Figure 3.

Relationship between the parameters of the three-component model and water temperature derived from sorting the dataset and conducted a running fit of the model (bin size 60 measurements) with increasing temperature. Average water temperature of each bin is on the abscissa and 13.6 and 86.4 % (darker shading) and 2.5 % and 97.5 % (lighter shading) confidence intervals of the parameters from a bootstrap fit (1,000 iterations) on the ordinate (confidence intervals are constrained to realistic values, 0 to 10 for $C_{p,n}^m$ and $C_p^m$ and <1 for D_p,n and D_p). (A) Shows the relationship between temperature and the parameters D_p,n and D_p. Solid black line is the model of Brewin et al. (2017b) tuned to the data (Equation 8) for cells <2 μm and dashed line for cells <20 μm (Equation 7). (B) Shows the relationship between temperature and the parameters $C_{p,n}^m$ and $C_p^m$. Solid black line is the model of Brewin et al. (2017b) tuned to the data (Equation 6) for cells <2 μm and dashed line (Equation 5) for cells <20 μm.