Programas hermanos de El Salvador y Panamá.

41 Data analysis

A mixed model approach was employed to describe temporal changes of incidence of B. cinerea and S. sclerotiorum over the flowering period using SAS Version 9.1 (SAS Institute, Cary, NC, USA). The incidence of fungi in surface sterilised flowers sampled on five occasions during the season was fitted to exponential, Gompertz, linear, logistic and monomolecular models to evaluate which population growth model most accurately described changes in the incidences of B. cinerea and S. sclerotiorum (Nutter, 1997). The goodness of fit of each models linear forms were compared using the root mean square error, coefficient of determination (R2), the F-statistic for linearity, visual comparison plotting predicted against residual values, and by comparing back transformed fitted disease values against original incidence data (Nutter, 1997). The mean incidence in nontreated plots at all five sampling times for B. cinerea and S. sclerotiorum were also compared against each other, using a paired sample Student’s t-test using GenStat Version 10.1 (VSN International Ltd.).

Survey of other mycoflora associated with pyrethrum flowers

These flower samples were collected as part of further studies described in chapter 2, with all flower sampling, surface sterilisation, incubation and microscopic assessment methodology used here described in chapter 2 materials and methods (Chapter 2.2 p55-57). Twenty five pyrethrum fields in total (Table 2.2.2) were surveyed throughout the flowering period over consecutive years for the presence or absence of fungi displaying sclerotial morphology consistent with S. minor on incubated flower samples. Flowers were sampled from non fungicide treated plots in fields during the flowering period as described previously (Fig. 1.2.1), and were sampled three times from ten fields during 2008-09 and twice from 15 fields in 2009-10 (Table 2.2.2, app. 2.2.2). Flower samples collected during 2008-09 were additionally surveyed for all identifiable mycoflora associated with flowers other than B. cinerea, S. sclerotiorum and fungi suspected of being S. minor. Flower samples were surface sterilised and incubated as described previously in materials and methods for this chapter.

Collection of fungal isolates for further study

During assessment of pathogen incidence in flowers, 73 isolates of S. sclerotiorum (Appendix 1.3.2 A) and 49 isolates of B. cinerea (Appendix 1.3.2 B) were collected from flower samples for further study. Each isolate was obtained from separate incubation boxes each containing 5 flowers (Plate 1.2.1 A-B), with all flowers sampled from nontreated plots in commercial fields.

42

Sclerotia of S. sclerotiorum were picked off flowers with sterile implements and surface sterilised in a solution of 2% sodium hypochlorite (Bleach, White King TM) in distilled water for five minutes before being rinsed with two consecutive washes of distilled water to remove excess bleach. After drying thoroughly overnight on filter paper in a laminar flower cabinet, sclerotia were bisected and plated onto petri plates containing PDA and incubated in the dark (20°C). Hyphal tips were removed from colonies after two days and placed onto fresh PDA plates. The resultant cultures were then incubated (20°C) for further production of mycelium and sclerotia. Cork plugs of mycelium 5 mm in diameter were cut from the actively growing outer edge of each isolate, dried overnight on sterile petri plates before being stored in cryogenic vials at -80°C under glycerol. Approximately two dozen dried mycelial plugs were stored under glycerol, with an equal number also stored as dry plugs in cryogenic vials at -80°C. Once mature, all sclerotia from one PDA plate per isolate were collected with aseptic implements onto sterile filter paper and air dried in a laminar flower cabinet overnight. Sclerotia were then transferred to autoclaved, air dried 5 ml glass jars and stored at 4°C.

Individual conidia of B. cinerea from masses of conidiophores present on incubating flowers (Plate 1.2.1 C) were removed with a sterile scalpel, plated onto PDA and incubated at 20° C until mycelial growth covered most of plate surface. Cork plugs 5 mm in diameter were then removed from the actively growing mycelial front of each isolate and dried overnight in a laminar flow cabinet, before storage at -80°C in sterile 5 ml cryogenic vials.

43 1.3 Results

1.3.1 Disease intensity and modelling

The incidence of B. cinerea throughout the flowering period in nontreated plots at field trial sites was best fitted to the Gompertz and logistic models at two and six fields respectively, indicative of a polycyclic disease epidemic (Table 1.3.1 A). The measured floral incidence of S. sclerotiorum over time was best explained by the exponential, logistic and Gompertz models at three, one, and three fields respectively (Table 1.3.1 A). Residual values were examined during regression analysis to determine the model of best fit for incidence changes of B. cinerea and S. sclerotiorum at each field trial site.

.

Initial flower infection and time of incidence in flowers reaching 50% was estimated from rising disease intensity and model analysis. Incidence of B. cinerea and S. sclerotiorum could not be fitted to models at 2 and 3 fields respectively due to decreasing disease incidence approaching the end of the flowering period. From sites where models could be fitted, initial infection by B. cinerea was estimated to have occurred between 11-16 November, while infection by S. sclerotiorum was estimated between 15 November and 1 December. Incidence of B. cinerea from flowers sampled between 13-16 November was ≤1% at 3 fields, while between 29 November and 4 December was ≥1% at 8 of 10fields. Across field sites where models were fitting, B. cinerea was estimated to reach 50% disease incidence on 11 December, with this time ranging between 30 November and 18 December among fields. Most fields had ≤1% incidence of S. sclerotiorum between 15-29 November. S. sclerotiorum was estimated to reach 50% disease incidence between 10-28 December, with a mean date of 18 December across fields. Changing incidences of both fungi in addition to flower maturity (FMS) over the flowering period for each site are shown (Fig. 1.3.1 A-J).

44 Pathogen Site Model of P Slope

Y

R2 Back RMSE Coef. of

50%