Molecular phylogeny and evolutionary history of Moricandia DC (Brassicaceae)

Introduction

The Brassiceae tribe of the Brassicaceae family includes many economically important species that are useful as vegetables, edible oils, crop forages, condiments and fuel crops (Zelmer & McVetty, 2009). For this reason, this crucifer tribe has been the focus of a vast amount of genetic, agronomic, and ecological research (Gómez-Campo, 1999; Gupta, 2009; Kole, 2011; Schmidt & Bancroft, 2011). Despite this interest, the phylogenetic relationship between the Brassiceae species is not yet resolved, and consequently, several well-established Brassiceae genera, such as Brassica or Diplotaxis, are probably polyphyletic and artificial. One small but economically important genus in this tribe is Moricandia DC. This genus probably originated in the Mediterranean Basin (Pratap & Gupta, 2009) and is currently distributed in North Africa, Middle East, Central Asia and Southern Europe, where it is associated mostly with arid and semi-arid environments (Tahir & Watts, 2011). Some Moricandia species have being extensively studied because they show intermediate C₃–C₄ photosynthetic metabolism (Apel, Horstmann & Pfeffer, 1997; McVetty, Austin & Morgan, 1989), a feature that improves carbon assimilation and water use efficiency under drought conditions (Apel, Bauve & Ohle, 1984; McVetty, Austin & Morgan, 1989). This so-called Moricandia syndrome may have great agronomic interest since it could be transferred to Brassica species by hybridization, increasing crop yield under extreme climatic conditions and in marginal areas (Apel, Bauve & Ohle, 1984). Disentangling the phylogenetic relationship between Moricandia species is key to understanding how these traits have evolved and to determine the placement of this genus inside the Brassiceae tribe.

Moricandia individuals mainly show erect and branched stems with simple, exstipulate leaves, usually with entire or pinnated lobes (Gupta, 2009). Their flowers are actinomorphic-disymmetric and mostly of lilac color, although range from almost white to deep purple depending on the species and weather conditions. Their fruits are dehiscent two-valves siliques with one or two seed series per valve (Gupta, 2009). Moricandia shows a high variability in the morphological characters used for identification, making the taxonomy of this genus complex and controversial (Jiménez & Sánchez Gómez, 2012). Eight species are currently recognised in the genus Moricandia (Warwick & Sauder, 2005; Warwick, Francis & Al-Shehbaz, 2006; Tahir & Watts, 2011): (1) M. arvensis (L.) DC, (2) M. moricandioides (Boiss.) Heywood, (3) M. foetida Bourg. ex Coss., (4) M. suffruticosa (Desf.) Coss. & Durieu, (5) M. spinosa Pomel, (6) M. foleyi Batt., (7) M. sinaica Boiss., and (8) M. nitens (Viv.) Durieu & Barr (Sobrino Vesperinas, 1984; Warwick, Francis & Al-Shehbaz, 2006; Tahir & Watts, 2011). Moricandia arvensis is an annual to perennial herb widely distributed in the northwest Africa, Iberian Peninsula and southern Italy, from where it has even invaded other parts of the planet (De Bolós, 1946; Gómez-Campo, 1999). It is mostly a ruderal species associated to cultivated areas, roadsides and other human disturbed habitats. Moricandia moricandioides and M. foetida are two herbaceous species endemic to the Iberian Peninsula. The former is distributed in semi-arid environments of the eastern Spain whereas the latter is a narrow endemism inhabiting arid habitats of the southeast Spain (Sobrino Vesperinas, 1993). Moricandia suffruticosa and M. spinosa are suffruticose species inhabiting Morocco, Algeria, Tunisia and probably Libya, whereas M. foleyi is an annual herb showing a very narrow distribution in desert areas of southern Morocco and Algeria. Moricandia sinaica is a herbaceous species located in desert areas from the Near East to Pakistan; and M. nitens is a suffruticose species distributed from North Africa to Middle East. In addition, several subspecies and varieties have been named, in particular for the widely distributed M. arvensis (Schulz, 1936; Heywood, 1964; Maire, 1967).

The phylogenetic relationships among the Moricandia species, and of these with the rest of the Brassiceae species, have not been well established yet. Moricandia was previously included into the subtribe Moricandiinae (Schulz, 1923; Schulz, 1936; Gómez-Campo, 1980), although molecular evidences have shown little support for this subtribe (Warwick & Black, 1994; Warwick & Sauder, 2005). Most family-wide phylogenies have included some Moricandia species (e.g., Warwick & Black, 1997; Warwick & Sauder, 2005; Bailey et al., 2006; Jiménez & Sánchez Gómez, 2012; Schlüter et al., 2016), and these large-scale studies have suggested that Moricandia belongs to the Rapa/Oleracea subtribe together with Brassica, Diplotaxis, Enartrocarpus, Eruca, Erucastrum, Morisia, Raphanus, Rapistrum and Rytidocarpus (Warwick & Hall, 2009). Although Moricandia has been considered a Brassica coenospecies on the basis of cytogenetic and morphological similarities (Gómez-Campo, 1999), molecular phylogenetic analyses suggest that Moricandia is closely related to the genus Rytidocarpus and less clearly to Eruca (Warwick & Black, 1994; Warwick & Sauder, 2005; Bailey et al., 2006; Couvreur et al., 2010). The monotypic genus Rytidocarpus has long been recognized as very close to Moricandia because it presents Moricandia-like cotyledons, similar purple flowers and the same chromosome (2n = 28) complement (Gómez-Campo, 1980; Prakash et al., 2009). Thereby, due to this incomplete and sometimes contradictory evidence, the phylogenetic position and evolutionary history of Moricandia is still unresolved, despite the importance this information could have to understand the evolution of agronomic traits.

The main goal of this study is to disentangle the phylogenetic relationships among the Moricandia species, using a nuclear and two plastidial regions. In addition, we have dated the main events in the evolution of this genus and determined the phylogenetic relationship of Moricandia with its closely related genera Rytidocarpus and Eruca. We show that one Moricandia species should be excluded of this genus and demonstrate that a population previously ascribed to Rytidocarpus moricandioides is indeed a new Moricandia species.

Materials and Methods

Results

Ribosomic 5.8S, and 18S and 28S flanking sequences did not show variation and were not included in the final alignment of the 24-samples set and the two outgroups, that spanned 4,055 bp. ITS1 extended over 274 bp and included 63 variable sites with 51 parsimonious informative positions. ITS2 spanned 194 bp and included 29 variable sites with 22 parsimonious informative positions. The amplified ndhF region spanned 2,004 bp with 87 variable sites and 58 parsimonious informative positions. No indels or stop codons were found on this region, which codifies for part of the NADH dehydrogenase subunit F. The trnT-trnF amplified region spanned 1,583 bp with 264 variable sites and 186 parsimonious informative positions.

Plastidial and nuclear phylogenetic trees

Figure 2 shows the Bayesian inference tree obtained with the combined cpDNA regions (ndhF and trnT-trnF) confronted with the tree obtained with the nuclear ITS regions using the same inference approach. The general pattern was maintained except for a lower resolution in the nuclear tree, where the Moricandia lineage appeared not so well resolved showing a basal trichotomy, and for minor rearrangements. For instance, M. spinosa and M. nitens appeared as sibling species in the plastidial tree but they were less closely related in the nuclear tree, although in both cases they belonged to the same African clade. It is also noticeable that M. foleyi samples appeared as monophyletic in the nuclear tree but interspersed with the Eruca samples in the plastidial tree. These trees were also obtained by ML inference and showed the same topologies (Fig. S2).

Chronogram

There were no significant differences between the trees obtained with or without assuming a clock (P = 0.92; LnL = −7851.663 enforcing a clock; LnL = −7835.925 without assuming a clock). Therefore, the null hypothesis of equal evolutionary rates (i.e., molecular clock) was not rejected and these sequences can be assumed to follow a global molecular clock. With this assumption, we inferred chronograms using a constant rate of 1.2 × 10⁹ substitution/site/year for the non-codifying plastidial DNA. The dating of the monophyletic Iberian clade was compatible with an origin during the end of the Messinian period (5.99–5.33 Ma). Other trees obtained with substitution rates reported in the literature (1.2–1.7  × 10⁹ s/s/y) were also compatible with this temporal window. Following, we produced a new chronogram constraining the origin of the Iberian clade to the Messinian salinity crisis period and including the ITS sequence of M. sinaica obtained from GenBank. This tree (Fig. 4) was congruent with that obtained with MrBayes (Fig. 1). The only topological differences were that the two samples of M. foleyi appeared to form a clearer lineage within the genus Eruca, and the position of M. sinaica, which was not included in the 24-samples set. In this chronogram the genus Moricandia separated 6.81 M years ago [5.83–7.89 Ma] from Rytidocarpus.

Discussion

The phylogenetic analyses presented here represent the evolution of the Moricandia genus clearly. Not only the Bayesian inference tree was concordant with the ML and timing trees, but also the cpDNA and ITS trees showed a fair agreement, with only minor discrepancies. In addition, all our samples were congruent with the GenBank accessions that were included in the ITS inclusive tree, although the latter tree showed lower resolution. In the trees of highest confidence, Moricandia appeared to be formed by two main lineages: the Iberian clade containing the Iberian species M. moricandioides, M. foetida and the new M. rytidocarpoides; and the African clade containing the species inhabiting the Southern Mediterranean region (M. sinaica, M. suffruticosa, M. nitens, M. spinosa and M. arvensis). The Bayesian inference trees showed high branch supports for this branching pattern and a good agreement with a molecular clock-like evolution. The timing of 7.19 Ma [5.18–9.23] for the separation of B. rapa and R. sativus, the two outgroup taxa in the chronogram, is congruent with the data previously reported (∼5.17 ± 2.5 Ma; Hohmann et al., 2015). This congruence is noteworthy since we used a strict clock that imposed a similar evolutionary rate in all lineages, which might not be realistic for the more distant (outgroup) taxa (Drummond et al., 2006). Therefore, we think that these phylogenetic trees are congruent and robustly represent the evolution of these taxa.

The genus Moricandia probably has a North African origin. Seven out of the previously eight recognized Moricandia species form a monophyletic group originated 6.81 Ma [5.83–7.89 Ma] after the separation from the Rytidocarpus lineage, their sister genera. Since Rytidocarpus is a North African endemism (Maire, 1967) and the genus Moricandia show a high number of species inhabiting North Africa (Maire, 1967), we support that Moricandia originated in North Africa, as was previously suggested (Sobrino Vesperinas, 1978). The colonization of the Iberian Peninsula probably occurred after a range expansion during the Messinian period, between 7.25 and 5.33 Ma (Fig. 5), coinciding with extensive African-Iberian floral and faunal interchanges (e.g., Fernández-Mazuecos & Vargas, 2011; Gibert et al., 2013). In that period, land bridges between North Africa and South Iberia appeared due to tectonic uplift coinciding with the isolation of the Mediterranean Sea from the Atlantic Ocean circa 6.3 Ma (Martín, Braga & Betzler, 2001). After the new aperture of the Strait of Gibraltar during the Zanclean period, the Iberian and African clades were separated and began to diverge (Figs. 4 and 5). A similar biogeographical pattern has been found for other plants. For instance, the genus Antirrhinum produced several lineages separated by the Straight of Gibraltar (Vargas et al., 2009), as well as the genus Hedera and the Saxifraga globulifera–reuteriana complex (Vargas et al., 1999; Vargas, Morton & Jury, 1999). However, this is not a general biogeographical pattern since other species have maintained genetic connections along the two sides of the Straight of Gibraltar (Rodríguez-Sánchez et al., 2008).

Moricandia sinaica is the most basal taxon of the African clade, arising circa 5.22 Ma (Fig. 4), although, given the moderate support of this branching event, its phylogenetic position should be accepted with cautiousness. In the inclusive ITS tree, which included the GenBank-ITS set plus the 24-samples set, M. sinaica (inhabiting Egypt and West Asia) and M. arvensis var. garamantum (from South Algeria) appeared as close relatives. This fact suggests that these two taxa could be the same species, as it has been previously postulated (Sobrino Vesperinas, 1984). The rest of the species forming the African clade (M. suffruticosa, M. nitens, M. spinosa and M. arvensis) can easily hybridize (Sobrino Vesperinas, 1997) and have been included in the same cytodeme (Prakash et al., 2009). Sobrino Vesperinas (1984) postulated that the now widely distributed M. arvensis is the same as the M. arvensis var. robusta from the Constantine area in Algeria (Gómez-Campo, 1999). Unfortunately, the ITS sequences included here did not show enough resolution to support or deny this claim. In the inclusive ITS tree (Fig. 3), M. arvensis var. robusta appeared in the same clade as the other M. arvensis samples, but also forming a polytomy with M. suffruticosa and M. spinosa. These two latter species show caryotypes with higher ploidy than the typical 2n = 28 (Sobrino Vesperinas, 1984) of this genus (namely, M. suffruticosa with 2n = 56 and M. spinosa with 2n = 84; Harberd, 1976), but in the same ploidy series than the other Moricandia species.

The caryotype of M. spinosa suggests that this species could be a stabilized hybrid. We have found that M. spinosa shows a cpDNA closely related to those of M. nitens but for the nuclear ITS markers shows a more distant genetic relationship (Fig. 2). For these last markers, M. spinosa is more similar to M. suffruticosa (Fig. 2). These facts suggest that M. spinosa (2n = 84) is an amphidiploid species produced by whole genome duplication after a hybridization event between M. suffruticosa (2n = 56) and M. nitens (2n = 28). These kind of events have been pervasive in the evolution of this tribe and family (Marhold & Lihová, 2006; Franzke et al., 2011). However, more genetic analyses are necessary to confirm this hypothesis.

After the Zanclean aperture of the Strait of Gibraltar, the Iberian Moricandia clade diverged into three well-supported lineages that can be now identified as three genuine species (Figs. 4 and 5). Of these species, the endemic M. foetida was the first to diverge and is now completely reproductively isolated from M. moricandioides. In fact, Sobrino Vesperinas (1993, 1997) experimentally demonstrated that they could not hybridize. The second lineage includes all the M. moricandioides samples in a monophyletic cluster. Five subspecies have been identified for this species based on morphological characters (Sobrino Vesperinas, 1993; Sánchez Gómez et al., 2001), with subspecies pseudofoetida showing intermediate characteristics with M. foetida (Sánchez Gómez et al., 2001). Jiménez & Sánchez Gómez (2012) based on ISSR markers proposed that M. moricandia pseudofoetida arose by reproductive isolation rather than hybridization between M. foetida and M. moricandioides moricandioides. Our current results do not refute that hypothesis and show that this subspecies, that inhabit similar environments that M. foetida, is clearly a M. moricandioides subspecies without showing signal of past hybridization.

The third species in the Iberian clade are plants growing in marls and calciferous substrates in a few places of the Guadiana Menor River Basin. Based mostly on fruit morphology, these plants were originally ascribed to the Moroccan-endemism Rytidocarpus moricandioides by Morillas-Sánchez & Fernández López (1995), and later accepted in the Vascular Flora of Eastern Andalusia (Blanca et al., 2011). These plants differ from other Moricandia in some peculiar morphological characters. Namely, they present sepals with scarious margins that persist during fruit development and fruits with two segments, the upper with beak shape, being similar to Rytidocarpus, whereas the other Moricandia species present non-persistent sepals without scarious margins and fruit in a siliqua. However, despite these morphological differences, our phylogenetic study clearly shows that they belong to the genus Moricandia (Figs. 1–4), inside the Iberian clade as a sister species to M. moricandioides. In addition, an exhaustive morphological comparison between the Jaén and the Morocco specimens clearly separated the Spanish samples from the genuine, Moroccan, Rytidocarpus moricandioides (see Tables S2, S3 and Figs. S3 S4). Therefore, these plants should be ascribed to the genus Moricandia and they deserve the taxonomic rank of species due to both their phylogenetic position and their distinctive morphological traits. We proposed the name Moricandia rytidocarpoides to denominate this new species Moricandia rytidocarpoides Lorite, Perfectti, Gómez, González-Megías & Abdelaziz sp.nov. urn:lsid:ipni.org:names:77166015-1 (see Appendix for a formal taxonomical description). Consequently, the siliqua fruits and the dehiscent scarious sepals are no longer defining characteristics (i.e., diagnostic traits) of the genus Moricandia. Several molecular analyses have demonstrated that fruit morphology shows homoplasy in Brassicaceae (Appel & Al-Shehbaz, 2003; Koch, Al-Shehbaz & Mummenhoff, 2003), and, therefore, fruit traits are not good indicators of phylogenetic relationships despite they have been widely used as taxonomic diagnostic traits (Hall et al., 2006).

The taxonomic status of Moricandia foleyi should be also amended. In all the phylogenetic analyses performed here M. foleyi appeared outside the Moricandia lineage and within the lineage of the Eruca species. Eruca is a genus of annual, non-hetero-arthrocarpic plants with a controversial taxonomy. Depending on the taxonomical treatment, Eruca includes from one to four species. Gómez-Campo (1999) considered this genus as monotypic, after E. loncholoma was ascribed to Brassica subgen. Brassicaria (Gómez-Campo, 1999) and E. setulosa was moved to the proposed genus Guenthera (Gómez-Campo, 2003). Pratap & Gupta (2009) also supported this view, meanwhile Warwick, Francis & Gugel (2009) recognized four taxa: E. loncholoma, E. pinnatifida, E. setulosa and E. vesicaria. However, after E. pinnatifida was classified as a subspecies of E. vesicaria, http://www.theplantlist.org/ accepted only three species in this genus: E. loncholoma, E. setulosa and E. vesicaria, with this last species presenting three subspecies. Of these subspecies, two (E. vesicaria vesicaria and E. vesicaria pinnatifida) are circumscribed to the West Mediterranean region, whereas the subspecies sativa shows a circum-mediterranean distribution (Pignone & Gómez-Campo, 2011), although it is currently cultivated in many other areas of the world (Gómez-Campo & Prakash, 1999). Our phylogenetic analyses support the inclusion of M. foleyi in the genus Eruca (Figs. 1–4). In addition, other evidences separate M. foleyi from the genus Moricandia. This species has been reported to be 2n = 14 (Sobrino Vesperinas, 1997), clearly different from the caryotipic values reported for the rest of the Moricandia species (2n = 28, 56, 84; Sobrino Vesperinas, 1984), but also a different value when compared to the Eruca species (2n = 22; Harberd, 1976). Recently, in an ITS phylogeny, Schlüter et al. (2016) separated M. foleyi of the Moricandia lineage, although unfortunately they did not include in their analyses any Eruca samples. Anecdotally, Maire (1967) described M. foleyi as a glabrous and robust green herb with a strong smell of E. vesicaria (“à forte odeur d’Eruca vesicaria”). In our nuclear-sequences trees (24-samples and the GenBank-ITS sets) M. foleyi consistently appeared in the Eruca clade. The same pattern appeared in the cpDNA tree (see Fig. 2 and Fig. S2), indicating that both nuclear and cytoplasmic genes ascribe M. foleyi to the genus Eruca and excluding a recent hybridization event. Therefore, M. foleyi should be removed from the genus Moricandia and ascribed, at least provisionally, to the genus Eruca: Eruca foleyi (new combination): Eruca foleyi (Batt.) Lorite, Perfectti, Gómez, González-Megías & Abdelaziz comb. nov. urn:lsid:ipni.org:names:77166016-1 (see Appendix for a formal taxonomical description).

Conclusions

We have reported an inclusive dated phylogeny of the genus Moricandia, showing that, after the adscription of M. foleyi to the genus Eruca, it is a recent monophyletic genus that evolved in North Africa between 5.81 and 7.89 Ma diverging from its sister monotypic genus Rytidocarpus. Following the new aperture of the Strait of Gibraltar during the Zanclean period, Moricandia diverged in two different lineages: the Iberian and Afrian clades. At a finer scale, future work should address the evolutionary relationships between the different subspecies of M. arvensis and confirm the phylogenetic position of M.  sinaica.

Supplemental Information