Short communication
Population patterns of
Aspergillus ¯avus
group and
A. niger
group in ®eld
soils
G.J. Grif®n
a,*, E.P. Smith
b, T.J. Robinson
baDepartment of Plant Pathology, Physiology and Weed Science, Virginia Polytechnic Institute and State University, Blacksburg, VA 24061, USA bDepartment of Statistics, Virginia Polytechnic Institute and State University, Blacksburg, VA 24061, USA
Received 9 June 1999; received in revised form 25 January 2000; accepted 20 June 2000
Abstract
The horizontal distribution of soil-borne fungi in ®eld soils is an important component of fungal ecology.Aspergillus ¯avusgroup and its antagonist, theA. nigergroup, had moderate and moderate to high degrees of aggregation (patchiness), respectively, in two peanut ®elds with a history of a¯atoxin contamination: Lloyd's index of patchiness values ranged from 2.32 to 6.55, where 1.0 is a random pattern. For theA. nigergroup, there was evidence that quadrats in close proximity had similar population densities. Clump size, rarely reported for soil-borne fungi, was found to be 13±52 m2forA. nigergroup in one ®eld. Mean population densities ofA. ¯avusgroup in both ®elds were higher than found previously in most Virginia peanut ®elds and, along with moderate aggregation, may be related, in part, to the occurrence of a¯atoxin in these ®elds.q2001 Elsevier Science Ltd. All rights reserved.
Keywords: Population patterns;Aspergillus ¯avus;A. niger
Information on population densities and horizontal patterns of soil-borne fungi is important to management of pathogens. While information on the degree of aggregation of soil-borne fungi has been presented in recent years (Campbell and Bensen, 1994; Grif®n and Baker, 1991), there is little or no information of the size and location of patches of aggregated (clumped) or high-density inoculum in ®eld soils (Gilligan, 1995; Mihail and Alcorn, 1987).
Aspergillus ¯avusgroup fungi (A. ¯avusLink andA. para-siticus Speare) have been responsible for production of
carcinogenic a¯atoxins in peanuts (Arachis hypogeae L.)
in soils worldwide. High levels ofA. ¯avusgroup
coloniza-tion of peanut fruits in soil and a¯atoxin produccoloniza-tion are favored by hot, dry conditions (Diener et al., 1987; Cotty
et al., 1994). BothA. ¯avusgroup fungi (primarilyA. ¯avus)
and A. niger group fungi may be isolated from peanut kernels, and both may colonize peanut fruits in ®eld soils. These fungi have been considered antagonists of each other by some workers (Ashworth et al., 1965; Joffe, 1969; Diener et al., 1987), and little or no information exists on the rela-tive spatial patterns of pathogens and their antagonists in ®eld soils (Gilligan, 1995). The present investigation was
undertaken to determine the horizontal pattern ofA. ¯avus
andA. nigergroups in two neighboring peanut ®elds with a
recent history ofA. ¯avusgroup colonization and a¯atoxin
contamination of peanuts in soil.
Field plots were established in two commercial ®elds, about 4 ha in size, in Virginia, USA, which contained peanuts that tested positive for a¯atoxin contamination in the previous crop. Peanuts and corn have been grown in these ®elds for several decades. The ®elds were on the same farm and were located about 300 m apart. The ®elds were sampled in June after peanuts had been planted in the ®rst week of May. Weed and disease control were good. A 25:4£25:4 m
2
lattice plot was established in each ®eld and
was divided into 49 blocks, each 3:6£3:6 m
2
:A soil-core
sample (2.5 cm diameter by 7.6 cm deep) was collected at the center of each block. These centric-systematic samples (Milne, 1959), or quadrat samples, were placed in plastic bags with pinholes for gas exchange and transported to the
laboratory. Soils were placed at 4±58C and assayed. Soil
samples were mixed by hand for 5 min previous to dilu-tion-plate assay on M3S2B medium, selective for low
popu-lation densities of A. ¯avus group and A. niger group, as
used previously (Grif®n et al., 1975, 1981). Undesired bacteria and fungi are completely or severely inhibited in
this isolation procedure, which uses Botran (2mg of
2,6-dichloro-4-nitroanaline per ml), NaCl (3%), high
tempera-ture (358C), and antibacterial antibiotics to achieve selective
inhibition. This medium was able to recover at 1021 soil
dilution very low population densities from soil [1±10
Soil Biology & Biochemistry 33 (2001) 253±257
0038-0717/01/$ - see front matterq2001 Elsevier Science Ltd. All rights reserved.
PII: S 0 0 3 8 - 0 7 1 7 ( 0 0 ) 0 0 1 3 3 - 4
www.elsevier.com/locate/soilbio
* Corresponding author. Tel:11-540-231-5049; fax:11-540-231-3221.
G.J.
Grif
®n
et
al.
/
Soil
Biology
&
Biochemi
stry
33
(2001)
253
±
257
Table 1
Advantages, disadvantages, and results obtained for ®ve statistical estimates, tests, or plots onAspergillus ¯avusgroup andA. nigergroup population densities in two peanut ®elds with a history of a¯atoxin contamination
Statistical estimate, test, or plot
Advantages Disadvantages Results
A. ¯avusgroup A. nigergroup
Field A Field B Field A Field B
Mean population density Provides information on whether high or low overall population densities are present in a ®eld soil
Does not provide information on aggregation, randomness, or location of fungal population densities in a ®eld soil
33.9 cfuaper g
soil
29.2 cfuaper g
soil
318.8 cfuaper g
soil
301.6 cfuaper g
soil
Lloyd's index of patchiness (or similar Morrisita's index of dispersion) (Pielou, 1977; Morrisita, 1959)
Indicates if propagules exist as aggregates or have a random pattern; 1.0random pattern and.1.0 indicates increased aggregation
Does not indicate the location or size (area) of aggregates (clumps, patches), if present in a ®eld soil
2.96 2.32 6.55 3.19
Moran'sIstatistic of autocorrelation (Upton and Fingleton, 1985)
Indicates if similar population densities are in close proximity in a plot of mapped data
Choice of scale used in sampling may markedly affect results
0.173 (P0.0336)b*
z1.823
0.117 (P0.0986)b*
z1.295
0.517
(P,0.0001)b*
z5.046
0.373
(P,0.0011)b*
z3.70 Plot of Morrisita's index of
dispersion versus number of samples in a cluster (Vandermeer, 1981)
Provides information on clump size (area) in a plot of mapped population density data
Does not provide information for speci®c population densities
13±26 M2 13±26 M2 13±26 M2 13±52 M2
Krishner Iyer test (Pielou, 1977)
Indicates if high population densities are randomly distributed in a plot of mapped data
Does not provide information for speci®c population densities
(NRJ)c (P.0.05)*
(NRJ)c (P.0.05)*
(RJ)c(P,0.05)* (RJ)c(P,0.05)*
a cfucolony-forming units. b Values for ranked data.P
,0:0001 andP,0:0011 indicate strong evidence,P0:0336 indicates marginal evidence andP0:0986 indicates very marginal evidence for samples in close proximity
having similar values.
colony-forming units (cfu) per g soil] ofA. ¯avusgroup in extensive evaluation tests (Grif®n and Garren, 1974; Grif®n
et al., 1975). Eight dilution plates at 1021or 1022dilution
were incubated at 358C for 3 days after which colony counts
ofA. ¯avusgroup andA. nigergroup (primarilyA. nigervan Tieghem) were made.
Several approaches have been used to study the pattern of soil-borne fungi in ®eld soils (Campbell and Bensen, 1994). Following soil assays, Lloyd's index of patchiness (Pielou, 1977) was calculated in the present study to give an estimate of the degree of aggregation of propagules in soil (Table 1).
Moran'sIstatistic (Cliff and Ord, 1973; Upton and
Fingle-ton, 1985) was used to evaluate whether similar population densities occurred in quadrats (samples) located in close
proximity in a lattice plot. Moran's I for ranked data,
found useful here, is given by:
I 12
where Xi is the rank of population density quadrat i, n
denotes the number of locations;
S04c c21;
where c is the number of columns, and Wij the weight
assigned to the join between quadratsiandjusing the rook's
de®nition (quadrats which have touching sides) of contigu-ity (Upton and Fingleton, 1985). The analyses for ranked
and unranked (population density) data are similar; as
indicated by Upton and Fingleton (1985), the equation for
Moran'sIusing ranked data is a simpli®cation of the
equa-tion for Moran'sI using unranked data. In this study, the
conversion of the population density data to ranked data was straightforward (the lowest population density has the lowest rank): thus, the quadrat with the lowest population
density value in the lattice plot for A. ¯avus group or A.
nigergroup was given the rank of 1; the quadrat with the second lowest population density value was given the rank of 2, and the quadrat with the third lowest population density value was given the rank of 3, etc. To estimate aggregate size, Morisita's index of dispersion or patchiness index (Table 1) was calculated (Morrisita, 1959; Vanderm-eer, 1981) for clusters of one to ®ve quadrats, taken down crop rows or across crop rows. Quadrat clusters of increas-ing size were thus obtained. For each data set, Morisita's index of dispersion was plotted against the number of quad-rats in a cluster. A rapid drop in Morisita's index was taken to indicate a fungal clump size that corresponds approxi-mately to the quadrat cluster size at which the rapid drop occurred (Vandermeer, 1981). The relative advantages and disadvantages of these tests for determining population pattern are indicated in Table 1. Other workers have also used Lloyd's index of patchiness (Hampson and Coombes, 1996; Grif®n and Baker, 1991), Morrisita's index of disper-sion (Mihail and Alcorn, 1987; Campbell and Bensen, 1994)
and Moran's I (Mihail and Alcorn, 1987; Campbell and
Bensen, 1994) to study the spatial patterns of soil-borne fungi.
Box plots of the population densities of A. ¯avusgroup
andA. nigergroup in the 49 quadrats of each ®eld were done in order to visualize the pattern and association of quadrat data. Using an approach suggested by Pielou (1977), that incorporated the Krishna Iyer (1949) test, population density data were classi®ed as high-valued and low-valued for each fungal group based on the median population density in each ®eld. The resulting maps for each ®eld were marked to show all horizontal, vertical and diagonal joins (contacts) between all high-valued quadrats, and eval-uated by the Krishna Iyer test for random distribution of high-value quadrats.
Aggregated population patterns were found for both fungal groups in both ®elds (Table 1). Lloyd's index of
patchiness values for A. ¯avus group in ®elds A and B
were 2.96 and 2.32. Lloyd's index of patchiness values for
A. niger group in ®elds A and B were 6.55 and 3.19. A Lloyd's index of patchiness value of 1.0 represents a random pattern (Pielou, 1977), with increasing values indi-cating increased degrees of aggregation.
For rankedA. nigergroup population densities in ®eld A,
there was strong evidence for similar population densities occurring in quadrats (samples) of close proximity (Table
1). ForA. nigergroup in ®eld B, for ranked data (Table 1),
and for unranked population density data [Moran's I
0:135; with a z-statistic of 1.87 P0:0307], there was
strong and marginal evidence, respectively, for quadrats in
close proximity having similar values. ForA. ¯avusgroup
ranked data in ®eld A, there was marginal evidence for quadrats in close proximity having similar values (Table
1). For A. ¯avus group ranked data in ®eld B, there was
very marginal evidence for quadrats in close proximity
having similar values (Table 1). For A. ¯avus group in
both ®elds and A. niger group in ®eld A, using unranked
data, no evidence was found for clustering of quadrats with similar values. As indicated (Table 1), choice of scale may markedly affect the values obtained for neighboring
popula-tion densities and the results obtained for Moran'sI. Ranked
data may be less sensitive to scale effects. Ranked data were examined in these tests because of the large differences in population density among the neighboring higher
popula-tion density quadrats. ForA. nigergroup in ®elds A and B,
plots of Morisita's index versus numbers of quadrats in a cluster, down the crop rows, indicated a cluster size of one
to two quadrats (13±26 m2) and one to four quadrats (13±
52 m2), respectively. Across the crop rows, a cluster size of
one to two quadrats (13±26 m2) was indicated in both ®elds.
The same was found forA. ¯avusgroup in both ®elds, either
down or across the crop rows.
Mean population densities forA. ¯avusgroup in ®elds A
and B were 33.9 and 29.2 colony-forming units (cfu) per g soil, respectively. Population densities ranged from 0 (one quadrat only) to 283.8 cfu per g soil for ®eld A and from 2.5 to 216.8 cfu per g soil for ®eld B. Mean population densities
for A. niger group in ®elds A and B were 318.8 and 301.6 cfu per g soil, respectively. Population densities ranged from 24.9 to 4788 cfu per g soil for ®eld A and from 35.5 to 2656 cfu per g soil for ®eld B. Repeat analysis of random soil samples gave similar results.
Box plots of the lattices in ®elds A and B for A. ¯avus
group andA. nigergroup population densities are shown in
Fig. 1. Population densities ofA. ¯avusgroup andA. niger
group are shown using a gray scale covering the range of population densities found in the ®eld. Using the approach recommended by Pielou (1977), the null hypothesis of
random mingling of high-population density quadrats was
not rejected P.0:05in either ®eld A or B for theA. ¯avus
group by the Krishna Iyer (1949) test (Table 1). However, the null hypothesis of random mingling of high-population
density quadrats ofA. nigergroup was rejected P,0:05
for both ®elds. All high-density quadrats ofA. nigergroup
in ®eld A were clustered into one large patch, measuring
316 m2, where the patch is the sum of the areas of all joined
high population density quadrats. Similarly, all high-popu-lation-density quadrats, except one, in ®eld B were
aggre-gated into one large patch measuring 303 m2.
G.J. Grif®n et al. / Soil Biology & Biochemistry 33 (2001) 253±257
Based on Lloyd's index of patchiness values, the results
indicate thatA. ¯avusgroup had a moderate level of
aggre-gation. Many soil-borne fungi examined previously had Lloyd's index of patchiness values of less than 2, indicative of low aggregation (Grif®n and Baker, 1991). Values greater
than 5, as found here forA. nigergroup in ®eld A, have not
been commonly reported. The spatial scale we used falls about mid-way in the table (for grain or quadrat size) given by Campbell and Bensen (1994) for 44 previous studies. Orum et al. (1997) recently pointed out the need
to determine the spatial pattern ofA. ¯avusin adjacent ®elds
in order to understand the factors that determine the spatial structure within and between ®elds. In the present study, the two ®elds were separated by about 300 m of woodland. Both ®elds had similar, moderately aggregated spatial patterns
and mean population densities ofA. ¯avusgroup.
The results obtained for Moran'sIstatistic, based on ranked
data, provided evidence that similar population densities ofA.
nigergroup tended to occur in close proximity in both ®elds. Similar ®ndings were found for unranked population density
data from ®eld B. The clump size range forA. nigergroup, 13±
52 m2,was found along the rows. Since this clump size was not
found across the rows, cultivation of the ®eld may have contributed to the movement of crop debris or fungal propa-gules in the ®eld and the orientation of clumps. Rarely has clump size in a ®eld been estimated for a soil-borne fungus (Mihail and Alcorn, 1987). The results of the Krishna Iyer test
indicated that the quadrats ofA. nigergroup, having greater
than the median population density, were not randomly distributed in each ®eld. Patch size estimates of 316 and
303 m2were relatively large.A. nigergroup is recognized as
a common competitor ofA. ¯avus group (Ashworth et al.,
1965; Joffe, 1969; Diener et al., 1987), and, like A. ¯avus
group, is favored by higher temperatures. However,A. niger
group may not be as competitive asA. ¯avusgroup at very low
water potentials (Diener et al., 1987). Conversely, somewhat moister areas or sites may favor the development of
high-population densities of A. nigerin both ®elds. Low, moist
areas were not apparent in either ®eld, but slight differences in topography may be dif®cult to detect. The texture of all soil samples appeared similar. Generally, areas in the ®elds with highA. niger group population densities did not have high
population densities of A ¯avus group. The high A. niger
group population densities found here may be a result of moist periods that occurred during crop debris decomposition.
Overall, the population densities found forA. ¯avusgroup
were higher than found previously in most Virginia peanut ®eld soils (Grif®n and Garren 1974; Grif®n et al., 1975; 1981), in which a¯atoxin occurrence is uncommon. These
higher population densities of A. ¯avus group, along with
the moderate aggregation found for this group, may be related, in part, to the occurrence of a¯atoxin in these two ®elds. Moderate aggregation could lead to patches in the ®eld with high population densities (greater than 200 cfu per g soil). For management considerations, the mixing effects of plowing and disking the ®eld may help break-up
aggregates and distributeA. ¯avusgroup propagules over a
greater volume of soil, possibly diluting these higher p-opulation densities.
Acknowledgements
We thank J. Ashley for help in locating the study ®elds.
References
Ashworth, L.T., Schroeder, H.W., Langley, B.C., 1965. A¯atoxins: envir-onmental factors governing occurrence in Spanish peanuts. Science 148, 1228±1229.
Campbell, C.L., Bensen, D.M., 1994. Spatial aspects of the development of root disease epidemics. In: Campbell, C.L., Bensen, D.M. (Eds.). Epidemiology and Management of Root Disease Epidemics, Springer, Berlin, pp. 195±243.
Cliff, A.D., Ord, J.K., 1973. Spatial Autocorrelation, Pion Ltd, London (178 pp).
Cotty, P.T., Bayman, P., Egel, D.S., Elias, K.S., 1994. Agriculture, a¯atox-ins andAspergillus. In: Powell, K.S. (Ed.). The Genus Aspergillus, Plenum Press, New York, pp. 1±27.
Diener, U.L., Cole, R.J., Sanders, T.H., Payne, G.A., Lee, L.S., Klich, M.A., 1987. Epidemiology of a¯atoxin formation byAspergillus ¯avus. Annual Review of Phytopathology 25, 249±270.
Gilligan, C.A., 1995. Modeling soil-borne plant pathogens: reaction±diffu-sion models. Canadian Journal of Plant Pathology 17, 96±108. Grif®n, G.J., Baker, R., 1991. Population dynamics of plant pathogens and
associated organisms in soil in relation to infectious inoculum. In: Katan, J., DeVay, J.E. (Eds.). Soil Solarization, CRC Press, Boca Raton, FL, pp. 3±21.
Grif®n, G.J., Garren, K.H., 1974. Population levels ofAspergillusandA. nigergroup in Virginia peanut ®eld soils. Phytopathology 64, 322±325. Grif®n, G.J., Ford, R.H., Garren, K.H., 1975. Relation ofAspergillus ¯avus
colony growth on three selective media to recovery from naturally infested soil. Phytopathology 65, 704±707.
Grif®n, G.J., Garren, K.H., Taylor, J.D., 1981. In¯uence of crop rotation and minimum tillage on the population ofAspergillus ¯avusgroup in peanut ®eld soil. Plant Disease 65, 898±900.
Hampson, M.C., Coombes, J.W., 1996. Spatial distribution ofSynchytrium endobioticum, the cause of potato wart disease in ®eld soil. Plant Disease 80, 1006±1010.
Krishna Iyer, P.V., 1949. The ®rst and second moments of some probability distributions arising from points on a lattice and their application. Biometrics 36, 135±141.
Mihail, J.D., Alcorn, S.M., 1987. Macrophomina phaseolina: spatial patterns in a cultivated soil and sampling strategies. Phytopathology 77, 1126±1131.
Milne, A., 1959. The centric systematic area sample treated as a random sample. Biometrics 15, 270±279.
Morrisita, M., 1959. Measuring the dispersion of individuals and analysis of the distributional patterns. Memoirs Faculty Science Kyushu University Series 2, 215±235.
Orum, T.V., Bigelow, D.M., Nelson, M.R., Howell, D.R., Cotty, P.J., 1997. Spatial and temporal patterns ofAspergillus ¯avusstrain composition and propagule density in Yuma County, Arizona, soils. Plant Disease 81, 911±916.
Pielou, E.C., 1977. Mathematical Ecology, Wiley, New York (385 pp). Upton, G.J.G., Fingleton, B., 1985. Spatial Data Analysis by Example,
Point Pattern and Quantitative Data, vol. 1, Wiley, New York (409pp.). Vandermeer, J., 1981. Elementary Mathematical Ecology, Wiley, New
York (294pp.).