Assessing the Effect of Diesel Fuel on the Seed Viability and Germination of Medicago sativa Using the Event-Time Model

The remediation of contaminated sites using plant-based techniques has gained increasing attention in recent decades. However, information on the effects of contaminant imbibition on seed viability and germination rates are often lacking in the literature. To this end, our research investigated, by means of an event-time model, the effect of diesel fuel imbibition on the seed viability and germination rate of Medicago sativa, a plant species with great potential for remediation of organic contaminants. The event-time model provided an accurate and biologically relevant method for analysing germination data. Our results reveal that the direct imbibition of diesel fuel by M. sativa seeds for ≤48 h, or their exposure to soil diesel fuel concentrations of 0–10 g/kg diesel fuel, affects their germination rates, as shown by increasing t50 values from 90.6 (±2.78) to 114.2 (±2.67) hours, without significantly affecting seed viability. On the other hand, diesel fuel imbibition of longer duration, or the exposure of M. sativa seeds to ≥20 g/kg diesel fuel-contaminated soils, leads to no further effect on time to seed emergence. Instead, these conditions compromise seed viability, resulting in a decrease in the proportion of germinated seeds from 0.91 (±0.03) in 10 g/kg diesel fuel contaminated soil to 0.84 (±0.04) and 0.70 (±0.05) in 20 and 30 g/kg diesel fuel-contaminated soils, respectively. The fact that low concentrations of diesel fuel and 0–48 h of direct imbibition delayed seed emergence without adversely affecting the percentage of viable seeds suggests that this inhibitory effect on germination at low diesel fuel exposure could be attributed more to physical constraints rather than biological damage on the seeds. The models used in this study provide an accurate and biologically relevant method for the analyses of germination data. This is vital since expensive germination experiments, be it in the field of toxicology or agriculture, deserve to be accurately analysed.


Introduction
There is a growing body of evidence that plant roots, in conjunction with their associated microbial communities, offer an effective treatment strategy for in situ remediation of contaminated soils [1,2]. Under a variety of environmental conditions, vegetation has been shown to enhance microbial degradation rates of organic chemical residues in soils [3]. Plant-based remediation (phytoremediation) is not a new concept. Constructed wetlands, reed beds and floating-plant systems have been widely

Effect of Diesel Fuel on Germination
The germination curves for M. sativa seeds sown in 0 (control), 5, 10, 20 and 30 g/kg diesel fuel contaminated soils are shown in Figure 1. When compared to nonlinear regression models, the event-time model provided the best fit for the germination data. As expected, the proportions of germinated seeds varied with time, being largest at intermediate monitoring intervals (48-120 h for 0 g/kg; 72-144 h for both 5 and 10 g/kg; 96-168 h for both 20 and 30 g/kg diesel fuel-contaminated soils) and smallest at the initial and final intervals when germination activity was low.  Table 1 provides a summary of important germination parameters from Figure 1. These parameters provide insight into the effect that diesel fuel imbibition had on the viability and germination rates of M. sativa seeds.

Concentration of Diesel Fuel in Soils (g/kg) b (Slope at t50) d (Upper Limit)
t50 (h) 0 (control) − 6. 16  The t50 values for seeds planted in 0 (control), 5 and 10 g/kg diesel fuel contaminated soils were 90.6 ± 2.78, 106.6 ± 2.74 and 114.2 ± 2.67 h, respectively ( Table 1). As shown in Table 1, the proportions of seeds in 5 and 10 g/kg diesel fuel-contaminated soils that germinated during the experimental period (indicated by the parameter d) were similar to that of the control samples. However, their germination slowed down, as indicated by the higher t50 values. On the other hand, while the t50 values for seeds in the 20 g/kg diesel fuel-contaminated soil remained the same as those for the 30 g/kg soil, the d value of the 20 g/kg diesel fuel-contaminated soil was greater than that of the 30 g/kg diesel fuel-contaminated soil. This indicates a possible significant effect of diesel fuel on M. sativa seed viability when sown in soils with diesel fuel concentrations of 20 g/kg or more. In addition, the value "1-d" indicates the proportion of the seeds that did not germinate during the experimental period owing to either non-viability or insufficient experimental duration. These seeds were considered to be right-censored.

Effect of Diesel Fuel Exposure on Seed Viability Using Triphenyltetrazolium Chloride
The effect of in vitro diesel fuel exposure on M. sativa seed viability can be seen in Figure 2. Diesel fuel imbibition reduced seeds viability, as shown by the number of seeds that were stained red  Table 1 provides a summary of important germination parameters from Figure 1. These parameters provide insight into the effect that diesel fuel imbibition had on the viability and germination rates of M. sativa seeds. The t 50 values for seeds planted in 0 (control), 5 and 10 g/kg diesel fuel contaminated soils were 90.6 ± 2.78, 106.6 ± 2.74 and 114.2 ± 2.67 h, respectively ( Table 1). As shown in Table 1, the proportions of seeds in 5 and 10 g/kg diesel fuel-contaminated soils that germinated during the experimental period (indicated by the parameter d) were similar to that of the control samples. However, their germination slowed down, as indicated by the higher t 50 values. On the other hand, while the t 50 values for seeds in the 20 g/kg diesel fuel-contaminated soil remained the same as those for the 30 g/kg soil, the d value of the 20 g/kg diesel fuel-contaminated soil was greater than that of the 30 g/kg diesel fuel-contaminated soil. This indicates a possible significant effect of diesel fuel on M. sativa seed viability when sown in soils with diesel fuel concentrations of 20 g/kg or more. In addition, the value "1-d" indicates the proportion of the seeds that did not germinate during the experimental period owing to either non-viability or insufficient experimental duration. These seeds were considered to be right-censored.

Effect of Diesel Fuel Exposure on Seed Viability Using Triphenyltetrazolium Chloride
The effect of in vitro diesel fuel exposure on M. sativa seed viability can be seen in Figure 2. Diesel fuel imbibition reduced seeds viability, as shown by the number of seeds that were stained red or pink during the triphenyltetrazolium chloride test (Figure 2). These results are summarized in Figure 3. The graph indicates that the imbibition of diesel fuel for between 0 and about 48 h had little effect on seed viability. The percentage viability of seeds exposed to 0, 24 and 48 h of diesel fuel imbibition were all greater than 90%. However, exposure of M. sativa seeds to 72 and 96 h of diesel fuel imbibition resulted in a decline in percentage seed viability to approximately 84 and 70%, respectively. This indicates that longer duration of direct exposure to diesel fuel affects the viability of M. sativa seeds.
Plants 2020, 9, x FOR PEER REVIEW 4 of 10 or pink during the triphenyltetrazolium chloride test ( Figure 2). These results are summarized in Figure 3. The graph indicates that the imbibition of diesel fuel for between 0 and about 48 h had little effect on seed viability. The percentage viability of seeds exposed to 0, 24 and 48 h of diesel fuel imbibition were all greater than 90%. However, exposure of M. sativa seeds to 72 and 96 h of diesel fuel imbibition resulted in a decline in percentage seed viability to approximately 84 and 70%, respectively. This indicates that longer duration of direct exposure to diesel fuel affects the viability of M. sativa seeds.

Discussion
This study demonstrates that diesel fuel exposure impacts on the seed viability and germination rate of M. sativa, and that these effects are dependent on contaminant concentration and/or duration of exposure. The study also reveals interesting agreement between the grouped-data event-time model and triphenyltetrazolium chloride-dependent viability tests.
To assess the ability of M. sativa plants to resist high levels of diesel fuel toxicity, the diesel fuel concentrations used in the germination study were comparable to, or higher than, those used in previous studies involving organic contaminants [19][20][21][22]. Similarly, diesel fuel imbibition was also or pink during the triphenyltetrazolium chloride test ( Figure 2). These results are summarized in Figure 3. The graph indicates that the imbibition of diesel fuel for between 0 and about 48 h had little effect on seed viability. The percentage viability of seeds exposed to 0, 24 and 48 h of diesel fuel imbibition were all greater than 90%. However, exposure of M. sativa seeds to 72 and 96 h of diesel fuel imbibition resulted in a decline in percentage seed viability to approximately 84 and 70%, respectively. This indicates that longer duration of direct exposure to diesel fuel affects the viability of M. sativa seeds.

Discussion
This study demonstrates that diesel fuel exposure impacts on the seed viability and germination rate of M. sativa, and that these effects are dependent on contaminant concentration and/or duration of exposure. The study also reveals interesting agreement between the grouped-data event-time model and triphenyltetrazolium chloride-dependent viability tests.
To assess the ability of M. sativa plants to resist high levels of diesel fuel toxicity, the diesel fuel concentrations used in the germination study were comparable to, or higher than, those used in previous studies involving organic contaminants [19][20][21][22]. Similarly, diesel fuel imbibition was also

Discussion
This study demonstrates that diesel fuel exposure impacts on the seed viability and germination rate of M. sativa, and that these effects are dependent on contaminant concentration and/or duration of exposure. The study also reveals interesting agreement between the grouped-data event-time model and triphenyltetrazolium chloride-dependent viability tests.
To assess the ability of M. sativa plants to resist high levels of diesel fuel toxicity, the diesel fuel concentrations used in the germination study were comparable to, or higher than, those used in previous studies involving organic contaminants [19][20][21][22]. Similarly, diesel fuel imbibition was also used to mimic extreme conditions of diesel fuel exposure [23], which enabled us to determine to what extent does extreme exposure to diesel fuel affect M. sativa seed viability.
Non-linear regression models are often used to model germination. However, in real life germination experiments, the underlying assumptions governing non-linear models (independence between proportions and variance homogeneity) are not satisfied. In contrast, the event-time model reflects the experimental design of germination experiments and allows meaningful biological interpretation of the germination data. Of course, the restriction F (0) = 0 in the event-time model indicates that the experiment is right-skewed [24]. Thus, it permits log-logistic, log-normal as well as Weibull-type models, which are all models with logarithm transformation [25]. It does, however, rule out models such as the Gompertz, logistic and normal models.
The right-skewness of this model reflects real-life situations in which some seeds do not germinate owing to non-viability or insufficient experimental duration. In these experiments, the seeds viability test using triphenyltetrazolium chloride revealed that M. sativa seeds exposed to 24 and 48 h of diesel fuel imbibition have similar percentage viability (92% and 91%, respectively) to that of the control samples. This indicates that the viability of M. sativa seeds was unaffected by up to 48 h of diesel fuel absorption. The ability of M. sativa to withstand diesel fuel-related biological damage for 48 h is an indication of its potential for biotechnological application in the phytoremediation of diesel fuel contaminated sites. On the other hand, diesel fuel imbibition for 72 h or more impacted seed viability, leading to a decline in the percentage of viable seeds (Figure 3).
The relative resistance of M. sativa to diesel fuel toxicity shows its suitability for the rhizoremediation of diesel fuel contaminated sites. It is worth noting that in actual field remediation approaches, seeds are not soaked in diesel fuel. Therefore, we expect that field lethal values (in hours) would be higher than the experimental values from this study. Similarly, actual percentage viability in the field would be expected to be higher than our experimental values. The implication of this is that M. sativa is potentially able to survive in diesel fuel contaminated sites. This ability of a plant to withstand contaminant toxicity or similar abiotic stress in the environment is an important factor in designing and establishing successful remediation and reclamation approaches [19].
The results of the germination experiment ( Figure 1 and Table 1) provide further biological details. Since the average germination time for M. sativa seeds is between two and four days at a temperature range of 18 to 30 • C, the experimental duration was set at nine days to enable the possible germination of all viable seeds. From the results, more than 90% of M. sativa seeds exposed to 0 to 10 g/kg diesel fuel contaminated soils germinated during the experimental period, indicating that up to 10 g/kg soil diesel fuel concentration did not affect viability. However, as revealed by the t 50 values, soil diesel fuel concentrations impacted significantly on time to germination (Table 1). This indicates increasing time to radicle emergence with increasing concentrations of diesel fuel.
Moreover, seeds exposed to 20 and 30 g/kg of soil diesel fuel concentrations gave lower d values ( Figure 1 and Table 1) than those in 0 to 10 g/kg soils. While the proportion of seeds in 20 g/kg and 30 g/kg diesel fuel contaminated soils that germinated during the experimental period varied, their t 50 values remained the same (approximately 136 h). This shows that higher concentrations of diesel fuel in soils affects seed viability rather than the time to germination. It can thus be concluded that up to 10 g/kg diesel fuel concentration affects the time to germination of M. sativa seeds without significantly affecting their viability. On the other hand, higher concentrations of diesel fuel result in a significant reduction in the viability of these seeds, without further affecting the time required for the seeds to emerge.
The mechanisms by which diesel fuel impacts on seed viability and germination rate can be classified into two: biological damage (toxicity), and physical constraints (oxygen and water repellence). Diesel fuel contains both volatile and non-volatile components [23,26,27]. Previous studies have shown that it is the volatile fraction, rather than the non-volatile components, that is primarily responsible for the inhibition of seed germination and plant growth [23,27], and that at temperatures of <20 • C, this effect is minimal, owing to reduced hydrocarbon volatility [28,29]. Medicago sativa seeds in diesel fuel tend to have a lag phase preceding germination (Figure 1), and this lag in germination increases with exposure to increasing diesel fuel concentrations. This can be attributed to the ability of the hydrophobic diesel fuel to create a water-repellent coating around the seeds. This consequently limits both oxygen and water absorption by M. sativa seeds, resulting in delayed germination. The fact that diesel fuel at low concentrations and 0-48 h of imbibition was delaying seed emergence without adversely affecting seed viability (as shown by the reduction of triphenyltetrazolium chloride to triphenylformazan) suggests that this inhibitory effect on germination could be attributed more to physical constraints rather than biological damage of the seeds. This is an important quality for biotechnological application since seeds used for phytoremediation purposes must be able to withstand biological damage.

Soil Preparation
The soil used for this experiment was a mixture of screened sand, soil, and composted organics, sold as "turf underlay" and obtained from Australian Native Landscapes Pty, Sydney, Australia. The soil was sieved using a 2 mm sieve to remove large particles. The soil textural class is dominantly sand (86.2% sand, 5.1% silt and 8.7% clay), with 9.3% organic matter content by loss on ignition and 0.18% total nitrogen content. The soil was then air-dried until a constant weight was achieved. Different concentrations of diesel fuel contaminated soils (0, 5, 10, 20 and 30 g/kg) were prepared by spiking the soil samples with appropriate amounts of diesel fuel. The diesel fuel used was petroleum diesel (also called petrodiesel), as opposed to synthetic diesel or biodiesel, and was obtained from a Shell service station along Epping Road, Macquarie Park, Sydney. The chemical composition is predominantly saturated hydrocarbons (C10 to C25 n-alkanes, iso-and cyclo-alkanes) and some aromatic hydrocarbons (e.g., alkylnaphthalenes and alkylbenzenes). The spiked soils were first mixed manually by hand, followed by a thorough mixing using a Sanfine portable electric 1800 W soil mixing machine (Model No. SF-HM1401/1401S, Taizhou, China). The mixing was performed for 15 min per pot (2 kg soil) with a break and manual shaking after every 5 min to achieve complete homogeneity.

Germination as Grouped Time-to-Event Data
One hundred seeds of M. sativa were placed in different 100 mm petri dishes containing 20 g of 0 (control), 5, 10, 20 and 30 g/kg diesel fuel-contaminated soils. The petri dishes were incubated at 20 • C in the Organic Geochemistry laboratory and monitored at 24 h intervals for a period of nine days. The initial emergence of radicle from the seed testa was used as evidence of germination. The seeds that did not germinate during the nine days were considered to be right-censored [17].

Seed Viability Test
In addition to the germination experiment, a triphenyltetrazolium chloride (TTC) test was conducted in petri dishes to estimate seed viability [23,30]. Triphenyltetrazolium chloride is a clear, water soluble compound (a salt) which is reduced by respiring tissues to yield triphenylformazan, a water-insoluble red pigment. Thirty M. sativa seeds were pre-soaked in petri dishes containing diesel fuel for 24, 48, 72 and 96 h in order to imbibe seeds prior to the test. At the end of the four respective periods, the imbibed seeds and the control samples were subjected to the triphenyltetrazolium chloride test as follows. A 1% triphenyltetrazolium chloride solution was prepared by dissolving triphenyltetrazolium chloride in distilled water. Each batch of seeds was placed in a beaker containing 50 mL of 1% triphenyltetrazolium chloride, and the beaker was covered. The beakers were placed in an incubator at 30 • C for 1 h. Following incubation, the liquid was decanted and the seeds were rinsed with distilled water until the wash water was clear. The seeds were blotted with a dry paper towel and the colour was observed. The seeds were classified into two categories according to their colour development, namely "red/pink" and "no colour", corresponding to "viable" and "not viable", respectively. This test was repeated to give five replicates per treatment concentration.

Statistical Analysis: Event-Time Model
Statistical analyses were performed using R [31]. Non-linear regression models are often used to model germination [32,33]. However, these models are problematic as they ignore the fact that successive observations on the germination curve are highly correlated. In other words, the total number of seeds that have germinated at a particular time is highly dependent on the number of seeds that germinated previously [34]. Moreover, variation in the proportions of germinated seeds will vary with time, being largest at intermediate monitoring intervals and smallest at the initial and final intervals when germination activity is low. This means that the fundamental assumptions underlying nonlinear regression, namely independence between proportions and variance homogeneity, are not satisfied [17]. Consequently, this results in overly precise parameter estimates of, for example, time to reach 50% germination (t 50 ), due to too small standard errors. Therefore, Ritz and Pipper [17] suggested a more appropriate approach where germination data were modelled as event times, that is, waiting times until germination no longer became possible due to termination of the experiment or non-viability of seeds. This approach provided a more adequate statistical description of the type of response that resulted from germination experiments [35] The present study used the drm package in library(drc) [31] to model the effect of diesel fuel exposure on seed viability as event-time data. This model is described using the following equation proposed by Ritz and Pipper [17]: The upper limit parameter d denotes the proportion of seeds that germinated during the duration of the experiment out of the total number of seeds present at the beginning of the experiment. The parameter b (excluding its sign) is proportional to the slope of F at time t equal to t 50 , while t 50 has the same interpretation as effective or lethal doses EC 50 or LC 50 but relative to d (the upper limit). Thus, t 50 refers to the time when 50% of the seeds that germinated during the experiment have germinated. This model reflects the experimental design of right-censored germination experiments while allowing meaningful biological interpretation of germination data. It also links the analysis of germination data with related dose-response analyses used in ecotoxicology where interest lies in obtaining a parametric model fit for an S-shaped curve.

Statistical Analysis: Viability Data
Viability data were analysed as binomial data using the 2-parameter log-logistic model in R [31]. As in the methods of Hose and Symington [36], a series of 2-parameter response curves including log-logistic, Weibull, log-normal and hormetic curves were fitted to the data, and the best fitting model based on Akaike information criterion was chosen. The e value from the model, also referred to as LC 50 , gives the duration of exposure (in hours) causing 50% reductions in viability of the test seed population. The 3-parameter log-logistic event-time model was fitted to the germination data. The event time considered is the time (hours) from sowing to germination, evidenced by the initial emergence of the radicle from the seed testa.

Conclusions
The examination of the effects of diesel fuel exposure on the viability and germination rate reveals that either the direct absorption of diesel fuel for up to two days or their exposure to soils contaminated with up to 10 g/kg diesel fuel affect germination rate, leading to delayed emergence of M. sativa radicle. However, these short durations and low concentrations of exposure to diesel fuel does not affect the viability of M. sativa seeds. This is an important quality for biotechnological application since seeds used for phytoremediation purposes must be able to withstand biological damage. Longer duration of diesel fuel imbibition and/or exposure to higher concentrations of soil diesel fuel results in a significant reduction in viable seeds. The event-time model used here provided an accurate and biologically relevant method for analysing germination data. These models incorporate the experimental design of right-skewness and non-continuous observation of germination process. We are confident that the results of this study will prove helpful in the design of plant-based remediation techniques.