POPULATION DYNAMICS OF (~OLLETOTRICHUM GLOEOSPORIOIDES ON DIFFERENT PARTS OF AN AVOCADO ( PERSEA AMERICANA) TREE

Abstmct: Colletolrichuh~. gloeosporioides in difyerent parts of a single avocado tree, and its dehl-is under the tree, was enumerated at monthly inte~lrals, during a period of two years. A positive cor~.elation between quiescent infections-QI on flowers and rainfall (RF), and a negative correlation of QI on twigs and RF were observed. The positive correlation hetween RF and QI on leaves was signiiicant only when RF was pre-lagged hy 3 months, and the level of significance increased progressively when RF was pre-lagged upto 6 months. Disease symptoms were never ohsellred on the vegetative parts ofthe tree and flowers, but a high numher of QI was observed on twigs and leaves. This is probably how C. gloeosporioides survives hetween fiviting seasons. Tllc different 1.elationships between QI on each part of the tree and RF, demonstrate how this pathogen may have adopted itself to survive on the tree. Thcreforc thc results suggest that onw the C. gloeosporioicles pl-opagules find their way to a tree, the pathogen remains there throughout its life time causing anthracnose on fruits. There was a significant drop in the density of QI of C. glomsl~orioides on larger, mature f i i t s compwed to that of smaller, immatwe fruits. Tluoughout this study, the free-living viable fungal propagule counts of dehl-is under the tree val-ied from 10' to 103 CW/g wllile those of washings of leaves, twigs and flowers varied fieom 0 to 10" CFUIg.


INTRODUCTION
The mechanism of infection of avocado (Persea anzericana Miller) by Colletotrichum. gloeosporioides (Penz.), the causal agent of anthracnose disease1;< and the biochemical basis of resistance of unripe avocado fruit to the path~gen".~have been well documented. This disease is one of the major reasons for the reduced shelf-life of avocado fruits in many parts of the w~r l d~.~ and in Sri Lanka6J.
As far back as 1971, it was demonstrated that decay spots could appear on softening fruits even if they have been inoculated with C. gloeosporioides as early as 3 months prior to harvest1. Seven years later it was observed that infections with C. gloeosporioides could occur on fruits from fruit set to harvestx. More recently it was shown, that the fungus was able to remain quiescent on the avocado fruit for periods of a t least 6 months!'.
In the literature there are no records of the presence of quiescent infections (QI) of C. gloeosporioides on the vegetative parts of the avocado tree. However in mango, it is reported that C. gloeosporioides survived actively in diseased leaves, defoliated branch terminals and old inflorescence^'^.
The present study was carried out on a single avocado tree. This was only an initial attempt to identify the intrinsic and extrinsic parameters to focus on, in studying the means of survival of C. gloeosporioides in between fruiting seasons, and in understanding its infection pattern on the developing avocado fruit.

METHODS & MATERIALS
The avocado tree used for santpling : A Persea antericana Miller tree of Carribean origin, which is about 15 years old, approximately 20ft tall and healthy, growing in a home garden in Kandy with no special attention given, was studied. The other trees within a 10 m radius were a mango tree and a jak tree. The avocado tree was well exposed to the sun and the wind. Several other avocado trees were present within a 100 m radius, and the tree sampled was the healthiest and robust of them all.

E7zumerating ir~oculum of C. gloeosporioides on avocado leaves, twigs, flowers
arzd debris under the tree: Samples of leaves, pieces of young green twigs, flowers (during seasons) from the tree, and debris under the tree were collected a t monthly intervals, and analysed for C. gloeosporioides counts immediately. This was done for a period of two years from 21 July 1991 and ending on 26 July 1993, where a total of 25 sets of counts for leaves and twigs, and 15 counts for flowers were taken. Samples of branches about l m in length from a t least five randomly selected locations around the lower region of t h e trunk were cut by standing under the tree using a knife fixed to a long pole. Samples of debris under the tree were collected into a clean plastic bag a t t h e same time the branches were collected. This consisted of a mixture of dry leaves, twigs and flowers. The samples were taken to the laboratory and the branches were kept fresh by immersing the cut end in water.

Estimation of &I:
Approximately 2x2 mm-ieces of green tender peelings of twigs to a depth of approximately 1 mm, leaves or fruit peel of about 1 mm thick were cut and surface sterilized by immediately dipping in 1% NaOCl solution for 2 minl1. Individual flowers were split in two before subjecting to surface sterilization. All surface sterilized tissues were plated on Cook's No. 2 agar medium12 (containing 200 ppm tetracycline). All culture plates made in these experiments were incubated a t room temperature (27+3"C) for 48 to 72h and the emergence of C. gloeosporioides colonies from the plated tissue was taken as indicative of quiescent infections. Every month, one hundred tissue pieces from each part of the avocado tree (twigs, leaves, flowers when available) were taken randomly (a total of 300 pieces a month, when flowers were available, or 200 pieces when the tree was not in flower), so that the pieces represented all the cut branches. The plating of all the samples was completed within 3 days of collection.
Estinzation offi-ee liviug viable fungal propagule FLWP) counts of C. gloeosporioides: Standard dilution plate technique was used. Randomly obtained pieces (approx. 2 x 2 cm9 of leaves, and twigs (15 cm length of approx. 3mm diameter), which made upto log fresh weight separately, representing all the branches cut as described above, and debris (log of a mixture of decaying twigs and leaves, of sizes described above) under the tree were separately placed in 90ml of sterile distilled water in a 250 ml capacity Erlenmeyer flask. The flasks were shaken vigorously for approximately 2 min after placing a previously sterilized rubber lid, and 1 ml from each was transferred to test tubes containing 9 ml sterile distilled water to prepare a series of 10 fold dilutions. One millilitre portions from the dilutions and the original suspension were pipetted out into separate petri dishes to which melted Cook's No. 2 agar medium containing 200 ppm tetracycline, was poured.
Plating was completed on the day the samples were obtained. The number of typical C. gloeosporioides colonies that grew on the agar plates (of no dilution to 10'~) after incubating at room temperature for 48-72 h, were counted and the values were expressed as the number of Colony Forming Units (CFU) per gram fresh weight of each part of the avocado tree plated.
At least one representative C. glocosporioidcs isolate from each part of the tree (twigs, leaves and flowers) and debris in both QI estimation stud.y, and in the FLVFP estimation study, were stored on agar slants a t 5" C until pathogenicity was established as described below.
Esttrnation of' &I of' C. gloeosporioidcs on fru,it.s at different maturity stages: Fruits at G different maturity stages (distinguished by their size) from very young upto those in harvesting maturity were sampled to determine the quiescent C. gloeosporioides infections on them. Soon after plucking from the tree, the average volumes were determined by immersing the entire fruit in water and noting the amount of water displaced by each fruit. The surface area of each fruit was estimated as the two-thirds power of the fruit volumel1. Each surface area value thus obtained was rounded off to the nearest integer. Pieces (2 x 2 m111) of peel tissues were obtained for surface sterilization followed by plating in such a manner that five pieces per unit surface area of each fruit were considered as the number of pieces to sample from each f r~~i t .
Therefore the mean number of pieces obtained from each fruit was: 5 pieces from stage 1, 9 pieces from stage 2, 12 pieces from stage 3, 56 pieces from stage 4, 95 pieces from stage 5, 125 pieces from stage 6 . At the first three stages of maturity, 5 fruits each were sampled, whereas at the last three stages of maturity 2 fruits each were sampled. The first three stages represent fruits pluclced within six weeks of anthesis and the last three represent fruits at stages 6 weeks prior to harvest.
Representative peel samples were obtained covering all sides of a fruit. These were surface sterilized and plated as described earlier. The plates were incubated a t room temperature for 72 h and the number of C. gloeosporioides colonies developed were counted, and this value was divided by 5 to obtain the number of quiescent infections per unit surface area (i.e. density) of the fruit. In this way an average density value was obtained for the total number of replicate fruits a t a particular maturity stage. The C. gloeosporioides isolates obtained were stored on agar slants a t 5°C and their pathogenicity on fruits was established as described below.
Deternrirr.ir~g pathogenicity of' isolated C. gloeosporioides culturc,s : C . gloeosporioides isolates obtained were subcultured on freshly prepared Cook's No.2 agar plates and allowed to incubate for 72h a t room temperature until the cultures were fully grown. The surface of healthy, mature but unripe avocado fruits were wiped with 70% ethyl alcohol. Suspensions of conidia (approx. 10" per ml) were prepared with each isolate in 2% sucrose solutions. Drops ( 20~1) of these suspensions were placed on 5 spots on fruits in one row along the longitudinal axis. A second parallel row of drops was placed on 5 spots which were wounded by piercing the skin to a depth of 2mm using a sterile needle. The inoculated fruits were kept inside moist chambers for 24 h and after t h a t they were allowed to ripen in trays in the laboratory a t room temperature until typical anthracnose lesions developed. The pathogen was reisolated into agar plates and the cultural characteristics were compared with the original isolates.
Rainfall (RF) was measured daily with the aid of a rain gauge (Yaganli International Trading, Japan), during the two year period of analysis. The monthly RF was determined by totalling the daily values from the day after the previous month's sample was taken, to the day the present sample was'taken. The RF values before the study started (i.e. in July 91) were determined by adding up the daily values for 30 day intervals ( a total of 6) prior to the first sampling.
Statistiwl anal,y.sis :Afive month moving average was used for the time series data of monthly RF and percentages of conidial counts of each part of the tree. The relationship between the independent variable (monthly RF) and each of the dependent variables (i.e. QI of percentage leaf, twig and flower) were -A@m ~1i,'1~& -zr3;ng ~f , C~L w f b ll;es, '~-~'thga, mF~J~str'U *dl -dieiz = ~~h m b e r 05 months) for RF. Regression analysis was performed for each of the dependant varjables with RF a t different lags and the most influential lags were determined.

RESULTS
Flowering was observed from January to June of 1992 and from December 1992 to July 1993 with a freak flowering period in between in October 1992 ( Figure 1). The estimated QIs fluctuated during this period of study and ranged from 3 to 93% on leaves, 13 to 86% on twigs, 1 to 9570 on flowers. The total FLVFP counts on the tree were low but consistent throughout the two year period except in July and September 1991, when the counts on the tree as a whole, were zero (Table 1).
Of the FLVFP counts taken, the highest were seen in washings of debris under the tree which varied from 0 to 10J CFUIg. The FLVFP counts of washings of leaves, twigs and flowers varied from 0 to lo2 CFU/g. Unlike QIs, there were 6 occasions on leaves, 5 occasions on twigs and 6 occasions on flowers where there were no FLVFP of C. gloensporioides detected. There was no instance where FLVFP counts on the tree and debris were zero in the same month ( Table 1). The C. gloeosporioides count in debris was zero only on one occasion, which corresponded to a sudden increase of RF in October 1991.
The graphical plots of monthly RF and QI on leaves, twigs, and flowers are given in Figure 1. No significant relationships were observed between the monthly RF values and each of the dependent variables. Therefore, 5 month moving averages were taken to study the trends. When moving averages of each of the dependent variables were compared with t h a t of RF values (using different lags for leaf infections) the relationships listed in Table 2 were observed. There was a significant positive correlation between RF (with no lag) and QI of flowers. The regression coefficient between RF (pre-lagged 3 months, up to 6 months) and QI on leaves was also positive. However, QI of twigs correlated negatively with RF. These statistically significant relationships between RF and each dependent variable are graphically represented in Figure 2. The two flowering seasons including the freak flowering observed in October 1992, a r e depicted continuously along wit11 . the corresponding RF values, disregarding the periods where flowering was absent, as the 5 month moving averages were determined in this manner.  The results of the regression analysis performed between the dependent variables and RF are diagramatically depicted in Figure 3. The positive linear correlations between RF and QI of flowers, RF (pre-lagged t-3) and QI of leaves and the negative linear correlation between RF and QI of twigs are clearly depicted in these figures. QI counts of C. gloeosporioides on peels of fruits a t different stages of maturity showed that the average conidial infections per unit area reduced sharply when the fruits enlarged with age ( Figure 4). ADuncan multiple range test (pe0.05) revealed that the readings taken from the 3 immature stages of fruits were not significantly different from each other, and the readings of t h e fruits close to harvesting were not significantly different from each other. However these two sets of three readings each, were significantly different from each other.
Irrespective of the part of the tree from which the isolates was obtained, all C. gLoeosportoides isolates tested caused typical anthracnose lesions on wound inoculated fruits. None of the inoculations made on unwounded skin of fruits developed into lesions.

DISCUSSION
The samples for enumerating QI and FLVFP of C. gloeosporioides, of leaves, flowers and twigs, were obtained only from the lower branches to eliminate variation caused by sampling from different heights. It is likely that the sunlight falling a t different intensities to branches a t different heights, among other environmental factors will affect the pathogen counts on branches a t different levels of the tree.
The influence of RF 011 QI in three distinct parts of the tree (i.e. leaves, twigs and flowers) appear to follow three different patterns, making suggestion of control measures (i.e. chemical sprays) complicated. An earlier study on avocado has reported that occurrence of quiescent infections of C. gloeosporioides on fruits reached a maximum during the wettest months, but decreased in drier months, and that the data best fitted a non-linear regression model". Although we did not correlate RF with fruit infections, the statistical relationships of QI observed on other parts of the tree were always linear.
The positive correlation of QI on flowers and leaves with RF pre-lagged (t-3), and the negative correlations seen with t h a t of twigs and RF, cannot be explained a t this stage. The presence of a thick cuticle on avocado leaves may perhaps be a lrey to explanation of regression coefficient becoming progressively more significant when the RF is lagged from t-3 to t-6 ( Table 2).
Initiation of flowering was observed in January and October respectively in the two years ( Figure I), which are dry months. Flowers have remained on the tree for G months in 1992, and 8 months in 1993. The QIs on flowers have increased with the rains. The QI of leaves also increased corresponding to RF of 3 months before (Figure 2). Fruit development occurs spread over a period of approximately 4 rnonths during and after the flowering season (unpublished observations ofA. M. Karunaratne ). Propagules of C. gloeosporioides either a s QI (Figure 1) or FLVFP (Table 1) appear to persist on the tree throughout the year with fluctuating numbers. The C. gloeosporioides inoculum appears to be nurtured by different parts ofthe tree, each part hosting the propagules through time in a different pattern, although the vegetative parts and flowers are not debilitated. If the pathogen caused disease in these parts the whole tree would get lrilled eventually, t h u s preventing a safe niche for the pathogen to survive in between fruiting seasons. This is different to the survival of C. gloeosporioides causing arithracnose of' mango, where the conidia were shown to be produced on leaf lesions'" throughout the year particularly duringflowering. In mango, anthracnose is a devastating field disease in addition to being a postharvest disease.
It is possible tshat the antifungal compounds reported to be present on leaves1" prevent them getting diseased by the pathogen. The role of antifungal compounds on different parts of the tree sustaining the propagules of the pathogen needs a detailed investigation. We have tried inoculating avocado leaves and twigs on cut branches with C. glocosporioides in the laboratory, by keeping the cut-end of the branches fresh, by immersing in water. Corlidial suspensions were prepared and drops (20111) were placed on the upper surface of intact and wounded leaves and l~orizontal tender twigs, by the method described above for fruits. They were enclosed inside plastic bags containing wet cotton wool to maintain moisture. None produced lesions although the same conidial suspensions were able to produce lesions on wounded spots of fruits a s described above (A. M. Karunaratne, unpublished data).
The FLVFP counts observed on the tree was low. The highest counts observed on t h e tree were 10"FUIg.
A slightly higher value of 10" CFUIg was observed in the debris. Due to this narrow variation and low counts in the FLVFP, they were not considered in interpreting data.
The low counts of FLVFP of C. gloeosporioides seen in this study may also indicate that the pathogen was unable to cause disease and increase propagule numbers while on the vegetative parts of the tree. Unlike in the present study, in an earlier study on citrus, washings of flower petals have yielded a high number of propagules (i.e. lo4 to 1O"er gram fresh weight) of C. gloeosporioides14, and this pathogen is known.to cause wither tip in citrus, in addition to anthracnose on fruits.
In avocado it is possible that the debris under the tree, acts as the ultimate source of inoculum. The survival of the conidia in the debris under the tree may also explain the spread of the disease to fruits of younger newly grown trees by rain splash from neighbouring older trees. An earlier study on strawberry anthracnose15, reported that C. acrLtatum survived in buried strawberry tissue for 9 months, and soil population densities gradually declined over a 11 month period.
As occurrence of anthesis appeared to spread for over a period of four months, we expected that relating the &I on fruits with their age, rather than with RF would give more meaningful data. Since the fruits for analysis were obtained from the same tree, fruit size was taken as a convenient index of maturity in this study although this would not be appropriate in a larger sample where genetical and environmental diversity would have an effect on fruit size. Quiescent infections were higher on fruits soon after anthesis. The levels of antifungal compounds are know11 to increase gradually and reach a maximum a t harvest7. However, there is no evidence that antifungal compounds affect the establishment of quiescent infections .
The significant reduction of the density of quiescent infections when the fruit matures (Figure 4), may be explained by any one or a combination of the following hypotheses: a majority of infections are able to establish a t an early stage of fruit maturity because of the peel anatomy; the rate of infection cannot keep up with the log phase of expansion of the fruit to maturity.

Several authors have observed t h a t penetration of C. gloeosporioides is inhibited by actively growing host
A histopathological study of C. gloeosporioides infections to avocado, reported that in fruit inoculated early in development when the epicuticular wax layer was very thin, many appressoria were seen later, resting on top of a thick epicuticular wax layerY. In addition, these authors have observed thin infection pegs which appeared to be physically broken in the region of the wax layer directly beneath these appressoria. It is not recorded how many of these detached infection pegs actually were able to cause disease on the ripe fruit. I t was observed in the present study and in several other studies, (A. M. Karunaratne, unpublished data), that when intact healthy skin of harvested mature unripe fruits were inoculated with viable C. gloeosporioides conidja, lesions did not form in these areas when the fruits ripened. I t is possible that the thick wax layer on unripe mature avocados acts as a deterrent to establishment of quiescent infections.

Anjani M. Karz~nnrat~w et al.
When fruits a t a later stage of development were inoculated with C . gloeosporioides infection pegs penetrated the wax layer and once through the wax layer the infection peg became thin and unmelanized2. While it is apparent that there is a difference in infection patterns when C. gloeosporioides inoculations occur a t early stages of fruit development and a t later stages, whether both types of infections result in lesion formation is not known. In an earlier study1, fruits covered early after fruit set showed no infections or lesions a t ripening, caused by C. gloeosporioides. However fruits inoculated with C . gloeosporioides a t different stages of development (as much as 3 months prior to harvest), developed anthracnose lesions a t ripening. These authors do not report how many lesions that formed on the fruit could be identified with the inoculations made by them earlier.
Several other environmental factors and physiological condition of the host will govern conidial counts and infections. An earlier study has shown that disease development on two weeds by C. gloeosporioides f. sp. malvae was affected by inoculum concentration and plant growth stage a t the time of inoculation, post inoculation air temperature and, dew-period duration and temperature1".
A study which would establish the presence of antifungal conlpounds on twigs and flowers and the influence of these compounds in different parts of the tree on the pathogen will help in understanding the pathogenicity of C. gloeosporioides on avocado. Eventually, control measures could be recommended, based on these findings.