Waterlogging stress induced more decreases in photosynthetic parameters of ‘LA4440’ than that of ‘MIX-002’ ( Figure 2 ). Net photosynthetic rate (P _N), stomatal conductance (Gs), intercellular CO₂ concentration (Ci), and transpiration rate (Tr) of ‘MIX-002’ significantly decreased by 42.2%, 55.0%, 17.3% and 42.3%, respectively under CW as compared with that under CC ( Figure 2 ). By comparison, the P _N, Gs, Ci, and Tr of ‘LA4440’ were significantly lower under CW than CC (50.8%, 75.6%, 58.0%, and 69.1%, respectively) ( Figure 2 ). The P _N, Gs, Ci, and Tr of both genotypes significantly increased under PW than CW, except the Ci of ‘MIX-002’ ( Figure 2 ). Moreover, it is notable that the four photosynthetic parameters showed no significant difference in both genotypes between PW and CC, except the increased Tr of ‘MIX-002’ ( Figure 2 ).

The SOD and POD activities of ‘LA4440’ were significantly lower and higher under CW than CC, respectively ( Figures 3A, C ). For ‘MIX-002’, the CAT and POD activities significantly decreased under CW than CC ( Figures 3B, C ). Priming induced lower SOD activity but higher CAT activity of ‘MIX-002’ as compared with CW ( Figures 3A, B ). The POD activity of ‘LA4440’ significantly dropped under PW as compared with CW ( Figure 3C ).

There was no significant difference between CW vs CC and PW vs CW in MDA content and O₂ ^•─ production rate of both genotypes ( Figures 4A, B ). However, the O₂ ^•─ production rate of ‘LA4440’ significantly increased by 74.5% under PW than CC ( Figure 4B ). Proline content of ‘LA4440’ significantly decreased by 23.0% under CW than CC, while that significantly decreased under PW as compared with both CC (69.4%) and CW (60.3%) ( Figure 4C ). Soluble protein content significantly decreased by 22.1% in ‘MIX-002’ but increased by 22.0% in ‘LA4440’ under CW as compare with the respective CC ( Figure 4D ). Priming induced a significant increase and decrease in soluble protein content of ‘MIX-002’ (14.3%) and ‘LA4440’ (13.3%), respectively, as compared with CW ( Figure 4D ).

The expression of SODCC2 was significantly upregulated, but the expressions of ARG2, 2-CP1, CAT2 and X92888 were significantly downregulated in ‘MIX-002’ under CW as compared with CC ( Figure 5 ). Concerning CW vs CC in ‘LA4440’, the expressions of ARG2 and X92888 significantly increased but the expressions of the other three genes significantly decreased ( Figure 5 ). The expression of SODCC2 was significantly downregulated in both genotypes under PW as compared with CW ( Figure 5A ). By contrast, the expression of ARG2 and X92888 significantly increased in ‘MIX-002’, which significantly decreased in ‘LA4440’ under PW as compared with CW ( Figures 5B, E ).

The size of plants under PW was smaller than CC and CW, demonstrating that Trichoderma harzianum did not show any alleviated effects on tomato phenotypes recovery ( Supplementary Figure S2A ). During the WL stage, the plant height of ‘LA4440’ under CW and PW showed no significant difference, which was significantly lower than CC (8.6% and 11.4%, respectively) ( Supplementary Figure S2B ). During the R2 stage, the plant height of ‘MIX-002’ significantly decreased under PWT as compared with CCC ( Supplementary Figure S2B ). In contrast, the plant height of ‘LA4440’ significantly decreased under CCT, CWT, PWC and PWT than CCC, where PWC and PWT showed the maximum deduction ( Supplementary Figure S2B ). There were no significant differences in stem diameter between all the treatments within each genotype in both WL and R2 stages ( Supplementary Figure S2C ). The shoot fresh weights of both genotypes were significantly lower under PWC and PWT than CCC ( Supplementary Figure S2D ). In contrast, the shoot fresh weights of both genotypes were significantly higher under PWC and PWT than CCC ( Supplementary Figure S2E ). By comparison, the shoot dry weight of ‘MIX-002’ was significantly lower under PWT than CCC, while the root dry weight of both genotypes showed no difference ( Supplementary Figure S2F, G ). The total fresh weight and dry weight of both genotypes showed no significant difference between all the treatments, except total dry weight of ‘MIX-002’ under PWT ( Figures 6A, B ). The decreased ratio of the shoot/root fresh weight under PWC and PWT than CCC was observed ( Figure 6C ). As compared with CCC, the ratio of the shoot/root dry weight significantly dropped under the other five treatments ( Figure 6D ).

The leaf SPAD (soil and plant analyzer development) is an important parameter to measure the relative content of chlorophyll in plants. The leaf SPAD of ‘MIX-002’ significantly decreased under PWC and PWT during the R2 stage as compared with PW during the P and R1 stages ( Figure 7A ). During the WL stage, the leaf SPAD of ‘LA4440’ was significantly higher under CW than CC ( Figure 7B ).

Generally, the H₂O₂ content of both genotypes under CCC kept constant ( Figures 7C, D ). Priming significantly increased the H₂O₂ content of ‘MIX-002’ during the R1 (40.4%), WL (30.0%) and R2 (129.1%) stages as compared the respective controls ( Figure 7C ). During the WL stage, the H₂O₂ content of ‘MIX-002’ significantly increased by 27.4% under CW as compared with CC ( Figure 7C ). During the R2 stage, the H₂O₂ content of ‘MIX-002’ significantly increased by 58.1%, 112.0%, 95.9%, 129.1%, and 95.6%, respectively, under CCT, CWC, CWT, PWC and PWT than CCC ( Figure 7C ). Moreover, priming significantly increased the H₂O₂ content of ‘LA4440’ during the P (77.1%), R1 (181.7%), WL (161.9%) and R2 (196.6%) stages as compared the respective controls ( Figure 7D ). During the R2 stage, the H₂O₂ content of ‘LA4440’ was significantly higher under CWC, CWT, PWC and PWT than CCC, with 162.2%, 132.9%, 196.6%, and 180.1% increased proportion, respectively ( Figure 7D ).

Waterlogging stress usually leads to chlorophyll degradation and decreased photosynthesis and ROS damage in plants (Pan et al., 2021). Waterlogging for three days significantly decreased the P _N and Gs of pepper (Ou et al., 2011). In accordance, the P _N of waterlogging-tolerant and waterlogging-sensitive tomato significantly decreased by 42.2% and 50.8%, respectively under CW than CC ( Figure 2 ). This indicated that the waterlogging stress inhibited photosynthesis of both genotypes being announced more in the sensitive genotype. Plant photosynthesis is often restrained due to stomatal and non-stomatal limitations (Flexas and Medrano, 2002). When Gs and Ci decrease simultaneously, the CO₂ levels in plant cells cannot meet the needs of photosynthesis that retrained P _N, being regarded as stomatal limitation (Farquhar and Sharkey, 1982; Zeng et al., 2021). By comparison, when Gs drops but Ci remains unchanged, it is the low activities of chloroplast and photosynthetic enzyme leading to the decreased P _N, known as non-stomatal limitation (Farquhar and Sharkey, 1982; Zeng et al., 2021). Here, both genotypes showed a simultaneous decrease in P _N and Gs, but only the waterlogging-sensitive tomato showed a significant decrease in Ci under CW than CC ( Figure 2 ). Therefore, the decreased photosynthesis of waterlogging-sensitive tomato under waterlogging stress was due to stomatal factors, while that of waterlogging-tolerant tomato was caused by non-stomatal factors ( Figure 8 ). X92888 (photosynthesis-related gene) was significantly downregulated in waterlogging-tolerant tomato but upregulated in waterlogging-sensitive tomato under CW as compared with CC ( Figure 5E ). Together, the two tomato genotypes exhibited different photosynthetic regulatory mechanisms in response to waterlogging stress, where the photosynthesis-related genes may play a critical role ( Figure 8 ).

ROS is a normal product of plant cell metabolism, and insufficient O₂ under waterlogging stress increases in intracellular ROS (Yan et al., 1996). MDA is the final decomposition product of membrane lipid peroxidation, caused by ROS with strong oxidative activities, which was applied to evaluate status of lipid peroxidation and reflect the degree of plant damage (Zhang F. et al., 2007). The MDA content and O₂ ^•─ production rate of both genotypes showed no significant difference between CC, CW and PW, except that the O₂ ^•─ production rate of the waterlogging-sensitive genotype was significantly higher under PW than CC ( Figures 4A, B ). This partially explained the sensitivity of ‘LA4440’ to waterlogging stress. Meanwhile, during the WL stage, waterlogging stress significantly increased the H₂O₂ content only in waterlogging-tolerant tomato ( Figures 7C, D ). The ARG2 (ROS-related gene) was significantly downregulated in waterlogging-tolerant tomato but significantly upregulated in waterlogging-sensitive tomatoes under CW than CC ( Figure 5B ). Plants can rely on the antioxidant enzyme system to dynamically maintain the ROS balance and reduce the degree of oxidative damage under waterlogging stress (Zhang G. et al., 2007; Hasanuzzaman et al., 2020). We found that the SOD and CAT activity decreased only in sensitive and tolerant tomato, respectively, while the POD activity decreased and increased in tolerant and sensitive tomato, respectively, under CW than CC ( Figure 3 ). The expression of SODCC2 was significantly upregulated in tolerant tomato but downregulated in sensitive tomato under CW than CC ( Figure 5A ). The expression of 2-CP1 (POD-related gene) significantly decreased by 17.3% and 80.4%, and that of CAT2 decreased by 33.1% and 78.9%, in tolerant and sensitive tomatoes under CW than CC ( Figures 5C, D ). Therefore, taking ROS and antioxidant enzymes changes into account, the ROS regulatory system of two tomato genotypes differentially responded to waterlogging stress ( Figure 8 ).

In addition to antioxidant enzymes, plants can also alleviate stress damage by adjusting osmotic regulatory substances such as proline and soluble protein (Ru et al., 2022; Zulfiqar and Ashraf, 2022). We found that only the waterlogging-sensitive tomato showed lower proline content under CW than CC ( Figure 4C ). The waterlogging-tolerant tomato exhibited lower soluble protein content, but the waterlogging-sensitive had higher soluble protein content under CW than CC ( Figure 4D ). Therefore, waterlogging tolerant and sensitive tomatoes under excess water exhibited differences in gas exchange, antioxidant system and osmotic regulatory mechanisms, and these differences are related to changes in crucial gene expression levels ( Figure 8 ).

The root fresh weight was enhanced in primed plants than non-primed plants at the WL stage ( Supplementary Figure S2E ). The possible reason for this could be that more sugars were transported to roots, which provided a metabolic substrate for anaerobic respiration and thereby produced more ATP to maintain the root growth in primed plants than non-primed plants (Feng et al., 2022). Even though there were no positive effects of priming on shoot weight ( Supplementary Figures S2D, F ), it is worth noting that the P _N, Gs, Ci and Tr of both genotypes were significantly higher under PW than CW, except the Ci of the tolerant genotype ( Figure 2 ). The P _N, Gs, Ci and Tr of both genotypes did not show significant difference between PW and CC, except the Tr of the tolerant genotype ( Figure 2 ). These indicated that the priming effectively alleviated the stress damage on photosynthesis when waterlogging reoccurred, which was consistent with the previous findings on wheat (Wang et al., 2016) and tomato (Zhou et al., 2022a). Basal heat tolerance refers to the heat tolerance of plants without adaptation or pre-adaptation; pre-exposing plants to appropriate levels of heat stress enhances heat tolerance of plants, which is called acquired heat tolerance (Mittler et al., 2012; Stief et al., 2014). If the plants can memorize the high temperature and the acquired heat tolerance can be maintained for a long time, it is called maintenance acquired heat tolerance (Stief et al., 2014; Sun et al., 2018). Similarly, we found that both tomato genotypes showed maintenance acquired waterlogging tolerance by memorizing the stress during the process waterlogging priming.

In plants, ROS can induce acclimatization, defense, and memory of stress (Mittler et al., 2022). Waterlogging priming significantly induced the processes of stress defense and energy metabolism to increase the waterlogging tolerance of wheat (Wang et al., 2016). On one hand, as a non-radical form of ROS, H₂O₂ can cause oxidative stress by inactivating enzymes in plants (Gill and Tuteja, 2010). Agualongo et al. (2022) suggested that waterlogging priming decreased H₂O₂ content, lipid peroxidation, and activity of antioxidant enzymes and thereby alleviated oxidative stress when soybean suffered the waterlogging again. On the other hand, H₂O₂ is an important signaling molecule of different regulatory or enzymatic targets in plant cells (Mittler et al., 2004; Stone and Yang, 2006; Mhamdi and Van Breusegem, 2018). For waterlogging-tolerant tomato, primed plants showed no difference in H₂O₂ content during the P stage, but exhibited significantly higher H₂O₂ content than the respective controls during the R1, WL, and R2 stages (40%, 30%, and 129%, respectively) ( Figure 7C ). By comparison, priming significantly increased the H₂O₂ content in waterlogging-sensitive tomato as compared with the respective controls, with 77.1%, 181.7%, 161.9%, and 196.6% increased proportion during the P, R1, WL, and R2 stages, respectively ( Figure 7D ). Sun et al. (2018) concluded that during the recovery from heat stress, tomato seedlings upregulated H₂O₂ content and enhanced the expression of heat responsive genes to improve the maintenance of acquired heat tolerance. Together with our findings, waterlogging priming induced waterlogging memory by upregulating the H₂O₂ content as a stress signal for stress defense induction, and a high H₂O₂ level was maintained when waterlogging reoccurred, resulting in enhanced maintenance of acquired waterlogging tolerance especially in sensitive tomato ( Figure 8 ).

We suggest that the high H₂O₂ level as a signal could enhance the regulation of stomatal conductance, antioxidant enzyme system and osmotic regulatory substances in tomato plants. Firstly, the Gs of the two tomato genotypes with priming were significantly higher than unprimed plants ( Figure 2B ). An et al. (2008) found that H₂O₂ was an important signal of ABA-induced stomatal closure in Vicia faba, where exogenous ABA stimulated extracellular copper-containing diamine oxidases (CuAO) activity in guard cells, increased H₂O₂ production and [Ca²⁺] cell level and induced stomatal closure (An et al., 2008). However, the upregulation of H₂O₂ induced by waterlogging priming may act as a signal to regulate stomatal aperture through ABA in tomato plants. Secondly, waterlogging priming altered the activity of antioxidant enzymes mediated by H₂O₂ in two tomato genotypes. The SOD activity was downregulated, while CAT activity was upregulated, resulting in stable H₂O₂ level in tolerant tomato under PW than CW ( Figures 3 , 7 ). By comparison, the POD activity of primed plants was significantly lower but the H₂O₂ level of primed plants was significantly higher than non-priming plants in sensitive tomato under waterlogging stress ( Figures 3 , 7 ).

Thirdly, the contents of proline and soluble protein can reflect the osmotic regulation ability of plants under stress (Shahnaz et al., 2011; Kavi Kishor and Sreenivasulu, 2014). In this study, the proline content of waterlogging-sensitive tomato significantly decreased, while the soluble protein content of waterlogging-tolerant tomato significantly increased under PW than CW ( Figure 4 ). In addition, ROS can trigger and modulate transcriptional regulation that enabled the plants to respond to stress and enhance plant resilience (Mittler et al., 2022). This explained why the sensitive tomato showed significantly decreased expression of SODCC2, ARG2, and X92888 under PW than CW ( Figure 5 ).

By regulating the distribution of metabolites, especially antioxidants, the redox states of plant cells can be balanced, thus achieving the regulation of metabolism and photosynthesis (Foyer and Shigeoka, 2011). We suggest that waterlogging priming enhanced the regulation of stomatal conductance, antioxidant enzyme activities and osmotic regulatory substances and key gene expressions mediated by H₂O₂. This contributed to keeping the ROS homeostasis, and promoting the maintenance of acquired waterlogging tolerance, thus alleviating the damage to tomato photosynthesis when waterlogging reoccurred ( Figure 8 ). However, the application of Trichoderma harzianum did not show alleviated effects. The potential reasons could be that: 1) commercially available powder of Trichoderma harzianum applied might not work as well as the cultured strain; and 2) foliar spraying may not be a good approach to apply Trichoderma harzianum in waterlogging stress study. Thereby, suitable waterlogging priming could be beneficial to improve the waterlogging tolerance of tomato plants, while the application of Trichoderma harzianum needs further investigation to work.

A total of 27 tomato genotypes were included in the first round of screening to evaluate their waterlogging susceptibilities. Among them, nine were provided by the Vegetable Physiological and Ecological Laboratory of Nanjing Agricultural University, and 18 were commercial varieties purchased from the market ( Supplementary Table S1 ). Tomato seedlings were sown in 72-hole plates (54-cm length and 28-cm width) filled with a mixture of substrates (the volume ratio of turf, perlite, and vermiculite = 2:1:1). The seedlings were grown in a climate chamber (RGD-1000C, Ningbo, China) with a 14 h light time (8:00-22:00) and a 10 h nighttime (22:00-8:00). The light intensity was 30,000 lux and a daytime/night temperature was 25°C in the chamber. The 21-day-old seedlings with three fully expanded leaves were transplanted into pot (6.5 cm height and 6.5-cm diameter). The 28-day-old seedlings with four fully expanded leaves were treated under control (C) and waterlogging (WL). The WL treatment was conducted based on Zhou et al. (2022b), where the plants in the above small pots were put in big pots (10.8-cm height, 11.1-cm diameter) being filled with water for 1 cm above the surface of the substrate. There were five plants per treatment per genotype.

On the 7^th day of treatments, plant height, stem diameter, fresh and dry weight of shoots and leaf MDA content were measured with three biological replicates. The plant height was investigated by measuring the vertical distance from cotyledonary node to growing point using a ruler. The stem diameter (1 cm above the cotyledonary node) was measured using a vernier caliper. The plant was cut from the cotyledonary node and weighed to obtain the fresh weight of shoot. The shoot samples were put at 105 °C for 30 min followed by 80 °C for two days and weighed, which was the dry weight of shoot.

The third fully expanded leaf from top to bottom was selected to measure MDA content. The determination of MDA content was based on thiobarbituric acid (TBA) method (Kumar and Knowles, 1993) with minor modification. The 0.2 g fresh leaves were weighed, added to 5% trichloroacetic acid (TCA) solution, ground to homogenate, and then centrifuged. The supernatant was mixed with 67% TBA solution by oscillating. The samples were incubated in boiling water for 30 min. The 200 uL supernatant per sample was taken after quickly cooling and centrifuging. The samples were put on enzyme-labeled plate with three times of technical repetition. The absorbance of the solution under the wavelength of 450 nm, 532 nm and 600 nm were measured using Microporous plate detecting instrument (Cytation3, BioTek USA).

According to the results from the first-round screening, three waterlogging-tolerant tomato candidates (No. 1, 7 and 9) and three waterlogging-sensitive tomato candidates (3, 10 and 16) were selected for the second-round screening. The six tomato candidates were grown and treated in the same way as the first-round screening, where the same parameters were investigated in the same way as the first round of screening.

The t-test was applied to compare the results under control and waterlogging in the screening experiments (P < 0.05) using SPSS Statistics (version 25.0, IBM, USA). Analysis of Variance (ANOVA) were used to test the significant differences of physiological and metabolic data between the two genotypes under all the treatments (P < 0.05) using SPSS Statistics (version 25.0, IBM, USA). The relative expression of genes was calculated using the 2^-∆∆Ct method with SLActin as an internal reference. All the Figures were made using Microsoft Excel 2019 and Origin Pro 2022.