Metabarcoding and Metagenomics :
Research Article
|
Corresponding author: Jens Boenigk (jens.boenigk@uni-due.de)
Academic editor: Agnès Bouchez
Received: 09 Oct 2017 | Accepted: 14 Dec 2017 | Published: 02 Jan 2018
© 2018 Jens Boenigk, Sabina Wodniok, Christina Bock, Daniela Beisser, Christopher Hempel, Lars Grossmann, Anja Lange, Manfred Jensen
This is an open access article distributed under the terms of the Creative Commons Attribution License (CC BY 4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Citation:
Boenigk J, Wodniok S, Bock C, Beisser D, Hempel C, Grossmann L, Lange A, Jensen M (2018) Geographic distance and mountain ranges structure freshwater protist communities on a European scalе. Metabarcoding and Metagenomics 2: e21519. https://doi.org/10.3897/mbmg.2.21519
|
Protists influence ecosystems by modulating microbial population size, diversity, metabolic outputs and gene flow. In this study we used eukaryotic ribosomal amplicon diversity from 218 European freshwater lakes sampled in August 2012 to assess the effect of mountain ranges as biogeographic barriers on spatial patterns and microbial community structure in European freshwaters. The diversity of microbial communities as reflected by amplicon clusters suggested that the eukaryotic microbial inventory of lakes was well-sampled at the European and at the local scale. Our pan-European diversity analysis indicated that biodiversity and richness of high mountain lakes differed from that of lowland lakes. Further, the taxon inventory of high-mountain lakes strongly contributed to beta-diversity despite a low taxon inventory. Even though ecological factors, in general, strongly affect protist community pattern, we show that geographic distance and geographic barriers significantly contribute to community composition particularly for high mountain regions which presumably act as biogeographic islands. However, community composition in lowland lakes was also affected by geographic distance but less pronounced as in high mountain regions. In consequence protist populations are locally structured into distinct biogeographic provinces and community analyses revealed biogeographic patterns also for lowland lakes whereby European mountain ranges act as dispersal barriers in particular for short to intermediate distances whereas the effect of mountain ranges levels off on larger scale.
protist, molecular diversity, biogeography, microbial ecology, algae
Protists are not only a very diverse group of organisms but also quantitatively and qualitatively important components of all ecosystems. Their key ecological role has become a paradigm in microbial ecology based on the concept of the microbial loop (
Increasing our knowledge of biodiversity distribution patterns at any level of the tree of life is of paramount ecological and evolutionary significance (
In contrast to ecological factors the biogeographic distribution patterns of protists and the significance of historical factors potentially structuring their distribution came only recently into focus. In contrast, for macroorganisms biogeography and the factors driving distribution patterns of life on Earth have attracted scientists already for centuries. The realization that different geographic regions, despite similar ecological conditions, may be inhabited by different organisms goes back at least to Buffon (1707-1788) (
While biogeographic theory has developed into a strong conceptual framework for understanding the distribution patterns of animals and plants, its applicability to microbes has remained controversial. There have been disputes on the biogeographic distribution of microbial organisms since at least the early 19th century (
Most, if not all, microbial ecologists meanwhile agree that at least some single-celled organisms have limited distribution patterns (
Here we focus on the importance of the recent geological history to the distribution patterns of protists in lakes. For protists, geographic gradients are generally known (e.g.
We sampled 218 European freshwater lakes and ponds from sites in Norway, Sweden, Germany, Poland, the Czech Republic, Slovakia, Hungary, Romania, Austria, Italy, France, Spain and Switzerland. Site selection focused on the European orogens, specifically the Alps, the Pyrenees, the Apennine, the High Tatras, the southern Scandinavian mountains and the connecting flatlands (Suppl. materials
Genomic DNA was extracted using the my-Budget DNA Mini Kit (Bio-Budget Technologies GmbH) following the protocol of the supplier with the following modifications: Filters were homogenized in 800 µl Lysis Buffer TLS within lysing Matrix E tubes (MP Biomedicals) using the FastPrep instrument (MP Biomedicals). Homogenization was run three times for 45 seconds each at a speed setting of 6 m/s and then incubated for 15 min at 55 °C. The next steps followed the standard protocol supplied by Bio-Budget Technologies GmbH. The quality and quantity of the DNA was checked using a Thermo Scientific NanoDrop® ND-2000 UV-Vis spectrophotometer (Thermo Fisher Scientifics). PCR amplifications targeted the SSU V9 region and ITS1. Briefly, in order to cover a broad taxonomic spectrum, two primers with different wobble positions were combined in a ratio of 10% : 90%: 5’-GCTGCGCCCTTCATCGKTG-3’ (ITS2_Dino; 10%) and 5’-GCTGCGTTCTTCATCGWTR-3’ (ITS2_broad; 90%). We used the Amplicon-Duo pipeline (
Adapter and quality trimming was performed by the sequencing company. Samples were demultiplexed by the sequencing company (Fasteris) using MID sequences. Base quality of the sequence reads was checked using the FastQC software (
Reads remaining after the AmpliconDuo filter were clustered via the software SWARM (version 2.1.9;
All sequences are deposited at Genbank (accession number PRJNA414052).
Enter section text
Sequencing of the European freshwater samples resulted in 323,430,198 cleaned V9 forward and reverse read pairs. We applied the AmpliconDuo filter (
The applied sequencing depth was sufficient to approach saturation of eukaryotic taxon richness both at the local and European scale (Fig.
Planktonic protist ribosomal diversity. (A) V9 rDNA OTUs rarefaction curve. (B) Saturation slope versus number of V9 rDNA reads for all of the 218 samples analyzed herein. A slope of 10-6 indicates that one novel sequence read can be recovered if one million new reads are sequenced. Overall, the saturation values indicate that the extensive sampling effort uncovered the majority of eukaryotic ribosomal diversity both at the European and the local level.
Phylogenetic affiliation of reads and OTUs to eukaryotic taxonomic groups. The area of the portion of the pie chart represents the number of reads, whereas the angle represents the relative contribution to OTU richness. Alveolates and in particular Ciliophora and Dinoflagellata contributed nearly 50 % to global OTU richness. Cryptophyta, Viridiplantae (without Embryophyta), as well as Chrysophyceae and Diatomea accounted for the majority of the remaining OTUs. Unlabeled pieces within the eukaryotic supergroups comprise all other taxa affiliated with that group.
The OTU pool showed a good fit to the log-normal distribution (Fig.
Abundance distribution and ubiquity of OTUs. A) Global OTU abundance distribution and fit to the Preston log-normal model. Quasi-Poisson fit to octaves (red curve) and maximized likelihood to log2 abundances (blue curve). Most OTUs were represented by 3 to 32 reads. Approximations fit to the Preston log-normal model. The maximized likelihood to log2 abundances (blue curve) fit was superior and subsequently used to calculate the Preston veil, which infers the number of OTUs that we missed during our sampling. Preston veil was 8,111, confirming that we captured most of the diversity as also indicated by the rarefaction slopes. Preston vail and rarefaction slopes confirm that holistic and general patterns of eukaryotic freshwater diversity can be extracted from our data. The unit of the y-axis is the number of OTUs B) Distribution of OTUs among European lakes. Few OTUs were ubiquitous (occurring in the majority of lakes) whereas most OTUs were specific to one or only a few lakes. Unit of y-axis as in a).
Delaunay triangulation plot based on the investigated lakes. The red squares indicate triangles of high dissimilarity (i.e. larger than mean + one standard deviation) between the three corner sampling sites. Community dissimilarities were calculated based on wheighted distances using the distance function x=((1-distance)/maximal triangle leg). The areas of high dissimilarity occur nearly exclusively along mountain ranges, in particular along the Alpes, the Pyrenees, the High Tatras, the Carpathian mountains and the Sierra Nevada.
Beyond a shift in taxon inventory with elevation, richness pronouncedly decrease around 1400m (Suppl. material
OTU distribution and protist biogeographic regions in Europe. A) Richness and contribution to beta diversity of individual lakes. Richness is illustrated by grey scale of the symbols: dark shades indicate a high richness whereas light shades indicate a low richness. The size of the symbols indicates the relative contribution to beta diversity, with large symbols indicating a high contribution. Richness of individual lakes varied between 15 and 4092. Lakes in high mountain regions tended to contribute at an above-average level to beta diversity. B) Protist biogeographic regions as calculated by geo-constrained hierarchical clustering. High mountain ranges appear as biogeographic islands; the geographic regions largely comprising high mountain protist communities are depicted in yellow. Furthermore, the lowland areas are biogeographically structured which is mainly due to an east- west-gradient. Even though mountain ranges affect protist community similarity between sites, they do not necessarily separate biogeographic regions.
When restricting the analysis to presence/absence data most of the beta-diversity was due to replacement and only a minor part due to loss of species (nestedness), i.e. 4,0 % for all lakes. However, the nestedness was somewhat higher within the northern low elevation (cis) group, i.e. 5.4%, possibly (even though hypothetical) to incomplete recolonization since the last glaciation events whereas in high elevation lakes which possibly act as refuge the nestedness is very low, i.e. 2.1 %.
On the European scale environmental factors slighty co-varied with distance for lowland lakes but did not systematically change with distance for high mountain lakes (Fig.
Effect of geographic distance on community dissimilarity. A) Habitat dissimilarity for high mountain lakes (above 1,400 m) and for lowland lakes (below 500 m). The latter were either separated (cis-trans pairs) or not separated by mountain ranges (cis-cis pairs, i.e. north of the Alps, Pyrenees and High Tatras and trans-trans pairs, i.e. south of the Alps, Pyrenees and High Tatras). Regression lines are shown for highland (blue), lowland cis-cis and trans-trans pairs (yellow), and lowland pairs separated by mountain ranges (orange). Habitat dissimilarity increased with distance for lowland lakes but did not change for high mountain lakes. B) Bray-Curtis community dissimilarity for the same groups as in A. Community dissimilarity in high mountain regions increased with distance even though habitat dissimilarity did not change with distance. For lowland lakes both, habitat characteristics and community composition, changed with distance. However, the effect of distance on habitat dissimilarity was independent on whether the lakes were separated by mountain ranges or not. In contrast, community dissimilarity was higher at low and intermediate distance for lakes separated by mountain ranges. Accordingly the slope of the regression line for cis-trans pairs was significantly lower as for cis- cis and trans-trans pairs (p = 0.0097). The effect of geographic barriers levels off for longer distances above approximately 1,500 km.
Beyond the general effect of geographic distance, geographic barriers such as mountain ranges contributed significantly to community dissimilarity between lowland sites on a European scale. Mean community similarity was higher for lakes within a given lowland area as compared to community similarities of lakes between areas which were separated by a mountain range (Fig.
Despite a generally low community similarity between different lakes, groups of lakes sharing a higher level of community similarity can be identified. Lakes sharing a higher degree of community similarity form clusters reflecting biogeographic provinces on a European scale (Fig.
The decreasing effect of mountain ranges in separating protist communities with distance between sites indicates that geographic barriers act only as relative dispersal barriers for protists; they do not inhibit dispersal, but rather slow it down and thereby largely act on short to intermediate distances. For larger distances, the separating effect of geographic barriers adding to that of geographic distance alone largely vanishes, as demonstrated by cis-cis and trans-trans (i.e. within a distinct lowland area) versus cis-trans (i.e. lowland sites separated by a mountain range) comparisons of habitat dissimilarities (Fig.
The temporal and spatial pattern of eukaryotic microbial diversity is underexplored and has only recently come into focus (
Protist diversity is undoubtedly tremendous (
Although comparatively smaller, freshwater systems are usually more heterogeneous than marine systems and offer a larger array and diversity of ecological niches (
Our survey aimed to cover all eukaryotic lineages, but universal primers may miss some relevant taxonomic groups (
Our analysis based on 218 European lakes yielded in 74,713 distinct unique V9 sequence SWARMs. Rarefaction analysis as well as Preston vein indicated that we captured most of the diversity. However, molecular diversity inventories based on high throughput sequencing techniques must be interpreted with caution as high throughput sequencing platforms are error prone and in particular for sequences exclusively present in one or very few samples PCR artefacts and sequencing errors must be taken into account (
For some samples the number of OTUs was considerably lower as expected. This may be due to several factors: First, our filter strategy, including the AmpliconDuo filter and SWARM reduced the number of OTUs considerably. In particular, AmpliconDuo filtered out many presumably artificial sequences most of which had low to medium read numbers. Therefore the total number of OTUs should be expected to be somewhat smaller as compared to studies which do not apply this or a comparable filter (
Ecological factors, in particular abiotic factors such as nutrient concentrations, temperature and light and UV intensity are well known to affect protist community composition (e.g
Based on the largest freshwater data set known to us, we demonstrate a significant contribution of geographical distance and historical factors to protist distribution patterns. We found spatial patterns of protist diversity with microbial community composition varying both with altitude in mountain ranges and with geographical distance. Seasonal taxon fluctuations influence protist community composition and may interfere with spatial analyses when samples originate from different seasons (
Recently,
For instance, we found a general trend of decreasing richness for high elevation lakes, i.e. high mountain lakes at elevations above approximately 1400 m showed a lower mean richness as compared to lakes at low elevations. This may be due to either the generally small size of mountain lakes as would be expected from species-area-relationships or indicate differential patterns of richness depending on elevation. As small lakes at low elevation can show a high richness we suspect that both, habitat size and elevation, or factors co-varying with elevation, may be repsonsible for the low richness.
Even though ecological factors are certainly of primary importance explaining for protist community composition, the contribution of geographic distance and historical factors (or geographic barriers) seems to have a stronger impact on protist freshwater communities than on marine communities (cf.
According to the generally high beta diversity, our study revealed a high proportion of taxa exclusively found in only one or a few lakes. The very low contribtion of nestedness of the high elevation lakes and the somewhat higher contribution of nestedness to beta diversity in the northern lowland lakes may be interpreted as historical signals of the last glaciation and the post-glacial colonisation. However, as these differences are weak the interpretation as a historical signal remains hypothetical. How far postglacial resettlement, geographical barriers, other historical reasons, ecological parameters like nutrient supply or just biotic influences by other organismic groups dominate the exceptionality of the protist communities within lakes, remains to be shown. Further, it is likely that ecological and historical factors interact in shaping protist distribution pattern (
Mountain ranges have long been accepted as dispersal barriers for larger animals and plants, but their significance as dispersal barriers for microbes remained unclear. We demonstrate a differential impact of geographic distance for lowland and for highland lakes. Biodiversity and richness of high mountain lakes differ from that of lowland lakes. In accordance with the patterns for animals and plants (
Furthermore, mountain ranges act as islands of high mountain protist biodiversity and geographic distance accordingly clearly contributed to community dissimilarity. However, it must be noted that the overall contribution of geography to protist distribution pattern in lakes was low but, nevertheless, significant. Other factors such as abiotic and biotic factors presumably contribute to community composition much stronger as geography as it has been suggested in numerous studies (
We thank the German Research Foundation (projects BO 3245/19-1 and BO 3245/17-2) for financial support. We thank Joachim Stadel, Verena Stadel, Susann Chamrad, Steffen Jost, and Geoffrey Ongondo for support during the sampling campaign. We thank the Department Ecology, Biodiversity & Evolution of Animals at the University Salzburg, the Department Plant Ecophysiology at the University Konstanz, the Institut de Ciències del Mar in Barcelona, the Division of Clinical Physiology at the University of Debrecen, the University of Potsdam and the IGB Neuglobsow for supplying liquid nitrogen to the teams during the sampling campaign.
Richness is shown for different elevations. The number of lakes within this elevation range is indicated. While mean richness ranges around 750 OTUs it drops to around 400 OTUs at high elevations. The transition seems to be around or slightly below 1400m.