Taxonomic investigations on Malva cretica s.l. (Malvaceae) ; Investigaciones taxonómicas en Malva cretica s.l. (Malvaceae)

A morphometric analysis has been carried out for the first time in order to investigate the variability of the infraspecific taxa included in the steno-Mediterranean Malva cretica s.l., i.e. M. cretica subsp. althaeoides, M. cretica subsp. cretica, and the poorly-known M. cretica var. montana, described from southern Italy. According to our results, M. cretica subsp. althaeoides and subsp. cretica are worth of taxonomic recognition at subspecies rank and occupy respectively the western and the central-eastern sectors of the species range. Concerning M. cretica var. montana, characterized by some intermediate features, it should be included in M. cretica subsp. cretica. Resumen. Se ha realizado, por primera vez, un análisis morfométrico con el fin de investigar la variación morfológica de los táxones infraespecíficos actualmente reconocidos en Malva cretica s.l., i.e. subsp. cretica y subsp. althaeoides, incluyendo la poco conocida var. montana del sur de Italia. Según nuestros resultados, la subsp. althaeoides y la subsp. cretica merecen el rango de subespecie y ocupan respectivamente los sectores occidentales y centro-orientales del área de distribución de la especie. Respecto a M. cretica var. montana, caracterizada por algunos rasgos intermedios, debe ser incluida en M. cretica subsp. cretica.


INTRODUCTION
Molecular studies by Ray (1995) and Escobar García & al. (2009) showed that the traditional separation of Malva L. and Lavatera L., based mainly on the degree of fusion of the epicalyx bracts, is artificial and cannot be maintained, while the taxonomic significance of the fruit morphology was emphasized. The overall evolution within the complex appears to be reticulate and the genus Malva in a broad sense (including Lavatera) is presumed to be monophyletic (Tate & al. 2005). Waiting for a comprehensive taxonomic study of Malveae J.Presl, several authors accept Malva in a broad sense (e.g., Ray 1998;Molero & Montserrat 2005, 2006Banfi & al. 2005;Iamonico 2010Iamonico , 2018Valdés 2011;Conti & Bartolucci 2012). However, some taxa (the "Lavateroid clade" sensu Ray, 1995) are still doubtfully placed and other authors (e.g., Bayer & Kubitzki 2003;Molero & Montserrat 2007) keep them in the genus Lavatera.
According to molecular data and morphological features of the fruit, Malva cretica Cav. is certainly referable to the "Malvoid clade" (sensu Ray 1995). Although it is well distinct as a species, its taxonomic placement in the genus Malva is controversial (see below). Moreover, the infraspecific variability of this species is remarkable and at least two geographical subspecies are currently recognized: M. cretica subp. cretica [Tunisia, Spain, France (Corse), Italy (including Sicily and Sardinia), Greece, East Egean islands, Crete, Cyprus and Turkey; naturalized in France] and M. cretica subsp. althaeoides (Cav.) Dalby [Spain, France (Corse) and doubtfully in Italy (including Sicily) and Malta; doubtfully native in Cyprus] (Valdés 2011). These two taxa have been treated differently over time, even at generic ranks. Candolle (1824) included both the taxa in the sect. Malvastrum DC., but placing M. cretica into the ser. Cymbalariae DC. and M. althaeoides into the ser. Bismalvae (Medik.) DC. Alefeld (1862: 258) included M. althaeoides in the genus Axolopha (DC.) Alef., while Krebs (1994) placed both taxa in the genus Dinacrusa G. Krebs. Nogueira & Paiva (2005)  and infra-specific ranks, Paoletti (1901) reported M. altheoides as a variety of M. cretica, but Dalby (1967) and Valdés (2011) accepted the subspecies rank.
The two taxa have been often confused and their current distribution should be verified (see e.g., Serra Laiga 2005). Moreover, the rarely cited Malva cretica var. montana Lacaita, described from southern Italy (Lacaita 1925), has also to be considered. Gavioli (1929), in a comparative study between the floras of Spain and Basilicata region (southern Italy) recognized M. cretica var. althaeoides, M. cretica var. cretica, and M. cretica var. montana. He regarded var. montana as endemic to southern Italy. Finally, according to Dalby (1968), intermediate plants between subsp. cretica and althaeoides would occur in southern Italy and Malta, while he excluded subsp. cretica from Spain. Later, Krebs (1994) indicated the occurrence of both Dinacrusa cretica (Cav.) G.Krebs subsp. cretica and D. cretica subsp. althaeoides (Cav.) G.Krebs in southern Italy. Interestingly, Gavioli (1929) identified as M. cretica var. althaeoides several specimens collected by himself in southern Italy, i.e. out of the currently accepted range of the taxon.
With the aim to clarify the taxonomy of the various taxa described under Malva cretica, here we present a biometric study as part of an ongoing research works on Malvaceae (see e.g. Iamonico 2010Iamonico , 2014Iamonico , 2016Iamonico & Peruzzi 2014) and the endemic flora of southern Italy (e.g., Vallariello & al. 2016;Santangelo & al. 2017;Erben & al. 2018). EDG, available at NAP) were examined for a preliminary screening.

Specimens of
Specimens from FI, NAP, VAL and Herb. EDG, preliminary re-identified following keys and descriptions in Lacaita (1925), Dalby (1968), Krebs (1994), and Nogueira & Paiva (2005), were employed for the analyses. A morphological analysis, based on 17 characters (14 quantitative and 3 qualitative, see Table 2) was performed. Nogueira & Paiva (2005) pointed out a presumed difference in corolla colour among the studied taxa. We decided not to include this character in the analyses, because it cannot be easily observed in sicco. However, according to Lacaita (1925), and our observations in the field (Italian material) and from photographs (whole range of variation), such character would seem not to be diagnostic. The data matrix so obtained was processed using software packages NCSS 2007 (NCSS LLC, Kaysville, UT) and SPSS v. 11.5 (SPSS Inc., Chicago, IL, USA). The variability of the characters has been examined by Discriminant Analysis (DA), bi-and univariate (box and scatter plots) analyses. Whenever correlation was  Basal leaves lenght (mm) 5 Basal leaves width (mm) 6 Cauline leaves lenght (mm) 7 Cauline leaves width (mm) 8 Ratio between characters 6 and 7 9 Peduncle length (mm) 10 Epicalyx segments length (mm) 11 Calyx segments length (mm) 12 Calyx segments width (mm) 13 Ratio between characters 11 and 12 14 Petal segments length (mm) 15 Petal segments width (mm) 16 Ratio between characters 14 and 15 17 Mericarps surface (glabrous or pubescent) > |0.7|, only one per group of correlate variables was employed in the further analyses; the selected character in these cases was that with the highest sum of loads (in absolute value) on all axes. Of the surviving characters, only those loading at least |0.7| on at least one axis were kept for DA. In the practice of excluding collinear variables and choosing the most explanatory variable for each group of correlates we followed previous literature (see  for further details), as when highly correlated variables are present, the discriminant function may not reliably predict grouping (Naes & Mevik 2001). A Discriminant Analysis (DA) was carried out, in order to clarify the relationships between the individuals, projecting Operational Taxonomic Units (OTUs) with population centroids in a two-dimensional space generated by canonical functions.

RESULTS AND DISCUSSION
The DA graph (Fig. 1) shows a rather compact ellipse, which apparently would suggest a Gaussian distribution. However, the OTUs on the right part of this cloud of points mainly represent specimens referable to ALTH and collected in Spain; while OTUs on the left can be mostly ascribed to CRET and to MONT, gathered in central and eastern Mediterranean. Therefore, a smooth West-East cline emerges. A percentage of 77.4% of cases resulted correctly classified in origin: 97.2% for CRET, 69.8 for ALTH, but only 50% for MONT. This is mainly due to specimens with some intermediate features typical of Malva cretica var. montana collected in Spain, but classified as M. cretica subsp. althaeoides on account of the absence of stellate hairs. In every case, the variability of MONT is mostly included in that of CRET, only showing sometimes a higher sepal/petal length ratio (Fig. 2).
The ALTH group includes plants with longer and wider petals and higher sepals/petals ratio [petals (12)19-34 mm long, ratio petals / sepals (1.3)1.5-2.5(2.9) in ALTH vs. petals (9-)11-20(-22) mm long, 0.8-1.4(-1.9) as long as the sepals in CRET + MONT; Fig. 3]. In addition, in the ALTH group, hairs are mostly simple, while the other group (CRET + MONT) corresponds to plants with mixed hairs, both simple and stellate (Fig. 4). On the basis of the results obtained [partial overlapping of the characters and geographic cline (Fig. 5) between western and centraleastern Mediterranean], two taxa can be recognized as partly distinct. The subspecific rank is therefore the most appropriate rank for them in our opinion, according to the treatment below: Malva cretica subsp. cretica, and M. cretica subsp. althaeoides. In addition, M. cretica var. montana should be regarded as a synonym of subsp. cretica. As said above, apparently intermediate populations between M. cretica subsp. cretica and M. cretica subsp. althaeoides occur in southern Italy (and we identified them with "M. cretica var. montana"), but possibly also in eastern Spain. Nevertheless, it is to be noted that some Spanish records of subsp. cretica could be due to introduction from eastern countries, as well as those known in France (Valdés 2011). Besides, no significant ecological difference can be highlighted from the information we have recorded. M. cretica subsp. cretica. These two taxa likely originated by an ongoing splitting of a single South-Mediterranean species into two groups of populations, after the rise of the sea level in the central part of its area (Fig. 5). Obviously, verifying our hypothesis would require a separate study on population dynamics on the whole. Finally, the inclusion of Malva cretica s.l. in the genus Dinacrusa (including annual taxa now located both in Malva and Althaea) by Krebs (1994) is a very interesting issue, but worth of further study with a larger sampling.   Krebs, 1994). Annual scapose herb, hairy. Stem 5-80 cm tall, up to 2.5 mm in diameter, erect or prostrate-ascending, hairy by pustulate simple hairs up to 2 mm and often with shorter stellate hairs. Basal leaves 5-40 × 5-41 mm, rotundate-lobate, crenate, with a peduncle up to 4 cm, hispid, pubescent or softly villous. Upper leaves similar to the basal ones but more divided, serrate, with 3-5 acute lobes, with a shorter peduncle. Stipels 5-6 mm long, linear-lanceolate. Flowers solitary, with peduncle longer than the axillary leaf. Epicalyx with 3 linear segments, 3-9 mm long, hispid to villous. Sepals 5, linear-lanceolate, up to 8-19 × 1-4 mm in fruit, hispid to villous. Petals 5, with a glabrous limb from obtuse to slightly retuse, 9-34 × 4-21 mm, and a ciliate awn. Mericarps glabrous and smooth or slightly rugose, 1.5-2.5 mm in diameter.  Figure 1: MA 73338!).