Diastatea (Campanulaceae-Lobelioideae) is a genus of small, herbaceous annuals found mostly in Mexico. Differences in the number of recognized species in the historical treatments, and the lack of a phylogeny, suggested that a revision of the genus was necessary. Species boundaries and species level relationships were tested based on a sampling of nearly 50 individuals. Sequences of the chloroplast spacer regions, atpB-rbcL and ndhF-rpl32, as well as the ITS, were generated and then analyzed using maximum likelihood and Bayesian inference methods. The analyses of the combined dataset produced congruent topologies. Two strongly supported clades of Diastatea are newly recognized as subgenera in the taxonomic treatment. Diastatea Scheidw. subg. Diastatea and Diastatea subg. Angustifolia E.P.Johnson and T.J.Ayers can be distinguished morphologically by leaf shape and pubescence, hypanthium shape, and the presence or absence of a gibbous projection on the ventral side of the corolla. A morphological treatment of the genus corroborated by the molecular analysis revealed one new species, Diastatea aptera E.P.Johnson & T.J.Ayers, one new subspecies, Diastatea tenera subsp. durangensis E.P.Johnson & T.J.Ayers, and elevated an existing subspecies, Diastatea ciliata (McVaugh) E.P.Johnson & T.J.Ayers, to specific rank.

Fig. S1 - Combined ML Tree with Outgroups Expanded:

IQ-Tree version 1.6.8 (Nguyen et al. 2015, Trifinopoulos et al. 2016) was used to conduct the phylogenetic analysis on a concatenated dataset that included atpB-rbcL, ndhF-rpl32, and the ITS. The alignment was partitioned as follows: ITS1: 1–397; 5.8S: 398–559; ITS2: 560–903; ndhF-rpl32: 904–2236; atpB-rbcL: 2237–3010. BIC scores were used to determine that IQ-Tree’s -spp partition model was the best fit for our data, allowing each partition its own evolutionary rate (Chernomor et al. 2016). ModelFinder (Kalyaanamoorthy et al. 2017) was used to find the best fit substitution model for each of five partitions and are as follows: ITS1 and ITS2: GTR+F+I+G4; 5.8S: JC+R2; ndhF-rpl32: TVM+F+G4; atpB-rbcL: TVM+F. Maximum likelihood (ML) analyses were run using ultrafast bootstrapping (UFBoot) (Hoang et al. 2017) for 1000 replicates. Phylogenetic trees were visualized using FigTree (v1.4.4). A Bayesian inference analysis was performed using MrBayes 3.2.2 (Ronquist et al. 2012). Two independent runs of 30,000,000 MCMC generations with four chains (one hot, three cold) using a chain temperature of 0.2 was run on the combined dataset. The trees were sampled every 100 generations and the first 25% of the trees were discarded (Hall 2011). Resulting posterior probabilities were appended to the ML tree.

Fig. S2 - ITS ML Tree:

IQ-Tree version 1.6.8 (Nguyen et al. 2015, Trifinopoulos et al. 2016) was used to conduct the phylogenetic analysis on the ITS dataset. The alignment was partitioned as follows: ITS1: 1–397; 5.8S: 398–559; ITS2: 560–903. BIC scores were used to determine that IQ-Tree’s -spp partition model was the best fit for our data, allowing each partition its own evolutionary rate (Chernomor et al. 2016). ModelFinder (Kalyaanamoorthy et al. 2017) was used to find the best fit substitution model for each of the partitions and are as follows: ITS1 and ITS2: GTR+F+I+G4; 5.8S: JC+R2. Maximum likelihood (ML) analyses were run using ultrafast bootstrapping (UFBoot) (Hoang et al. 2017) for 1000 replicates.

Fig. S3 - Combined chloroplast ML Tree:

IQ-Tree version 1.6.8 (Nguyen et al. 2015, Trifinopoulos et al. 2016) was used to conduct the phylogenetic analysis on a concatenated chloroplast dataset that included atpB-rbcL and ndhF-rpl32. The alignment was partitioned as follows: ndhF-rpl32: 1–1332; atpB-rbcL: 1333–2106. BIC scores were used to determine that IQ-Tree’s -spp partition model was the best fit for our data, allowing each partition its own evolutionary rate (Chernomor et al. 2016). ModelFinder (Kalyaanamoorthy et al. 2017) was used to find the best fit substitution model for each partition and are as follows: ndhF-rpl32: TVM+F+G4; atpB-rbcL: TVM+F. Maximum likelihood (ML) analyses were run using ultrafast bootstrapping (UFBoot) (Hoang et al. 2017) for 1000 replicates. 

Fig. S4 - SEM Images of Seeds:

Seeds were removed from dried specimens, mounted on stubs and sputter-coated for 30 s with gold palladium. The images were taken at the NAU Imaging and Histology Core Facility on a Zeiss Supra 40 VP Scanning Electron Microscope.

Fig. S5 - Chromosome Images:

Chromosome counts were made from germinated seeds with 2–3 mm long root tips treated with 0.05% colchicine for 90 min followed by fixation in formalin-acetic acid-alcohol (1:1:3) at room temperature for 2 hrs. The root tips were then stored at 4°C for 1–4 hrs before processing. After fixation the tips were treated with 60°C 1M HCl for 2–3 min to loosen tissue, rinsed in distilled water, and then stained for 1 hr in 1% aceto-orcein before seed coat was removed and root tip mounted on slide in 45% acetic acid-glycerol (9:1) and flattened. Photos were taken at 1000x using MagnaFire 2.0 imaging software on a Leica DMLD compound microscope.

Table S1 - Samples Per Locus: The number of samples representing each individual locus both from Diastatea (ingroup) and Palmerella, Lobelia, Campanula, and Trachelium (outgroup), before concatenation of the dataset.

Alignment of Total Concatenated Dataset:

Sequences were assembled and edited using Lasergene-SeqMan Pro (DNASTAR, Madison, Wisconsin). They were aligned in MAFFT (Katoh and Standley 2013) using the L-INS-i refinement method for atpB-rbcL and ndhF-rpl32 regions, and the Q-INS-i method for the ITS region. Q-INS-i was chosen to align the ITS region due to the presence of secondary structures in that region (Katoh and Toh 2008). The datasets were concatenated using SeaView (Gouy et al. 2010). The alignment was partitioned as follows: ITS1: 1–397; 5.8S: 398–559; ITS2: 560–903; ndhF-rpl32: 904–2236; atpB-rbcL: 2237–3010. 

Alignment of ITS Dataset: 

Sequences were assembled and edited using Lasergene-SeqMan Pro (DNASTAR, Madison, Wisconsin). They were aligned in MAFFT (Katoh and Standley 2013) using the Q-INS-i method. Q-INS-i was chosen to align the ITS region due to the presence of secondary structures in that region (Katoh and Toh 2008). The alignment was partitioned as follows: ITS1: 1–397; 5.8S: 398–559; ITS2: 560–903.

Alignment of cpDNA Dataset: 

Sequences were assembled and edited using Lasergene-SeqMan Pro (DNASTAR, Madison, Wisconsin). They were aligned in MAFFT (Katoh and Standley 2013) using the L-INS-i refinement method for both the atpB-rbcL and ndhF-rpl32 regions. The datasets were concatenated using SeaView (Gouy et al. 2010). The alignment was partitioned as follows: ndhF-rpl32: 1–1332; atpB-rbcL: 1333–2106.

ReadMe file can be found with data download.

Figure Captions for Supplemental Figures and Table:

Fig. S1.  Maximum likelihood tree for the concatenated dataset with outgroup nodes expanded to show specimen information. Numbers above branches represent bootstrap values followed by posterior probabilities. Posterior probabilities below 0.95 are shown as dashes. 

Fig. S2. Maximum likelihood tree for the ITS dataset. Numbers above branches represent bootstrap values. 

Fig. S3. Maximum likelihood tree for the concatenated chloroplast dataset which includes ndhF-rpl32 and atpB-rbcL. Numbers above branches represent bootstrap values. 

Fig. S4. Scanning electron micrograph images of seed coat surface in Diastatea. A. D. aptera (Ayers et al. 1767).B. D. tenera subsp. durangensis (Jiménez 186). C. D. ciliata (Reveal et al. 4183). D. D. micrantha (Ayers et al. 1766)with striations obvious due to incomplete filling of seed.

Fig. S5. Photos of chromosomes obtained from root tips. A. Diastatea aptera, 2n = 14. B. D. virgata, 2n = 14. C. D. micrantha, 2n =14.

Table S1. Number of specimens out of total available where data were obtained. 

Alignment of Total Concatenated Dataset: Sequences obtained from atpB-rbcL and ndhF-rpl32 regions of the chloroplast, and the nrITS region, were concatenated into one alignment for use in the phylogenetic analysis. 

Alignment of ITS Dataset: Sequences obtained from the nrITS region were aligned to perform a phylogenetic analysis. 

Alignment of Concatenated Chloroplast Dataset: Sequences obtained from atpB-rbcL and ndhF-rpl32 regions of the chloroplast were concatenated and aligned to perform a phylogenetic analysis.