Introduction

The recent revision of the genus Tamarix in Crete (Villar & al. 2014) confirmed the presence of at least four species on that island: T. hampeana Boiss. & Heldr., T nilotica (Ehrenb.) Bunge, T. parviflora DC. and T. smyrnensis Bunge. Among these, the record of T. hampeana was considered to be controversial and its identity was left in need of further clarification. It was difficult to obtain a precise identification of the specimens initially assigned to T. hampeana (Turland 2010; Villar & al. 2014). In fact, the different taxonomic works consulted (Boissier 1849, 1867; Bunge 1852; Baum 1968, 1978) did not lead to satisfactory results in respect to the morphology of the studied specimens or their identity with T. hampeana or any other described taxa. The specimen MO6207620 (herbarium codes according to Thiers 2015+) was first identified as T. africana Poir. However, on B. Baum's advice, it was finally identified and reported as T. hampeana (Turland 2010; Villar & al. 2014). On the other hand, the specimens ABH54194, ABH54195 and ABH54196 (see Appendix) were first identified as T. tetragyna Ehrenb., due to the high proportion of tetramerous and pentamerous flowers present in the late-season racemes. Nevertheless, after examining several herbarium specimens of T. tetragyna, its morphological features showed clear differences from the Cretan specimens and, therefore, that identification was proven to be clearly erroneous. These specimens were finally assigned, with reservations, to T. hampeana (Villar & al. 2014).

On account of the doubts held about the taxonomic identity of these specimens from Crete, a deeper morphological study was conducted. In addition, a preliminary phylogeny, based on three plastid DNA regions (intergenic spacers trnS-trnG, ndhF-rpl32 and trnQ-rps16), was conducted to identify which of the existing taxa could be genetically closest to the controversial Cretan samples.

The combination of morphological and molecular evidence allows the description of a new species, Tamarix minoa, apparently endemic to Crete. The morphological features of the new species, as well as its taxonomic relationships, are described below.

Material and methods

The material on which this study is focused was collected at Georgioupoli, on the N coast of W Crete. The precise locations for all samples are not more than 1 km apart, so they can be considered to belong to a single population. The specimens collected by Turland and Bareka are kept at UPA, with duplicates at B and MO; the ones collected by Villar, Alonso and Vicente are conserved at ABH.

In order to clarify the morphological identity of the Cretan specimens, over 120 herbarium specimens of the morphologically closest taxa (Tamarix africana and T. hampeana) were studied, including type material (Villar & al. 2015) kept at G, K, MA and P. Moreover, different European and American herbaria (ABH, B, G, MO, MPU, P, PR, PRC, VAL and W) (Thiers 2015+) were visited in order to obtain a wider view of the morphological characterization of the whole genus. In addition, field collections were made around the W Mediterranean region, where T. africana is mainly distributed, as well as around the coasts of continental Greece, where T. hampeana is widespread. Those field collections are kept at ABH. Detailed observations and morphological measurements of vegetative and flower parameters were undertaken according to Villar & al. (2012). The main floras and taxonomic works dealing with Tamarix in the Mediterranean region were consulted (Candolle 1828; Boissier 1849, 1867; Bunge 1852; Baum 1966, 1968, 1978).

A total of 23 samples of Tamarix, belonging to 10 species from 11 countries, were used for the molecular analyses (Table 1). Reaumuria alternifolia (Labill.) Britten, Myricaria bracteata Royle and Myrtama elegans (Royle) Ovcz. & Kinzik., all belonging to Tamaricaceae, were also included as the outgroup. Regarding Tamarix samples, in addition to three of the Cretan samples, we included the vaginate-leaved T. aphylla (L.) H. Karst, and T. usneoides E. Mey. ex Bunge; the generic type species T. gallica L.; species occurring geographically nearby (T. dalmatica B. R. Baum and T. parviflora); and those species once thought to match with the Cretan samples (Villar & al. 2014): T. africana, T. hampeana and T. tetragyna (along with its close relative T. boveana Bunge). All the studied samples are kept at the herbaria ABH or MO, with the exception of the outgroup samples corresponding to R. alternifolia and M. bracteata, which belong to John F. Gaskin, kept at the USDA ARS at Sidney (MT, U.S.A.), soon to be deposited at MO (Table 1).

Genomic DNA was extracted from silica-gel-dried leaf material (Chase & Hill 1991), following a modified 2xCTAB method (Doyle & Doyle 1987). Total DNA is kept at the ABH-DNA bank. The species-level phylogeny is based on three plastid intergenic spacers (trnS-trnG, trnQ-rps16 and ndhF-rpl32), which were amplified using the primers described in Hamilton (1999) and Shaw & al. (2007). The amplifications were performed in a reaction volume of 25 µl containing 22.5 µl AB Gene 1.1× Master Mix 2.5 mM MgCl₂ (Thermo Scientific Waltham, MA, U.S.A.), 0.5 µl of 0.4% bovine serum albumin (BSA), 0.5 µl of each primer (10 pmol/µl) and 1 µl of template DNA. The PCR programme used for all three regions was an initial denaturation of 94°C for 2 min, followed by 35 cycles at 94°C (1 min 15 sec), 55°C (1 min 30 sec) and 72°C (2 min); and a final elongation at 72°C for 10 min. PCR products were purified using Ultraclean^® PCR Clean-Up Kit (MOBIO, Carlsbad, CA, U.S.A.) micro-columns, following the instructions provided by the manufacturer. Both strands were sequenced with the same primers for each region and for all samples at Macrogen Inc. Korea ( www.macrogen.com).

Complementary strands were assembled using Sequencher 4.1 (Gene Codes Corp., Ann Arbor, MI, U.S.A.). The three plastid regions were aligned automatically by ClustalW, conducted in MEGA 5.05 (Tamura & al. 2011). Minor manual corrections were made to obtain the final alignments, and the three regions were joined in a single matrix. Maximum parsimony analyses (MP) and Bayesian analyses were performed on the plastid dataset.

MP analyses were conducted in PAUP v.4.0b10 (Swofford 2002), using heuristic search options. Searches included 1000 random addition replicates and tree-bisection-reconnection (TBR) branch swapping, with MULTREES in effect (keeping multiple most-parsimonious trees). All characters were treated as having equal weight. The shortest trees held in the heuristic search were used as initial trees for a final heuristic analysis, with the previously commented options. MP support was assessed by 1000 bootstrap replicates, TBR branch swapping, simple addition sequence and MULTREES on, keeping 10 trees per replicate (Salamin & al. 2003). For the MP analyses, the consistency index (CI) and retention index (RI) were calculated excluding uninformative characters. Clades showing bootstrap (BS) values between 50%–74 % were considered as weakly supported, 75%–89% moderately supported and 90%–100% strongly supported.

A Bayesian Inference (BI) analysis was conducted using MrBayes v.3.1.2 (Huelsenbeck & Ronquist 2001; Ronquist & Huelsenbeck 2003). The most accurate evolutionary model required for Bayesian estimation was selected using the AIC (Akaike Information Criterion) (Posada & Crandall 1998; Posada & Buckley 2004), conducted in JMODELTEST 2.1.5 (Darriba & al. 2012). Two simultaneous and independent analyses were performed. For each analysis, the Markov and Monte Carlo chains were run for 2 × 10⁶ generations and sampled every 100 generations. The chains became stationary and the average standard deviation of the split frequencies stabilized in 250 000 generations. Thus the first 250 trees were excluded (“burn-in”) and the remaining trees were used to compile a posterior probability (PP) distribution using a 50% majority-rule consensus.

Table 1.

Materials used in the molecular analysis.

Results and Discussion

Tamarix minoa J. L. Villar, Turland, Juan, Gaskin, M. A. Alonso & M. B. Crespo, sp. nov. — Fig. 1 & 2.

Holotype: Greece, Crete, Nomos Chanion, Eparchia Apokoronou, Georgioupoli beach, by river mouth at E edge of village, 35°21′34.3″N, 24°15′59.6″E, 0 m, sandy river bank on beach with springs emerging beneath the Tamarix trees, 28 Mar 2009, N. J. Turland 1778 & P. Bareka (UPA-Herb. Phitos & Kamari; isotypes: B, MO6207620).

— Tamarix hampeana sensu Turland in Willdenowia 40: 198. 2010, non Boiss. & Heldr. 1849.

Diagnosis — Species distincta primo aspectu ad Tamaricem africanam Poir. aemulans et cum T. hampeana Boiss. & Heldr. plerumque confusa, sed eis singulari characterum combinatio valde diversa et multo differt; racemis vernalis e ramorum vetustiorum productis aliquando junioribus hornotinis ramulis alternantibus; bracteis lineari-oblongis, concavis, subpatulis, calyces paulo superantibus; floribus plerumque pentameris, sed ad postremum tempum florendi cum aliis tetrameris floribus simul commixtis; sepalis acutis, marginibus denticulatis; petalis oblongo-obovatis vel obovatis, concavis vel leviter carinatis; antheris mucronatis.

Description — Shrubs or low shrubby trees, up to 5 m tall. Bark reddish brown to dark burgundy. Leaves green to glaucous green, lanceolate, (1-)1.5-2.5(-3.5) mm long, surface slightly papillose, bearing salt glands, more evident and abundant in older leaves, base narrow, slightly decurrent, broader and auriculate in leaves remaining from previous year, apex acuminate, slightly incurved. Racemes loosely arranged on previous year's branchlets, sometimes alternating with new-leaved twigs (Fig. 2B), 2.5-4.5(-6) cm long, 6–7.5 mm wide. Peduncle 3–5 mm long, with 4–8 scarious bracts grouped at base, sometimes showing 2–3 mm of naked rachis before first floral bract; rachis lacking papillae on all studied specimens. Bracts patent, yellowish brown, linear-oblong, narrowing at apex, concave, (2-)2.5-3.5(-4.5) mm long, usually slightly surpassing calyx, sometimes not reaching it, base narrow, slightly decurrent, sometimes showing 2 small auricles, margin sparsely papillose-ciliate, apex obtuse to acute, slightly incurved. Pedicels 0.5–0.75 mm long, always shorter than sepals. Calyx of 5 sepals (sometimes 4, very rarely 6)*. Sepals green, ovate-triangular, 1.5–2 mm long, 1–1.2 mm wide, 2 outer ones slightly larger than 3 inner ones, margin broadly membranous, markedly denticulate, apex acute; pentamerous calyx sepals almost alike; tetramerous* calyx outer sepals sometimes with a marked central nerve and apex apiculate, inner sepals evidently smaller and apex obtuse. Corolla of 5 petals (sometimes 4, very rarely 6)*. Petals deciduous at fruit maturity, white, oblong-obovate to obovate, concave to keeled, recurved at late anthesis, 2.8–3.2 mm long, 1–1.5 mm wide, margin sometimes irregular, base sometimes cuneate, not unguiculate. Staminal disk dark brown, 5-lobed (rarely 4-lobed, exceptionally 6-lobed)*; lobes antesepalous, broadly ovate, apex somewhat progressively narrowed into filaments; filaments white, 1.5–2 mm long; anthers pale yellow to light pink, 0.8–1 mm long, 0.5–0.6 mm wide, apex mucronate. Gynoecium of 3 styles and 3 carpels, (rarely 4)*; styles nearly 0.6 mm long.

*More common in late-season racemes.

Phenology — Flowering mainly in March and April, though some late-season racemes may extend the flowering period until the end of May.

Distribution and biogeography—At present, Tamarix minoa is known only from the type locality at Georgioupoli on the N coast of W Crete in the S Aegean region, Greece. Further exploration in the field and examination of herbarium material are needed to clarify if its distribution is confined to Crete or if it extends to other territories. According to Rivas-Martínez & al. (2004), the type locality of T. minoa is located in the Cretan Subprovince, which belongs to the Graeco-Aegean Province, Eastern Mediterranean Subregion and Mediterranean Region. The biogeographical unit Cretan Subprovince matches the “Cretan area” (“Cr”) of Flora europaea (Tutin & al. 1964–1980) and the Euro+Med PlantBase (Euro+Med 2006+) as well as the floristic region “Kriti+Karpathos” (“KK”) of Flora hellenica (Strid & Tan 1997, 2002) and the recent checklist of Greek vascular plants (Dimopoulos & al. 2013).

Ecology — Growing on sandy river banks close to the sea, with springs of percolating (fresh or brackish?) groundwater emerging beneath the Tamarix trees, exposed to inundation by seawater during storms (Fig. 2A).

Etymology — The epithet minoa makes reference to the Minoan civilization, which existed in Crete in the Bronze Age between 3500–2500 BCE and c. 1100 BCE. The term “Minoan” itself is derived from Minos (Mívω;ς;), a mythical king of Crete.

Fig. 1.

Tamarix minoa—A: young branchlet with leaves; B: portion of branchlet with bracts and flower (petals removed); C: bracts, adaxial (left) and abaxial (right) surfaces; D: flower, basal view, showing pedicel and calyx; E: sepals, abaxial surface, outer sepal (left), inner sepal (right); F: petals, adaxial surface; G: staminal disk, apical view with filaments and anthers; H: anthers, lateral views, with distal part of filament (left), with apex at bottom (right); I: gynoecium, lateral view. — Drawn by María García Rivas from the isotype MO6207620.

Fig. 2.

Tamarix minoa — A: tree growing on sandy ground beside river (left), springs emerging beneath tree, sea in distance; B: flowering branchlets. — Type locality, 28 Mar 2009, photographs by N. J. Turland.

Fig. 3.

One of the six most parsimonious trees obtained (774 steps, CI 0.935, RI 0.916). Branch lengths for parsimony analysis are shown above branches. Bootstrap values (BS) > 50% and Bayesian posterior probability (PP) > 0.90 are shown below branches (BS/PP). The outer vertical bars represent the outgroup and the two first split clades in Tamarix. The inner vertical bars represent the sections accepted by Baum (1978).

Remarks — According to our observations, Tamarix minoa is mostly a pentamerous-flowered species, with 5 sepals, 5 petals, 5 stamens and 3 styles. A few rare flowers, though, can be found to be tetramerous or hexamerous in some of the floral whorls, more frequently on late flowering racemes. Often tetramery or hexamery does not affect all whorls at the same time. Most of the examined flowers with a tetramerous calyx had a pentamerous staminal disk. Most of the hexamerous staminal disks were found in pentamerous flowers in relation to sepals and petals. In a few cases, tetramery of sepals was related to hexamery of petals (or vice-versa). No special pattern was found in the occurrence of a 4-styled/-valved gynoecium.

Additional specimens seen (paratypes) — Greece, Crete, Chania, Georgioupoli, river mouth, 30S KV 508171 [35°22′00″N, 24°15′28″E], 12 Jul 2009, Alonso, Vicente & Villar T2CR (ABH54194); ibid., 30S KV 512172 [35°22′04″N, 24°15′42″E], 12 Jul 2009, Alonso, Vicente & Villar T3CR (ABH54196); Georgioupoli, 30S KV 516163 [35°21′34″N, 24°15′59″E], 12 Jul 2009, Alonso, Vicente & Villar T4CR (ABH54195).

Morphological comparisons

Despite the taxonomic complexity of the genus, and the diffuse borders between similar species, we can highlight some morphological features to separate Tamarix minoa from the morphologically close T. africana, as well as from its previous identification of T. hampeana (Table 2).

Leaves — Leaves are similar between Tamarix minoa and T. africana, though those of T. africana tend to show a broader auriculate base. In both species the longer leaves rarely reach 3.5 mm. The leaves of T. hampeana are narrower, and it is common to find some longer ones (4–7 mm) at the base of the first young twigs of the year.

Racemes — Tamarix hampeana racemes are among the larger ones in the genus (usually 4-10 cm long × 7-12 mm wide), some reaching up to 16 cm long × 14 mm wide. The racemes of T. minoa and T. africana, are smaller (usually less than 5 cm long × 8 mm wide), although the smallest T. hampeana racemes can overlap in size with T. africana and T. minoa. However, those of T. minoa show the peculiarity of frequently alternating with new leafy twigs.

Bracts — The bract size of the three species falls into a similar range. However, the bracts of Tamarix minoa are patent, linear-oblong, only narrowing at the apex, with a narrow slightly decurrent base sometimes with two small auricles. On the other hand, the bracts of T. africana are triangular-oblong, always narrowing progressively toward the apex, with a narrow decurrent base sometimes calcarate with two small auricles that can even be slightly decurrent. The bracts of T. hampeana can be ovate-oblong, oblong or narrowly triangular; they are so variable that they are not very reliable for comparison with the other two species.

Pedicels — The pedicels of Tamarix hampeana (usually 1–4 mm long) are frequently longer than the sepals and significantly larger than those of T. africana and T. minoa (in both species subsessile to 0.75 mm long).

Calyx — Tamarix africana and T. minoa show essentially a pentamerous calyx, although both species can show exceptions. Tetramerous calyces are more common in T. minoa than in T. africana. Tamarix hampeana shows flowers with normally 4 or 5 sepals, but sometimes 6 or 7 or even 8 or 9. This variability makes it difficult to identify T. hampeana as tetramerous or pentamerous. The sepals of T. minoa look similar, the two outer ones are slightly larger than the three inner ones, all are clearly denticulate at the margin and most are acute at the apex. The two outer sepals of T. africana tend to be narrower and usually show a marked central nerve, and all sepals usually have a subentire margin and obtuse apex. The two outer sepals of T. hampeana tend to be narrower and sometimes show a marked central nerve, and all sepals have a subentire margin and acute to obtuse apex.

Corolla — Tamarix africana and T. minoa are essentially pentamerous, with rare exceptions. In T. hampeana, the number of petals is as variable as the number of sepals (4-6(-9)). The petals of T. hampeana tend to be ellipticoblong, whereas those of T. minoa are oblong-obovate to obovate and concave to keeled, and those of T. africana tend to be broadly ovate and flat (in the typical variety) to oblong-unguiculate. In a single flower, for all three species, all petals are the same size.

Staminal disk — The configuration of the disk is very similar in Tamarix minoa and T. africana, both showing disks with the filaments rising progressively from the apex of disk lobes. They are also quite stable in showing 5 stamens, rarely 6, exceptionally 7 or 8 (Villar & al. 2012). On the other hand, the filaments of T. hampeana rise abruptly from the apex of the disk lobes. Moreover, T. hampeana shows an extremely high variability in the staminal disk features: disks with 4 or 5 staminal filaments are not always easy to find, whereas disks with 6-8(-10) well-developed stamens are common.

Anthers — Only the anthers of Tamarix minoa are clearly mucronate. Those of T. africana and T. hampeana are not, although submucronate anthers can sometimes be found.

Table 2.

Some key morphological features for distinguishing among Tamarix minoa and the related T. africana and T. hampeana.

According to the molecular and morphological results obtained, we can conclude that Tamarix minoa is indeed an independent species well separated from any other taxa of Tamarix. The closest species with respect to morphology is T. africana. However, T. minoa can be separated from T. africana by the linear-oblong concave bracts, acute and denticulate sepals, concave-keeled petals and apiculate anthers. Our phylogenetic results have pointed out the well-supported, isolated position of the studied samples, clearly separated from the morphological close taxa T. africana and T. hampeana, as well as from geographically close taxa such as T. dalmatica and T. parviflora. Regarding morphology, the correct identification of Tamarix collections is known as a troublesome matter, and T. africana is one of the most widely distributed and morphologically variable species in the Mediterranean region. Taking these points into consideration, we would recommend being careful with future records of apparent T. minoa from outside the Aegean region, in order to avoid mistaken identifications, as happened with T. dalmatica in the Iberian Peninsula (Villar & al. 2012).