All 169 currently recognised Cyrtandra species found in Borneo were included in this key (Atkins et al. 2024). The two subspecies and six varieties are currently not included, but users are provided with further resources for species with intraspecific classifications on the relevant species pages. We followed the taxonomic concepts and character terminology of Bornean Cyrtandra by O.M. Hilliard and B.L. Burtt who, together, described 116 out of the 169 recognised Cyrtandra species (Burtt 1970, Burtt 1978, Burtt 1982, Burtt 1990a, Burtt 1990b, Burtt 1993, Burtt 1997, Burtt 1999, Hilliard and Burtt 2002, Hilliard et al. 2003, Hilliard and Burtt 2004, Burtt 2005, Hilliard 2005, Hilliard and Burtt 2005a, Hilliard and Burtt 2005b, Hilliard and Burtt 2006a, Hilliard and Burtt 2006b) and several in regional works on Brunei (Burtt 1997) and Mt Kinabalu (Beaman et al. 2001).

Botanical terminology can be subjective and used differently by different authors and within different taxonomic groups. This can lead to inconsistent descriptions of characters and a failure to correctly identify character states. To ensure consistent use of botanical terminology in this key, the species descriptions of Hilliard and Burtt were carefully studied to identify terminology used and morphological features associated with each term (Fig. 1). This terminology was then applied to species described by other authors to obtain consistent descriptions across species.

Figure 1.  

Workflow for homogenising botanical terminology across various source materials.

A character matrix was constructed to form the backbone of the key using all available data. The data sources used in this process were the original species descriptions, revised species descriptions by Hilliard and Burtt (Burtt 1970, Burtt 1978, Burtt 1982, Burtt 1990a, Burtt 1990b, Burtt 1993, Burtt 1997, Burtt 1999, Hilliard et al. 2003, Hilliard and Burtt 2004, Burtt 2005, Hilliard 2005, Hilliard and Burtt 2005a, Hilliard and Burtt 2005b, Hilliard and Burtt 2006a, Hilliard and Burtt 2006b), Hilliard’s unpublished illustrations of floral morphology of each Cyrtandra species, type specimens and further specimens identified by Hilliard and Burtt at the Royal Botanic Garden Edinburgh. Herbarium specimens identified by other authors were not included to stay consistent with the species concepts of Hilliard and Burtt.

The final set of characters used in the key was selected, based on taxonomic significance, discriminatory power and user-friendliness. Characters that were taxonomically insignificant (consistent across all species), highly subjective to describe, difficult to accurately assess or had limited data availability were eliminated from the dataset. The final dataset includes both geographic and morphological characters (Table 1). Most morphological characters can be easily observed on both living and herbarium materials, with a few characters of floral morphology requiring a 10x hand lens, microscope or a dissection kit to assess. Each species was scored for all characters for which the data were available. Some characters, such as the leaf shape, show considerable intraspecific variation and these characters were scored for all applicable character states to reflect the morphological diversity.

The character states within each character were carefully assessed and defined to clearly communicate the variation present. This included the definition of boundaries between character states, addition of character states to reflect morphological variation and reduction of complicated character states into more clearcut, broad categories. For example, the botanical terminology used to describe the shape of the bracts is highly inconsistent within the literature and the shapes do not form clear natural categories. Consequently, the relative length and width of the bracts (length to width ratio) was used as a proxy for bract shape to create consistent and objective categorisation.

While the absolute measurements of various organs can be a useful and objective character to record, it is difficult to estimate the amount of intraspecific variation present in a poorly-known species. Measurements presented in protologues and currently known herbarium specimens offer a good indication of the general size of plant organs, but do not necessarily capture the degree of variation present in natural populations. Vegetative characters appear particularly variable in certain Cyrtandra species. To take this into account, measurements were used in a very conservative manner as broad categories with the aim of filtering out the species with unusually large or small organs.

The final data matrix of the species scored for their character states was exported into the Xper3 platform (Vignes-Lebbe et al. 2016). The descriptive model was finished by adding detailed descriptions and illustrations for each character and character state to aid with identification. Characters were then manually assigned a weight between 1 to 5 to reflect their taxonomic significance, which prompts the programme to present these characters at the top of the key. Using the descriptive model and species data, Xper3 will automatically create a multi-access identification key.

Illustrations were provided as a visual aid for accurately interpreting characters and character states, as well as to support correct identification of species. Original illustrations were used where available; however, most Bornean Cyrtandra species were not published with illustrations. The remaining illustrations consist of photographs sourced from Royal Botanic Garden Edinburgh (RBGE) archives, photographs provided by Cyrtandra experts and collaborators, photographs of Cyrtandra growing in RBGE living collections and herbarium specimens, Hilliard’s unpublished illustrations of floral morphology (adapted and rescaled) and original illustrations, based on living materials, herbarium specimens and photographs. Having expert-verified illustrations for each species is crucial for confirming the identification. In groups where there are potentially undescribed species, the extra step of confirming the determination is essential.

The key was tested by the following user groups: the key authors, Cyrtandra experts and inexperienced users without expertise in Cyrtandra. The testing was conducted by participants using the key on a set of photographs taken in the field and of herbarium specimens. The participants recorded their identification results, as well as feedback on the usefulness and user experience on various aspects of the key. Based on this feedback, some of the character descriptions were updated and some ineffectual characters were removed to promote more accurate identification process.

There is a need for user-friendly taxonomic resources to facilitate biodiversity research within various scientific disciplines (Gotelli 2004). Online multi-access keys can answer this need by providing globally available open-access identification materials with features that make them particularly well-suited for non-specialist users. Online platforms allow easy incorporation of hyperlinks and multimedia, which can enrich the descriptions provided and facilitate knowledge sharing. Additional materials, such as detailed descriptions of specialist terminology used, are particularly important for the creation of user-friendly taxonomic resources and can be easily included in online materials.

Multi-access keys have several benefits over traditional dichotomous keys. Firstly, multi-access keys allow users to start the identification process at their character of choice, rather than following a set list of questions in a linear fashion. For example, characteristics of reproductive organs (flowers and fruits) usually bear high diagnostic significance in plants and, consequently, they are widely used in taxonomic keys. However, due to the seasonal and ephemeral nature of these organs, they can often be missing from specimens. Multi-access keys allow users to avoid getting stuck with questions they cannot answer due to the absence of specific characters in their data by allowing them to choose which questions to answer. This can also increase the accuracy of the identification by allowing preferential use of characters which the user can identify with high confidence and which can be particularly beneficial for non-specialist users.

Based on the questions the users choose to answer, this process may not result in species-level identification. However, in the Xper3-platform, it is possible to constantly see the names and characteristics of the remaining species while using the key, which allows the user to easily manually assess the remaining species. In our key, this includes additional information such as illustrations and hyperlinks to the type specimens that are available online, which can support the identification process beyond the scored data.

The non-linear process also allows users to select multiple character states at any point. While this increases the number of questions required for identification, the lack of mutually exclusive selection facilitates the incorporation of highly variable characters, as taxonomic units do not need to fit to a single character state. In Cyrtandra, for example, leaf shape can either be relatively constant within species or show a high degree of intraspecific variability. By scoring a species’ leaf shape for all matching shapes, the diversity can be represented without having to create a separate category for each unique combination of shapes. Thus, multi-access keys offer an alternative to traditional dichotomous keys and may be particularly well suited as a tool for studying diverse taxonomic groups and for creating accessible user-friendly resources.

Highly diverse taxonomic groups such as large plant genera can present considerable challenges to taxonomists because of the unmanageable numbers and potential for complex relationships between species due to rapid diversification and they have often been neglected in the past for this reason (Frodin 2004, Moonlight et al. 2018, Atkins et al. 2021). It is anticipated, however, that new methods and data availability combined with global collaborations may make it more feasible to tackle these large groups (Moonlight et al. 2024).

The process of constructing taxonomic keys traditionally heavily relies on having a well-resolved taxonomy of the target group. This may not be realistic for many highly diverse taxa due to the limits of our current knowledge, which may create the question of the usefulness of identification resources in such groups. Here, we argue that an identification key in the absence of a well-resolved taxonomy can be a stepping stone in taxonomic research, as it allows easy comparison between species and faster identification of specimens, in the context of our current taxonomic understanding of them. By systematically checking all species names in a group, a key can also aid in recognising undescribed species and provide a foundation for their formal description. However, we emphasise that, without a well-resolved taxonomy, the purpose of an identification key needs to be redefined and the key construction must clearly identify the source information of its taxonomic backbone. In this key, for example, the taxonomy is based on the species concepts of Hilliard and Burtt.

When used conservatively, a multi-access taxonomic key can be an excellent aid in research. The creation of such a key itself collates a dataset of taxonomically informative characters. In the Xper3 platform, the authors of a key can access other functions, such as comparing different pairs or groups of species using these data. The key can be used to rapidly narrow down species based on characters of interest, without necessarily attempting to identify specimens to species level. In a genus with more than 100 species, narrowing down to a manageable number will still significantly speed up the process of identification.

Finally, the Xper3 platform allows updating the dataset as new information accumulates, which is a useful feature in a group in which new discoveries are expected. The species delimitations or descriptions of our key may change in the future to reflect this progress in taxonomic research, making it a dynamic tool for both research and identification.