RESEARCH ARTICLE

 

Models for predicting pseudothecium maturity and ascospore release of Phyllosticta spp. in South African citrus orchards

 

 

Providence Moyo^I; Susan du Raan^II; Paul H. Fourie^{I, III}

^ICitrus Research International, Nelspruit, South Africa
^IIQMS Laboratories, Letsitele, South Africa
^IIIDepartment of Plant Pathology, Stellenbosch University, Stellenbosch, South Africa

Correspondence

 

 

---

ABSTRACT

Ascosporic infection plays a major role in the epidemiology of citrus black spot (CBS) in South Africa, a disease caused by Phyllosticta citricarpa. Phyllosticta pseudothecium maturation and ascospore release models have been integrated in infection models to predict the availability of the primary inoculum source. However, these models have not been validated on a broader data set and this study aimed to validate and improve these epidemiological models. New pseudothecium maturation and ascospore release models for P. citricarpa were developed, based on weather and ascospore trap data from 13 locations and up to five seasons. From the 29 data sets analysed, 3775 3-hourly periods with ascospore events were recorded on 1798 days; 90% of these events occurred between 16.0 °C and 32.1 °C (daily T_min and T_max
of 15.4 °C and 33.5 °C, respectively) and 75% occurred above a relative humidity (RH) of 55.9% (daily RH > 47.9%). Rain was recorded during 13.8% of these ascospore events and 20.0% of ascospore days. Using logistic regression, a Gompertz model that best predicted pseudothecium maturation, or the probability of onset of ascospore release, was developed and was markedly more accurate than the previously described models. The model consisted of DDtemp [cumulative degree-days from midwinter (1 July) calculated as (minimum + maximum daily temperature) / 2 - 10 °C] and DDwet (DDtemp accumulated only on days with >0.1 mm rain or vapour pressure deficit <5 hPa) as variables in the formula: probability of first ascospore event = exp(-exp(-(-3.131 + 0.007 x DDtemp - 0.007 x DDwet))). A Gompertz model [PAT = exp(-2.452 x exp(-0.004 x DDwet2))] was also developed for ascospore release; DDwet2 = DDtemp accumulated, from first seasonal ascospore trap day, only on days with >0.1 mm rain or vapour pressure deficit <5 hPa. Similar to the DDwet2 model described in a previous study, this model adequately predicted the general trend in ascospore release but poorly predicted periods of daily, 3-day and 7-day ascospore peaks.
SIGNIFICANCE:
• We developed a new pseudothecium maturation model from 29 data sets, comprising different climatic regions in South Africa, and validated previously published models. The new model was markedly more accurate in predicting the onset of ascospore release and can be used to improve existing CBS epidemiological models and improve risk assessment and management of CBS in South African citrus orchards.

Keywords: ascosporic infection, pseudothecium maturation, epidemiology, temperature, moisture

---

 

 

Introduction

Citrus black spot (CBS), caused by Phyllosticta citricarpa (McAlpine) van der Aa, is the most important fungal disease of citrus in South Africa, specifically due to the quarantine status of this pathogen in certain fruit export markets. The disease does not affect the internal fruit quality, but rather causes cosmetic lesions that reduce the fruit quality standard.^1,2 Fruit lesions form largely on maturing fruit from latent infections that occurred when fruit was immature.^1-4 The critical period for fruit infection in South Africa and Australia is the first 4-5 months after fruit set, whereafter fruit becomes more tolerant to infection.^1,5,6 In South Africa, Australia and Argentina, protective fungicide sprays are only required during this critical fruit infection period for effective control^3-8, but longer periods of protection are required under the highly CBS conducive conditions in São Paulo, Brazil⁹. Leaves are susceptible to latent infection during the 10 months after unfolding, but rarely show symptoms.¹⁰

Infection is caused by asexual pycnidiospores and sexual ascospores.⁴ Pycnidiospores are produced in pycnidia formed in leaf litter and certain fruit, leaf or twig lesions. Pycnidiospores ooze from pycnidia in a gelatinous mass and are typically washed down, leading to infections occurring relatively short distances (<80 cm) from the source.^1,11-13 However, in regions with frequent storms such as Florida (USA), pycnidiospores have been reported to contribute to the dispersal of CBS across tree rows.^14,15 Ascospores, on the other hand, are formed in pseudothecia from which they are forcibly ejected and are wind-dispersed.¹⁶ Whilst conditions required for germination are similar for both spore types (>12 h wetness at optimal temperature of 25-27 °C), ascospores play a more prominent role in CBS epidemiology in South Africa and Australia.^1,4,17

Most citrus leaves drop naturally after 2 years on the tree, predominantly at the end of winter and in early spring.¹⁸ Phyllosticta citricarpa is heterothallic^19,20 and mating occurs on decomposing leaf litter on the orchard floor to form pseudothecia^21,22. Alternating wet and dry conditions at mild temperatures (21-28 °C) are required for pseudothecium maturation, whereas long wet periods are detrimental.^1,10,23 Ascospore discharge occurs after the onset of pseudothecium maturity, with ascospore peaks typically occurring during summer months, declining into early autumn.^3,24-26 Rainfall as little as 3 mm triggers ascospore release^3,4, but dew is also considered to trigger ascospore maturity and discharge²⁷. Fourie et al.²⁵ reported ascospore release events of Phyllosticta spp. in the absence of a rainfall trigger and noted that other wetness factors, such as relative humidity, dew or irrigation, should be investigated.

Quantification of pseudothecium maturation and availability of P. citricarpa ascospores in orchards can be achieved by use of volumetric spore traps. This method can provide accurate measurement of cumulative ascospore release, but it is labour intensive and time consuming. An important consideration when using ascospore trap data is the fact that P. citricarpa ascospores cannot morphologically be distinguished from those of the common endophyte Phyllosticta capitalensis.^28-30 P. citricarpa appears to prevail over P. capitalensis in South African citrus orchards in CBS prevalent areas³⁰, but further research is required to elucidate the relative prevalence of these species in citrus orchards in different climatic regions. Recently described species of Phyllosticta²⁹ are currently unknown in South African citrus orchards, but their relative proportion will also need to be investigated if they are found to exist.

Effects of environmental factors on pseudothecium maturation have been studied in different pathosystems, including apple scab (Venturia inequalis) and pear scab (Venturia pirina), as a basis for development of systems to forecast release of ascospores.^31-33 Models that relate pseudothecium maturation and cumulative ascospore release to cumulative degree-days have effectively been in use in many countries for V. inequalis.³⁴In South Africa, results from Phyllosticta ascospore trapping by means of volumetric spore traps are routinely used by certain growers for decision support, to assess risk and improve CBS management. Ascospore trap data and weather data obtained for three areas over three seasons in the Limpopo Province of South Africa were previously used to model the effect of temperature and wetness on pseudothecium maturation and ascospore release.²⁵ These degree-day models were integrated into infection models used in pest risk assessment for P. citricarpa¹¹³⁵, as well as a web-based decision-support platform (www.cri-phytrisk.co.za) used by citrus growers in South Africa. The pseudothecium maturation and ascospore release models reported by Fourie et al.²⁵ were, however, built on a limited data set and needed to be validated using data from different geographical areas. In the present study, therefore, we aimed to validate and/or improve the models described by Fourie et al.²⁵ by using an extensive data set obtained from a diverse range of climatic regions in South Africa.

 

Materials and methods

Monitoring of ascospore release and weather parameters

The natural release of ascospores was recorded in 15 localities belonging to three provinces in South Africa: eight localities in Limpopo Province, six localities in the Eastern Cape Province and one locality in Mpumalanga Province. Ascospore release was monitored at 3-hourly intervals by use of volumetric spore traps (Interlock Systems, Pretoria, South Africa) as described by Fourie et al.²⁵ Monitoring of ascospore release was conducted over five seasons (2012-2016) in five localities in Limpopo (Letsitele A, Letsitele B, Letsitele C, Hoedspruit A and Hoedspruit B), three seasons (2014-2016) for the rest of the localities in Limpopo (Burgersfort, Ohrigstad, Musina A and Musina B), and over two seasons (2015-2016) in the Eastern Cape (Addo A, Addo B Sunland, Kirkwood A, Kirkwood B and Kirkwood C) and Nelspruit (Mpumalanga). Information on citrus type, GPS coordinates and prevalence of CBS at each location is presented in Table 1. In each location, hourly recordings of rainfall (mm), temperature (°C) and relative humidity (%) were provided by weather stations located in close proximity (<1 km) to the spore traps.

To investigate the relationships between the weather variables and the presence of ascospores (i.e. during the 3-hourly periods in which Phyllosticta ascospores were trapped), the hourly weather data were transformed into 3-hourly data as total rainfall, average temperature and relative humidity (RH). Thereafter, quantiles were estimated using the empirical distribution function in XLSTAT (version 2019.1.2; www.xlstat.com). Likewise, the data were summarised as daily data [minima, averages and maxima for temperature (T_min, T_avg, T_max) and RH (RH_min, RH_avg, RH_max), total rainfall and total number of ascospores trapped] and quantiles estimated.

Prediction of pseudothecium maturity and onset of ascospore release

Degree-day accumulation was used to determine the influence of weather variables (temperature, rainfall and relative humidity) on pseudothecium maturity and the onset of seasonal ascospore discharge. Onset of seasonal ascospore discharge was regarded as the date of the first meaningful discharge of Phyllosticta ascospores (>5 ascospores trapped per day). Cumulative degree-days were computed from daily weather data beginning on 1 July (biofix) as DDtemp = (T_min + T_max) / 2 - base temp, with a base temperature of 10 °C.²⁵ Degree-day accumulation was also calculated for rainy [DDrain = DDtemp accumulation only on days with measurable rainfall (>0.1 mm)], humid [DDvpd = DDtemp accumulation only on days with vapour pressure deficit (VPD) <5 hPa], as well as for rainy or humid days [DDwet = DDtemp accumulation only on days with measurable rainfall (>0.1 mm) or VPD <5 hPa].²⁵ Daily VPD was calculated as (1 - RH/100) x 6.11 X exp[(17.47 x T_avg)/(239 + T_avg].^25,33

Similar to Rossi et al.³³ and Fourie et al.²⁵, logistic regression analysis was performed on a subset of data for rainy or humid days (rainfall >3 mm or VPD <5 hPa) from 1 July to first meaningful ascospore discharge to model degree-day variables most predictive of onset of ascospore dispersal. The values 0 and 1 were used as dependent variables for when no ascospores were trapped, and when ascospores were trapped on that day, respectively. Independent variables were DDtemp, DDrain, DDvpd and DDwet. Best model was selected based on the coefficient of determination, adjusted following Nagelkerke, and root-mean-square error (RMSE). Model building was performed using data (594 cases in total) from the following locations and seasons: Letsitele C (2014 and 2015); Letsitele A, Letsitele B and Hoedspruit A (2012, 2014, 2015, 2016); Hoedspruit B (2012-2016); Ohrigstad and Musina B (2015 and 2016); Nelspruit (2015); Musina A, Addo A, Kirkwood C, Kirkwood A, Kirkwood B (2016). Data in Fourie et al.²⁵ were used for model evaluation (117 cases in total). Due to missing weather data and/or ascospore trapping data, data sets from the following areas and seasons were not considered in this analysis: Letsitele C (2012, 2013 and 2016); Letsitele A, Letsitele B and Hoedspruit A (2013); Musina A (2014 and 2015); Musina B and Ohrigstad (2014); Nelspruit (2016); Addo B and Sunland (2015 and 2016); Addo A, Kirkwood B, Kirkwood A and Kirkwood C (2015). The accuracy of the predictive model in distinguishing between true and false first ascospore events was determined by a receiver operating characteristic curve, which plots model sensitivity against specificity.

Modelling of ascospore release

Modelling of ascospore release was performed as described by Rossi et al.³³ and Fourie et al.²⁵ The relative ascospore dose was expressed as the daily proportion of ascospores trapped (PAT) and cumulated on a 0-1 scale.^33,36 The non-linear regression procedure in XLSTAT using a Gompertz function was then used to model PAT against DDtemp2, DDrain2, DDvpd2, or DDwet2 data, which were calculated as described for DDtemp, DDrain, DDvpd, and DDwet but using the first seasonal ascospore trap day as biofix.²⁵ Non-linear regression was conducted for the complete data set (data of all locations combined) with the various parameters. The best model (generic model) was selected using the coefficient of determination and RMSE. The generic model was compared with the respective data set specific models (site-specific models), as well as the ascospore release model proposed by Fourie et al.²⁵ The site-specific models were built by modelling PAT of each site against DDtemp2, DDrain2, DDvpd2, or DDwet2 data using non-linear regression. Following Fourie et al.²⁵, Pearson's correlation analyses of predicted and measured PAT were conducted to compare model performance. Additionally, daily, 3-day and 7-day ascospore peaks (accumulation in PAT) were correlated with predicted ascospore peaks for all data sets using Pearson's correlation analyses.

 

Results

Monitoring of ascospore release and weather parameters

Onset of ascospore release was generally earlier in the Northern parts of the country (Limpopo and Mpumalanga) in comparison to the Eastern Cape Province. The earliest ascospore release was recorded 62 and 83 days after 1 July in Limpopo and Mpumalanga, respectively, in comparison to 115 days in the Eastern Cape. The onset of release of Phyllosticta ascospores occurred as early as 1 September at Letsitele B during the 2016/2017 season and as late as 10 November at Kirkwood C during the 2016/2017 season (Table 2). DDtemp accumulated from 1 July until the first day of ascospore release ranged between 362.30 (Ohrigstad in 2015/2016 season) and 895.60 (Kirkwood C in 2016/2017) (Table 2), with a mean of 638.96. There were many days with measurable rain before first ascospore release in the Eastern Cape (ranged from 31 to 54) in comparison to 0 to 19 for the Northern areas (Table 2).

Ascospores were trapped throughout the day and night in this study. Greater numbers were captured between 9:00 and 15:00, but not at significantly higher levels (results not shown). Ascospore release was observed from September through to March, but large differences were observed in the number of ascospores trapped between localities and seasons (Table 3). Markedly higher numbers of ascospores were recorded in Hoedspruit A, particularly during the 2014/2015 season. Hoedspruit B had the second highest number of ascospores trapped, while the lowest number of ascospores was recorded in Ohrigstad, followed by Musina A during the 2016/2017 season. More ascospore events were recorded in Hoedspruit B than in Hoedspruit A.

From the 29 data sets analysed, a total of 3775 3-hourly periods with ascospore events were recorded; these were analysed separately for the 13 different locations before averages of the weather variables were calculated. The average median number of ascospores trapped per 3-h event was 510.0 spores/m³, up to a 95th percentile of 3769.6 spores/m³ and an average maximum of 36 997.2 spores/m³ (Table 4). The average first and fifth percentiles for temperature at which ascospores were trapped were 14.0 °C and 16.0 °C, respectively. The average first and fifth percentiles for RH at which ascospores were trapped were 20.7% and 34.0%, and 25th percentile 55.9% (Table 4). Rainfall was sporadically (13.8%) measured during the 3-hourly ascospore release events.

Ascospore events were recorded on 1798 days. The average median for number of ascospores trapped per day was 875.9 spores/m³, and the average maximum was 57 352.8 spores/m³ (Table 5). Daily minimum temperature and relative humidity values recorded during ascospore days were lower than those observed for 3-hourly intervals (Table 4). The average first and fifth percentiles for T_min on days when ascospores were trapped were 13.7 °C and 15.4 °C, respectively. The 25th percentile values recorded on ascospore days for RH_min, RH_avg and RH_max were 47.9%, 58.5% and 64.1%, respectively (Table 5). Median values for daily T_min, T_avg and T_max were 20.6, 22.1 and 23.3 °C, respectively. Rainfall was measured on 359 days (20% of cases), and in most cases was <5 mm/ day (the 95th percentile was 4.8 mm) (Table 5).

Prediction of pseudothecium maturation and onset of ascospore release

The logistic regression model that best predicted the probability of onset of ascospore release had an R²(Nagelkerke) value of 0.699 and consisted of DDwet and DDtemp as variables in the formula: probability of first ascospore event = exp(-exp(-(-3.131 + 0.007 x DDtemp - 0.007 x DDwet))). Using a probability of 0.5 to predict onset of ascospore release, this model (herein referred to as the DDwet pseudothecium maturation model) gave a true positive proportion of predicted first ascospore events (sensitivity) of 0.55, i.e. the model accurately predicted 21 of 38 actual first ascospore release events (Table 6). The model displayed a very high true negative proportion (specificity) of 0.98 as it predicted 544 of the 556 events without ascospore release. A sensitivity value of 0.95 (36 of the 38 actual ascospore discharges were accurately predicted) and specificity value of 0.81 (correctly predicting 64 of 79 events without ascospore release) were achieved by the model in the validation data set (Table 6). The area under the receiver operating characteristic curve was 0.975 (results not shown).

 

 

When compared with the temperature and temperature/moisture pseudothecium maturation models, described by Fourie et al.²⁵, in predicting the actual pseudothecium maturation date (i.e. first meaningful ascospore release date per season) using a probability of 0.5, the DDwet pseudothecium maturation model was generally more accurate. It accurately (within 14 days) predicted 19 of 29 actual ascospore release events, across all locations and years tested; on average across data sets, the DDwet pseudothecium maturation model predicted onset of ascospore release 1 day later than the actual. In cases in which the model was not very accurate, differences of up to 27 days occurred between the predicted and observed times of onset of pseudothecium maturity (Table 7). On the other hand, the temperature and temperature/ moisture models²⁵ predicted 18 and 16 of the 29 actual ascospore release events, respectively; however, these models' predictions were on average, respectively, 10 and 16 days later than the actual (Table 7).

Modelling of ascospore release

The use of Gompertz equations in the non-linear regression analysis of PAT against DDrain2, DDwet2, DDvpd2 or DDtemp2 in the complete data set, revealed DDwet2 as the most suitable predictor of seasonal Phyllosticta ascospore release trends. Although the highest R²value of 0.820 (RSME = 0.148) was achieved in the non-linear regression analysis of PAT against DDtemp2, the model poorly predicted periods of ascospore release or their absence, due to the consistent increase in DDtemp2 (results not shown). PAT was poorly predicted from DDvpd2 (R² = 0.420; RMSE = 0.271). The DDrain2 (R² = 0.716; RMSE = 0.186) and DDwet2 (R² = 0.746; RMSE = 0.176) models, on the other hand, adequately predicted the general trend in ascospore release, with events predicted when DDrain2 or DDwet2 increased. The DDwet ascospore release model using DDwet2 as an explanatory variable was chosen as the best model based on its higher R² value and lower RMSE and also because it supports observations made during ascospore trapping, i.e. rain was not always a prerequisite for ascospore release: PAT = exp(-2.452 (standard error 0.0372) X exp(-0.004 (standard error 0.0005) x DDwet2)).

Non-linear regression of PAT against DDwet2 for each site and year resulted in good fits with coefficients of determination ranging from 0.821 to 0.993. The end values of PAT predicted by the site-specific models ranged from 0.811 to 1.000, and generally were >0.815 for the generic and published models; however, in two cases, the predicted final PAT values were as low as 0.569 and 0.528 (Letsitele A in 2014/2015 season) and 0.666 and 0.655 (Table 3). In both these cases, the PAT was predicted from markedly lower DDwet2 values (final DDwet2 values of 371.8 and 454.2), compared with the other data sets (627.4-1644.6). Final DDwet2 values did not correlate with cumulative ascospore counts, even when comparing per location across seasons.

The newly described generic DDwet ascospore release model behaved similarly in predicting PAT to the DDwet model described by Fourie et al.²⁵, as can be observed in Figure 1 (a-c), which displays the onset of ascospore release as predicted by the DDwet pseudothecium maturation model, observed seasonal ascospore data, daily rainfall and PAT predicted by both the generic and site-specific DDwet ascospore release models, as well as the published DDwet model²⁵. Lag phases following onset of ascospore release until PAT began to increase to more than 0.1 ranged from 0 to 6 weeks. Onset of ascospore release was generally predicted during these lag phases by the DDwet pseudothecium maturation model (e.g. Figure 1a, b), and in some cases not (Figure 1c). At a probability of 0.5, the DDwet pseudothecium maturation model predicted onset of ascospore release when actual PAT was less than 0.1 in all cases, except for Addo A, Kirkwood B and Hoedspruit A (2012/2013 season) (Table 7, Figure 1). The trends of the lag phases and subsequent exponential increase in ascospore release were in most cases accurately predicted by the site-specific and generic DDwet ascospore release models, as well as the published model (Figure 1). The three DDwet ascospore release models followed the trend of measured ascospore release fairly accurately, but generally predicted ascospore peaks poorly. In all cases, the models correctly predicted ascospore peaks during certain days, missed ascospore peaks on others and also predicted false peaks (Figure 1). Graphs of the results from the remaining locations and/or seasons are not shown.

The models predicted trends in seasonal ascospore dispersal accurately: Pearson correlations between actual and predicted daily PAT ranged from 0.906 to 0.996 for site-specific models, whereas those for the generic DDwet ascospore release model and the model described by Fourie et al.²⁵ ranged from 0.829 to 0.995 and 0.789 to 0.995, respectively. Prediction of the actual daily ascospore peaks by the site-specific models was poor (0.018-0.448) (Table 3), and daily peak predictions were even poorer for the DDwet ascospore release model and the model described by Fourie et al.²⁵ (results not shown). The sum of rolling 3-day (each particular day plus previous 2 days accumulation in PAT) and 7-day ascospore peaks were also correlated with these ascospore peaks predicted by the models. This slightly improved the outcome of the correlations for some locations but correlation coefficients were poor in most cases, ranging from -0.007 to 0.594 and 0.039 to 0.784 for 3- and 7-day peaks for the site-specific models, respectively, and even poorer for the other models (Table 3).

Further ascospore peak prediction comparisons involved classifying each day as '1' if one or more ascospore events occurred or as '0' if no ascospore event occurred. These binary data were then used to calculate 3-day and 1-day ascospore peaks. Pearson's correlation coefficients between actual PAT data and predicted PAT data were calculated, and similar to the previous peak prediction analysis, the correlation coefficients were generally poor (results not shown).

 

Discussion and conclusions

In South Africa, CBS is generally controlled by the repeated application of fungicides, targeted at the primary inoculum (ascospores). The use of mathematical models to estimate the maturity of pseudothecia of P. citricarpa is therefore important in the management of CBS because they predict the start of ascospore release, which is key in determining when fungicide applications need to begin in the field. Information on ascospore availability combined with infection model output better informs the decision on whether a protective or curative fungicide should be applied, and the number of infection periods and inoculum pressure informs the general CBS infection risk, as is contemplated in the CRI-PhytRisk application (www.cri-phytrisk.co.za). To date, the Phyllosticta ascospore availability models were published by Dummel et al.²⁶ and Fourie et al.²⁵, of which the models described by Fourie et al.²⁵ were subsequently used in CBS risk assessment studies^11,35 and in CRI-PhytRisk.

The present study evaluated the performance of models described by Fourie et al.²⁵ against new data obtained from several geographical locations with differing climatic conditions, and also described a more accurate pseudothecium maturation model. This newly described model considers both wetness and temperature as the two main weather factors that influence the maturation of pseudothecia of Phyllosticta spp., which is consistent with published literature.^{1,3,4,10,23,25,26} The temperature model described by Fourie et al.²⁵ uses DDtemp as the sole variable and predicts pseudothecium maturation in the absence of wetness. This model was favoured for use in pest risk assessment studies^11,35, largely due to some aberrant predictions from the related temperature/moisture model (PH Fourie, personal observation). The model developed in the present study considers that the pseudothecium maturation process progresses when wet conditions occur in combination with moderate spring temperatures above a baseline of 10 °C. Alternate wetting and drying at temperatures between 21 °C and 28 °C is required for maturation of the pseudothecium of P. citricarpa.^{1,3,4,10,23,25,26}The DDwet pseudothecium maturation model described here is a significant improvement on the temperature and temperature/moisture models described by Fourie et al.²⁵ and more accurately predicted onset of ascospore release.

Ascospore release occurred at lower temperatures in this study, compared to the values reported by Fourie et al.²⁵ Fourie et al.²⁵ reported that 90% of ascospore events occurred at temperatures between 11.8 °C and 33.0 °C (daily T_min and T_max of 15.1 °C and 35.5 °C), while 16.0 °C to 32.1 °C (daily T_min and T_max of 15.4 °C and 33.5 °C) is the range of temperatures at which 90% of ascospores were trapped in the present study. Reports on the relationship between ascospore trapping and rainfall have also been inconsistent. Previous studies found that rainfall was a requirement for ascospore release.^3,24 In this study, ascospore release did not always coincide with rainfall periods, which is in agreement with observations made by Fourie et al.²⁵ This indicates that other sources of moisture such as irrigation, dew and relative humidity may be playing a role in ascospore discharge.^1,26,21,31 Reis et al.³⁸ reported that ascospore release was more related to the duration of leaf wetness than the amount of rainfall. Similar to the 59.3% RH_avg reported by Fourie et al.²⁵, more than 15% of ascospores were released during 3-hourly periods with an RH_avg above 55.9% (and days with RH_min >41.9%), which supports the possible role of high RH in triggering ascospore release.²⁵ High humidity can prolong wetness of leaf surfaces which accelerates the maturation and opening of pseudothecia.²⁶ Contrary to our findings, Dummel et al.²⁶ reported that ascospore release started after a drop in RH after midday and postulated that leaf litter surfaces need to dry for a period of time to allow ascospores to be successfully ejected into the air.

Higher numbers of ascospores were captured during the day, reaching a peak at 12:00 to 15:00. Fourie et al.²⁵ and Dummel et al.²⁶ found greater ascospore numbers from 12:00 to 21:00 and 16:00 to 20:00, respectively, while no differences were found in the pattern of ascospore release during the day and night in Brazil³⁸. No correlations were found between more humid seasons and the number of ascospores trapped, when comparing cumulative DDwet2 and ascospore trap numbers. Pseudothecium maturation is hindered in areas where the leaf litter is constantly dry or wet.^1,23 CBS is a polyetic epidemic, i.e. inoculum builds up over time, and the inoculum pressure and disease incidence is expected to differ among orchards and years. This could further explain the differences observed in the number of ascospores trapped and ascospore release events between seasons and localities in this study.

As expected, higher numbers of ascospores and ascospore events were observed in areas of high CBS prevalence, i.e. Hoedspruit A, Hoedspruit B, Letsitele B and Letsitele C compared to areas with moderate CBS prevalence (locations in the Eastern Cape) as well as areas of low CBS prevalence (Ohrigstad and Musina A). Ascospore release was observed from September through to March, but peaks were observed at different times among the years and locations, but generally followed trends reported previously.^3,25,26,38 There was no direct relationship between rainfall and number of ascospores captured, as was also found in previous studies.^25,26,38 Ascospore release is triggered by small amounts of rainfall and as long as leaf litter surfaces remain moist, a few ascospores will continue to be released.^25,31 This may explain the release of ascospores in small numbers, but with occasional considerable increases in numbers (peaks), often observed in this study.

The ascospore release model developed in this study, as well as that of Fourie et al.²⁵, used mild to warm temperatures on humid or rainy days (DDwet2) as the climatic driver of ascospore release and were accurate in predicting the general trends in ascospore release, and are useful to predict the lag phases at the start and end of the ascospore release cycle, as well as the period of exponential increase. However, the models poorly predicted daily, 3- and 1-day ascospore peaks, which limits their potential use, for example, in integration in infection models or forecasting platforms. It is possible that ascospore release patterns are influenced by microclimatic weather variables (including leaf wetness^26,21,38), which are not necessarily correlated with mesoclimatic data, and this possibility should be investigated in future studies.

The DDwet pseudothecium maturation model, developed in this study, was markedly more accurate in predicting the onset of ascospore release and will undoubtedly benefit existing CBS epidemiological models and improve risk assessment and management of CBS in South Africa.

 

Acknowledgements

We thank Citrus Research International and the Department of Science and Innovation (South Africa) for financial support.

 

Competing interests

We declare that there are no competing interests.

 

Authors' contributions

P.M. was responsible for data analysis and the first draft of the paper. S.d.R. was responsible for ascospore trapping, and the compilation and preparation of data sets. PH.F. conceptualised the study, and participated in data analyses and finalisation of the paper.

 

References

1. Kiely TB. Preliminary studies on Guignardia citricarpa n. sp., the ascigerous stage of Phoma citricarpa McAlp., and its relation to blackspot of citrus. Proc Linn Soc NSW. 1948;13:249-292.         [ Links ]

2. Kotzé JM. Black spot. In: Timmer LW, Garnsey SM, Graham JH. Compendium of citrus diseases. 2nd ed. St. Paul, MN: The American Phytopathological Society; 2000. p. 23-25.         [ Links ]

3. McOnie KC. Orchard development and discharge of ascospores of Guignardia citricarpa and the onset of infection in relation to the control of citrus black spot. Phytopathology. 1964;54:1148-1453.         [ Links ]

4. Kotzé JM. Epidemiology and control of citrus black spot in South Africa. Plant Dis. 1981;65:945-950. https://doi.org/10.1094/pd-65-945        [ Links ]

5. Schutte GC, Mansfield RI, Smith H, Beeton KV. Application of azoxystrobin for control of benomyl-resistant Guignardia citricarpa on 'Valencia' oranges in South Africa. Plant Dis. 2003;87:784-788. https://doi.org/10.1094/pdis.2003.87.7.784        [ Links ]

6. Schutte GC, Kotze C, Van Zyl JG, Fourie PH. Assessment of retention and persistence of copper fungicides on orange fruit and leaves using fluorometry and copper residue analyses. Crop Prot. 2012;42:1-9. https://doi.org/10.1016/j.cropro.2012.04.015        [ Links ]

7. Miles AK, Willingham SL, Cooke AW. Field evaluation of strobilurins and a plant activator for the control of citrus black spot. Australas Plant Path. 2004;33:371-378. https://doi.org/10.1071/AP04025        [ Links ]

8. Fogliata GM, Munoz ML, Rojas AA, Ploper LD. Eficacia de tres estrobilurinas para el control de mancha rojiza (Guignardia mangiferae) y mancha negra (Guignardia citricarpa) en frutos de limon, en Tucuman, Republica Argentina [Efficacy of three strobilurins for the control of red spot (Guignardia mangiferae) and black spot (Guignardia citricarpa) on lemon fruits, in Tucuman, Argentine Republic]. Rev ind Agric Tucumán. 2011;88:37-15. Spanish.         [ Links ]

9. Lanza FE, Metzker TG, Vinhas T, Behlau F, Silva Junior GJ. Critical fungicide spray period for citrus black spot control in São Paulo state, Brazil. Plant Dis. 2018;102:334-340. https://doi.org/10.1094/PDIS-04-17-0537-RE        [ Links ]

10. Truter M. Epidemiology of citrus black spot disease in South Africa and its impact on phytosanitary trade restrictions. [PhD thesis]. Pretoria: University of Pretoria; 2010.         [ Links ]

11. Wager VA. The occurrence of the black-spot fungus in the citrus areas of South Africa. Farming in South Africa. 1949;24:367-374.         [ Links ]

12. Whiteside JO. Sources of inoculum of the black spot fungus, Guignardia citricarpa in infected Rhodesian citrus orchards. Rhodesia, Zambia, Malawi J Agric Res. 1967;5:171-177.         [ Links ]

13. Spósito MB, Amorim L, Bassanezi RB, Yamamoto P Felippe MR, Czermainski ABC. Relative importance of inoculum sources of Guignardia citricarpa on the citrus black spot epidemic in Brazil. Crop Prot. 2011;30:1546-1552. https://doi.org/10.1016/j.cropro.2011.08.007        [ Links ]

14. Hendricks KE, Christman M, Roberts PD. Spatial and temporal patterns of commercial citrus trees affected by Phyllosticta citricarpa in Florida. Sci Rep. 2017;7, Art. #1641. https://doi.org/10.1038/s41598-017-01901-2        [ Links ]

15. Hendricks KE, Christman MC, Roberts PD. The effect of weather and location of fruit within the tree on the incidence and severity of citrus black spot on fruit. Sci Rep. 2020;10, Art. #1389. https://doi.org/10.1038/s41598-020-58188-z        [ Links ]

16. Spósito MB, Amorim L, Ribeiro PJ Jr., Bassanezi RB, Krainski ET. Spatial pattern of trees affected by black spot in citrus groves in Brazil. Plant Dis. 2007;91:36-40. https://doi.org/10.1094/PD-91-0036        [ Links ]

17. Magarey RD, Hong SC, Fourie PH, Christie DN, Miles AK, Schutte GC, et al. Prediction of Phyllosticta citricarpa using an hourly infection model and validation with prevalence data from South Africa and Australia. Crop Prot. 2015;75:104-114. https://doi.org/10.1016/j.cropro.2015.05.016        [ Links ]

18. Erickson LC, Brannaman BL. Abscission of reproductive structures and leaves of orange trees. P Am Soc Hortic Sci. 1960;75:222-229.         [ Links ]

19. Amorim R, Savi DC, Ferreira-Maba L, Aluizio R, Goulin EH, Takita MA, et al. MAT gene idiomorphs suggest a heterothallic sexual cycle in the citrus pathogen Phyllosticta citricarpa. Eur J Plant Pathol. 2017;147:325-337. https://doi.org/10.1007/s10658-016-1005-8        [ Links ]

20. Wang N-Y Zhang K, Huguet-Tapia JC, Rollins JA, Dewdney MM. Mating type and simple sequence repeat markers indicate a clonal population of Phyllosticta citricarpa in Florida. Phytopathology. 2016;106:1300-1310. https://doi.org/10.1094/phyto-12-15-0316-r        [ Links ]

21. Tran NT, Miles AK, Dietzgen RG, Dewdney MM, Zhang K, Rollins JA, Drenth A. Sexual reproduction in the citrus black spot pathogen, Phyllosticta citricarpa. Phytopathology. 2017;107:732-739. https://doi.org/10.1094/PHYTO-11-16-0419-R        [ Links ]

22. Guarnaccia V, Gehrmann T, Silva-Junior GJ, Fourie PH, Haridas S, Vu D, et al. Phyllosticta citricarpa and sister species of global importance to citrus. Mol Plant Pathol. 2019;20:1619-1635. https://doi.org/10.1111/mpp.12861        [ Links ]

23. Lee YS, Huang CS. Effect of climatic factors on the development and discharge of ascospores of the citrus black spot fungus. J Taiwan Agric Res. 1973;22:135-144.         [ Links ]

24. Kotzé JM. Studies on the black spot disease of citrus caused by Guignardia citricarpa Kiely, with particular reference to its epiphytology and control at Letsitele C [dSc (Agric.) thesis]. Pretoria: University of Pretoria;1963. https://doi.org/10.2903/j.efsa.2009.925        [ Links ]

25. Fourie P Schutte T, Serfontein S, Swart F. Modeling the effect of temperature and wetness on Guignardia pseudothecium maturation and ascospore release in citrus orchards. Phytopathology. 2013;103:281-292. https://doi.org/10.1094/phyto-07-11-0194        [ Links ]

26. Dummel DM, Agostini JP, Ricardo M. Predictive model for ascospore release of Guignardia citricarpa using climatological data. Acta Hortic. 2015;1065:953-963. https://doi.org/10.17660/actahortic.2015.1065.119        [ Links ]

27. Kiely TB. Control and epiphytology of black spot of citrus on the central coast of New South Wales. NSW Dept Agric Sci Bull. 1950;71:4-88.         [ Links ]

28. Baayen RP, Bonants PJM, Verkley G, Carroll GC, Van der Aa HA, De Weerdt M, et al. Nonpathogenic isolates of the citrus black spot fungus, Guignardia citricarpa, identified as a cosmopolitan endophyte of woody plants, G. mangiferae (Phyllosticta capitalensis). Phytopathology. 2002;92:464-477. https://doi.org/10.1094/PHYTO.2002.92.5.464        [ Links ]

29. Guarnaccia V Groenewald JZ, Li H, Glienke C, Carstens E, Hattingh V et al. First report of Phyllosticta citricarpa and description of two new species, P. paracapitalensis and P. paracitricarpa, from citrus in Europe. Stud Mycol. 2017;87:161-185. https://doi.org/10.1016/j.simyco.2017.05.003        [ Links ]

30. McOnie KC. The latent occurrence in citrus and other hosts of a Guignardia easily confused with G. citricarpa, the citrus black spot pathogen. Phytopathology.1964;54:40--43.         [ Links ]

31. Gadoury DM, MacHardy WE. A model to estimate the maturity of ascospores of Venturia inaequalis. Phytopathology. 1982;72:901-904. https://doi.org/10.1094/phyto-72-901        [ Links ]

32. Rossi V Ponti I, Marinelli M, Giosuè S, Bugiani R. Environmental factors influencing the dispersal of Venturia inaequalis ascospores in the orchard air. J Phytopathol. 2001;149:11-19. https://doi.org/10.1046/j.1439-0434.2001.00551.x        [ Links ]

33. Rossi V Salinari F, Pattori E, Giosuè S, Bugiani R. Predicting the dynamics of ascospore maturation of Venturia pirina based on environmental factors. Phytopathology. 2009;99:453-146. https://doi.org/10.1094/phyto-99-4-0453        [ Links ]

34. Stensvand A, Eikemo H, Gadoury DM, Seem RC. Use of a rainfall frequency threshold to adjust a degree-day model of ascospore maturity of Venturia inaequalis. Plant Dis. 2005;89:198-202. https://doi.org/10.1094/pd-89-0198        [ Links ]

35. EFSA PLH Panel (EFSA Panel on Plant Health). Scientific opinion on the risk of Phyllosticta citricarpa (Guignardia citricarpa) for the EU territory with identification and evaluation of risk reduction options. EFSA Journal. 2014;12(2), Art. #3557. http://doi.org/10.2903/j.efsa.2014.3557        [ Links ]

36. Rossi V Ponti I, Marinelli M, Giosuè S, Bugiani R. Field evaluation of some models estimating the seasonal pattern of airborne ascospores of Venturia inaequalis. J Phytopathol. 1999;147:567-575. https://doi.org/10.1046/j.1439-0434.1999.00436.x        [ Links ]

37. Smith JH. A study of the effect of various disease control programs on spore releases of the citrus black spot pathogen Guignardia citricarpa Kiely. Proceedings of the International Society of Citriculture. 1996;1:351-352.         [ Links ]

38. Reis RF, Timmer LW, De Goes A. Effect of temperature, leaf wetness, and rainfall on the production of Guignardia citricarpa ascospores and on black spot severity on sweet orange. Fitopatol Bras. 2006;31:29-34. https://doi.org/10.1590/s0100-41582006000100005        [ Links ]

 

 

Correspondence:
Providence Moyo
Email: pmoyo@cri.co.za

Received: 12 Feb. 2020
Revised: 11 Aug. 2020
Accepted: 17 Aug. 2020
Published: 26 Nov. 2020

 

 

Editors: Teresa Coutinho, Salmina Mokgehle
Funding: Citrus Research International, South African Department of Science and Innovation