Reassessing the link between soluble solids and sugars in melon: FTIR models as an alternative tool

Soluble solids content (SSC, °Brix) is widely used to estimate melon sugar content. However, in a diverse set of Ibericus melons, it correlated moderately (R² ≈ 0.60) with total sugars and sucrose, and not with hexoses content or the hexose-to-sucrose ratio. Moreover, the sensitivity of SSC under high selection pressure for high sugar (10%-20%) remained low (55%-60%). In contrast, FTIR trial specific models predicted sucrose (s), hexoses (h), citric acid (not malic), total sugars, sucrose equivalents, and the h to s ratio with mean prediction R² ranging 0.73-0.88 and errors ranging 3.5-8.9%. A general model, performed similarly to the weakest specific one, strongly affected by low fructose levels, resulting in mean prediction R² of 0.63 (external validation). Externally validated specific models achieved prediction R² values of 0.79-0.94 (mean 0.84) and errors ranging 3.6-28.7% (mean 11%). FTIR represents a good alternative to SSC for high-throughput phenotyping.