A linear sequence to facilitate curation of herbarium specimens of Annonaceae

Summary. This paper provides a linear sequence of four subfamilies, 15 tribes and 106 genera of the magnoliid family Annonaceae, based on state-of-the-art and stable phylogenetic relationships. The linear sequence facilitates the organisation of Annonaceae herbarium specimens.


Introduction
Plant taxonomy is a scientific expression for one of the defining characteristics of the human species: observing, assembling and classifying. The ordering of the plant world has been attempted ever since the origin of modern man, and even before. Neanderthals were able to distinguish fruit, nuts, roots, bulbs and tubers that were exploited as food resources (Henry et al. 2011), and upper palaeolithic hunter-gatherers recognised and categorised plants for economic and ritual uses (Nadel et al. 2013;Power et al. 2014). From Theophrastus onwards, in the fourth century BC, botanists have attempted to organise plants into classification systems. The sexual system published by Linnaeus (1753) in his Systema Naturae is the classical example of a classification system that has been designed for convenience, most notably to facilitate plant recognition and identification, and unequivocal communication about plants.
In Linnaeus's time the practice of drying and conserving plants for future study was well established. In the first half of the 16 th century, the Bolognese botanist Luca Ghini introduced a new way of studying plants by making the earliest hortus siccus. By pressing plants and storing them in a book, he invented the herbarium. It is in this era that botanic gardens, illustrated botanical publications, and herbaria were established as a trinity of resources for botanical sciences, a foundation that is still fundamental to botanical research today.
Botany in the eighteenth and nineteenth century to a large extent involved the rejection of Linnaeus' artificial system, replacing it with classifications that reflected supposed evolutionary relationships based on careful observations of plant characters. This endeavour was greatly facilitated by collections that arrived in Europe from all over the world and were kept in newly established flourishing herbaria. Up to that point, private ownership of plant collections had been common practice. In the mid-19 th century, however, these collections were often sold to the burgeoning herbaria, with the specific goal of making the collections available for study by staff and visitors. Until then, classifications such as those by Linnaeus and de Jussieu (1789) had primarily been based on European temperate plants. The influx of samples representing the wide plant diversity in colonial territories challenged these classification systems, with many non-temperate plant groups such as Annonaceae that were largely unknown to European botanists. Proliferating collections and botanical studies resulted in natural classifications by, e.g., Bentham & Hooker (1862-1883 and Engler & Gilg (1924), which have been the basis for taxonomic literature and for the arrangement of herbaria and botanic gardens for a long time. Given the Herculean task of changing the classification system followed in any sizeable herbarium (e.g. Wearn et al. 2013;Le Bras et al. 2017), many herbaria are still organised to date on the basis of outmoded classification systems going back in time a century or more.
Over the past two decades or so, phylogenetic systematics has resulted in a notable transformation of the classification of plants, and of angiosperms in particular. Based on the results of phylogenetic analyses, initially the delineation of angiosperm orders and families was evaluated and changed if necessary to make plant families comply with the prime guiding criterion of monophyly (APG I 1998;APG II 2003;APG III 2009;APG IV 2016). Subsequently, working groups of systematists have applied the results of phylogenetic analyses to revise infrafamilial classifications (e.g. Schneider et al. 2014;Bone et al. 2015;Chacón et al. 2016;Claudel et al. 2017;De Faria et al. 2017;Simões & Staples 2017), an endeavour that is still ongoing. Systematists have spent great effort in revising the classification of angiosperms because of the awareness that phylogenetics has brought methodological rigour to systematics and predictivity to classifications, which enabled the treatment of phylogenetic relationshipsand therefore of classificationsas testable hypotheses, rather than opinions of scientists, however scholarly they might be.
Recently, the herbarium of the Royal Botanic Gardens, Kew, was reorganised following the APG III system at the family level and taking phylogenetic classifications into account at the infrafamiliar level. Linear sequences of plant taxa enable curators to curate herbarium collections in accordance with phylogenetic relationships among genera. Linear sequences reflect the order of names attached to the tips of a phylogenetic tree, after the branches in the tree have been ordered according to some projection method. Alternatively, herbarium collections may be organised alphabetically, and the choice between an arrangement based on alphabet or on classification has been cause for debate (Funk 2003;Burger 2004). Storing collections according to any organising system remains indispensable as herbaria have retained their historic functions, being the basis for plant systematics and taxonomy, floristics and identification, assessment of botanical diversi ty, and teach ing . In additi on, s ci ent i fi c developments have unlocked new applications of herbarium collections, such as the characterisation of phenological responses to climate change (Willis et al. 2017), the assessment of global rarity of plant species to guide conservation (bioquality; Marshall et al. 2016), the sequencing of near-complete plastomes (Bakker et al. 2016;Hoekstra et al. 2017) and the targeted enrichment of nuclear genes (Hart et al. 2016), both for phylogenetic and evolutionary studies.
In this paper, we present a linear sequence of genera of Annonaceae. Generally, the family is among the most species-rich and abundant families in tropical rain forest communities (e.g. Cardoso et al. 2017;Sosef et al. 2017;Turner in press) and is amply represented in major herbaria. Haston et al. (2007Haston et al. ( , 2009) published a simple methodology for translating tree-like relationships into a linear sequence, and applied this to a phylogenetic tree of angiosperm families. Similarly, linear sequences have been produced for gymnosperms (Christenhusz et al. 2011a) and lycophytes and ferns (Christenhusz et al. 2011b). In order to extend the phylogenetic arrangement of collections to the level of genera, linear sequences that translate family phylogenies are indispensable. So far, linear sequences are available for Fabaceae (Lewis et al. 2013), and monocots excluding Poaceae and Orchidaceae (Trias-Blasi et al. 2015).

Linear sequences
The assembly of the phylogenetic tree underpinning the linear sequence, and the translation of the tree into the sequence consisted of the following steps.
Y a summary tree showing relationships of all genera of Annonaceae was assembled. Details are given below, in the section 'Annonaceae classification'. Nodes that did not receive significant support (parsimony or maximum likelihood bootstrap percentages, Bayesian posterior probabilities) in any of the published studies were resolved according to the topology most frequently inferred in all used publications. Y we defined clade size in terms of number of species, and not number of higher taxa (e.g. number of genera to define the size of tribes) as the former estimate of clade size can be expected to be more stable than the latter, i.e. more robust to changing taxonomic concepts (Hawthorne & Hughes 2008). Y species numbers for all genera were taken from Annonbase (Rainer & Chatrou 2006). Y following Haston et al. (2007), nodes of the phylogenetic tree were rotated in such a way that clades with fewer species were placed before clades with more species. This clade size criterion was applied subsequently to all nodes in the tree, starting from the root node (Fig. 1). The names along the tips, reading down from the top, represent the linear sequence.

Annonaceae classification
Historically, botanists have been reluctant to provide a classification for genera in the magnoliid family Annonaceae. Even though subfamilies and tribes were described by eminent botanists such as Rafinesque (1815), Endlicher (1839), Hooker & Thomson (1855) and Baillon (1868), these were hardly used by Annonaceae workers at the end of the 20 th century, just before the breakthrough of phylogenetic methods. The classification most frequently referred to was the one by Fries (1959), who identified informal groups of genera but was reluctant to solidify his arrangement into a formal classification.

Linear sequence of Annonaceae
Accepted names are listed in bold and synonyms in italics. We listed Unona L.f. and Uva Kuntze as synonyms of Xylopia and Uvaria respectively, as the type specimens of the former two genera have been put into synonymy of the latter two. Note, however, that species previously classified in Unona or Uva can now be found in dozens of genera of Annonaceae. We considered it beyond the scope of this paper to include details on revisions and other taxonomic information, for which we refer to a recent overviews (e.g. Maas et al. 2011;Erkens et al. 2012) and continuously updated taxonomic data in Annonbase (Rainer & Chatrou 2006) (Fig. 1a) and Guatteria aeruginosa Standl. (Fig.  1b). PHOTOS