Annals of Botany

Data mining reveals silicon extractors

The paper of Hodson et al. (Oxford Brookes, Warwick and Nottingham Universities, UK, pp. 1027–1046) illustrates clearly that, on occasions, new insights may be gained from comprehensive analysis of previously published data. They point out that silicon is one of the most abundant elements in soil; a small proportion of this Si exists as soluble silicic acid and many plants are able to take up Si from this source. It is already known that there is variation between different plant groups in respect of Si accumulation but the authors have provided a much more detailed picture of that variation. To make comparisons meaningful, strict criteria were applied to the selections from the primary literature: the data came from papers ‘that reported Si concentrations… in at least two species growing in the same environment and which contained at least one species in common with another study’. The data were subjected to a residual maximum likelihood (REML) analysis to ‘adjust for differences in between-study variances and means’. The authors were then able to rank plant groups according to their shoot Si content. Of the groups analysed, ferns have the lowest Si content (a mean of <1 % of dry weight); just above them are the gymnosperms (approx. 1 %) and then the angiosperms (approx. 1.5 %). At the other end of the scale are the horsetails (approx. 3.5­%) and liverworts (approx. 5.5 %). Further analysis of the angiosperm data revealed extensive variation, both within the dicots and the monocots, with the highest values being seen in two commelinoid monocot orders, the Arcales and Poales, the latter point being clearly relevant to agricultural use of cereals and other grasses. Further, the likelihood that, in the near future, a gene encoding an Si transporter will be identified raises the possibility of manipulating plant Si content, an endeavour that will be aided by this analysis.