Corresponding author: Gabriela Zuquim ( firstname.lastname@example.org )
Academic editor: Angelo Troia
© 2017 Gabriela Zuquim, Hanna Tuomisto, Jefferson Prado.
This is an open access article distributed under the terms of the Creative Commons Attribution License (CC BY 4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Citation: Zuquim G, Tuomisto H, Prado J (2017) A free-access online key to identify Amazonian ferns. PhytoKeys 78: 1-15. https://doi.org/10.3897/phytokeys.78.11370
There is urgent need for more data on species distributions in order to improve conservation planning. A crucial but challenging aspect of producing high-quality data is the correct identification of organisms. Traditional printed floras and dichotomous keys are difficult to use for someone not familiar with the technical jargon. In poorly known areas, such as Amazonia, they also become quickly outdated as new species are described or ranges extended. Recently, online tools have allowed developing dynamic, interactive, and accessible keys that make species identification possible for a broader public. In order to facilitate identifying plants collected in field inventories, we developed an internet-based free-access tool to identify Amazonian fern species. We focused on ferns, because they are easy to collect and their edaphic affinities are relatively well known, so they can be used as an indicator group for habitat mapping. Our key includes 302 terrestrial and aquatic entities mainly from lowland Amazonian forests. It is a free-access key, so the user can freely choose which morphological features to use and in which order to assess them. All taxa are richly illustrated, so specimens can be identified by a combination of character choices, visual comparison, and written descriptions. The identification tool was developed in Lucid 3.5 software and it is available at http://keyserver.lucidcentral.org:8080/sandbox/keys.jsp.
Amazonia, Ferns, Identification key, Indicator species, Lucid3, Free-access key, Online identification tools, Pteridophyte, Tropical forests
“Science is a sequence of generating new ideas, detailed explorations, incorporation of the results into a toolbox for understanding data, and turning them into useful knowledge.”
The first species identification key was published more than 330 years ago (
The rise of the Internet has been a significant development for systematics (
With the rapid progress in online approaches, taxonomy has become more accessible and integrated to the society. Newly developed software like Lucid (http://www.lucidcentral.com/), FRIDA (
In the last 20 years, a wide range of tools has been developed to address different societal demands on species identification. Crowd source identification (http://www.inaturalist.org/) and taxonomic platforms (http://scratchpads.eu/), mobile apps (http://bien.nceas.ucsb.edu/bien/tools/plant-o-matic/, http://leafsnap.com/), and even games (
With all these facilities, the contribution of non-experts to the accumulation of biological data is increasing, and has already been remarkably successful in providing data on species occurrences and distribution (
Biodiversity research and conservation actions are heavily dependent on the availability of adequate species identifications, whether the aim is to understand species origins, patterns in spatial distribution, or responses of organisms to human impact (
It has been suggested that ferns are good indicators of environmental conditions, forest types and general floristic patterns (
There is an urgent need for more biological data to improve conservation planning in poorly known areas (
We developed a free-access online identification tool for Amazonian ferns in order to provide a relatively easy and up-to-date resource for species identification. Our aim was to specifically address the problems mentioned above, and thereby to stimulate the collection of ferns and accumulation of data about them.
We started the key development by compiling a preliminary list of species and the morphological characters we thought are most useful when identifying them. Preference was given to such external characters that can be observed in the field or in the herbarium with the naked eye. The next step was to decide how the characters and their states are communicated to the user. Rather than programming a key platform from scratch, we decided to use the specialized key design program Lucid v. 3.5 for this purpose. Lucid v. 3.5 treats the taxa as entities and morphological characters as features that can have two or more states. The characters are implemented in Lucid by scoring, for each feature, the states that are present in each taxon. For example, the feature “nodes on the petiole” may be scored to have the state “present” for one species (entity) and “absent” for another. A single taxon can have more than one state scored in a feature, so if a species has nodes on some petioles but not all, both states can be scored. Another example from the present key is the feature “leaf architecture” of Asplenium pearcei Baker, which is scored for the states “entire”, “lobed” and “pinnate” (Figure
Morphological variation in Asplenium pearcei Baker. Three fertile individuals of Asplenium pearcei Baker showing different patterns in leaf dissection. A entire leaves B pinnate leaves and C lobed leaves. All states were scored in the features table.
Lucid allows scoring features in two different layout modes. The spreadsheet mode shows all features and all entities at the same time, and is useful for mass scoring of features, as well as to get an overview of the data (Figure
Structure of the key under development in LucidBuilder. Screenshots of the key been developed in Lucid Builder V3.5. A spreadsheet scoring table view B tree view mode of the scores for a single species C an example of features positive relationship: the plus sign associated with “number of included free veinlets” and “orientation of included free veinlets” shows that these features are controlled by the highlighted state (“usually present” under “included free veinlets”). All controlling states for some feature in the key are marked in red D frequency plot of the number of differences among pairs of entities and a list of those entities that have at most 10 different scores among them.
There are more than 500 described species in the paraphyletic group known as pteridophytes (ferns and lycophytes) in Brazilian Amazonia (
The list of taxa in the key is not complete yet, but accommodating more families, genera, and species is straightforward. The features can be applied to any fern species, and more states can be added if the new species are not adequately described by the existing states. For example, at the moment the key contains no species that are hemidimorphic (only a part of the leaf is fertile). When a hemidimorphic species (such as Anemia) gets added, a state “leaf divided into a fertile and sterile part” could be added to the feature about the morphology of fertile leaves in relation to sterile ones.
Features that are relevant to a few taxa only were included in separate subkeys. For example, the number of cells in the hairs on the lamina is a diagnostic feature in species of the genus Triplophyllum Holttum, but this is a rather difficult character to observe, and in most other genera it is uninformative. Therefore, the feature was only included in the Triplophyllum subkey, which is then embedded in the main key. The subkey can be accessed either as link through the main key or independently from the key portal (http://keyserver.lucidcentral.org:8080/sandbox/keys.jsp). More subkeys are under development for other taxa.
Subkeys can also be used more generally to provide modular structure to the key. They can be developed independently for a taxonomic group of interest or for a geographic region and then linked to the main key. If new geographical areas were to be included in the key, a practical option might be to add region and/or biome as one of the features, so the user can subset the key to the geographic area of interest (for example, Amazonia, Panama, Neotropics, Andes, etc.).
Regardless of the identification key format, some morphological features are only meaningful in the presence of a specific state of another feature. For example, “number of pinna pairs” is a relevant feature only among those ferns that have pinnately divided leaves in the first place. A conditional list of features is implemented in Lucid by ascribing logical relationships among characters (or dependencies, following Lucid´s terminology). In other words, Lucid keys can be designed such that the feature “number of pinna pairs” is unavailable (i.e., invisible to the user) until the user scores the state “pinnate” in the feature “leaf architecture”. At that moment, features that are relevant for pinnate leaves automatically appear in the features list. This is called positive dependency (a dependent feature appears when a controlling state is selected). Lucid also allows for negative dependency, which means that the dependent feature is initially available, but disappears from view when a controlling state is selected in another feature that makes it irrelevant. The planning of dependencies is an important part of key design, because when they are well used, they help to keep the features list concise and the appearance of the key more inviting. Our fern key uses positive dependencies but not negative ones. For example, the feature “number of veinlets” is not visible when the key is launched, but it has a positive dependency with two of the states in the feature “included veinlets”, namely “usually present” and “rarely present” (Figure
Each species in our key is scored for a higher taxon membership, i.e. genus and family. This provides a quick possibility to subset the key when the user already knows the genus or family. Scoring one of these taxa for the specimen to be identified causes the key to work as a key to species within that taxon only. Taking into account that the key is still missing many species known to occur in Amazonia, we recommend that the user checks from other sources if it seems that none of the species of the key matches the plant needing identification.
We used a template to transform the scores of each entity into automatically generated text descriptions (Figure
Available contents of the online fern keys. Screenshots of the online key. A Entrance page with a list to all available keys B an overview of the key after two features have been scored, (C) link to possible identification occurrence map in GBIF D description of the entity that possibly matches the plant needing identification.
We included more than 5,000 illustrations (photos and drawings) in the key to document the morphological variation within each species and to illustrate the features and their states that were used in the character list. Some of the photos are associated with voucher specimens deposited in TUR, INPA, or SP Herbaria. Species names were updated according to PPG I (2016).
We tried to avoid botanical jargon, but sometimes the precise scientific terms were needed. Then we illustrated them and/or defined them in a glossary that can be accessed in the following link: http://www.utu.fi/en/sites/amazon/publications/Pages/Glossary.aspx
Lucid identification keys are structured in four tiles containing (a) available features, (b) chosen features, (c) entities remaining, and (d) entities discarded (Figure
Of the 45,451 possible combinations of pairs of taxa, 99.3% pairs were differentiated from each other by more than 10 features (not counting “Family” and “Genus”) (Figure
Since ours is a free-access key, the user can select and score the features in any order, and not all features need to be scored. This represents an important improvement in relation to the usual fern identification keys, which have a single-entry structure. Having to identify a specimen using a pre-determined sequence of features is problematic especially because most keys start with reproductive characters, which are absent in many (if not most) of the individuals one is likely to encounter in the field. The present Amazonian fern key includes features related to both fertile and sterile leaves independently, which makes it possible to identify individuals regardless of its reproductive state. The key can be used to identify fertile fern individuals of any size, but sterile individuals only if they have leaves longer than 10 cm. Although it is possible to identify even smaller juveniles than this, their features are difficult to express, and adding them could make the key confusing.
One possible disadvantage of free-access keys is that the high number of choices can confuse the user (
A subkey to identify Neotropical Triplophyllum (http://keyserver.lucidcentral.org:8080/sandbox/player.jsp?keyId=5&thumbnails=true&gallery=true) based on (
Preliminary versions of our key were tested both by pteridologists and by non-specialists, and several changes were done based on their feedback. Workshops were carried out in Finland, USA, Peru, and Brazil (https://amazonkey.wordpress.com/). In Peru and Brazil, university students went to the field to collect ferns and the fresh specimens were identified using the key. The workshop in Finland focused on preserved material. In USA, the workshop was arranged during the international conference “Next Generation Pteridology (2015)” and the focus was on discussing the structure of the free-access key with fern specialists.
Some outcomes of the workshops were: 1) Even though the order in which the features are presented is not relevant in a free-access key, users tended to score the features in the order that they were listed. Therefore, features were re-sorted such that they start from the easiest and most informative ones. 2) Among the material collected in the field, juvenile individuals often remained unidentified or were even misidentified. This was especially the case with species whose habit and/or laminar dissection characters are very different in juveniles and adult plants. For example, juvenile Lomariopsis, which have simple entire leaves, were sometimes misidentified as Asplenium serratum or Elaphoglossum. Therefore, we added juveniles as separate entries that can be identified independently of the adult plants. For example, Lomariopsis prieuriana Fée now appears twice in the entities list of the key: one entry refers to the pinnate adult form and the other to entire-leaved juveniles.
Ferns can be used as indicators of environmental conditions in Amazonia, so mapping fern species distribution can contribute to producing habitat maps, to describing biogeographical patterns and to conservation planning. For all these purposes, a limiting factor is the poor availability of georeferenced and accurately identified species observations. In order to assist in the species identification problem, we have developed a user-friendly free-access key that is available online and summarises some of the existing taxonomic information about Amazonian ferns.
Online keys can be designed such that they require little taxonomic background knowledge, if they focus on intuitive and/or well-explained morphological characters. Our key is almost entirely based on features that are easily observable with the naked eye, and we avoided jargon as much as possible and clarified the terms when needed. Especially the taxa but also the features are richly illustrated, which allows visual comparison between the specimen and the candidate species. We also tried to keep the appearance of the key simple by taking advantage of tools that allow keeping the visible feature list concise and to quickly subset the features and species lists. More species and features can be added to the key any time. Subkeys to the larger and more difficult families and genera can be embedded, similarly to the current Triplophyllum subkey. The structure of the key is dynamic and flexible, so in addition of being immediately useful, there is ample room for further development. The next phases will be to expand the key to cover the rest of Amazonian fern flora, and preferably species of adjacent areas as well and to develop more subkeys. We would like to invite colleagues to collaborate on this endeavor.
We thank the participants and organizers of the workshops in Brazil, Peru and the USA for their feedback and Robbin C. Moran, Michael Sundue, Gabriel Moulatlet, Kari Kaunisto, Denise Sasaki, and Weston Testo for providing pictures for the key. We thank the Brazilian and Peruvian institutions for collection permits and assistants for invaluable help during fieldwork. We also thank Leif Schulman, Flávia Costa, Glenda G. Cárdenas, and Samuli Lehtonen for invaluable help in developing the key and Sandra Knapp, Alan Smith and Stefano Martellos for valuable comments to the manuscript. The project was funded by Academy of Finland (grants to HT) and Brazilian National Council for Scientific and Technological Development/PPG7.