Results from CART analysis of bacterial production rates (mgC m<sup>−3</sup> h<sup>−1</sup>, thymidine method) K SorrellBrian HawesIan SafiKarl 2013 <p><strong>Figure 4.</strong> Results from CART analysis of bacterial production rates (mgC m<sup>−3</sup> h<sup>−1</sup>, thymidine method). Threshold values of predictors are shown below each split, and variance explained for predictors is shown above each split. Mean values ±1SD and the number of samples (<em>n</em>) are shown for each subset of data, with the three terminal groups identified by numbers in <strong>bold</strong>. The total variation explained by this CART model was 56%. Scatterplot of relationships of BPthy and BPleu to PP, the only significant parameter selected by regression tree analysis for both methods (CART for BPleu not shown). The vertical lines identify the values of PP that best explained variation in BPthy (dashed lines) and BPleu (<strong>bold</strong> dotted line).</p> <p><strong>Abstract</strong></p> <p>We compared planktonic primary and secondary production across twenty meltwater ponds on the surface of the McMurdo Ice Shelf in January 2007, including some ponds with basal brines created by meromictic stratification. Primary production ranged from 1.07 to 65.72 mgC m<sup>−3</sup> h<sup>−1</sup> in surface waters. In stratified ponds primary production was always more than ten times higher in basal brines than in the corresponding mixolimnion. Regression tree analysis (<em>r</em><sup>2</sup> = 0.80) identified inorganic nitrogen (as {\mathrm{NH}}_{4}^{+}) as the main factor limiting planktonic primary production. However, there was also evidence of inorganic carbon co-limitation of photosynthesis in some of the more oligotrophic waters. Neither C nor N limited carbon fixation at [NH<sub>4</sub>–N] > 50 mg m<sup>−3</sup>, with photoinhibition the factor most likely limiting photosynthesis under such conditions. Primary production was the only factor significantly correlated to bacterial production and the relationship (<em>r</em><sup>2</sup> = 0.56) was non-linear. Nitrogen limitation and tight coupling of planktonic primary and bacterial production is surprising in these ponds, as all have large pools of dissolved organic carbon (1.2–260 g m<sup>−3</sup>) and organic nitrogen (all >130 mg m<sup>−3</sup>). The dissolved pools of organic carbon and nitrogen appear to be recalcitrant and bacterial production to be constrained by limited release of labile organics from phytoplankton.</p>