-47-
where f(a) = the relative frequency of aggregates with size a and Ns is the number of spores counted.
2.3.10. Culture of microorganisms from filter samples.
The cellulose support pad was first wetted by adding 1ml of suspension fluid (0.1% bacteriological peptone, 0.05% Tween 80, 2% inositol) through the exit port before a further 5 to 8 ml of suspension fluid was added to the membrane filter through the inlet port. The whole monitor was then shaken on a Whirlimixer (Fisons) for 1 min. Serial dilutions were then prepared by taking 1 ml of suspension, first from the monitor and then from successive dilutions after shaking on a Whirlimixer, and adding them to 9 ml 1/4-strength Finger's solution. Dilutions down to 10-7 to 10-11 were prepared and at least two agar plates of each dilution/medium/temperature combination were then inoculated with 0.1 ml of suitable dilutions. This was spread over the agar surface with a bent glas6 rod using a turntable. For isolating fungi, 2% malt extract agar containing penicillin (20 I.U. ml- 1) and streptomycin (40 units ml-1) was used; for xerophilic fungi, dichloran-glycerol agar (DG18, Oxoid);
for bacteria and actinomycetes, 1/2-strength nutrient agar (Qxoid) plus actidione (50 mg ml-i); and for thermophilic bacteria and actinomycetes, 1/2-strength tryptone-soya agar (Qxoid) plus 0.4% casein hydrolysate (Qxoid) and actidione (50 mg m l- 1) . Plates were incubated at either 55 (thermophilic actinomycetes only), 40 or 25°C.
Concentration (xlO-7 spores m-3)
Laboratory1
Method2
Saw mill, sorting & trimming Wood chip handling Straw handling Grain elevator Pig house
IAS SEM
1.2 34 15 100 N/D
AMY SEM
1.4 (0.88)4 55 387
130 N/D
RES LM
1.3 52 11 88 0.18
SLU FM
0.24 195
4.8 165
3.9 IAU FM
0.16 205,6 5.9 165
1.56
RES culture3
(0.40)4 6.1 3.3 55
0.042
1, See Fig. 1 for abbreviations.
2, SEM, scanning electron microscopy; LM, light microscopy; FM, fluorescence microscopy.
3, Samples for culture were sampled simultaneously with those assessed by SEM but outside the sampling apparatus used for parallel samples.
4, Sampled during transportation study.
5, Short filter holders used.
6, Only one filter sample
7, One filter yielded 8 5 % of spores in large aggregates.
skin scales. These obscured microorganisms and made counting difficult, especially by SEM so that these samples could not be counted at IAS and AMY.
Preparation for FM allowed particle density to be adjusted by varying the volume of suspension filtered and counted while fluorescent staining of the microorganisms drew attention to them on skin scales. It thus helped in the counting of pig house samples. Without staining, estimates by LM were only 3 - 16% of those by FM. FM thus appears the most satisfactory method when microorganisms account for only a small proportion of the particles or when they are carried on other particles. Because results were incomplete and the estimates of microbial numbers were variable, the pig house samples were omitted from further comparisons.
By contrast with the pig house samples, samples collected in the saw mill and during the handling of wood chips consisted mostly of fungus spores while those collected in the grain elevator and during the handling of straw contained large numbers of actinomycete spores. The relative counts obtained by each method and laboratory are compared in Table 3. Total spore concentrations as estimated by LM and SEM were usually in good agreement but
-49-
Table 3. Relative spore concentrations obtained by different laboratories with four series of parallel samples using three methods of assessment
Lab.1 Method1 All spores Single spores SCU
IAS SEM 104 (16) 164 (39) 126 (28) AMY SEM 182 (58) 166 (35) 146 (25)
126 (36)2
RES LM 100 100 100 SLU FM 29 (6) 81 (17) 45 (7) IAU FM 26 (10) 90 (32) 50 (16)
1, For abbreviations, see text; SCU include single spores, aggregates of spores and other particles carrying microorganisms.
Results are expressed as percentages of the mean results obtained by LM.
Figures in brackets indicate the standard deviations.
2, Results omitting one sample giving an exceptionally large count.
estimates by FM were only 10 - 46% of these, although the two laboratories using this method agreed well. This suggests that particles were lost during preparation for FM. As expected, aggregates were dispersed during sample preparation for FM, substantially decreasing the number of aggregates and increasing numbers of single spores and SCU counted as well as increasing precision. Indeed, single spores, in FM preparations, formed 75 - 100% of the total SCU (Table 4 ) , significantly more than by other methods, but this only increased the numbers of single spores to a similar level to those found in samples for LM and SEM (Table 3), in which 40 - 85% of particles were single spores, and even exceeded LM and SEM counts of single spores with samples taken during straw handling. The precision of the spore count was decreased with LM and SEM by the numbers of large SCUs present in samples. It is surprising that there were consistently more single spores in SEM samples (median 58 and 64%) than in LM samples (median 50%). The liquid mountant used for LM might have been expected to disperse some aggregates but instead the reverse appears to have happened. Other possible explanations are disruption of SCUs by electrostatic charges during preparation for SEM and aggregation of single spores caused by differences in polarity between the glycerol
Laboratory1
Method1
Saw mill, sorting
& trimming Wood chip handling Straw handling Grain elevator
MEDIAN
Wilcoxon'
AMY RES SLU IAU
IAS SEM
83 84 74 71 45 54 57 56
64
Percentage
AMY SEM
84 86 69 70 41 43 46 43
58
s signed rank test of
IAS
*
*** ***
***
AMY
*** *
***
of single spores in SCU
RES LM
55 59 60 68 36 43 43 46
50
statistical
RES
*** ***
SLU FM
89 85 84 95 87 83 97 86
86
significance
SLU
n.s.
IAU FM
100 100 100 N/D 80 70 72 88
88
1, For abbreviations, see text
2, n.s., not significant; *, p<0.05; **, p<0.01; ***, p<0.001
triacetate and spores.
The accuracy of microscopic estimates could have been affected by a number of factors including uneven distribution of particles on the filters because of heavy deposition of large particles at the centre beneath the inlet port (unfavourable effect on SEM and LM); staining of cells (favourable effect on FM); the difficulty of counting small spores close to the limits of resolution by LM and FM; and losses during the resuspension, filtering and staining of spores for FM. Differences between LM or SEM and FM appear to be
-51-
due to negative factors affecting the latter method rather than positive factors affecting SEM and LM. The most likely cause seems to be the adsorption of spores onto filter-holders and other laboratory ware during resuspension and filtering. It was evident that aggregates were disrupted during the preparation for FM since estimates of single spores differed little from those by LM and SEM. Further study is necessary to determine the reasons for differences between LM and SEM and the sources of loss during FM preparation.
Differences between the two laboratories employing SEM were unexpectedly large for samples from the wood chip handling series. The samples were exchanged and recounted and also one of the samples prepared at IAS was recounted by W.E. using the SEM at IAS and compared with filters prepared at AMY. Estimates differed markedly for filters prepared at IAS because of charge effects in the SEM that resulted in decreased counts. Samples prepared at AMY showed no charging during counting and counts of these at the two laboratories agreed fairly well. However, IAS had changed microscopes after the first counts. The initial counts were used in comparisons made in this paper.
Estimates of airborne spore concentrations obtained from the numbers of colonies grown in culture (Table 2) were invariably smaller than those obtained by LM and SEM but sometimes exceeded those obtained by FM. Samples from different sites yielded only 12 - 62% (mean 32%) of the colonies expected from LM estimates. The smallest proportion of spores that yielded colonies came from samples from wood chip handling (12%) and most from the grain elevator (62%) with the pig house (23%), straw handling (30%) and saw mill
(45%) samples intermediate. The grain elevator and saw mill samples also yielded more colonies than spores counted by FM. By contrast, only about 2%
of the spores seen by FM in pig house samples were accounted for by colonies.
Losses of spores during preparation of samples for plating ought to be similar to those prepared for FM and other explanations must be sought. Possibly microorganisms in pig houses were exposed to inhibitors, such as ammonia or
urea, that prevented their outgrowth in culture but, otherwise, choice of agar medium or of incubation conditions may have been unsuitable for some organisms or viability was lost during prolonged storage of some substrates. However, it seems unlikely that the delay between sampling and plating was a factor since two of the samples yielding few colonies accompanied samples giving high yields and collected with only short intervals between.
The excess of colony counts over FM counts in two series is not easily explained. More spores, especially larger SCU, may have been collected in the cultured samples outside the sampling apparatus or there may be differences in spore losses due to differences in the number of dilutions used for analysis, allowing more contact with glassware or other containers, differences in detergent concentration, or fixing and staining may alter adsorption characteristics. The loss of more spores during FM preparation than during culturing could result in larger counts by culturing when spore viability is high as was indeed found.
2.4.2. Classification of spores.
Different spore types could be recognised by both SEM and LM but reference material of identified spore types is required to enable their recognition in samples. However, sufficient reference material and experience was available to enable spore types seen by LM to be identified mostly to genus. The concentrations of different spore types recognised by LM are shown in Table 5. In saw mill samples, concentrations of different types estimated by LM were similar to those obtained by SEM with Rhizopus the predominant spore type. The mould flora in the saw mill contrasted with the with the wood chip handling environment where Penicillium/Aspergillus, Mucoraceae- and Aspergillus glaucus-type spores were all more numerous than in the sawmill. Straw handling and barley grain elevator air sporas were both characterised by the predominance of bacteria + actinomycetes. Although concentrations associated with straw handling were much smaller than those in
Table 5. Classification of spores from different environments determined by light (LM) and scanning electron (SEW) microscopy.
Bacteria + actinomycetes Penicilliiun/
Aspergillus Aspergillus glaucus Rhizopus
Mucoraceae Wallemia sebi Humicola lanuginosa Botrytis
Paecilomyces Other
Spore concentration (xlO-7
Saw mill LM
0.072 0.017 0.002 1.2
- - - -
0.06
-
SEW
0.056 1.1
-
- - - -
0.11 0.02
Wood chip handling
LM
3.1 46
2.2 7.1
- - -
0.38
-
0.43 -f m-3)
Straw Grain handling Elevator
LM
10 0.23 0.13 0.06 0.04 0.02
-
0.01 0.02
-
LM
870 9.3 0.13 1.4
-
3.3 0.29
- -
0.08
Pig house LM
0.15 0.03
- - - - - - - -
the elevator, similar spore types were found at both sites and the relative proportions were generally similar. However, Wallemia sebi was only found in the barley elevator, in numbers nearly as large as Penicillium/
Aspergillus-type spores, in a total fungal spore concentration numbering almost 108 m-3. Fewest airborne spores were found in the pig house and these were mostly bacteria + actinomycetes.
2.4.3. Spore aggregation.
Aggregation of spores was noted by both LM and SEW with actinomycetes + bacteria more often aggregated and in larger aggregates than fungal spores.
Although most particles of each of the fungal 6pore types recognised by LM consisted of single spores, more than half the actinomycete + bacteria particles were of two spores or larger and several exceeded 200 spores in size. As a consequence, only 5.7% of actinomycete spores + bacteria, compared
Bacteria + ) LM actinomycetes) SEM Penicillium/ LM
Aspergillus
Aspergillus LM glaucus
Rhizopus (LM (SEM Mucoraceae LM Wallemia sebi LM
1
491 267 912 64 162 606 100 131
2
289 117 159 22 41 51 19 16
No.
3
111 58 66 14 28 11 6 3
particles in each size
4
75 30 44 6 14 9 3
-
5
40 10 11 3 15 1 3
-
6
26 10 8 1 5 1 1
-
7
29 8 2 3 2
- - -
8
21 6 8 1 3 1 1
-
9
8 9 1
-
4
- - -
10
17 5 4
-
2
- - -
class
11-20
43 14 7 2 11 1 1
-
21-50
12 4
- -
4
- - -
>50
17 14
- - - - - -
Mean size (spores
SCU-1)
7.15 8.65 1.58 2.18 2.92 1.19 1.54 1.15
LM, counted by light microscopy; SEM, counted by scanning electron microscopy
to more than 50% of fungal 6pores, were in single spore particles (Table 6;
Fig. 3). Of the fungi, Rhizopus spores were most highly aggregated and Wallemia sebi least. The mean particle sizes for these two types were 2.92 and 1.15 spores respectively while there was an average of 7.15 actinomycetes + bacteria per particle. The particle size distribution was close to log-normal, at least for particles of up to 20 actinomycetes + bacteria or up to 10 fungal spores (Fig. 3 ) . However, larger particles tended to deviate from this distribution probably due to the small numbers of these particles in the estimates, their lower efficiency of collection and the absence of fungal spore particles larger than 10 (Mucor) to 30 spores (Rhizopus). W. sebi yielded no particles larger than 3 spores so that it is uncertain whether the log-probability plot of particle size against cumulative percentage shows a linear relationship indicative of a log-normal distribution or not. Aggregate size distributions prepared by SEM showed less aggregation of Rhizopus spores than LM in saw mill samples (Table 6; Fig. 3). Aggregation of actinomycete spores + bacteria determined by SEM was similar to that determined by LM.
Particle size is important because it may affect the site of deposition of spores in the lungs. To be able to cause allergic alveolitis, deep
Fig. 3. Size distributions of particles of different spore types counted by light and scanning electron microscopy.
- 5 5 -
penetration is necessary so that the disease is usually associated with individual spores smaller than 5 um (LACEY et al. , 1972). Aggregation could cause earlier deposition but its effects on aerodynamic size of spore particles have received little study. However, FERRANDINO & AYLOR (1984) showed that the terminal velocities of particles composed of Uromyces phaseoli urediospores, L y c o p o d i u m spores or Ambrosia elatior pollen could be described by the expression:
(Vs)cluster = 0.98 (V8)single N 0.53 . . . . (5) where Vs is the terminal velocity of clusters and single spores, respectively, and N is the number of spores in a cluster. The spores were all in the size range 20 - 30 mm diameter and the aggregates were relatively compact. If this relationship holds for other spores, the aerodynamic diameter of particles consisting of two spores is about 1 8 % greater than that of the single spore and that of particles of five spores about 5 0 % greater. To double the aerodynamic diameter of the single spore requires about 15 spores in a particle. However, many chains of mould spores and loose aggregates of actinomycetes were observed in our study which could probably be expected to have a smaller aerodynamic diameter than predicted by equation ( 5 ) . Airborne bacteria in stables were found by MULLER et al. (1977) to be in clusters or attached to particles with aerodynamic diameters averaging 4.5 mm. Since bacteria are individually only about 1 mm diameter, such clusters could be composed of several hundred cells.
2.4.4. Frequency of isolation of different colony types.
Counts of individual colony types confirmed suggestions from LM that there were considerable differences in the species composition of the air spora between the different sites (Table 7 ) . The colonies isolated generally supported classifications of spore types obtained by LM but additional
Table 7. Classification of colonies from different environments grown in culture.
B a c t e r i a + a c t i n o m y c e t e s Bacillus licheniformis Other b a c t e r i a
Faeniia r e c t i v i r g u l a Thermoactinomyces
vulgaris T. thalpophilus Saccharomonospora
viridis
Saccharopolyspora hordei Streptomyces albus S. griseus Grey Streptomyces s p p . T h e r m o m o n o s p o r a s p p .
T o t a l b a c t e r i a + a c t i n o m y c e t e s
Fungi
Absidia corymbifera Mucor s p p .
Rhizomucor pusillus R. miehei
Rhizopus microsporus s s p . rhisopodiformis Acremonium s p p .
Aspergillus candidus A. flavus
A. fumigatus A. nidulans A, niger A. terreus C h r y s o s p o r i u m s p p . Cladosporium s p p . Eurotium s p p . Humicola lanuginosa Paeci1omyces
Penicillium s p p . Thermoascus crus taceus Y e a s t s
Wallemia sebi O t h e r s
T o t a l fungi
Spore c o n c e n t r a t i o n ( x l 0 - 5 m - 3)
Saw m i l l
0.011
- - - - -
- - - -
"
0.011
- - - -
3 . 8
- -
36
-
0.006
- - - - -
0.055
-
0.027
- -
0.002
-
3 9 . 8
Wood c h i p h a n d l i n g
210
-
1.0 1.0
- -
- - - - -
212.00
4.00
- - - - - - - - - - - - - - - -
150 160
-
73
-
393
Straw Grain h a n d l i n g E l e v a t o r
0.08 0.92 200
64 1.9 1.0 0.56
-
0.87 0.29 9.8
276
0.33 6 . 3
-
1.7
-
1.7 0.59 0.13 10
8.3 0.13 0.13 0.17 0.05 14
1.7 0.40
-
0.33
- -
3.50
49.9 2 . 3 6 . 8 3400 1700 1.6 1.4 68
2 . 3 2 . 3
- -
5120
20 20
-
9 . 1
- - -
9 . 1
- - - - -
4 . 5 6 . 8
- -
39 4 . 6
-
220
-
334 P i g house
4 . 0
-
0.032
-
0.005
-
0.005
-
0.005
- -
4.08
0.002 0.002
- - -
0.007 0.007
- - - -
0.007
-
0.002 0.002 0.002
-
0.047
-
0.015
-
0.093
categories could be recognised, usually with identification to genus and often to species. However, there were some suggestions of misclassification by LM.
For instance, yeasts isolated from samples taken during wood chip handling could have been classified as Mucoraceae by LM. Differences between numbers of Pe n i c i l l i u m / A s p e r g i l l u s - and Rhizopus-type spores in saw mill samples result from sampling on different occasions.
Viability, as calculated from the difference between LM and culturing, appeared to differ greatly between samples. Only 0.1 to 68% of bacteria + actinomycetes and 3 to 98% of fungal spores counted by LM grew in culture.
Fewest bacteria + actinomycetes grew from saw mill (0.14%) and barley elevator (5.9%) samples, although actinomycetes especially were most numerous in the latter. Most grew from wood chip handling samples (68%) where most colonies were bacteria. However, samples from the pig house (mostly bacteria) and from straw handling (mostly actinomycetes) showed similar viabilities (both about 27%, respectively). Fewest of the fungal spores counted by LM grew in culture from samples from the pig house (3.1%) and from wood chip handling (7.1%). In the pig house, spores may have been exposed to inhibitors, such as ammonia and urea, that decreased viability but surprisingly, only 1.5% of Penici11ium/Aspergillus-type spores grew from wood chips and only 5.0% from barley elevators produced colonies, perhaps the result of a long period of storage or susceptibility to spontaneous heating resulting from rapid microbial growth. By contrast, 87.0% grew from straw with, overall, 98% of fungal spores viable.
Aspergillus fumigatus predominated in the saw mill with R hi z o p u s microsporus ssp. rhizopodiformis; bacteria, yeasts and Penicillium chrysogenum with wood chip handling; Faenia rectivigula, Thermoactinomyces vulgaris, Thermomonospora spp., Eurotium spp. (Aspergillus glaucus group), A.
fumigatus, A.nidulans and Rhizomucor pusillus with straw handling; and F.
rectivirgula, T. vulgaris, Saccharopolyspora hordei, W a l l e m i a sebi, Penicillium spp. , R. pusillus and Absidia corymbifera in the grain elevator;
but only bacteria in the pig house. The thermophilic and thermotolerant
-59-
actinomycetes and fungi from the straw and barley suggest that they had heated spontaneously through storage while too damp (LACEY, 1980). Of these, F.
rectivirgula and T. vulgaris are well known as causes of farmer's lung (PEPYS et al. , 1963) while A. fumigatus, JR. pusilliis and A. corymbifera may be pathogenic to animals and man (AINSWORTH and AUSTWICK, 1973). R. m i c r o s p o r u s ssp. rhizopodifoimis has also been implicated in allergic alveolitis of wood chip handlers (BELIN, 1983). Possible health hazards from bacteria are not known as they were not identified but some may be sources of endotoxins.
Health effects attributable to endotoxin inlialation have been noted in pig farmers (BROUWER et al. , 1986; CLARK, 1986; DONHAM, 1986; DONHAM et al. , 1986a,b; DUTKIEWICZ, 1986; LUNDHOLM et al.,1986; OLENCHOCK et al., 1986)
2.4.5. Precision of counting.
The precision of counting samples was calculated from the results of duplicate samples from each laboratory (Table 8). The precision of of counting single spores or SCU was better than that of counting total spores.
This was in agreement with EDUARD and AALEN (1988) who found that spore aggregates had a great influence on the counting precision of mould spores.
Table 8. Relative precision of microscopic counts, based on counts of four duplicate samples per laboratory
Laboratory Method Relative precision (%)
Spores1 SCU2
Total Single
IAS SEM 15 16 14 AMY SEM 46 5 7 RES LM 22 20 21 SLU FM 22 23 20
IAU3 FM 12 13 13
Overall 26 17 16 1, Spores includes airborne bacterial cells, where present.
2, SCU includes single spores, aggregates of spores and other particles carrying microorganisms.
3, Three pairs of samples only.
The counting precision of the total spore count could be predicted from the aggregate size distribution. To obtain a relative counting precision of 10%, 200 to 300 fungal spores, and more than 500 bacteria + actinomycete spores have to be counted. Alternatively, 150-200 SCU, whether of actinomycetes + bacteria or fungal spores must be counted.
Prediction of the counting precision of the samples analysed in this study was not attempted because of differences in the numbers of spores counted, the aggregate size distributions and in counting procedure.
The quality of parallel samples, estimated from the coefficient of variation of all samples collected during the study of losses during transportation, was 9.0%. With a counting precision of ca. 1 0 % and possible variation due to losses during transportation, the coefficient of variation of parallel samples must be substantially smaller than 9%.
2.4.6. Distribution of spores on filters.
It was noticeable that filters exposed in the pig house had much denser deposits directly under the inlet port than elsewhere on the filter. To assess the effect of uneven distribution on estimates of spore concentration, the radial distribution of fungus spores on filters was determined from the counts by SEM of samples collected with the long, 37 mm diameter, closed-face filter-holders (Fig. 4 ) . Fungus spores were counted to decrease variability resulting from the presence of large SCUs. Spore density was greatest near the centre of the filters and then declined towards the edge. Thus, making counts of only part of a filter may lead to errors if even distribution is assumed. Viewing fields must be distributed over either the whole filter or a segment of it systematically or at random. Alternatively, the distance from the centre can be taken into account when calculating the spore load on the filter. Earlier studies (EDUARD and AALEN, 1988) showed a similar, although