Predicted over observed ratio for L_IR for all the galaxies modeled and for the different stellar population scenarios we have explored (dashed symbols stand for declining SFH models including nebular emission, solid lines neglecting this effect, dot-dashed cyan and magenta consider rising SFH scenarios including nebular emission). We can see that the SMC-based predictions underpredict L_IR in almost all cases. The Calzetti-based models, although more degenerate, achieve a match with most of the objects’ observed L_IRwithin the 68% confidence range. The rising SFH models predict globally at least as much or more L_IR than their corresponding (in terms of extinction law) declining SFH ones (as shown in Schaerer et al. 2013), pushing in particular the Calzetti-based models to overpredict the observed quantities. The effect is similar but smaller for the SMC-based solutions, and allows a perfect match in the case of C0. We note that for MACS0451N the observed L_IR of the northern part, representing ~1/3 of the total, was adopted here and the predicted L_IR compared here are also derived for this same region, for coherence. This means that, depending on the model, the predicted L_IR used are ~35−40% of the values listed in Tables 4 and 5. If one plotted the same for the total arc in the eventuality of a negligible AGN contribution the ratios shown would be scaled down by ~0.15 dex at most.