Samples from three localities of the putative new taxon, and three samples of D. pubipetala from three different localities (represented by accessions nos. 1 to 6 in Table 1), were collected for morphological, anatomical, phytochemical, and molecular investigation. Voucher specimens were deposited in the BKF herbarium and duplicates were distributed to other herbaria. Three mature leaflets per accession were fixed in 70% ethyl alcohol for anatomical study. Young leaves of the three putative new taxon samples were collected and stored in silica gel for later DNA extraction. Stems and roots were also collected, cleaned, cut into smaller pieces and then dried at 50–60 °C in a hot air oven, and then stored at room temperature for phytochemical study.

Voucher specimens were examined using a stereomicroscope. The species description was prepared following the format of the Flora of Thailand (Sirichamorn 2020). Morphological measurements and comparison of specimens with (type) specimens of Derris-like plants housed in Thai herbaria (BK, BKF, and PSU) or available as online digital images (K, L, and P) were carried out.

Three lateral leaflets of mature leaves for each accession were used. Samples of 5 × 5 mm were taken from the center of the leaflets (from midrib to margin, including the midrib). Sections were cleaned, dehydrated in a series of ethyl alcohol, dried by critical point drying (CPD) in liquid CO₂, and preserved in a desiccator for subsequent observation by SEM. Samples were mounted directly on aluminum stubs using double-sided carbon tape, then sputter-coated with gold using an SPI module sputter coater. The samples were photographed using a Tescan Mira3 scanning electron microscope (SEM) at the Scientific and Technological Equipment Center, Faculty of Science, Silpakorn University.

Leaf epidermis was studied using the leaf scraping technique (modified from Johansen 1940). Mesophyll was scraped from the upper and lower surfaces with a razor blade, followed by epidermal bleaching using 10% sodium hypochlorite. Samples were cleaned three times in distilled water, then stained with 1% Safranin-O before washing again and dehydrating in a graded series of ethyl alcohol and a series of xylene. Samples were mounted on slides using DePeX mounting media (VWR international Ltd., England).

Three leaflets from 3 mature leaves for each accession were sectioned by hand to produce transverse sections, then stained with 1% Safranin-O, dehydrated in a graded series of ethyl alcohol and series of xylene. Finally, the sections were mounted on slides using DePeX. All leaf epidermal surfaces and sectioned parts were digitally photographed with an Olympus BX53 microscope with a DP27 camera attachment. Each leaf anatomical character was measured using ImageJ (Rueden et al. 2017).

Dried stems and roots of each accession were ground separately into a powder using a high-speed blender. The maceration process (modified from da Costa et al. 2016) involved soaking 5 grams of each plant sample in 250 mL of Dichloromethane in the dark at room temperature for 72 hrs. After maceration, the solvent extract of each accession was then filtered and subsequently evaporated in a dark fume hood at room temperature for 24 hrs. The crude extract was transferred into a sealed plastic tube, protected from light and stored in a refrigerator at 0–5 °C for further analysis.

The qualitative phytochemical analysis of rotenone and deguelin was analyzed using high-performance liquid chromatography (Agilent Technologies, Germany) consisting of a 1260 Infinity II LC system controller fitted with 1260 Infinity II Quaternary, 1260 Infinity Binary pump, 1260 Infinity II degasser, and 1260 Infinity II Diode Array UV-Visible detector. An Agilent 5 TC-C18 (4.6 × 150 mm, particle size 5 µm) reversed-phase analytical column was used. The isocratic method was performed for separating chemical substances using a mobile phase of acetonitrile/water (60:40; v/v) for 35 min. The injection volume was 10 µL, and the flow rate was 1.0 mL/min. The liquid chromatography system had to be stabilised for 15 min with the mobile phase before injecting the analyte. The analysis was performed at a wavelength of 294 nm. The control and data elaboration used Agilent OpenLAB ChemStation Edition.

Approximately 0.01 g of crude extract was dissolved and diluted with 1 mL of acetonitrile in a 2 mL microcentrifuge tube. The extracted solution was filtered using a syringe filter nylon membrane of 0.22 µm pore size. The chemical patterns of the two mains chemical markers were carried out using the HPLC system.

Taxon sampling for this study were selected based on the phylogeny reconstructed by Sirichamorn et al. (2012a) (Suppl. material 1). Additional DNA samples were extracted from dried young leaves of the three populations of the putative new species using the DNeasy Plant mini kit and modified protocol (Qiagen, Hilden, Germany). Two chloroplast regions [trnL-F intergenic spacer (IGS) and trnK-matK] and one nuclear region [the ribosomal internal transcribed spacer (ITS/5.8S)] were amplified using universal primers (Table 2). PCR reagents were carried out in a 25 µL reaction mixture which contained 1 µL (10 µM) of each forward and reverse primer, 12.5 µL GoTag Green Master Mix (Promega), 2 µL (< 250 ng) of total DNA and Nuclease-free water to 25 µL. PCR conditions for trnL-F IGS and trnK-matK followed a modified protocol of Hu et al. (2000). The ITS/5.8S region was amplified following Wojciechowski et al. (1993, 1999). Quality, quantity, and size of PCR products were tested by gel electrophoresis and genomic absorption. Samples were sent to Celemics, Inc. (http://www.celemics.com) for purifying and sequencing. Each purified fragment was treated using a Barcode-Tagged Sequencing (BTSeq) technique of Next-Generation Sequencing (NGS) (CELEMICS, Seoul, Republic of Korea) technology for automated dsDNA sequencing. All newly generated sequences in this study have been deposited in GenBank (Suppl. material 1).

Sequence alignments were performed using the program Bioedit v. 7.0.9 (Hall 1999) with the CLUSTAL W multiple alignment (default settings; Thompson et al. 1994) together with subsequent manual adjustment. Sequences of each marker were firstly aligned separately. Then, the combined data matrix was made by concatenating those already aligned datasets. Phylogenetic relationship was reconstructed using Maximum Parsimony (MP), Maximum Likelihood (ML), and Bayesian Inference (BI). For MP analyses, phylogenetic trees were constructed with the program PAUP* v. 4.0a169 (Swofford 2002) under heuristic search with 10,000 replicates of random taxon additions. The process employed Tree Bisection-Reconnection (TBR) branch swapping and Multrees activated, with all parsimonious trees saved. Bootstrap percentage analysis was used for the evaluation of MP clade support and calculated using the same settings (Felsenstein 1985). Maximum Parsimony Bootstrap Support (MPBS) is described as high (85–100%), moderate (75–84%), low (50–74%), or none (< 50%). The jModelTest v. 2 (Darriba et al. 2012) on the CIPRES web portal was used to find the best-fit substitution model chosen by the Akaike Information Criterion (AIC) scores (Akaike 1974). The General Time Reversible (GTR) (Tavaré 1985) nucleotide substitution model with a gamma distribution for among-site rate variation was selected for all DNA regions. ML analyses were performed for the combined data sets using the program IQ-TREE v. 2.2.0 (Nguyen et al. 2015) under GTR+G partition models implemented with the “-p” command. Bootstrap (Felsenstein 1985) clade support was calculated using a non-parametric bootstrap resampling with 2,000 replicates. Maximum Likelihood Bootstrap Support (MLBS) values are described as high (85–100%), moderate (75–84%), low (50–74%), or none (< 50%). Bayesian Markov Chain Monte Carlo (MCMC) (Yang and Rannala 1997) phylogenetic analyses were reconstructed using the program MrBayes v. 3.2.7a (Ronquist et al. 2012) via the CIPRES Science Gateway v.3.3 (Miller et al. 2010). The majority-rule consensus tree was started from random trees and run for 10,000,000 generations until stationarity, with MCMC sampled every 1,000 generations. The first 25% of all trees were discarded as burn-in and each BI clade was supported by posterior probabilities (PP) estimation from the remaining sampled trees. PPs are described as high (0.95–1), moderate (0.9–0.94), low (0.5–0.89), or none (< 0.5) support.

Three putative new Derris samples were morphologically studied and photographed as shown in Figs 1–3. They were collected from three localities in Songkhla and Nakhon Si Thammarat provinces (Fig. 4, a yellow square and 2 light-blue triangles). Their overall vegetative morphology is the same and they presumably represent a single species. The most remarkable characteristic found in all three samples is the reddish midribs of the mature leaflets. Only the sample collected from Pha Dam Forest Ranger Unit in Songkhla province had flowers. Morphologically, the most similar species in Peninsular Thailand is D. pubipetala, although this differs by several morphological characters. In general, the putative new Derris samples have paler bark and more leaflets per leaf, and the inflorescence is longer than that of D. pubipetala. Shape, curvature, and the lower auricle of the wing petals are also different. Stamens of the new Derris are hairy on the free part of their filaments and, more importantly, hairs at the base of the anthers are not present in D. pubipetala or any other species of Derris. Comparisons between the morphological characters of the new Derris and D. pubipetala are summarised in Table 3.

Figure 1. 

Derris rubricosta, sp. nov. A habit and habitat B leaves C stem D young reddish leaflets E reddish midribs of mature leaflets F, G close-up of midrib on adaxial and abaxial surface respectively H inflorescences and foliage I, J close-up of inflorescence, showing flower buds and brachyblasts K close-up of flower and buds. All photos were taken at Pha Dam Forest Ranger Unit in Songkhla province. Photos by Punvarit Boonprajan (A, B, D–G) and Charan Leeratiwong (C, H–K).

Figure 2. 

Comparative macro- and micro-morphological characters of leaflets and flowers of Derris rubricosta (A, A1–A10) and D. pubipetala (B, B1–B10) A, B a branch with leaves and inflorescences A1, B1 outer surface and A2, B2 inner surface of standard petal A3, B3 outer surface and A4, B4 inner surface of keel petals A5, B5 outer surface and A6, B6 inner surface of wing petals A7.1, B7.1 stamens and A7.2, B7.2 close-up stamens A8.1, B8.1 pistil and A8.2, B8.2 close-up pistil apex A9, B9 adaxial and A10, B10 abaxial leaflet surfaces. Photos by Charan Leeratiwong (A, B) and Punvarit Boonprajan (A1–A10, B1–B10). Scale bars: 1 mm.

Figure 3. 

D. rubricosta A inflorescence with leaf B close-up of inflorescence C flowers (top and side view) D outer surface and E inner surface of standard petal F outer surface and G inner surface of keel petals H outer surface and I inner surface of wing petals J staminal sheath K anthers L pistil M adaxial and N abaxial leaflet surfaces. Drawn by Punvarit Boonprajan from C. Leeratiwong 19-1666 (BKF). Scale bars: 5 mm (A, B); 1 mm (C–N).

Figure 4. 

Collection sites of the new Derris samples, represented by a yellow square (flowering specimen) and light-blue triangle (specimen without flowers) and D. pubipetala (orange circles) in Peninsular Thailand.

As seen in D. pubipetala, all samples of the putative new Derris (Table 1) showed no significant differences in their leaf anatomy. Micro-morphological and anatomical characteristics are summarised in Table 4 and presented in Figs 5–7.

Figure 5. 

Scanning electron microscope (SEM) photographs showing leaflet micromorphology A–D) adaxial and I–L abaxial surfaces of leaflet lamina E–H adaxial and (M–P) abaxial surfaces of midrib Q–T non-glandular and glandular trichomes A, B, E, F, I, J, M, N, Q, R the new Derris samples and C, D, G, H, K, L, O, P, S, T D. pubipetala; bt = bicellular non-glandular trichome; ep = epidermal cell; gt = glandular trichome; mr = midrib; nt = non-glandular trichome; st = stoma. Scale bars: 5 µm (T); 10 µm (Q, S); 20 µm (J, L, R); 50 µm (B, D, N, P); 100 µm (F, I); 200 µm (E); 500 µm (A, C, H, K, M, O); 1000 µm (G).

Figure 6. 

Comparative anatomical characters of the leaflet epidermis A–D adaxial leaflet epidermis E–H abaxial leaflet epidermis A, B, E, F, I, J the new Derris samples and C, D, G, H, K, L D. pubipetala. as = anomocytic stomata; bc = basal cell of trichome; nt = non-glandular trichome; ps = paracytic stomata. Scale bars: 50 µm (B, D, F, H–L); 100 µm (A, C, E, G).

Figure 7. 

Comparative anatomical characters of leaflet transverse sections A–L midrib M, O leaflet blade and N, P leaflet margin A, B, E, F, I, J, M, N the new Derris and C, D, G, H, K, L, O, P D. pubipetala. ar = adaxial ridge; cc = cortical cells; nt = non-glandular trichome; pf = perivascular fibers; pr = prism. Scale bars: 50 µm (M, N (below), O, P (above)); 100 µm (I–L); 200 µm (B, D, E–H, N (above), P (below)); 500 µm (A, C).

Leaf epidermis (Fig. 6A–L); Leaf epidermal cells of the three new Derris samples and those of D. pubipetala were quite similar on both leaflet surfaces, i.e., lobed or jigsaw-like in shape because of their undulate anticlinal cell walls (slightly more deeply undulate in the new Derris samples). The width and length of the epidermal cells on both surfaces ranged between 21.91–41.53 µm and 31.15–56.60 µm, respectively. Ratios of epidermal cell size (width: length) of the new Derris species were 0.77 (±0.20) for the adaxial and 0.66 (±0.12) for the abaxial surface, whereas the ratios of D. pubipetala were 0.83 (±0.09) for adaxial and 0.78 (±0.36) for abaxial surfaces, respectively. Leaflets of all taxa were hypostomatic with commonly paracytic and rarely anomocytic stomata. The width and length of stomata in the new species of Derris (w = 17.47±1.01 µm; l = 22.95±1.03 µm) was greater than in D. pubipetala (w = 12.43±0.73 µm; l = 15.04±0.52 µm), whereas the stomatal density of D. pubipetala (230±15.62 per mm²) was higher than for the new Derris (104.66±3.05 per mm²). The stomatal index of the new species was also higher (17.02±0.75) than that of D. pubipetala (16.52±1.04). Guard cell size of the new Derris was c. 22.65 µm long and c. 8.24 µm wide (ratio = 0.36±0.01). For D. pubipetala, guard cells were clearly smaller, c. 15.41 µm long and c. 5.17 µm wide (ratio = 0.33±00). Only unicellular non-glandular trichomes occurred on both leaflet surfaces in the new Derris. In D. pubipetala, unicellular non-glandular trichomes occurred only on the adaxial side and three different lengths (large: more than 400 µm; medium: > 100–400 µm, and small: less than 100 µm) of bicellular non-glandular trichomes were found only on the abaxial surface. No significant differences in leaflet epidermis were found among the three accessions of the new Derris or among populations of D. pubipetala.

Transverse sections (Fig. 7A–P); the adaxial leaflet surface midrib of the new Derris samples is convex, while it is flat or slightly concave in D. pubipetala. The midrib transverse section of the new Derris species is larger (w = 1200.15±234.88 µm; h = 1241.50±244.61 µm) than in D. pubipetala (w = 616.44±85.36 µm; h = 593.48±80.23 µm). The height of the epidermal cells on the adaxial and abaxial surface of the midrib ranged from 8.41–13.9 µm. in both species. The midrib of both taxa shows an elliptic or semi-circular shaped vascular bundle, surrounded by perivascular fibers. Inside the vascular bundle, vascular tissue is presented in two groups; smaller and fan-shaped on the adaxial surface and larger and horseshoe-shaped on the abaxial surface. The vascular bundle of all samples of the new Derris was larger (w = 1067.22±240.35 µm; h = 923.20±240.47 µm) than was found in D. pubipetala (w = 481.48±64.20 µm; h = 406.98±52.31 µm). Bicellular non-glandular and glandular trichomes were restricted to the midrib of the new Derris (Fig. 5E, F, M, N). In D. pubipetala, on the other hand, only bicellular non-glandular trichomes were found (Fig. 5O, P, 7C, D).

All taxa had dorsiventral leaves (Fig. 7M, O), with palisade mesophyll adaxially and spongy mesophyll abaxially. The new Derris samples have a greater leaflet thickness (204.27±37.90 µm) than in D. pubipetala (150.35±8.43 µm). The epidermal cell heights on the adaxial and abaxial surfaces of leaflet blades in all studied samples were between 9.22–23.49 µm. Furthermore, the mesophyll of the new Derris has a thicker 2–3-layered palisade parenchyma (105.14±19.70 µm) than in D. pubipetala (59.62±7.51 µm). However, spongy parenchyma in D. pubipetala was thicker (69.85±5.19 µm) than in samples of the new Derris (64.48±20.24 µm). Spongy mesophyll of D. pubipetala consists of variously-shaped, loosely arranged cells with larger intercellular air spaces than in the new Derris, which possessed fewer and smaller intercellular air spaces.

Transverse sections of the leaflet margin in all samples revealed them as being slightly revolute (Fig. 7N, P). The leaflet margin of the new Derris was thicker (143.98±5.86 µm) than in D. pubipetala (138.04±4.19 µm). Height of the epidermal layer on the adaxial and abaxial surfaces ranged from 7.73–14.10 µm. Thickness of palisade and spongy parenchyma at the leaflet margin of the new Derris samples and D. pubipetala were (74.57±6.01 vs. 31.88±2.27 µm) and (49.54±7.35 vs. 54.42±11.82 µm), respectively.

Leaflet transverse sections of all samples demonstrated the accumulation of prismatic crystals, generally associated with vascular bundles, and particularly in the leaflet midribs. Rhomboidal prisms and styroid prisms were two types of crystal observed in the leaflets of the new Derris species, whereas only rhomboidal di-pyramid prisms were seen in D. pubipetala. The reddish substance present in cortical parenchyma cells of the leaflet midrib on the abaxial leaflet surface is a unique characteristic observed only in the new Derris.

HPLC chromatograms (Fig. 8) revealed intra- and inter-specific differences in chemical compounds of the root and stem crude extracts. Under optimal HPLC conditions, the peaks of two chemical markers showed acceptable resolution (Fig. 8A). The standard peaks of rotenone and deguelin were detected at 6.1 (peak I) and 6.7 (peak II) min, respectively. Most of the HPLC chromatograms (8 of the 12) of the new Derris (Fig. 8B) and D. pubipetala (Fig. 8C) displayed two common chemical markers, i.e., Rotenone (I) and Deguelin (II). Exceptions were noted in the Derris new species’ accession no. 2 (both root and stem extract), in the stem extract of the new Derris accession no. 1, and in the stem extract of D. pubipetala accession no. 5. Peaks of unknown chemical compounds (III–XI) were detected only in D. pubipetala, and were absent in all samples of the new Derris.

Figure 8. 

High-performance liquid chromatography (HPLC) chromatograms of the new Derris samples and D. pubipetala roots (the lower, blueish lines) and stems (the upper, reddish lines) extracts A two standard compounds B The new Derris species samples C D. pubipetala; Each Arabic numeral represents the accession number of each sample presented in Table 1. Each Roman numeral represents each peak of HPLC chromatograms (chemical compound) with its retention time, i.e., (I) RT, 6.1 min; (II) DG, 6.7 min; (III) UK1, 13.3 min; (IV) UK2, 14.4 min; (V) UK3, 15.3 min; (VI) UK4, 24.0 min; (VII) UK5, 26.1 min; (VIII) UK6, 29.5 min; (IX) UK7, 32.2 min; RT = rotenone; DG = deguelin; UK = unknown.

A phylogeny based on combined nuclear and chloroplast sequences using three analyses (MP, ML, and BI) demonstrated different degrees of resolution and support, but with compatible tree topologies. According to the cladogram of the Bayesian analysis (Fig. 9), the clade of the genus Derris has very high support (PP 1.00, MLBS 99%, MPBS 98%) and is sister to the clade consisting of members of the genus Brachypterum including Millettia pinnata (L.) Panigrahi and Fordia splendidissima (Blume ex Miq.) Buijsen. The Derris clade is divided into several subclades with varying clade support.

Figure 9. 

Consensus tree from Bayesian Inference analysis (BI) of concatenated two plastid and one nuclear dataset. The posterior probabilities (PP) are shown above the branches. Bootstrap percentage support values of Maximum likelihood (MLBS) and Maximum parsimony (MPBS) are shown below the branches, respectively (- = MPBS and/or MLBS < 50%). Abbreviations and numbers after the scientific names indicate locality codes following Table 1: NN = Nopphitam district, Nakhon Si Thammarat province; RS = Rattaphum district, Songkhla province; SS = Sadao district, Songkhla province, Thailand. Red and blue lettering highlight the position of the three samples of the new Derris and D. pubipetala, respectively.

Samples of the putative new Derris species from three separate populations, however, did not form a clade in the Maximum Likelihood and Bayesian analyses because the sample from Nakhon Si Thammarat province (NN in Fig. 9) was separate from the other two samples from Songkhla province (SS and RS). Those two samples formed a strongly supported clade in all three analyses (PP 1.00, MLBS 100%, MPBS 100%). The clade emerged as sister to the clade consisting of D. elegans Benth., D. ferruginea Benth., D. glabra Sirich., D. laxiflora Benth., D. laotica Gagnep., D. laxiflora Benth., D. longiracemosa (sp. nov., in press), D. reticulata Craib, D. rubrocalyx Verdc., D. trifoliata and the new taxon sample from Nakhon Si Thammarat province, with high PP (0.97) but low MLBS (51%) and MPBS (< 50%) clade support. A slightly different topology emerged in the Maximum Parsimony (MP) analysis, where the three new species samples formed a moderately supported clade (MPBS 72%, see Suppl. material 2).

The new taxon is inserted as couplet 7 in a modified key to species of Derris in the Flora of Thailand (Sirichamorn 2020; 391–392).

6 Brachyblasts variable in shape and length, usually with more than 3 flowers. Standard less than 10 mm long, rarely with basal callosities

7 Mature leaflets with reddish midribs. Stamen filament sparsely hairy. Anther base with a tuft of hairs 18. D. rubricosta

7’ Mature leaflets without reddish midribs. Stamen filament glabrous. Anther base glabrous

8 Pods single winged or wingless

9 Leaflets hirsute to velvety underneath; stipels present 4. D. elegans

9’ Leaflets glabrous underneath; stipels usually absent

10 Leaflets 3.3–7.5 × 0.9–3.5 cm; petiolules 3–5 mm long 8. D. laotica

10’ Leaflets 3.5–16 × 1.5–8.5 cm; petiolules 5–10 mm long 17. D. trifoliata

8’ Pods two-winged

11 Pods velvety or sericeous

12 Leaflets slightly strigose to velvety below, apex rounded, obtuse, or cuspidate to short-acuminate. Pods with upper wing 4–10 mm wide, lower wing 4–7 mm wide. North-eastern Thailand 6. D. ferruginea

12’ Leaflets usually strigose to almost glabrous below, apex distinctly acuminate. Pods with upper wing 5–9 mm wide, lower wing 2–4 mm wide. Southern Thailand 13. D. pubipetala

11’ Pods mostly glabrous

13 Leaflets 9–11 per leaf, narrowly obovate, base narrowly cuneate to attenuate 11. D. monticola

13’ Leaflets 3–7(–9) per leaf, elliptic, ovate, or obovate, base cuneate to obtuse

14 Leaflets sometimes glaucous below. Lateral veins reaching leaflet margin 2. D. amoena

14’ Leaflets never glaucous below. Lateral veins not reaching leaflet margin but curving toward leaf apex, sometimes forming an intramarginal vein

15 Leaflets 3–5 per leaf. Terminal leaflet distinctly longer and wider than lateral ones. Calyx glabrous outside 7. D. glabra

15’ Leaflets 5–7 per leaf. Terminal leaflet slightly longer but not wider than lateral the others. Calyx thinly sericeous outside 12. D. pseudomarginata

6’ Brachyblasts slender with 2 or 3 flowers at apex. Standard usually more than 10 mm long, usually with basal callosities

The authors have declared that no competing interests exist.

No ethical statement was reported.

This research was financially supported by the Faculty of Science, Silpakorn University, grant number SCSU-STA-2564-02 and SRIF-JRG-2564-06.

Conceptualization: Punvarit Boonprajan and Yotsawate Sirichamorn. Fieldwork and specimen collection: Punvarit Boonprajan, Charan Leeratiwong, and Yotsawate Sirichamorn. Methodology and Experimental Work: Punvarit Boonprajan. Original Draft Preparation: Punvarit Boonprajan and Yotsawate Sirichamorn. Response to Reviewers and Manuscript Revision: Punvarit Boonprajan and Yotsawate Sirichamorn. Funding Acquisition: Punvarit Boonprajan and Yotsawate Sirichamorn

Punvarit Boonprajan https://orcid.org/0009-0006-7251-1928

Charan Leeratiwong https://orcid.org/0000-0002-0752-4275

Yotsawate Sirichamorn https://orcid.org/0000-0002-3026-3894

All of the data that support the findings of this study are available in the main text or Supplementary Information.