Morphological characteristics and genetic evidence reveals a new species of Manihot (Euphorbiaceae, Crotonoideae) from Goiás, Brazil

Abstract During botanical expeditions between 2010 and 2015, as part of a taxonomic study of Manihot in the Midwest region of Brazil, approximately 500 specimens of the genus were collected. Some of these specimens presented similarities to Manihot irwinii. However, after careful morphological analyses, associated with genetic evidence, we propose here Manihot pulchrifolius as a new species. The new species is described, illustrated, and compared to Manihot irwinii, its most similar species. Furthermore, geographic distribution, conservation status, and period of flowering and fruiting of the novel species are also provided.


Introduction
Manihot Mill. encompasses over 100 Neotropical species, and therefore stands out as one of the largest genera of Euphorbiaceae in Brazil, with ca. 80 species (Silva 2014). In the Cerrado Biome, over 50 species of Manihot have already been documented, among which 40 are endemic . Nevertheless, recent studies of the genus in the Chapada dos Veadeiros region, in the state of Goiás, Brazil, revealed some new species , Silva 2014, Silva and Sodré 2014, Silva 2015a, b, Silva et al. 2016a, which demonstrates that there is still a lot to discover about this genus, whose taxonomy remains relatively poorly known in the Cerrado Biome. Some species of Manihot endemic to the Cerrado Biome have leaves that are considerably diverse morphologically, a fact that has aroused the interest of botanists and geneticists (Duputié et al. 2011, Silva et al. 2016b During botanical expeditions to the Serra Dourada State Park, in the state of Goiás, Brazil, since October 2010, as part of a floristic survey of Euphorbiaceae, approximately 500 specimens of Manihot were collected, some of them showing similarities to M. irwinii D.J. Rogers & Appan regarding habit and foliage type. After careful morphological analyses of these collections, associated with genetic evidence, we propose herein the new species Manihot pulchrifolius. A detailed description, comments on flowering, fruiting, distribution, environmental preferences, conservation status, and comparisons with morphologically similar species are provided.

Morphological Studies
The description of the new taxa was based on observations of populations in the field since 2010, analyses of available specimens from herbaria UFG, NY, K, RB, and UB (acronyms follow Thiers continuously updated), and review of the literature (Rogers andAppan 1973, Allem 1989). The terminology used to describe the types of inflorescences and leaves follows Rogers and Appan (1973). The illustrations were based on fresh material fixed in alcohol 70% during collection in the field. Holotypes of the new species are deposited at UFG, and isotypes are going to be sent to NY, K, RB, and UB. Photographs of natural populations were taken in the field. The conservation status of the species follows IUCN (2016).

Data analysis
The genetic diversity of the populations studied was assessed based on estimates of the average number of alleles per locus (A), rarefied allelic richness (AR), observed heterozygosity (Ho), expected heterozygosity under Hardy-Weinberg equilibrium (He), and intrapopulation fixation index (f ). The genetic structure of the populations was evaluated according to Weir and Cockerham (1984). These analyses were conducted using the package Hierfstat for the statistical software R (Goudet 2005). The genetic structure was assessed by a Bayesian approach, conducted using the software STRUCTURE 2.3.4 (Pritchard et al. 2000), assuming a model that allows mixing alleles between populations for four independent runs, with K values ranging from one to ten. The tests were performed using the Markov Chain Monte Carlo (MCMC) method, with periods of burn-in of 10,000 and 1,000,000 replicas. The average of likelihood values for each K for all runs was determined by the statistical ΔK developed by Evanno et al. (2005).
The pattern of differentiation among populations was evaluated by calculating the genetic distance between pairs of populations, based on estimates of pairwise by fixation index (F ST ). To visualize the pattern of differentiation among populations, the genetic distance matrix was subjected to a cluster analysis using the unweighted pair- group method with arithmetic averages (UPGMA). To assess the degree of representativeness of the dendrogram, the cophenetic correlation coefficient was estimated with 10,000 permutations. These analyses were conducted using R Hierfstat (Goudet 2005) and adegenet (Jombart 2008 Diagnosis. Shrubs up to 2.5 m tall, erect, glabrous; young branches and young leaves reddish to purplish, green-vinaceous to violet; adult leaves 5-lobed at the plant base, 3-lobed along the stem, or rarely unlobed near inflorescence; long racemes or panicles (up to 27 cm long), erect to pendent, axes reddish to purplish; calyx of staminate flowers reddish or purplish with yellow margins, filaments pubescent; bracts and bracteoles of flowers of both sexes reddish to purplish; fruits dark green with violet to purplish wings.
Distribution and Ecology. Manihot pulchrifolius is endemic to the state of Goiás, where it was found growing in Serra Dourada (Figure 3), one of the most beautiful and preserved mountainous areas in the state. This mountain range encompasses the Serra Dourada State Park, an area of over 30,000 hectares protected by law since 1965. The species grows in Cerrado sensu stricto, on rocky outcrops, rocky slopes, and Cerrado rupestre, in clayey, clayey-stony, and sandy soils, or even on rocky crevices, between 900 m and 1,000 m.
Phenology. The species has been collected with flowers and fruits from November to July. However, the flowers are more usual from January to March, whereas the fruits are more abundant from April to July.
Etymology. The specific epithet "pulchrifolius" alludes to the beautiful foliage of the species, especially in the leaf flushing stage, when the leaves are reddish or purplish to green-vinaceous to violet.
Conservation status. Given that the populations have more than 50 individuals and the vegetation where they grow is commonly found in the central part of the state of Goiás (in the municipality of Goiás and neighboring municipalities), we consider M. pulchrifolius as Least Concern (LC) according to IUCN (2016).

Remarks
Manihot pulchrifolius was identified by Rogers and Appan (1973) Rogers and Appan (1973), probably because they were not aware that M. irwinii is a little known species, scarcely represented in Brazilian herbaria (UB, CEN, HPB, and UFG), and with distribution restricted to the Serra dos Pireneus and neighboring areas where it grows in open areas of Cerrado sensu stricto in clayey soils. However, in the last two years, morphological and genetic studies developed by the authors of this paper have shown that populations from Serra dos Pireneus and the Serra Dourada State Park previously identified as M. irwinii present differences regarding leaf, inflorescence, and flower morphology, as well as genetic structure as described below. Therefore, we concluded that the specimens from the Serra Dourada State Park belong to a new species, herein named M. pulchrifolius. Both species share a shrubby habit, leaves with lobes overlapping basally, midrib veins thickened, secondary veins subparallel, flowers of both sexes with sepals pubescent internally, and winged fruits.
Systematically, the new species can be situated in Manihot section Quinquelobae Pax according to Silva (2014), by having a shrubby habit, leaves widely spaced along the branches, basal petiole attachment, deeply lobed leaf blade, lobes of various shapes (but not linear), monoecious racemose or panicled inflorescences, foliaceous or setaceous bracts and bracteoles, winged or wingless fruits. However, since M. quinquelobae belongs to a polyphyletic group (Silva et al. unpublished), we prefer not to ascribe M. pulchrifolius to any sections of the genus.

Genetic studies
A total of 50 alleles were found for the seven loci evaluated in populations of M. irwinii sensu lato, ranging from three to eleven alleles per locus. The populations studied showed high and similar genetic diversity, which was also observed in other species of the genus (Roa et al. 2000). Only the population of the municipality of Corumbá de Goiás presented genetic diversity values significantly smaller than the others, suggesting that the population may be suffering from a fragmentation of its habitat since it grows in an area surrounded by agriculture and disturbed by anthropic actions related to tourism. The inbreeding values estimated in the four populations were not significant, indicating adherence to the frequencies expected by Hardy-Weinberg equilibrium for the evaluated loci ( Table 2).
The population genetic structure analysis showed an estimated value of θ of 0.363, indicating that 36.3% of the genetic diversity is in the component between populations. This level of genetic structure is considered very strong (Wright 1978), especially taking into account the geographical distance among the assessed populations ( Figure 3) and    the fact that M. irwinii sensu lato is probably allogamic (Loveless and Hamrick 1984).
The estimated values of F (0.359) and f (-0.006, not significant) show that the observed genetic structure is related to the effect of genetic drift and low gene flow between populations and not to the reproductive system of the plant. A strong pattern of genetic structure was also observed in the analysis using the Bayesian approach, which was determined in the formation of two genetic groups (K = 2, Figure 4). The first group determined by STRUCTURE is restricted to the population found in Serra Dourada, municipality of Mossâmedes, whereas the second group contains the other three populations studied ( Figure 5). The pattern of attribution of the individuals to the groups was very similar to the patterns observed for different species of the genus Quercus that present lack of gene flow between groups (Valencia-Cuevas et al. 2015). This pattern of genetic differentiation between populations can be observed in the dendrogram constructed from the F ST pairwise matrix, which clearly points out the separation from the population collected in Serra Dourada in a distinct group ( Figure  6). This result, associated with the strong genetic structure detected, suggests that populations of M. irwinii sensu lato are quite distinct from each other, particularly from M. pulchrifolius (population from Serra Dourada), which seems to be genetically isolated.