Molecular Ecology (2015) 24, 1236–1247
doi: 10.1111/mec.13116
Evidence for an active rare biosphere within freshwater protists community N E H U G O N I * † and I S A B E L L E D O M A I Z O N ‡ DIDIER DEBROAS,*† MYLE *Laboratoire ‘Microorganismes: Genome et Environnement’, Clermont Universite, Universite Blaise Pascal, BP 10448, F-63000 Clermont-Ferrand, France, †CNRS, UMR 6023, LMGE, Aubiere F-63171, France, ‡INRA, UMR 42 Centre Alpin de Recherche sur les Reseaux Trophiques et Ecosystemes Limniques, F-74200 Thonon Les Bains, France
Abstract Studies on the active rare biosphere at the RNA level are mainly focused on Bacteria and Archaea and fail to include the protists, which are involved in the main biogeochemical cycles of the earth. In this study, the richness, composition and activity of the rare protistan biosphere were determined from a temporal survey of two lakes by pyrosequencing. In these ecosystems, the always rare OTUs represented 77.2% of the total OTUs and 76.6% of the phylogenetic diversity. From the various phylogenetic indices computed, the phylogenetic units (PUs) constituted exclusively by always rare OTUs were discriminated from the other PUs. Therefore, the rare biosphere included mainly taxa that are distant from the reference databases compared to the dominant ones. In addition, the rarest OTUs represented 59.8% of the active biosphere depicted by rRNA and the activity (rRNA:rDNA ratio) increased with the rarity. The high rRNA:rDNA ratio determined in the rare fraction highlights that some protists were active at low abundances and contribute to ecosystem functioning. Interestingly, the always rare and active OTUs were characterized by seasonal changes in relation with the main environmental parameters measured. In conclusion, the rare eukaryotes represent an active, dynamic and overlooked fraction in the lacustrine ecosystems. Keywords: activity, lakes, protists, rare biosphere Received 3 August 2014; revision accepted 13 February 2015
Introduction The application of high-throughput sequencing technologies has aided in the discovery of a tremendous microbial diversity and the existence of a rare microbial biosphere. For the first time, Sogin et al. (2006) have provided evidence of a myriad of bacterial OTUs found to exist in low abundance. These findings have raised questions about the ecological significance of these rare micro-organisms, and this has become one of the hottest topics in microbial ecology, with many questions remaining unanswered. The presence of rare taxa within microbes is not new, and it was observed for various protistan taxa identified by their morphology, but generally for specific taxonomic groups (e.g. ciliates; Dunthorn et al. 2014). The concept of rare biosphere coming Correspondence: Didier Debroas, Fax: +33 4 73 40 76 70; E-mail: [email protected]
from the molecular inventories depicting complex microbiological assemblages is quite different, with rare taxa potentially belonging to various taxonomic and functional groups and not only to a specific population. This highly diversified rare biosphere is generally associated with the idea of functional redundancy, that is a reservoir of dormant taxa that provides biological buffering capacity to withstand environmental changes (Caron et al. 2009). However, we still do not know to what extent rare taxa play significant roles in community function, that is as ribosomally active organisms important to the ecosystem function and not just when they become abundant (Bachy & Worden 2014). According to a previous study by Pedr os-Ali o (2012), rare taxa have been hypothesized to consist of a seed bank (i.e. dormant micro-organisms) that could potentially be resuscitated under favourable environmental conditions. Indeed, for bacteria, it was observed that many taxa switch between abundant and rare (Brown © 2015 John Wiley & Sons Ltd
A C T I V E R A R E P R O T I S T S I N L A K E S 1237 et al. 2005; Campbell et al. 2011). Such changes/rearrangements have also been observed in Archaea, for instance the clade MGII.a (Euryarchaeota) (Hugoni et al. 2013). Thus, some members of microbial communities are always present but vary in population size, switching from rare to abundant, due to environmental changes (Caporaso et al. 2012). However, the hypothesis of a seed bank is contradicted by some results: (i) some Bacteria and Archaea taxa are always rare, and a significant portion of these rare communities is actually active (Campbell et al. 2011; Hugoni et al. 2013) and (ii) the growth rates can decrease as abundance increases (Campbell et al. 2011). The role of rare and active microbes in terrestrial systems was indirectly observed via the manipulation of the trophic network because the reduction of rare soil microbes modified plant–herbivore interaction (Hol et al. 2010). Studies about the rare microbial biosphere are mainly focused on Bacteria and Archaea and fail to include the protists, which are involved in the main biogeochemical cycles of the earth covering all of the functional roles (phototrophs, parasites, saprotrophs, phagotrophs). The first attempt to study the dynamics of the rare protists was based on the analyses of DNA sequences obtained from the Sanger method (Caron & Countway 2009), and it is likely that the rare microbial biosphere was underestimated. Recently, the few high-throughput sequencing approaches in freshwater ecosystems demonstrated a stable predominance of a few highly abundant taxa at different temporal scales in lakes (Nolte et al. 2010; Mangot et al. 2013) and a biogeography of the rarest taxa, which represent a huge diversity of taxa (Lepere et al. 2013). As described for prokaryotes, the rare protists could be inactive (i.e. dormant) and become dominant with changing environmental conditions according to the model of Jones & Lennon (2010). These authors show that the variation of total phosphorus had no effect on the proportion of dormant protists that were detected by T-RFLP in lakes, and the transition between activity and dormancy could play a more important role in shaping bacterial communities than in eukaryotic communities. The specificity of the eukaryotic domain could be due to the ability to form resting stages or to the differential sinking rates. The hypothesis is that the switch between active and dormant stages could affect a low proportion of protists. Some protists could also be active as rare taxa, as demonstrated recently by Logares et al. (2014) in marine environments; however, to our knowledge, no other study has demonstrated the existence of rare and active protistan taxa. A low population density could be an advantage for some taxa because it could reduce parasitic infection, zooplankton predation, and could aid in the ability of that taxa to persist in the ecosystem. © 2015 John Wiley & Sons Ltd
Finally, there are various hypotheses on the rare microbial biosphere, which carries out keystone metabolic pathways and metabolizes rare energy sources. However, essential information to discuss the role of rare taxa is determining their level of activity, for instance by studying the structure community at both the DNA and RNA level (Jones & Lennon 2010; Campbell et al. 2011; Hugoni et al. 2013; Logares et al. 2014); such studies are still very rare. An overview of the dynamics and role of the rare microbial biosphere requires additional knowledge, particularly on the rare microbial eukaryotes whose dynamic and activity could differ from other microbes. We focused our study on small protists (0.2–5 lm) that can be considered to be an excellent model to study the rare biosphere in this kingdom by molecular approaches because the rank–abundance curve and the definition of the rarer fraction could be less biased than for larger eukaryotes. Indeed, the 18S gene copy number is correlated to genome size in eukaryotes (Prokopowich et al. 2003) and seems to be low in these small protists (0.1%). The data were normalized to have an equal number of rRNA and rDNA reads per sample in order to infer on the activity of eukaryotic taxa (rRNA:rDNA).
Taxonomic affiliation The representative OTUs were used to build phylogenetic trees, with FASTTREE (Price et al. 2010), for main taxonomic groups with reference sequences. These references were extracted to the SSUREF SILVA database (Pruesse et al. 2007) according to the following criteria: length >1200 bp, quality score >75% and a pintail value >50. In addition, the taxonomy of this reference database was modified to take account the typical freshwater lineages previously defined. The taxonomic annotation was conducted in two ways: the nearest neighbour (NN) and the last common ancestor (LCA) methods. A phylogenetic unit (PU) is a cluster of OTUs that is phylogenetically closest to a taxa from the reference database meaning that the cut-off is dependent on the closest relative in this database (Fig. 1). This process was implemented in the pipeline Phylogenetic Analysis of Nextgeneration AMplicons (PANAM http://code.google.com/ p/panam-phylogenetic-annotation/downloads/list) and described more in detail in the related study (Taib et al. 2013) and in the supporting information (Fig. S1, Supporting information).
Comparing the experimental OTUs with reference sequences Two methods were used for comparing the OTUs defined in this work with the 18S sequences in the database. First, OTUs were compared with the SSUREF SILVA © 2015 John Wiley & Sons Ltd
A C T I V E R A R E P R O T I S T S I N L A K E S 1239 package. To test the putative variables controlling the dynamics of OTUs, a canonical correspondence analysis (CCA) was conducted by linking the environmental parameters (temperature, oxygen, nutrients, chlorophyll a, abundances of Cyanobacteria, diatoms and Cryptophyta) selected by a forward analysis, to the OTUs (expressed in read abundances). This procedure was performed with the Vegan package (Dixon 2003).
Results Structure of the small protists communities
Fig. 1 Explanation of phylogenetic indices computed in this study from the trees generated from the experimental OTUs and reference sequences. In this example, 3 OTUs were gathered in a phylogenetic unit (PU) (e.g. PU_1) with a bootstrap equal to 99%. The nearest neighbour among the reference sequences in the tree, FJ153652, is affiliated to the Choricystis (Chlorophyceae), and this taxonomy is propagated to all the OTUs of this PU.
database (NR 117) restricted to the eukaryotes using UCLUST with a threshold of 95%. Second, various phylogenetic indices were computed from the trees generated (Swenson 2009) (Fig. 1). Phylogenetic diversity (PD) corresponds to the sum of branch lengths from the different PUs or all the OTUs to the root of the trees. Mean patristic distance (MPD) corresponds to the mean PD between each OTU and each other among the PUs. Mean nearest neighbour distance (MNND) is defined as mean phylogenetic distance from each species to its closest relative in the PUs. This index was computed by taking into account the reads abundance (MNND ab) or not (MNND pa). The ‘X depth/deeper’ is defined as the average distance between OTUs in the PUs and the closest reference sequence in the tree (Pommier et al. 2009). A higher MPD and MNND suggests a higher diversity within the PU, and a higher X depth/deeper suggests a higher phylogenetic distance between the PU and the closest relative in the reference database. These various indices were computed using R with the packages picante (Kembel et al. 2010), geiger and ape (Paradis et al. 2004).
Statistical analysis The relation between the proportion (%) of rRNA and rDNA determined for each OTU in the three fractions defined was tested by a linear regression. Linear models (regression and ANOVAs) were computed using R © 2015 John Wiley & Sons Ltd
All samples were processed together to display a general view of the protists community structure. After the cleaning procedures described in the Experimental Procedures and discarding the singletons and OTUs without rDNA reads, we defined 6446 OTUs delineating 686 PUs. In term of OTUs number, the main taxonomic groups were the Alveolata (1923 OTUs), Fungi (1564), Stramenopiles (1165) and Viridiplantae (776) (Table S1). This order changes if we consider the abundances of the rDNA reads; the Viridiplantae slightly dominated the Stramenopiles. Rank–abundance curves were created from the rDNA reads (Fig. 2), showing that only seven OTUs had an abundance >1% according to criteria used sometimes for defining the dominant ones (Pedr os-Ali o 2006). These OTUs belonged to three taxonomic groups: Haptophyceae, Chlorophyta and Fungi. The OTUs always rare and abundant (i.e. >0.1%) represented 77.2% and 2.4%, respectively, of the 6446 OTUs. In this abundant fraction, we did not detect any OTU belonging, for example, to Choanoflagellida, Euglenozoa or Rhodophyta, whereas OTUs from these groups were in the two other fractions.
Similarity and phylogenetic indices By clustering the OTUs (95%) with the SILVA database, we found that 26.9%, 8.2% and 2.5% of the abundant, cycling and always rare fractions, respectively, were included in a cluster with at least one sequence of the public database. In the always rare fraction, 13.9% of OTUs have an identity 1%). In the panel C, the outside of the circle corresponds to the active OTUs (OTUs with rRNA reads) and inside the inactive OTUs (including only rDNA reads).
(B)
(A)
Table 1 Similarity and phylogenetic indices determined on the different phylogenetic units (PUs). The mix PUs includes OTUs among different fractions always rare, cycling and abundant, whereas the other PUs include OTUs belonging only to one fraction
Num PUs Active PUs† BLAST
‡
MNND ab MNND pa MPD X depth/ deeper
Abundant PUs
Cycling PUs
Always rare PUs
Mix PUs
22 22
48 40
309 236
307 301
96.64a 0.14ac 0.17ac 0.30bcd 0.11ac
96.65a 0.25bc 0.25bc 0.25d 0.13bc
95.81 0.57 0.59 0.66ac 0.32
95.28 0.23ab 0.29ab 0.62ab 0.22ab
P*
doi: 10.1111/mec.13116
Evidence for an active rare biosphere within freshwater protists community N E H U G O N I * † and I S A B E L L E D O M A I Z O N ‡ DIDIER DEBROAS,*† MYLE *Laboratoire ‘Microorganismes: Genome et Environnement’, Clermont Universite, Universite Blaise Pascal, BP 10448, F-63000 Clermont-Ferrand, France, †CNRS, UMR 6023, LMGE, Aubiere F-63171, France, ‡INRA, UMR 42 Centre Alpin de Recherche sur les Reseaux Trophiques et Ecosystemes Limniques, F-74200 Thonon Les Bains, France
Abstract Studies on the active rare biosphere at the RNA level are mainly focused on Bacteria and Archaea and fail to include the protists, which are involved in the main biogeochemical cycles of the earth. In this study, the richness, composition and activity of the rare protistan biosphere were determined from a temporal survey of two lakes by pyrosequencing. In these ecosystems, the always rare OTUs represented 77.2% of the total OTUs and 76.6% of the phylogenetic diversity. From the various phylogenetic indices computed, the phylogenetic units (PUs) constituted exclusively by always rare OTUs were discriminated from the other PUs. Therefore, the rare biosphere included mainly taxa that are distant from the reference databases compared to the dominant ones. In addition, the rarest OTUs represented 59.8% of the active biosphere depicted by rRNA and the activity (rRNA:rDNA ratio) increased with the rarity. The high rRNA:rDNA ratio determined in the rare fraction highlights that some protists were active at low abundances and contribute to ecosystem functioning. Interestingly, the always rare and active OTUs were characterized by seasonal changes in relation with the main environmental parameters measured. In conclusion, the rare eukaryotes represent an active, dynamic and overlooked fraction in the lacustrine ecosystems. Keywords: activity, lakes, protists, rare biosphere Received 3 August 2014; revision accepted 13 February 2015
Introduction The application of high-throughput sequencing technologies has aided in the discovery of a tremendous microbial diversity and the existence of a rare microbial biosphere. For the first time, Sogin et al. (2006) have provided evidence of a myriad of bacterial OTUs found to exist in low abundance. These findings have raised questions about the ecological significance of these rare micro-organisms, and this has become one of the hottest topics in microbial ecology, with many questions remaining unanswered. The presence of rare taxa within microbes is not new, and it was observed for various protistan taxa identified by their morphology, but generally for specific taxonomic groups (e.g. ciliates; Dunthorn et al. 2014). The concept of rare biosphere coming Correspondence: Didier Debroas, Fax: +33 4 73 40 76 70; E-mail: [email protected]
from the molecular inventories depicting complex microbiological assemblages is quite different, with rare taxa potentially belonging to various taxonomic and functional groups and not only to a specific population. This highly diversified rare biosphere is generally associated with the idea of functional redundancy, that is a reservoir of dormant taxa that provides biological buffering capacity to withstand environmental changes (Caron et al. 2009). However, we still do not know to what extent rare taxa play significant roles in community function, that is as ribosomally active organisms important to the ecosystem function and not just when they become abundant (Bachy & Worden 2014). According to a previous study by Pedr os-Ali o (2012), rare taxa have been hypothesized to consist of a seed bank (i.e. dormant micro-organisms) that could potentially be resuscitated under favourable environmental conditions. Indeed, for bacteria, it was observed that many taxa switch between abundant and rare (Brown © 2015 John Wiley & Sons Ltd
A C T I V E R A R E P R O T I S T S I N L A K E S 1237 et al. 2005; Campbell et al. 2011). Such changes/rearrangements have also been observed in Archaea, for instance the clade MGII.a (Euryarchaeota) (Hugoni et al. 2013). Thus, some members of microbial communities are always present but vary in population size, switching from rare to abundant, due to environmental changes (Caporaso et al. 2012). However, the hypothesis of a seed bank is contradicted by some results: (i) some Bacteria and Archaea taxa are always rare, and a significant portion of these rare communities is actually active (Campbell et al. 2011; Hugoni et al. 2013) and (ii) the growth rates can decrease as abundance increases (Campbell et al. 2011). The role of rare and active microbes in terrestrial systems was indirectly observed via the manipulation of the trophic network because the reduction of rare soil microbes modified plant–herbivore interaction (Hol et al. 2010). Studies about the rare microbial biosphere are mainly focused on Bacteria and Archaea and fail to include the protists, which are involved in the main biogeochemical cycles of the earth covering all of the functional roles (phototrophs, parasites, saprotrophs, phagotrophs). The first attempt to study the dynamics of the rare protists was based on the analyses of DNA sequences obtained from the Sanger method (Caron & Countway 2009), and it is likely that the rare microbial biosphere was underestimated. Recently, the few high-throughput sequencing approaches in freshwater ecosystems demonstrated a stable predominance of a few highly abundant taxa at different temporal scales in lakes (Nolte et al. 2010; Mangot et al. 2013) and a biogeography of the rarest taxa, which represent a huge diversity of taxa (Lepere et al. 2013). As described for prokaryotes, the rare protists could be inactive (i.e. dormant) and become dominant with changing environmental conditions according to the model of Jones & Lennon (2010). These authors show that the variation of total phosphorus had no effect on the proportion of dormant protists that were detected by T-RFLP in lakes, and the transition between activity and dormancy could play a more important role in shaping bacterial communities than in eukaryotic communities. The specificity of the eukaryotic domain could be due to the ability to form resting stages or to the differential sinking rates. The hypothesis is that the switch between active and dormant stages could affect a low proportion of protists. Some protists could also be active as rare taxa, as demonstrated recently by Logares et al. (2014) in marine environments; however, to our knowledge, no other study has demonstrated the existence of rare and active protistan taxa. A low population density could be an advantage for some taxa because it could reduce parasitic infection, zooplankton predation, and could aid in the ability of that taxa to persist in the ecosystem. © 2015 John Wiley & Sons Ltd
Finally, there are various hypotheses on the rare microbial biosphere, which carries out keystone metabolic pathways and metabolizes rare energy sources. However, essential information to discuss the role of rare taxa is determining their level of activity, for instance by studying the structure community at both the DNA and RNA level (Jones & Lennon 2010; Campbell et al. 2011; Hugoni et al. 2013; Logares et al. 2014); such studies are still very rare. An overview of the dynamics and role of the rare microbial biosphere requires additional knowledge, particularly on the rare microbial eukaryotes whose dynamic and activity could differ from other microbes. We focused our study on small protists (0.2–5 lm) that can be considered to be an excellent model to study the rare biosphere in this kingdom by molecular approaches because the rank–abundance curve and the definition of the rarer fraction could be less biased than for larger eukaryotes. Indeed, the 18S gene copy number is correlated to genome size in eukaryotes (Prokopowich et al. 2003) and seems to be low in these small protists (0.1%). The data were normalized to have an equal number of rRNA and rDNA reads per sample in order to infer on the activity of eukaryotic taxa (rRNA:rDNA).
Taxonomic affiliation The representative OTUs were used to build phylogenetic trees, with FASTTREE (Price et al. 2010), for main taxonomic groups with reference sequences. These references were extracted to the SSUREF SILVA database (Pruesse et al. 2007) according to the following criteria: length >1200 bp, quality score >75% and a pintail value >50. In addition, the taxonomy of this reference database was modified to take account the typical freshwater lineages previously defined. The taxonomic annotation was conducted in two ways: the nearest neighbour (NN) and the last common ancestor (LCA) methods. A phylogenetic unit (PU) is a cluster of OTUs that is phylogenetically closest to a taxa from the reference database meaning that the cut-off is dependent on the closest relative in this database (Fig. 1). This process was implemented in the pipeline Phylogenetic Analysis of Nextgeneration AMplicons (PANAM http://code.google.com/ p/panam-phylogenetic-annotation/downloads/list) and described more in detail in the related study (Taib et al. 2013) and in the supporting information (Fig. S1, Supporting information).
Comparing the experimental OTUs with reference sequences Two methods were used for comparing the OTUs defined in this work with the 18S sequences in the database. First, OTUs were compared with the SSUREF SILVA © 2015 John Wiley & Sons Ltd
A C T I V E R A R E P R O T I S T S I N L A K E S 1239 package. To test the putative variables controlling the dynamics of OTUs, a canonical correspondence analysis (CCA) was conducted by linking the environmental parameters (temperature, oxygen, nutrients, chlorophyll a, abundances of Cyanobacteria, diatoms and Cryptophyta) selected by a forward analysis, to the OTUs (expressed in read abundances). This procedure was performed with the Vegan package (Dixon 2003).
Results Structure of the small protists communities
Fig. 1 Explanation of phylogenetic indices computed in this study from the trees generated from the experimental OTUs and reference sequences. In this example, 3 OTUs were gathered in a phylogenetic unit (PU) (e.g. PU_1) with a bootstrap equal to 99%. The nearest neighbour among the reference sequences in the tree, FJ153652, is affiliated to the Choricystis (Chlorophyceae), and this taxonomy is propagated to all the OTUs of this PU.
database (NR 117) restricted to the eukaryotes using UCLUST with a threshold of 95%. Second, various phylogenetic indices were computed from the trees generated (Swenson 2009) (Fig. 1). Phylogenetic diversity (PD) corresponds to the sum of branch lengths from the different PUs or all the OTUs to the root of the trees. Mean patristic distance (MPD) corresponds to the mean PD between each OTU and each other among the PUs. Mean nearest neighbour distance (MNND) is defined as mean phylogenetic distance from each species to its closest relative in the PUs. This index was computed by taking into account the reads abundance (MNND ab) or not (MNND pa). The ‘X depth/deeper’ is defined as the average distance between OTUs in the PUs and the closest reference sequence in the tree (Pommier et al. 2009). A higher MPD and MNND suggests a higher diversity within the PU, and a higher X depth/deeper suggests a higher phylogenetic distance between the PU and the closest relative in the reference database. These various indices were computed using R with the packages picante (Kembel et al. 2010), geiger and ape (Paradis et al. 2004).
Statistical analysis The relation between the proportion (%) of rRNA and rDNA determined for each OTU in the three fractions defined was tested by a linear regression. Linear models (regression and ANOVAs) were computed using R © 2015 John Wiley & Sons Ltd
All samples were processed together to display a general view of the protists community structure. After the cleaning procedures described in the Experimental Procedures and discarding the singletons and OTUs without rDNA reads, we defined 6446 OTUs delineating 686 PUs. In term of OTUs number, the main taxonomic groups were the Alveolata (1923 OTUs), Fungi (1564), Stramenopiles (1165) and Viridiplantae (776) (Table S1). This order changes if we consider the abundances of the rDNA reads; the Viridiplantae slightly dominated the Stramenopiles. Rank–abundance curves were created from the rDNA reads (Fig. 2), showing that only seven OTUs had an abundance >1% according to criteria used sometimes for defining the dominant ones (Pedr os-Ali o 2006). These OTUs belonged to three taxonomic groups: Haptophyceae, Chlorophyta and Fungi. The OTUs always rare and abundant (i.e. >0.1%) represented 77.2% and 2.4%, respectively, of the 6446 OTUs. In this abundant fraction, we did not detect any OTU belonging, for example, to Choanoflagellida, Euglenozoa or Rhodophyta, whereas OTUs from these groups were in the two other fractions.
Similarity and phylogenetic indices By clustering the OTUs (95%) with the SILVA database, we found that 26.9%, 8.2% and 2.5% of the abundant, cycling and always rare fractions, respectively, were included in a cluster with at least one sequence of the public database. In the always rare fraction, 13.9% of OTUs have an identity 1%). In the panel C, the outside of the circle corresponds to the active OTUs (OTUs with rRNA reads) and inside the inactive OTUs (including only rDNA reads).
(B)
(A)
Table 1 Similarity and phylogenetic indices determined on the different phylogenetic units (PUs). The mix PUs includes OTUs among different fractions always rare, cycling and abundant, whereas the other PUs include OTUs belonging only to one fraction
Num PUs Active PUs† BLAST
‡
MNND ab MNND pa MPD X depth/ deeper
Abundant PUs
Cycling PUs
Always rare PUs
Mix PUs
22 22
48 40
309 236
307 301
96.64a 0.14ac 0.17ac 0.30bcd 0.11ac
96.65a 0.25bc 0.25bc 0.25d 0.13bc
95.81 0.57 0.59 0.66ac 0.32
95.28 0.23ab 0.29ab 0.62ab 0.22ab
P*
