{"paper_id":"09f690da-85cc-4005-86b2-ec670bfd2693","body_text":"Two simple ways to make taxonomic diagnoses more useful. \nLaurence Packer1 \n1 Department of Biology, York University, 4700 Keele St., Toronto, ON M3J 1P3, Canada \nCorresponding author: Laurence Packer (xeromelissa@gmail.com) \nhttps://orcid.org/0000-0002-1711-8203 \n \n  \nAuthor-formatted, not peer-reviewed document posted on 15/11/2024. DOI:  https://doi.org/10.3897/arphapreprints.e141840\n\nAbstract \nDiagnoses are important aspects of description and identification processes, but they often do not \nmake it clear whether they are useful for a particular specimen. I suggest that diagnoses always \nbe accompanied by overt statements as to what taxonomic group the new taxon is being \ncompared with, and whether it is a geographically, morphologically or ecologically restricted \nsubset. I term the group to which the diagnosis overtly relates the reference group. Further, a \npaper that provides the features that the reference group must possess has the potential for being \nmore broadly useful. For example, if the reference group is a subtribe but the authors explain \nhow to separate that subtribe from all others in the subfamily, then a user must be able to identify \nonly the subfamily before finding potentially useful information in the paper. I term this often \nmore expansive taxon the recognition group.  \nFor 313 newly described insect genera for which diagnoses were provided, I assess the \ntaxonomic level and number of genera in both the reference and recognition groups. The two \ngroups were identical in almost half of the cases, were at the same taxonomic level but \ngeographically or morphologically restricted in less than 9% and at a higher taxonomic level in \nthe remainder. When authors explained how to identify the reference group from a larger \nrecognition group, the number of genera from which the new one could be differentiated \nincreased by a factor of more than four. \nI make a series of recommendations on how diagnoses can be improved based upon \nanalyses of reference and recognition groups. \n \nKey words: Best practices, ease-of-use, identification, taxonomy, insects, higher-level \nclassification \n \n  \nAuthor-formatted, not peer-reviewed document posted on 15/11/2024. DOI:  https://doi.org/10.3897/arphapreprints.e141840\n\nIntroduction \n \nThere is increasing demand for tools to aid in the identification of the world’s biota and while \nnumerous modalities have been employed, at their description, new taxa are still mostly only \nidentifiable using traditional morphological approaches. The most common of these methods are \nidentification keys and diagnoses (Winston 1999; Borkent 2021; Packer 2024). Among the \nalmost 600 newly described genera, less than 40% were associated with a key to aid in their \nidentification (Packer 2024). Thus, diagnoses are critically important for biodiversity research, \nespecially for newly described taxa for which they may be the easiest available tool permitting \nidentification. Despite this, there is little advice associated with constructing diagnoses and even \ntheir inclusion is merely a recommendation of the Code of the International Commission on \nZoological Nomenclature (ICZN 1999) which states:  \n \n “Recommendation 13A: Intent to differentiate. When describing a new nominal  \ntaxon, an author should make clear his or her purpose to differentiate the taxon  \n by including with it a diagnosis, that is to say, a summary of the characters that  \ndifferentiate the new nominal taxon from related or similar taxa.” \n \nThis leaves open the question of which taxa might be “related or similar” and how \nbroadly those terms are to be interpreted. Both might best be understood in terms of the \ncharacteristics an organism possesses that indicate that the diagnosis might be relevant to a \nspecimen at hand. This information is often not presented in a clear fashion. To improve how \ndiagnoses may be formulated, I define two terms: reference group and recognition group.   \nThe reference group is the taxonomic group to which the diagnosis directly relates.  For \nexample, if the diagnosis states: “the new genus can be differentiated from others in the subtribe” \nthen the reference group is the relevant subtribe. Similarly, if the diagnosis is a comparison to \nonly one other genus (as is commonly the case when a new genus is separated from the one to \nwhich its species previously belonged) then the reference group is just the one genus in the \ncomparison.  \nThe recognition group is the highest level in the taxonomic hierarchy for which \nidentifying features were provided by the author(s)—the taxonomic group to which a user must \ndetermine that their specimen belongs for the information provided in the paper to be worth \ninvestigating further. The recognition group is frequently at a higher taxonomic level than the \nreference group. For example, if the diagnosis is to genera within a subtribe but elsewhere in the \npaper the authors explain how to differentiate the subtribe within the subfamily but not how to \nseparate that subfamily from others in the family, the recognition group is the relevant subfamily, \nbut the reference group remains the subtribe. The reference and recognition groups are often at \nthe same taxonomic level but may contain different numbers of genera. For example, the \nreference group may be a geographically, morphologically or ecologically restricted subset of a \nnamed taxon at the same level in the hierarchy (western hemisphere Orientalidae, members of \nthe Filiantenninae with swollen flagellomeres or members of the Asterophagini that form galls \non rhododendron, to give some imaginary examples).  \nAuthor-formatted, not peer-reviewed document posted on 15/11/2024. DOI:  https://doi.org/10.3897/arphapreprints.e141840\n\nInclusion of a recognition group that is more expansive than the reference group \nincreases the number of genera from which the new genus is discriminated from. As this is a \ndesirable component of new taxon descriptions, the larger the number of taxa included in the \nrecognition group the better. Additionally, the higher up the taxonomic hierarchy the recognition \ngroup is, the easier it will be for users to know that the content of the paper might be relevant to \nmaterial in their collection because higher level taxa are easier to identify.  \nIn this paper, I provide results of assessments of reference and recognition groups \nassociated with the description of new insect genera published over an 18-month period starting \nin January 2021. I chose the genus level to ensure that the dataset was both tractable in size while \nalso including a broad range of insects. The conclusions reached, however, should be applicable \nto any taxonomic rank for all groups of organisms. \n \nMethods \n \nIn early 2022 I searched for newly described insect genera through Scopus searches using ordinal \nnames combined with terms such as “new gen*” or “gen* n*” filtered for the year 2021. \nSearches were repeated in July and August 2022 but with 2022 as the filter. Papers dealing only \nwith fossil taxa were removed from the sample. The resulting papers were downloaded where \navailable online or through the library resources available to me and I attempted to obtain pdfs of \npapers published in Zootaxa (where the largest number of new genera were described) that were \nbehind a paywall by contacting the senior authors and requesting a pdf (see Packer 2024 for \nmore details).   \nFor each newly described genus, I looked for a section in the paper that had the word \ndiagnosis (or a derivative such as “diagnostic”) as a heading or subheading. Papers lacking this \nwere not assessed further. For each newly described genus associated with a diagnosis, I \ngathered the following information in addition to standard reference data: a) the reference group \nto which the diagnosis applied, b) the information that enabled a user to identify the reference \ngroup; c) where such information was placed; d) the recognition group; e) the reason for the \ndifference between reference and recognition groups, when these were different, f) the number of \ngenera in the reference group; g) the number of genera in the recognition group and h) what \nreference or recognition groups were implied by the paper’s title. Some of these variables require \nadditional explanation as follows.  \nReference and recognition groups were initially classified as given in each paper. The \ncategories found were—a single genus, two or more genera (including genus complex, “phyletic \nseries”, clade), subtribe, tribe, supertribe, subfamily, family, superfamily and order. In some \ncases, it was not possible to decide at which taxonomic level the diagnosis applied and in two \ncases the reference group was ecologically defined (Dorchin et al., 2021 and Gaimari et al., \n2021). These levels were simplified to four categories for visualisation due to negligeable sample \nsizes for the more rarely used taxonomic levels: i) genus plus genus group, ii) subtribe to \nsupertribe, iii) subfamily and iv) family level and above.  \nWhen both reference and recognition groups could be detected and were different, I \nsearched for rationales for their differences as stated in the papers. The following categories were \nfound: i) geographic, ii) descriptions, iii) diagnoses, iv) keys, v) one or morphological \nAuthor-formatted, not peer-reviewed document posted on 15/11/2024. DOI:  https://doi.org/10.3897/arphapreprints.e141840\n\ncharacteristics, vi) “size”, vii) “appearance”, viii) morphology-based phylogenies, ix) \n“taxonomic differences”. Categories ii) through viii) could also be lumped together under the \nbroader category of morphological differences. For morphological differences, the number of \nfeatures given differentiating reference from recognition groups was estimated for each genus. \nWhen data on number of genera in the reference or recognition groups were not provided \nin the paper, I searched elsewhere for the information. Taxon-specific databases such as the \nOrthoptera Species File (Cigliano et al. 2024) were particularly useful, but in some cases \nWikipedia or Wikispecies pages had to be consulted. All such online searches were conducted in \nSeptember 2024 and the sources of information used are indicated in Suppl. material 2 columns J \nand K. I attempted to correct the numbers of genera to those that would have applied at the time \nthe paper was written: all new genera described after 2020 were subtracted from the totals \nwhenever this information could be readily located. I used the numbers given by the authors \nwithout cross-correction as the information presented in the paper would relate to how the author \nunderstood the group at the time and it was not possible to decide which viewpoint might be \nmost broadly accepted by the relevant taxonomic community. Also, it seemed likely that online \nsources might not reflect the authors’ understanding of the size of the reference and recognition \ngroups. Online sources may be inaccurate for various reasons (for example, the wikispecies page \nfor Gigantometopini (Hemiptera: Miridae) does not include the genus that is imaged on the page \n– Megalofaciatus, in the list of genera). Thus, I analysed group size data in two ways—first with \nonly information provided by the authors and second with data from additional sources also \nincluded. When the number of genera in the reference and/or recognition group was \ngeographically or morphologically restricted, it was often not possible to discover how many \ngenera were relevant as such information is often not available unless stated by the authors. \nThe distribution of number of genera in reference and recognition groups had very long \nright-hand tails. As averages are strongly influenced by relatively few large numerical outliers, I \noften provide the median which is a more accurate reflection of differences among subcategories.   \nWhere in the paper the information permitting separation of the reference group from the \nrecognition group was provided was noted. Ideally this would be in the diagnosis itself, indicated \nby the term “diagnosis” in Suppl. material 2 (column I) or close by, either above it or following \nthe description “close”. In more extensive treatments such as in revisions of higher-level taxa, \nthis information might be given at the beginning of a taxonomy section and although perhaps at \nsome remove from a new genus’ diagnosis, its placement in such cases was considered \n“sensible”. Another useful place for this information to be provided would be associated with a \nkey to genera if one was provided “key”. Less useful were instances where the required \ninformation was far ahead of the diagnosis in an introduction “far” or near the end of the paper, \nusually in a discussion or concluding remarks “end”.  \nAnalyses are presented on a per-genus basis irrespective of the number of new genera \ntreated in the paper as it is the genus that users need to identify. Additionally, sometimes authors \nused different levels in the taxonomic hierarchy for different genera described in the same paper. \nFor example Lee et al. (2021) had one genus as the reference group for Palumbina but seemingly \nthe entire subfamily for Tenupalpa (Lepidoptera: Gelechiidae: Thiotrichinae).  \n \n \nAuthor-formatted, not peer-reviewed document posted on 15/11/2024. DOI:  https://doi.org/10.3897/arphapreprints.e141840\n\nResults \n \nThree hundred and thirteen genera were described associated with one or more diagnosis in my \nsample. The papers are listed in Suppl. material 1 and the raw data associated with them in \nSuppl. material 2. Below, I note in which column(s) the relevant raw data for the various \nanalyses can be located. I could not determine the reference and/or recognition groups \nunambiguously in two papers (Goncalves and Domahovski 2021; Motamedinia et al. 2021).  \nColumns E and F in Suppl. material 2 show the taxonomic levels of the reference and \nrecognition groups as noted by the authors. The two groups were identical as originally stated in \n148 cases (47.6%), were at the same taxonomic level but the former was a morphological or \ngeographical subset of the latter in 26 (8.4%). The reference group was one level below the \nrecognition group in 111 instances (35.7%, four of these were further geographically restricted), \ntwo levels below it in 15 (4.8%), three levels in 10 (3.2%) and four levels in one (0.3%). \nReasons for differentiating the two groups were found for 160 genera (51.1% of the total, \ndata in Suppl. material 2 column H).  More than one category of reason (as many as four) was \ngiven for 14 genera; thus, there were 177 rationales in total. Most of these (136) were \nmorphological in nature (76.8% of the total number of reasons), 34 (19.2%) were geographic, 4 \n(2.3%) were phylogenetic and 3 (1.7%) were based upon taxonomic differences that were not \ndetailed. \nData for taxonomic levels based on the simplified four categories are presented in Fig. 1 \nfor each order with at least 17 genera in the dataset as well as for the sum of all orders with fewer \ndata (the largest of which was Dictyoptera with nine entries). Overall, reference groups around \nthe tribal level were the most common (50.2% of instances) with the other three categories \nranging from 14.5% (family level and above) to 18.1% (one or more genera) (Fig. 1). The \nnumber of cells in the table with entries less than 5 was too large for statistical analysis. \nNonetheless, it is clear that Lepidoptera and the combined data for the smaller orders both stood \nout from the rest. For Lepidoptera, the percentage of reference groups at the generic level was \n60%—more than twice that for the next largest value (29.4% for Orthoptera). This may reflect \nthe image-based identifications commonly applied to members of this group and the greater \ndiscriminatory power expected of lepidopterists with decisions often based on wing patterns. For \nthe smaller orders, the use of reference groups at the family level and above was substantially \ngreater than in any of the larger orders (50.0% compared to an average of 11.8% among the \nothers). This is perhaps unsurprising given that, by definition, the smaller orders have less \ndiversity. \nFor the recognition group, the taxonomic level data range from 10.9% for genus group to \n38.8% for subfamily. Again, Lepidoptera and the combination of smaller orders are outliers \nrequiring genus group-level identification and family or higher-level respectively for over half of \nthe new genera respectively (Fig. 1).  \n \n  \nAuthor-formatted, not peer-reviewed document posted on 15/11/2024. DOI:  https://doi.org/10.3897/arphapreprints.e141840\n\n \nFig. 1. Distribution of reference (above) and recognition (below) groups  among four taxonomic \nlevels each expressed as a percentage ofthe total number of diagnoses provided separately for the \nsix major orders as well as for the minor orders combined. \n \n0\n10\n20\n30\n40\n50\n60\nrecognition group \n>=family subfamily tribe genus\n0\n10\n20\n30\n40\n50\n60\n70\n80\nreference group\n%  \nAuthor-formatted, not peer-reviewed document posted on 15/11/2024. DOI:  https://doi.org/10.3897/arphapreprints.e141840\n\nThe number of genera in the reference group was stated in the paper in 257 cases \n(82.1%), are presented in Suppl. material 2 column J and ranged from one to 257, averaging 27.1 \nand with a median value of 13. When data from additional sources were included, the reference \ngroup was given or could be estimated for 299 genera (95.5%), ranged from 1 to over 2400, \naveraged 43.7 and had a median of 17.  \nThe authors provided the number of genera in the recognition group in 156 cases (49.8%) \nas shown in Suppl. material 2 column K. This number ranged from one to 265, with an average \nof 41.7 and median of 25. The number of genera in the recognition group could be estimated for \nan additional 138 genera (for a total of 294, 93.9%) when data from additional sources were \nincluded. The number of genera ranged from one to ~28000 (all beetles), averaged of 476.9 and \nhad a median of 45.5.  \nThe increase in number of genera in the recognition group compared to the reference \ngroup could be estimated from the information presented by the authors for 150 genera (47.9%) \nand ranged from zero to 253 with an average of 18.3 and median of zero due to the large number \nof instances where the two groups were identical (91 instances - 60.7% of the genera where the \ndatum could be calculated). With zeros removed, the average became 45.9 and the median value \n32. Again, with zeros removed the number of genera from which the new one could be \ndifferentiated increased by an average of 8.4 times and a median of 4.1 times.  \nWhen other sources of information were included, the increase in size of the recognition \ngroups could be estimated in 289 cases (92.3%) and ranged from zero to almost 24000, averaged \n440.7 and with a median of 9 which became 56 when zeroes were removed. Again, when \nidentical recognition and reference groups were removed from the data where group size(s) \nestimated from extraneous information were included, discrimination of the recognition group \nresulted in an average increase of 127 times more genera, with a median of five times. \nWhen morphological features were given to separate reference and recognition groups, \nthe number of features requiring observation varied from 1 to 49 with an average of 12.1 and \nmedian of 7 (Suppl. material 2 column L). One or more of the features required dissection (of \ngenitalia or digestive system) in 47 cases ranging in relative proportion from one out of 26 \nfeatures to both of two. The average proportion of features that required dissection among all \nfeatures was 10.5%. Both sexes were required to identify the reference group in six cases and \nboth sexes as well as the pupa in one more. Thus for 5.3% of the genera for which the relevant \ndata were presented, it was seemingly not possible for a user to embark on checking the \ndiagnosis for the new genus unless they either had both sexes, or a juvenile stage in addition to \nboth sexes. \nPlacement of the information that permitted identification of the reference group was in \nthe diagnosis itself in 24 instances (14.8%), in a sensible position within the overall taxonomic \ntreatment in 89 (54.9%) and close to the diagnosis in 21 (13.0%) (usually immediately above it \nor in a remarks section beneath the description). Less usefully, these data were given two or \nmore pages above the diagnosis in 21 instances (13.0%), near the end of the paper in six (4.3%) \nand in a phylogenetic tree in one (0.6% - see recommendation #6 below).  \nKeys were available to aid in the identification of 122 of the new genera (38.6%; Suppl. \nmaterial 2, column M). Papers lacking a key had significant differences in the taxonomic levels \nof their reference and recognition groups compared to those with keys, but the patterns were not \nAuthor-formatted, not peer-reviewed document posted on 15/11/2024. DOI:  https://doi.org/10.3897/arphapreprints.e141840\n\nsimple. Reference groups without a key were disproportionately less represented at intermediate \ntaxonomic levels (subfamily and tribe) (χ2 = 16.7, p < 0.001) whereas recognition groups had \nfewer subfamilies for those genera not made identifiable with a key compared to all other \ntaxonomic levels (χ2 = 27.1, p < 0.00001). The numbers of genera in reference or recognition \ngroups did not differ significantly whether an identification key was, or was not, provided when \nbased upon data provided by the authors (z = 0.75, p = 0.46 and z = 1.6, p = 0.11 respectively). \nWhen data from extraneous sources were included, there were significantly fewer genera in both \nreference and recognition groups in papers without a key compared to those with one (in both \ncases z=2.31, p=0.021). Genera not made identifiable with a key were more likely to have the \nreference and recognition groups identical (χ2 = 20.2, p < 0.00001), thus making the paper \ndescribing them less useful in two ways: lacking a key and having the smallest possible \nrecognition group. \n The taxonomic information provided in the paper’s title (Suppl. material 2 column E) \nindicated the reference group precisely in 64 cases (20.8%), differed only in terms of geographic \nor ecological delimitation in 95 (30.9%), indicated a higher taxonomic level in 116 (37.8%) and \na lower one in 24 (7.8%). The title reflected the recognition group in 25 cases (8.2%), differed \nonly in ecological or geographic delimitation in 95 (30.9%), a higher taxonomic level in 79 \n(25.7%) and a lower taxonomic level in 107 (34.9%). Thus, the title of the paper may be \nconsidered misleading as to the taxonomic relevance of the information within (through over – or \nunderstating its generality) for ~80% of the genera. \n \nDiscussion \n \nDiagnoses are important components of the description of new taxa and their inclusion has been \nrecommended by the ICZN since 1930 (Rheindt et al., 2023). But how might a user decide \nwhether the diagnosis is relevant to a specimen they wish to identify?  \nI have provided a framework whereby this information might be conveyed: the reference \ngroup being that to which the diagnosis applies directly, and the recognition group the taxon \nwhich the user needs to be able to identify before using the more finely discriminatory \ninformation in the publication. While these were often the same, there are two main advantages \nto proscribing a recognition group that is at as-high-a taxonomic level as possible, or at least as-\nhigh-a-level as might be reasonable depending on the audience. First, the more expansive the \nrecognition group the easier it will be for a user to determine whether their specimen belongs to \nit or not and thereby decide that further investigation of the paper is warranted. Second, the \nlarger the recognition group the more likely it is that users will already have the information to \nidentify it. For example, if the diagnosis for the new genus compares it to only one other genus \n(especially if the shared features that differentiate them from higher taxonomic levels are not \nprovided), the user needs to be able to identify the original genus. Precisely this situation was \nfound in ten instances (where entries in Suppl. material 2, columns F and G both state “1 genus”. \nThis does a disservice both to taxonomy and to those that use the results of taxonomic research: \nsuch papers will rarely be cited because so much additional information is required of a user for \nthe paper to be readily comprehensible to them.  \nAuthor-formatted, not peer-reviewed document posted on 15/11/2024. DOI:  https://doi.org/10.3897/arphapreprints.e141840\n\nAt the opposite extreme, if a new genus is diagnosed within a family but how to tell \nmembers of that family from all others in the order is outlined, the user needs only to be able to \nidentify a specimen at the ordinal level for the information in the paper to be potentially useful. \nA good example of this was Lôpez-Pérez and Zaragoza-Caballero’s (2021) description of two \nnew genera of beetles of the family Telegeusidae in which they also explain how to differentiate \nthe family from all other beetles. In this instance, the user can get from more than 28000 genera \nto just one from a single research article (which included a global key to genera in the family). \nRecognition groups at the ordinal level were found in six instances (Suppl. material 2 column H). \nThis approach is useful as even beginner entomologists would likely be able to recognise most \ninsects at this level. Recognition groups at the family level were found in 58 instances. This \nmight be considered the lowest level at which a useful recognition group can be defined for a \nwide range of users in that keys to families within orders (either globally or geographically) \nshould be available to any committed entomological taxonomist. \n How expansive the information associated with a new taxonomic description is should \ndepend on the target clientele of the paper. Braby et al. (2024) state that a diagnosis “should \nprovide sufficient information to allow anyone to distinguish the new species from similar or \nclosely related species” [italics mine]. By extension, diagnoses of genera should also allow \nanyone to identify the genus, indeed, their higher taxonomic rank should make their \nidentification easier. Clearly these statements are too optimistic (not everyone has access to a \nmicroscope or literature for example), but it does seem that most new genus descriptions are \naimed at an audience comprising no more than the few other taxonomic experts in the same \ngroup. This is a pity given the increasing global interest in, and acknowledged importance of, the \nworld’s biota – especially given the repeated statements that the field of taxonomy is in crisis \n(e.g. Löbl et al. 2023). \n Of course, the title of a journal article should provide the reader with the information \nrequired to decide whether to proceed with the text of the paper as an identification tool relevant \nto specimens at hand. But in almost 40% of cases, the title suggested a broader (higher \ntaxonomic level) relevance than was provided, potentially wasting user’s time. This doubtless \nresults from the requirements of most journals to state the order and family that are the topic of \nthe paper. Loosening of that regulation so that either reference or recognition groups or both can \nbe provided, or overtly added, would seem useful and it would be preferable if stating what these \nare is mandatory. \n Some journals require a complete taxonomic breakdown for the new taxa at the \nbeginning of the taxonomy section. For example, the instructions to authors for the European \nJournal of Taxonomy state with respect to the results “[T]this section should start with \na contextual account of the current taxonomic hierarchy of the target taxon”. However, this can \nbe misleading as it suggests that the subsequent information is relevant to the lowest level \nprovided in that hierarchy. For example, in Biscaccianti et al (2022), the taxonomic hierarchy \nsuggest the diagnosis might differentiate the new genus from those within its tribe, whereas it is \nseparated from “all other known genera” of its family, which is much more useful. In the other \ndirection, Chang et al. (2021) describe three new genera, each of which is diagnosed only with \nrespect to one other genus, yet both the title and the taxonomic hierarchy suggest that tribal level \nAuthor-formatted, not peer-reviewed document posted on 15/11/2024. DOI:  https://doi.org/10.3897/arphapreprints.e141840\n\ndifferentiation might be expected and no key is provided to assist. These issues could easily be \navoided through overt statement of the reference and recognition groups. \n \nRecommendations \n \nBased upon the deliberations above, I provide the following eight recommendations. Although \nthe data upon which these ideas were generated were from newly diagnosed insect genera, I \nbelieve they should apply to all taxa at all levels in the hierarchy, including the species level. \n \n1. State reference and recognition groups overtly: “the diagnosis contrasts the new taxon with all \nothers in the subtribe Diagnostina (reference group), this subtribe can be differentiated from all \nothers in the family (recognition group) by….”. In some cases, there was some doubt as to \nwhether the authors meant what they actually said – for example if a genus is separated from all \nothers in its group but the paper’s title suggests that the taxa being considered are for a \ngeographically restricted subset of that taxon. \n2. State how both groups are understood. This would be analogous to the taxon concept (Meier \n2017; Packer et al. 2018). This recommendation is necessitated by the seemingly disparate ways \ndifferent authors sometimes understood the same taxon. In instances where there is taxonomic \ncontroversy as to group membership, authors should state which classification they are using. \nFor example, Scheffrahn et al. (2021) and Arias et al. (2021) place the Termitidae into different \norders and Ramos and Melo (2021) and Wood et al., (2022) put closely related bees into \ndifferent families. The best way to do this would be to cite a reference that defined the group – \n“sensu Melo and Gonçalves (2005) or Michener (2007) respectively for the previous example.  \n3. State the number of taxa that are in the reference and recognition groups. This not only \nprovides an additional potential check for the groups’ taxon concept but also gives the reader an \nidea of the complexity of the task of separating a specimen to be identified from others in the \ngroups.  \n4. State the reference and recognitions groups in the (sub)heading to the diagnosis and start the \ndiagnosis with information that permits differentiation of the reference group from the \nrecognition group. Alternatively, provide the characteristics separating reference from \nrecognition groups in an obvious position, immediately above the diagnosis perhaps. It was often \nnecessary to search the entire paper to find out whether a recognition group could be detected \nand to find the features that separate the reference group from it: why hide the information that \nenables a potential citer to know whether the paper is worth further study? [Of course, authors \nmay have mentioned the recognition group only “in passing” as their focus would have been on a \nmore narrowly defined taxon.] \n5. Provide information to differentiate reference from recognition groups briefly and use features \nthat are decisive. As with keys (Packer 2024) as well as diagnoses (Packer unpublished data) \nfeatures with exceptions or overlapping states are of limited utility as they are indecisive \nalthough if carefully worded, especially in the context of a combination of features, they may \nhelp exclude some taxa. Similarly, long descriptive text explaining how to identify the reference \ngroup will be more onerous for a user to get through: perhaps only a few features in combination \nwill be sufficient.  \nAuthor-formatted, not peer-reviewed document posted on 15/11/2024. DOI:  https://doi.org/10.3897/arphapreprints.e141840\n\n6. Discriminatory information should be easy for a user to access. As a counterexample, to \ndiscover the defining features to differentiate the reference group from the recognition group in \nChen et al (2021), the user must locate the characters that define the group from the phylogeny \ngiven in a supplementary figure (where the states are not provided though character numbers \nare), find the relevant matrix from among a set of supplementary tables wherein the 56 column \nand 55 row data matrix is provided to locate the relevant state numbers and then look at the \ncharacter state descriptions elsewhere in the supplementary material. In this example, users must \ngo through three unnecessarily time-consuming steps before obtaining an answer that could \neasily have been given in the body of the paper. \n7. The information provided to differentiate reference from recognition groups should be as easy \nto assess as possible. If the features provided include information from both sexes, then those \nfeatures will not be assessable from a single specimen or presumes users have associated the \nsexes correctly. If one or more of the features listed require dissection, unless they are the only \nsuitable features, the information will be unnecessarily time-consuming to evaluate.  \n8. Be consistent with respect to the reference or recognition group being used. For example, \nCrispolon et al., (2021) listed their new genus under a subtribe but in the discussion stated that it \ncould not be placed in a subtribe, causing some confusion for the reader.   \n \nConclusion \n \nDiagnoses will be made considerably more useful if authors, reviewers and editors ensure that \ntwo simple facts are overtly provided for each new taxon: a) what group is the diagnosis directly \nrelated to (the reference group) and b) how can that group be identified from among a broader \nrange of organisms that might be relatively easy for a user to distinguish (the recognition group).   \n \nAcknowledgements \n \nI thank Thomas Onuferko and Doug Yanega for comments on an earlier draft of this manuscript \nand Czarina Ortega for typing the list of references provided in Suppl, material 1. The impetus \nfor this paper came partly from comments received on an earlier manuscript of mine from \nWojciech Pulawski for which I remain thankful. \n \nReferences \n \nArias J, Boom A, Wang M, Clitheroe C, Šobotník J, Stiblik P, Bourguignon T, Roisin Y (2021) \nMolecular phylogeny and historical biogeography of Apicotermitinae (Blattodea: Termitidae). \nSystematic Entomology 46: 741–756. DOI: 10.1111/syen.12486 \n \nBiscaccianti AB, Esser J, Cuoco S, Grimaldi EG, Audisio PA (2022) West Palaearctic taxa \nformerly connected to the ‘old’ genus Atritomus Reitter, 1877 (Coleoptera, Mycetophagidae): \ntaxonomy, distribution, and description of a new genus. European Journal of Taxonomy 828: 61–\n74. http://dx.doi.org/10.5852/ejt.2022.828.1853 \n \nAuthor-formatted, not peer-reviewed document posted on 15/11/2024. DOI:  https://doi.org/10.3897/arphapreprints.e141840\n\nBorkent A (2021) Diagnosing diagnoses – can we improve our taxonomy? ZooKeys 1071: 43–\n48. https://zookeys.pensoft.net/article/72904/  http://dx.doi.org/10.3897/zookeys.1071.72904 \n \nBraby MF, Hsu Y-F, Lamas G (2024) How to describe a new species in zoology and avoid \nmistakes. Zoological Journal of the Linnean Society, advance access publication: \nhttps://doi.org/10.1093/zoolinnean/zlae043 \n \nChang YL, Wang T, Shi FM (2021) Three new genera and five new species of the tribe \nMeconematini (Ortchoptera: Tettigoniidae: Meconematinae) from Southwestern China. \nEuropean Journal of Taxonomy 751: 140–158. https://doi.org/10.5852/ejt.2021.751.1385 \n \nChen X, Freidberg A, Islam MS, Zhu, C (2021) Phylogeny of Adramini (Diptera, Tephritidae) \nbased on integrative evidence. Zoologica Scripta 50: 71–83. DOI: 10.1111/zsc.12436 \n \nCigliano MM, Braun H, Eades DS, Otte D (2024) Orthoptera Species File: \nhttp://orthoptera.speciesfile.org/ \n \nCrispolon ES, Guilbert E, Yap SA, Soulier–Perkins A (2021) New genus and new species of \nspittlebugs (Hemiptera: Cercopidae) from the Philippines. European Journal of Taxonomy 778: \n90–135. https://doi.org/10.5852/ejt.2021.778.1571 \n \nDorchin N, van Munster S, Klak C, Bowie RC, Colville JF (2022) Hidden diversity—A new \nspeciose gall midge genus (Diptera: Cecidomyiidae) associated withs succulent Aizoaceae in \nSouth Africa. Insects 13 75. https://doi.org/10.3390/insects13010075 \n \nGaimari SD, Havill NP (2021) A new genus of Chamaemyiidae (Diptera: Lauxanioidea) \npredaceous on Adelgidae (Hemiptera), with a key to chamaemyiid species associated with \nPinaceae-feeding Sternorrhyncha. Zootaxa 5067: 1–39. \nhttp://dx.doi.org/10.11646/zootaxa.5067.1.1 \n \nGonçalves CC, Domahovski AC (2021) A new genus and five new species of Neocoelidiinae \nleafhoppers from Brazil (Insecta: Hemiptera: Cicadellidae) with a key to males. European \nJournal of Taxonomy 775: 34–61.  https://doi.org/10.5852/ejt.2021.775.1545 \n \nHuang SY, Volynkin AV, Miao ZP, Tan SY, Wang M, Fan XL (2021) Molecular phylogeny and \nclassification of Nudariina (Lepidoptera: Erebidae) with emphasis on the genera Barsine Walker, \nAmmatho Walker and Ovipennis Hampson. Systematic Entomology 46: 1045–1059. \nhttp://dx.doi.org/10.1111/syen.12509 \n \nICZN [International Commission on Zoological Nomenclature] (1999) International code of \nzoological nomenclature. Fourth edition. London: International Trust for zoological \nNomenclature.  \n \nLee GE, Han T, Park H, Qi M, Li H (2021) A phylogeny of the subfamily Thiotrichinae \n(Lepidoptera: Gelechiidae) with a revision of the generic classification based on molecular and \nmorphological analyses. Systematic Entomology 46: 357–379. DOI: 10.1111/syen.12466 \nAuthor-formatted, not peer-reviewed document posted on 15/11/2024. DOI:  https://doi.org/10.3897/arphapreprints.e141840\n\n \nLöbl I, Klausnitzer B, Hartmann M, Krell FT (2023) The silent extinction of species and \ntaxonomists—An appeal to science policymakers and legislators. Diversity 15: 1053. \nttps://doi.org/10.3390/d15101053 \n \nLópez-Pérez S, Zaragoza-Caballero S (2021) Two new genera of Telegeusidae (Coleoptera) \nfrom Mexico. Revista Mexicana de Biodiversidad 92: 1–10. \nhttp://dx.doi.org/10.22201/ib.20078706e.2021.92.3613 \n \nMeier, R. (2017) Citation of taxonomic publications: the why, when, what and what not. \nSystematic Entomology 42: 301–304. http://dx.doi.org/10.1111/syen.12215 \n \nMelo GA, Gonçalves RB (2005) Higher-level bee classifications (Hymenoptera, Apoidea, \nApidae sensu lato). Revista Brasileira de Zoologia 22: 153–159.  https://doi.org/10.1590/S0101-\n81752005000100017 \n \nMichener CD (2007) The bees of the world.  Second edition. John Hopkins University Press, \n953pp. \n \nMotamedinia B, Skevington JH, Kelso S, Kehlmaier C (2021) First comprehensive, multigene \nmolecular phylogeny of big-headed flies (Diptera: Pipunculidae). Zoological Journal of the \nLinnean Society (accessed as a preprint in 2022) published in 2022: 195: 1200-1218 \nhttp://dx.doi.org/10.1093/zoolinnean/zlab094  \n \nPacker L (2024) A survey of keys for the identification of newly described insect genera: \nrecommendations for authors, reviewers, editors, and publishers. ZooKeys 1215: 65-90. \nhttps://doi.org/10.3897/zookeys.1215.130416 \n \nPacker L, Monckton SK, Onuferko TM, Ferrari RR (2018) Validating taxonomic identifications \nin entomological research. Insect Conservation and Diversity 11: 1–12. \nhttp://dx.doi.org/10.1111/icad.12284 \n \nRamos KS, Melo GAR (2021) Three new genera of Protandrenini bees from South America \n(Hymenoptera, Apidae, Andreninae). Revista Brasileira de Entomologia 65: e20210096. \nhttps://doi.org/10.1590/1806-9665-RBENT-2021-0096 \n \nRheindt FE, Bouchard P, Pyle RL, Welter-Schultes F, Aescht E, Ahyong ST, Ballerio A, \nBourgoin T, Ceríaco LM, Dmitriev D, Evenhuis N (2023) Tightening the requirements for \nspecies diagnoses would help integrate DNA-based descriptions in taxonomic practice. PLoS \nbiology 21(8):e3002251. https://doi.org/10.1371/journal.pbio.3002251 \n \nScheffrahn RH, Roisin Y, Akama PD and Šobotnik, J, (2021) Ebogotermes raphaeli, new genus \nand new species, an African soldierless termite described from the worker caste (Isoptera, \nTermitidae, Apicotermitinae). Zootaxa 5067: 279–284. \nhttps://doi.org/10.11646/zootaxa.5067.2.10 \n \nAuthor-formatted, not peer-reviewed document posted on 15/11/2024. DOI:  https://doi.org/10.3897/arphapreprints.e141840\n\nWanat M. (2021) New basal taxa of South African Apioninae (Coleoptera: Curculionoidea: \nBrentidae). Zootaxa, 5035: 1–60. https://doi.org/10.11646/zootaxa.5035.1.1 \n \nWinston JE (1999) Describing species: practical taxonomic procedure for biologists. Columbia \nUniversity Press, New York, 518pp. \n  \nWood TJ, Patiny S, Bossert S (2022)  An unexpected new genus of panurgine bees \n(Hymenoptera, Andrenidae) from Europe discovered after phylogenomic analysis. Journal of \nHymenoptera Research 89: 183–210. https://doi.org/10.3897/jhr.89.72083 \nAuthor-formatted, not peer-reviewed document posted on 15/11/2024. DOI:  https://doi.org/10.3897/arphapreprints.e141840","source_license":"CC-BY-4.0","license_restricted":false}