Quantitative resistance differences between and within natural populations of Solanum chilense against the oomycete pathogen Phytophthora infestans

Abstract The wild tomato species Solanum chilense is divided into geographically and genetically distinct populations that show signs of defense gene selection and differential phenotypes when challenged with several phytopathogens, including the oomycete causal agent of late blight Phytophthora infestans. To better understand the phenotypic diversity of this disease resistance in S. chilense and to assess the effect of plant genotype versus pathogen isolate, respectively, we evaluated infection frequency in a systematic approach and with large sample sizes. We studied 85 genetically distinct individuals representing nine geographically separated populations of S. chilense. This showed that differences in quantitative resistance can be observed between but also within populations at the level of individual plants. Our data also did not reveal complete immunity in any of the genotypes. We further evaluated the resistance of a subset of the plants against P. infestans isolates with diverse virulence properties. This confirmed that the relative differences in resistance phenotypes between individuals were mainly determined by the plant genotype under consideration with modest effects of pathogen isolate used in the study. Thus, our report suggests that the observed quantitative resistance against P. infestans in natural populations of a wild tomato species S. chilense is the result of basal defense responses that depend on the host genotype and are pathogen isolate‐unspecific.


| INTRODUC TI ON
Plant immunity can be observed in a qualitative manner, in which plants are either completely resistant or susceptible, or in a quantitative way, in which certain plants of the same species are more resistant and others are less resistant. Such intraspecific quantitative variation in resistance was formally introduced as horizontal resistance in the field of phytopathology by Vanderplank in 1963, and he further extended his concept in 1968. By his definition, horizontal resistance is a product of multiple underlying genes, that all have minor effects on the amount of resistance observed. This may explain why quantitative resistance is observed to be pathogen genotype independent in many cases. This is in contrast to vertical resistance, also called qualitative resistance, that depends on a single major resistance (R) gene and is pathogen genotype dependent.
Here, the product of specific R gene recognizes a specific pathogenic protein to deliver the resistance. Thus, plants with quantitative or horizontal resistance show a range of intermediate infection phenotypes but no complete resistance. Today the theories and concepts proposed by Vanderplank are well integrated in plant pathology and in use by plant disease epidemiologist in the field (Coutinho et al., 2020).
For natural populations, knowledge on quantitative resistances is limited, when compared to agriculture. However, a number of species have been reported to show quantitative resistance in their natural populations against several pathogens as highlighted in a review by Laine et al. (2011). Burdon (1980) was one of the first to show in detail the differences in resistance to the fungal pathogen Cymadothea trifolii and Pseudopeziza trifolii in a population of white clover, Trifolium repens, with a significant difference observed among the least and most resistant genotypes. Another notable example is a study on two populations of Senecio vulgaris. The species was shown to have up to 10 different resistance phenotypes against the powdery mildew fungus Erysiphe fischeri within an area of 1 m 2 of the population (Bevan et al., 1993). Some recent reports also highlight interesting aspects underlying the quantitative resistance observed.
A study on six natural genotypes of bittersweet, Solanum dulcamara, showed differences in tolerance and resistance against P. infestans and interestingly highlights a mechanism of overcompensation in some genotypes with increase in shoot length and number of flowers in plants with higher disease levels (Masini et al., 2019). Ultimately, these studies highlight host genotype-specific differences in resistance to biotic stress in wild plant species and show the potential of natural populations to be a valuable resource to explore and exploit quantitative resistance for evolutionary and economical aspects.
Here, we investigate phenotypic variation in resistance of the wild tomato species Solanum chilense. The natural habitat of S. chilense is southern Peru and northern Chile. The demography of this species can be described by four clearly defined genotype groups based on four geographical regions: one northern, one central, and two southern groups (Böndel et al., 2015). The regions are geographically separated from each other, and thus, all have their own specific climate. As a result, the populations growing in these four regions are genetically different and abiotic stress adaptations have been reported for example for genes involved in cold-and droughtstress responses (Böndel et al., 2018;Fischer et al., 2013;Nosenko et al., 2016). An advantage of working with S. chilense is that it is perennial. Hence, each plant represents a unique genotype from a distinct population but can be kept in the greenhouse and regularly cut back to maintain identical genetic material for years and comparable experiments.
The species has also been established as a genetic model for studies on plant-pathogen interactions. Some accessions of the species possess resistance genes against the fungal pathogen Verticillium dahlia or to tomato mottle virus (Griffiths & Scott, 2001;Tabaeizadeh et al., 1999). Genes of the Pto resistance pathway, conferring resistance to Pseudomonas sp., are shown to be under balancing selection (Rose et al., 2011).  previously showed that also R genes of the nucleotide-binding site, leucine-rich repeat receptor (NLR) gene family, show signatures of adaptation across the species. Using targeted resequencing of these important receptor genes, they showed that it depends on the population whether NLRs are under positive or balancing selection and that only a few NLRs show consistent positive or balancing selection signatures across the species. This led to the conclusion that indeed S. chilense is a highly variable host species, but that some adaptations, also in pathogen defense-associated genes, depend on or even drive colonization of new habitats . Most recently, we showed that genetic diversity in the species also has immediate effect on the observed resistance phenotypes of the host populations. Two specific Cladosporium fulvum resistance gene-dependent responses show presence-absence variations between and within populations of the central geographical group of S. chilense (Kahlon et al., 2020).
Interestingly, in this study, we also found complete loss of major R gene-mediated resistance responses in the southern part of the species range, which we hypothesize to be related to the extreme environments and probably absence of the pathogen in these regions.
The presence and absence of these R genes results in qualitative differences in resistance. However, as mentioned above, many interactions in nature are of quantitative nature. Populations from geographically distinct locations in the S. chilense species range show such quantitative variation in the resistance against three common filamentous pathogens of tomatoes, P. infestans, Alternaria solani, and a Fusarium species (Stam et al., 2017). The resistances against these three pathogens do not follow geographic patterns, but instead show a mosaic pattern of resistance quantities, which led to the conclusion that local pathogen pressures and separation of individual niches likely differ even within geographical regions.
The oomycete P. infestans belongs to the Peronosporaceae family and has large economical relevance (Kamoun et al., 2015). It was the cause of the Irish potato famine, and it causes significant losses as late blight pathogen on tomatoes and other Solanaceae species (Kaushal et al., 2020). It was one of the first oomycetes to have its genome sequenced (Haas, 2009), making it one of the best studied oomycetes at the genomic level. P. infestans contains more than 500 genes coding for so-called effectors, which are small specific proteins with recognizable amino acid domains that can contribute to the virulence of the pathogen (e.g., Bos et al., 2010). It is also known that the effector repertoire varies between isolates, with complete presence-absence variations as well as point mutations occurring in important effector genes (Thilliez et al., 2019). P. infestans occurs globally, with large variation between field populations: some being highly diverse, some being clonal, and some following sexual and some asexual reproduction mode (Fry et al., 2015). The origin of P.
infestans has been either in the Andes (Gómez-Alpizar et al., 2007;Martin et al., 2014) or in the Toluca Valley in Mexico (Goss et al., 2014).
P. infestans has likely been present in areas home to wild tomato species for a long time. A recent publication confirms that P. infestans can regularly be detected on wild potato and several tomato populations, growing in close proximity to cultivated potato fields in the Peruvian Andes (Lindqvist-Kreuze et al., 2020). Where cultivated potato and tomato crops are decimated when infected with these P. infestans isolates, these wild plants show often limited infection symptoms. Infection lesions appear to be smaller and do not spread over the whole plant (Lindqvist-Kreuze et al., 2020), suggesting a high level of quantitative resistance to be present in the wild plant populations.
To evaluate above hypothesized presence of quantitative resistance from former mentioned studies, in this study, we looked in detail at the phenotypic variation observed in natural populations of S. chilense, tested the quantitative resistance against various isolates of P. infestans, and elucidated the role of plant genotype and the pathogen isolate in determining quantitative resistance.

| Pathogen material and maintenance
Phytophthora infestans isolates EC1, NL88069, 06_3928A, and T30-4 were obtained from Wageningen University, Pi100, Pi078, Pi054, were obtained from INRAE Avignon. We used these P. infestans isolates as they have been reported to have differences in effector diversity, originating from different geographical locations and different hosts shown in Table S1 (Caromel et al., 2015;Thilliez et al., 2019). All the isolates were maintained on long-term maintenance medium, Rye A Agar (Stam, 2016a), and were subcultured on sporulation medium, Rye B Agar (Stam, 2016b), before experiments and incubated 9 days in the dark at 16°C and 1 day in the light at room temperature. The sporangia were scratched from the plates after flooding with water, with the help of a pipette tip and stored at 4°C for 1-2 hr prior to infection assays. The final concentration used for the infection assays and pathogen phenotype assays was 3,000 sporangia/ml.

| Detached leaf infection assays with isolate Pi100
The infection assays were performed on 85 plants representing nine populations of S. chilense using a drop inoculation method with sporangia of P. infestans isolate Pi100 on the detached leaf as described in Stam, Silva-Arias, et al. (2019). Leaves on our glasshouse plants differed in age and size. Therefore, the assays were performed in three biological replicates, each several weeks apart. For each biological replicate, seven to eight randomly selected leaves were harvested per plant from each of the mother plants (making up to 24 technical replicates per plant) and mother plants were cut back between experiments. Due to large number of plants and technical replicates, experiments were performed in two sets; in the first set, we tested five populations (LA1963, LA2932, LA3111, LA3786, and LA4330), and in the second set, we tested four populations (LA1958, LA2747, LA4107, and LA4117A). Population LA1963 was included in the second set as an internal calibrator to test potential effects of different experimental dates (Table S2).
To avoid influence of epiphytes in our infection outcome, we first evaluated presence of epiphytes in our mother plants in 10 randomly selected leaves. Each leaf was cut into four explants and transferred on four plates with commonly used culture medium for microorganism propagation, Luria-Bertani agar, V8 agar, 0.25 potato dextrose agar, and Spezieller Nahrtoffarmer agar, compositions shown in Table S3, and we used 70% surface sterilization to check the effect on reducing the commonly present epiphytes. We visually evaluated the presence of contaminants 7 days postexplant transfer ( Figure S2). As surface sterilization proved to be an effective method to avoid influence of fungal epiphytic contaminants, we performed surface sterilization of leaves with 70% ethanol for further assays, by dipping the leaves in the solution briefly, rinsing with water and gently dabbing with a dry tissue to remove the excessive water. The leaves were then placed in plastic boxes (50 × 32.5 × 6.5 cm) containing bedding of two layers of wet tissue paper as shown in schematic in Figure S1, with the abaxial site facing upward. Drop inoculation with 5 µl volume of liquid containing 3,000 sporangia/ml was performed on the individual leaflets of each leaf (up to 18 leaflets per leaf), and boxes were kept in the dark at 18-20°C. We observed still minor amount of bacterial contamination after surface sterilization ( Figure S2). To allow ruling out its influence in the outcome of our infection assays, we simultaneously performed control inoculations with water on four leaves for each plant in each biological replicate and we kept the plants out of the analysis that showed clear presence of the contaminants on either control or inoculated leaves ( Figure S3). Seven days postinoculation, the infection frequency was recorded per individual leaf by calculating the ratio of the number of leaflets showing mycelia growth divided by the total number of leaflets inoculated. A leaflet was counted as infected when we observed lesions that were larger in size than the originally placed droplet. Every infection event has a certain amount of stochasticity, as spore germination rates are not uniform, even under optimal conditions (Minogue & Fry, 1981). To reduce the variance caused by this effect, we performed a minimum of 1,588 drop inoculations per population. This number accumulates from a total of three independent experiments with inoculation of all individual leaflets of each seven to eight leaves from nine to ten individual plants per population.

| Phenotyping different P. infestans isolates on culture medium and plants
Differences in mycelial growth type of seven isolates were first confirmed on Rye A agar plates when the mycelial fully covered the plates after agar plug transfer ( Figure S4). To investigate the growth rate of these isolates, drop inoculations of the same sporangia solution used for the infections were done on Rye B Agar culture medium and incubated at 18-20°C in the dark. Phenotyping was performed with up to five technical replicates and three biological replicates per strain. Each biological replicate was performed at least 10 days apart, with freshly propagated starter cultures. The diameter of the mycelia growth (in cm) was taken 10 days postinoculation. Additional infection assays were performed on the population LA3111 (nine plants), which is a population from the central region of the species demography. We choose this population due to availability of a reference genome from this population . The same drop inoculation method was performed with 3,000 sporangia/ml of seven isolates of P. infestans (Pi100, Pi078, Pi054, EC1, NL88069, 06_3928A, and T30-4) on the detached leaves. The infection frequency was obtained from individual plants, using the same methodology as described above.

| Quantitative resistance differences between populations are independent of the pathogen isolate
To know whether populations of S. chilense possess quantitative resistance differences upon inoculation with a very aggressive P.
infestans isolate Pi100, originating from Solanum lycopersicum in Europe, we performed drop inoculations on the detached leaves.
We observed differences in quantitative resistance in the different populations upon infection with P. infestans isolate Pi100 (Figure 1).

LA4107 was most resistant population and LA4330 most susceptible.
We evaluated the effect of date of experiment using ANOVA with post hoc Tukey honest significant difference test and found no significant differences among the different dates of experiments (Table S2).
In our study, all plants were infected. ANOVA with post hoc Tukey honest significant difference tests showed significant differences between populations in 23 out of 36 pairwise comparisons (Table S4). Previously, Stam, Silva-Arias, et al. (2019) showed similar quantitative resistance differences when infected with P. infestans isolate EC1. We calculated the Pearson's correlation coefficient between the overlapping population's mean score of infection frequency from data on EC1 from the previous study (Stam, Silva-Arias, et al., 2019) and our infection frequency data on Pi100 from this study. This coefficient of 0.724162 indicates a positive correlation between the two data sets, which means that the S. chilense populations behave similarly to the two isolates. Due to a limited sample size (n = 6 overlapping populations), we cannot reasonably judge the significance of this correlation. We further had up to six nonoverlapping plants per population between the experiments accounting for additional variance within our samples. However, these results suggest that the variation in the infection frequency is the result of genetic differences in the host plants.

| Individual plant genotypes appear to drive resistance against single P. infestans isolate
To test whether the observed variance in the infection frequencies  (Tables 1 and S5). This indicates that the differences in successful infection of P. infestans can at least partly be attributed to genetic differences within the host populations; however, other factors beyond our control cannot be disregarded in this outcome (i.e., physiological differences between individual leaf explants from perennial plants).
To test whether the observed differences in infection frequency for each of the plants were statistically significant, we performed an ANOVA, with post hoc Tukey honest significant difference test for all plant-plant comparisons within a population (Table S6)

| Growth rate of different isolates of P. infestans on culture medium is significantly different
To understand whether resistance was isolate-specific, we selected seven diverse isolates of P. infestans. To first judge general differences of these seven isolates apart from the host plant, we evaluated their growth rate in vitro. To measure phenotypic growth differences, we performed drop inoculation of sporangia solution  Table S7).  (Table S8). To evaluate the differences in infection frequency correlation to growth rate of the P. infestans isolates, we compared the mean growth of P. infestans isolates on the plates with the mean of infection frequency of these isolates on the plants. We saw no correlation upon evaluation with a Pearson's correlation test (coefficient of −0.064, n = 7).

| Infection with seven different isolates of P. infestans supports host genotype dependency
Finally, we tested the effect of the plant genotype and the pathogen isolate on the infection frequency via multiway ANOVA.
We found both, the effect of the plant genotype and the pathogen TA B L E 1 Variance observed in infection frequency in S. chilense populations against P. infestans.

| D ISCUSS I ON
Solanum chilense is known for its resistance against various pathogens (Tabaeizadeh et al., 1999;Griffiths & Scott, 2001;Ji et al., 2007;Verlaan et al., 2013). It has also been studied extensively in relation to habitat adaptation (Böndel et al., 2018;Fischer et al., 2013;Nosenko et al., 2016;Xia et al., 2010) and as a model for the evolution of pathogen defense mechanisms in plants (Kahlon et al., 2020;. Previously, S. chilense populations from different geographical regions were shown to express different levels of resistance against three filamentous tomato pathogens (Stam, Silva-Arias, et al., 2019). In this report, we further characterize the resistance in this species against the oomycete pathogen P.
infestans as quantitative resistance which is largely determined by host genotypes.
In our previous work, we found significant differences in resistance of the different S. chilense populations but also had some very important remaining questions (Stam, Silva-Arias, et al., 2019). We  Figures 1 and 2), and the populations, but also significant differences between the populations (Figure 1) that correlated with our previous results (Stam, Silva-Arias, et al., 2019). The variance was reduced when looking into individual plants per population (Figure 2), which suggests that the resistance outcome is at least in part dependent on the plant genotype under consideration. One possible explanation for the observed variance between plants is the age and morphology of the used leaves, as older, thicker leaves might be harder to penetrate by the pathogen. The effect of plant age on potato leaflets has been highlighted and shown to be variable among flowering plants or nonflowering plants (Stewart, 1990). We tried to minimize this effect by regularly cutting back our plants and randomly selecting leaves for infection, but it remains possible that the selected leaves from some plants were less uniform than for other plants, because S. chilense is known to be a morphologically diverse species, with large variation between and within populations (Raduski & Igić, 2020) and many quantitative resistance phenomena have been described to be age and physiology-dependent (Niks et al., 2015). We also observed high variability of infection frequency on individual leaves from the same population or even from the same plant. This may have limited the number of statistical significant differences between our genotypes. However, due to the high number of repetitions, we consider these results robust and valuable. They reflect not just a single developmental stage or specific leaf level as considered in most model plant experiments but rather the spectrum of pathogen resistance responses in each given plant. Infections of wild and cultivated tomato with the pathogen A. solani also often resulted in no visible lesions on some of the leaves of generally susceptible accessions (Chaerani et al., 2007).
That within individual plants the variance observed was still high indicates other effects, such as stochastic effects affecting spore germination rates. Variation in spore germination rate is commonly observed with P. infestans, also under conductive laboratory conditions (Minogue & Fry, 1981). However, with the high spore concentrations used in our experiments, such variation should not have a strong effect on the outcome of our results.
Another limitation of our previous study was that we tested only a single isolate of P. infestans. It is well known that P. infestans strains are genetically diverse, with complete presence-absence polymorphisms of certain key virulence effector genes (Thilliez et al., 2019).
Also, different isolates of P. infestans have been shown to be tremendously different in their aggressiveness on a single host (Cooke et al., 2012). Testing a single isolate prevented drawing general conclusions on the observed resistance in S. chilense, as outcomes could have been isolate specific.
To address this, we looked into the infection frequency of the  (Tables 2 and 3). This does not need to depend on isolate germination and growth rate on agar plates, but can be related to various other factors that contribute to the general aggressiveness of individual isolates (Pariaud et al., 2009).
Even though we used seven different isolates, from different hosts, we used only one Latin American isolate from a clonal lineage that is spreading throughout Peru (Lindqvist-Kreuze et al., 2020). Therefore, there is a possibility that adapted pathogens (e.g., collected directly from S. chilense) would show different results, which would be interesting to explore in future. As shown by Thrall and Burdon (2003), in more resistant flax populations more virulent rust was observed. This mechanism is also shown in resistant plants of Plantago lanceolata and powdery mildew Podosphaera plantaginis (Laine, 2005 showed that all the plants possess a basal level of resistance against Phytophthora uniformis before invasion by the pathogen, but also that seedlings from pathogen-invaded locations were more resistant than seedlings from uninvaded locations. Exapted resistance is also shown in the case of powdery mildew, Erysiphe alphitoides, and Phytophthora cinnamomi in European oak (Quercus rubur) where multiple genes were shown to be responsible for the variation in resistance between the trees of naive, not previously infected, populations (Bartholomé et al., 2020).
The molecular mechanisms underlying the observed phenotypes in our study remain thus far unknown. We did not observe any host genotype-pathogen isolate combination that led to complete resistance or showed hypersensitive reactions, suggesting that no major R genes against the used isolates are active in those populations.
It has been shown that the perception of conserved M/PAMPs can differ within and between wild tomato species (Roberts et al., 2019). the activated defense mechanism was associated with geographical origin of the population (Ogran et al., 2016).
In a study with a panel of six domesticated and six wild tomato accessions, infection severity of 97 Botrytis cinerea isolates was shown to be determined by complexity of both host and pathogen genotype and a few pathogen genes associated with host preferences (Soltis et al., 2019). Similar multigenic effects has been indicated in a genome-wide association mapping analysis of 88 genotypes of quinoa, Chenopodium quinoa, showing large variation in disease traits upon inoculation with one isolate of the downy mildew pathogen Peronospora variabilis (Colque-Little et al., 2021). With the recent availability of a reference genome sequence for S. chilense , studies to identify the multiple genes underlying the observed differences in quantitative resistance will be the next step.
Our findings show an example of intraspecific variation in quantitative resistance in S. chilense plants from selected populations, in which we earlier reported variation in qualitative resistance against another pathogen (Kahlon et al., 2020). We show that this quantitative variation exists between and within the natural populations of the species and is mainly driven by the plant genotype. The resistance can potentially be defined as exapted resistance as it appears to protect against nonlocal, un-adapted pathogen isolates. Overall, we show that the plants of the wild tomato species S. chilense have their own unique resistance and that the species is amenable for detailed studies on quantitative resistance in the laboratory. In future studies, it would also be interesting to look into structural and molecular factors underlying these quantitative differences and possibly associate this with genetic differences.

CO N FLI C T O F I NTE R E S T
The authors declare that no competing interests exist.

O PE N R E S E A RCH BA D G E S
This article has been awarded Open Data, Open Materials Badges.
All materials and data are publicly accessible via the Open Science Framework at https://github.com/remco -stam/Kahlo nEcoEvol.

DATA AVA I L A B I L I T Y S TAT E M E N T
Supplemental tables are made available as supporting information.