Getting science priorities straight: how to increase the reliability of specimen identification?

Abstract

‘We advise the authors to find a native English speaker to proofread the manuscript’. This is a standard feedback journals give to non-native English speakers. Journals are justifiably concerned with grammar but do not show the same rigour about another step crucial to biological research: specimen identification. Surveying the author guidelines of 100 journals, we found that only 6% of them request explicitly citation of the literature used in specimen identification. Authors hamper readers from contesting specimen identification whenever vouchers, identification methods, and taxon concepts are not provided. However, unclear taxonomic procedures violate the basic scientific principle of reproducibility. The scientific community must continuously look for practical alternatives to improve taxonomic identification and taxonomic verification. We argue that voucher pictures are an accessible, cheap and time-effective alternative to mitigate (not abolish) bad taxonomy by exposing preventable misidentifications. Voucher pictures allow scientists to judge specimen identification actively, based on available data. The popularization of high-quality image devices, photo-identification technologies and computer vision algorithms yield accurate scientific photo-documentation, improving taxonomic procedures. Taxonomy is timeless, transversal and essential to most scientific disciplines in biological sciences. It is time to demand rigour in taxonomic identifications.

‘We advise the authors to find a native English speaker to proofread the manuscript’. Non-native English speakers commonly receive this feedback during the publication process (although being a native is not the sine qua non for academic English proficiency, see [1]). This concern is justified because poor writing may confound readers and overshadow the findings. However, just like words need expert assessment, so does specimen identification. Since species are the fundamental units of biology, accurate specimen identification underpins all biological research. Unlike grammar mistakes which may compromise a sentence's meaning, species misidentification compromises the whole study, weakening scientific integrity. A misidentification triggers a cascade effect in the worst-case scenario, accumulating spurious information around a subject (see examples in [2]).

The ‘author guidelines’ for most journals describe strict instructions to improve the readability and impact of papers (e.g. English language editing, academic illustration, figure formatting, and graphical abstract design). Surveying submission guidelines and editorial policies of 100 journals from different biological areas and publishers (electronic supplementary material, table S1), we found that most of them (83%) encourage authors to deposit the data from which published results are derived (e.g. trees, scripts, nucleotide sequences) in public databases. Unfortunately, few instructions related to the taxa per se are provided: 32% of the journals require the author to deposit voucher specimens in scientific collections, whereas only 6% request explicitly the literature used to identify the taxa of study. Moreover, Packer et al. [3] found that just 50% of the papers complied with vouchering recommendations. This lack of explicit editorial policies raises the question of whether journals are sufficiently committed to providing information on specimen identification.

Concerning the taxonomic treatment of zoological papers, a survey conducted by Monckton et al. [4] showed that only 10.7% of papers cited taxonomic identification methods, 6.9% indicated taxon concepts and 29.2% made available vouchers—that is, 70.8% of papers excluded the possibility for double-checking taxonomic identifications permanently. Thus, authors hamper readers from contesting specimen identification whenever vouchers, identification methods and taxon concepts are not provided. Taking a step back, reviewers and editors do not have access to the raw data (i.e. specimens) during the review process for a taxonomic verification. In cases where specimens are vouchered, a researcher may have partial access to the vouchers (e.g. by mail, photographs or visiting the collection where they are deposited), although it demands funding and goodwill. In other words, neither the editor, the reviewers, nor the readers have easy access to the specimens used in the papers to double-check identification.

The lack of supporting information justifying or guaranteeing a careful identification procedure is a usual failure in biological sciences [2] that plays against the basic scientific principle of reproducibility [5]. Thus, suggestions to attenuate the subjectivity of specimen identification and good taxonomic practices have been under debate (e.g. [6–10]). For instance, Meier [11] suggested explicit taxonomic identification methods and taxon concept statements supported by references. Similarly, Bortolus [6] advocated adoption of a ‘Taxonomic Verification’ section by journals, allowing future taxonomic validation of the specimens under study. These suggestions take little printed-space and can be readily implemented, unless the researcher thinks of specimen identification as an unimportant step—a much bigger problem [5,11]. Besides violating scientific principles, omission to cite taxonomic papers used for specimen identification is also disrespectful to others' intellectual production, an ethical dilemma. Verifying and using the updated literature as a regular procedure for researchers may avoid mistakes beyond the published papers, improving the quality of data deposited in public databases, a key source of information for current research (see also [12,13]).

In the past decades, online databases have expanded and transformed scientists’ use of research data [14]. Along with a straightforward process to deposit data and promising sources for users, public databases have their inherent pitfalls. Inadequate taxonomic procedures (e.g. [15–18])—that is, bad taxonomy sensu Winston [19]—may now spread misidentifications at a fast rate. For instance, misidentifications and problematic taxonomic meta-data are a recurrent issue in the Global Biodiversity Information Facility (GBIF) [20,21]. Furthermore, in the Barcode of Life Data System (BOLD), misspellings and invalid names may exceed 10% of deposited sequences for a taxon, outnumbering the records with poor-quality sequences and compromising the integrity of databases (see [22]). Published/public misidentifications affect many biological fields, such as conservation [23], invasive species management [24], product traceability [25], and evolutionary biology [26].

Specimen identification requires fundamental expertise since the taxonomic literature may be hostile owing to the technical language and the laboured methodological procedures, even for taxonomists. Thus, the mention of a taxonomist could be expected in papers not strictly taxonomic. But, for example, more than half (62.5%) of studies on community ecology neither have a taxonomist among the authors, nor acknowledge taxonomists for specimen identification or cite taxonomic literature to support taxa identification [2]. It brings into question the quality of the identification of those specimens deposited in public databases owing to the unclear identification process. Again, the end-users are often tried to contest the identification. Most databases provide tools to flag incorrect information and deter bad taxonomy, but this process is possible only when supporting information—such as collection site, voucher pictures or literature used in the identification—is provided.

Similar to how journals request deposition of nucleotide sequences in public databases to enable the reliability and replicability of studies, an additional simple request could improve taxonomic verification: voucher pictures. Images are always powerful allies of taxonomy, and the inclusion of voucher pictures allows a first taxonomic verification in a few mouse-clicks. Researchers should be aware of what (e.g. specimen view, emphasis of structure) maximizes the usefulness of voucher pictures for the taxa under study. As long as the data (e.g. nucleotide sequence, occurrence record) are linked to voucher pictures, the end-user can contest the identification (see [27]) and then opt for using or excluding those data from future analyses.

The popularization of high-quality image devices has changed how science records data, and photographs have been used beyond taxa descriptions in a myriad of different studies. With the current availability of electronic image devices and basic training on specific taxa, accurate scientific photo-documentation can even be conducted in the field by amateur (or nonprofessional) scientists (see [28,29]). One may argue that taking specimen pictures is not feasible for researchers who access large amounts of specimens—but taking scientific photographs is cheaper (most of the times costless) and much less time-consuming than other standard research procedures, such as DNA extraction, preparation for scanning electron microscopy and micro-CT (see examples in [30,31]). Furthermore, time-efficient systems for automatic specimen digitization have been designed to take pictures of thousands of specimens in a few minutes (e.g. [32–34]). Pictures may mitigate (not abolish) bad taxonomy by exposing preventable misidentifications that would not be noted if images were not provided (e.g. BOLD record ASAHE106-12 is identified as Euschistus tristigmus (Say) (Hemiptera: Pentatomidae), but it is clearly a non-pentatomid immature [35]).

Presuming that voucher pictures are a panacea for misidentification is a naive position. Photo-identification may be puzzling, sometimes impossible, for reaching lower taxonomic levels (e.g. nematodes [36], insects [37], chiropterans [38]; but see [8]). But another reason to embrace specimen photographs is to develop and popularize photo-identification technologies and computer vision algorithms, already used for diverse taxonomic groups (e.g. [39–42]). Moreover, machine learning and neural networks have improved specimen identification accuracy even using those photos considered of low-quality under human eyes [32]. Undoubtedly, pictures will soon be evaluated by databases' algorithms similarly to how nucleotides sequences are routinely analysed in BLASTn (http://www.ncbi.nlm.nih.gov/) (e.g. [31]). New possibilities to aid taxonomic identification are welcome; here, we encourage the use of voucher pictures—an overlooked, low-cost and feasible alternative—to lessen the problems mentioned above.

Taxonomy is timeless, transversal and essential to most scientific disciplines in biological sciences and demands immediate strict rigour in taxonomic identifications. Journals should explicitly acknowledge the importance of taxonomic identification and verification, being even tougher than they are on grammar, for instance. We argue that the availability of voucher pictures of specimens used in scientific research would increase reliability and allow identifications to be contested. Our suggestion strengthens the many other practicable alternatives already proposed to cope with this problem (e.g. [9,11,27]). We must continuously look for practical alternatives to improve taxonomic identification and taxonomic verification. Reviewers and editors may encourage authors to make their taxonomic identification clear, replicable and verifiable while pushing forward the proposal of including these directions in the author guidelines of the journals they work for.

Data accessibility

The data are provided in electronic supplementary material.

Authors' contributions

All authors contributed equally to the conception and preparation of the article.

Competing interests

We declare we have no competing interests.

Funding

F.M.B. is supported by a postdoctoral fellowship from Coordenação de Aperfeiçoamento de Pessoal de Nível Superior—Brasil (PNPD/CAPES)—Finance code 001; L.T.G. is supported by a doctoral fellowship from Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq).