linear distances of observations in the Sea of Japan and along the European coast. One of the weaknesses of using linear distance off Davidson Seamount as a proxy for range size is that remote Pacific data are intermixed with Atlantic data but represent two entirely different patterns. The Sea of Japan records probably indicate a Rim of Fire' Pacific distribution, while the few species that have been recorded along the Atlantic represent a larger dispersal and should probably be in the context of my next observation. Two fifths of the twenty five percent of megafauna found outside the coast of North America and Hawaii (or ten percent of all Davidson megafauna) has been witnessed at geographic ranges greater than 13000 km, either in the Antarctic or Indian Ocean. Given their presence in multiple oceans, these should be considered cosmopolitan species, and will probably be observed elsewhere with increased sampling. The data above present strong evidence against the endemicity hypothesis, and a locally distributed megafauna in general. However, if Davidson Seamount is roughly forty-two by thirteen kilometers, then taxa endemic to Davidson Seamount must have linear distances less than approximately forty-four kilometers (maximum length using the Pythagorean theorem). A relatively small percent of the assemblage falls within this category. In contrast, in examples of species-range size distributions of beetles, wildfowl, and primates from the terrestrial literature (Gaston 1996, 1998) the bin that has most species is the smallest. This may be because the deep ocean is not a dispersal barrier in and of itself (McClain 2007). The presence of a high number of cosmopolitan species in the deep sea may result from the absence of sharp abiotic gradients such as light, temperature, and salinity that figure prominently in, for example, intertidal zonation patterns. The physical difference between two points with widely differing latitudes or longitudes is likely to vary less than in intertidal or terrestrial environments, leading to wider distributions in the deep sea. Normally, biogeography is analyzed at the species level using the classic definition of a group of individuals comprising a breeding population. Among fauna with limited distribution on Davidson, there is reason to suggest that we might not be getting a complete picture. Of the nine instances of fauna only seen at Davidson Seamount, as well as some organisms that appear on the far left-hand side of Fig. 2a, the taxonomy is not known with great specificity, since the phylogeny of many organisms on Davidson has not been worked out to a species level. These labels are only consistent within MBARI's VARS database. This allows one to probe for other observations within the Northeast Pacific deep sea, but cannot be used in a broader literature search, placing an artificial limit on these organisms' distributions. The difference in level of confirmed taxonomic identification may underlie the trends seen in fig. 3 between phyla. Of the Chordates included in this study, twenty-four of the thirty-two species (mostly fish), had both a genus and species name. In contrast, only five of the twenty-two Porifera had both names. The biggest area for improvement within this study lies in the refinement of the faunal list of Davidson Seamount. Further resolution may remove the limitation on the size of organisms that can be detected by frame grabs, but this cannot be used as a substitute for rigorous taxonomic analysis. Standardized identification procedures will be instrumental in ensuring seamount researchers from disparate geographic localities can make equivalent comparisons. Additionally, this effort has the potential to detect the presence of morphologically similar species that remain cryptic because differences are visually imperceptible. This would also critically alter our knowledge of distribution patterns. Both taxonomic and sampling improvements have the potential to shift this distribution to the right. CONCLUSION This study meshes two older fields of biology, taxonomy and biogeography, with the emerging field of macroecology to provide a snapshot picture' of what we know about a very specific faunal group— those organisms found on Davidson Seamount. However, because seamount biology is in its infancy, some sacrifice of "the apparent precision and analytical elegance of ecologists" (Jackson et al. 2001) is necessary. Combining datasets from different researchers will allow us to answer fundamental questions within the field that would be difficult, if not impossible to do otherwise. As sampling increases, we can expect these ranges to expand as more information becomes known. This study demonstrates that most of the megafaunal community of Davidson Seamount has been found at least a thousand kilometers away. This information can be used to reevaluate the role of seamounts in conservation. Protecting a single seamount with the expectation that this will encompass the entire geographic range of many novel species is not a valid conservation strategy because it is based on speculation. My study is the most comprehensive among a growing body of knowledge that is demonstrating that few organisms are localized to this scale. Instead conservation proposals should emphasize the high number of species representing a wide continuum of geographic ranges that are concentrated within a relatively small area on seamounts. ACKNOWLEDGEMENTS The author was grateful for the chance to collaborate with the Monterey Bay Aquarium Research Institute, and the assistance of Craig McClain and Lonny Lundsten. LITERATURE CITED Davis, A.S., D.A. Clague, W.A. Bohrson, G.B. Dalrymple and H.G. Greene (2002) Geol. Soc. Amer. Bull. 114 316-333. Gaston, K.J. (1996) Species-range-size distributions: patterns, mechanisms and implications. Trends Ecol. Evol. 11 197-201. Gaston, K.J. (1998) Species-range size distributions: products of speciation, extinction and transformation. Phil. Trans. R. Soc. Lond. 353 219-230. Gage, J.D. and P.A. Tyler (1991) Depth related patterns in community composition, pp. 229-247 in Deep- Sea Biology. Cambridge University Press, New York, NY. Jackson, J.B.C., M.X. Kirby, W.H. Berger, K.A. Bjorndal, L.W. Botsford, B.J. Bourque, R.H. Bradbury, R. Cooke, J. Erlandson, J.A. Estes, T.P. Hughes, S. Kidwell C.B. Lange, H.S. Lehihan, J.M. Pandolfi, C.H. Peterson, R.S. Steneck, M.J. Tegner, and R.R. Warner. (2001) Historical Overfishing and the Recent Collapse of Coastal Ecosystems. Science 293 629-638. McClain, C.R. (2007) Seamounts: identity crisis or split personality? J. Biogeogr. 34 2001-2008. O'Hara, T.D. (2007) Seamounts: centres of endemism or species richness for ophiuroids? Global Ecol. Biogeogr. 16 720-732. Richer de Forges, B., Koslow, J.A., & G.C.B Poore (2000) Diversity and endemism of the benthic seamount fauna in the southwest Pacific. Nature 405 944-947. Roberts, C.M. and J.P. Hawkins (1999) Trends Ecol. Evol. 14 214-246. Rogers, A.D. (1994) The Biology of Seamounts. Adv. Mar. Bio. 30 306-349. Samadi, S., Bottan, L., Macpherson, E., Richer De Forges, B., & M-C Boisselier (2006) Seamount endemism questioned by the geographic distribution and population genetic structure of marine invertebrates. Mar. Biol. 149 1463-1475. Whittaker, R. J. and J.M. Fernandez-Palacios (2007) Endemism, pp. 63- 72 in Island Biogeography. Oxford University Press, New York, NY. Wilson, R.R. and R.S. Kaufmann (1987) Seamount Biota and Biogeography. Geophys. Monogr. 43, 355-377. 10 FIGURE LEGENDS Fig. 1 A topographic map showing the geographic location of Davidson Seamount. Depth isomers and transect locations from ROV dives are overlaid. The inset shows Davidson's position relative to the California coast. Fig. 2a The frequency distribution of published maximal linear distances in kilometers (km) for Davidson Seamount fauna. The orange dotted line at 1879 km indicates the median value for this dataset. Fig. 2b Transformation of the frequency distribution of published maximal linear distances into a probability distribution. This figure plots the percentage of species found on Davidson Seamount with a range less than the x-axis value. Lines representing the approximate location from Davidson Seamount of select localities have been overlaid. Fig. 3 Boxplots of the maximally published linear distances (in kilometers) by phylum for each phylum found on Davidson seamount. The lines of the boxplot represent the first quartile, the median, and the third quartile. The whiskers were calculated by multiplying the difference between the third and first quartiles by 1.5. FIG. 1 3500 E an Eraneiseo Monterey 122°50W 122°45W 122°40W 12 FIG. 2a M 0 1000 3000 5000 7000 9000 11000 13000 15000 17000 19000 Linear Distance From Davidson (km) 20 10 FIG. 21 1.0 0.9 0.8 0.7 0.6 0.5 2 0.4 0.3 0.2 0.1 — 0 1000 3000 3000 7000 9000 11000 13000 15000 17000 19000 Lincar Distance From Davidson (km) FIG. 3 20000 15000 10000 5000 Archrapnde Churdnia Cnidarin Clempheures Echinnekernin alliunen Perikern Phylum