A diversity of traits contributes to salinity tolerance of wild Galapagos tomatoes seedlings

Traits of modern crops have been heavily selected in agriculture, causing the commercial lines to be more susceptible to harsh conditions, which their wild relatives are naturally better able to withstand. Understanding the developed mechanisms of tolerance present in wild relatives can enhance crop performance under stress. In this study, salinity tolerance traits of two species of wild tomato endemic to the Galapagos Islands, Solanum cheesmaniae and Solanum galapagense, were investigated. Since these tomatoes grow well despite being constantly splashed with seawater, they could be a valuable genetic resource for improving salinity tolerance in commercial tomatoes. To explore their potential, over 20 traits reflecting plant growth, physiology and ion content were recorded in 67 accessions of S. cheesmaniae and S. galapagense and two commercial tomato lines of Solanum lycopersicum. Salt treatments of 200 mM NaCl were applied for ten days, using supported hydroponics. Great natural variation was evident in the responses of the Galapagos tomatoes to salt stress and they also displayed greater tolerance to salt stress than the commercial lines tested, based on multivariate trait analyses. Although Galapagos tomatoes in general exhibited better tolerance to salt stress than the commercial lines tested, the accessions LA0317, LA1449 and LA1403 showed particularly high salinity tolerance based on growth maintenance under stress. Thus, Galapagos tomatoes should be further explored using forward genetic studies to identify and investigate the genes underlying their high tolerance and be used as a resource for increasing salinity tolerance of commercial tomatoes. The generated data, along with useful analysis tools, have been packaged and made publicly available via an interactive online application (https://github.com/mmjulkowska/La_isla_de_tomato) to facilitate trait selection and the use of Galapagos tomatoes for the development of salt tolerant commercial tomatoes.

Introduction transparent, then they were washed several times in tap water. This treatment was necessary 123 for the germination of most of the Galapagos tomato seeds (Rush and Epstein, 1976). 124 Although the commercial varieties did not require bleaching to germinate, the same treatment 125 was applied to all the seeds.

126
Experimental hydroponics setup 127 Eight-centimeter square pots were filled with plastic pellets as a substrate to support the 128 roots. The pellets were chosen for their inert quality and dark color to protect the roots from 129 light. The plastic pellets were made of 20% talc-filled polypropylene, black, and with a 130 density of 1.05 g/cm 3 to sink in water (Edwards Industrial Repair, Robards, KY). Treated 131 seeds were germinated directly in the pots, on 0.8% agar plugs (8 mm diameter, 12 mm deep) 132 containing ¼ Murashige and Skoog (MS) salts inserted in the plastic pellets ( Figure 1A). 133 Two agar plugs, each with one seed, were placed in each pot to increase the chances of 134 germination of at least one seed per pot. Sown pots were placed in nursery trays filled with 135 fresh water and covered with a transparent plastic cover, then kept at 26°C. Germination took 136 between three and eight days. After germination, the two seedlings per pot were thinned to 137 one, by choosing plants with even size and healthy appearance. Treatment with 10% bleach 138 was repeated for those seeds that did not germinate one week after the first treatment 139 (Darwin, 2009). Six biological replicates were used for each control and salt treatment. 140 Because different species with different growth habits were being compared, another six 141 replicates were included, to be harvested before the salt treatment was started -thus, the 142 effects of salinity on growth that occurred only during the time of the salt treatment could be 143 calculated, correcting for differences in growth that occurred prior to the salinity treatment. 144 When the cotyledons had emerged fully and the radicle was long enough, the pots were 145 transferred from the nursery to a supported hydroponics system (EconoTray by American 146 system consists of a grow tray, a tray frame, a 100 L nutrient reservoir tank, and a 148 submersible aquarium-pump (ViaAqua 360 by Commodity Axis, Inc.). The tray frame height 149 was modified, from 0.5 to 1.06 m, to hold the plants above the greenhouse walls and thus 150 avoid shading. Each growth tray was able to hold up to 108 pots (8 cm 2 ) with seedlings at the 151 cotyledon stage and 96 pots (8 cm 2 ) growing tomato plants up to the 7 th or 8 th leaf stages. The 152 growth tray rests on the tray frame above the reservoir tank, which contains the nutrient  Salt stress treatment and considerations 167 The use of 8 units of the hydroponics system allowed the salt treatment of a total of 69 168 accessions of three different tomato species (S. cheesmaniae, S. galapagense, and S. 169 lycopersicum), with 6 biological replicates per accession per treatment , plus 6 seedlings per 170 accession that were harvested as a baseline before treatment. The plants were subjected to salt 171 stress treatment at the same developmental stage, when the 4 th leaf started to emerge. Given that S. lycopersicum plants were bigger than Galapagos tomatoes throughout all 173 developmental stages (as leaves were larger and stems were thicker), they were treated when 174 the 3 rd leaf started to emerge, to compensate for the size difference ( Figure S1), since bigger 175 plants are often better able to tolerate salt stress than smaller plants. 176 The salt stress treatment was administered gradually to the plants. As each plant reached the 177 desired developmental stage (i.e. the 4 th or 3 rd leaf emergence), it was moved from the 178 nutrient-only hydroponics systems to the systems supplemented with NaCl. Plants were first 179 moved to a 75 mM NaCl hydroponic system and 12 hours later, moved to a 200 mM NaCl 180 hydroponic system. Seedlings remained in the 200 mM NaCl hydroponic system for 10 days.

181
Supplemental CaCl 2 was added to the solutions to compensate for the decrease in Ca 2+ 182 activity arising from the addition of NaCl (Tester and Davenport, 2003). The amount of 183 CaCl 2 added to the NaCl solutions was calculated using GEOCHEM-EZ software (Shaff et 184 al., 2010) to maintain Ca 2+ activity at 0.4 mM, which was the normal Ca 2+ activity in the 185 nutrient solution prior to NaCl addition (Table S1). images were used to test a non-destructive approach to estimate the salinity tolerance of 194 Galapagos tomato seedlings. The photographs were processed using a Matlab script for green 195 pixel count (green_finder_V2.m), which can be found in the Supplementary Data. For 196 destructive sample harvesting before and after salt treatment, the plant was carefully 197 extracted from the pot, roots were rinsed in 10 mM MgCl 2 solution with the excess solution 198 dried off using tissue paper. Plant root, shoot, and 3 rd or 4 th leaf tissues were each weighted 199 separately. The 3 rd and 4 th leaves, taken from S. lycopersicum and Galapagos accessions 200 respectively, were further characterized as they had developed under salt stress conditions.

201
Each leaf was scanned using an EPSON scanner to calculate leaf area, perimeter, height and 202 width using the WinFolia software (Regent Instruments Inc.). In terms of physical 203 measurements, stem thickness was measured at the base using a caliper, while the stem and 204 root length were measured using a ruler. The different tissues were then stored in paper 205 envelopes and dried at 60°C for three days to then measure their dry mass. Dry leaf samples 206 (without petiole) and root samples were digested in 50 mL Falcon tubes with 5 mL of 1%

207
(v/v) nitric acid in a HotBlock TM (Environmental Express) at 80°C for four hours. Sodium 208 content was measured in leaf and root samples using a flame photometer (model 420;   The variability in traits related to salinity tolerance was described using a principal  The two Galapagos tomato species were observed to have distinctly different morphologies, 231 hence the phenotypic data for each species were analyzed separately. The K-means clustering 232 method (MacQueen, 1967) was used to reveal groups within the data. The clustering method 233 was run with different numbers of clusters (2 to 6) and it was found that two clusters 234 provided the most interpretable output, in terms of accession clustering by traits. The 235 accessions were grouped based on 11 traits related to salinity tolerance: Na and K 236 concentration in root and leaf, leaf area, leaf elongation (width/length), leaf succulence (leaf 237 water/leaf area), leaf number, stem and root length, and total fresh mass. K-means were 238 calculated using the stats package in R (R Core Team, 2017).

239
To identify possible tolerance mechanisms, all trait measurements from the accessions of 240 each species were compared using a heat map, drawn by the function heatplot of the made4 R  Figure S4). Plant growth-related traits, 267 such as shoot fresh and dry mass, root fresh and dry mass, total fresh and dry mass and total water content, are also positively and significantly correlated (correlation coefficients 0.69-269 1.00, p-value=0.001) ( Figure S4).

270
Interestingly, leaf K concentration in salt-treated plants relative to control plants is negatively 271 correlated with all the leaf traits, some plant growth-related traits, and Na and K in the root.

272
On the other hand, leaf Na concentration in salt-treated plants, relative to control plants, did

289
A PCA was performed to reduce data dimensionality and reveal the potential relationships 290 among representative salinity-tolerance traits. In this study, the four main PCA axes had 291 eigenvalues larger than 1 (Table S2), which indicates that each principal component (PC) 292 accounts for more variance than accounted-for by one of the original variables in the 293 standardized data. This was used as a cut-off to determine the number of PCs to retain.

294
The PC1 explained 33.8% of the total variability between traits/individuals and was 295 associated with most traits, except leaf Na concentration and leaf succulence (Table 1 and 296 Figure 4). The most significant trait for PC1 was the total fresh mass ( Table 1). The 297 accessions at the lower end of PC1 are those whose growth was most affected by salinity but 298 were still able to retain high levels of K in the leaf, while at the higher end, there are the 299 accessions with higher levels of plant growth, leaf area, leaf number, and stem and root 300 length.

301
PC2 accounted for an additional 15% of the total variability among seedling traits and 302 appeared to be related to the ion content and some growth traits (Table 1 and Figure 4). The 303 accessions with succulent leaves and higher accumulation of Na in the leaf were located at 304 the lower end of PC2, while those with increased leaf number and K retention in the leaf 305 were located at the higher end of PC2. The PC2 also divided the root Na concentration and 306 root length, where those accessions with high Na concentration in the root had the shortest 307 root.

308
PC3 accounted for 11.7% of the total variability among salinity tolerance-related traits. It was 309 significantly associated with total fresh mass but had a stronger association with leaf traits, 310 such as elongation factor (length/width) and Na concentration (Table 1). This could suggest 311 that Na concentration in the leaf is independent of the other traits.

312
PC4 accounted for an additional 10% of the total variability and is significantly associated 313 with Na and K accumulation in the leaf and root, but also, with leaf number and area, and 314 root length (Table 1).
Overall, the PCA indicates that in this experiment, the primary traits that varied and 316 correlated with each other were total fresh mass, ion content and some leaf traits, such as 317 elongation factor.

318
Cluster analysis suggests that salinity tolerance at the seedling stage is best defined by

332
Considering the values of the selected traits, the Euclidean distance between each accession 333 and the cluster mean was calculated to assign the accession to the nearest cluster. A new 334 mean value of each cluster was calculated after an accession was assigned to it and every 335 accession was checked again to see if they were closer to a different cluster. These steps were 336 iteratively repeated until convergence was achieved.

337
Bar plots were used to visualize the distribution of the accessions by cluster for each specific 338 trait, a similar visualization strategy is commonly used when plotting Q-matrices and 339 identifying K clusters in population structure studies (Pritchard et al., 2000). The accessions 340 were arranged in descending order and the bars are colored by cluster ( Figure S6). By 341 visualizing bar plots for all traits, it was easy to identify that the plant fresh mass was 342 predominantly defining the clustering by K=2. From this, it was observed that the accessions 343 of both species were best grouped by their fresh mass production under salt stress relative to 344 control conditions ( Figure S6). Thus, the two clusters divide the accessions of each species of 345 Galapagos tomato into those with high tolerance and low tolerance to salinity, in terms of Phenotypic data were also analyzed using a hierarchical clustering approach (Figure 6), 353 which was found to be more complex, but more informative, than the K-means clustering

365
In S. cheesmaniae, the hierarchical clustering method ( Figure 6A) separated the accessions 366 into two main clusters, which display contrasting values of fresh mass, leaf area, stem length, 367 leaf number, root length, leaf elongation, leaf Na and K concentration and green pixel count.

368
Within the cluster of accessions with high relative fresh mass, five different clusters could be 369 distinguished, that differed in leaf succulence, root K concentrations, root length, leaf Na + 370 concentration and leaf elongation (Table 2).

371
The S. galapagense accessions also separated into two clusters ( Figure 6B), based on their 372 relative fresh mass, leaf area, root Na, root K, leaf K, and green pixel count. The cluster with 373 high relative fresh mass was divided into two clusters differing in leaf succulence, leaf Na 374 concentration and root length (Table 2).

375
The phenotypic data collected were integrated into a Shiny App: Isla_Tomate, supported the roots, and protected them from breaking and allowing uncontrolled Na + influx 408 into the root system (Miller, 1987). An opaque substrate was preferred to simulate the light 409 blocking properties of soil and to limit algal growth. other aspects of cellular function, such as protein synthesis (Flowers and Dalmond, 1992).

474
The four most salt tolerant accessions of S. cheesmaniae and the two most salt tolerant 475 accessions of S. galapagense showed high K in their roots, which could indicate that they are 476 good at maintaining K uptake under salt stress. In some accessions, higher K in the roots 477 seems to go hand in hand with low K in the leaves, which could be explained by higher K + 478 re-translocation, useful to assist NO 3 uptake and distribution (Taleisnik and Grunherg 1994), 479 or lower K + translocation from roots to shoot. In addition, bigger leaves appeared to have a 480 lower K concentration compared to smaller leaves, which could be due to a dilution effect, 481 e.g. having a similar amount of K to that of the smaller leaves but more water content (Jarrell 482 and Beverly, 1981).

483
Increase in leaf succulence (measured as water per unit leaf area), a strategy to reduce salt 484 concentrations in photosynthetic tissues (Han et al., 2013), is another known mechanism of salinity tolerance in some plants, including tomato (Cuartero and Fernández-Muñoz, 1998).

486
The hierarchical clustering of accessions and traits showed that both S. cheesmaniae and S. 487 galapagense accessions each formed a cluster of accessions with increased leaf succulence 488 and low leaf Na concentrations. This might be caused by the succulence increasing cell size, 489 thereby diluting the salt without increasing the leaf (Munns et al., 2016).

490
The accumulation of Na + relative to biomass can also be an indicator of salinity tolerance.

491
However, Na + is toxic when it accumulates in the cell cytosol, resulting in ionic 492 disequilibrium (Hanin et al., 2016). Additionally, Na + reduces the availability of K + binding 493 sites for important metabolic processes in the cytoplasm (Wei et al., 2017). For the plant to 494 protect itself when exposed to salt stress, it has to either limit the entry of Na + through the 495 roots, or to control Na + concentration and distribution once inside (Tester and Davenport,              L e a f _ K . r e l   LA3124  LA0528B  LA1406  LA0422  LA0437  LA1404  LA0434  LA0166  LA1041  LA0429  LA1412  LA0927  LA1040  LA0428  LA0524  LA1043  LA1138  LA1427  LA1414  LA1409  LA1447  LA0531  LA1037  LA1448  LA1450  LA1042  LA0522  LA0746  LA1139  LA0749  LA0521  LA0932  LA1035  LA0529  LA1407 Plant_fresh_mass 0.0

Tables
Plant fresh mass in saline conditions relative to control conditions Plant fresh mass in saline conditions relative to control conditions