SEARCH

SEARCH BY CITATION

Keywords:

  • Phycodnaviridae;
  • marine viruses;
  • North Atlantic;
  • virus phylogeny

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

Studies of the Phycodnaviridae have traditionally relied on the DNA polymerase (pol) gene as a biomarker. However, recent investigations have suggested that the major capsid protein (MCP) gene may be a reliable phylogenetic biomarker. We used MCP gene amplicons gathered across the North Atlantic to assess the diversity of Emiliania huxleyi-infecting Phycodnaviridae. Nucleotide sequences were examined across >6000 km of open ocean, with comparisons between concentrates of the virus-size fraction of seawater and of lysates generated by exposing host strains to these same virus concentrates. Analyses revealed that many sequences were only sampled once, while several were over-represented. Analyses also revealed nucleotide sequences distinct from previous coastal isolates. Examination of lysed cultures revealed a new richness in phylogeny, as MCP sequences previously unrepresented within the existing collection of E. huxleyi viruses (EhV) were associated with viruses lysing cultures. Sequences were compared with previously described EhV MCP sequences from the North Sea and a Norwegian Fjord, as well as from the Gulf of Maine. Principal component analysis indicates that location-specific distinctions exist despite the presence of sequences common across these environments. Overall, this investigation provides new sequence data and an assessment on the use of the MCP gene.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

Phycodnaviridae are eukaryotic algal viruses that have large ∼100–560 kb genomes, 100–220 nm diameter capsids and have only been studied in recent decades (Brussaard, 2004). The activities of marine Phycodnaviridae are particularly important when considering their top-down potential as predators of photosynthetic hosts (Wilhelm & Suttle, 1999), which have a dramatic global-scale influence on biogeochemical cycles (Redfield, 1958; Fuhrman, 1992; Falkowski et al., 1998; Field et al., 1998). In general, aquatic viruses are known to play a variety of roles and are key members of aquatic ecosystems (Suttle, 2007; Brussaard et al., 2008; Wilhelm & Matteson, 2008).

Commonly, researchers have focused on the ‘total’ abundance and the total production rate of viruses in aquatic systems, while virus community composition has been examined only through a series of a few conserved molecular markers. To a vast extent, this is due to the need for more genomically characterized virus–host model systems to use as a basis for probe development. Other issues, including low concentrations of virus DNA, avoidance of restriction enzyme activity due to viral DNA modifications, and difficulties in cloning sequences that encode products detrimental to the cloning hosts, have also hindered efforts (Paul et al., 2002).

In spite of these challenges, researchers have made significant advances in the last decade towards resolving aquatic viral diversity. An examination of virus metagenomes from four different ocean environments has suggested that the viral population of the world's oceans is universally distributed, but that regional selection factors control the overall composition (Angly et al., 2006). Additionally, RNA–virus population metagenomics (Culley et al., 2006) and virioplankton genetic diversity (Comeau et al., 2006) have also been investigated and yielded similar conclusions.

To date, most advances in our understanding of virus diversity, however, have been made by exploiting conserved functional genes associated with certain virus families. The phylogenetics of the Phycodnaviridae have been primarily based on the DNA polymerase gene (DNA pol) sequence (Chen & Suttle, 1995; van Etten, 2006). However, because DNA pol cannot always resolve phylogenetic needs (Sandaa et al., 2001; Schroeder et al., 2002), the major capsid protein (MCP) gene has recently been examined as a potential biomarker for Phycodnaviridae phylogeny. For Emiliania huxleyi-infecting Phycodnaviridae [E. huxleyi viruses (EhV), genus Coccolithovirus], which were isolated from the Western English Channel, it was found, based on the host-range of the individual isolates, that MCP gene sequences better resolved virus isolates than DNA pol (Schroeder et al., 2002). It was proposed that the MCP gene could be a useful second biomarker for resolving genotypes, especially because it appears to be more varied in EhV than DNA pol (Schroeder et al., 2002). In addition, examinations of natural viral assemblages from Raunefjorden and Puddefjorden, and of virus isolates of different non-EhV Phycodnaviridae members, indicated that the MCP gene could also be considered as a genetic marker for Phycodnaviridae (Larsen et al., 2008). Moreover, while several studies (Chen et al., 1996; Schroeder et al., 2002; Short & Suttle, 2002; Brussaard et al., 2004; Larsen et al., 2008) have now described the phylogenetics of Phycodnaviridae sampled from coastal and offshore waters, ocean-wide diversity of eukaryotic algal viruses (namely Phycodnaviridae) has yet to be examined.

Our goals for the current study were to expand our knowledge of MCP gene diversity and assess the use of MCP as a biomarker for EhV in the open ocean. The diversity of EhV was examined in the North Atlantic to assess spatial changes across >6000 km of open ocean. This diversity was compared with previous coastal and isolate examinations (Schroeder et al., 2002), newly obtained sequences from the Gulf of Maine, and to amplicons from lysates generated after exposing virus concentrates from the North Atlantic to host cultures in the laboratory.

Materials and methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

Sample collection

Virus concentrates were generated while onboard the RV Seward Johnson during the NASB 2005 cruise (Rowe et al., 2008b) as described by Wilhelm & Poorvin (2001). Figure 1 shows the location of these stations. Water samples (20 L) were prefiltered through 142-mm-diameter GF/F filters (Whatman) and then through 0.45- and 0.22-μm nominal pore-size, 142-mm-diameter polycarbonate filters (Millipore). The virus-size fraction was concentrated to ∼500 mL using Millipore's M12 ProFlux tangential flow filtration with a 30-kDa cutoff filter cartridge (Millipore). Virus concentrates were stored in the dark at 4 °C until their use in the laboratory.

image

Figure 1.  Station locations in the North Atlantic from where virus concentrates were generated.

Download figure to PowerPoint

To expand our spatial comparison of viral diversity, we also obtained MCP sequences from the Gulf of Maine that had been prepared using prior protocols (Martinez-Martinez et al., 2007). Samples were collected during transects in the Gulf of Maine from Portland, ME to Yarmouth, Nova Scotia (Canada), between August 2007 and September 2008.

Lysate generation

Virus concentrates from the North Atlantic were screened for viruses infectious to E. huxleyi strains CCMP373 and DWN 61/87/10 (DWN61). Cultures were grown in L1 medium (Guillard, 1975; Guillard & Hargraves, 1993) at ∼21 °C on a 12/12 day/night cycle. Cultures were inoculated with virus concentrate and in vivo fluorescence was monitored for a minimum of 14 days using a Turner Designs TD-700 fluorometer equipped with an in vivo chlorophyll a (chl a) filter set (ex. λ=340–500 nm; em. λ≥665 nm). The presence of a lytic agent was deemed positive when significant decreases in fluorescence (relative to unexposed controls) were observed (Rowe et al., 2008a). MCP genes were amplified and clone libraries generated for sequence analysis for the following: all 20 North Atlantic virus concentrates, the lysates generated from inoculating host cultures with those same concentrates, and water samples taken from the Gulf of Maine.

DNA extraction

For the North Atlantic samples, viral DNA was extracted from concentrates and lysates following a protocol adapted from Smith & Burgoyne (2004). First, the salt concentration was reduced by placing a sample on top of a premoistened 0.025-μm nominal pore-size, mixed cellulose filter (Millipore) and incubated at room temperature for 30–40 min. Sample was then spotted onto a FTA CloneSaver card (Whatman) and dried for 40 min at room temperature. Approximately 2.0-mm-diameter samples were punched from the sample dried on the FTA card and washed twice with 500 μL of 10 mM NaOH for 15 min, then twice with 200 μL of Tris-EDTA (pH 8) for 5 min, and finally with 500 μL of 1 : 5, Tris-EDTA : bovine serum albumin (1 mg mL−1) for 10 min. Punches were dried for 10–15 min at 56 °C. DNA was eluted by submersing punches in 30 μL of sterile water and incubating for 10 min in a water bath at 90 °C. Samples were briefly centrifuged at a high speed and then stored at −20 °C. FTA cards were stored at room temperature, away from light, and with a desiccant.

For the Gulf of Maine samples, 1-L seawater samples were filtered onto 0.45-μm pore-size Supor-450 47-mm-diameter filters (Pall Corp.). This method collects both free viruses as well as cells infected by viruses. The filters were transferred to 2-mL cryotubes, snap frozen in liquid nitrogen, and stored at −80 °C. Genomic DNA was isolated using a phenol/chloroform method, where filters were cut into small dissolvable pieces and placed in a 2-mL microcentrifuge tube (Eppendorf). Following the addition of 800 μL of GTE buffer (50 mM glucose, 25 mM Tris-Cl pH 8.0, and 10 mM EDTA), 100 μL of lysozyme (10 mg mL−1), and 100 μL of 0.5 M filter-sterilized EDTA, samples were incubated at room temperature for 1–2 h with gentle shaking. Sodium dodecyl sulfate was added (200 μL of a 10% stock) and the mixture was incubated for 10 min at room temperature. DNA was then purified by phenol extraction. Air-dried DNA pellets were resuspended in 50 μL of TE buffer.

PCR

The MCP gene was amplified from extracted North Atlantic samples using the primers MCP-F and MCP-R as described by Schroeder et al. (2002), yielding amplicons of ∼238 bp. Annealing temperature was optimized at 65 °C and a final extension step of 74 °C for 5 min was included to facilitate cloning. Reaction mixes were set up using PuRe Taq Ready To-Go PCR beads (GE Healthcare), with ∼1–3 μL of DNA, 20 pmol primer concentration, and sterile water up to 25 μL. Reaction mixes minus template served as negative controls for all PCRs. All PCRs were performed on an Eppendorf Mastercycler Gradient thermocycler, with the product stored at 4 °C until electrophoresis: 2% agarose gel in 1 × TBE buffer at ∼100–120 V, with subsequent visualization and documentation on a Fotodyne Investigator digital imager with ethidium bromide filter using foto/analyst®pc image v. 9.0.4 software.

Gulf of Maine MCP amplicons were obtained in two stages using two pairs of MCP-specific oligomers (Table 1 in Martinez-Martinez et al., 2007). Initial amplification was conducted using the primer pair MCP-F1/MCP-R1, and nested PCR of the resulting product was conducted with MCP-F2/MCP-R2. Denaturing gradient gel electrophoresis of the second-stage nested PCR product and sequencing of excised bands were conducted as described in Schroeder et al. (2003).

Table 1.   Measurements of the total chl a concentrations, total viral abundance, salinity, and temperature that were observed at each of the stations from where virus concentrates were generated
Sampling locationTotal chl a (μg L−1)Total viral abundance (VLP L−1× 109)SalinityTemperature (°C)
  1. Virus concentrates were generated from surface (5 m) water samples, unless otherwise specified. VLP, virus-like particle; DCM, deep chlorophyll max; ND, not determined.

10.441.1NDND
2 (25 m, DCM)NDNDNDND
30.561.735.3513.34
40.84ND35.4113.64
5 (10 m)0.792.035.4213.46
60.672.335.3611.89
71.012.135.3311.59
81.273.035.3011.07
92.961.035.2711.13
101.781.235.2810.66
112.131.135.2710.89
120.751.335.2011.34
13NDND35.1711.93
141.721.235.2511.26
151.861.435.208.99
161.082.435.1910.12
171.613.035.139.72
182.324.234.719.46
192.102.233.282.6
201.668.134.889.3

Cloning and sequencing

For North Atlantic samples, bands of interest were excised from agarose gels and DNA eluted using Promega's Wizard SV gel and PCR clean-up system. Eluted DNA was cloned using Invitrogen's TOPO TA cloning kit and the resulting plasmids were extracted and purified using the QIAprep Spin Miniprep Kit from Qiagen. In total, 151 quality sequences from the amplicons were determined at the Clemson University Genomic Institute. Sequences were deposited in the NCBI under accession numbers FJ543391–FJ543465.

Sequence and phylogenetic analyses

North Atlantic sequences were first assessed for quality and clarification of read, and primers were removed using 4peaks. North Atlantic sequences were initially aligned with clustalx 2.0 and Bayesian inference trees with the following settings: HKY85 nucleotide substitution model, unconstrained branch lengths, a subsampling frequency of 200, and 1.1 × 106 generations (Huelsenbeck & Ronquist, 2001) were built using geneious 4.0.4 (Drummond et al., 2009), with known sequences (Schroeder et al., 2002). Phylogenetic reconstructions were also completed for the entire 238 bp amplicons using the neighbor-joining method in bioedit (ver 7.0.9.0, Hall, 1999) to compare the results of model reconstructions. Rarefaction curves were generated using the analytical rarefaction program provided by Steven Holland (University of Georgia, Athens), available at http://www.uga.edu/strata/software

To examine the relationship between different communities based on sequence, as well as to compare the sequences derived from lysates with those directly amplified from the water column, we obtained previously published sequences related to a conserved 99-bp region of the MCP (Schroeder et al., 2002) and also included sequences from the Gulf of Maine (this study). Sequences were realigned and phylogenetic reconstructions were created for the community using the neighbor-joining method in bioedit (ver 7.0.9.0, Hall, 1999). Comparisons between stations were completed using the PCA tool within the unifrac analysis service (http://bmf2.colorado.edu/unifrac/index.psp) as described previously for weighted environments (Lozupone et al., 2006, 2007). The robustness (node reoccurrence as a percentage of resamplings) was determined using the Jackknife analysis tool in the unifrac service. The Chao 1 statistic was calculated for the combined data set as per Kemp & Aller (2004).

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

Table 1 presents an abbreviated list of data ranges for chl a, total viral abundance, salinity, and temperature (Rowe et al., 2008b, and this study) observed at the sampling stations shown in Fig. 1. The results highlight an increase in phytoplankton biomass (as estimated by chl a concentrations) in conjunction with a decrease in temperature during the northward transect. The total virus densities ranged from ∼1 to 8 × 109 L−1. Sequence information was used to generate rarefaction curves, with all libraries grouped as one to assess our overall sampling of the viral population across the North Atlantic: nucleotide sequences (Supporting Information, Fig. S1a) and amino acid sequences (Fig. S1b), with sequences grouped by 99.5% and 98% identity, respectively. Both curves demonstrate a positive linear trend without ‘leveling off,’ indicating that a saturated sampling of richness was not reached. Rarefaction analyses for individual libraries also failed to reach saturation (data not shown). This trend reflects the sampling pattern observed, wherein most sequences were detected only once, a few detected <5 times, and a few more detected multiple times (≥10). This included a number of sequences (12) that were an identical match for the well-studied virus isolate EhV-86. These results lead to an overall estimate of ∼280 different OTUs (a Chao 1 estimate of 279.5) within the samples.

As suggested by initial Bayesian inference phylogenetic trees (not shown), phylogenetic reconstruction using the neighbor-joining approach demonstrates that much of the diversity within our samples fell outside the range identified in previous studies (Fig. S2). To better compare our new sequences with those determined previously, we completed a principal component analysis of reconstructed phylogenies for each set of amplicons (Fig. 2). More than 86% of the variation between the amplicons was explained by the first two principal components, which effectively separated sequences from the open ocean (the NASB cruise) from the sequences determined in the North Sea, a Norwegian Fjord, and the Gulf of Maine. Amplicons from virus lysates generated in the laboratory grouped exactly with the environmental amplicons from the same location of collection (CCMP1516 with amplicons directly from fjord samples, DWN61 and CCMP373 with amplicons directly from North Atlantic waters).

image

Figure 2.  Principal component analyses of samples obtained by PCR amplification of the 99-bp conserved region of the MCP from viruses infecting Emiliania huxleyi. Libraries for PCR amplicons generated directly from seawater [GOM, Gulf of Maine (this study); NASB, North Atlantic Sea Bloom (this study); NF, Norwegian Fjord; NS, North Sea]. Sequences of isolated viruses from Norwegian Fjord propagated on a specific host (CCMP1516) were taken from Schroeder et al. (2002). Comparisons with PCR-amplified MCP sequences from lysates of E. huxleyi DWN61 and CCMP373 samples (from the North Atlantic, this study) demonstrate that amplification on a host does not change the diversity of the community as determined by MCP sequencing. Accumulated variation in the first two components (86.07%) demonstrates populations separating by geographic location (with both direct PCR- and host-amplified viruses clustering most closely in both components).

Download figure to PowerPoint

To determine the robustness of the groupings, we completed a Jackknife resampling of the data in order to determine how often groups clustered when subsets of the data were sampled (Fig. 3). Consistently (100% of all resamplings), the Gulf of Maine amplicons grouped away from those of other amplicon libraries, implying that there was a distinct virus population in that region relative to the other locations. The relationship between the amplicons generated directly from the open North Atlantic waters (NASB) and from virus lysates of the E. huxleyi hosts was also consistent, implying that the direct PCR amplification did not augment the richness of the samples generated by the lysis of the host cells.

image

Figure 3.  Jackknife analyses of clusters (using five of eight libraries for 1000 resampling permutations). Percent reoccurrence within sampling given at nodes as determined using the fast unifrac service. Sequences were directly amplified from the environment (NASB and GOM, this study; NF and NS, Schroeder et al., 2002, 2003) or first from lysates (DWN61, CCMP373 – this study) or purified viruses (CCMP1516, Schroeder et al., 2002).

Download figure to PowerPoint

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

The current study serves several purposes independent of our ability to saturate rarefaction. Firstly, it has confirmed the utility of the MCP gene as a valuable phylogenetic marker within a group of viruses infecting the cosmopolitan phytoplanktor, E. huxleyi. Our results suggest that the EhV community is dominated by several clones that are represented by a high abundance of conserved sequences. The results also demonstrate, however, that the community contains numerous unique viral MCP sequences, implying that significant genetic richness persists. The identification of MCP sequences from viral lysates of laboratory cultures that reside in previously unrepresented clades suggests that the diversity of these viruses is much broader than seen in the original, more regional, studies, and documents that this diversity is real and functional. In total, the work validates the use of the MCP gene marker for studies of EhV richness. It also provides an insight into the distribution of virus communities across different environments of the Atlantic Ocean, suggesting that there may be regional differences in EhV populations.

In the current study, we detected a total of 35 unique (only seen once) sequences across all stations, 17 of which appeared from direct amplification of environmental samples, with the rest arising from the lysates of host cells. That the viruses in these lysates were brought into culture demonstrates they are functional, broadly enriching the known phylogeny of the group. In contrast, several MCP gene sequences were detected multiple times. For example two sequences were detected 10 and 37 times within the environmental samples, but a total of 31 and 56 times, respectively, when environmental and culture-generated lysate samples were combined. Their detection multiple times suggests that they may be quantitatively important representatives of the natural population (at least at the time of sampling). Such a disproportionate evenness within our libraries, especially when considering that sampling occurred during the North Atlantic Spring Bloom event, suggests that these sequences were from viruses that may have recently infected resident host species, likely bloom-formers high in abundance, in a manner in concordance with the ‘kill the winner’ hypothesis (Thingstad & Lignell, 1997). Moreover, the distributions of sequences from both sets of samples (virus concentrates and lysates) agree with previous observations suggesting that environmental conditions select from a universal ‘pool’ of viruses (Angly et al., 2006). Rarefaction curves for both DNA and protein sequences (Fig. S1a and b) grouped by 99.5% and 98% identity, respectively, demonstrate, however, that our sampling efforts did not fully cover the viral richness available, suggesting that there is likely more sequence diversity to be sampled. It is possible that lower percentages of identity for OTUs may have increased the probability of fully covering sample richness; however, given the relatively new use of MCP as a marker, we felt that the higher (more conservative) values were more relevant.

When comparing amplicons generated directly from the North Atlantic with those generated from lysates, only four sequences were common between these sets of MCP gene sequences. These results are in agreement with the rarefaction curves (i.e. true richness in this region is higher than sampled). Examinations of aquatic viral diversity have suggested that changes can occur rapidly (Bratbak et al., 1990; Steward et al., 2000; Marston & Sallee, 2003) and we felt it would be likely that exposure to specific hosts may select for the production of specific viruses (given the assumption that there are degrees of resistance between related viruses and hosts, with respect to cross-infectivity – e.g., see Table 2 in Schroeder et al., 2002). However, analysis of the richness within the communities by the unifrac service provides results to the contrary: whether amplified through hosts cells or not, amplicon libraries clustered by source water geography. Phylogenetic differences between natural populations and laboratory culture collections are well documented for viruses (Suttle, 2005), but these are primarily based on the premise that all PCR amplicons represent actual functional viruses. The many sequences detected by direct amplification from our virus concentrates, but not from the lysates, serve as a reminder that host ranges of EhV in the North Atlantic may not be fully represented by the range of E. huxleyi strains in culture and that the choice of host strain may have a considerable impact on the selection of viruses amplified. This observation also highlights one important caveat: PCR products from in situ seawater amplifications may be from virus particles that are noninfectious/nonfunctional. Although virus infectivity decays quickly in seawater (Weinbauer et al., 1999; Wilhelm et al., 2003), it is not known how long the noninfectious particles (and nucleic acids within them) last before they decay to a point where molecular targets cannot be PCR amplified. Moreover, the frequency with which nonfunctional particles are produced during the lytic process is also unknown. As such, microheterogeneity within the virus community must be considered with caution. Thus, it is important that culture-based techniques continue to confirm the viability of phylogenetic reconstructions via the specific testing of known infectious particles.

In addition to assessing the diversity of EhV MCP gene sequences across the North Atlantic and between natural samples and lysates, comparisons were made between open and coastal ocean waters. MCP gene sequences from EhVs isolated in the English Channel (UK) and Raunefjorden (Norway) (Schroeder et al., 2002) were included in phylogenetic analyses, as was a new set of environmental amplicons generated from the Gulf of Maine. It is interesting to note that of these 10 coastal isolates, only one, EhV-86, was represented in the North Atlantic nucleotide sequences, and was only found in the lysate samples that were generated. Examination of these communities clearly demonstrates, however, that, while there are similarities, there are also sufficient differences so that the libraries are statistically distinct from the open ocean NASB amplicons and lysate isolates. To this end, while we see to some degree that while many sequences are represented at all locations represented in this study, there are sufficient qualitative and quantitative attributes to this richness that makes the communities unique.

Although it is not unexpected that several sequences detected from the open ocean are similar to those from coastal waters of the same ocean, it is very clear that the richness of EhV is higher than previously known. The continued use of laboratory-dependent and -independent techniques, in conjunction with the impending development of higher throughput and quantitative molecular tools coupled to informatics, suggests that in the near future, we will be in a position to carefully tease apart the dynamics of these host pathogen communities across previously unprecedented spatial scales.

Acknowledgements

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

We thank G. DiTullio for providing E. huxleyi strains. We also thank the captain and crew of the RV Seward Johnson, scientific colleagues onboard during NASB 2005, G. LeCleir and A. Buchan for recommendations, P. Dubus (University of Bordeaux) for his advice on DNA extraction, I. Gilg for technical help with the Gulf of Maine samples, and J. Troutman for persistence in sequencing. This work was supported by grants from the National Science Foundation (OCE 0452409 and EF 0949120) to S.W.W. and (EF 0723730) to W.H.W.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

Supporting Information

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

Fig. S1. Rarefaction curves generated for (a) nucleotide sequences observed from total samples after sequences ≥99.5% identity were grouped together, (b) amino acid sequences observed from total samples after sequences ≥98% identity were grouped together.

Fig. S2. Neighbor-joining reconstruction (unrooted) of phylogenies for MCP amplicons.

Please note: Wiley-Blackwell is not responsible for the content or functionality of any supporting materials supplied by the authors. Any queries (other than missing material) should be directed to the corresponding author for the article.

FilenameFormatSizeDescription
FEM_1055_sm_suppl_figure-s1-s2.doc52754KSupporting info item

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.