El cine propagandístico

The SST diagnostic experiments run during the Mediterranean model development (section 4.4.1) indicated that the Δ47-G. ruber (w) SST reconstruction (Rodríguez-Sanz

et al., 2017) best represents monsoon season SST in the eastern Mediterranean during S5. This strongly supports that G. ruber (w) inhabited a thinner, fresher layer at the top of the summer mixed layer (SML) during the monsoon season. In this case, δ18Oruber reflects the

full extent of both the direct and indirect influences of increased freshwater influx to surface waters. Directly, the introduction of a large volume of isotopically light freshwater into surface waters dilutes seawater and lowers its δ18_{O. Indirectly, the} increased buoyancy of fresher surface waters acts to increase stratification so that vertical

mixing is inhibited and freshwater is concentrated in thin, less saline surface layers at the top of the SML. It has been proposed that these layers are susceptible to a ‘temperature concentration’ effect, where solar insolation excessively warms thin, isolated surface layers (Emeis et al., 2003). This inference was supported by the difference between the Δ47-G. ruber (w) and the UK’37-based SST reconstructions (Fig. 5.1c; Rodríguez-Sanz et

al., 2017). The higher temperatures in the fresher surface waters further lower the δ18O of calcifying foraminifera through their impact on water-to-calcite isotope fractionation. This hypothesis implies that during times of increased freshwater influx, the SML stratifies into a fresher ‘upper summer mixed layer’ (upper SML), and a ‘lower summer mixed layer’ (lower SML) underlying it.

In my box model, excess freshwater runoff input during S5 is defined by the volume and δ18_{O of monsoonal runoff (section 4.3.3). The freshwater runoff effect on seawater δ}18_O of the fresher, surface layer is controlled via the defined depth of the upper SML (i.e., the volume into which freshwater input is initially mixed). The resultant seawater δ18_{O and} SST during the monsoon season (i.e., the temperature of the upper SML) are then used to calculate the expected δ18O for calcifiers in the upper SML (i.e., G. ruber (w)) over the S5 interval. In the model, the excess freshwater runoff, SST, and depth of the upper SML during the monsoon season are distinct, separately defined parameters. However, in the real world, these parameters would be considerably inter-related. Of these, SST is the only parameter quantified for S5, while monsoonal runoff volume and depth of the upper SML have only been loosely approximated using a simpler, earlier version of the model applied here (Rohling et al., 2004). Using observations made in the model development (Chapter 4), I can start to define some of the relationships between these parameters. As inferred from the SST diagnostics (section 4.4.1), while the Δ47-G. ruber (w) SST record (Rodríguez-Sanz et al., 2017) reflects the SST of the upper SML, the UK’37-based SST with a superimposed seasonal deviation (+3°C) (Rohling et al., 2002, 2004; Marino

et al., 2007) reflects the lower SML temperature. Building on this concept of Mediterranean surface water stratification during S5, I calculate the difference between the Δ47-based and UK’37-based (+3°C summer deviation) SST records over the study interval (ΔT; Fig. 5.1d). The calculated ΔT approximates the difference in temperature between the upper and lower SMLs.

As ΔT increases, I interpret this as a greater temperature concentration effect in the upper SML, i.e. a thinner, fresher, more stratified layer over which insolation warming is concentrated. Hence, I set up a correlation between ΔT and the depth of the upper SML (z), which allows the upper SML depth to be estimated throughout S5 from the SST data (Eq. 5.1; Fig. 5.1e):

𝒛 = (−𝟐. 𝟕𝟖𝟖𝟔) × ∆𝑻 + 𝟑𝟎 (Equation 5.1) This relationship assumes that if ΔT = 0, when both the Δ47 and UK’37 records show the same SST, there is no separation of the SML into an upper and lower SML (i.e., the SML is one homogeneous water mass). This is due to theΔ47-G. ruber (w)exhibiting the same temperature as the lower SML (UK’37 +3°C), suggesting that G. ruber (w) no longer has the fresher surface water niche to exploit, and is residing in the main body of the SML. Hence, when ΔT = 0, the depth of the upper SML is set to 30m, i.e., same as the full SML depth. However, as ΔT increases, the depth of the upper SML is set to decrease, representing an increasing stratification within the SML, and a shift in G. ruber (w)’s depth habitat towards the surface. The maximum ΔT observed (10.22°C; Fig. 5.1d, upper 95% confidence interval) is taken to correspond to an upper SML depth of 1.5m. A linear relationship is then assumed between these two end members.

A minimum upper SML depth of 1.5 m was originally selected as Rohling et al. (2004) calculated 1.5 m as the shallowest upper SML depth before salinity becomes too low to be able to sustain foraminiferal life (~22 p.s.u). In the box model, salinities do not reach this low for an upper SML depth of 1.5 m, as evaporation over a very thin, warm surface layer removes much of this freshwater. However, reducing upper SML depths to less than 1.5 m causes the model’s evaporation-precipitation-runoff balance to become unstable, and produce extremely unrealistic δ18_{O estimates. Therefore, 1.5 m is kept as the} minimum upper SML depth.

By using the relationship with ΔT to define the upper SML depth in the model (Eq. 5.1), the only remaining major unquantified parameter that exerts a control over the expected δ18_{O of calcifiers in the upper SML is the actual volume of monsoonal freshwater runoff.} To calculate this, I have inverted my box model, as described in section 5.2.2 below. By using the δ18Oruber records as an input to constrain the δ18O of upper SML calcifiers, the

equations within the model were rearranged, so that the volume of monsoonal freshwater runoff into the Mediterranean became the calculated output.