Paleozoic Roof Pendants Impart Subtle Geochemical Differences in Similar Springs
The ten springs along the foot of the eastern Sierra Nevada provide a broad geographic and geologic representation of springs along the west side of Owens Valley (Figure 1A). These rheocrene type springs[3] were physicochemically similar, being characterized by cool, fresh water (temperature: 11.2-18.7 ℃; specific conductance: 66.1-470.7 µS/cm), circumneutral pH (pH: 6.94-8.16), and oxygen undersaturation (dissolved oxygen: 2.9-8.7 mg/L). Detailed physicochemical data and representative photos of each spring are included in Supplementary Tables 1-4 and Supplementary Figure 7.
An HCA incorporating major ions, Si, and specific conductance data divided spring waters into the two major geochemical groups (Figure 1B). Student’s t-tests were utilized to assess specific geochemical differences between the groups for ~60 physical and chemical analytes. This revealed that the two spring groups possessed significantly different Ca2+/Na+ molar ratios (p-value = 3.1 * 10−5), Ca2+/Mg2+ molar ratios (p-value = 0.05), divalent to monovalent cation ion ratios (D/M; p-value = 0.001), and slopes of the spring emergences (p-value = 0.004) (Supplementary Table 1). Spring geochemistry in both groups reflected recharge through granitoid rock, which predominates the high-elevation recharge zones in the Sierra Nevada. All spring waters also contained an excess of calcium (Ca2+/Na+ > 0.6) over that expected for plagioclase weathering alone (Ca2+/Na+ of 0.4-0.6; Figure 1C), consistent with dissolution of disseminated calcite during recharge[41, 74, 75, 76, 77]. However, only one spring group contained an “excess of excess calcium”, with significantly higher Ca2+/Na+ ratios (Ca2+/Na+ > 1.5). This “excess of excess calcium” signature has recently been interpreted to reflect roof pendant weathering[41], which is consistent with the location of the five springs with these signatures. Two emerge downgradient from the Pine Creek Pendant units near Mt. Tom (Elderberry Canyon spring, IES-033) and Wheeler Crest (Wells Meadow B spring, IES-038), whereas the three remaining springs emerge downgradient of unnamed roof pendants near Lookout Point (Supplementary Figure 8). For simplicity, the spring group with “excess of excess calcium” signatures is hereafter referred to as “roof pendant springs”, whereas those with lower Ca2+/Na+ ratios are hereafter referred to as “granitic springs”.
Roof Pendant Springs Host More Diverse Benthic Macroinvertebrate Communities
Following quality filtering, 843,111 total 16S rRNA gene sequences from bacteria and archaea were recovered from 39 benthic samples, with an average of 21,618 reads per sample. A total of 14,523 unique ASVs were recovered, approximating the species richness of prokaryotes in these samples, and accounting for 52 bacterial phyla and nine archaeal phyla. Rarefaction curves for all benthic microbial community samples reached an asymptote between 10,000-15,000 sequences, indicating adequate sequencing depth for robust comparisons among springs and spring groups (Supplementary Figure 9). A total of 2,660 individual BMIs were identified across ten composite benthic samples with an average of 266 BMIs per sample. From the BMI community data, four phyla, seven classes, 16 orders, 47 families, and 85 genera were identified across the ten springs.
Several α-diversity metrics were calculated to determine whether geochemical grouping influenced diversity (Figure 2; Supplementary Tables 5-7). No significant differences in benthic microbial diversity were observed between roof pendant and granitic springs (Figure 2A-D; Supplementary Table 5; Supplementary Table 7). In contrast, BMI α-diversity was higher in roof pendant springs for richness (observed count ANOVA, p-value = 0.009), evenness (Gini-Simpson Index, Kruskal-Wallis Test, p-value = 0.01), and diversity (Shannon Index, Kruskal-Wallis Test, p-value = 0.01) (Figure 2E-G; Supplementary Tables 6 + 7).
Subtle Geochemical Differences Alter Community Composition of Benthic Microbes and Macroinvertebrates
NMDS ordinations based on Bray-Curtis dissimilarity for both prokaryotes and BMIs demonstrated that spring groups hosted unique communities (Figure 3). Significant differences were observed in both microbial (R-value = 0.28, p-value = 0.001) and BMI (R-value = 0.65, p-value = 0.009) communities, with samples clustering on the ordination according to geochemical grouping. Geographic distance was not correlated with Bray-Curtis community dissimilarity of microbial (Mantel test, p-value = 0.11) or macroinvertebrate communities (Mantel test, p-value = 0.08) (Supplementary Table S8).
Covariates of community dissimilarity (p-value ≤ 0.05) identified among the ~60 physical and chemical analytes were fitted to the NMDS ordinations to visualize correlates of community dissimilarity (Figure 3). Ca2+/Na+, Ca2+/Mg2+, and D/M ratios correlated with microbial and BMI communities from roof pendant springs, reflecting weathering of the roof pendants during recharge. Additional vectors corresponding to spring or sediment characteristics and landscape placement were also significant. In roof pendant springs, current velocity correlated with both communities, whereas slopes of the spring emergences and cobble percentage correlated only with microbial communities and gravel percentage only with BMI communities. In granitic springs, spring water residence time and muck percentage correlated with both communities.
Many analytes correlated only with microbial community dissimilarity (Figure 3A). Temperature and various ions/trace elements (vectors #12-#27) correlated with the two oldest springs, specifically, Spring along Hogback Ck. A (IES-024, 1,850 years in age) and Lubken Canyon Spring 2 (IES-054. 1,270 years in age) of the granitic springs group. Mo, Al, and sand percentage correlated with microbial communities in the granitic spring Unnamed Pine Creek Spring (IES-039). Wetted width and V correlated with microbial communities in the granitic spring Unnamed Spring north of Red Mountain (IES-029).
Roof Pendant Springs are Enriched with Physiologically Distinct Biological Communities
Benthic microbial communities were compromised predominantly of Proteobacteria (46.1% of the microbial community on average), Bacteroidetes (10.5%), Acidobacteria (6.9%), Actinobacteria (5.1%), Verrucomicrobia (5.1%), Chloroflexi (4.1%), Planctomycetes (3.8%), Nitrospirae (3.7%), and Cyanobacteria (3.6%) and other phyla at varying relative abundances across individual springs (Supplementary Figure 10). Relative abundance bar plots representing benthic microbial communities at various taxonomic levels are included in the supplement (Supplementary Figures 11-13).
LEfSe analysis of benthic microbial communities revealed 408 taxa (8 phyla, 33 classes, 69 orders, 111 families, and 187 genera) that were significantly (p-value ≤ 0.05) enriched in either roof pendant springs or granitic springs (Figure 4). Several phyla and one class with multiple child taxa were enriched in roof pendant springs: phyla Acidobacteria (four classes, three orders), Planctomycetes (three classes, four orders), Gemmatimonadetes (three classes, three orders), and Actinobacteria (one class, four orders), and the class Alphaproteobacteria (six orders) (Figure 4A). In granitic springs, only the archaeal phylum Euryarchaeota (two classes, three orders) had multiple classes or orders enriched. Granitic springs were also enriched with several bacterial classes with fewer child taxa, namely Clostridia (one order), Endomicrobia (one order), and Gracilibacteria (one order), and bacterial orders Bacteroidales, Beggiatoales, Chromatiales, Desulfobacterales, Desulfuromonadales, and Methylococcales.
The enriched families and genera suggest a unique set of physiological traits associated with each geochemical group (Figure 4B-C). Roof pendant springs were enriched with aerobic/facultatively anaerobic genera predominantly from Alphaproteobacteria (eight families, 11 genera), Actinobacteria (eight families, eight genera), Planctomycetes (three families, two genera), and Acidobacteria (two families, two genera). These taxa included those known to produce prosthecae/stalks and holdfasts in the Alphaproteobacteria and Planctomycetes, such as Bauldia, Hirschia, Gemmata, Pedomicrobium, Pir4 lineage (family: Pirellulaceae), Schlesneriaceae, Planctomyces, Pedomicrobium, Hyphomicrobiaceae, Micropepsaceae, SWB02 (family: Hyphomonadaceae), and Rhodomicrobium. Granitic springs were instead enriched with anaerobes, including fermenters in the Clostridia (one family, four genera), sulfate reducers in the Deltaproteobacteria (five families, eight genera), and methanogenic archaea within the Euryarchaeota (three families, three genera). Aerobic bacteria included methylotrophic bacteria in the Gammaproteobacteria (six families, four genera) and sulfide oxidizers in the Beggiatoaceae (Figure 4C). A full list of significant results from the LEfSe analysis can be found in the supplement (Supplementary Table 9).
BMI communities differ in composition at multiple taxonomic levels (Supplementary Figure 14; Figure 5). At the genus level, communities in roof pendant springs were relatively even and dominated by vagile insect taxa, such as Lepidostoma, Optioservus, Gumaga, Malenka, Ironodes, Baetis, Argia, and Enchytraeidae with additional populations of lower abundance taxa. These taxa are primarily shredders, collector-gatherers, and predators. Pyrgulopsis was present in only one roof pendant spring, Elderberry Canyon Spring (IES-033), where it was the most abundant taxon. In contrast, granitic spring BMI communities were dominated by Pyrgulopsis and stress-tolerant, non-insect taxa, mainly Pisidium and Hyalella, with less abundant populations of the insect taxa Argia, Optioservus, and Tricladida.
Extensive Network Connectivity Between Microbes and BMIs in Roof Pendant Springs
A co-occurrence network revealed 112 significant (p-value < 0.05) positive correlations between microbes and BMIs (Figure 6). Cluster 1 is a large network highlighting positive correlations between microbial and BMI genera in roof pendant springs. BMI genera belonging to shredder, collector-gatherer, or predator feeding groups comprised central nodes with high connectivity to microbial taxa, inferring key roles for these taxa in trophic interactions (Figure 6A): Ironodes (14 network connections), Gumaga (13), Baetis (11), Malenka (7), and Pentaneura (7). Prosthecate/stalked bacteria in Cluster 1 included Hyphomicrobium, Pedomicrobium, Pirellula/Pirellulaceae, Gemmataceae, and SWB02 (Hyphomonadaceae). Almost all bacteria in this cluster were aerobic or facultatively anaerobic, while most BMIs were vagile insects with low to moderate stress tolerance values (Figure 6B).
Much less connectivity between microbes and BMIs was observed in granitic spring communities (Clusters 2-4), inferring looser trophic associations (Figure 6). In Clusters 2-4 no stalked/holdfast bacteria were observed, unlike Cluster 1 (Figure 6A). BMI genera in Cluster 2 included three functional feeding groups, collector-gatherers (Hyalella), collector-filterers (Pisidium), and predators (Tricladida); both Cluster 3 and Cluster 4 were represented by the scrapers Pyrgulopsis and Zaitzevia. No shredders were present in the granitic spring clusters. The largest network was Cluster 2, with correlations between the stress-tolerant BMI genera Pisidium, Hyalella, and Tricladida and a mixture of aerobic/facultatively anaerobic (Dechloromonas and Lacunisphaera), methylotrophic (Crenothrix and Methyloglobulus), and anaerobic (Anaeromyxobacter and Geobacter) bacteria (Figure 6B). In Cluster 3, Zaitzevia was positively associated with the aerobic, ammonia-oxidizing archaeon Candidatus Nitrosoarchaeum and an unclassified genus in the Anaerolineaceae. Cluster 4 included Pyrgulopsis and the cyanobacterium Pleurocapsa PCC-7319 and an unclassified genus of Chitinophagaceae.