A new large-flowered species of Andeimalva (Malvaceae, Malvoideae) from Peru

Abstract Andeimalvaperuviana Dorr & C.Romero, sp. nov., the third Peruvian endemic in a small genus of five species, is described and illustrated from a single collection made at high elevation on the eastern slopes of the Andes. Molecular phylogenetic analyses of nuclear ribosomal ITS sequence data resolve a group of northern species of Andeimalva found in Bolivia and Peru from the morphologically very different southern A.chilensis. The new species bears the largest flowers of any Andeimalva and is compared with Bolivian A.mandonii. A revised key to the genus is presented.


Introduction
The genus Andeimalva J.A. Tate (Malvaceae, Malvoideae) was created to accommodate four species found in the Andes of South America from northern Peru to central Chile and includes three species previously placed in Tarasa Phil. and one in Malacothamnus Greene (Tate 2003). Krapovickas (1965) described two of these species, T. machupicchensis Krapov. and T. spiciformis Krapov., and allied them with a third, T. mandonii (Baker f.) Kearney, because he observed that the three differed from all other species of Tarasa in having entire lanceolate leaves more than three times as long as wide and awned mericarps with smooth (versus reticulate) sides. Despite the distinctiveness of this group, Krapovickas (1965) chose not to recognize it formally. The fourth species of Andeimalva, endemic to Chile, was long known as M. chilensis (Gay) Krapov. even though its placement in Malacothamnus was problematic based on its morphology, cytology, and geography (Bates 1963;Bates and Blanchard 1970;Tate and Simpson 2003). Tate (2003), combining molecular, cytological and morphological data, transferred these three species of Tarasa and one species of Malacothamnus to the newly described genus Andeimalva. The Chilean species of Andeimalva is anomalous morphologically in that it differs from the group recognized by Krapovickas (1965) in having suborbicular, slightly 3-5-lobed leaves and awnless mericarps. Nonetheless, molecular and cytological data indicate that it is closely related to the other three species found in Bolivia and Peru (Tate and Simpson 2003).
Recent floristic exploration in the northern Andes of Peru yielded a Malvaceae collection (Vega Ocaña 419) of uncertain initial generic affiliation. With further study and molecular phylogenetic analyses, we determined it to represent a new species of Andeimalva, which is described here as A. peruviana Dorr & C.Romero, and the third one endemic to Peru. We are amazed that a shrub with such conspicuous flowers is not known from earlier collections and can only speculate that either its range is very restricted in a botanically poorly explored area or it might simply have passed unnoticed due to a limited flowering period.

Materials and methods
We sought molecular data both to confirm the generic identity of the new collection and to provide a phylogenetic context within the ITS (Internal Transcribed Spacer region of the nuclear ribosomal repeat) data set of Tate and Simpson (2003). Leaf tissue with abaxial trichomes scraped off was ground in a bead mill and extracted with the DNeasy Plant Mini Kit (Qiagen Inc., Valencia, California, USA) following an extended digestion with the AP1 buffer emended with 1.0 mg proteinase K (PCR grade solution, Roche Applied Science, Germany) and 1.8% v/v 2-mercaptoethanol. To reduce systemic contamination risk, all equipment was dedicated to "antique DNA" in a pre-PCR extraction laboratory and no other Malvales were co-extracted or amplified at the same time. Despite being a 2.5-year old collection with reasonably good preservation (i.e., retaining some color and alcohol treated), the resulting extraction only yielded highly degraded DNA (< 400 bp on an agarose test gel). Amplifications of ITS were attempted using a series of nested primers for decreasing amplicon sizes, and only a 302 bp fragment of ITS1 could be recovered with primers 1m (CGTAG-GTGAACCTGCGGAAGGATC, newly modified from ITS1) and P2 (GCCRAGA-TATCCGTTGCCGAG; Cheng et al. 2016). Amplification reactions were conducted in a final volume of 15 µl with 1× Bioline reaction buffer, 8 mM dNTPs, 1.75 mM magnesium chloride, 2 nM of each primer, 0.1 mg/ml of bovine serum albumin, 0.025 U/µl of Biolase DNA polymerase (Bioline USA, Taunton, Massachusetts, U.S.A.), 1 M betaine, and 1 µl of DNA extraction. Sequencing was with BigDye Terminator v3.1 chemistry (Thermo Fisher Scientific, Waltham, Massachusetts) on an ABI 3730xl DNA Analyzer (Thermo Fisher Scientific). The new ITS1 sequence (GenBank MK044847, from Vega Ocaña 419) was manually aligned using a similarity criterion with the Tate and Simpson (2003) data set archived in TreeBASE (http://purl.org/ phylo/treebase/phylows/matrix/TB2:M516). Small changes were made to the source matrix including converting mixed bases to ambiguity codes and improving the alignment within Fuertesimalva Fryxell (i.e., inconsistencies relative to positions 114-155 in our alignment), which reduced local branch lengths relative to prior analyses. The 53-tip matrix has an aligned length of 696 bp and relatively few, mostly small 1-4 bp indels (except Fuertesimalva spp. sharing a 42 bp deletion); we did not exclude any data in the analyses. We could not obtain any new plastid data and thus did not analyze the plastid data set of Tate and Simpson (2003), which the authors did not combine with ITS due to incongruence. Parsimony (MP) analyses were conducted with PAUP* 4.0a build 163 (Swofford 2002) using 1000 random addition replicates with TBR branch swapping and no search limits, and 1000 bootstrap (MP-BS) replicates, each with two random additions and a search limit of 10,000 trees per iteration (MulTrees = 10,000). Maximum likelihood (ML) analyses were conducted under a GTR + I + Γ model with GARLI 2.01 (Zwickl 2006) on CIPRES XSEDE (https://www.phylo.org/), using 100 replicates for optimal trees, and 1000 ML bootstrap replicates (ML-BS), each with two random additions and the automated stopping criterion.

Data resources
The data underpinning the analyses reported in this paper are deposited in the Dryad Digital Repository at: https://doi.org/10.5061/dryad.44dm150
Distribution and ecology. At present, known only from the type locality where it occurs in alpine bogs at 3750 m.
On the label of the type specimen there is a slight discrepancy regarding the political subdivision for the locality. Google Earth Pro mapping (8 June 2011 imagery, https:// www.google.com/earth/) places the coordinates cited on the label (presumably derived from a GPS device and for a logical locality near a road crossing of a wet drainage) in San Martín, and just over the eastern border delineated for La Libertad. However, we cannot verify how well Google Earth Pro finely draws political boundaries in relation to those that might appear on official government maps of such remote regions.
Etymology. The specific epithet is derived from the name of the country (Peru) where the new species is found.
Preliminary conservation status. Following the criteria and categories of IUCN (2017), Andeimalva peruviana is given a preliminary status of Vulnerable (VU D2) due to population very small or restricted (area of occupancy < 20km 2 and number of locations < 5). While the type of locality is quite remote and unexplored habitat occurs nearby, the general area has no protected regions and might be subject to habitat degradation from grazing or changes in hydrology. The latter includes human modification to its drainage system or climate change affecting montane precipitation.

Discussion
Morphologically Andeimalva peruviana appears to be closer to the three anomalous species of Tarasa first united by Krapovickas (1965) and later transferred to Andeimalva by Tate (2003), which share leaf morphology and mericarps with apical awns. Andeimalva chilensis, sister to the crown group (Fig. 1), has greater morphological divergence. However, its lack of awns relative to the other species of Andeimalva (uncertain in A. peruviana) may be an autapomorphic loss given that related genera such as Tarasa mostly possess awns (see Tate 2003). The chromosome number of our new species is unknown, but it would prove informative for generic affiliation. Optimal trees place A. peruviana as sister to A. mandonii, which has ML but not parsimony bootstrap support. This affinity is reflected in the morphology with shared generally larger floral features relative to the other species, but notably differs in stipule morphology (see Key below and Diagnosis). The large stipules of A. peruviana with prominent parallel major veins differ from those of the other species that are narrow in width and attachment.
Most species of Andeimalva are geographically well separated. However, A. spiciformis (Krapov.) J.A. Tate and A. peruviana are nearly sympatric on a gross scale with both occurring at high elevation in La Libertad, Peru. Nonetheless, the two species appear to be separated by altitude, hydrology, and a physical barrier. Andeimalva spiciformis with its scattered distribution in northwestern Peru at 2400-3200 m, is known in La Libertad from a single collection (A. López M. 1439, US;see Tate 2003: 15) made at 2600 m. In contrast, the type locality of A. peruviana, about 70 km distant near the border of La Libertad, was made at 3750 m and is physically well separated by an east-west divide created by the deep valley of the Rio Marañon. The boggy habitat of A. peruviana in a region rich in small alpine lakes near the wet eastern side of the Andes generally appears much wetter than where A. spiciformis grows on drier mountain areas to the west where the Andean rainfall shadow has greater influence.