The material was collected as part of a larger study focused on the biodiversity of JTNP biological soil crusts (Pietrasiak et al. 2011; Flechtner et al. 2013). The sample WJT24 was collected at coordinates 33°45’47”N, 115°47’46”W on 13 Jun. 2006. The site was located north of Cottonwood, CA, U.S.A. and described as having sandy, gravelly soil derived from a granitic outcrop west of Eagle Mountains. The site was vegetated with desert plants typical for the Colorado Desert and also exhibited well developed biological soil crusts. A composite sample of the topsoil (0–1 cm soil depth) was collected systematically every 5 m along a 50 m transect (Pietrasiak et al. 2011).

A 1 g subsample of the composite soil was rehydrated in 100 ml Bold’s Basal Medium (BBM) (Bischoff and Bold 1963) and dilution-plated on an agar plate, and algal colonies were picked as described in Flechtner et al. (1998, 2013). The culture, labelled WJT24VFNP31, was maintained on agarized BBM under 16:8 light:dark cycle at 18°C. Prior to lipid extraction, the cultures were placed in a growth rig with 40 W fluorescent lights at 15°C and 2000 to 4000 lux, 12:12 light:dark cycle, for three weeks or until reaching sufficient biomass for lipid extraction.

The morphology of the strain WJT24VFNP31 was first assessed using light microscopy to survey and measure cell size, shape, and internal structures visible under a light microscope. The Olympus BH-2 photomicroscope with Nomarski DIC optics was used for observations and photographs were obtained using an Olympus DP25 camera (Olympus, Center Valley, CA, USA). Cells from 2–4-week-old cultures were observed. To determine the number of nuclei in young and mature vegetative cells, the cells were first left overnight in a permeation buffer containing 0.1% Triton X and 0.05% Tween 20. Cells were washed three times with phosphate buffer saline (PBS), then fixated using 4% formalin for at least 2 hours. Samples were washed and combined with DAPI (diluted from stock 1:1000) for 7–10 minutes. Samples were washed again five times with PBS in order to remove any residual DAPI that would cause background staining. Cells were mounted onto a microscope slide in distilled H₂O and were visualized using a Nikon Eclipse 90i microscope (Nikon Instruments Inc., Melville, NY, USA) with an Andor camera (Andor Technology Ltd., Belfast, Northern Ireland).

DNA was extracted from WJT24VFNP31 using MO BIO PowerPlant Pro DNA Isolation Kit (Qiagen, Germantown, MD, USA) following the manufacturer’s protocol with the exception of using distilled water instead of Buffer 7. The total genomic DNA was sequenced using Illumina MiSeq (Illumina Inc., San Diego, CA, USA) technology (150 bp paired reads). Two million reads (2,078,788) were obtained in total from two separate MiSeq runs.

The sequences of the 18S and 28S genes were initially obtained using PCR and Sanger sequencing, following the protocols described in Saber et al. (2018) and Shoup and Lewis (2003). The Sanger results were consistent with the Illumina data and are not discussed in detail in this article. However, the fact that they were obtained from a different DNA extraction and matched the high-throughput sequencing data helps rule out potential sequencing artifacts and contamination. A low-coverage region in the Illumina assembly of the rRNA genes was well covered by five Sanger reads, which further adds confidence to our results.

The Illumina reads were paired, trimmed and assembled de-novo using Geneious v.6 (Biomatters Inc., Boston, MA, USA; details in Fučíková et al. 2014c, 2016) and inspected by eye. Chloroplast tRNA and rRNA genes and intron conserved elements were identified using the RNAweasel tool (Lang et al. 2007, http://megasun.bch.umontreal.ca/cgi-bin/RNAweasel/RNAweaselInterface.pl), and their boundaries were edited based on alignment with other Chlorophycean sequences. Conserved protein-coding genes were identified by reference assembly of published plastid genes (Ankyra judayi NC_029735.1) to the WJT24VFNP31 plastid contig. Highly variable genes (e.g. ycf1, ftsH) were initially identified by the ORF prediction tool in Geneious and subsequent BLAST searches. All gene boundaries were verified based on alignment with the data set of Fučíková et al. (2019). The genome map (Supplementary material 1) was drawn using OGDRAW v.1.3.1 (Greiner et al. 2019).

After the full contigs for rDNA and the chloroplast genome were obtained, a reference assembly was performed in Geneious to visually inspect coverage and its distribution along the contigs. Stringent conditions were applied in the reference assembly, allowing no more than 1% mismatches and 1% gaps, with a word size of 20 nucleotides. Areas of low coverage were inspected by eye to rule out mis-assemblies.

The NCBI nucleotide database was searched for closest matches for the rRNA gene sequences of WJT24VFNP31 using BLAST (Altschul et al. 1990). The results, in combination with existing 18S and 28S alignments, were used to determine the boundaries of the genes and internal and external transcribed spacers in the rDNA.

The secondary structure of the 18S gene was visualized using the Chlamydomonas reinhardtii P.A.Dangeard model from https://bioinformatics.psb.ugent.be/webtools/rRNA/secmodel/Crei_SSU.html (Van de Peer et al. 2000). The WJT24VFNP31 18S sequence was manually mapped onto this structure and from the mapping, compensatory base changes (CBCs) in paired regions were determined, as well as other substitutions in the nucleotide sequence and changes in the predicted secondary structure. Structure was verified using Mfold (Zuker 2003) in stems and loops with many nucleotide substitutions.

The secondary structure of the internal transcribed spacer 2 (ITS2) was predicted using the ITS2 Database (Koetschan et al. 2010, http://its2.bioapps.biozentrum.uni-wuerzburg.de), implementing the Identity matrix (because the ITS2 PAM 50 model yielded no results) with otherwise default settings for template-based modelling. Spermatozopsis exsultans Korshikov and Ankyra judayi (G.M.Smith) Fott were selected as templates after the initial search for templates yielded no results. Alternative structures of the individual helices were explored using Mfold (Zuker 2003).

Nucleotide sequences of 59 chloroplast genes were aligned with the existing data from Fučíková et al. (2019), concatenated (69 taxa, 33,483 nucleotide positions) and analysed using MrBayes v.3.2 (Ronquist et al. 2012), applying the GTR+I+Γ model over 6,250,000 generations and two MCMC chains in each of two parallel runs. Sites were partitioned by codon position across the entire data set. The first 20% of each run was discarded as burn-in. An analysis of the translated amino acid data set was also conducted, applying a mixed AA model (averaging over ten amino acid models) in MrBayes. The OCC clade was used to root the tree, consistent with previously published literature (Fučíková et al. 2019 and references within).

The evolution of the 18S GC content was reconstructed using fastAnc in the R package phytools v.1.2 (Revell 2012). The trait (18S GC content) was mapped onto the chloroplast nucleotide phylogeny. Additionally, the GC content of the coding chloroplast regions (including exons of protein-coding genes, rRNA and tRNA genes but not intronic or other hypothetical or unidentified ORFs) was mapped onto the chloroplast phylogeny.

To examine whether GC content in the 18S gene is correlated with the temperature of the locality of origin, we first conducted a Pearson correlation test in base R on the uncorrected temperature and 18S GC data, which were available for all but two of the Chlorophycean species included in our phylogenetic data set. We then regressed the 18S GC and average temperature of the warmest month at the locality of origin and corrected for the effect of phylogenetic relatedness using the phylogenetically independent contrasts (PIC) method (Felsenstein 1985) in phytools. The complete annotated R code and the underlying data set is available at https://github.com/KarolinaFucikova/GC_mapping, with a permanent DOI https://doi.org/10.5281/zenodo.8011298.

Three replicate cultures were grown for three weeks or longer until sufficient biomass was produced for extraction. A minimum of 15 mg of wet algal biomass was harvested and suspended in gas-chromatography vials with methanol. The open vials were placed in a glass desiccator overnight with Drierite desiccant. After weighing the vials for their dry mass the next day (optimum dry mass being above 5 mg), transesterification was conducted using 200 μl of 2:1 CHCl₃:methanol, 300 μl of 0.6 M HCl:methanol, and 25 μl of C13:0 methyl ester internal standard. The vials were tightly sealed with parafilm to prevent evaporation. After heating the samples at 85°C for an hour in a water bath, the vials were cooled and 1 ml of hexane was added to each vial. The vials were then vortexed, allowed to rest for 1–4 hours, then 500 μl of the top layer was removed and placed in a new vial (Van Wychen et al. 2015).

All lipids were quantitated by a Varian CP-3800 Gas Chromatograph using a Saturn 2000 Mass Spectrometry detector (Varian, Palo Alto, CA, USA). The gas chromatography analysis was conducted using a Phenomenex ZB-WAX capillary column (Phenomenex, Torrance, CA, USA) with helium as carrier gas with the following parameters: helium flow 1 mL/min, temperature program: 100°C for 1 min, 25°C/min heating up to 200°C, hold for 1 min, 5°C/min heating up to 250°C, hold for 7 min. For mass spectrometry, a two-minute solvent delay was followed by electron impact ionization and detection of masses from 45–400 m/z. Quantitation was accomplished with the Total Ion Chromatogram for all lipids except for C22:5 (quant ions 79 + 91 m/z) and C24:0 (quant ion 382 m/z) because their peaks overlapped.

The lipid content, composition, and saturation indexes were calculated in MS Excel. The percent total lipid content of the samples was calculated using the initial dry mass of the algae after they were desiccated. Based on total lipid content, estimated percent composition of the various fatty acids were determined.

Cells were solitary, spherical or nearly so (Fig. 1A–E). Young cells in two-week and one-month-old cultures were 4–6 μm in diameter. Mature cells were up to ca 12 μm in diameter. Older cells had visible cytoplasmic granules/deposits that were dark reddish-brown and occurred outside of the plastids (Fig. 1E, highlighted with arrowheads). Young cells contained at first a single cup-shaped chloroplast, which then became lobed and broke up into multiple parietal plastids as the cells aged. No pyrenoid was observed. Cell wall did not appear to thicken appreciably with age. Reproduction via four autospores was observed (Fig. 1C), but additional sporangia-like cells were seen, which were large and irregular in shape (Fig. 1F, G). Senescent cultures of 6+ months were deep orange in colour. DAPI staining demonstrated that young cells have a single nucleus (Fig. 2A, B) and maturing cells can have two or more (Fig. 2C, D). Chloroplast DNA stains with DAPI as well, which makes the precise determination of the number of nuclei difficult.

Figure 1. 

Light micrographs capturing the morphology of Johansenicoccus eremophilus gen. et sp. nov. A, B. Young and mature vegetative cells. C. Vegetative cells and autospore formation. D. Mature vegetative cells with clear multiple chloroplasts and extra-plastidic inclusions. E. Detail of vegetative cell; arrows point to unidentified extra-plastidic inclusions. F, G. Irregularly shaped cells, possibly in the process of spore production. Scale bars represent 10 µm, second scale bar pertains to E–G.

Figure 2. 

Micrographs of DAPI-stained cells of Johansenicoccus eremophilus gen. et sp. nov. A, B. Young vegetative cells. C, D. Mature vegetative cells. A, C. Bright field. B, D. Fluorescence micrographs corresponding to A and C, respectively.

The phylogenetic relationships among Chlorophyceae, derived from the sequences of 59 chloroplast genes, were largely consistent with the phylogenies of Fučíková et al. (2019), and similar discrepancies were found between the nucleotide and amino acid topologies. In both analyses (nucleotide and amino acid), the position of the strain WJT24VFNP31 was sister to Spermatozopsis similis H.R.Preisig & M.Melkonian with absolute support, and the two taxa formed a clade sister to the family Sphaeropleaceae (Ankyra + Atractomorpha), also with absolute support of BPP = 1.0 (Fig. 3). The position of this four-taxon clade with respect to the main SV lineages (Volvocales, Scenedesminia, Treubarinia, Microsporaceae, and several incertae sedis taxa) differs between analyses, as it did in Fučíková et al. (2019). Specifically, in the amino acid analysis the Sphaeropleaceae + Spermatozopsis + WJT24VFNP31 appeared sister to Volvocales, consistently with Fučíková et al. (2019).

Figure 3. 

Bayesian consensus phylogeny based on an analysis of 59 concatenated protein-coding chloroplast genes, with the nucleotide data partitioned by codon position. Numbers on branches represent Bayesian Posterior Probability. Solid branches represent relationships that were also recovered in the corresponding Bayesian analysis of the amino acid data. Dashed branches indicate relationships that were not recovered in the amino acid analysis. Scale bar represents the estimated number of nucleotide substitutions per site.

Under the stringent reference assembly conditions, which gave a conservative (low) estimate, the chloroplast genome coverage ranged from 12 to 139 (mean 75.9). The chloroplast genome was assembled in two equally probable structures: either in a single circular contig with two direct (not inverted) repeats containing ribosomal genes, or in two circular contigs containing one of the repeats each. Without long-read sequencing data, it is not possible to determine the true structure of the genome, and it is also possible that the genome exists in both forms. Nevertheless, all expected genes were detected in the contig(s) and upon visual inspection there were no areas of suspected mis-assembly. Therefore, we concluded that the chloroplast genome sequence is likely complete, despite the uncertainty about its structure.

The chloroplast genome of WJT24VFNP31 (GenBank OQ849777) is 235,854 bp in size. Of the two possible structural configurations we present the single circular molecule with two direct repeats. The repeats were confirmed by their twofold read coverage compared to the rest of the plastome, and contain ribosomal RNA genes, two tRNA genes (trnA-UGC and trnI-GAU), and part of the petA gene. No introns were found in the repeats (Supplementary material 1).

All genes expected in Sphaeropleales and Chlorophyceae incertae sedis were detected in the WJT24VFNP31 plastome. In addition, the uncommon (in Chlorophyceae) trnR-UCG gene was present, as was a trnT-AGU gene. The genome also contained a trans-spliced psaA gene, split into two exons at position 269, as is common in the SV clade, but contiguous at position 89, a condition only reported from a handful of SV taxa. A full account of gene and intron content across analysed chlorophyceans is presented in Supplementary material 2.

The full nuclear ribosomal gene set is deposited in GenBank under accession number OQ849776. The top BLAST match for the 18S gene was 84.1% similar (Heterochlamydomonas inaequalis Ed.R.Cox & T.R.Deason), and others in the 83–84% range included other Volvocales, Sphaeropleales (e.g., Mychonastes P.D.Simpson & S.D.Van Valkenburg), and even Trebouxiophyceae (Neocystis Hindák), further underscoring the uniqueness and uncertain phylogenetic position of WJT24VFNP31 and its rDNA sequence.

The GC content in the 18S gene was 54.9%, which is 3% higher than the next highest GC in the data set, which is in Jenufa perforata Němcová, M.Eliáš, Škaloud & Neustupa (51.9%). The third highest was Borodinellopsis texensis Dykstra (51.1%), illustrating that high GC content is not exactly clade-specific, although high 18S GC seems to be more common among incertae sedis taxa (lineages outside of OCC, Volvocales, and Scenedesminia, Fig. 4). The full compilation of GC content in the rRNA genes and chloroplast coding regions of Chlorophyceae is available in our GitHub repository (https://doi.org/10.5281/zenodo.8011298). The 18S secondary structure, compared to that of Chlamydomonas reinhardtii, seemed largely maintained despite the numerous substitutions and elevated GC content (Supplementary materials 3, 4). Many of the substitutions and some structural changes were unique to WJT24VFNP31.

The ITS2 structure was significantly different from those currently deposited in the ITS2 Database. The Database tools readily detected the proximal stem and the borders of the spacer, however, the initial search for suitable templates did not yield any results. Two of the putatively closest relatives of the strain WJT24VFNP31 were therefore selected as templates for structural folding: Spermatozopsis (represented in the database only by S. exsultans), and Ankyra judayi. No high-quality model could be derived from the Spermatozopsis template. Percentages of helix transfer were 64, 35, 50, and 22 for the four helices, respectively. Mapping was also performed onto the Ankyra template and yielded helix transfer percentages of 66, 37, 60, and 50. Ankyra judayi was thus a better fit as a template for WJT24VFNP31. The predicted ITS2 structure modelled onto the Ankyra template contained many extra bulges (unpaired regions), and therefore alternative structures of individual helices were explored. Mfold (Zuker 2003) yielded some structures visually more similar to the helices of the templates and other algae (Supplementary material 5), but the full secondary structure of WJT24VFNP31’s ITS2, as well as its functionality remains unresolved.

Figure 4. 

Evolution of GC content in the 18S nuclear ribosomal gene mapped onto an independently derived phylogeny (based on data from 59 chloroplast genes). Darker colour signifies higher 18S GC content.

Figure 4 shows how high the 18S GC content is compared to other Chlorophyceae, and that this feature evolved along the long branch subtending WJT24VFNP31 from an ancestor with low to moderate GC content. The sister taxon to WJT24VFNP31 is Spermatozopsis similis, which has a low GC content in its 18S gene (47.8%). The figure also shows the uniquely high GC content in Mychonastes among the otherwise low-GC Scenedesminia. This could explain the previous studies’ difficulty with phylogenetically placing this genus. The 28S gene generally follows the trend of the 18S gene across Chlorophyceae – taxa with high 18S GC also have a high 28S GC and vice versa. WJT24VFNP31 has extremely high 28S GC (58.1%) compared to all other taxa examined.

In contrast to the nuclear ribosomal GC content, the GC content of the chloroplast coding regions in WJT24VFNP31 is 38.5%, which is not remarkably high compared to other Chlorophyceae. Both species of Mychonastes have over 40% GC, and the volvocalean Stephanosphaera pluvialis Cohn has the highest chloroplast coding GC of 44%. Supplementary material 6 shows both mappings in colour and our GitHub repository contains the full data set for further exploration (https://doi.org/10.5281/zenodo.8011298).

The Pearson correlation test yielded a significant positive correlation between the 18S GC content and maximum temperature at the locality of origin (p = 0.0014), and the corresponding plot showed the strain WJT24VFNP31 as an outlier with extremely high 18S GC, even given the warm, desert climate at its locality of origin (Fig. 5). When WJT24VFNP31 was excluded from the data set, the Pearson correlation was still significantly positive (p = 0.0054, r = 0.3607). After correction for the effect of phylogeny (WJT24VFNP31 included), the 18S GC content still correlated positively and significantly with temperature (p = 0.0004, adjusted R² = 0.1838, slope of 0.0842, Supplementary material 7). No significant correlation with temperature was detected in the GC of the chloroplast coding regions (https://doi.org/10.5281/zenodo.8011298).

Figure 5. 

Regression analysis of the relationship between temperature at the site of origin and GC content in the 18S gene across 59 species of green algae. Temperature refers to the average temperature in the warmest month at or near the site of the species/strain’s collection. Algae from different habitats (aquatic, snow, soil) are represented with differently shaped symbols. The strain WJT24VFNP31 (Johansenicoccus eremophilus gen. et sp. nov.) is labelled. Grey line represents the fitted regression line.

The chloroplast ribosomal genes were not investigated in depth, as data for several taxa are not available due to the partial nature of the published genome sequences. In most algae in our data set, the GC content in chloroplast ribosomal genes was 10–20% higher than the general coding GC in the chloroplast. The strain WJT24VFNP31 and Bracteacoccus giganteus H.W.Bischoff & Bold had the highest GC content in the rrs gene: 53.7%. All three included species of Bracteacoccus had rrs GC over 53%, while most other taxa had rrs GC < 52%. There was a weak positive correlation between temperature and rrs GC: r = 0.2733 and p = 0.0499. The data set is available in our GitHub repository for further exploration (https://doi.org/10.5281/zenodo.8011298).

The most abundant fatty acid in the strain WJT24VFNP31 was alpha-linolenic acid (C 18:3 n-3, 56.0% on average), followed by palmitic (C16:0, 24.7%) and vaccenic (C18:1 trans-11, 9.7%) acids (Supplementary material 8). Oleic (C18:1 cis-9), linoleic (C18:2), and gamma-linolenic (C18:3 n-6) acids were also present in detectable amounts. On average, 73% of the fatty acids detected by the GC-MS method were unsaturated.

The nuclear ribosomal region of Johansenicoccus eremophilus contains the 18S gene, 5.8S gene, 28S gene, and two internal transcribed spacers in the expected order, but the rRNA genes are extremely GC-rich compared to other Chlorophyceae (Figs 4, 5). The maintenance of 18S secondary structure (Supplementary materials 3, 4) suggests that the unusual 18S sequence codes for a functional ribosomal subunit. This is consistent with the fact that in our genomic data the ‘abnormal’ ribosomal region is abundant and therefore unlikely to be a deteriorating pseudogene. Additional reference assemblies have hinted at the presence of another low-copy rDNA region (also very divergent from other GenBank sequences), but it was represented only by a few reads and we were unable to assemble it fully. We have confirmed the unusual 18S (main copy) with PCR and Sanger sequencing to rule out chimeric assemblies in the high-throughput data. The PCR primers did not amplify any other 18S copy that could be a putative paralog. Therefore, while it is possible that other, paralogous copies of the gene and spacer set exist in the genome of J. eremophilus, all of our evidence currently suggests that the unusual rDNA genes discussed here code for the components of functional ribosomes in the species.

One previously explored hypothesis is that high GC content may be an adaptation to hot climates because of the higher thermal stability of the G-C pairing compared to the A-T/A-U pairing. While there does not seem to be a correlation between GC content and temperature for whole genomes or the (more or less) freely evolving third codon positions, the pattern has been consistently recovered for structural RNA in prokaryotes (Hurst and Merchant 2001), where species living in higher temperatures tend to have higher GC content in their ribosomal genes. This could be due to the high selective pressure to maintain the ability to translate proteins, which is contingent of a functional structure of ribosomes and therefore the secondary structure of rRNA after transcription. Such an issue would not necessarily arise in chromatin (DNA). Somewhat contrastingly, however, Šmarda et al. (2014) reported higher genomic GC content as indicative of adaptation to cold and drought stress in monocots. The drivers of GC content are therefore likely quite complex.

In our study we uncovered a significant positive correlation between 18S GC content and warmest-month temperature at the locality of origin (Fig. 5), consistent with Hurst and Merchant (2001). However, temperature or desert habitat alone do not explain the unusual and GC-rich DNA sequence of the rRNA genes in J. eremophilus. Other desert algae, such as Rotundella rotunda (also from JTNP), do not have nearly such extreme GC levels in their ribosomal genes. Therefore, our current data only provide tentative support for the thermal adaptation hypothesis, and more research is needed to properly explain the unusual nature of the J. eremophilus nuclear ribosomal loci.

This uniqueness in ribosomal genes all but precludes phylogenetic analysis using 18S or 28S, which are otherwise very commonly used in algal systematics. Using secondary structure to guide the alignment and analysis, as recommended by e.g. Wolf et al. (2008) and Czech and Wolf (2020), does not remedy the extremely long branch subtending J. eremophilus in any attempted 18S-based phylogenies (not shown) and the placement of the species in such phylogenies does not receive high support. ITS2, another commonly used marker, faces similar problems (being very divergent from other green algal taxa, Supplementary material 5), though this marker is more appropriate for genus- and species-level resolution and we currently only have one strain and species of Johansenicoccus. If additional representatives of the genus are found in the future, ITS2 may still be of use for species-level taxonomy.

The chloroplast genome of Johansenicoccus eremophilus is nearly 236 kbp, which makes it the largest among its closest relatives, Spermatozopsis (135 kbp, Fučíková et al. 2019) and Sphaeropleaceae (157 kbp in Ankyra, Fučíková et al. 2016). The gene content in the plastome of J. eremophilus is nearly perfectly consistent with its phylogenetic position in the Chlorophyceae (details in Supplementary materials 1, 2). Like its closest relatives, J. eremophilus has very few introns in its plastome. Like other non-OCC chlorophyceans, J. eremophilus lacks the gene psaM and has a trans-spliced psaA. The two-exon configuration of psaA (as opposed to a three-exon configuration) is not common, and is known from four divergent non-OCC chlorophyceans, J. eremophilus, Golenkinia longispicula Hegewald & Schnepf (incertae sedis), Chloromonas radiata (T.R.Deason & Bold) T.Pröschold, B.Marin, U.W.Schlösser & M.Melkonian (Volvocales), and Tetradesmus obliquus (Turpin) M.J.Wynne (Scenedesminia), suggesting that the two-exon state may be convergently evolved.

Unlike the unsurprising gene content, the structure of the J. eremophilus plastome is noteworthy, being either a single circular molecule with two large direct repeats containing blocks of rRNA genes (Supplementary material 1), or two smaller circular molecules with one rRNA block present in each circle. While our data cannot confidently resolve which of the alternative configurations is the true one, both are unusual in Chlorophyceae. Multiple chloroplast chromosomes have been reported in the OCC taxon Koshicola spirodelophila Shin Watanabe, Fučíková & L.A.Lewis (Chaetopeltidales, Watanabe et al. 2016) and outside of green algae it is known in dinoflagellates (Zhang et al. 1999). A similar situation is known for some mitochondrial genomes as well (Smith et al. 2010). In contrast, the single-chromosome configuration would be unique among green algae because of the orientation of the rRNA gene-containing repeats. Most plants and green algae contain two copies of the ribosomal RNA gene set too, but it is positioned on an inverted repeat (IR) segment in their chloroplast genome. In the Chlorophyceae, species in Chaetophorales and Chaetopeltidales lack an IR and only possess a single copy of each of their rRNA genes. A genome with “un-inverted” repeats, however, is not known from any green algal lineage, though this configuration was recently reported in two species of the vascular plant genus Selaginella P.Beauv. (Xu et al. 2018; Mower et al. 2019), and outside of green plants also in the heterokont alga Rhizochromulina marina D.J.Hibberd & Chrétiennot-Dinet (Han et al. 2019). A similar orientation (tandem, or direct repeats positioned immediately next to each other on the genome) was also found in several euglenophytes (Wiegert et al. 2012; Maciszewski et al. 2022). Further, Wang and Lanfear (2019) reported that in most plants (though not Selaginella tamariscina (P.Beauv.) Spring) the plastome exists in two equally frequent versions, which have the short single copy region oriented in the opposite directions. Given these new discoveries, the unusual plastome structure of J. eremophilus is perhaps less surprising. More in-depth investigations could include long-read sequencing data and the statistical approach of Wang and Lanfear (2019).

Microalgal lipid production is subject to a growing body of literature, as microalgal cultivation promises advances in biofuels, feedstock, and nutritional supplements. Green algal biomass tends to be rich in 16- and 18-carbon fatty acids, and often contains significant amounts of unsaturated lipids. Our results are consistent with this body of literature. For example, Tetradesmus obliquus, a commonly used algal model, was found to contain similar fatty acids as found in our study (Choi et al. 1987), though in a different order of abundance: linoleic (only found in small amounts in our study), palmitic (second most-abundant and the only significantly represented saturated fatty acid in both studies) and alpha-linolenic (most abundant in our study). Unsaturated fatty acid composition was higher in Tetradesmus G.M.Smith than in J. eremophilus (80% vs 73%). Saber et al. (2018) analysed the lipid composition of two Sphaeroplealean desert algae and cited fatty-acid composition roughly in agreement with Hong et al. (2012), with the abundant alpha-linolenic and palmitic acids being consistent with our study.

The evolutionary significance of lipid composition has been well studied in vascular plants. Such studies have shown that plant cells can desaturate their membrane lipids in response to lower temperature and vice versa. The mechanisms of the maintenance of membrane fluidity are complex especially in environments where temperature fluctuates rapidly on a daily basis, as can be true in alpine environments and deserts (Zheng et al. 2011 and references within). Therefore, a comparative study would be needed to determine whether the lipid content and composition in J. eremophilus reflects any adaptations to its desert lifestyle. Such a follow-up study would also need to explore different culturing conditions. Breuer et al. (2013), for example, demonstrated that nitrogen deprivation can alter the fatty acid composition in Tetradesmus significantly, and Nzayisenga et al. (2020) showed that heterotrophic growth with varying availability of organic carbon also affects the lipid production in several green algal species. Perhaps most importantly, the study would have to grow the algae at different temperatures, and ideally directly compare J. eremophilus to other desert and non-desert strains.