1. Introduction
Physic nut (
Jatropha curcas L.), a tropical plant native to Mexico and Central America, is notable for its seeds, which contain approximately 36% oil [
1]. The agricultural potential of
J. curcas in tropical regions is affected by several diseases [
2,
3], including rust caused by
Phakopsora species. This basidiomycete was first identified in Puerto Rico in 1915 as
Uredo jatrophicola Arthur [
4] and was definitively reclassified as
P. arthuriana, with reports in the United States of America [
5], Brazil, the West Indies [
6], Mexico [
7], and Thailand [
8]. Physic nut leaf rust causes necrotic lesions with a chlorotic halo on the upper surface and a reddish coloration with eruptions (uredia) that form uredospores on the lower epidermis. These lesions lead to increased colonization and reproduction of the pathogen, which causes severe defoliation and drastically reduces the photosynthetically active rate of the plants [
9,
10].
In Ecuador, the National Institute of Agricultural Research (INIAP) began researching physic nut in 2007, aiming to preserve the genetic diversity and select genotypes with high productive and agroindustrial potential for the Ecuadorian coast. In late 2011, the Ecuadorian government launched an initiative to generate electricity for Floreana Island, one of the Galapagos Islands [
11]. In 2021, characteristic rust symptoms caused by
Phakopsora sp. were observed in physic nut hybrids and promising genotypes at the Technical University of Manabí (UTM). To the best of our knowledge, this disease has been minimally studied in Ecuador; thus, its etiology is poorly understood and its temporal progression within the crop remains unexplored.
Rust intensity can escalate under both intensive and extensive monoculture practices [
9,
12,
13] and is influenced significantly by environmental factors [
14,
15,
16]. The incidence of the disease ranges from 30 to 70% [
7,
9]. To date, no alternative hosts for physic nut rust have been identified, and a comprehensive description of its life cycle, which includes five stages, remains incomplete, with only the uredinial and telial stages documented in
J. curcas [
9]. Although rust infections can diminish the leaf area index, the yield, and some yield components in certain short-cycle crops [
13,
17], the impact of
P. arthuriana on physic nut is transient and heavily dependent on its epidemiology and the prevailing environmental conditions.
The modeling of rust’s temporal progress requires distinct approaches for perennial crops compared with annuals [
18]. Given the perennial nature of physic nut, the pathogen finds ample opportunity for survival, spread, infection, colonization, and reproduction across all seasons. In contrast, the progression of rust in annual plants is typically analyzed using Exponential, Logistic, and Gompertz nonlinear models [
17,
19,
20]. These models, however, are less suitable for perennials due to the disease’s variable or non-uniform behavior over time. The Weibull model, known for its flexibility, is more adept at capturing the temporal dynamics of diseases in perennial plants [
21]. This model’s effectiveness was demonstrated by García-López et al. [
22], who employed it to examine the temporal patterns of incidence and severity of malformation disease (
Fusarium spp.) in mango (
Mangifera indica) cultivars. Its adaptability allows for a more accurate representation of diverse disease progression curves, unlike more rigid models [
23].
The Weibull model is characterized by the following three parameters: the location parameter (‘a’), indicating the time of disease onset; the scale parameter (‘b’), which is inversely related to the rate of disease increase; and the shape parameter (‘c’), which determines the curve’s asymmetry and the inflection point’s position [
24,
25]. Its widespread use in life testing, even with small sample sizes, stems from the dramatic variations in the density function based on the ‘c’ value [
26]. The simplicity, flexibility, and precision of the Weibull distribution facilitate fitting diverse disease progression curves, making it an invaluable tool for modeling plant disease epidemics.
Based on the described antecedents, our research questions were as follows: (1) What is the fungal pathogen that causes rust on J. curcas? (2) How do physic nut genotypes respond to rust? (3) What is the temporal progress of the disease on J. curcas under field conditions? and (4) What is the correlation of the disease with some of the environmental conditions? Thus, the objective of this work was to characterize morphologically and to evaluate the intensity and etiology of the physic nut rust present in Ecuador under both semi-controlled and field conditions and analyze its temporal progress in the lower, middle, and upper canopy of adult plants of six genotypes of physic nut. In addition, we studied the correlation between environmental conditions and the disease under field conditions.
2. Materials and Methods
Research under semi-controlled (net house) and field conditions was conducted at the experimental site “La Teodomira” of the Faculty of Agricultural Engineering (FIAG) of UTM, located in the municipality of Santa Ana, province of Manabí, Ecuador (01°09′51″ S; 80°23′24″ W), at an altitude of 60 masl. This area is classified as having a tropical savanna climate (Aw), with clay loam soil and flat topography with slight undulations. The average dew point is 20.6 °C, with temperatures averaging between 22.1 °C and 26.7 °C (
Supplementary Figure S1). Furthermore, the area receives on average 81.8 daylight hours, with a relative humidity of 83.0% and an accumulated precipitation of 485.3 mm.
2.1. Morphological Characterization of the Causal Agent of Rust
Symptomatic leaves from established plants under field conditions and artificially inoculated seedlings were used for the morphological characterization of the rust’s causal agent. The rust populations, designated as FiagR1 (field conditions) and FiagR2 (semi-controlled conditions), collected in 2022 were analyzed both macroscopically and microscopically. Uredia on the leaves were scraped with a needle to primarily obtain uredospores, while leaf tissues were histologically sectioned using a classic edge blade to observe all potential pathogen structures. Samples were prepared on wet mounts with lactophenol and examined using a light optical microscope (CX22LED, Olympus, China). The dimensions of each structure were measured across 30 uredia and telia and 100 uredospores and teliospores using micrographs captured with a 10 Mpx camera (Better Scientific, Germany) attached to the microscope and analyzed with ImageJ version 1.52v software.
2.2. Artificial Inoculation of Seedlings and Evaluation of Rust under Semi-Controlled Conditions
Ten physic nut seeds from each of the six genotypes, previously established in the field, were planted in a sandy loam substrate. This substrate had been sterilized before use and the seeds were placed in 3 L volume polyethylene bags. These seedlings were then kept in a greenhouse, where the average temperature was maintained at 24.1 ± 2.5 °C and the relative humidity was kept above 90%. The seedlings were watered thrice weekly for 30 days. The inoculum, derived from a single fungal population, was prepared using symptomatic leaves that produced a large number of uredospores. These leaves were collected from adult plants cultivated at the Faculty of Agronomic Engineering. In the laboratory, the leaves were chopped into approximately 1 cm pieces and then blended into a homogeneous mixture with a sterile distilled water solution containing 0.1% Tween 20. The uredospore suspension was then extracted by filtering the mixture twice through a double layer of cheesecloth. The concentration of the suspension was adjusted to 1 × 106 uredospores mL−1 using a Neubauer counting chamber.
The seedlings, with their leaves fully expanded (between 20 and 23 days after sowing), from each physic nut genotype (hybrids JAT 001100, JAT 001103, JAT 001164, and JAT 001165 and the two promising genotypes CP-041 and CP-052) were inoculated using a modified version of the spray method described by [
27]. Five seedlings of each
Jatropha genotype were inoculated with the uredospore suspension using an airbrush attached to a compressor (1/5 HP, 110V, Truper, Mexico), ensuring all leaves were sprayed to the runoff point. Seedlings that were only sprayed with sterile distilled water served as controls and were kept in the same greenhouse but in a separate growth chamber from the inoculated seedlings to prevent the spread of and contamination by uredospores.
At 20 days after inoculation, disease intensity was assessed. Disease incidence (%) was calculated by counting the number of leaves with or without lesions (pustules). Rust severity was visually estimated by determining the percentage (%) of leaf area covered by rust pustules. Additionally, the number of lesions (pustules) per cm
2 was quantified. Each leaf was longitudinally divided using the central midrib as a dividing line (left and right side), the area was defined, and, with a cork borer (Ø 10 mm), the presence of lesions on both parts of the leaf was quantified [
27] with the assistance of a binocular stereoscopic microscope (model SMZ-168 TLED, Motic, Hong Kong, China).
2.3. Rust Evaluation in Adult Plants Established under Field Conditions
Adult physic nut plants of four imported hybrids (JAT 001100, JAT 001103, JAT 001164, and JAT 001165 from JatroSolutions GmbH, Stuttgart, Germany) and two promising genotypes (CP-041 and CP-052) were established at the experimental site “La Teodomira” in September 2020 (
Supplementary Table S1). Plants of each genotype were transplanted with 2 m between plants and 4 m between rows, covering a total area of 5376 m
2 (96 m × 56 m), under a completely randomized block experimental design. This setup was distributed across 24 plots, each consisting of 4 rows of 6 plants (24 plants per plot) and organized into four blocks.
Plants were fertilized with a source of urea (CH
4N
2O) at 16 g per plant, phosphorus pentoxide (P
2O
5) at 5 g per plant, and potassium oxide (K
2O) at 25 g per plant following a soil analysis conducted before transplantation (
Supplementary Table S2). A drip irrigation system was employed only during the initial months post-transplantation (September–December 2020); thereafter, no irrigation was provided. Weed control was conducted regularly using a motorized mower.
Three leaves from four randomly selected physic nut plants were collected from the two central rows of each plot (four plants), representing the lower, middle, and upper canopy of each plant (one leaf per canopy level). The plant tissues were placed in pre-labeled bags and transported to the phytopathology laboratory for rust evaluation. Disease intensity was assessed monthly over a nine-month period (from 29 November 2021 to 26 August 2022). The rust evaluation was not conducted beyond nine months due to the leaf drop and intense defoliation observed across all physic nut genotypes, particularly in the lower and middle canopy, starting on 12 September 2022. All monthly disease assessments consistently focused on the four designated plants. The incidence, severity (percentage), and number of rust lesions per cm2 on leaves of adult physic nut plants were evaluated using the same methods applied to seedlings established under semi-controlled conditions.
2.4. Statistical Analysis
Disease intensity datasets obtained in physic nut plants over time for each genotype were compiled into the area under the disease progress curve (AUDPC), calculated according to [
24]. After verifying the homogeneity of the variance and the normality of the residuals for the datasets obtained under field and semi-controlled conditions using the Bartlett and Shapiro–Wilk tests, respectively, the data were analyzed by Analysis of Variance (ANOVA). Subsequently, means were separated using either the Scott–Knott or the Kruskal–Wallis test (
p ≤ 0.05).
For the estimation of the maximum likelihood parameters of the three-parameter Weibull distribution, we used Function (1):
where
k is the shape parameter (c),
λ is the scale parameter (b),
η is the threshold parameter (a), and
x represents the observed data. The shape parameter determines the shape of the distribution, the scale parameter determines the dispersion of the distribution, and the threshold parameter represents the minimum value at which the distribution is defined [
25].
To obtain the maximum likelihood estimates (MLEs) of a, b, and c, we used the R package “weibullness” (
https://www.r-project.org/, accessed on 15 June 2023). First was calculated the likelihood of the sample data for one set of parameter values (a, b, and c), and then this process for different sets of parameter values was repeated. The Mean Squared Errors (MSEs) were then used to make inferences about the population from which the sample was drawn, and their validity was confirmed by checking that the model met the regularity conditions for maximum likelihood estimation.
The cumulative distribution function (CDF) of the three-parameter Weibull distribution is as described in Function (2):
where
x is the random variable, k is the shape parameter,
λ is the scale parameter, and
η is the threshold parameter. The CDF gives the probability that a random variable will take a value less than or equal to
x. It can be useful to provide insight into the probability of failure or the probability of a variable being in a certain range. Using the MLEs obtained from the likelihood function, the CDF can be used to make predictions about the probability of certain events occurring in the population and to estimate the reliability of a system or a product [
28].
Pearson correlations were obtained to examine the relationship between rust severity (%) and the number of lesions per square centimeter (cm2) under field conditions. These correlations were obtained in order to assess the relationship between the disease variables and environmental parameters, including dew point (Dew), maximum temperature (Tmax), mean temperature (Tmed), minimum temperature (Tmin), and precipitation (Prec). Statistical significance levels were established at p ≤ 0.001, p ≤ 0.01, and p ≤ 0.05. Also, the disease severity was correlated with two epidemiological parameters, namely a (the location parameter or the initial disease amount) and b (the scale parameter or the disease progress rate), using the same levels of statistical significance. All these statistical analyses were performed with Rstudio.
4. Discussion
Physic nut is an oleaginous plant with high oil content, making it ideal for industrial use, especially in biodiesel production. Nevertheless, few studies have been carried out on rust in this crop species worldwide [
7,
9,
29]. In this study, the rust pathogen associated with this problem in Ecuador was morphologically characterized and the rust intensity on six physic nut genotypes was evaluated under field and semi-controlled conditions on adult plants and seedlings, respectively. We also epidemiologically modeled the disease using the Weibull nonlinear model in adult plants and correlated some climatic parameters with the disease and the epidemiological parameters of the nonlinear model with the disease intensity.
Although the species
Uromyces amapaensis and
P. jatrophicola have been reported to cause rust on
Jatropha spp. (Euphorbiaceae), only the latter has been described on diseased leaves of physic nut (
J. curcas) plants [
7,
9,
29]. The shape, pigmentation, and size (length and width) of the uredia, uredospores, telia, and teliospores observed by us in physic nut plants and seedlings are similar to those obtained by Díaz-Braga et al. [
9], Nolasco-Gúzman et al. [
7], and Haituk et al. [
8], who describe
P. arthuriana as the causal agent of rust in physic nut plants. Although it would be interesting in the future to conduct a molecular characterization of physic nut rust populations in Ecuador, the morphological characterization and the pathogenic characterization conducted in this research allow us to indicate for the first time that the basidiomycete that causes physic nut rust in Ecuador is
P. arthuriana.
Seedling responses of hybrids and physic nut genotypes to rust under semi-controlled conditions differed from those observed under field conditions. For example, in greenhouse assays, the hybrid JAT 001165 exhibited higher severity, and this material, along with hybrids JAT 001103 and JAT 001164, displayed a higher number of lesions per cm
2 compared with the other genotypes. However, hybrids JAT 001100 and JAT 001103 showed a higher disease intensity under field conditions. Conversely, the promising genotypes CP-041 and CP-052 exhibited higher disease intensity under both environmental conditions. Nevertheless, all of the physic nut genotypes were susceptible to rust. Only two reports demonstrate the response of physic nut to rust under field conditions [
30] and with detached leaves [
9]. In the former, a maximum severity of 13% was recorded, while, in the latter, an average of 15.5 ± 2.3 lesions per leaf, 181.2 ± 19.5 pustules per leaf, and 12.6 ± 2.4 pustules per lesion were observed on the leaves of the abaxial side. Neither of these two studies mentions the specific genotype of physic nut used. Therefore, we cannot directly compare the response of our genotypes to rust with those of the other authors.
Under field conditions, rust lesions were observed on the leaves of all physic nut genotypes across the three canopy levels, reaching an incidence of 100%. Interestingly, these plants were just a little over a year old when we began to evaluate them, so the time it took for them to become diseased was relatively short. The incidence of rust observed in our genotypes is higher than the maximum found by Díaz-Braga et al. [
9] and Nolasco-Gúzman et al. [
7], at 45% and 70%, respectively, and it could even be the highest incidence of the disease reported in physic nut plants worldwide.
The rust’s temporal progress and intensity (severity and number of lesions per cm
2) differed between genotypes and canopies. For instance, in the upper canopy, we observed a higher variation between plants within each genotype and where the disease progress was practically linear in almost all genotypes. Only the hybrid JAT 001100 and the genotype CP-041 experienced an increase in rust intensity between March and June 2022 compared with the rest of the genotypes, while the rust decreased progressively in the hybrids JAT 001103 and JAT 001164 and the genotype CP-041, reaching values between 6 and 10%. Also, the AUDPC for the severity of rust was higher in the middle canopy and in the average of the canopies of the hybrid JAT 001100, and the AUDPC for the number of lesions per cm
2 was higher in the same material and the hybrid JAT 001103 in the average of the canopies, when compared with the rest of the genotypes. In general, hybrids JAT 001100 and JAT 001103 were the physic nut genetic materials most susceptible to the rust caused by
P. arturiana considering that the occurrence and severity of rust can vary over time and depend on the source of the inoculum, the environmental conditions, and the genetic base of the host [
2,
30].
The dew point is among the climatic variables most associated with the advancement of the pathogen infection. This factor favors the spread of the disease in plants along with precipitation due to the increase in humidity in the environment. Physic nut is a perennial plant, so the disease would be present in the plant when its leaves are attached. However, the rust infection observed in any plant stratum could also be affected by host defense mechanisms, the temperature (especially the minimum temperature), and the dew point. Due to their being members of a perennial cycle species, when the genotypes are exposed to the climatic conditions of the tropics the disease has enough leaf area to spread when climatic conditions are favorable. Identifying defense mechanisms against rust in physic nut plants will be an exciting challenge for future research.
The intensity and progression of rust were influenced by environmental factors such as dew point, minimum and mean temperature, precipitation, and relative humidity, but not by maximum temperature. In this research, we found positive and (mostly) negative correlations between some meteorological variables and severity (%) and the number of lesions per cm
2 in almost all canopy levels of the physic nut plants, except in the lower canopy. It is known that there is a genotype–environment interaction that affects the intensity of rust in plants [
31]. However, a genotype–environment interaction was not observed; instead, the rust intensity was influenced only in some genotypes and canopies. For example, in some physic nut genotypes, the disease was positively induced by dew point (JAT 001164 and JAT 001165) and relative humidity (JAT 001103) in the middle and upper canopy and negatively affected (all genotypes) by dew point, precipitation, and minimum and mean temperature in the middle canopy, upper canopy, and average of the canopies. Our results show that the rust infection in physic nut plants was variable throughout the year and highly dependent on the environmental conditions observed during the evaluation period. These results differ from those expressed empirically by Nolasco-Gúzman et al. [
7], who mentioned that a humid and temperate climate with rainfall throughout the year favors the progress of rust in physic nut plants. On the other hand, the infective potential of some rusts like
Uropyxis petalostemonis on the legume
Dalea candida decreases sharply at temperatures > 25 °C [
32]. Nevertheless, in our field experiment, we found that maximum temperature (between 25.8 and 28.8 °C) did not affect the severity and number of lesions per cm
2 associated with the infection of
P. arturiana in physic nut.
Under field conditions, we observed negative correlations between the number of lesions per cm
2 of the promising genotypes CP-041 and CP-052 in the lower canopy and between precipitation and the genotypes JAT 001103 and CP-052 in the middle canopy. The probability of pathogen dispersal is higher under a large amount of precipitation in soybean rust [
15]. However, under the conditions in which we conducted the experiment, it appears that physic nut rust is favored more by dew point and relative humidity than parameters such as precipitation. In fact, dew point is important for other rusts, such as soybean rust [
33].
As we mentioned earlier, we found no significant correlations between meteorological variables and rust severity in the lower canopy and the mean of all canopies and negative correlations with some parameters such as dew point, precipitation, and maximum and minimum temperature. Both pathogens and plants have an optimal environmental condition for their growth and reproduction. An optimal environmental condition is better for the outbreak of a disease, and if more climatic factors deviate from this “disease optimum”, a lower intensity of rust will occur on the plant [
34]. Interestingly, for physic nut, negative correlations were found only when the environmental conditions were correlated with the number of lesions per cm
2, but not with rust severity. In more-studied rusts, such as soybean rust (
P. pachyrhizi), the disease severity increases when increasing the number of lesions [
27]. In several diseases, the presence of a small number of lesions would be associated with few sites (anatomical features or host resistance), pathogen characteristics (a low number of infectious units or low infection efficiency), or microenvironmental constraints [
35]. Although microenvironmental constraints are related to our result, further studies are needed to understand how environmental factors negatively affect the number of physic nut leaf lesions.
The Weibull distribution analysis conducted in this research showed that most of the disease and environmental data fit this model of distribution well, suggesting that this model can be used to model rust in the physic nut crop or in another perennial crop. Almost all of the observed data were close to the fitted distribution. In addition, the estimated parameters, such as the initial disease amount and disease progress rate, agreed with the disease intensity data. For instance, high mean values were found for both epidemiological components in the lower and middle canopy of the hybrid JAT 001100, perhaps explaining why this material had the highest disease severity and number of lesions per cm
2. Coincidentally, in our study, the disease progress rate was significantly correlated with the severity and number of lesions per cm
2 in the middle and upper canopy. The Weibull model also has an additional parameter, the shape parameter (c), which can be used in the analysis of disease progression data. Perhaps the shape of the disease progression curve is indicative of a currently undefined type of host resistance [
21,
24].
The Weibull model can effectively describe the progress of rust in red raspberry (
Rubus idaeus L.); however, despite its flexibility, the model seems to have limitations in describing epidemics when the severity values are less than 5% [
23]. In contrast, the rust severity observed in the physic nut cultivars exceeded this threshold. On the other hand, not all disease progression curves are well or easily described by a growth curve model, especially in perennial crops [
36]. Alternative methods for quantifying the development of epidemics include the AUDPC and nonlinear models such as the Weibull model. Our results confirm the usefulness of the AUDPC in terms of the severity and number of lesions per cm
2 and the Weibull distribution model in terms of determining disease development in different physic nut cultivars because they integrate factors such as host response and pathogen aggressiveness as described by García-López et al. [
22] in a study on a malformation disease caused by
Fusarium sp. in mango. Although these authors did not correlate the disease or the AUDPC with the epidemiological parameters ‘a’ and ‘b’ as we did in this research, they found a relationship between the AUDPC, the final infection of the disease, and the apparent infection rate of the malformation disease in mango cultivars.
This research demonstrated that the Weibull model can effectively model rust and compare rust epidemics among physic nut genotypes. Moreover, the Weibull analysis allows for a better understanding of plant diseases from an epidemiological perspective, which has important implications for the future management of rust in physic nut crops. To the best of our knowledge, this is the first report on the modeling of the temporal progress of rust in physic nut. This is particularly important considering that physic nut could contribute to the sustainable production of food and bioenergy, the rehabilitation of degraded lands, and the reduction of atmospheric carbon dioxide in Ecuador. Therefore, the evaluation of rust on physic nut genotypes under both field and greenhouse conditions, along with adequate epidemiological modeling of the disease, is crucial for the development of improved cultivars in the field.