Synopsis on the knowledge and distribution of the family Bougainvilliidae ( Hydrozoa , Hydroidolina ) Latin

The family Bougainvilliidae comprises a group of anthoathecate hydrozoans that is biologically, ecologically and biogeographically poorly understood, and consequently, poorly taxonomically organized. Here, our goal is to synthesize knowledge of the family from an historical perspective, and to analyze their potential distribution based on their ecology. We analyzed all the available information on the family (based on 303 articles and databases), comprising 15 genera and 97 valid species in five oceans. Two temporal peaks (1900 and 2000) in publications are dominated by records of meroplanktonic species. The coastal zone has the most frequently reported occurrences. The widest latitudinal ranges are found in the genera Bimeria and Bougainvillia. Ecological niche modeling of 25 species (MaxEnt algorithm) finds that chlorophyll is the most important variable that influences the distribution of the family. Five possible latitudinal distributional patterns are derived from the model, dominated by the subtropical-polar distribution.


INTRODUCTION
The "hydroid" family Bougainvilliidae Lütken, 1850 pertains to a group of historically and conveniently (i.e., not monophyletically) described of cnidarians in the orders Leptothecata and "Anthoathecata" (the latter, also not monophyletic, Cartwright et al., 2008) in the class Hydrozoa.As with other anthoathecates, the family Bougainvilliidae has two phases in its life cycle: fixed to a substrate as a polyp and as free-swimming medusae (which may also have a reduced form that remains with the polyp, Russell, 1953).The pseudohydrotheca characterizes this family and it is defined as an external, glycosaminoglycan covering, with or without detrital incrustations, that envelops the hydrants and which can be reduced in some genera (cf., Calder, 1988;Schuchert, 2007).
Here, we will summarize the deficiencies of understanding the Bougainvilliidae, so that we may then proceed to build a context in which we can better understand the group and suggest avenues for their future study in biogeography, ecology and taxonomy.More specifically, we will synthesize current knowledge of the family based on historical and distribution perspectives, with which we will carry out an ecological analysis to predict potential occurrence of this interesting and diverse family.

MATERIALS AND METHODS
Basic taxonomy of the Bougainvilliidae and its place within the Hydrozoa follow Marques & Collins (2004), Bouillon et al. (2006); Schuchert (2007) and Cartwright et al. (2008).To determine historical series and distributions, we gathered 303 articles about the family published between 1890 and 2013.With this information, and databases, such as Genetic Sequence (GenBank) and WoRMS, we developed a data base for each genus and species, with a total of 1290 records having the following information: author, lifecycle phase, reproductive state, sex, location (body of water, country, state, region, latitude and longitude), date of collection, water depth (minimum, maximum), substrate type, salinity, temperature, synonymy and references cited.We then examined the historical and geographical information compiled.
To generate potential distribution models for valid species, we used the information from the data base described above, separating by life-cycle phase (polyp and medusa).Species were plotted using ArcGIS vers. 10 (ESRI, 2010).Environmental descriptive variables were: average surface water temperature, average surface salinity, and average concentration of nitrates, phosphates, oxygen and chlorophyll, from the BioOracle database (Tyberghein et al., 2011).We modeled using the Maxent algorithm (maximum entropy approach; Phillips et al., 2006).We used a threshold-independent measure, the area under the curve (AUC), to evaluate models and a Jackknife method (Sokal & Rohlf, 1995;Phillips et al., 2006) was used to examine the importance of each variable to the final model.The results of this evaluation are expressed as a measure known as gain, which shows the ecological requirements of each species, and probably determine their potential distribution.Classification of AUC values followed Metz (1986), as excellent (0.90-1.0), good (0.80-0.90), average (0.70-0.80), bad (0.60-0.70) and very bad (0.50-0.60).

RESULTS
The Bougainvilliidae comprise a total of 19 genera and 170 named species, of which 54 species were synonymized.Most of these are the cosmopolitan species Bougainvillia muscus (Allman, 1863).Another 19 species were doubtful.Thus, a total of 15 genera and 97 species are valid (Table 1) representing about 3% of the Hydrozoa species in the world (3,140 species, cf.Bouillon et al., 2006).
Species in the Bougainvilliidae tend to be meroplanktonic genera (e.g., Bougainvillia, Koellikerina and Nubiella, Table 2), and are not the most studied.About 65% of publications are about the genera Bimeria (which is benthic) and Bougainvillia.While reports of new species has been continuous, the species accumulation curve shows that about six new species per decade are being described, and therefore the asymptotic number of species is far from being reached (Fig. 1).
Historically, studies of the Bougainvilliidae have had several temporal hiatuses.Beginning in 1900, with very few studies from 1920-1950, increasing in the 1950s and still doing so (Fig. 1), with most of these studies (3-10 papers each) being by Calder, Galea, Genzano, Marques, Pagés, Palma, Schuchert, Segura-Puertas and Xu.Most of these were studies in tropical and southern Atlantic, northeast and northwest Atlantic, southwestern and central eastern Pacific oceans, in contrast with previous studies, which were mostly in European and North American waters.
Species in the remaining genera have fragmented distributions (Fig. 5): -Chiarella is in the Gulf of California (Brinton et al., 1986), and subequatorial zone of the Pacific Ocean; -Dicoryne is antitropical at depths of 1-400 m (Hirohito, 1988;Schuchert, 1996); -Garveia is most commonly reported from subequatorial and tropical regions in the northern hemisphere in the Atlantic, Indian and Pacific oceans and in deeper waters (2100 m, Ramil & Vervoort, 1992); -Koellikerina has a tropical or subtropical distribution, with the exception of K. massi, whose    -Pachycordyle is found in the north Atlantic and Pacific oceans, except for P. navis found in the Indian Ocean near South Africa, at depths between the surface and 30 m (Calder, 1991;Kramp, 1959b); -Parawrightia is found in northern and southern tropical waters of the western Atlantic and sub equatorially in the eastern Indian Ocean, at depths of 0-12 m (Kelmo & Santa-Isabel, 1998;Grohmann et al., 2003), and salinity of 36.5 (Calder, 1993); -Rhizorhagium has a wide latitudinal distribution in all oceans, with the majority of records from sub polar regions, at depths to 890 m (Schuchert, 2007); -Silhouetta is tropical in the Atlantic, Pacific and Indian oceans, and also in the Caribbean Sea, at a maximum depth of 30 m (Millard & Bouillon, 1973), and salinity of 36.5 (Calder, 1993); -Thamnostoma is common in warm tropical waters, including reefs in the Atlantic, Pacific and Indian Oceans, at depths of up to 100 m (Goy, 1979;Riera et al., 1986;Segura-Puertas, 1991); -Velkovrhia is the only genus restricted to fresh waters and caverns at up to 1830 m altitude (Tvrtković & Veen, 2006) near the Adriatic Sea.Modeling of ecological niche for a total of 25 species (with a minimum of 10 location records) resulted in AUC values >0.81, which indicates that classifications were good to excellent.In general, chlorophyll is the most important variable in 69% of the models, with phosphates important in 17%, temperature in 10% and salinity in 3%.The Jackknife test indicated that chlorophyll (implicating high primary productivity) is the greatest contributor (gain) to the models (Table 3).
These models suggest five possible patterns to describe potential latitudinal and longitudinal limits for species distributions: equatorial-tropical, tropicalsubtropical, subtropical-subpolar, subtropical-polar and widespread (from the equator to subpolar waters; Table 3, Figs.6-10).The subtropical-polar category dominates, with 31% of the species and includes peaks in species richness.Coastal species are also more abundant, with 90% of the species (Table 3).Predictions of distributions from the models concur with those observed in most life-phases in the Bougainvilliidae.However, for the species B. vestita, B. carolinensis (polyp), B. fulva, B. muscus (polyp), B. rugosa, G. franciscana, N. alvarinoae and R. roseum some occurrence points do not correspond to those predicted as most adequate.

Considerations about the distribution of the Bougainvilliidae
For a comprehensive analysis of aquatic species distributions, one must consider the three dimensions: latitude, longitude and depth (Miranda & Marques, 2011;Bentlage et al., 2013).In the case of the Bougainvilliidae, due to lack of information, those three dimensions are not always known, as most studies have focused on the epipelagic zone (0-200 m), as it is typical for Medusozoa (Marques et al., 2003;Segura-Puertas et al., 2003).The epipelagic is the zone in which we expect to find most species because it is known to be most diverse for zooplankton (Angel, 1994) as well as benthic organisms (Genzano et al., 2009).
Despite the lack of data, we carried out a distributional analysis in two dimensions.In latitude, the first dimension, the number of species declines (from 44 to 3) with increasing latitude, as is the case with respect to other groups of planktonic (Boltovskoy et al., 1999;Genzano et al., 2008) and benthic (Fautin et al., 2013) invertebrates.Longitudinally, the second dimension, we find that the majority of species are coastal, which shows the importance of substrates for development during the polyp phase, even for those species in which only the medusa phase is known.This suggests that the ancestral niche that was for species with both polyp and medusa phases has been maintained, despite subsequent evolutionary modifications with respect to dispersal and development.
As mentioned, models were somewhat limited by the sparse available information, both about species distribution and environmental variables.Also, alternating generations (found in some genera during their life cycles) are a problem for this type of analysis, because there are no methods that consider the influence of dispersal along with biotic and abiotic factors on each developmental phase (Pearson & Dawson, 2003).Regardless of these potential problems, the distributions generated by the models appear to be robust, in that the predicted distributions are very similar to those observed today.In the few cases with apparent incongruence's between observed and predicted distributions, these may not be errors but they rather may represent samples from sink populations of Bougainvilliidae (as in the cubomedusa Chironex fleckeri, Bentlage et al., 2009).These kinds of populations tend to exist temporarily in marginal habitats where they may often disappear if dispersal does not supply more individuals, such as when mortality exceeds natality (Dias, 1996;Palmer et al., 1996).
Hydrozoans distribution are known to be influenced by a variety of environmental factors (Arai, 1992), yet most models suggest that chlorophyll is the most important.Chlorophyll plays an important role and is a proxy of high primary productivity and nutrient-rich waters for some regions, hence indicating abundance of food resources for pelagic and benthic communities (Acha et al., 2004).It is also evident that other biotic factors can be important, yet they were not considered in the analysis, such as epibiosis (e.g., Oliveira & Marques, 2007, 2011), and those associated with human activities (e.g., Rocha et al., 2013).
Temperature is important for the distribution of some species in the genera Dicoryne and Garveia.Dicoryne occurs where temperatures oscillate between 2.7-23.6ºC(Okolodkov, 2010), such as <20 o C for D. conferta (Broch, 1916), while Garveia is found in waters with temperatures >25ºC (Okolodkov, 2010).Temperature is often considered a key factor in determining the distribution of benthic (Boltovskoy & Wright, 1976) and planktonic (Beaugrand et al., 2013) organisms.Of course, local and historical factors are considered to be important as well.
The potential distribution for some species includes regions in which important studies were carried out without noting the presence of those species.This may be due to the fact that modeling does not include biological interactions, geographic barriers and history, which clearly are also important and may explain why species seldom occupy all favorable environments (Anderson & Martínez-Meyer, 2004).An example in point is found in the latitudinal distribution of B. superciliaris and Nemopsis bachei.Both have been found in the northern hemisphere and their potential distribution due to modeling includes both hemispheres, which suggests dispersal limitation.Specifically, N. bachei is considered to be euryhaline (Calder, 1971) (salinity 3-35, Denayer, 1973;Mendoza-Becerril et al., 2009) and usually present at temperatures <26°C (Cronin et al., 1962;Denayer, 1973;Marshalonis & Pinckney, 2007).
Of the species with sufficient information for polyp and medusa phases, models can be particular to the life phase.In two species (B.carolinensis and B. muscus) patterns are similar for both phases, and in others, one phase generally has a larger predicted distribution.In B. muscus the medusa phase has the greatest predicted distribution while in B. superciliaris it is the polyp phase.Similarly, in at least B. superciliaris, temperature is most important for the polyp phase, while chlorophyll is most important for the medusa phase.This reinforces the idea that the influence of the environment can vary by life phase and so coupling of phases is not trivial.As a consequence, modelling without respect to life phase, for organisms with distinct phases, may overlook very important and complex patterns.Adding to uncertainty is the dispersal potential of each phase, with some considering the medusa phase as the principal dispersal agent, which is reflected by a widespread distribution, including pelagic (Bouillon, 1981;Leclère et al., 2009).This is exemplified in B. muscoides.On the other hand, some researchers suggest the contrary (Cornelius, 1992), as seen in B. superciliaris.Most likely is that while some patterns predominate, there will always be exceptions.
The lack of information from a variety of regions is also clear in this analysis, especially in the southeastern Atlantic, southern Gulf of Mexico, Central America, southwest Pacific and northeastern Indian Ocean.These regions also have relatively high AUC values (0.65-0.99) for several species and would be interesting places to increase sampling effort.However, the number of studies seems to be declining with respect to the polyp phase (Table 2).This trend and the species accumulation curves illustrate that more studies are needed, especially of benthic and planktonic faunas and at a variety of depths and aquatic environments, in addition to revisions based on morphological, reproductive and molecular characters to respond questions of evolutionary and geographic patterns.

Figure 1 .Figure 2 .
Figure 1.Cumulative number of species described since 1809 to date (2013) and number of studies that include the family Bougainvilliidae since 1809, by decade.The line indicates the cumulative number of species and the bars indicate of number the articles.

Figure 3 .
Figure3.Percent of the total number of currently valid species in the family Bougainvilliidae found in each ocean.The total adds to greater than 100% because several species are found in more than one ocean.

Figure 4 .
Figure 4. Frequency distribution over latitude of the number of currently valid species in the family Bougainvilliidae.

Figure 5 .
Figure 5. Latitudinal range for genera in the family Bougainvilliidae.indicates genera in polyp phase, indicates genera in medusa phase.The number above the bars is the number of valid species for each genus.Fragmented bars indicate that the latitudinal distribution is not continuous.

Table 1 .
Species described in the literature of the family Bougainvilliidae with life-cycle phase indicated and distribution.P: polyp, M: medusa.

Table 2 .
List of valid species in the genera of Bougainvilliidae, 1809 to 2013.M: medusa, P: polyp.

Table 3 .
Modeled classification for distribution of the 25 valid species with >10 records.