Fruit water content as an indication of sugar metabolism improves simulation of carbohydrate accumulation in tomato fruit

Abstract

Although fleshy fruit is mainly made up of water, little is known about the impact of its water status on sugar metabolism and its composition. In order to verify whether fruit water status is an important driver of carbohydrate composition in tomato fruit, an adaptation of the SUGAR model proposed previously by M. Génard and M. Souty was used. Two versions of the model, with or without integrating the influence of fruit water content on carbohydrate metabolism, were proposed and then assessed with the data sets from two genotypes, Levovil and Cervil, grown under different conditions. The results showed that, for both genotypes, soluble sugars and starch were better fitted by the model when the effects of water content on carbohydrate metabolism were taken into consideration. Water content might play a regulatory role in the carbon metabolism from sugars to compounds other than sugars and starch in Cervil fruit, and from sugars to starch in Levovil fruit. While water content influences tomato fruit carbohydrate concentrations by both metabolism and dilution/dehydration effects in the early developmental stage, it is mainly by dilution/dehydration effects in the late stage. The possible mechanisms underlying the effect of the fruit water content on carbohydrate metabolism are also discussed.

Keywords: Solanum lycopersicum; Carbohydrate metabolism; genotype×environment interaction; modelling; soluble sugars; starch; water status.

Fig. 1.

Diagram of the SUGAR model illustrating the sugar metabolism and carbon balance in tomato fruit. Arrows represent carbon flows, and the parameters [k₁(t), k₂(t), and k₃(t)] indicate the reaction rate constants related to carbon conversion among soluble sugars, starch, and other compounds. Boxes represent the carbon components in the fruit, and the two ellipses indicate carbon supply and losses through respiration, respectively.

Fig. 2.

Measured FW (open circles, left axis), DW (closed circles, right axis), and their corresponding fitted curves (FW, dashed line; DW, solid line) for the different treatments for Cervil (A) and Levovil (B). The Gompertz function was used to fit the observed data. The data sets are provided in Supplementary Table S1. (HL, high crop load; LL, low crop load; WW, well-watered; WD, water deficit.)

Fig. 3.

Time-courses of the fruit water content for Cervil (a) and Levovil (b). The filled circles are the maximum values at each sampling date from the pooled measurements of all the treatments. The upper boundary line of fruit water content is fitted with these points by an exponential function using non-linear least-squares estimate. The data sets are provided in Supplementary Table S3. (HL, high crop load; LL, low crop load; WW, well-watered; WD, water deficit.)

Fig. 4.

Seasonal variations of mean observed data and simulated values by Tom_SUGAR and by Tom_SUGAR_WC for soluble sugar and starch concentration on a DW basis (SS and ST, g 100 g^–1 DW) and on a FW basis (SSC and STC, g 100 g^–1 FW) under different treatments (2003, 2007_HL, 2007_LL, 2014_WW, and 2014_WD) for Cervil (A) and Levovil (B). The bars indicate the SDs of the measurements (n=3–5). The data sets are provided in Supplementary Table S2. Each line represents the simulation result derived from a set of parameters estimated by the genetic algorithm, and 2×100 simulations were implemented. The bold lines represent simulation results with the best parameter set shown in Table 4. (HL, high crop load; LL, low crop load; WW: well-watered; WD: water deficit.)

Fig. 5.

Variations in the calculated reaction rate constant related to carbon conversion from soluble sugars to other compounds [k₁(t)] and from soluble sugars to starch [k₃(t)] depending on fruit age (degree-days after anthesis, dd), hourly mean air temperature (Temp, °C), and relative growth rate of fruit DW (RGR, h^–1). Their values are calculated using Equations (9) and (10) with the measurements (Supplementary Tables S1, S2) from the treatments 2003, 2007_HL, 2007_LL, 2014_WW, and 2014_WD for Cervil (A) and Levovil (B). (HL, high crop load; LL, low crop load; WW: well-watered; WD: water deficit.)

Fig. 6.

The time-course variation of the reaction rate constants related to carbon conversion from soluble sugars to other compounds [k₁(t)] and from soluble sugars to starch [k₃(t)] under different treatments for Cervil and Levovil. k₁(t) and k₃(t) were calculated using Equations (17) and (18), respectively, with 100 sets of parameters, and each line represents the result calculated from a set of parameters. (HL, high crop load; LL, low crop load; WW, well-watered; WD, water deficit.)

Fig. 7.

Simulated seasonal variations of carbon amount in the form of soluble sugars (b), starch (c), and other compounds (d), as well as soluble sugar and starch concentration on a FW basis (SSC and STC) (e, f) under four scenarios of fruit water content dynamics (a) for Cervil (A) and Levovil (B). For all the scenarios, the fruit DW curves derived from the treatment of 2003 for both genotypes. The fruit water content for scenario C was kept the same as that of the 2003 treatment. The fruit water content for WS was 2% less than that for C during the whole simulation period. The fruit water content for CWS was the same as that for C in the early stage (e.g. from 5 daa to 30 daa for Cervil and from 9 daa to 45 daa for Levovil), and decreased to that for WS during the late stage (e.g. from 31 to 45 daa for Cervil and from 46 to 60 daa for Levovil). In contrast to CWS, the fruit water content for WSC was the same as that for WS in the early stage and increased to that for C during the late stage. For each scenario, each line represents the simulation result derived from a set of parameters estimated by the genetic algorithm, and 100 simulations were implemented.