Field work and plant material

Plant materials were collected during two field expeditions in 2011 and 2013. The herbarium specimens and the type specimens are preserved in the herbarium of the Halophytes and C₄ plants Research Laboratory (herb. Akhani), housed at the School of Biology, University of Tehran and duplicated in the herbaria of Iranian Research Institute of Plant Protection (IRAN) and the Botanic Garden and Botanical Museum Berlin (B). The habitat of the species was studied by recording cover-abundance and vegetation data in a plot according to Braun-Blanquet (1964). The Electric Conductivity (EC) of the water in the river was measured by a commercial EC-meter Windaus (Model WinLab Data Line).

Morphology

The morphological descriptions are provided by detailed examination and measurements of the specimens. The details of floral parts were first photographed by an Olympus stereomicroscope (SZX 12) coupled with a digital camera DP 12. The details were then measured using the software Olysia Bioreport 3.2 (Build 670).

Karyotype

Young flowers were fixed in Pinar solution (ethanol: chloroform: propionic acid; 6: 3: 2) in the field in March 2013 for at least 48 hours at 4°C, and were then stored in 70 % ethanol in a refrigerator. Slides were prepared by squashing anthers in 2 % acetocarmin. All slides were examined under a Nikon OPTOPHOT-2 and photographed by a Moticam 2300 digital camera.

Anatomy

Segments of first-year stems (3–4 mm long) were fixed in AFE (acetic acid glacial- formalin- ethanol 70 %, 1: 1: 9). They were dehydrated in an ethanol series and embedded in Araldite resin (TAAB, Berkshire, U.K.) based on Millonig (1976). Ultra-thin transverse sections (1.8 µm) were made using a glass knife and a Leica Ultracut UCT microtome (University of Tehran). Sections were stained using 0.5 % (w/v) Toluidine blue in 1% (w/v) Na₂CO₃. Images were taken with DP12 Olympus digital camera mounted on an Olympus BX51 microscope.

For epidermal studies we followed slightly modified procedures in Bokhari (1971) and Akhani & al. (2013). Leaves were stored in KOH 10% plus a few drops of H₂O₂ for 12 hours, then were washed 3 times with distilled water and immersed in H₂O₂ + distilled water for two hours. After washing with distilled water, leaves were soaked in Javel water for an hour. Macerated epidermis was separated and stained with Toluidine blue for a minute. Permanent slides were prepared using Entellan mounting media (Merck, Darmstadt, Germany). Quantitative traits were measured using image processing software ImageJ ( http://rsb.info.nih.gov/ij/).

Molecular phylogeny

DNA isolation and sequencing of ITS followed the approach described in Malekmohammadi & al. (2017). PCR conditions for the trnGtrnS spacer were as described in Schäferhoff & al. (2009) with the primers trnS and trnG (Hamilton 1999) for amplification and sequencing and for the rpl16 intron followed Sánchez-del Pino & al. (2012). The sequences were edited manually and aligned with a motif-based approach (Löhne & Borsch 2005) using PhyDE (Müller & al. 2012). Simple indel coding (Simmons & Ochoterena 2001) was implemented with SeqState (Müller 2005) to generate a presence-absence matrix that was added to the sequence matrix. Parsimony analysis was done with PAUP* v.4.0b10 (Swofford 2002), employing random stepwise addition with 1 tree held at each step and TBR branch swapping. Jackknifing employed 10,000 replicates with 36.788% of the characters deleted and one tree held at each replicate. MrBayes v. 3.2.5 (Ronquist & al. 2012) and a GTR+G model was used for Bayesian inference (BI) as determined with jModeltest2 (Darriba & al. 2012). Four runs with four chains each were carried out for 10 million generations and the first 25% of the trees discarded as burn-in. This fitted with convergence diagnostics (standard deviation of split frequencies), effected sampling sizes and detection of stationarity based on log-likelihood plots. Myricaria Desv. and Reaumuria L. constituted the out-groups in both data sets. Trees were than illustrated with TreeGraph2 (Stöver & Müller 2012). Our phylogenetic analysis focuses on the samples relevant to the description of this new species but is excerpted from a larger data set generated in an ongoing phylogenetic analysis of the genus Tamarix. We have deliberately included several accessions of T. kotschyi to ensure that our new species is not just a hybrid or an aberrant population of this species.

Description of new species

Tamarix humboldtiana Akhani, Borsch & N. Samadi, sp. nov. – Fig. 1, 2.

Holotype: Iran, Hormozgan, c. 100 km ENE of Bandar Abbas, 2 km S of Bikah village, along Minab river near Deh Gel Kan village, 27°19′56″N, 57°12′07″E, 153 m, 8 Mar 2011, H. Akhani 21693 (IRAN; isotypes: B [bar-code B 10 0465441; stable identifier  http://herbarium.bgbm.org/object/B100465441], herb. Akhani).

Morphological diagnosis — Based on molecular phylogenetic data, Tamarix humboldtiana is closely related to T. kotschyi and T. tetrandra. It differs from both species by its pentamerous flowers. It differs further from T. kotschyi by its clearly pedicellate (vs ± sessile) flowers. Anatomically, T. humboldtiana differs from T. kotschyi by the absence of ingrowing papillae that overarch the stomatal pores.

Molecular diagnosis — Rpl16 intron, positions in the sequence of the type specimen, upstream of the large 3′ exon: nucleotide character states “C” in pos. 135, “G” in pos. 195, “C” in pos. 323, “G” in pos. 896. TrnGtrnS spacer, positions in the sequence of the type specimen, downstream of the trnG exon: nucleotide character state “T” in pos. 550.

Morphological description — Shrubs, to 2 m tall; bark reddish, glabrous; inflorescence parts indistinctly papillose, irregularly powdery on surface under high magnification (×100). Leaves subvaginate, 1.1–2 mm long; young stems 5–7 mm in diam., covered by amplexicaul leaves. Racemes appearing in winter, prior to development of leaves, ± perpendicular to slightly oblique to inflorescence axis, 1–2.5 × 0.6–0.7 cm, 7–20-flowered; pedicels of all flowers 1–1.5 mm long; bracts shorter than pedicels, amplexicaul-vaginate, triangular, 0.75–1 mm long, margin membranous, entire or irregularly crenate, apex acute. Flowers pentamerous. Calyx ovate, (1–)1.1–1.4(–1.5) × 0.8–1 mm, margin with a narrow, membranous band 0.18–0.23 mm wide, apex ± obtuse to ± acute. Petals white, elliptic, (1.6–)2–2.2 × 0.8–1.1 mm, apex obtuse. Staminal disk synlophic (peridiscal and lacking intermediate lobes), dark red, c. 0.8 mm in diam.; filaments 1.7–2.2 mm long; anthers c. 0.9 mm long, apex minutely apiculate. Ovary (including style) reddish in living state, brownish to yellowish when dry, conical, to 4.5 mm long; stigmas 3, 0.2–0.4 mm long. Seeds unknown.

Anatomical description — Leaves are dorsiventral (bifacial), attached with adaxial surface in direct contact with stem and abaxial surface exposed to sunlight (Fig. 5A). Epidermal cells are narrowly oblong, polygonal or irregular with straight anticlinal walls (Fig. 5C). Epidermal surface is dotted with stomata and 8–celled salt glands. Stomatal type is brachyparacytic. Stomatal pores are transversely arranged to course of vein. Both stomata and salt glands are sunken. Young stems include 6 or 7 discrete vascular bundles; outer parts contain massive sclerenchyma fibres (perivascular fibres) functioning as mechanical tissue (Fig. 5A). Pith cells are round or elliptic with narrow walls. Chlorenchyma tissue occurs in outer cortex of stem. Epidermis of sun-exposed stems has stomata and salt glands similar to leaves but fewer in number.

Karyological description — Meiotic study in this species shows a diploid chromosome number with n = 12, as in most other Tamarix species (see Samadi & al. 2013). Twelve bivalents are vividly distinguishable in diakinesis and metaphase I (Fig. 4).

Molecular description — Sequences describe the type specimen (code T282) and are available in EMBL/GenBank/DDBJ under accession numbers LR583807 (rpl16 intron), LR584012 (trnGtrnS spacer) and LR583687 (nrITS).

Phenology — Flowers February–March; fruits March–April.

Eponymy — The epithet “humboldtiana” commemorates the great German phytogeographer Alexander von Humboldt (1769–1859) whose 250^th birthday is in the year of publication of this species. The name was also inspired by the Alexander von Humboldt Foundation having supported a research internship on diversity and phylogeny of the genus Tamarix.

Additional specimens examined — Iran: Hormozgan: c. 100 km ENE of Bandar Abbas, 2 km S of Bikah village, along Minab river near Deh Gel Kan village, 27°19′38″N, 57°12′05″E – 27°19′52″N, 57°12′10″E, 153 m, 21 Feb 2013, H. Akhani, M. Dehghani, M. Doostmohammadi & A. Noormohammadi 23716 & 23717 (herb. Akhani).

Fig. 1.

Tamarix humboldtiana. – A: habit and habitat along Minab river; plants grow on the river side and many of them are damaged by flooding; B: part of inflorescence with young flowers; C: part of inflorescence, showing whitish petals and reddish ovary and anthers. – All photographs taken at the type locality on 21 Feb 2013 by H. Akhani.

Fig. 2.

Details of leaf, inflorescence and flower parts of Tamarix humboldtiana. – A: part of leafy young branch showing vaginate leaves; B: raceme, note long pedicels; C: calyx outside (left), calyx inside (centre), ovary and stigma (right); D: petals; E: staminal disk, showing 5 peridiscal filaments attached to synlophic disk. – Scale bars: A–E = 1 mm.

Sequence characters and phylogenetic relationships

The trnG-trnS spacer contained two poly A/T microsatellites, the latter of which was excluded from analysis as a mutational hotspot because of unclear homology of sequence elements in particular with respect to Myricaria and Reaumuria. Another highly variable poly AT/ TA satellite-like region located approximately 700 positions downstream of trnG was also excluded because of unclear homology. Another long poly A/T microsat- ellite was found close to the trnS exon but the matrix was trimmed upstream because pherograms of the trnG primer after the poly A/T stretch were not readable and also pherograms of primer trnS were not reliable in the initial positions. The trnGtrnS matrix had 993 positions and contained 14% variable characters of which 5% were informative. Two mutational hotspots located 345 bp downstream of trnG and 692 bp downstream of trnG were excluded because of uncertain homology (variable polyA microsatellite and AT-rich sequence elements in other genera; variable AT satellite-like elements). The rpl16 intron data set began approximately 70 bases downstream of the small rpl16 5′ exon. Sequences were completely alignable and yielded a matrix of 1121 positions of which 9.4% were variable and 3.3% informative. A total of 59 indels were coded, of which 12 were informative. The ITS pherograms were without polymorphic sites except the sequence of the type specimen of Tamarix humboldtiana, which had a double signal of “A” and “G” in position 3 of ITS2 downstream of 5.8S.

Fig. 3.

Distribution map of Tamarix humboldtiana ().

The phylogenetic trees based on combined plastid rpl16 and trnGtrnS sequence data (Fig. 6A) as well as nuclear ITS (Fig. 6B) both show that Tamarix humboldtiana belongs to a well-supported clade with T. kotschyi and T. tetrandra. The trees from the different genomic compartments are incongruent regarding the position of the new species. In the plastid tree, T. humboldtiana is sister to T. tetrandra (0.9 PP), and this lineage again is sister to a well-supported monophyletic group of all individuals of T. kotschyi (1.0 PP, 95% JK). To the contrary, the ITS topology depicts sequences of T. humboldtiana and T. kotschyi in an also well-supported clade (0.97 PP, 97% JK). The ribotype obtained from the type specimen of T. humboldtiana even appears nested among T. kotschyi ribotypes. However, this position is not well supported and is based on very few sequence differences. Since deviant ITS sequences of two different ancestral species can also evolve in a pattern that is biased toward one parent (Wendel & al. 1995; Winterfeld & al. 2009), the observed incongruent gene trees are in line with a case of hybrid speciation. Interestingly, the only polymorphic site identified in the pherograms of the T. humboldtiana sample exhibits both “A” and “G”. The “A” is the common state in Tamarix, whereas “G” occurs in T. meyeri Boiss. and T. tetrandra. This would be in line with a scenario of hybrid speciation including ancestors of T. kotschyi and T. tetrandra. Nevertheless, there are many more differences between the ITS sequences of T. kotschyi and the new species, suggesting largely biased ITS evolution towards the T. kotschyi ribotype. Further phylogenomic analysis and the inclusion of more samples will likely shed more light on this in the future. Villar & al. (2019) sampled neither T. kotschyi nor T. tetrandra. The rather distant cp haplotype of sample Akhani 21693 (the type specimen of T. humboldtiana) was not found in more extensive plastid analyses of Tamarix (Akhani & Borsch, pers. comm.), indicating that the maternal parent remains unknown and possibly extinct. A further difference of the rpl16 sequence of the sample constituting the type in comparison with the other sequences is a longer polyA-stretch upstream of the large 3′ exon in positions 496–504 with 9 A/Ts only in the sample of the type specimen. Nevertheless, we did not use it as diagnostic character because microsatellites are highly variable, making their diagnostic potential at species level questionable (González Gutiérrez & al. 2013).

Fig. 4.

Meiosis in Tamarix humboldtiana, n = 12. – A: diakinesis; B: metaphase I. – Scale bar = 10 µm.

Table 1.

Morphological, micromorphological and anatomical comparison of Tamarix humboldtiana and T. kotschyi. The morphometry of anatomical data of T. kotschyi are the average of five specimens (Akhani & al. 21834, 21977, 22265, 22267 and 22395). The leaf and stem anatomical measurements of T. humboldtiana refer to Akhani & al. 23716 and the epidermis of Akhani 21693.

Interestingly, none of the closely related putative species or their close ancestors occurred sympatrically with the new entity found along the Minab river. The observed sympatric species were Tamarix mascatensis Bunge and T. stricta Boiss., both of which are very distantly related (Akhani & Borsch, unpubl. data). Tamarix mascatensis appears close to T. meyeri in the plastid tree (Fig. 6A). The phylogenetic trees therefore reject any close relationship between our new species and T. mascatensis, although both have pentamerous flowers.

Fig. 5.

Cross-section of stem and surrounding leaf and epidermal micromorphology of Tamarix humboldtiana and T. kotschyi. – A: stem and surrounding leaf of T. humboldtiana. The vaginate part of leaf and stem tissues can be distinguished by the collapsed parenchyma cells, which can be distinguished from the grooves on the lower corners of the section, which separate leaf and stem; note the slightly sunken stomata and salt glands; B: stem and surrounding leaf of T. kotschyi; note that leaf encircles more than half of the stem in both species but in T. humboldtiana much larger parts of the stem are covered; C: leaf adaxial epidermis of T. humboldtiana showing brachyparacytic stomata and subsidiary cells lacking ingrowing papillae; D: T. kotschyi, brachyparacytic stomata characterized by two ingrowing papillae. – Abbreviations: S = sclerenchyma; S.C. = subsidiary cells; P = phloem; X = xylem. – Scale bars: A, B = 200 µm; C, D = 20 µm.

Our plastid and nuclear trees show the Tamarix aphylla clade (including T. aphylla (L.) H. Karst. and T. usneoides E. Mey), which was also found by Villar & al. (2015a, 2019) based on combined trnGtrnS, ndhFrpl32 and trnQrps16 spacers with high support. They also both reveal a clade including several species (e.g. T. meyeri) to which T. ramosissima Ledeb. is congruently resolved as sister. This clade represents most of the diversity of Tamarix in this study, but its position is inconsistent between the ITS and plastid trees (Fig. 6). The earliest studies of Gaskin & Schaal (2003) also found T. usneoides as sister to the remainder of the genus based on ITS, but this position in their tree based on sequences of the trnGtrnS spacer was not supported. Further work is therefore needed to establish a robust plastid tree for Tamarix and to test for possible deep reticulations.

Morphological and anatomical characters

Superficially, Tamarix humboldtiana is also similar to T. mascatensis but differs clearly by having different chloroplast and nuclear sequences, subvaginate leaves (not only amplexicaul leaves), and pedicellate flowers. Both T. humboldtiana and T. kotschyi have a brachyparacytic stomatal type, although the laterocytic type rarely occurs in T. kotschyi. The subsidiary cells in T. kotschyi are associated with ingrowing papillae that overarch the stomatal pores, which might have a role in minimizing water loss (Fig. 5C). This character is absent in T. humboldtiana (Fig. 5D). Additional differences in epidermal cells and stomatal density between T. humboldtiana and T. kotschyi are given in Table 1.

Fig. 6.

Trees showing the position of the type specimen of Tamarix humboldtiana. – A: majority rule phylogram from Bayesian analysis of combined plastid rpl16 intron and trnG-trnS spacer sequences; B: of nrITS sequences. Values above branches are bootstrap posterior probabilities, and below branches are jackknife percentages. Note that plastid haplotypes of T. kotschyi are resolved as monophyletic, whereas the sequence of T. humboldtiana is sister to T. tetrandra. In the tree inferred from nuclear ITS, T. humboldtiana appears among the sequences of T. kotschyi.

Table 2.

List of voucher specimens used in phylogenetic reconstruction and relationships of Tamarix humboldtiana.

Continued

Table 3.

Vegetation characteristics and species composition of a stand of Tamarix humboldtiana at the type locality.

Morphologically and geographically Tamarix humboldtiana shows some similarity to T. kermanensis Baum (Baum 1967), but T. humboldtiana has subvaginate leaves, shortly pedicellate flowers, and white petals. Tamarix kermanensis occurs commonly on sandy dunes and near saline rivers in Kerman, Hormozgan, Sistan and Baluchestan Provinces in Iran and adjacent Pakistan (Qaiser 1981). In contrast to T. humboldtiana, T. kermanensis has distinctly vaginate leaves, subsessile flowers, red petals, and triploid as well as tetraploid chromosome compliments (Samadi & al. 2013). Villar & al. (2019) found T. kermanensis in a clade together with T. nilotica (Ehrenb.) Bunge and T. senegalensis DC. in ITS trees and sister to T. canariensis Willd. in their plastid trees, albeit without statistical support. Our own unpublished data point to relationships of T. kermanensis to T. arceuthoides Bunge and T. ramosissima (H. Akhani & T. Borsch, pers. comm.), which are completely different lineages than the T. kotschyi clade depicted here (Fig. 6).

In his monograph of Tamarix, Baum (1978) synonymized the pentamerous T. leptopetala Bunge with the tetramerous T. kotschyi, a view accepted by Villar & al. (2015a). The type locality of T. leptopetala from the Lar valley in the Alborz Mountains (in valle Loura in montium Elbrus, 1813, Kotschy 728, GOET!, W!) has been examined in various seasons during our project. The only species found in the area is T. ramosissima. However, the specimen matches well T. arceuthoides and T. mascatensis. The presence of a few pentamerous flowers, as exhibited by the type specimen of T. kotschyi, is very rare in natural populations of this species, unless it hybridizes with its sympatric species T. arceuthoides and T. mascatensis. Whereas the type specimen of T. leptopetala clearly differs from T. humboldtiana by having leaves not subvaginate, flowers without distinct pedicels, and moreover a completely different distribution and ecology. However, a definite answer regarding the conspecifity of T. leptopetala with T. kotschyi can only be given through DNA sequence data from the type.

Habitat and conservation

A small thicket of Tamarix humboldtiana was found along the margin of the Minab river on sandy-gravelly soils (Fig. 1A). This patch grows on the innermost vegetation zone of the very wide river bed. Tamarix mascatensis and T. stricta are the most common species in the area, particularly at some distance from the river margin. In dense stands dominated by these two species T. humboldtiana is rare. The river is a freshwater river with an electric conductivity (EC) of 1644 µs and 0.8 salinity. The habitat of the species is very fragile, because it occurs in the innermost vegetation zone of the riverside subjected to frequent floods (Fig. 1A). The associated species with T. humboldtiana are T. mascatensis, T. stricta, Phragmites australis (Cav.) Steud., Saccharum sp., Juncus sp. and Nerium oleander L. (Table 2).

Most species of Tamarix are locally very common. They are successful plants by both vegetative and generative reproduction. The riparian habitat and the very tiny, pappus-bearing seeds are advantages for effective dispersal. The question is, therefore, if the species is really as rare as the currently available data suggest. There are many rivers in the area that need to be searched for this species. Our propagation experiments showed that production of adventitious roots via propagation is very poor in T. humboldtiana in comparison with many other species, such as the co-occurring T. mascatensis and T. stricta. Several of the propagated branches died off after we transferred them into soil. These results and the restriction of the patches of T. humboldtiana to the fragile zone of the river indicate that the populations are driven by other competitive species. Therefore, we could suggest that natural impacts play a major role in threatening the survival of this species. These impacts are accelerated by the current water abstraction in the area and frequent flooding in the fragile marginal habitat along the rivers. Based on T. humboldtiana being known only from one small site, it meets the criteria B1ab(iii) of the category Critically Endangered (CR) of the IUCN (2012) for assigning threat categories. We strongly recommend further research and implementation of measures to ensure in situ conservation and cultivation activities for ex situ conservation of the species.