• Colletotrichum;
  • Cryptosporiopsis;
  • inoculum source;
  • latent infection;
  • sadie


  1. Top of page
  2. Abstract
  3. Introduction
  4. Material and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

The incidence of sooty blotch/flyspeck (SBFS) and bitter and bull’s eye rots were assessed in a Fuji apple orchard during two seasons. Using a regular sampling design, 252 trees were selected and 20 fruits per tree were sampled at harvest and scored for disease incidence. For bitter and bull’s eye rots, additional assessments were made on all symptomless fruit after a 30-day period of storage. Randomness in the spatial pattern was assessed using beta-binomial analysis of incidence data for three sampling scales (one, three or six adjacent trees as sampling units) and using Spatial Analysis by Distance Indices (sadie) for disease counts for the 3-tree sampling scale. sadie was also used for testing spatial associations between a pair of diseases, between years for the same disease or between rotted and latently infected fruit. Using a toroidal-shifts procedure, 360 maps of disease counts were created based on the observed data, which were further analysed using sadie. Most datasets showed an aggregated spatial pattern, which was more consistent for the two fruit rots than SBFS, which showed distinct patterns depending on the year or method of analysis. The two fruit rots were spatially associated in most situations but SBFS and bull’s eye rot were dissociated in one season. Results from virtual orchards showed that the patterns observed in the original maps may accurately represent those in similar apple-growing areas. Hypotheses regarding aspects of ecology and epidemiology of pathogens studied and potential efficacy of control measures in the region are discussed.


  1. Top of page
  2. Abstract
  3. Introduction
  4. Material and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

In southern Brazil, apple scab and fruit rots (especially bull’s eye rot and bitter rot) are of paramount importance to commercial production, while other diseases, particularly the sooty blotch and flyspeck disease complex (SBFS) can reach epidemic and economically significant levels (Spolti et al., 2011).

Bitter rot is common in several regions of the world (Jones et al., 1996) but is more important in humid, temperate climates where it can lead to yield losses >40% (Crusius et al., 2002; González et al., 2006). In southern Brazil and the southeastern United States, three species are commonly associated with bitter rot (Colletotrichum gloeosporioides, C. acutatum and Glomerella cingulata) (Crusius et al., 2002; González et al., 2006). Inoculum of G. cingulata and C. gloeosporioides survives on shoots, buds, mummified fruit and fallen leaves, and can be dispersed by rain splash (C. gloeosporioides) or wind (G. cingulata) (Crusius et al., 2002).

Bull’s eye rot (Neofabraea perennans (anamorph: Cryptosporiopsis perennans)) was identified in Brazil during the mid-1990s and is distributed throughout the apple growing regions of the country causing losses varying from 16 to 45% under disease-conducive weather conditions (Valdebenito-Sanhueza et al., 2010). In North America, three Neofabrea species have been identified (N. perennans, N. malicorticis and N. alba) (Jong et al., 2001), as well as Cryptosporiopsis kienholzii (Spotts et al., 2009). In Brazil, only the anamorphic phase has been identified (Bogo et al., 2008). Conidia of Cryptosporiopsis spp. can survive in cankers, shoots, buds and fallen fruit (Sharples, 1959; Grove, 1992; Valdebenito-Sanhueza et al., 2010) but cankers are probably the primary inoculum source from which spores are dispersed by rain splash onto fruit. Infections may be initiated any time after petal fall and fruit susceptibility increases with maturation (Sharples, 1959). In Brazil, bull’s eye rot symptoms are detected in the field only when injured fruit become infected (Valdebenito-Sanhueza et al., 2010).

A complex of more than 60 fungal species (Díaz Arias et al., 2010) has been associated with sooty blotch and flyspeck disease complex (SBFS) which blemishes the cuticle of apple fruit (Williamson & Sutton, 2000). Although their colonization is superficial, the economic damage they cause can be substantial because fruit may be downgraded from premium fresh-market grade to processing use. Several species within the SBFS complex appear to be cosmopolitan, such as Schizothyrium pomi, usually associated with the flyspeck-type of symptom on fruit, and Peltaster fructicola, Leptodontium elatius and Geastrumia polystigmatis associated with the punctate, fuliginous, and ramose-type symptoms, respectively (Johnson et al., 1997). Although SBFS has been investigated in North America for more than 90 years (Williamson & Sutton, 2000; Gleason et al., 2011), studies elsewhere in the world are relatively recent. In Brazil, SBFS was reported two decades ago, but its aetiology remains poorly understood (Valdebenito-Sanhueza et al., 2009). Epidemiological studies have been conducted for elucidating aspects of the ecology of S. pomi in Massachusetts (Cooley et al., 2007) and North Carolina (Brown & Sutton, 1993), and P. fructicola in North Carolina (Johnson et al., 1997). These examples provide the most comprehensive information about the epidemiology of SBFS fungi (Williamson & Sutton, 2000).

Analysis of spatial patterns and associations of diseases may help to provide insight and hypotheses on aetiology and epidemiology of diseases and ecology of plant pathogens, potentially leading to improved disease management strategies (Benson et al., 2006). A variety of statistical tools are available to quantify within-field spatial disease patterns, allowing the assessments either below or above the sampling unit (Madden & Hughes, 1995). For example, distribution fitting is commonly used to assess disease incidence heterogeneity for data grouped into sampling units, assess disease pattern at a given scale, as well as improve sampling procedures (Hughes et al., 2002). However, distribution fitting does not take into account the spatial location of the sampling units. On the other hand, spatial autocorrelation allows the analysis of spatial relationships among sampling units with similar disease status and takes the relative position of the sampling units into account (Gottwald et al., 1992). Spatial Analysis by Distance Indices (sadie) can detect clusters in count data independent of their numerical properties (Perry, 1998). sadie has been used to quantify the spatial patterns of fungal and viral diseases (Dallot et al., 2003; Bonnot et al., 2010), and relationships between airborne inoculum and disease intensity during the course of aerially spread epidemics (Carisse et al., 2008).

sadie has been applied to characterize spatial associations between diseases in the same area (Turechek & Madden, 2000). Determining the scale of spatial association (or lack of) is as necessary as determining the nature of an association (positive, negative or neutral) and may provide substantial information on the epidemiology of the species complex (Nelson & Campbell, 1992). Spatial associations of diseases at the field level may result from environmental (e.g. climatic) and/or biological conditions affecting both diseases over large scales (Nelson & Campbell, 1992). Dissociations at the fruit level, for example, could be due to antibiosis or competition of the pathogens for infection sites (Nelson & Campbell, 1992).

A number of studies have evaluated within-field spatial patterns of apple diseases (Johnson et al., 1982; van Leeuwen et al., 2000; Xu et al., 2001; Valdebenito-Sanhueza et al., 2005; Biggs et al., 2008). The objectives of this study were to determine (i) the spatial patterns of incidence, at various sampling scales, and counts of SBFS and bitter and bull’s eye rots; and (ii) the spatial association between pairs of diseases, between years for one disease and between rotted and latently infected fruit.

Material and methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Material and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References


Trials were conducted in a commercial apple orchard (cv. Fuji on rootstock MM-116) located at Vacaria, Rio Grande do Sul State, Brazil (28°29′44·50′′ S 50°50′19·31′′ W). The orchard was established in 1987 with 5 m between rows and 2 m between trees within each row. The number of trees in the orchard at the beginning of the observation period was 1000. The orchard had regular boundaries and a wind-break (Pinus sp.) on the south and east sides. Apple scab (Venturia inaequalis) was controlled using captan (120 g a.i. ha−1) (Orthocide 500, Arysta LifeScience do Brasil) applied prophylactically at 14-day intervals beginning after petal fall until harvest. Sprays (1000 L ha−1) were made using a turbo atomizer pulled by tractor.

Sampling design and data collection

Intensity of naturally occurring epidemics of bitter rot, bull’s eye rot and SBFS in the orchard were assessed 1 week prior to commercial harvest on 23 March 2007 and 17 March 2008. The sampling area was a rectangular subsection within the orchard (around 8960 m2). The smallest sampling unit was composed of a tree in a group of three adjacent trees spaced 2 m apart in the row, and each group was 10 m apart from the other group in the row. Seven groups of three plants were established in each of 12 rows assessed (Fig. 1). At harvest, 20 fruit were arbitrarily taken from both the bottom and top positions of the canopy of each tree. Samples were taken from a total of 252 trees and 5040 fruit were harvested. In each 20-fruit sample, the number of fruit showing signs of SBFS, and symptoms of bitter and bull’s eye rots, were assessed visually at harvest. All symptomless fruit were incubated at 25°C in the dark and re-assessed after 30 days in storage for the presence of rot symptoms (Henríquez et al., 2008). Hence, for bitter and bull’s eye rots, total counts were calculated by adding the counts of the rotted and latently infected fruit.


Figure 1.  Sampling design and three scales (1, 3 or 6 adjacent trees) used in a spatial analysis of datasets of incidence of sooty blotch and flyspeck and bitter and bull’s eye rots of apples assessed in a total sample of 5040 fruits sampled in a cv. Fuji orchard in southern Brazil.

Download figure to PowerPoint

Spatial analyses

Heterogeneity analysis

Disease incidence data (proportion of diseased fruit in the sample) were grouped into three sampling scales (groups) with an individual tree as the smallest sampling unit. An intermediate size sampling scale was composed of three adjacent trees in the row and the largest sampling scale was formed by six trees (3 adjacent trees in a row × 2 adjacent rows) (Fig. 1). For the smallest sampling scale, there were 252 groups; for the intermediate there were 84 groups (seven subgroups of three adjacent trees along each of 12 rows), and for the largest there were 42 groups (seven subgroups of trees along each of six rows) (Fig. 1).

Randomness of disease incidence data within sampling units was assessed via distribution and indices analysis. The beta-binomial and the binomial distributions were fitted to the data using the computer program BBD (Madden & Hughes, 1994) for each assessment. The beta-binomial has two parameters, P, which is the expected probability of disease, and θ, a measure of the variation in disease incidence per sampling unit. While a good fit of the binomial distribution, which has a single parameter (π), is suggestive of a random spatial pattern of disease incidence, the beta-binomial is indicative of an aggregated spatial pattern. Using the BBD, the maximum likelihood estimates and standard errors of P and θ of the beta-binomial distribution were calculated for each year and sampling scale (= 1, 3 and 6 trees) for the disease incidence data as well as the latent infection data. t-tests were used to determine if the maximum likelihood estimates of θ differed significantly from zero (when θ = 0, the beta-binomial is reduced to the binomial distribution) (Madden & Hughes, 1995). The C(α) test (Z-statistic) was used to verify the null hypothesis of overdispersion described specifically by the beta-binomial distribution (Madden & Hughes, 1994). The index of dispersion (D) was also calculated. Values of > 1 suggest aggregation because the observed variance of disease incidence is inflated above the expected binomial variance.

Correlation-based analysis

The spatial arrangement of the disease counts (number of diseased fruit) in the sampling points was evaluated using sadie methodology (Perry, 1995). sadie uses the location of the sampling units and the counts of diseased individuals per unit to evaluate the spatial arrangement of diseased individuals. Similar to a correlation-based analysis, sadie results reflect the spatial arrangement of disease at the sampling unit level and above but depend on the pre-existing level of heterogeneity in the dataset (Madden et al., 2007). In sadie, the distance to regularity, Dr, is the minimum total distance that diseased fruit would need to move to achieve the same number in each quadrat. The degree of nonrandomness within a set of data is quantified by comparing the observed spatial pattern with rearrangements obtained after random permutations of the individuals among the quadrats. Pa, defined as the proportion of randomized samples with distance to regularity as large as or larger than the observed value Dr, can be used for a one-sided test of spatial aggregation (at the significance level of 5%). An overall index of aggregation is given by Ia = Dr/Ea, where Dr is the distance to regularity for the observed data, and Ea is the mean distance to regularity of the randomized samples. Index of aggregation (Ia) > 1 indicates an aggregated pattern. The organization of clusters into patches (neighbourhoods of units with counts larger than the average density m) or gaps (neighbourhoods of units with counts <m) was analysed by mapping clustering indices attributed to each sampling scale (Perry, 1995). The index vi measures the degree to which the unit contributes to a patch, whereas vj is defined similarly but for a gap and takes by convention a negative value. As a general rule, large vi values were those >1·5 and small vj values those <1·5, and they were members of a patch or gap, respectively. A more formal test of degree of clustering was provided by comparing the average values of vi and vj with their corresponding values from randomizations (Perry et al., 1999). Contour maps were drawn from these values provided by sadie. For each disease, crop year and status of fruit rot infection (visible, latent or total rot), sadie analysis was performed using the intermediate sampling scale (1 × 3) dataset with a total of 2028 randomizations for each plot.

Spatial association analysis

Two populations (in this case, two diseases or the same disease assessed at two different times), may be spatially positively associated, disassociated or occur randomly with respect to each other (Perry, 1998). In this study, local spatial association was measured using a sadie association index (Ita), which is based on similarities between the clustering indices of two populations measured at the kth sample unit (Carisse et al., 2008). A coincidence of two patches or two gaps indicates positive values of Ita (association) while a negative association (disassociation) results from a patch coinciding with a gap in both populations. The overall spatial association Pa was calculated as the mean of local values of the two populations (Carisse et al., 2008). The significance of Ita was tested by randomizations, with values reassigned among sample units, after a small-scale autocorrelation in cluster indices from either population. At the 5% level, the statistic < 0·05 indicated significant association while  0·975 indicated significant disassociation. A total of 29 spatial associations were analysed as follows: (i) pair-wise comparisons of the three diseases, (ii) infection with symptoms versus latent infections for each fruit rot, and (iii) first versus second year for each disease dataset.

Toroidal shifts

In order to quantify whether the spatial disease patterns depicted in the orchards could be replicated, virtual samplings were simulated using a toroidal shifts procedure based on the disease counts in each observed epidemic. This was necessary because individual samples were not independent in spatially structured data, hence some relationships could derive from accidental occurrence in neighbouring sample sites. Toroidal shifts is a common procedure for conducting randomization tests in ecological studies, assuming that the spatial process is stationary inside and outside the target area (Fortin & Dale, 2007). It has also been reported in a theoretical simulation-based study on spatiotemporal disease progress (Thébaud et al., 2005). Whereas sadie uses a complete randomization process, toroidal shifts provide restricted permutations, thus assuring that the main spatial structure of the original maps is retained (Fortin & Dale, 2007). In the approach here, a bi-dimensional torus was computationally built by connecting the North to the South parts of the original maps, as well as the West and East margins. The simulated orchards were drawn by randomly sliding a box over the torus as many times as necessary. The procedure was implemented using toroids (F.F. Laranjeira, Embrapa Mandioca e Fruticultura, Brazil, personal communication), a computer program written in Python to support application of randomization tests for maps of plant diseases.

In this study, 30 virtual patterns were constructed based on data from each observed epidemic and each virtual pattern was analysed individually using sadie to determine the spatial pattern based on Ia statistics as described previously. Given the lower contribution of visible rot datasets to total incidence, virtual datasets were generated only for latent infection and total rot incidence. Hence, a total of 300 simulation runs were generated: 60 for SBFS (30 maps × 2 years); 120 for bull’s eye rot (30 maps × 2 years × 2 rot status − latent and total rot) and 120 for bitter rot (30 maps × 2 years × 2 rot status).


  1. Top of page
  2. Abstract
  3. Introduction
  4. Material and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Disease intensity

The mean incidence of fruit diseases was variable across the years. Mean highest incidence of total bitter rot and SBFS were 61·7 and 26·2%, respectively, measured in the 2007 season and these values were about two to three times higher than those measured in the 2008 season (Table 1). Conversely, total bull’s eye rot incidence was slightly higher in the 2008 season (30·7%) than in the previous season (24·8%) (Table 1). For both fruit rots and both years, latent infections were detected and contributed the most (61–80%) to total rot incidence. In general, incidence of latent infections for fruit rots was higher than visible rots at harvest. For example, bull’s eye rot increased from 4·8 to 23·8% when both visible and latent infections were added together in the 2007 season (Table 1).

Table 1.   Beta-binomial and index of dispersion analysis of sooty blotch and flyspeck (SBFS) and bitter and bull’s eye rots assessed in a cv. Fuji apple orchard at three sampling scales (1, 3 or 6 adjacent trees) in two consecutive growing seasons
DatasetDisease incidence (%)Beta-binomial parameter (θ)aDispersion index (D)b
1 × 11 × 32 × 31 × 11 × 32 × 3
  1. aMaximum likelihood estimate of the beta-binomial aggregation parameter θ.

  2. bIndex of dispersion (D) values for the indicated sampling scale by plot and assessment date for SBFS and bitter and bull’s eye rots in apple. Tests for aggregation were performed by comparison of (− 1) × D with the chi-square distribution and with the C(α) test (Z-statistic). This test was highly significant (< 0·001) for all dispersion indices except those marked as n.s., indicating aggregation in most cases.

2007 SBFS61·70·1130·0370·0052·941·70 1·22n.s.
2008 SBFS21·10·1170·0320·0072·841·51 1·12n.s.
2007 Bitter rot
 Visible at harvest 8·60·0590·0390·0261·623·13 3·71
 Latent infection17·60·0910·0570·0412·774·33 5·94
2008 Bitter rot
 Total11·90·0850·0400·0322·493·32 4·75
 Visible at harvest 3·00·0490·0120·0101·791·75 2·17
 Latent infection 8·90·1040·0350·0222·173·01 3·43
2007 Bull’s eye rot
 Total24·80·0950·0760·0454·375·25 6·28
 Visible at harvest 4·80·0380·0280·0201·672·63 3·46
 Latent infection20·00·0650·0370·0412·133·13 5·94
2008 Bull’s eye rot
 Total30·70·0550·0440·0472·753·51 6·50
 Visible at harvest11·70·0620·0250·0152·142·52 2·92
 Latent infection19·00·0740·0290·0152·322·71 3·93

Spatial analyses

Distribution fitting analysis

The beta-binomial distribution provided the best fit to the majority of disease incidence datasets. Aggregation parameter (θ) values varied between 0·032 and 0·145 in all situations for the smallest sampling scale. For the intermediate scale, aggregation parameter values decreased and they were lowest for SBFS data (0·005 and 0·007), indicating lower evidence of aggregation. The index of dispersion (D) varied between 1·51 and 7·97, and 1·12 and 10·34 for the intermediate and largest scale, respectively (Table 1). Evidence of significant spatial aggregation (< 0·001) was observed for datasets of all sampling scales, excepting SBFS incidence for the intermediate scale (> 0·05). For the fruit rots, the highest D values (5·25 and 7·97) were observed for the total incidence of the rots. When total incidence was partitioned into visible or latent infection datasets, D values were consistently higher for the latent infection datasets for both diseases (Table 1).

Correlation-based spatial pattern analysis

Spatial analysis using distance indices showed patterns that were not fully consistent with patterns defined by BBD analysis; this was most evident for bitter rot. For SBFS, both the aggregation statistic (Ia index) and clustering indices (vi and vj) suggested a random spatial pattern of disease incidence in the 2007 season and an aggregated pattern in the following season (Table 2). Spatial associations indices revealed that SBFS counts were dissociated between the two growing seasons (Iat = −0·014; = 0·538) (Table 3).

Table 2. sadie statistics for the spatial analysis of counts of sooty blotch and flyspeck (SBFS) and bitter and bull’s eye rots assessed at an apple orchard in southern Brazil over two consecutive growing seasons
DatasetIndex of aggregationaIndex of clusteringb
  1. aIa is the overall index of aggregation and Pa is the proportion of the 2028 random cases larger than Dr. Significant aggregation is indicated when Pa < 0·05.

  2. bvj and vi correspond to the average values of the indices of clustering vivI (patch) and vJ (gap). P values correspond to the proportion of randomized I or J that exceed the observed values. Significant clustering is indicated when < 0·05.

2007 SBFS1·260·077−1·261·190·0890·141
2008 SBFS2·080·013−1·801·960·0000·000
2007 Bitter rot
 Visible at harvest1·260·090−1·211·180·0000·141
 Latent infection1·230·103−1·121·130·1150·103
2008 Bitter rot
 Visible at harvest1·480·013−1·461·500·1530·012
 Latent infection1·530·013−1·521·610·0120·012
2007 Bull’s eye rot
 Visible at harvest2·030·013−1·951·890·0000·000
 Latent infection1·910·013−1·962·020·0000·000
2008 Bull’s eye rot
 Visible at harvest1·560·013−1·511·570·0120·012
 Latent infection1·480·013−1·39 1·370·0640·051
Table 3.   Spatial association measured by spatial analysis by distance indices (sadie) for pairwise datasets of sooty blotch and flyspeck (SBFS) and bitter and bull’s eye rots assessed at an apple orchard in southern Brazil over two consecutive growing seasonsa
Dataset 1Dataset 2ItaP value
  1. aIndex of association calculated by sadie and associated probability for the association (< 0·05). Data are from a 3-tree sampling unit.

Bitter rot × Bitter rot
 2007 total2008 total−0·0320·603
 2007 visible2007 latent infection0·761<0·0001
 2008 visible2008 latent infection0·4110·0002
 2007 visible2008 visible−0·1680·914
 2007 latent infection2008 latent infection0·0680·317
Bull’s eye rot × Bull’s eye rot
 2007 total2008 total0·2470·012
 2007 visible2007 latent infection0·596<0·0001
 2008 visible2008 latent infection0·432<0·0001
 2007 visible2008 visible0·2550·013
 2007 latent infection2008 latent infection0·1110·158

For bitter rot, distribution of disease counts was found to be random for any incidence measure in the 2007 season. However, the spatial pattern of bitter rot changed to predominantly aggregated in the 2008 season for all incidence variables. Rotted and latently infected fruit at harvest were spatially associated for any situation (fruit rot and year); Iat values were >0·411 (Pa < 0·001) (Table 3). Spatial dissociation of total bitter rot, as well as for one of its components (visible bitter rot), was detected when comparing the growing seasons (Iat ≤ 0·032;  0·603).

Bull’s eye rot disease patterns were aggregated in all datasets, consistent with results obtained from distribution fitting. For both seasons and all incidence variables, Ia values ranged from 1·48 to 2·05 (Pa ≤ 0·013). Significant spatial associations (< 0·0001) were detected between total bull’s eye rot incidence (as well as for visible rot) between years (Ita = 0·247, = 0·012) and between visible rotted and latently infected fruit at harvest in the two seasons (Table 3). Patches and gaps were coincident for the visible and latent infection of bull’s eye rot (Fig. 2).


Figure 2.  Maps of clustering indices estimated using the spatial analysis by distance indices (sadie) for visible and latent infected bull’s eye rot of apple at harvest. Axes show distances in sampling units of the intermediate sampling scale (3-trees) with y-axis in the direction of rows of trees. Areas within dark contours indicate strong clustering as patches, those within white contours strong clustering as gaps.

Download figure to PowerPoint

Bull’s eye and bitter rots were spatially associated in all situations (Table 4). SBFS was not associated spatially with either bitter or bull’s eye rots in all situations. In fact, dissociation of visible rots was detected between total bull’s eye rot and SBFS at harvest in the 2008 season (Iat = −0·286; Iat = −0·342, < 0·05).

Table 4.   Spatial association measured by spatial analysis by distance indices (sadie) for pairwise datasets of sooty blotch and flyspeck (SBFS) and bitter and bull’s eye rots incidence from epidemics in 2007 and 2008 seasons at a cv. Fuji apple orchard with datasets from harvest time and after latent infection detection (rots)a
Dataset 1Dataset 2ItaP value
  1. aIndex of association calculated by sadie and associated probability < 0·05 for the association and > 0·975 for dissociation. Data are from 3-tree sampling unit.

SBFS × Bitter rot
 20072007 total0·0690·303
 20072007 visible0·0670·304
 20072007 latent infection0·0930·238
 20082008 total−0·0630·708
 20082008 visible0·0230·419
 20082008 latent infection−0·1220·841
SBFS × Bull’s eye rot
 20072007 total0·0110·471
 20072007 visible−0·0680·699
 20072007 latent infection0·0370·373
 20082008 total−0·2860·994
 20082008 visible−0·3420·998
 20082008 latent infection−0·1620·932
Bitter rot × Bull’s eye rot
 2007 total2007 total0·412<0·0001
 2007 visible2007 visible0·3010·004
 2007 latent infection2007 latent infection0·3810·003
 2008 total2008 total0·456<0·0001
 2008 visible2008 visible0·2670·009
 2008 latent infections2008 latent infection0·2470·015

Toroidal shift analysis

The proportion of orchards showing aggregated spatial patterns, based on sadie statistics, was higher for SBFS in the 2008 season (66%) compared to the previous season (46·6%). The greater proportion of virtual orchards showing aggregated pattern of SBFS incidence in the 2008 season is in agreement with that defined for the two orchards sampled (Table 1). For bitter rot, a random spatial pattern was predominant in the dataset of virtual orchards for most situations. For total bitter rot as well as its latent infection dataset, the proportion of datasets showing aggregation was lower in the 2007 season than the following season, when around 46% of the virtual orchards showed aggregation for total bitter rot. In general, results for bitter rot for the virtual orchards were consistent with the predominant random pattern defined by sadie analysis of the sampled orchards (Table 2). For bull’s eye rot, 100 and 50% of the virtual orchards showed aggregated patterns of total and latent infections of bull’s eye rot, respectively, in the 2008 season (Table 5).

Table 5.   Proportion of datasets (toroidal shift) (= 30) of sooty blotch and flyspeck (SBFS) and bitter and bull’s eye rots of apple in which the null hypothesis of randomness was rejected (< 0·05) using sadie analysisa
DatasetNumber (%) of orchards showing aggregated pattern of disease counts
2007 season2008 season
  1. asadie (Perry, 1998) is based on the number of moves to regularity of the observed data (Dr) relative to the moves to regularity for randomizations of the observed data. Pa is the proportion of sadie randomizations that are larger than Dr and was used to determine the significance (P value) of a two-sided hypothesis test of no randomness.

SBFS14 (46·6%)20 (66·6%)
Bitter rot
 Total count 3 (10·0%)14 (46·6%)
 Latent infection 7 (23·3%)11 (36·6%)
Bull’s eye rot
 Total count 3 (10·0%)30 (100·0%)
 Latent infection10 (33·3%)15 (50·0%)


  1. Top of page
  2. Abstract
  3. Introduction
  4. Material and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Most datasets of disease incidence or counts for the three apple fruit diseases simultaneously evaluated in this work showed evidence of an aggregated spatial pattern. This pattern was more consistent for bitter and bull’s eye rots than for SBFS, with the exception of the bitter rot counts during the 2007 season, when the null hypothesis of randomness of the fruit counts was accepted by sadieIa. The SBFS spatial pattern was either random or aggregated, depending on the year, and randomness was detected even when SBFS incidence was highest, reaching 61% in the 2008 growing season.

When using distribution fitting analysis it was observed that SBFS incidence changed from random at the largest scale (six trees) to an aggregated pattern with increasing levels of aggregation when the spatial scale decreased from three to one plant. This may be related to a possible secondary spread of SBFS accounting for the aggregated pattern in the smallest (single tree) plots.

An aggregated pattern of all diseases was suggested by distribution fitting, irrespective of the sampling scale for the two fruit rots. Moreover, there was also a tendency for the aggregation parameter and dispersion index to increase with the decrease of scale of sampling. This may be due to the influence of the sample size (n, e.g. number of fruit per sample) in which θ is usually reduced with increasing n values (Madden & Hughes, 1995). Nevertheless, similar results of increasing aggregation for smaller quadrat size (and higher sample size) were reported by Dallot et al. (2003), when studying spatial patterns of sharka disease (Plum pox virus strain M) in peach orchards in France.

In another spatial analysis study, Scott et al. (2003) observed that sadie analysis tended to produce more variable results when determining spatial disease patterns (random or aggregated) in comparison to other correlation-based methods (e.g. median run analysis). When using sadie for SBFS counts, the pattern was random for the 2007 dataset, while SBFS incidence was aggregated when using BBD for the same sampling scale. Also, unlike the BBD results, bitter rot spatial patterns for fruit counts in the 2007 harvest season were random based on the sadieIa. There were four disagreements in 14 direct comparisons of spatial patterns between BBD and sadie, with the latter being less sensitive in detection of aggregation. Since these techniques measure different properties and the patterns are influenced by spatial scale and disease intensity, it is not unusual to obtain different results regarding spatial patterns when using more than one method of analysis (Madden & Hughes, 1995; Dallot et al., 2003; Pethybridge et al., 2010).

Latent infections are very common for the fruit rots evaluated in this study (Prusky, 1996), and the manifestation of the disease during postharvest storage is typical. Latent infection levels were observed that were two to four times higher than the incidence of visible rotted fruit, and hence they were the primary contributors to the total incidence of fruit rots.

The role of latent infections in influencing the pattern of disease progress curves has been discussed (Bergamin Filho & Amorim, 2002). However, the authors are not aware of any literature depicting the spatial patterns of fruit with latent infections, particularly studying its spatial association with rotted fruit from the same sample. Considering that fruit with latent rot infection does not contribute inoculum for new infections in the field, it can be inferred that spatial patterns of primary inoculum in the orchard may explain the typical patterns of infections in the rotted fruit in the field. Incidence levels of rotted fruit were at a maximum of 11·7% during 2008 for bull’s eye rot, which increased up to 30% when including latent infections. The successful use of eradicant chemical treatments during tree dormancy to reduce losses from bull’s eye rot (Valdebenito-Sanhueza et al., 2010) also indicates the importance of the role of primary inoculum in the bull’s eye rot epidemics in the region. Fruit showing bitter and bull’s eye rot symptoms before harvest are a likely source of inoculum for infections that are detected postharvest (Spotts, 1985). These conidia on fruit can be dispersed a short distance by rain splash and thus generate aggregated disease patterns, especially during the initial epidemic phase (Xu & Madden, 2004).

This study is in agreement with a previous analysis of another rot disease affecting Fuji apples in the region, white rot caused by Botryosphaeria dothidea (Valdebenito-Sanhueza et al., 2005). The spatial pattern of the incidence of white rot for both the trees showing fruit infection and the rotted fruit assessed in sampling units of six plants tended to aggregate slightly as suggested by BBD analysis and Taylor’s power law. The epidemiology of white rot in the region may be more similar to bitter rot because the sexual stage of the fungus is present (Valdebenito-Sanhueza et al., 2005). The likely contribution of ascospores to epidemics may lead to less aggregated or random patterns as was shown for bitter rot in this study using sadie analysis.

The strong evidence for the spatial association found between visibly rotted and latently infected fruit at harvest in the same year, for both rots, is indicative of the contribution of both primary and secondary inoculum sources to the epidemics. Scott et al. (2003), analysing spatial associations of oilseed rape mildew between lagged time periods (t vs. t-1) suggested that newly infected sites contributed less to the epidemics than expansion of previously infected sites, probably the result of localized conidial inoculum from secondary sources. The results here suggest that considerable levels of rot incidence at harvest may indicate that there will be even greater levels after harvest. However, no or very low rot incidence at harvest may not indicate that there will be no or low postharvest incidence because latent infection contributed mostly to total incidence.

For bull’s eye rot, there was evidence of spatial association of rotted fruit between years, which suggests that inoculum sources overwinter in survival sites such as cankers within the apple trees themselves, and from where it can be dispersed, especially the conidia. Such association was not detected for bitter rot, and this may be related to other factors influencing epidemics, such as varying environmental conditions or inoculum sources that are relatively distant from the infected trees (Turechek & Madden, 2000). Reductions of bull’s eye rot up to 40% have been observed in Brazil with the application of sulphur-based treatments during the dormancy period (Valdebenito-Sanhueza et al., 2010). Such evidence suggests that a substantial portion of the primary inoculum for bull’s eye rot comes from within orchards, while the failure of dormant sprays to suppress SBFS and bitter rot suggests that primary inoculum for these diseases comes from outside orchards.

When spatial associations between different diseases were measured, there was a significant spatial association between bitter rot and bull’s eye rot. Similarities in means of spore dispersal, overwintering sites, and environmental requirements for the establishment and development of these rots may explain the positive spatial association between these diseases. Both fruit rot diseases have rain splash as their principal means of inoculum dispersal and similar requirements of temperature and wetness for infection (Noe & Starkey, 1982; Henríquez et al., 2008).

Bull’s eye and bitter rots differ substantially in terms of environmental requirements for infection from SBFS. For example, SBFS epiphytic fungi usually require relatively long (>175 h) wetness periods to colonize the epicuticular layer of apple fruit and become visible in the field (Brown & Sutton, 1993). In one of the disease associations tested, SBFS was dissociated from bull’s eye rot, which may be related to combined effects that include differences in inoculum sources and/or varying micro-environmental conditions inside the orchard. Alternatively, Venkatasubbaiah et al. (1995) reported that Peltaster fructicola, one of the fungal species causing SBFS, was capable of producing mycotoxins that inhibited growth of fruit rot pathogens –Colletotrichum gloeosporioides, C. acutatum, B. dothidea and B. obtusa. The lack of dissociation found between SBFS and bitter rot, which may be expected based on the results by Venkatasubbaiah et al. (1995), reinforces the need for a better understanding of the ecology of SBFS fungi in the studied region.

The variation in weather between the seasons cannot be neglected as a factor influencing the spatial patterns and associations found in this study as steps in the disease cycle are influenced differently by a given type of environmental condition (Spotts, 1985; Henríquez et al., 2008). In the first year, wetter conditions were observed during the entire period of fruit formation (November to March) compared to the second year. Accumulated rainfall was 825 and 580 mm in the successive years (data not shown). More disease-conducive weather during extended time periods may have favoured infections of SBFS (Williamson & Sutton, 2000; Spolti et al., 2011) and bitter rot fungi in the first growing season (Noe & Starkey, 1982), thus explaining the higher incidence of both diseases in that year.

Fruit are susceptible to bull’s eye rot pathogen during the entire period of fruit formation (Henríquez et al., 2008), but the maximum susceptibility occurs at the end of the maturation period, approximately 45 days prior to harvest. The 2008 season was characterized by drier conditions during the beginning of the fruit formation period (November to December), but by higher rainfall between January and March (468 mm) as fruit began to mature. This may partly explain the slightly higher incidence of bull’s eye rot in the second year, especially for visible rot at harvest, and not in the first year as noticed for other diseases.

Using the toroidal shifts to generate virtual maps of disease incidence, there were both random and aggregated patterns in the 30-orchard sample for the situations tested for all diseases. For SBFS, the higher proportion of scenarios were random patterns of the counts of fruit with symptoms in 2007 and aggregation in the 2008 season, which agrees with the actual observation. For bitter rot, the higher frequency of random patterns in the virtual orchards did not match with the observed scenario. For total disease counts of bull’s eye rot, random patterns predominated in the 2007 season, but all virtual orchards showed aggregated patterns of total bull’s eye rot incidence in 2008. In general, the sadie analysis for the simulated orchards tended to generate a higher proportion of random spatial patterns. This was expected since toroidal shifts are used to simulate maps of a given ecological variable keeping most, but not all of the spatial structure of data (Fortin & Dale, 2007). These results also show that the patterns observed in the original maps were not a happenstance, but represent an accurate picture of what should occur in other, similar apple-growing areas.

From this study, several hypotheses can be made regarding key aspects of the ecology and epidemiology of the pathogens studied. Firstly, there is little evidence for activity of a sexual stage in bull’s eye rot epidemics because splash dispersed conidial inoculum surviving in cankers on the trees primarily produces aggregated disease patterns (Madden & Hughes, 1995). Secondly, for SBFS, the fact that patches of disease were not similar between the years suggest that inoculum may originate from sources that are both internal and external to the orchard and that micrometeorological conditions may have varied within the orchard across the years. Finally, for bitter rot, random or weakly aggregated patterns may be due to the contribution of both the airborne sexual spores and the rain-splashed conidia to the epidemics. Other factors such as residue management, winter treatments, pruning, wind-breaks, and physical damage on fruit can generate complex scenarios that lead to various and dynamically changing disease patterns.


  1. Top of page
  2. Abstract
  3. Introduction
  4. Material and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

The authors thank Turner Sutton (North Carolina State University) for critical review of a preliminary version of the manuscript. Thanks are also given to the Programa de Pós-graduação em Fitotecnia (UFRGS) and CAPES – Coordenação de Aperfeiçoamento de Pessoal de Nível Superior – for providing a scholarship to PS. RMVS and EMDP are grateful to the Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq) for a research fellowship. This work was partially funded by grants from Inova Maçã project.


  1. Top of page
  2. Abstract
  3. Introduction
  4. Material and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  • Benson DM, Grand LF, Vernia CS, Gottwald TR, 2006. Temporal and spatial epidemiology of Phytophthora root rot in fraser fir plantations. Plant Disease 90, 117180.
  • Bergamin Filho A, Amorim L, 2002. Doenças com período de incubação variável em função da fenologia do hospedeiro. Fitopatologia Brasileira 27, 5615.
  • Biggs AR, Turechek WW, Gottwald TR, 2008. Analysis of fire blight shoot infection epidemics on apple. Plant Disease 92, 134956.
  • Bogo A, Maffioletti MA, Valdebenito-Sanhueza RM, Casa RT, 2008. Morphological characterization of Cryptosporiopsis perennans isolates in different culture media. Tropical Plant Pathology 33, 24851.
  • Bonnot F, de Franqueville H, Lourenço E, 2010. Spatial and spatiotemporal pattern analysis of coconut lethal yellowing in Mozambique. Phytopathology 100, 30012.
  • Brown EM, Sutton TB, 1993. Time of infection of Gloeodes pomigena and Schizothyrium pomi on apple in North Carolina and potential control by an eradicant spray program. Plant Disease 77, 4515.
  • Carisse O, Savary S, Willocquet L, 2008. Spatiotemporal relationships between disease development and airborne inoculum in unmanaged and managed Botrytis leaf blight epidemics. Phytopathology 98, 3844.
  • Cooley DR, Lerner SM, Tuttle AF, 2007. Maturation of thyriothecia of Schizothyrium pomi on the reservoir host Rubus allegheniensis. Plant Disease 91, 13641.
  • Crusius LU, Forcelini CA, Valdebenito-Sanhueza RM, Fernandes JMC, 2002. Epidemiology of apple leaf spot. Fitopatologia Brasileira 27, 6570.
  • Dallot S, Gottwald T, Labonne G, Quiot J-B, 2003. Spatial pattern analysis of Sharka disease (Plum pox virus strain M) in peach orchards of southern France. Phytopathology 93, 154352.
  • Díaz Arias MM, Batzer JC, Harrington TC et al. , 2010. Diversity and biogeography of sooty blotch and flyspeck fungi on apple in the Eastern and Midwestern United States. Phytopathology 100, 34555.
  • Fortin MJ, Dale M, 2007. Spatial Analysis: a Guide for Ecologists. Cambridge, UK: Cambridge University Press.
  • Gleason ML, Batzer JC, Sun G et al. , 2011. A new view of sooty blotch and flyspeck. Plant Disease 95, 36883.
  • González E, Sutton TB, Cortell JC, 2006. Clarification of the etiology of Glomerella leaf spot and bitter rot of apple caused by Colletotrichum spp. based on morphology and genetic, molecular, and pathogenic tests. Phytopathology 96, 98292.
  • Gottwald TR, Richie SM, Campbell CL, 1992. LCOR2 - Spatial correlation analysis software for the personal computer. Plant Disease 76, 2135.
  • Grove GG, 1992. Perennial canker of apple: seasonal host susceptibility, spore production, and perennation of Cryptosporiopsis perennans in infected fruit in eastern Washington. Plant Disease 76, 110914.
  • Henríquez JL, Sugar D, Spotts RA, 2008. Effects of environmental factors and cultural practices on bull’s eye rot of pear. Plant Disease 92, 4214.
  • Hughes G, Gottwald TR, Levy L, 2002. The use of hierarchical sampling in the surveillance program for Plum pox virus incidence in the United States. Plant Disease 86, 25963.
  • Johnson DL, Doust JL, Eaton GW, 1982. The effect of European canker and its spatial pattern on four apple cultivars in British Columbia. Journal of Applied Ecology 19, 6039.
  • Johnson EM, Sutton TB, Hodges CS, 1997. Etiology of apple sooty blotch disease in North Carolina. Phytopathology 87, 8895.
  • Jones AL, Ehret GR, Meyer MP, Shane WW, 1996. Occurrence of bitter rot on apple in Michigan 1996. Plant Disease 80, 12947.
  • Jong SN, Levesque CA, Verkley GJM, Abeln ECA, Rahe JE, Braun PG, 2001. Phylogenetic relationships among Neofabraea species causing tree cankers and bull’s-eye rot of apple based on DNA sequencing of ITS nuclear rDNA, mitochondrial rDNA and the β-tubulin gene. Mycological Research 105, 6589.
  • van Leeuwen GCM, Stein A, Holb I, Jeger MJ, 2000. Yield loss in apple caused by Monilinia fructigena (Aderh. & Ruhl.) Honey, and spatio-temporal dynamics of disease development. European Journal of Plant Pathology 106, 51928.
  • Madden LV, Hughes G, 1994. BBD - Computer software for fitting the beta-binomial distribution to disease incidence data. Plant Disease 78, 53640.
  • Madden LV, Hughes G, 1995. Plant disease incidence: distributions, heterogeneity, and temporal analysis. Annual Review of Phytopathology 33, 52964.
  • Madden LV, Hughes G, van den Bosch F, 2007. Study of Plant Disease Epidemics. St. Paul, MN, USA: American Phytopathological Society.
  • Nelson SC, Campbell CL, 1992. Incidence and patterns of association of pathogens in a leaf spot disease complex on white clover in the Piedmont region of North Carolina. Phytopathology 82, 101321.
  • Noe JP, Starkey TE, 1982. Relationships of apple fruit maturity and inoculum concentration to infections by Glomerella cingulata. Plant Disease 66, 37981.
  • Perry JN, 1995. Spatial analysis by distances indices. Journal of Animal Ecology 64, 30314.
  • Perry JN, 1998. Measures of spatial patterns for counts. Ecology 79, 100817.
  • Perry JN, Winder L, Holland JM, Alston RD, 1999. Red-blue plots for detecting clusters in count data. Ecology Letters 2, 10613.
  • Pethybridge SJ, Hay FS, Gent DH, 2010. Characterization of the spatiotemporal attributes of Sclerotinia flower blight epidemics in a perennial pyrethrum pathosystem. Plant Disease 94, 130513.
  • Prusky D, 1996. Pathogen quiescence in postharvest diseases. Annual Review of Phytopathology 34, 41334.
  • Scott JB, Hay FS, Wilson CR, Cotterill PJ, Fist AJ, 2003. Spatiotemporal analysis of epiphytotics of downy mildew of oilseed poppy in Tasmania, Australia. Phytopathology 93, 7527.
  • Sharples RO, 1959. Further orchard sources of infection by Gloeosporium spp. Plant Pathology 8, 712.
  • Spolti P, Schneider L, Valdebenito-Sanhueza RM, Batzer JC, Gleason ML, Del Ponte EM, 2011. Improving sooty blotch and flyspeck severity estimation on apple fruit with the aid of standard area diagrams. European Journal of Plant Pathology 129, 219.
  • Spotts RA, 1985. Environmental factors affecting conidial survival of five pear decay fungi. Plant Disease 69, 3912.
  • Spotts RA, Seifert KA, Wallis KM et al. , 2009. Description of Cryptosporiopsis kienholzii and species profiles of Neofabraea in major pome fruit growing districts in the Pacific Northwest USA. Mycological Research 113, 130111.
  • Thébaud G, Peyrard N, Dallot S, Calonnec A, Labonne G, 2005. Investigating disease spread between two assessment dates with permutation tests on a lattice. Phytopathology 95, 145361.
  • Turechek WW, Madden LV, 2000. Analysis of the association between the incidence of two spatially aggregated foliar diseases of strawberry. Phytopathology 90, 15770.
  • Valdebenito-Sanhueza RM, Duarte V, Amorim L, Porto MDM, 2005. Detecção e epidemiologia da podridão branca da maçã. Fitopatologia Brasileira 30, 21723.
  • Valdebenito-Sanhueza RM, Nachtigall GR, Kovaleski A, Santos RS, Spolti P, 2009. Manual de Identificação e Controle de Doenças, Pragas e Desequilíbrios Nutricionais da Macieira. Bento Gonçalves, Brazil: Embrapa Uva e Vinho.
  • Valdebenito-Sanhueza RM, Spolti P, Del Ponte EM, 2010. Control of initial inoculum for reducing losses by bull’s eye rot on apples. Revista Brasileira de Fruticultura 32, 104454.
  • Venkatasubbaiah P, Sutton TB, Chilton WS, 1995. The structure and biological properties of secondary metabolites produced by Peltaster fructicola, a fungus associated with apple sooty blotch disease. Plant Disease 79, 115760.
  • Williamson SM, Sutton TB, 2000. Sooty blotch and flyspeck of apple: etiology, biology, and control. Plant Disease 84, 71424.
  • Xu X-M, Madden LV, 2004. Use of sadie statistics to study spatial dynamics of plant disease epidemics. Plant Pathology 53, 3849.
  • Xu X-M, Robinson JD, Berrie AM, Harris DC, 2001. Spatio-temporal dynamics of brown rot (Monilinia fructigena) on apple and pear. Plant Pathology 50, 56978.