Plant material.

During the 2019–20 season (season 1), 17 romaine lettuce breeding lines and cultivars and one crisphead cultivar (control) were evaluated for two plantings (Table 1). Because pink rib severity of romaine lettuce was only studied in season 1, an expanded follow-up study was conducted during the 2020–21 season (season 2), which included the previous romaine lines and cultivars plus nine crisphead, five butterhead, eight leaf, and three Latin accessions in three plantings (Table 1). These accessions represented elite breeding lines from the University of Florida Institute of Food and Agriculture Sciences lettuce breeding program, commercial cultivars currently used by growers in the Everglades Agricultural Area in Florida, and commercial and noncommercial cultivars not adapted to Florida. The crisphead cultivar Salinas was used as a control for this study because it consistently develops “moderate” to “moderately severe” pink rib (Artés and Martínez 1996).

Table 1. Lettuce accessions of five morphological types and seasons planted near Belle Glade, FL, USA, and accessed for susceptibility to pink rib disorder in two planting seasons (2019–20 and 2020–21). Breeder and pedigree information, when available, and zone of adaptation of lettuce accessions used in the study are provided.

View Fullscreen

Sample preparation and evaluation.

For both growing seasons, all lettuce accessions involved in this research were planted in a commercial field in the Everglades Agricultural Area near Belle Glade, FL, USA, and grown using local commercial practices (Sandoya et al. 2021). There were three 6-m-long plots per accession that were assigned using a completely randomized design. In each plot, lettuce seeds were manually planted in double-row beds, with 0.20-m spacing between rows; plants were thinned to 0.43-m in-row spacing at the three true-leaf stage. During season 1, seeds were planted on 13 Nov and 10 Dec 2019; during season 2, seeds were planted on 8 Oct 2020, 16 Dec 2020, and 3 Mar 2021 (Table 2).

Table 2. Planting and harvest dates for lettuce accessions accessed for susceptibility to pink rib disorder in five morphological lettuce types grown in the Everglades Agricultural Area in two seasons 2019–20 and 2020–21 and types. Minimum, medium and maximum temperatures averages during the crop cycle on each of the experiments planted during two seasons and the growing degree days calculated from University of Florida, Agroclimate Project (2023).

View Fullscreen

At harvest, in both seasons, outer leaves not folded into the lettuce heart were discarded and whole heads (n = 3 heads/accession) were placed in commercial waxed, corrugated cartons (60 cm length × 39 cm width × 33 cm height, 24 romaine heads/carton). Lettuce was commercially vacuum-cooled within 2 hours of harvest to an internal (core tissue) temperature of ∼5 °C and then stored in a commercial cold room (5 °C) in Belle Glade, FL. The experimental setup began within 1 h at 5 °C, noted as “day 0”.

Whole lettuce heads (n = 3) were randomly selected; one to two outer leaves with no visual sign of decay or mechanical damage were selected per head (n = five leaves total). The leaves were prepared using a procedure modified from Cantwell (2016). A midrib segment (12.5 cm) was excised from the center of each leaf (n = 5), and excess leaf blade tissue was trimmed parallel to the midrib, leaving 1- to 2-cm-wide midrib tissue. Each midrib was oriented with the abaxial surface (leaf underside) facing upward, and 9 or to 10 perpendicular slices (∼0.8 cm depth) were made at 1.3-cm increments along the midrib, leaving only the midrib cuticle layer attached. Then, the midribs were placed with the surface cuts facing up in an unsealed polyethylene bag and stored at 5 °C.

A subjective scale was used to rate pink rib severity for the cuts on each midrib every 2 d, starting on day 2 during season 1 (total = 6 d) or day 1 during season 2 (total = 9 d), based on a 5-point discoloration scale, where 1 = none, 2 = slight, 3 = moderate and the limit of marketability, 4 = moderately severe, and 5 = severe (Cantwell 2016) (Fig. 1).

View Figure

• Download as Powerpoint
• Download Figure

Experimental design and statistical analysis.

Pink rib severity was analyzed using a nonparametric statistical method of the ranked means (Brunner et al. 2002; Shah and Madden 2004). Because pink rib severity was evaluated during a time-point storage period, the data were analyzed as repeated measures; the analysis of variance (ANOVA) was calculated using PROC MIXED in SAS (version 9.4; SAS Institute Inc., Cary, NC, USA) according to Shah and Madden (2004). Replicate, accession (A), storage time (ST), experiment (E), and their interactions (A × ST, A × E, ST × E, and A × ST × E) were considered as the fixed statement, and an “r” and “corr” covariance structure was used as the repeated statement. The macros LD_CI and F1_LD_F1 were used to calculate the relative marginal effects (RMEs) and their confidence interval (CI) at 95% for each accession through the storage periods (Shah and Madden 2004). The RMEs are a transformation of the rating scale used to measure severity. The method transforms the ranked means into a relative value of the ordinal scale used to measure a trait. Differences during storage periods were considered significant when the CI of the RME between a given accession did not overlap with the CI of the control cultivar Salinas. Additionally, linear contrasts were calculated between each accession and the commercial control cultivar (Salinas) during the storage period.

Furthermore, to perform a better assessment of the overall pink rib development for each of the accessions, weighted means and RMEs were calculated for each season. The subjective ratings recorded every other day for 9 d were used to calculate weighted means across the evaluation dates to obtain a single mean value per accession using the following formula (Hills et al. 1980):$$WM=\frac{{\mathrm{\Sigma }}_{i=1}^n(R_i{W}_i)}{{\mathrm{\Sigma }}_{i=1}^n(W_i)}$$ where WM is the weighted mean for each accession, n is the number of evaluation dates to be averaged, R_i is the rating value at the i^th evaluation date, and W_i is the weight applied to each rating value at the i^th evaluation date (1, 2, 3, 4, and 5 correspond to days 1, 3, 5, 7, and 9 of storage, respectively).

During this analysis, an ANOVA-F was calculated as previously described following the methodology of Shah and Madden (2004), except for the use of the LD_CI macro. This analysis was conducted considering the weighted means as a randomized complete block design. Similarly, RMEs and 95% CIs were calculated (Shah and Madden 2004).

Genotype × environment analysis.

For season 2, pink rib severity data of a discriminant temperature (5 °C) were subjected to genotype + genotype × environment (GGE) biplot analysis to identify the winning accessions (accessions with less pink rib severity) in a mega-environment and the most stable lettuce accessions in terms of pink rib across the three harvest dates. The three harvest dates for season 2 are herein referred to the following three environments as proposed by Yan et al. (2000): Expt. 1 = S2H1; Expt. 2 = S2H2; and Expt. 3 = S2H3. For the GGE biplot analysis, values recorded using the pink rib rating severity scale (range, 1–5) were proportionally inverted because higher values of pink rib severity were less desirable (5 = no discoloration to 1 = severe pink rib). A “which-won-where” biplot analysis was used to identify accessions (which) that are the best performers (won) in a specific mega-environment (where) (Yan and Kang 2003). In the “which-won-where” biplots, a polygon is graphed by connecting the most responsive accessions located at the farthest positions from the origin. Perpendicular lines are drawn to divide the polygon into sectors called mega-environments that may group one or more of the actual tested environments (e.g., Expts. 1, 2, and 3). The subjective ratings recorded every other day for 9 d were used to calculate weighted means across the evaluation dates to obtain a single mean value per accession using a previously described formula (Hills et al. 1980).

Then, PBTools software (version 14; International Rice Research Institute, Los Baños, Laguna, Philippines) was used to perform the GGE biplot analysis and generate which-won-where biplots for pink rib severity.

In addition, the average environment coordination (AEC) view of the biplot was graphed to rank accessions based on the mean performance and stability of each accession across environments. The single-arrow line in the graph refers to the AEC abscissa and points to the highest PC2 value, which signifies the least pink rib severity across environments (harvests). The AEC ordinate (y-axis) points to the variability of the trait toward both directions. The accessions, experiments, and their respective codes used in the GGE biplots are listed in Tables 1 and 2.

Pink rib severity.

During both seasons 1 and 2, there were significant differences in pink rib severity (P < 0.001) among accessions.

For season 1, romaine cultivars Hialeah and Holbrook had the lowest rated pink rib severity, as demonstrated by the average RME values (0.07 and 0.15, respectively), and a “slight” rating for the weighted means (rating of 2.3 and 2.5), independent of the harvest and storage period (Fig. 2, Table 3). In contrast, 50098, 60182, 60184, and ‘Tall Guzmaine’ had the highest average RME values (0.85, 0.72, 0.79, and 0.89, respectively) and “moderately severe” ratings for the weighted means (ratings >4). The remaining 11 romaine accessions and one crisphead accession developed “moderate” pink rib severity, as indicated by the RME values (from 0.21 to 0.71) and weighted means (ratings from 2.8 to 4.0) (Fig. 2, Table 3).

View Figure

• Download as Powerpoint
• Download Figure

Table 3. Pink rib severity and the weighted means of 17 romaine and one crisphead (control) lettuce accessions during storage (days 2, 4, and 6) at 5 °C in season 1. Weighted means were calculated using the subjective ratings recorded every other day for 6 d to obtain a single mean value per accession.

View Fullscreen

For season 2, ‘Holbrook’, ‘Hialeah’, ‘Manatee’, and 50100 had the lowest pink rib severity among romaine types, as demonstrated by the average RME values (0.25, 0.34, 0.47, and 0.48, respectively), and a “slight” rating for the weighted means (ratings of 2.3, 2.3, 2.4, and 2.7, respectively) across the harvest and storage periods (Fig. 3A, Table 4). The romaine accessions Floricos 83, 70096, 60183, and 60182 had the highest RME values (0.67, 0.72, 0.86, 0.68, respectively) and “moderately severe” ratings for the weighted means (ratings >4). ‘Terrapin’, ‘Tall Guzmaine’, and 43007 also had high average RME values (0.72, 0.73, and 0.76, respectively), but the pink rib severity was not severe; therefore, the weighted means were “moderate” (3.5, 3.7, and 3.5, respectively). The remaining six accessions developed “slight” to “moderate” pink rib severity, as shown by the RME values (0.48–0.70) and weighted means (ratings from 2.4 to 3.8).

View Figure

• Download as Powerpoint
• Download Figure

Table 4. Pink rib severity and the weighted means of 43 lettuce accessions of five morphological types (romaine, crisphead, butterhead, leaf, and Latin) during storage (days 1, 3, 5, 7, and 9) at 5 °C in season 2. Weighted means were calculated using the subjective ratings recorded every other day for 9 d to obtain a single mean value per accession.

View Fullscreen

For the crisphead type, H1098 had the lowest pink rib severity, as indicated by the low RME value (0.43), and a “slight” rating for the weighted mean (rating of 2.8) in season 2 (Fig. 3B, Table 4). The accessions ‘Salinas’, 60167, and H1078 had the highest average RME values (0.62, 0.69, and 0.73, respectively) and “moderate” ratings for the weighted means (ratings of 3.2, 3.3, and 3.4, respectively). The remaining six accessions developed “slight” pink rib severity, as shown by the RME values (0.43–0.56) and weighted mean (rating of 2.9).

The butterhead (B1190 and ‘Palmetto’), leaf (‘Cordoba’ and ‘Galactic’), and Latin (45060) accessions developed the lowest pink rib severity within these individual types, as shown by the average RME values (0.35, 0.21, 0.13, 0.38, and 0.34, respectively), and “no” to “slight” ratings for the weighted means (ratings of 1.9, 2.1, 1, 1, and 2.1, respectively) (Fig. 3C–E, Table 4). The remaining eight accessions among the butterhead, leaf, and Latin types developed “slight” pink rib severity, as shown by the RME values (0.17–0.56) and weighted means (ratings from 2.1 to 2.7).

In season 1, there were significant differences for storage period (P < 0.001), accession × storage period interactions (P < 0.001), and accession × harvest (P < 0.001) interactions, but not for harvest (P = 0.089). The susceptible control ‘Salinas’ developed “slight to moderate” (rating of 2.5) pink rib by day 2 of storage, whereas 5 of the 17 accessions had ratings of 1.0 to 1.5 (Table 3). The linear contrasts, which compare lettuce accessions with ‘Salinas’, support this finding with significant differences for pink rib by day 2 for six breeding lines (50100, 60182, 60183, 60184, 70096, and C1145) and seven cultivars (Green Towers, Hialeah, King Henry, Manatee, Okeechobee, PIC, and Tall Guzmaine) (Supplemental Table 1). By day 4, 9 of the 17 accessions were significantly different from ‘Salinas’ (rating of 3.5) (Supplemental Table 1), either because of less pink rib (ratings ≤3.0), as for ‘Hialeah’, ‘Holbrook’, ‘Green Towers’, and C1145, or because of greater pink rib (ratings ≥3.8), as for 50098, 60182, 60183, 60184, and ‘Tall Guzmaine’ (Table 3, Supplemental Table 1). On day 6, only ‘Hialeah’ and ‘Holbrook’ were rated as “slight” to “moderate” (ratings of 2–3), with significantly lower severity than ‘Salinas’, which developed “moderately severe” (rating of 4) pink rib (Table 3, Supplemental Table 1). Accessions 50098, 60182, 60184, and ‘Tall Guzmaine’ developed “severe” (ratings of 4.8–5) pink rib. RME values were low for ‘Hialeah’ (0.07), ‘Hialeah’ (0.25), and ‘Holbrook’ (0.38), and they were high for ‘Salinas’ (0.38), ‘Tall Guzmaine’ (0.88) and ‘Tall Guzmaine’ (0.94) at days 2, 4, and 6, respectively (Supplemental Table 2).

In season 2, linear contrasts between individual accessions and ‘Salinas’ for each storage evaluation estimated these differences (Supplemental Table 3). On day 1, most accessions had no pink rib development; therefore, very few significant contrasts were detected (Table 4, Supplemental Table 3). By day 3 of storage, ‘Salinas’ developed “moderate” (rating of 3) pink rib; 10 of the accessions had significant contrasts with ‘Salinas’, with ratings of 2.5 or less, whereas four had ratings of 3 (Supplemental Table 3). The number of significant contrasts increased to 15 by day 5, with nine accessions significantly lower than ‘Salinas’ (rating of 3), with ratings of 2.5 or less, whereas two accessions (43007 and ‘Tall Guzmaine’) were rated as “moderately severe” (rating of 4) compared with the “moderate” (rating of 3) rating of ‘Salinas’. By day 7, there were 20 significant contrasts, and the findings were similar to those on day 5; eight accessions had significantly lower pink rib ratings of 2 or less, and eight accessions had higher ratings than ‘Salinas’ (rating of 3). By day 9, 22 of the 43 accessions had significant linear contrasts because of less (ratings ≤3) pink rib severity than ‘Salinas’ (rating of 4) for 50113, 60157, 60159, 70882, B1190, C1148, ‘Chosen’, ‘Cordoba’, H1098, H2020, ‘Hialeah’, ‘Manatee’, ‘Palmetto’, ‘RSX743’, ‘Tehama’, and ‘Two Star’. Greater pink rib severity (ratings of 4 or 4.5) was found for 43007, C1145, ‘PIC’, ‘Tall Guzmaine’, ‘Terrapin’, and ‘Floricos 83’ (Supplemental Table 3). RME values were as low for ‘Salinas’ (0.12) and for 16 other accessions at day 1, including ‘Galactic’ (0.30) at day 3, ‘Holbrook’ (0.36) at day 5, ‘Galactic’ (0.32) at day 7, and ‘Cordoba’ (0.33) at day 9. RME values were high for 50098, 60184, and 70202 (0.30) at day 1, 60182 (0.67) at day 3, 60183 (0.83) at day 5, 60183 (0.91) at day 7, and 60183 (0.91) at day 9 (Supplemental Table 4).

Genotype × environment interaction for pink rib severity.

The GGE biplot analysis allowed dissection of the effects of the genotype and G × E interaction on the total phenotypic variation observed for the trait of interest. In addition, the principal component (PC) scores derived from singular value decomposition allowed the construction of “which-won-where” biplots and, consequently, the identification of superior accessions in specific environments (Yan et al. 2000).

During this research, PC1 and PC2 accounted for 75.7% and 16.3%, respectively, of the total variability of the visual ratings that estimated pink rib severity in season 2 (Fig. 4A). Significant correlations were observed between Expt. 1 (S2H1) and Expt. 2 (S2H2) (r = 0.62; n = 170; P < 0.001), Expt. 1 and Expt. 3 (S2H3) (r = 0.53; n = 170; P < 0.001), and Expt. 2 and Expt. 3 (r = 0.52; n = 215; P < 0.001).

View Figure

• Download as Powerpoint
• Download Figure

Two mega-environments were identified for the three experiments during this research (Fig. 4A). The first mega-environment comprised only Expt. 3, in which the leaf cultivar Cordoba (G33 in Fig. 4) was considered the “winning accession,” or the accession that consistently had the lowest pink rib severity. Expt. 1 and Expt. 2 were placed within a second mega-environment, and the leaf cultivar Galactic (G34) was the “winning accession,” with the lowest pink rib severity for both environments (Fig. 4A). Although ‘Galactic’ had less pink rib severity when grown in the environmental conditions for Expt. 1 (S2H1) and Expt. 2 (S2H2), the mega-environment was positioned in the negative y-axis, suggesting that selection of accessions with lower pink rib severity during both early-season and mid-season plantings might not be simultaneously possible (Fig. 4A).

According to the AEC view of the GGE biplot, accessions on the right side of the AEC ordinate (double-arrow line) had below-average pink rib severity, whereas those accessions on the left side had above-average pink rib severity ratings. Cordoba (G33) was confirmed to be the accession closest to the ideal genotype; namely, it is the one that developed the least severe pink rib across the three experiments (plantings) during this analysis (Fig. 4B). Germplasm such as Cordoba would be better-suited (i.e., would develop less pink rib) for environments similar to the one in Expt. 1, because the AEA (x-axis) of the “average environment” was positioned closer to that of Expt. 1 (Fig. 4B).

Pink rib variation among accessions.

Reducing pink rib severity of whole-head lettuce is a focus for multiple breeding programs; however, there have been no reports of long-term solutions, such as new cultivars or exogenously applied compounds that can reduce or eliminate the development of pink rib (Luna et al. 2016; Saltveit et al. 2005; Tomás-Barberán et al. 1997). Therefore, the identification of accessions with lower susceptibility to pink rib is essential for the long-term mitigation of this disorder. In the current study, lettuce accessions with less susceptibility to pink rib severity were identified from germplasm adapted to a humid subtropical climate. Variations in mid-rib discoloration severity have been reported for numerous lettuce types and accessions (Atkinson et al. 2013a; Cantos et al. 2001; Tudela et al. 2016) that are not adapted to the subtropical climate of Florida.

Atkinson et al. (2013a) reported a distinct rate of discoloration for each lettuce type, with romaine and butterhead types showing the quickest onset of pink rib. Similar findings were reported after 7 d of storage at 5 °C, when the highest a* values of affected tissues, representative of pink to red discoloration, were noted for crisphead cultivars, a butterhead type, a romaine type, and a romaine cultivar (Atkinson et al. 2013a). These findings align with the present study, in which romaine accessions developed the highest pink rib severity, followed by a crisphead type, a butterhead type, a leaf type, and a Latin type, respectively. Similarly, all lettuce accessions developed pink rib after 9 d of storage at 5 °C; the lone exceptions were the two red-leaf cultivars, Cordoba and Galactic. The accessions studied that showed “no” to “slight sensitivity” to pink rib disorder were not previously reported (Atkinson et al. 2013a, 2013b; Cantos et al. 2001; Teng et al. 2019; Tudela et al. 2016). As a consequence, multiple potential parents incorporated into breeding lettuce with tolerance to pink rib for the humid subtropics.

Latin, butterhead, and leaf types had lower susceptibility to pink rib than the other morphological types. Several studies found that the romaine cultivar Green Towers had the least discoloration after 5 d at 7 °C (Teng et al. 2019). Similar findings were presented by Peng et al. (2021), who found that ‘Green Towers’ had little midrib discoloration, whereas ‘King Henry’ and ‘Tall Guzmaine’ developed the most severe discoloration among the 10 accessions evaluated. A similar severity pattern was found during the present study in season 1; pink rib development of ‘Green Towers’ was less severe than that of ‘King Henry’ and ‘Tall Guzmaine’. Three romaine cultivars, Hialeah, Holbrook, and Okeechobee, were the only commercial cultivars with lower pink rib than Green Towers in seasons 1 and 2, and they are commonly planted by Florida growers. However, in season 2, an additional 12 romaine accessions were identified that developed less pink rib than ‘Green Towers’, suggesting that there is wide genetic variation within the lettuce germplasm, and that future studies should evaluate more romaine accessions to identify germplasm with the least pink rib severity.

Salinas, a crisphead lettuce, has been a cultivar of interest for many researchers because of its many positive postharvest characteristics, including relative tolerance to the development of pink rib (Artés and Martínez 1996; López-Gálvez et al. 1996) and longer shelf life. ‘Salinas 88’, a highly related progeny of Salinas (Ryder 1991), was reported to have reported “moderate” to “moderately severe” pink rib severity after 2 weeks at 2 °C Artés and Martínez (1996). Although the value of ‘Salinas’ was confirmed during the present study, seven other crisphead accessions developed less pink rib than ‘Salinas’. Furthermore, leaf lettuce had the least pink rib severity; therefore, further screenings should be conducted with more accessions of this type. For each lettuce type evaluated, there are multiple lettuce accessions that can be used as a source of less pink rib severity in breeding programs.

Pink rib variation during storage.

In the present study, the overall pink rib severity was determined within 5 d of storage. Lettuce accessions with moderate pink rib severity (>3 rating) by storage day 5 either maintained this severity rating or progressed to a higher rating by day 9. Additionally, accessions with less severity (<3 rating) by day 5 maintained less pink rib severity throughout the study. Based on these results, the susceptibility of lettuce germplasm to pink rib appears to be genetically controlled until day 5 of storage at 5 °C; after that time, the accessions differed in their capacity to develop pink rib, which could be attributed to the environment. It was confirmed that this variation in pink rib severity can be attributed to the genetics of lettuce accessions, and not to the environment, because multiple environmental factors, including larger wounds or higher storage temperatures, that normally induce pink rib (Pereyra et al. 2005) were avoided during the present study. In this study, lettuce was not stored under modified atmosphere conditions (low O₂ and/or increased CO₂ levels) that normally suppresses the development of pink rib (Luna et al. 2016; Singh et al. 1972). Therefore, pink rib development was not masked by modified atmosphere conditions.

Pink rib variation among planting seasons.

The G × E (accession × harvest) interactions for pink rib severity were influenced by the harvest time, as demonstrated in romaine lettuce during season 1 and in multiple lettuce types during season 2. For example, the cultivar Cordoba had the least pink rib severity overall, whereas the cultivar Galactic had the least in the first and second plantings. ‘Cordoba’ was determined to be the most stable germplasm, with less pink rib severity, and the best-suited for environmental conditions similar to those during Expt. 3 (spring season), with increasingly higher temperatures and longer daylengths.

Although pink rib severity has been positively correlated to overly mature heads at harvest (Hilton et al. 2009), it is unlikely that the higher severity of pink rib during Expt. 3 was attributable to lettuce being overly mature because the lettuce was harvested at proper market maturity for the commercial cultivars commonly grown in Florida. There were variations in growing degrees in different seasons and different plantings that may have influenced the differences in pink rib severity across these experiments (Table 2).

The significant positive correlations for pink rib severity among experiments suggested noncrossover G × E interactions. Noncrossover G × E interactions indicate that germplasm selection within a single representative environment should be sufficient for breeders to identify lettuce with less pink rib severity (Jenni 2005; Jenni and Hayes 2010; Jenni and Yan 2009). Previously, pink rib in romaine lettuce accessions were reported to have high broad sense heritability, indicating a small environmental effect (Peng et al. 2021). However, as plant breeders evaluate progeny during multilocation trials for traits with potential G × E crossover interactions, inclusions of ratings of accessions for susceptibility to pink rib would be useful when assessed in a wide range of environments (locations).

This is the first study to identify genetic variation of pink rib severity among diverse lettuce accessions adapted to humid subtropical growing conditions. Although the cultivars Cordoba and Galactic had less pink rib severity across all the tested accessions among all types, accessions with the least pink rib severity were identified within each type. For instance, ‘Hialeah’ and ‘Holbrook’ in romaine, H1098 in crisphead, B1190 in butterhead, and 45060 in Latin types showed less pink rib severity. Developing mapping and breeding populations that combine low susceptibility to pink rib of these accessions with other desirable quality and horticultural characteristics should be a priority for genetics studies and breeding purposes.