CAPACIDADES MOTRICES

A 16S rRNA gene fragment amplicon possessing a unique base sequence has been defined as a “ribotype” (Section 1.2.3) and in this study it is the basic unit of taxonomy, phylogenetically indivisible for the purposes of studying bacterial diversity

(Ferrari and Hollibaugh, 1999). In this sense a ribotype differs from an operational taxonomic unit (OTU) which is commonly defined as a grouping of ribotypes sharing 97 % or more sequence homology (Giovannoni and Rappé, 2000; Pommier et al.,

2006).

There are several ribotypes, equating to discrete DGGE gel bands, that responded to increased pCO2 in mesocosms 1 and 2, and did not respond likewise in

ambient CO2 mesocosms 5 and 6. Those bands showing a decline in intensity in

successive samples in high CO2 mesocosms include band 10 in the

“Alphaproteobacteria” gel (Rhodobacteraceae, Figure 4.7a), band 23 (Nitrosococcus

sp., 100 % BLASTn similarity); and band 27 in the “Betaproteobacteria” gel

(Burkholderia, Figure 4.7b). Also temporally increasing in intensity were band 36

(SAR86 clone, 100 % BLASTn similarity, Pham et al., 2008), band 37 (Acinas et al.,

2004), band 42 (SAR86 clone, 99 % BLASTn similarity) and band 43 in the

“Gammaproteobacteria” gel (Figure 4.7c). No isolates were matched to these

sequences by BLAST, and in the Greengenes database they were all aligned with the genus Pseudomonas. Band 76 aligned with the Rhodobacteraceae, and with 88 %

similarity by BLASTn to Sulfitobacter, in the “Bacteria” gel (Figure 4.8d). The

organisms possessing all these ribotypes may be becoming less abundant, within the mesocosm, relative to others better able to respond positively to the increased pCO2.

Other ribotypes increase in abundance following the mid-term gas bubbling. Examples of these include those excised from the “Bacteroidetes” gel (Figure 4.7d):

band 72 (aligned with Brumimicrobium in the Greengenes database, and with

Marinoxanthimonasophiurae by BLASTn) and band 73 (Flavobacteriaceae,

specifically Olleyamarilimosa, 96 % BLASTn similarity). Band 65 - Mesoflavibacter

zeaxanthinifaciens in the phylum Bacteroidetes with 90 % BLASTn similarity - from

the “Planctomycetes” gel (Figure 4.8c). Bands increasing in intensity, or becoming

present, with time include three from the “Bacteria” gel (Figure 4.8d): band 81

(weakly aligned with the Sphingobacterium genus, but having greater similarity to an

Algibacter according to BLASTn), band 101 (Polaribacter), and 106 (the

FlavobacterialesWinogradskyella sp., 90 % BLASTn similarity), band 111

(Algibacter sp., 93 % BLASTn similarity) and 114 (Flavobacterium sp., 94 %

BLASTn similarity). These last two are only classified as far as the family

Flavobacteriaceae by Greengenes database alignment. The organisms possessing

being more refractory than others to the fall in pH. Alternatively they may have responded well to the organic carbon “windfall” caused by gas bubbling.

There are other ribotypes in each group that are dominant throughout the experiment, unaffected by either elevated CO2, or gas bubbling. These include those

from bands 11, 12, 15, 16, 17 and 18 (all with > 98 % sequence homology to

Synechococcus sp., Cyanobacteria, Figure 4.8, Table 4.4), 25 and 36 (Bacillus

pseudofirmus, 98 %, Firmicutes, Table 4.4). Synechococcus sp. may be small enough

to resist the cavitational shock of bubbling and, as already mentioned, they may be growth-limited in this setting by N or P, rather than inorganic C.

There are limitations to identifying the organism from which a ribotype originates by means of a BLASTn search (Ashelford et al., 2005). Firstly the

BLASTn results are defined by the query base sequence, the ribotype, which, as well as varying in length, is subject to the biases of the PCR listed above in Section 4.5.3.1 and discussed fully in Chapter 6. The longer a ribotype sequence is the more

taxonomic information it can yield about the organism from which it was amplified. The ribotype sequences used here for BLASTn queries were generally short (~ 180 bp). However, the group-specific primer pairs were originally designed to produce BLASTn-appropriate length DGGE-ready PCR products with GC-clamps.

Subsequent performance of a nested PCR on this template necessarily reduced the size of ribotype sequences for DGGE. However there is a balance to be struck, as both PCR and DGGE are compromised with longer sequences (Neufeld and Mohn, 2006).

The results of a BLASTn search are also dependent on the quantity and quality of the homologous sequences residing in the database. Submission to public

repositories is subject to limited quality control, with the result that many sequences may be of low fidelity, for example those containing primer sequences, and those that are chimaeric constructs (Ashelford et al., 2005; Neufeld and Mohn, 2006). The

homology between a query sequence and its closest aligned database equivalent is an aspect of BLASTn data which needs to be considered carefully. Characteristics attributed to the database ribotype are not necessarily attributable to the query ribotype, especially below the different arbitrary sequence homology cut-off points for bacterial “species.” These are 94 % sequence similarity, equivalent to 70 % DNA- DNA hybridisation (Wayne et al., 1987), the more widely accepted 97 % sequence

similarity (Venter et al., 2004), and the strict 99 % sequence similarity unless niche-

The alignment of sequences in the Greengenes database, performed for

sequences in Tables 4.2, 4.3, and 4.4, is more reliable than a simple association with a sequence that aligns closest following BLASTn analysis. With Greengenes, sequences are placed in a phylogenetic tree according to maximum parsimony. The classification obtained is only given to the lowest attributable taxonomic level (e.g.: class, order, family, or genus). The nearest cultured isolate imparts a weighting to the classification proportional to its phylogenetic distance from the query sequence. The results are thus more robust than those sometimes obtained through BLASTn alignment.

Although many taxonomic demarcations in 16S rRNA gene bacterial

phylogenetics are arbitrarily based on sequence differences, and there is clearly a need for quality policing of databased sequences, the system is currently the only basis for widespread application of bacterial phylogenetics. For the time being the approach used here to acquire 16S rRNA gene sequences, and to infer from their analysis patterns and developments within in situ bacterial communities, is valid.