Development of an interactive identification key for Oncaeidae (Copepoda: Cyclopoida)

The current state of development of an interactive electronic identification key for females of the marine pelagic microcopepod family Oncaeidae is presented. To date, 112 species have been described, allocated to seven genera (Archioncaea, Conaea, Epicalymma, Monothula, Oncaea s.str., Spinoncaea, and Triconia) and 16 species groups within the speciose paraphyletic taxon Oncaea s.l. Only two species of Oncaea s.l. show an uncertain group status. Of all described oncaeids, seven species are not yet included in the key due to inadequacies in morphological description and/or lack of type material for re-examination. In addition, seven morphologically distinct species and two new species groups are considered that have not yet been formally described. The key uses easily distinguishable morphological characters as far as possible (e.g. body size, proportional lengths of urosomites, length of P5 exopod) and includes more difficult characters (e.g. spine counts on swimming legs, proportional spine lengths) where required for unequivocal identification. The characters used in the key allow the identification of genera and species groups by investigators with limited taxonomic experience; species separation in most cases requires some advanced dissection skills. Potential identification problems caused by difficulties discerning minute morphological features or intraspecific morphometric variability are discussed. The present version of the identification key still requires complementary illustrations of character states and inclusion of fact sheets summarizing the characteristics of each species with notes on taxonomic uncertainties and links to databases providing zoogeographical and genetic information.


Introduction
The microcopepod family Oncaeidae is one of the most abundant copepod taxa in marine pelagic habitats. It occurs in all climates from tropical to polar seas and in all vertical layers from the surface to the deep sea (e.g. Nishibe 2005 and literature cited therein; Koppelmann et al. 2009;Kosobokova et al. 2011;McKinnon et al. 2013). The family is very diverse, with more than 100 species described to date (Böttger-Schnack and Boxshall 2014). Most of them are smaller than 600 µm in size as adults and are thus challenging to adequately consider in studies on species diversity. The role of oncaeid species in the marine pelagic environment is also largely unknown (reviewed by , because studies on their species specific requirements are hampered by identification problems in part caused by their small size. Over the past three decades, oncaeid morphology has been studied in great detail. The mouthparts, in particular, are now known much better and have provided an important basis for the definition of genera and species groups (Böttger-Schnack and Huys 1998;Böttger-Schnack and Schnack 2013). The number of described oncaeid species has significantly increased, and genetic studies using barcoding have proven the importance of minor morphological details for separating sibling species (Böttger-Schnack and Machida 2011). Published identification keys for Oncaeidae in the printed literature are all regionally limited: they often do not include the many small-sized species less than 500 µm, and do not include recent taxonomic progress (e.g. Shmeleva 1969;Bradford-Grieve et al. 1999). The few available regional online keys are of restricted use because they do not include all species known for the region (e.g. Marine Species Identification Portal 2014, http://species-identification.org; Zooplankton Identification Manual for North European Waters 2014, http://192.171.193.133) and/or are in part erroneous (e.g. Swadling et al. 2013 (ed.) Australian Marine Zooplankton, http://www.imas.utas. edu.au).
At present, the generic system of Oncaeidae is only partly resolved. This represents a fundamental problem for the development of a general identification key. So far, seven genera are recognized (Boxshall and Halsey 2004): Archioncaea Böttger-Schnack and Huys, 1997;Conaea Giesbrecht, 1891;Epicalymma Heron, 1977;Monothula Böttger-Schnack and Huys, 2001;Oncaea Philippi, 1843;Spinoncaea Böttger-Schnack, 2003 andTriconia Böttger-Schnack, 1999. The large genus Oncaea is a paraphyletic taxon (Böttger-Schnack and Huys 1998) that comprises still another 17 morphologically distinct lineages. One of these includes the type species of the family, Oncaea venusta Philippi, and has been formally separated as Oncaea sensu stricto . All other lineages of Oncaea sensu lato will be called 'species groups' or 'groups' hereafter (cf. Böttger-Schnack and Schnack 2013, table 1). The identification of oncaeid genera and groups is difficult because the characters used to define them are mainly based on details of the mouthparts (cf. Böttger-Schnack and Schnack 2013, tables 2 and 3). Observation of mouthparts in microcopepods requires a high level of dissection skill and therefore these characters were not considered practical for non-specialists. Instead, characters had to be found that are easier to discern or recognize.
The aim of this key is to facilitate the identification of species and genera or species groups from throughout the global ocean and to primarily use easily accessible characters. In cases where practical species determination is difficult or distinctions are not yet fully resolved, the key should nevertheless allow the identification of genera or species groups.

Methodological background of the key
The key is built on Lucid 3.5 software, developed for the construction of identification keys at the University of Queensland, Australia, and now distributed by the company Identic Pty Ltd. (www.lucidcentral.org). At present, a total of 112 oncaeid species are considered in this key; additional species can be easily included when new descriptions are provided. Of the valid species described in the literature to date (Böttger-Schnack and Boxshall 2014), seven species have not been included because the descriptions are not sufficiently clear and require revision (cf. Böttger-Schnack and Schnack 2013, table 3B). In addition, seven yet undescribed species from the personal copepod collection of the first author have been included, each of which show distinct characters and can be clearly separated from all other known species (cf. Böttger-Schnack and Schnack 2013, table 3C). Within the key, genera and species groups are listed in alphabetical order, as are the species within their respective genus or group. Those species which could not yet be allocated to groups are presented at the beginning, and the new, as yet undescribed species at the end of the list.
The morphological data used in this key are taken from reliable taxonomic publications (for references see Appendix Table A1) and for about half of the species personal descriptions, unpublished observations and line drawings from the first author are used. Discrete characters such as numbers of spines on the legs are generally preferred because they can be clearly defined, but continuous characters such as body length and size ratios of body parts had to be used as well, though these can be less clearly defined. Primary characters have been considered for all species in the family, whereas secondary characters are considered only for selected species or groups as required. Within the key, morphological features are arranged from the anterior to the posterior end of the body. Figure 1 shows the Lucid identification screen, with features presented in the top left window, and the copepod taxa (called 'entities') in the top right window. Features and entities (taxa) are organized in several levels of subgrouping and presented as trees, which can be expanded or collapsed as required.
Descriptive terminology follows that of Huys and Boxshall (1991), and the abbreviations P1-P5 are used to represent the first to fifth thoracopods.

Development of key matrix
The primary characters considered for identification are presented in Table 1, grouped according to the skill required for their observation. For some characters the determination is easy because no or only simple dissection is required, but others require more advanced dissection skills and are thus more difficult to consider. Total length, length proportions of urosomites and the shape of the maxilliped basis are examples of easy characters. The number of spines on the swimming legs is more difficult to observe, but is of high diagnostic value. The combination of spine counts on the exo-and endopods of P1-P4 can be grouped into 14 different armature types (Figure 2), most of which represent a single genus, group or even species (cf. Böttger-Schnack and Schnack 2013, tables 5 and 6) and can be used for identification. However, one armature type is shared by a large number of different genera and groups (cf. Böttger-Schnack and Schnack 2013, table 4), and for these taxa other features had to be found.
[Note: In table 5 of Böttger-Schnack and Schnack (2013), the presentation of leg armature type 9 is missing erroneously; it refers to a new and yet undescribed morphospecies, Oncaea sp. K from Australian waters, which showed deviation from the typical oncaeid spine count on the endo-and exopods. Leg armature type 9 is given in Appendix Table A2.] Among the primary characters, a small subset of preview features has been defined (marked in bold in Table 1), which are typical of individual species or groups. These preview features include body length, because size alone will considerably narrow down the number of species to be further considered (Figure 3). The cephalosome may have a very group specific shape (Figure 4) and the somite carrying P1 or P2 may have a characteristic dorsal protrusion or projection ( Figure 5). The size of P5 exopod segment (Figure 6), the shape of the maxilliped basis (Figure 7), and the presence or absence of a conical process between the two distal spines of P4 endopod (Figure 8) in addition to the corresponding process on P2 and P3, are easily discernible features indicative of specific genera and groups or even species. In several cases some additional, secondary characters are required for a final identification of species (Table 2). For example, the length to width ratio of the caudal ramus can be used to separate species within the genus Oncaea s. str. or the length of the outer basal seta on P5 can separate the five described species belonging to the dentipes-subgroup of Triconia.

Discussion
With the use of the characters outlined in this contribution, oncaeid specimens can be allocated to genera or species groups independent of any remaining identification problems. Within genera and groups species separation is possible in most cases, with the exception of Epicalymma species, or the many undescribed species in the Oncaea atlantica-vodjanitskii-group (cf.       The usefulness of the key may be hampered by problems of observation: e.g. it might be difficult to count the endopodal spines when they are very small. However, possible misinterpretations of feature states are accommodated in the key. Some problems may also be encountered when measuring continuous characters such as the spine lengths on the endopods. Besides difficulty making a correct measurement of spines in a three-dimensional space under the microscope, the intraspecific variability of a certain spine length is not well known . Thus, at present, fairly wide ranges of possible values have been defined for these features. Another problem for identification keys in general is the morphological uncertainty of species, particularly those that have not been described in sufficient detail or that are new to science. The validity of species names also has to be taken in mind (e.g. Böttger-Schnack and Huys 2004). Finally, it remains uncertain whether species descriptions for particular regions can be transferred to other areas of the world ocean. There may be slight differences between populations of the same species, or sibling species, which have developed in different areas (e.g. Ueda and Bucklin 2006;Sakaguchi and Ueda 2010;Ueda et al. 2011) and that at first sight look morphologically very much alike.

Conclusion
It is urgently recommended to use a genus, group-or subgroup-name in cases where species identification remains uncertain. Presenting uncertain species names will lead, and has already led, to erroneous information about zoogeographic distribution patterns, e.g. the distribution pattern reported for the well-known species Triconia conifera (Giesbrecht, 1891) (Razouls et al. 2005(Razouls et al. -2014a, which actually represents a subgroup of 13 closely related species ) each with a different distribution pattern, e.g. T. antarctica  or T. derivata   (Razouls et al. 2005(Razouls et al. -2014b(Razouls et al. , 2005(Razouls et al. -2014c. Identification keys using relational databases, such as the Lucid 3 program, avoids the need to review states of features in a fixed sequence, but rather allows the consideration of features according to their accessibility, so that even damaged specimens or parts of specimens may lead to successful identification. The system also allows an easy addition of new species and updating as improved taxonomic knowledge becomes available. The software itself supports the identification process by providing suggestions for the best or next best feature to look at for a quick identification by eliminating features no longer relevant, and by providing a list of differences relevant for the separation of remaining species. The final key will have to include illustrations for all feature states considered and 'fact sheets' summarizing the characteristics of each species, including notes on controversial taxonomic issues of the respective taxon and links to databases providing species specific zoogeographical and ]. After completion, the key will be made available for general use and will be announced on the web site of the first author (http://rb-schnack.de) Table A1. (Continued).