Terrestrial Callitriche specimens were collected from various locations, such as Hyogo (HN2), Osaka (OH1), Fukuoka (FF2), Ibaraki (IbTkb1 and IbTcu1), Tokyo (TKoi1) and Kanagawa (KKmz1, KHd1, KHsm1) Prefectures (Table 1). The plant strain from Hyogo Prefecture (HN2) has been previously used in developmental biology studies (Doll et al. 2021a, Doll et al. 2021b, Koga et al. 2021). Some specimens co-occurred with C. japonica, but they could be easily distinguished by their shoot forms and fruit morphologies (Katsuyama 2018). The localities of KKmz1 and KHsm1 in Kanagawa Prefecture matched with the information of voucher specimens for C. terrestris, M. Matsumoto NA0112762 (KPM) and C. Hasekura NA0112764 (KPM), respectively (Katsuyama and Kawasumi 1999).

To observe the flower and fruit morphologies, the collected plants were cultured in soil (Mizukusa-ichiban sand, GEX, Japan) in a growth chamber with a long-day condition (16 h light and 8 h dark) at 22˚C. Some specimens were plants that were inbred for 1–5 generations after collection. The voucher specimens were deposited to Herbarium of the Department of Botany, University of Tokyo (TI; Table 1). Herbarium collections of Kanagawa Prefectural Museum of Natural History (KPM) were also examined for both morphological and genetic investigations (Suppl. material 1).

To obtain genomic DNA for analysis, we extracted DNA from the shoot tips of herbarium specimens and living plants. For herbarium specimens, we first homogenised the dry samples by using TissueLyser II (Qiagen, Hilden, Germany) with zirconia beads and then performed DNA extraction by using the cetyl trimethyl ammonium bromide method, followed by purification with AMPure XP beads (Beckman Coulter, Brea, California, USA). For fresh samples, we used a DNeasy Plant Mini Kit (Qiagen). We amplified matK and rbcL regions of the extracted DNA by using PCR with specific primer sets: matK_390_F (5′-CGATCTATTCATTCAATATTTC-3′) and matK_1326R (5′-TCTAGCACACGAAAGTCGAAGT-3′) (Cuénoud et al. 2002); rbcL_5F (5′-ACCACAAACAGARACTAAAGC-3′) and rbcL_1210R (5′-AAGGRTGYCCTAAAGTTCCTCC-3′) (Les et al. 1993, Philbrick and Les 2000). In cases where PCR with the primer sets for fresh samples was unsuccessful for herbarium samples, we designed a new primer set that specifically targeted a short (199 bp) region of matK in the examined species: matK_497F (5′-ATTGGGTAAAAGATGCCTCTTCTTTGC-3′) and matK_695R (5′-TGAGAAGATTGGTTACGTAGAAAGACG-3′). We performed PCR with PrimeSTAR GXL Polymerase (Takara Bio, Shiga, Japan) or Ex Premier Polymerase (Takara Bio) and purified the PCR products with AMPure XP beads (Beckman Coulter). Sanger sequencing was used to sequence the purified products with the primers used for PCR. The resulting sequences were deposited in NCBI/DDBJ/EBI (accession numbers listed in Table 2 and Table 3).

To evaluate the phylogenetic positions of the samples, rbcL and matK sequences of Callitriche and some outgroup species were obtained from NCBI/DDBJ/EBI. The accession numbers are listed in Suppl. material 2. We used the data from Ito et al. (2017), which provided both rbcL and matK sequences from a broad range of Callitriche species. DNA alignment was conducted by MAFFT (Katoh and Standley 2013). The Maximum Likelihood tree was reconstructed using concatenated sequences of matK and rbcL by IQ-TREE 2 (Minh et al. 2020). The confidence values of each node were calculated using both ultrafast bootstrapping and Shimodaira–Hasegawa-like approximate likelihood ratio test (Hoang et al. 2017).

On the basis of recent descriptions (Lansdown et al. 2017a, Lansdown et al. 2017b, Hassemer and Lansdown 2018, Lansdown and Hassemer 2021), morphological traits of the collected plants, particularly flowers and fruits, were examined. The plants were raised in a growth chamber and harvested before observation. The specimens were divided into two types. One type, represented by specimens KKmz1, KHsm1 and IbTcu1, displayed traits consistent with those of C. terrestris, as described in an original report from Japan (Katsuyama and Kawasumi 1999), including short pedicels (≤ 1 mm) that point downwards when the fruits mature (Fig. 1B, C). It is common to find one female and one male flower in an axil, with a solitary female flower in the opposite axil (Fig. 1A, M). The fruit wings are narrow and not easily noticeable until the fruit is fully mature and dry (Fig. 1C–E). The styles, which persist when the fruit is fully mature, are typically shorter than the fruit height (Fig. 1A–C). Mature fruits are blackish, 0.5–0.7 mm in length and 0.6–0.8 mm in width (Fig. 1D, E). These traits also match the description of C. terrestris in recent taxonomic papers from Europe and South America (Hassemer and Lansdown 2018, Lansdown and Hassemer 2021). Callitriche turfosa Bertero ex Hegelmaier, which was previously referred to as subspecies Callitriche terrestris subsp. turfosa (Bertero ex Hegelm.) Bacigalupo and is reportedly very similar in morphology to C. terrestris, has a convex fruit surface (Lansdown and Hassemer 2021), whereas specimens KKmz1, KHsm1 and IbTcu1 have a relatively flat surface. Therefore, they were identified as C. terrestris rather than C. turfosa.

Figure 1.  

Flower and fruit morphologies of C. terrestris and C. deflexa. (A–E) Callitriche terrestris (KKmz1 and IbTcu1) and (F–L) C. deflexa (HN2 and FF2). A A node with two young female flowers and one male flower; B An immature fruit with a short deflexed pedicel; C A node with mature fruits; D, I Fully matured schizocarp abscised from the pedicel; E, J Mericarps; F A node with one young female flower and one male flower in an axil; G Young female flowers with elongated pedicels; H Mature fruit with an elongated pedicel; K A node with two female flowers and one male flower; L Mature fruits with short pedicels; M The proportion of flowering patterns of nodes. Scale bars: 1 mm.

The other type of specimens showed a similar shoot appearance, but had different flower and fruit morphologies. These plants often produced fruits with elongated pedicels (>1 mm) usually directed downwards (Fig. 1G, H) and short pedicels were sometimes observed (Fig. 1K, L). Male and female flowers were often found on the same axil, whereas the opposite axil often lacked a flower (Fig. 1F, H, M). Mature fruits had clear wings on all sides (Fig. 1I, J). Styles were either erect or weakly recurved and persisted on fully mature fruits and were often longer than the fruit height (Fig. 1G, L). Mature fruits were brown, measuring 0.5–0.7 mm in length and 0.7–0.9 mm in width (Fig. 1I). These characteristics were consistent with the keys for C. deflexa in recent literature (Lansdown et al. 2017a, Hassemer and Lansdown 2018, Lansdown and Hassemer 2021), leading us to conclude that these specimens were C. deflexa.

The shoots of both C. terrestris and C. deflexa are similar in appearance, but some differences become apparent when they are grown under the same conditions (Fig. 2). For example, the stem of C. deflexa tends to be longer and the mature leaves are rounder in shape when compared with C. terrestris (Fig. 2). These differences in shoot traits are stable when the plants are grown in a controlled environment, such as a growth chamber, but they may vary in different natural environments and are, thus, not suitable for rigid identification in the field. Previous studies have shown that the stomatal development of C. deflexa (denoted as C. terrestris in these papers) has some specific characteristics, such as amplifying division of stomatal lineage (Doll et al. 2021b) and stomata that differentiate on both sides of the leaf (amphistomy) (Doll et al. 2021a). Here, we confirmed that these characteristics were also shared by genuine C. terrestris (Fig. 2E, F).

Figure 2.  

Shoot morphologies of C. terrestris and C. deflexa. A A whole-plant image of C. terrestris; B A whole plant image of C. deflexa; C Images of shoots; D Leaf shapes; E Abaxial epidermis of C. terrestris leaf. Guard cells indicated by an arrowhead are surrounded by three cells of heterogeneous size, suggesting that this lineage has undergone amplifying division; F Stomatal index (stomatal number per epidermal cell) on the adaxial and abaxial sides of C. terrestris leaves (n = 5 each). Circles indicate the values of each leaf and crosses indicate mean values. Scale bars: A–C 1 cm, D 1 mm, E 100 µm.

Although the collected specimens showed differences in the traits described above, these traits were the keys to identifying C. deflexa and no traits could be used to identify C. terrestris. For example, flowering nodes in which a female flower is opposed by another female flower are also often observed in C. deflexa (Fig. 1G, K, M). Furthermore, the pedicel length of C. deflexa is variable and sometimes as short as that of C. terrestris (Fig. 1L). Thus, multiple traits of many flowers and fruits need to be comprehensively analysed for morphological identification.

Next, we analysed DNA sequences to evaluate the genetic divergence between the two specimen types. According to previous studies that extensively examined the phylogenetic relationships of the genus Callitriche (Ito et al. 2017), matK and rbcL sequences were sequenced from the specimens. All sequences from specimens identified as C. deflexa on the basis of morphology were identical to C. deflexa analysed by Ito et al. (2017) (in the paper, this specimen was denoted as Callitriche compressa N.E. Br. on the basis of the original identification, but the identification of the specimen was corrected to C. deflexa by Lansdown et al. (2017a)) (Tables 2, 3). In contrast, the rbcL sequence of a specimen assigned to C. deflexa by Philbrick and Les (2000) (Table 3) did not match the rbcL sequence of C. deflexa. As some traits of the specimen used by Philbrick and Les (2000) are inconsistent with the above-mentioned description, it is highly likely that the specimen is not C. deflexa, but a different species from the current taxonomic point of view. All three specimens morphologically identified as C. terrestris shared identical sequences, with 16 variant sites in matK and four variant sites in rbcL when compared with C. deflexa (Table 2 and Table 3). The rbcL sequence was identical to that of C. terrestris analysed by Philbrick and Les (2000).

As the phylogenetic position of C. terrestris was not examined even in previous broad phylogenetic analyses (Ito et al. 2017, Prančl et al. 2020), we performed a phylogenetic tree reconstruction by combining matK and rbcL sequences (Fig. 3). The result placed C. terrestris in clade VI, according to Ito et al. (2017). It is not a sister group to the morphologically similar C. turfosa and the results do not support the systematics that placed the two taxa as subspecies. However, the validity of the identification of specimens whose sequences were analysed as C. turfosa must be reviewed using the current taxonomic view. Callitriche deflexa belonged to clade VII, which is consistent with the results of previous studies (Ito et al. 2017, Lansdown et al. 2017a).

Figure 3.  

Phylogenetic tree of a plastid gene dataset (matK and rbcL). A Maximum Likelihood tree reconstructed by concatenated alignment of matK and rbcL DNA sequences. Node values represent Shimodaira–Hasegawa-like approximate likelihood ratio test (SH-aLRT) support (%)/ultrafast bootstrap (UFboot) support (%). Node values with either SH-aLRT support < 80% or UFboot support < 95% were omitted as they were not reasonably supported. Clade names (I to IX) were assigned according to Ito et al. (2017). Due to there being no sequence differences amongst the specimens obtained for this study, we used the sequence of only one specimen for each species.

We re-examined herbarium specimens from KPM, which has the largest collection of specimens previously identified as C. terrestris in Japan. We confirmed that most of the herbarium specimens were C. terrestris, but some specimens showed the characteristics of C. deflexa: extensive variation in pedicel length, sometimes reaching more than 3 mm; clear wings on fruit edges (Fig. 4). This type of plant has been recorded in Kanagawa, Mie, Osaka and Okinawa Prefectures. The oldest specimen was collected in 1999 in Okinawa Prefecture, indicating that C. deflexa has been misidentified as C. terrestris in Japan for at least 24 years.

Figure 4.  

Herbarium specimens of C. deflexa. A representative specimen with a magnified image of the female flowers with elongated pedicels. A M. Matsumoto NA0116625 (KPM) from Okinawa Pref.; B M. Matsumoto NA0203540 (KPM) from Mie Pref.; C M. Matsumoto NA0208310 (KPM) from Kanagawa Pref. Scale bars: 1 cm for the whole plant panels and 1 mm for the magnified panels.

We also tried to amplify and sequence plastid DNA from some herbarium specimens deposited in KPM (Suppl. material 1). Due to the difficulties in DNA extraction from herbarium specimens, we could obtain only short fragments of matK sequences from limited specimens (Suppl. material 1). Nevertheless, we confirmed clear genotypic differences that were consistent with the morphological differences. All examined voucher specimens of NA1004372, NA1104918 and NA0112762; KPM and additional recently-collected specimens (NA0162807 and NA0216556; KPM) had the same sequences as C. terrestris, which we confirmed using the freshly-collected specimens. We also succeeded in obtaining amplicons from three out of four specimens that are probably C. deflexa on the basis of the morphological traits and found that they have sequences distinct from those of C. terrestris. Except for one single nucleotide variant (SNV) site (position 562) found in the specimens from Mie and Kanagawa, the sequences were identical to that of C. deflexa collected in this study (Table 2). In conclusion, on the basis of both morphology and genetics, two clearly different types of specimens were found in this study: C. terrestris, as identified previously and C. deflexa.