Directory UMM :Data Elmu:jurnal:S:Scientia Horticulturae:Vol86.Issue1.Sept2000:
Scientia Horticulturae 86 (2000) 33±47
Evidence for the variation in susceptibility
of bananas to wound anthracnose due to
Colletotrichum musae and the in¯uence
of edaphic conditions
M. Chilleta,*, L. de Lapeyre de Bellairea, M. Dorelb,
J. Joasc, C. Duboisb, J. Marchalb, X. Perrierb
a
Cirad ¯hor, Station de Neufchateau, Sainte Marie, 97130 Capesterre Belle Eau,
Guadeloupe, FWI
b
Cirad ¯hor, BP 5035, 34032 Montpellier cedex 1, France
c
Cirad ¯hor, Station de Moutte, 97200 Fort de France, Martinique, FWI
Accepted 11 February 2000
Abstract
Wound anthracnose is a post-harvest disease which develops during storage and ripening of
bananas. In the French West Indies, it mainly occurs on fruits coming from plantations situated on
soils at low-altitude, during the second half of the year. It is caused by a pathogenic fungus,
Colletotrichum musae. A diagnostic survey was carried out on 106 plots representative of all the
soil/climatic conditions and techniques in Guadeloupe in order to assess the variability of fruit
susceptibility to wound anthracnose. Secondly, the effect of mineral nutrition on this susceptibility
was analysed for the soil/climatic zone where the anthracnose problems are most serious. For this
purpose, 54 plots on halloysitic and ferrallitic soils were chosen by including in the selection plots
from all cultural situations. This study has brought to light a wide variation in the susceptibility of
bananas to Colletotrichum musae. Fruits from high-altitude plantations are the least susceptible. On
low-altitude soils, where the most variability is observed, a relationship was found between the Mn
content of fruit and susceptibility to anthracnose; the plants producing the most susceptible fruit had
high foliar Mn concentrations and low Ca concentrations, and had grown on rather acid soils.
*
Corresponding author. Tel.: 590-86-17-69; fax: 590-86-80-77.
E-mail address: [email protected] (M. Chillet)
0304-4238/00/$ ± see front matter # 2000 Elsevier Science B.V. All rights reserved.
PII: S 0 3 0 4 - 4 2 3 8 ( 0 0 ) 0 0 1 3 8 - 2
34
M. Chillet et al. / Scientia Horticulturae 86 (2000) 33±47
Hypotheses for the physiological mechanisms involved in the sensitisation of the fruit are discussed.
# 2000 Elsevier Science B.V. All rights reserved.
Keywords: Banana; Musa; Wound anthracnose; Colletotrichum musae; Manganese; Calcium
1. Introductionz
Wound anthracnose is a disease that is caused by a fungus, Colletotrichum
musae, which infects the fruit in the ®eld in the ®rst month after ¯owering (de
Lapeyre de Bellaire and Mourichon, 1997). The conidia reach the surface of the
fruit in rainwater trickling over the bunch (de Lapeyre de Bellaire, 1999). They
germinate quickly and form a melanic appressorium which is a dormant structure
of the pathogen. The melanic appressoria only germinate during fruit ripening, to
form an infection hypha which will colonise the peel, and then the pulp of the fruit
(Muirhead and Deverall, 1981; Swinburne and Brown, 1983). When the fruits are
damaged, lesions develop though they are still green, and the lesions develop larger
(Meredith, 1960). It is this form of the disease, called wound anthracnose, which is
the cause of the damage during transport and storage of the bananas and which
seriously detracts from the quality of the product from Guadeloupe (FWI).
This disease is usually controlled by means of treatments applied just before
packing using a fungicide with an antimitotic action, thiabendazole. However,
anthracnose problems appear every year at the same time (fruit exported between
August and January) in the middle of the rainy season (the climate of Guadeloupe
is of the tropical humid type), mainly on plantations situated in low-altitude zones.
Strains of C. musae exist which are resistant to thiabendazole. Moreover, the timing
of the appearance of the disease cannot be explained by resistance to the fungicide.
It has, therefore, been suggested that this seasonal phenomenon is more likely to be
the result of variation in the level of the quality potential of the fruit (Chillet and de
Lapeyre de Bellaire, 1996). This quality potential is de®ned partly by a physiological
component which conditions the susceptibility of fruit and partly by a parasitic
component which appears as the level of contamination of the bananas. Accordingly
to this hypothesis, the quality potential depends on the soil and climatic conditions of
the production zone, and the cultural techniques used by the growers. These factors
have an in¯uence on the physiological component of the potential via the mineral
nutrition and water supply to the banana plants.
To identify the factors which might explain variation in fruit susceptibility to
wound anthracnose, a diagnostic survey was conducted in all of the soil and
climatic conditions of Guadeloupe. Firstly the variation in fruit susceptibility to
C. musae over the whole of the zone studied was observed, and the relationships
between this variability and the mineral nutrition of the bananas for the lowaltitude soils of Basse-Terre, region where the anthracnose incidence is most
serious, was also explored.
M. Chillet et al. / Scientia Horticulturae 86 (2000) 33±47
35
2. Materials and methods
2.1. Soil characteristics of the study zone
2.1.1. Andosols of the windward coast
These soils, developed on a volcanic material, are characterised by the presence
of gibbsite and allophanes. They are situated in high-altitude, high-rainfall areas,
and presented physical properties which are generally favourable to the crop
(large water reserve, good internal drainage and structural stability).
2.1.2. Andosols of the leeward coast
Developed on more recent volcanic material than the corresponding soils on the
windward coast, these soils are less weathered and more fertile. They are also
distinguished by the absence of gibbsite.
2.1.3. Vertisols
These soils, developed on coral limestone, contain large amounts of expanding
clays, smectites, which confer on them particular hydrodynamic properties. Water
in®ltration is notably controlled by swelling and shrinking phenomena, which
causes the appearance and disappearance of large cracks.
2.1.4. Halloysitic and ferrallitic soils
These two soil types contain a large percentage of clay (80%). The dominant
clay mineral is halloysite in both soil types (Van Oort, 1988).
Their structural stability and hydraulic conductivity are high. However, in
certain topographic conditions one can ®nd signs of hydromorphy in agricultural
soils which have been subjected to intense mechanical cultivation.
2.2. Plot sampling
The survey of wound anthracnose incidence took place during the period of
poor banana quality in 1996. One hundred and six plots of banana plants (six on a
vertisol, 54 on halloysitic and ferrallitic soil, 35 on windward coast andosols and
11 on leeward coast andosols) of the Musa acuminata triploid (Grande Naine and
Poyo cultivars) were chosen in the four types of soil described above. For the
second part of the study, the 54 plots on the halloysite and ferrallitic zone soils
were chosen according to the technical level of the farm. The plots were selected
by the level of intensi®cation, notably irrigation, fertilisation, and preceding crop.
2.3. Methodology
On each selected plot, twenty banana plants at the ¯owering stage (all female
hands and 0±1 male hands exposed) were chosen. These plants were as uniform
36
M. Chillet et al. / Scientia Horticulturae 86 (2000) 33±47
as possible as regards to stage of development and bunch shape and were
representative of the plot.
2.3.1. Diagnosis of mineral nutrition
A sampling of the third leaf was collected for mineral analysis (Martin-PreÂvel,
1974). The elements analysed were N, P, K, Ca, Mg, Na, Mn, Fe, Cl, Zn, B, Cu and S.
A soil sample was taken for chemical analysis. This was analysed for total C and
N content, Olsen±Dabin-available P, exchangeable K, Ca, Mg, Na and Mn, analysed
after extraction with cobaltihexamine, pH and CEC (cation exchange capacity).
A median internal fruit of the third hand was sampled at the harvest stage for
mineral analysis of the pulp and peel. The harvest stage of the bunches is reached
when the median external ®nger of the fourth hand reaches the 34 mm grade.
2.3.2. Assessment of the susceptibility of fruit to anthracnose
At ¯owering, two median internal ®ngers of the third hand were inoculated by
placing 25 ml of a suspension of C. musae containing 106 conidia/ml. The strain
of C. musae used for these inoculations is sensitive to thiabendazole.
At the harvest stage, the two inoculated fruits were sampled and wounded by
applying pressure to the inoculation zone. To create a uniform bruise and to take
account of possible grade variations, each fruit was compressed to 15% of its grade
by means of a piston with a rounded end, guided by a texture analyser. One of the two
fruits was treated with thiabendazole by soaking it in a 500 ppm solution for 2 min.
The fruits were stored at 13.58C for 10 days to simulate transport conditions. They
were then placed for 10 days in a room at 218C, where they ripened naturally.
After these 20 days of storage, the length and breadth of the lesions were
measured. The area of the lesion was estimated using the formula for the area of
an ellipse. For each plot, the mean area of the lesions on untreated (SDN0) and
treated (SDNZ) fruits was calculated.
2.4. Statistical analysis
The data were subjected to principal component analysis (PCA) and analysis of
variance using the WINSTATic (1998) and SAS (1996) software.
Two PCAs were done; the ®rst with the soil mineral analysis data (PCA soil)
and the second from the leaf analysis data (PAC leaf).
3. Results
3.1. Evidence for variation in fruit susceptibility
Fig. 1 shows the distribution of fruit susceptibility by soil type in the absence of
fungicidal treatment. These graphs illustrate the wide dispersion of the variable
SDN0; in fact, the areas of the lesions ranged from 156 to 1089 mm2. The fruits
M. Chillet et al. / Scientia Horticulturae 86 (2000) 33±47
37
Fig. 1. Numbers of samples in the classes for the mean area of the lesion on fruit not treated with
thiabendazole (SDN0) for plots sited on a given soil type (&) in relation to the number for the
whole of the 106 plots studied (&). The ®gure above each bar represents the total number of plots
studied in the corresponding SDN0 class.
from the andosols of the leeward coast had smaller lesion areas; they seem to be
less susceptible to wound anthracnose. A more erratic distribution occurs in the
low-altitude soil (halloysitic and ferrallitic), with generally higher lesion areas.
This variability is less for the fruit produced on the andosols of the windward
coast and on the vertisols.
Fig. 2 shows the distribution of susceptibility of fruit treated with thiabendazole
by soil type. These graphs illustrate the wide dispersion of the variable SDNZ; the
lesion areas of treated fruit ranged from 48 to 370 mm2. As in the absence of
fungicide treatment, the fruit produced on the andosols of the leeward coast
seems to have less developed lesions. Similarly, most of the variability is found
on the halloysitic and ferrallitic soils. For the fruit produced on the andosols of
the windward coast and on the vertisols, this variability is less.
38
M. Chillet et al. / Scientia Horticulturae 86 (2000) 33±47
Fig. 2. Numbers of samples in the classes for mean lesion area on fruit treated with thiabendazole
(SDNZ) for plots sited on a given soil type (&) in relation to the number for the whole of the 106
plots studied (&). The ®gure above each bar represents the total number of plots studied in the
corresponding SDNZ class.
3.2. Halloysitic and ferrallitic soils
Fig. 3 presents the lesion sizes on treated fruit (SDNZ) in relation to the lesion
areas on untreated fruit (SDN0) for the halloysitic and ferrallitic soil types. This
graph shows that there is no correlation between the two variables. When the
lesions on untreated fruits are larger than 600 mm2, the lesions on treated fruit
can be very large (more than 300 mm2) or very small (less than 100 mm2).
3.2.1. Effect of mineral nutrition on fruit susceptibility
Fig. 4 shows the correlation pattern for the primary factorial plane of the PCA
of soil variables, which accounts for 60% of the total inertia.
Axis 1 opposes pH, Ca content and CEC to the Mn content. Axis 2 opposes the
C and N contents to those of potassium and phosphorus. The lesion areas, SDN0
and SDNZ, which enter as supplementary variables, tend to follow the alignment
M. Chillet et al. / Scientia Horticulturae 86 (2000) 33±47
39
Fig. 3. Distribution of the variable lesion area for untreated fruit (SDN0) related to that for treated
fruit (SDNZ). The values are in mm2. Each point represents the mean area of the lesion for a plot.
of the variable Mn and to be inversely related to the variables pH and Ca. This
graph seems to indicate a relation between the soil chemical characteristics and
fruit susceptibility; the most susceptible fruits were produced on low-pH soils,
with little Ca and a lot of available Mn.
Fig. 5 shows the correlation pattern for the primary factorial plane of the leaf
PCA, which accounts for 48% of the total inertia.
All the variables are located on the same side of axis 1: most of the correlations
are positive or absent. On axis 2, the Mn and Ca contents are opposed. As in the
case of the soil, the supplementary variables relating to fruit susceptibility, SDN0
and SDNZ, tend to follow the line of the variable Mn and to be inversely related
to Ca. Banana plants low in Ca and with high Mn contents thus produce the most
susceptible fruit and vice versa.
3.2.2. Relationship between Mn content and fruit susceptibility
The PCA suggests the existence of a relationship between the plant Mn and Ca
content and fruit susceptibility.
The leaf Mn contents (L_Mn) were grouped into three classes with equal
numbers of samples. The mean Mn contents of the soil, the peel and the pulp
and the mean lesion areas were calculated for each of these three classes (Tables 1
and 2).
40
M. Chillet et al. / Scientia Horticulturae 86 (2000) 33±47
Fig. 4. Primary factorial plane of the correlation pattern of the principal component analysis (PCA)
of soil variables. The soil variables marked S_xxx are active; the variables SDN0 and SDNZ are
supplementary.
The data in Table 1 show that there is a close relationship between leaf and
green fruit Mn content, both for peel and pulp. There is no evidence for
signi®cant differences between soil Mn contents, but nevertheless there appears
to be a steady trend.
The analysis of the values for lesion area in relation to leaf Mn content shown
in Table 2 con®rms the trends apparent from the PCA. For fruit not treated with
thiabendazole (SDN0), analysis of variance indicates a signi®cant difference
between the third class of L_Mn and the two other classes. The banana plants
with the highest Mn levels have high SDN0 values. For fruit treated with
thiabendazole (SDNZ), the differences are not signi®cant. However, the trend is
still consistent: the highest SDNZ values correspond to the third class of
L_Mn. The banana plants with the highest Mn contents thus produce the most
susceptible fruit.
3.2.3. Relationship between Ca contents and fruit susceptibility
The PCA also suggests the existence of a relation between leaf Ca contents and
lesion area (both for SDN0 and SDNZ).
M. Chillet et al. / Scientia Horticulturae 86 (2000) 33±47
41
Fig. 5. Primary factorial plane of the correlation pattern of the principal component analysis (PCA)
of leaf variables. The soil variables marked L_xxx are active; the variables SDN0 and SDNZ are
supplementary.
The leaf Ca contents (L_Ca) were grouped into three classes of equal sample
numbers. The mean Ca content of the soil, the peel and the pulp and the mean
lesion area were calculated for each of the three classes (Tables 3 and 4).
Table 1
Relation between leaf Mn content and Mn content of soil (S_Mn), pulp (CV_Mn) and peel
(PV_Mn) of green fruit (means within a column followed by the same letter in brackets are not
signi®cantly different according to Newman±Keuls' test of multiple comparison of means at the 5%
probability level)
Class limits
for L_Mn (ppm)
Class
mean
S Mna
(meq/100 g)
CV Mnb
(in ppm)
PV Mnc
(in ppm)
Mn(1): [194; 795]
Mn(2): ]795; 1340]
Mn(3): ]1340; 2508]
Prob>F
576
1010
1623
0.213 (a)
0.252 (a)
0.293 (a)
0.541
24.1 (a)
55.4 (b)
83.2 (c)
F
0.48
0.60
0.71
3.10 (b)
4.79 (b)
7.67 (a)
Evidence for the variation in susceptibility
of bananas to wound anthracnose due to
Colletotrichum musae and the in¯uence
of edaphic conditions
M. Chilleta,*, L. de Lapeyre de Bellairea, M. Dorelb,
J. Joasc, C. Duboisb, J. Marchalb, X. Perrierb
a
Cirad ¯hor, Station de Neufchateau, Sainte Marie, 97130 Capesterre Belle Eau,
Guadeloupe, FWI
b
Cirad ¯hor, BP 5035, 34032 Montpellier cedex 1, France
c
Cirad ¯hor, Station de Moutte, 97200 Fort de France, Martinique, FWI
Accepted 11 February 2000
Abstract
Wound anthracnose is a post-harvest disease which develops during storage and ripening of
bananas. In the French West Indies, it mainly occurs on fruits coming from plantations situated on
soils at low-altitude, during the second half of the year. It is caused by a pathogenic fungus,
Colletotrichum musae. A diagnostic survey was carried out on 106 plots representative of all the
soil/climatic conditions and techniques in Guadeloupe in order to assess the variability of fruit
susceptibility to wound anthracnose. Secondly, the effect of mineral nutrition on this susceptibility
was analysed for the soil/climatic zone where the anthracnose problems are most serious. For this
purpose, 54 plots on halloysitic and ferrallitic soils were chosen by including in the selection plots
from all cultural situations. This study has brought to light a wide variation in the susceptibility of
bananas to Colletotrichum musae. Fruits from high-altitude plantations are the least susceptible. On
low-altitude soils, where the most variability is observed, a relationship was found between the Mn
content of fruit and susceptibility to anthracnose; the plants producing the most susceptible fruit had
high foliar Mn concentrations and low Ca concentrations, and had grown on rather acid soils.
*
Corresponding author. Tel.: 590-86-17-69; fax: 590-86-80-77.
E-mail address: [email protected] (M. Chillet)
0304-4238/00/$ ± see front matter # 2000 Elsevier Science B.V. All rights reserved.
PII: S 0 3 0 4 - 4 2 3 8 ( 0 0 ) 0 0 1 3 8 - 2
34
M. Chillet et al. / Scientia Horticulturae 86 (2000) 33±47
Hypotheses for the physiological mechanisms involved in the sensitisation of the fruit are discussed.
# 2000 Elsevier Science B.V. All rights reserved.
Keywords: Banana; Musa; Wound anthracnose; Colletotrichum musae; Manganese; Calcium
1. Introductionz
Wound anthracnose is a disease that is caused by a fungus, Colletotrichum
musae, which infects the fruit in the ®eld in the ®rst month after ¯owering (de
Lapeyre de Bellaire and Mourichon, 1997). The conidia reach the surface of the
fruit in rainwater trickling over the bunch (de Lapeyre de Bellaire, 1999). They
germinate quickly and form a melanic appressorium which is a dormant structure
of the pathogen. The melanic appressoria only germinate during fruit ripening, to
form an infection hypha which will colonise the peel, and then the pulp of the fruit
(Muirhead and Deverall, 1981; Swinburne and Brown, 1983). When the fruits are
damaged, lesions develop though they are still green, and the lesions develop larger
(Meredith, 1960). It is this form of the disease, called wound anthracnose, which is
the cause of the damage during transport and storage of the bananas and which
seriously detracts from the quality of the product from Guadeloupe (FWI).
This disease is usually controlled by means of treatments applied just before
packing using a fungicide with an antimitotic action, thiabendazole. However,
anthracnose problems appear every year at the same time (fruit exported between
August and January) in the middle of the rainy season (the climate of Guadeloupe
is of the tropical humid type), mainly on plantations situated in low-altitude zones.
Strains of C. musae exist which are resistant to thiabendazole. Moreover, the timing
of the appearance of the disease cannot be explained by resistance to the fungicide.
It has, therefore, been suggested that this seasonal phenomenon is more likely to be
the result of variation in the level of the quality potential of the fruit (Chillet and de
Lapeyre de Bellaire, 1996). This quality potential is de®ned partly by a physiological
component which conditions the susceptibility of fruit and partly by a parasitic
component which appears as the level of contamination of the bananas. Accordingly
to this hypothesis, the quality potential depends on the soil and climatic conditions of
the production zone, and the cultural techniques used by the growers. These factors
have an in¯uence on the physiological component of the potential via the mineral
nutrition and water supply to the banana plants.
To identify the factors which might explain variation in fruit susceptibility to
wound anthracnose, a diagnostic survey was conducted in all of the soil and
climatic conditions of Guadeloupe. Firstly the variation in fruit susceptibility to
C. musae over the whole of the zone studied was observed, and the relationships
between this variability and the mineral nutrition of the bananas for the lowaltitude soils of Basse-Terre, region where the anthracnose incidence is most
serious, was also explored.
M. Chillet et al. / Scientia Horticulturae 86 (2000) 33±47
35
2. Materials and methods
2.1. Soil characteristics of the study zone
2.1.1. Andosols of the windward coast
These soils, developed on a volcanic material, are characterised by the presence
of gibbsite and allophanes. They are situated in high-altitude, high-rainfall areas,
and presented physical properties which are generally favourable to the crop
(large water reserve, good internal drainage and structural stability).
2.1.2. Andosols of the leeward coast
Developed on more recent volcanic material than the corresponding soils on the
windward coast, these soils are less weathered and more fertile. They are also
distinguished by the absence of gibbsite.
2.1.3. Vertisols
These soils, developed on coral limestone, contain large amounts of expanding
clays, smectites, which confer on them particular hydrodynamic properties. Water
in®ltration is notably controlled by swelling and shrinking phenomena, which
causes the appearance and disappearance of large cracks.
2.1.4. Halloysitic and ferrallitic soils
These two soil types contain a large percentage of clay (80%). The dominant
clay mineral is halloysite in both soil types (Van Oort, 1988).
Their structural stability and hydraulic conductivity are high. However, in
certain topographic conditions one can ®nd signs of hydromorphy in agricultural
soils which have been subjected to intense mechanical cultivation.
2.2. Plot sampling
The survey of wound anthracnose incidence took place during the period of
poor banana quality in 1996. One hundred and six plots of banana plants (six on a
vertisol, 54 on halloysitic and ferrallitic soil, 35 on windward coast andosols and
11 on leeward coast andosols) of the Musa acuminata triploid (Grande Naine and
Poyo cultivars) were chosen in the four types of soil described above. For the
second part of the study, the 54 plots on the halloysite and ferrallitic zone soils
were chosen according to the technical level of the farm. The plots were selected
by the level of intensi®cation, notably irrigation, fertilisation, and preceding crop.
2.3. Methodology
On each selected plot, twenty banana plants at the ¯owering stage (all female
hands and 0±1 male hands exposed) were chosen. These plants were as uniform
36
M. Chillet et al. / Scientia Horticulturae 86 (2000) 33±47
as possible as regards to stage of development and bunch shape and were
representative of the plot.
2.3.1. Diagnosis of mineral nutrition
A sampling of the third leaf was collected for mineral analysis (Martin-PreÂvel,
1974). The elements analysed were N, P, K, Ca, Mg, Na, Mn, Fe, Cl, Zn, B, Cu and S.
A soil sample was taken for chemical analysis. This was analysed for total C and
N content, Olsen±Dabin-available P, exchangeable K, Ca, Mg, Na and Mn, analysed
after extraction with cobaltihexamine, pH and CEC (cation exchange capacity).
A median internal fruit of the third hand was sampled at the harvest stage for
mineral analysis of the pulp and peel. The harvest stage of the bunches is reached
when the median external ®nger of the fourth hand reaches the 34 mm grade.
2.3.2. Assessment of the susceptibility of fruit to anthracnose
At ¯owering, two median internal ®ngers of the third hand were inoculated by
placing 25 ml of a suspension of C. musae containing 106 conidia/ml. The strain
of C. musae used for these inoculations is sensitive to thiabendazole.
At the harvest stage, the two inoculated fruits were sampled and wounded by
applying pressure to the inoculation zone. To create a uniform bruise and to take
account of possible grade variations, each fruit was compressed to 15% of its grade
by means of a piston with a rounded end, guided by a texture analyser. One of the two
fruits was treated with thiabendazole by soaking it in a 500 ppm solution for 2 min.
The fruits were stored at 13.58C for 10 days to simulate transport conditions. They
were then placed for 10 days in a room at 218C, where they ripened naturally.
After these 20 days of storage, the length and breadth of the lesions were
measured. The area of the lesion was estimated using the formula for the area of
an ellipse. For each plot, the mean area of the lesions on untreated (SDN0) and
treated (SDNZ) fruits was calculated.
2.4. Statistical analysis
The data were subjected to principal component analysis (PCA) and analysis of
variance using the WINSTATic (1998) and SAS (1996) software.
Two PCAs were done; the ®rst with the soil mineral analysis data (PCA soil)
and the second from the leaf analysis data (PAC leaf).
3. Results
3.1. Evidence for variation in fruit susceptibility
Fig. 1 shows the distribution of fruit susceptibility by soil type in the absence of
fungicidal treatment. These graphs illustrate the wide dispersion of the variable
SDN0; in fact, the areas of the lesions ranged from 156 to 1089 mm2. The fruits
M. Chillet et al. / Scientia Horticulturae 86 (2000) 33±47
37
Fig. 1. Numbers of samples in the classes for the mean area of the lesion on fruit not treated with
thiabendazole (SDN0) for plots sited on a given soil type (&) in relation to the number for the
whole of the 106 plots studied (&). The ®gure above each bar represents the total number of plots
studied in the corresponding SDN0 class.
from the andosols of the leeward coast had smaller lesion areas; they seem to be
less susceptible to wound anthracnose. A more erratic distribution occurs in the
low-altitude soil (halloysitic and ferrallitic), with generally higher lesion areas.
This variability is less for the fruit produced on the andosols of the windward
coast and on the vertisols.
Fig. 2 shows the distribution of susceptibility of fruit treated with thiabendazole
by soil type. These graphs illustrate the wide dispersion of the variable SDNZ; the
lesion areas of treated fruit ranged from 48 to 370 mm2. As in the absence of
fungicide treatment, the fruit produced on the andosols of the leeward coast
seems to have less developed lesions. Similarly, most of the variability is found
on the halloysitic and ferrallitic soils. For the fruit produced on the andosols of
the windward coast and on the vertisols, this variability is less.
38
M. Chillet et al. / Scientia Horticulturae 86 (2000) 33±47
Fig. 2. Numbers of samples in the classes for mean lesion area on fruit treated with thiabendazole
(SDNZ) for plots sited on a given soil type (&) in relation to the number for the whole of the 106
plots studied (&). The ®gure above each bar represents the total number of plots studied in the
corresponding SDNZ class.
3.2. Halloysitic and ferrallitic soils
Fig. 3 presents the lesion sizes on treated fruit (SDNZ) in relation to the lesion
areas on untreated fruit (SDN0) for the halloysitic and ferrallitic soil types. This
graph shows that there is no correlation between the two variables. When the
lesions on untreated fruits are larger than 600 mm2, the lesions on treated fruit
can be very large (more than 300 mm2) or very small (less than 100 mm2).
3.2.1. Effect of mineral nutrition on fruit susceptibility
Fig. 4 shows the correlation pattern for the primary factorial plane of the PCA
of soil variables, which accounts for 60% of the total inertia.
Axis 1 opposes pH, Ca content and CEC to the Mn content. Axis 2 opposes the
C and N contents to those of potassium and phosphorus. The lesion areas, SDN0
and SDNZ, which enter as supplementary variables, tend to follow the alignment
M. Chillet et al. / Scientia Horticulturae 86 (2000) 33±47
39
Fig. 3. Distribution of the variable lesion area for untreated fruit (SDN0) related to that for treated
fruit (SDNZ). The values are in mm2. Each point represents the mean area of the lesion for a plot.
of the variable Mn and to be inversely related to the variables pH and Ca. This
graph seems to indicate a relation between the soil chemical characteristics and
fruit susceptibility; the most susceptible fruits were produced on low-pH soils,
with little Ca and a lot of available Mn.
Fig. 5 shows the correlation pattern for the primary factorial plane of the leaf
PCA, which accounts for 48% of the total inertia.
All the variables are located on the same side of axis 1: most of the correlations
are positive or absent. On axis 2, the Mn and Ca contents are opposed. As in the
case of the soil, the supplementary variables relating to fruit susceptibility, SDN0
and SDNZ, tend to follow the line of the variable Mn and to be inversely related
to Ca. Banana plants low in Ca and with high Mn contents thus produce the most
susceptible fruit and vice versa.
3.2.2. Relationship between Mn content and fruit susceptibility
The PCA suggests the existence of a relationship between the plant Mn and Ca
content and fruit susceptibility.
The leaf Mn contents (L_Mn) were grouped into three classes with equal
numbers of samples. The mean Mn contents of the soil, the peel and the pulp
and the mean lesion areas were calculated for each of these three classes (Tables 1
and 2).
40
M. Chillet et al. / Scientia Horticulturae 86 (2000) 33±47
Fig. 4. Primary factorial plane of the correlation pattern of the principal component analysis (PCA)
of soil variables. The soil variables marked S_xxx are active; the variables SDN0 and SDNZ are
supplementary.
The data in Table 1 show that there is a close relationship between leaf and
green fruit Mn content, both for peel and pulp. There is no evidence for
signi®cant differences between soil Mn contents, but nevertheless there appears
to be a steady trend.
The analysis of the values for lesion area in relation to leaf Mn content shown
in Table 2 con®rms the trends apparent from the PCA. For fruit not treated with
thiabendazole (SDN0), analysis of variance indicates a signi®cant difference
between the third class of L_Mn and the two other classes. The banana plants
with the highest Mn levels have high SDN0 values. For fruit treated with
thiabendazole (SDNZ), the differences are not signi®cant. However, the trend is
still consistent: the highest SDNZ values correspond to the third class of
L_Mn. The banana plants with the highest Mn contents thus produce the most
susceptible fruit.
3.2.3. Relationship between Ca contents and fruit susceptibility
The PCA also suggests the existence of a relation between leaf Ca contents and
lesion area (both for SDN0 and SDNZ).
M. Chillet et al. / Scientia Horticulturae 86 (2000) 33±47
41
Fig. 5. Primary factorial plane of the correlation pattern of the principal component analysis (PCA)
of leaf variables. The soil variables marked L_xxx are active; the variables SDN0 and SDNZ are
supplementary.
The leaf Ca contents (L_Ca) were grouped into three classes of equal sample
numbers. The mean Ca content of the soil, the peel and the pulp and the mean
lesion area were calculated for each of the three classes (Tables 3 and 4).
Table 1
Relation between leaf Mn content and Mn content of soil (S_Mn), pulp (CV_Mn) and peel
(PV_Mn) of green fruit (means within a column followed by the same letter in brackets are not
signi®cantly different according to Newman±Keuls' test of multiple comparison of means at the 5%
probability level)
Class limits
for L_Mn (ppm)
Class
mean
S Mna
(meq/100 g)
CV Mnb
(in ppm)
PV Mnc
(in ppm)
Mn(1): [194; 795]
Mn(2): ]795; 1340]
Mn(3): ]1340; 2508]
Prob>F
576
1010
1623
0.213 (a)
0.252 (a)
0.293 (a)
0.541
24.1 (a)
55.4 (b)
83.2 (c)
F
0.48
0.60
0.71
3.10 (b)
4.79 (b)
7.67 (a)