What Drives the Diversity of the Most Abundant Terrestrial Cercozoan Family (Rhogostomidae, Cercozoa, Rhizaria)?

Environmental sequencing surveys of soils and freshwaters revealed high abundance and diversity of the Rhogostomidae, a group of omnivorous thecate amoebae. This is puzzling since only a few Rhogostomidae species have yet been described and only a handful of reports mention them in field surveys. We investigated the putative cryptic diversity of the Rhogostomidae by a critical re-evaluation of published environmental sequencing data and in-depth ecological and morphological trait analyses. The Rhogostomidae exhibit an amazing diversity of genetically distinct clades that occur in a variety of different environments. We further broadly sampled for Rhogostomidae species; based on these isolates, we describe eleven new species and highlight important morphological traits for species delimitation. The most important environmental drivers that shape the Rhogostomidae community were soil moisture, soil pH, and total plant biomass. The length/width ratio of the theca was a morphological trait related to the colonized habitats, but not the shape and size of the aperture that is often linked to moisture adaption in testate and thecate amoebae.


Introduction
Environmental sequencing surveys challenge traditional beliefs of protist biodiversity and its environmental drivers. Amplicon-based metagenomics (also called metabarcoding) with different underlying bioinformatic pipelines and various primer sets all report consistent findings: (i) Cercozoa (Rhizaria) are one of the most abundant groups of protists in soils [1][2][3] and (ii) among the cercozoan sequence reads, operational taxonomic units (OTUs) assigned to the Rhogostomidae (Thecofilosea) appear to dominate cercozoan communities in a wide variety of terrestrial and some aquatic habitats [4][5][6][7][8][9][10]. However, aside from such molecular surveys, the Rhogostomidae are hardly mentioned in scientific literature and only a handful of species have been described until today [11][12][13]. Consequently, there appears to exist high untapped biodiversity of potentially important microbial key players, which we here investigated comprehensively taking the family Rhogostomidae as an example.
Theodosius Dobzhansky, half a century ago, made an iconic statement, i.e., "nothing in biology makes sense except in the light of evolution" [14]. With decreasing costs, environmental sequencing produces huge amounts of microbial data and the meaningful interpretation of the data becomes the main effort. The correct exploitation of environmental data, including the assignment of genetic data, in this case OTUs, to existing species is crucial for linking microbial sequence diversity to functioning [15]. In other words: Nothing in environmental sequencing makes sense unless it is based on species in culture flasks (SARESTEDT AG & Co. KG, Nümbrecht, Germany) and stored at 14-16 • C. Strains were sub-cultivated monthly and deposited in the Culture Collection for Algae and Protozoa (CCAP). Respective accession numbers are given in Table 1.

Microscopic Observation
Light microscopic observations were made with a Nikon Eclipse TS100 inverted microscope (up to 400× magnification, phase contrast). Pictures and videos were taken with a Nikon Eclipse 90i microscope (DIC, up to 600× magnification) with a Nikon digital sight DS-U2 mounted camera (program: NIS-Elements V4. 13.04). Picture assemblies were made with Adobe Photoshop CC 2014 (Adobe Systems, Munich, Germany).
Hapantotypes were dried and submitted to the Upper Austrian State Museum Invertebrate Collection as Inv. Nr. 2019/71-80.

Phylogenetic Analyses
A nearly full-length SSU rDNA sequence was amplified from each isolate in two overlapping fragments, using EukA and EukB as universal eukaryotic primers and different combinations of cercozoan specific primers (Tables 1 and 2). PCR amplifications were preferably done with single cells in a total volume of 17 µL. The mixture included 1x Thermo Scientific Dream Taq Green Buffer, 1 µM forward and 1 µM reverse primer, 0.2 mM dNTPs, and 0.01× DreamTaq polymerase (Thermo Fisher Scientific, Dreieich, Germany) and dd H 2 0 to fill until 17 µL. An amplification profile consisting of 34 cycles with 32 s at 95 • C, 36 s at 50 • C, and 2 min at 72 • C, followed by 7 min at 72 • C for the final extension was conducted. PCR products were purified by adding 1 U/mL of Exonuclease, 0.3 U/mL FastAP and to 8 µL PCR product, then heating for 30 min at 37 • C, and subsequently for 20 min at 85 • C. For sequencing, the Big dye Terminator Cycle sequencing Kit and an ABI PRISM automatic sequencer were used.
For the phylogenetic analyses, a reference database containing sequences with >93% similarity to the type species Rhogostoma schuessleri were obtained from the NCBI GenBank database. These sequences most likely represent the full genetic diversity of Cryomonadida present in the NCBI GenBank database (last date of accession: 26/07/19). Subsequently, mismatches and erroneous sequences were deleted. Eleven sequences of the Tectofilosida were added as outgroup. They were aligned in mafft using the linsi algorithm [30]. The sequences of all strains obtained in this study (Table 1) were manually checked for sequencing errors using Chromas (V2.6.5) and assembled in SeaView (V4.6, [31]). Additionally, we added subsets of large-scale environmental survey data marking the hypervariable region V4 of the SSU rDNA (18S rDNA) of Cryomonadida from following published datasets: Degrune et al. [9], Fiore-Donno et al. [5,6], Heger et al. [25] and Jauss et al. [24]. Accordingly, an alignment with 752 sequences, containing 501 NCBI database sequences and an additional 251 OTUs from the previously mentioned datasets were used for phylogenetic analysis. Maximum likelihood trees were calculated in RAxML (Randomized Axelerated Maximum Likelihood; [32]). The best scoring tree was used to report the confidence values as percentages obtained through 200 non-parametric bootstraps under the GTRCAT model. The same method was used to obtain the 18S rDNA phylogenetic analysis of Rhogostomidae. There, an alignment with all described Rhogostomidae spp. of which SSU rDNA data was publicly available was conducted and eight of the same Tectofilosida spp. as mentioned above were used as outgroup.

Statistical Analyses
All statistical analyses were carried out in R (V3.6, [33]). To determine significant differences between the morphological features of Rhogostoma spp. strains, one-way ANOVA followed by Turkey's HSD test were conducted. Results were expressed by means and standard deviation.
Shifts in the magnitude of morphological traits between isolated Rhogostomidae strains were shown in box plots (see Supplementary Materials). To identify the most significant discriminating morphological traits between strains the traits were normalized by cell size and analyzed by principal component analysis (PCA) using Euclidean distances [34]. Scaling 1 was used to visualize distances among normalized morphological measures in the biplot as approximations of their Euclidean distances in multidimensional space. The radius of the circle of equilibrium contribution represents the length of the vector representing a variable that would contribute equally to all the dimensions of the PCA space ( Figure 1; [34]). Non-metric multidimensional scaling (NMDS) using Bray Curtis distances was used to visualize the relationships among beta diversity of Cryomonadida and environmental factors. We used Permutational Multivariate Analysis of Variance (PERMANOVA; [35]) to test if these environmental factors were significantly influencing the Cryomonadida community. Each test was permuted 999 times. All ordinations were calculated with the package vegan [36]. the length of the vector representing a variable that would contribute equally to all the dimensions of the PCA space ( Figure 1; [34]). Non-metric multidimensional scaling (NMDS) using Bray Curtis distances was used to visualize the relationships among beta diversity of Cryomonadida and environmental factors. We used Permutational Multivariate Analysis of Variance (PERMANOVA; [35]) to test if these environmental factors were significantly influencing the Cryomonadida community. Each test was permuted 999 times. All ordinations were calculated with the package vegan [36].

Morphological Analysis
We were able to culture 16 Rhogostoma strains out of approximately 50 screened samples from Germany, the Netherlands, and Austria. All isolated thecate amoebae were bilateral symmetric, bore a smooth hyaline theca, and the cells moved with aid of filopodia which originated from a basal slit-like aperture ( Figure 1). Filopodia branched and anastomosed in all directions during the early stages of culturing, but cells in older cultures often ceased to extend their filopodia. According to our analyses, four general Rhogostoma morphotypes can be distinguished despite high morphological variability: i.e., cells that were elongated and lateral flattened (R. kappa), cylindrical/conical (R. cylindrica, R. pseudocylindrica, R. florae, R. absidea, R. karsteni), angular (R. kyoshii, R. medica, R. radagasteri) and lateral compressed spherical (R. schuessleri, R. micra, R. minus, R. epiphylla, R. leviosa, R. bowseri). See Supplementary Table S1 for detailed morphological measurements; and Taxonomic Appendix A for detailed species descriptions.
To determine important morphological traits for species delimitation, we analyzed traits of Rhogostoma spp. via a principal component analysis. Cell length to width ratio, relative cell outer length, and invagination of the aperture turned out to be the morphological traits that were best suited for separating the strains (Figure 2; PCA axis PC1 = 46%; PC2 = 19%). Cell length to width ratio and aperture depth were negatively correlated (R 2 = 0.49, p < 0.001). Aperture invagination was positively correlated with cell outer length (R 2 = 0.94, p < 0.001). The remaining morphological traits did not contribute sufficiently in explaining variation among strains (see the circle of equilibrium contribution, Figure 2).

Phylogeny
To illustrate morphological adaption during the evolution of Rhogostomidae, 16 SSU rDNA sequences were obtained and subjected to phylogenetic analyses. The sequences ranged from 909 to 1767 nucleotides (Table 1). A combined analysis of these sequences and sequences from all currently sequenced and described Rhogostomidae revealed the freshwater genus Sacciforma (Rhogostomidae) to group basal to Rhogostoma spp. with full support (Figure 3). All Rhogostoma spp. group in a highly supported monophylum. Species which originated from freshwater are found in two distinct groups one being basal to (S. sacciformis and R. kappa) and another one being high in the Rhogostoma radiation (R. radagasteri, R. micra and R. minus). The latter intermingles with sequences that originated from soil taxa (Figure 3). Morphological variation did not show a clear pattern with phylogenetic branching, but it must be addressed that some branches within the Rhogostomidae were only poorly supported.

Phylogeny
To illustrate morphological adaption during the evolution of Rhogostomidae, 16 SSU rDNA sequences were obtained and subjected to phylogenetic analyses. The sequences ranged from 909 to 1767 nucleotides (Table 1). A combined analysis of these sequences and sequences from all currently sequenced and described Rhogostomidae revealed the freshwater genus Sacciforma (Rhogostomidae) to group basal to Rhogostoma spp. with full support (Figure 3). All Rhogostoma spp. group in a highly supported monophylum. Species which originated from freshwater are found in two distinct groups one being basal to (S. sacciformis and R. kappa) and another one being high in the Rhogostoma radiation (R. radagasteri, R. micra and R. minus). The latter intermingles with sequences that originated from soil taxa ( Figure 3). Morphological variation did not show a clear pattern with phylogenetic branching, but it must be addressed that some branches within the Rhogostomidae were only poorly supported. A large-scale phylogenetic analysis was conducted to reveal the genetic diversity of the Rhogostomidae and clades of which strains have not yet been cultured and novel clades that await exploration. Accordingly, we included as many OTUs and their respective metadata as we could gather from published data sets including the NCBI to shed light on the potential distribution of respective clades. The analysis revealed four novel clades at the base of Cryomonadida with sequences originating from diverse environments (Figure 4). Sequences from the genera Protaspa and Cryothecomonas intermingle with each other and a large variety of environmental OTUs that exclusively originated from saline habitats. Similar to the phylogenetic analyses presented in Figure  3, the genus Sacciforma groups basal to Rhogostoma. The genus Rhogostoma shows a large variety in SSU rDNA sequences and numerous environmental clades without cultured representatives. A large-scale phylogenetic analysis was conducted to reveal the genetic diversity of the Rhogostomidae and clades of which strains have not yet been cultured and novel clades that await exploration. Accordingly, we included as many OTUs and their respective metadata as we could gather from published data sets including the NCBI to shed light on the potential distribution of respective clades. The analysis revealed four novel clades at the base of Cryomonadida with sequences originating from diverse environments (Figure 4). Sequences from the genera Protaspa and Cryothecomonas intermingle with each other and a large variety of environmental OTUs that exclusively originated from saline habitats. Similar to the phylogenetic analyses presented in Figure 3, the genus Sacciforma groups basal to Rhogostoma. The genus Rhogostoma shows a large variety in SSU rDNA sequences and numerous environmental clades without cultured representatives. The bootstrap values are visualized in a colour flow from black (low support) to purple (high support); end nodes are also given in black. Environmental sequences of the in-depth analysed datasets [5,6,9,24,25] are indicated in purple font, NCBI database environmental sequences and sequences derived from isolated strains are indicated in black. Isolation habitats are highlighted and visualized in the outer colour stripe.

The Phylogenetic Position of Capsellina
The poor quality of the only available sequence of the Rhogostomidae genus Capsellina (GQ377676) was already discussed in Dumack et al. [12]. Now we provide further evidence that the name of this sequence is based on misidentification: Rhogostoma strains K8 and K9, here described as R. florae have a highly similar SSU rDNA sequence to the claimed "Capsellina sp.".

Environmental Drivers of Species Turnover in Rhogostomidae
Based on previously published datasets that come with extensive metadata [6,24,25], we investigated the effects of soil physicochemical parameters on the beta diversity of Cryomonadida (Rhogostomidae and relatives). We decided to include whole Cryomonadida diversity to avoid exclusion of OTUs with unknown family-affiliation on the basis of an arbitrary threshold. On large scale, the composition of Cryomonadida communities differed among four dominant ecosystem types along a bog-woodland transect (i.e., bog woodland, bog forest, blanket bog, and zonal forest; PERMANOVA F3,33 = 4.03, p = 0.001). In general, differences in soil physicochemical parameters in respect to soil contents of iron (Fe), sodium (Na), manganese (Mn), total carbon (Total C), available P, and pH influenced species turnover in Cryomonadida ( Figure 5A; PERMANOVA F1,34 = 8.03, p = 0.003). Chemical elements including iron and sodium were among the strongest predictors of . SSU rDNA phylogeny of Cryomonadida (Rhogostomidae and relatives) with a focus on Rhogostomidae diversity (highlighted in orange for Rhogostoma and cyan for Sacciforma) with chosen Tectofilosida as outgroup. Shown is the maximum likelihood tree obtained by RAxML. The bootstrap values are visualized in a colour flow from black (low support) to purple (high support); end nodes are also given in black. Environmental sequences of the in-depth analysed datasets [5,6,9,24,25] are indicated in purple font, NCBI database environmental sequences and sequences derived from isolated strains are indicated in black. Isolation habitats are highlighted and visualized in the outer colour stripe.

The Phylogenetic Position of Capsellina
The poor quality of the only available sequence of the Rhogostomidae genus Capsellina (GQ377676) was already discussed in Dumack et al. [12]. Now we provide further evidence that the name of this sequence is based on misidentification: Rhogostoma strains K8 and K9, here described as R. florae have a highly similar SSU rDNA sequence to the claimed "Capsellina sp.".

Environmental Drivers of Species Turnover in Rhogostomidae
Based on previously published datasets that come with extensive metadata [6,24,25], we investigated the effects of soil physicochemical parameters on the beta diversity of Cryomonadida (Rhogostomidae and relatives). We decided to include whole Cryomonadida diversity to avoid exclusion of OTUs with unknown family-affiliation on the basis of an arbitrary threshold. On large scale, the composition of Cryomonadida communities differed among four dominant ecosystem types along a bog-woodland transect (i.e., bog woodland, bog forest, blanket bog, and zonal forest; PERMANOVA F 3,33 = 4.03, p = 0.001). In general, differences in soil physicochemical parameters in respect to soil contents of iron (Fe), sodium (Na), manganese (Mn), total carbon (Total C), available P, and pH influenced species turnover in Cryomonadida ( Figure 5A; PERMANOVA F 1,34 = 8.03, p = 0.003). Chemical elements including iron and sodium were among the strongest predictors of Cryomonadida communities, followed closely by soil pH, which ranged in this study from 3.68 to 4.93.

Discussion
In this study, we followed an integrative approach to facilitate species delimitation of morphologically quite similar, but genetically distinct minute thecate amoebae. This approach allowed us to identify specific morphological traits and biotic and abiotic environmental drivers to differentiate species in Rhogostomidae and appears as a promising strategy for the reliable species On a small scale, biotic factors like total plant biomass, plant litter biomass, microbial biomass carbon content (Cmic), and numbers of bacteria, as well as soil physicochemical parameters, such as bulk density, soil moisture, soil C/N ratio, organic carbon content (organic C), extractable organic carbon phosphate (PO 4 2 ), and nitrate (NO 3− ) content, had significant effects on the beta diversity of Cryomonadida ( Figure 5B; PERMANOVA F 1,175 = 19.8, p = 0.005). Compared to the large-scale bog-woodland transect, soil pH and clay content did not show an effect in the small-scale study in grasslands. Here the strongest predictors of beta diversity of Cryomonadida were soil moisture and its negative correlation with plant biomass (R 2 = 0.28, p < 0.001). In a data set comparing protists from tree canopies with those on the forest ground, the Cryomonadida communities differed clearly between canopy habitats and the forest floor, but not among canopy microhabitats such as fresh leaves, bark, deadwood, and detritus between tree branches ( Figure 5C; PERMANOVA F 8,72 = 5.42, p = 0.001).

Discussion
In this study, we followed an integrative approach to facilitate species delimitation of morphologically quite similar, but genetically distinct minute thecate amoebae. This approach allowed us to identify specific morphological traits and biotic and abiotic environmental drivers to differentiate species in Rhogostomidae and appears as a promising strategy for the reliable species delimitations in cryptic species complexes of protists. We follow the premise that undoubtedly the best taxonomic concept is the one that reflects evolution. At the core of taxonomy lies the notion of the species. Since sex in most protists is decoupled from reproduction, the 'biological species concept' cannot be applied [37][38][39]. Following an integrative approach settles the dispute whether morphology or genetic data should be used to delimit species. We show that a combination of both is needed, especially in taxa with few morphological characters, but high genetic diversity, like in Rhogostomidae. Based on the combination of morphological investigations and genetic analyses, we were able to identify reliable morphological traits differentiating species in Rhogostomidae like cell length-to-width ratio and aperture invagination.
We show that the Cryomonadida are surprisingly diverse and can be isolated from various habitats. We found a high genetic diversity in a single protist genus and almost every isolated individual varied in numerous morphological aspects and thus we conclude the Rhogostomidae to be highly species-rich, confirming our first hypothesis. The consensus of all available environmental sequencing data of the NCBI database and the fact that we were not able to isolate Rhogostomidae from our marine samples indicates a prevalence of Rhogostomidae for terrestrial (soil, plant surfaces) and some freshwater microhabitats (ponds and wastewater treatment plants). According to the wealth of ribosomal sequences in the NCBI database, the Rhogostomidae largely lack representation in marine environments in our data, although marine habitats were in comparison to terrestrial environments highly sampled in recent years [40,41]. According to the branching patterns of the phylogenetic analysis, the Rhogostomidae may have derived from marine ancestors (related to Protaspa, Crypthecomonas) and subsequently adapted over freshwater (Sacciforma and Rhogostoma kappa) to terrestrial habitats (most other Rhogostoma spp.), leading us to reject our second hypothesis that Rhogostomidae are ubiquitously distributed, but supporting our third hypothesis.
Several large-scale studies have tried to identify environmental drivers of protistan diversity and community assembly [6,24,25,[42][43][44]. However, the complexity of their diverse sizes, morphotypes, and diverse feeding modes complicates the detection of specific environmental drivers for a group of taxa. Species turnover (beta diversity) of Cryomonadida was high between large-scale contrasting habitats along a transect from blanket bog to a zonal forest. In particular, the strong gradient of soil acidity from bog to forest structured the Cryomonadida community. In contrast, gradients of physicochemical parameters of soil samples within a 10 × 10 m grassland site were rather low [6], but seasonal differences in plant biomass production and associated soil moisture were the main environmental factors structuring the small-scale community composition of Rhogostomidae. Although scientific reports of Rhogostomidae are scarce, the genus Rhogostoma was repeatedly found to be associated with plants, where it seems to feed on fungi, algae, and bacteria [10,11,[45][46][47][48]. The affinity to plants also triggered the description of the first plant surface-associated Rhogostoma species, R. epiphylla [11]. Our phylogenetic analyses show an intermingling of soil and plant surface inhabiting Rhogostoma species and we further show a negative correlation of soil moisture and plant biomass to predict the Rhogostomidae community. Both of these findings may indicate that a diverse array of Rhogostoma species may be attracted to plant surfaces due to a certain degree of water availability in its vicinity. The organic theca of Rhogostomidae may protect the cells from water loss to a certain degree and may explain their increased dominance in drier conditions [6]. Additionally, the aperture of the theca of Rhogostomidae is particularly slim, potentially reducing water loss from the cell as an adaption to terrestrial life. Therefore, we expected clear differences in aperture depth between freshwater and terrestrial Rhogostomidae. However, we could not find clear evidence for this hypothesis since our phylogenetic analyses showed freshwater Rhogostomidae occasionally intermingling with terrestrial clades of Rhogostomidae. Considering their morphology (see Supplementary Materials for details), we further could not find a clear distinct difference of aperture depth or cell shape between freshwater and terrestrial Rhogostomidae. Instead, we found a negative correlation between aperture depth and the cell length/width ratio, which shows that the aperture depth is rather connected to overall cell size and shape and does not show direct adaption to environmental differences like water availability. Altogether this may indicate that Rhogostoma individuals independently of aperture depth and cell shape can tolerate drier conditions.

Conclusions
The Rhogostomidae are a particularly abundant taxon in terrestrial ecosystems and we show that they represent a surprisingly diverse family in the Cercozoa. Reliable species delimitation was possible by an integrative taxonomic approach. Clear patterns in beta diversity along environmental gradients indicate species turnover due to taxa specifically adapted to specific biotic and abiotic environmental conditions. However, specific morphological adaptions to microhabitat conditions could not be found in these character-poor and minute organisms. The invagination of the aperture and cell length to width ratio turned out to be the best morphological characters for species delimitation in Rhogostomidae.