APPLIED AND ENVIRONMENTAL MICROBIOLOGY, Feb. 1996, p. 420–428 Vol. 62, No.

2
0099-2240/96/$04.0010
Copyright q 1996, American Society for Microbiology

Phospholipid Fatty Acid Composition and Heavy Metal

Tolerance of Soil Microbial Communities along
Two Heavy Metal-Polluted Gradients
in Coniferous Forests
TAINA PENNANEN,1* ÅSA FROSTEGÅRD,2 HANNU FRITZE,1
2
AND ERLAND BÅÅTH

Finnish Forest Research Institute, 01301 Vantaa, Finland,1 and Department of Microbial Ecology,
Lund University, 22362 Lund, Sweden2
Received 1 September 1995/Accepted 28 November 1995

The effects of long-term heavy metal deposition on microbial community structure and the level of bacterial
community tolerance were studied along two different gradients in Scandinavian coniferous forest soils. One
was near the Harjavalta smelter in Finland, and one was at Rönnskär in Sweden. Phospholipid fatty acid
(PLFA) analysis revealed a gradual change in soil microbial communities along both pollution gradients, and
most of the individual PLFAs changed similarly to metal pollution at both sites. The relative quantities of the
PLFAs br18:0, br17:0, i16:0, and i16:1 increased with increasing heavy metal concentration, while those of 20:4
and 18:2v6, which is a predominant PLFA in many fungi, decreased. The fungal part of the microbial biomass
was found to be more sensitive to heavy metals. This resulted in a decreased fungal/bacterial biomass ratio
along the pollution gradient towards the smelters. The thymidine incorporation technique was used to study
the heavy metal tolerance of the bacteria. The bacterial community at the Harjavalta smelter, exposed mainly
to Cu deposition, exhibited an increased tolerance to Cu but not to Cd, Ni, and Zn. At the Rönnskär smelter
the deposition consisting of a mixture of metals increased the bacterial community tolerance to all tested
metals. Both the PLFA pattern and the bacterial community tolerance were affected at lower soil metal
concentrations than were bacterial counts and bacterial activities. At Harjavalta the increased Cu tolerance of
the bacteria and the change in the PLFA pattern of the microbial community were found at the same soil Cu
concentrations. This indicated that the altered PLFA pattern was at least partly due to an altered, more
metal-tolerant bacterial community. At Rönnskär, where the PLFA data varied more, a correlation between
bacterial community tolerance and an altered PLFA pattern was found up to 10 to 15 km from the smelter.
Farther away changes in the PLFA pattern could not be explained by an increased community tolerance to metals.

Disturbances caused by heavy metals to microbial biomass the effects of alkaline pollution and different management
and activity are known to be reflected in decreased litter de- practices have been detected (7, 8, 25).
composition and subsequently less-efficient soil nutrient cy- Often the effect of a toxic substance on the soil microbial
cling (see reviews in references 3, 13, 17, and 44). Several community is measured by standard techniques, such as mea-
studies have also demonstrated heavy metal-induced changes surements of activity and biomass. However, such measure-
in specific parts of the soil microbial community. Traditional ments have the disadvantage that they will not be specific for
ways of studying the soil microbial community are time-con- the toxic substance studied. For example, although soil respi-
suming and dependent on the success of culturing the organ- ration has often been found to be a sensitive measurement to
isms. In the case of soil bacteria the culturable portion is reveal heavy metal toxicity under natural conditions (3), soil
known to represent only a limited part of the total bacteria respiration will be affected by other toxic substances as well as
(37). An approach to detect possible changes in the soil mi-
environmental factors such as soil moisture content, tempera-
crobial community, in a nonselective way, is analysis of the
ture, carbon availability, and pH. In laboratory studies, where
phospholipid fatty acid (PLFA) composition in the soil (42).
such environmental factors can be standardized, this is usually
PLFAs are located in membranes of the cells, which implies a
a lesser problem. In a more complex field situation this will
rather fast turnover of these compounds (42). Different subsets
of microorganisms have different PLFA patterns, and it is increase the variability in the data, thus decreasing the possi-
possible to characterize the features of the microbial commu- bilities to detect toxic effects. This criticism would hold for the
nity directly in a natural habitat, without an initial isolation PLFA analysis also. However, the enhanced tolerance of a
step. The PLFA profile does not give an actual species com- community to a toxic compound suggests that a selective pres-
position but instead gives an overall picture of the community sure has been exerted by the compound in question (9, 10).
structure. However, in some cases, changes in the concentra- The measurement of community tolerance would therefore be
tions of certain PLFAs may be correlated to changes in more a more direct technique to indicate toxicity in nature. Such
specific groups of organisms. On the basis of PLFA analysis, measurements would also be less affected by the environmen-
tal factors mentioned above and would therefore be a more
sensitive technique for use in field studies.
* Corresponding author. Mailing address: Finnish Forest Research Increases in the numbers of metal-tolerant bacteria (14, 15,
Institute, P.O. Box 18, 01301 Vantaa, Finland. Fax: 358-0-8572575. 18, 30, 38), actinomycetes (32), and fungi (2, 32, 46) have been
Electronic mail address: [email protected]. seen as the results of heavy metal pollution by the traditional

420
VOL. 62, 1996 EFFECTS OF HEAVY METALS ON SOIL MICROBIAL COMMUNITIES 421

plate count technique. Recently, the radioactive thymidine in-

corporation technique, which estimates the growth rate of bac-
teria by the rate of thymidine incorporation into bacterial cells,
was modified to allow measurements on bacteria extracted
from soil by homogenization-extraction (4). It has successfully
been adapted to estimate the community heavy metal toler-
ance of bacteria exposed to artificial pollution (5, 12). PLFA
analysis and the thymidine incorporation technique (to mea-
sure community tolerance) have previously been shown to be
effective methods to reveal the changes in the microbial com-
munity in soils artificially polluted with heavy metals (12, 26).
They have not, however, been applied to a more complex field
situation. In the present study we analyzed two different heavy
metal pollution gradients, one around a smelter in Finland
with a rather simple pollution pattern (mainly Cu emitted) and
one around a primary smelter in Sweden, where a mixture of
metals is emitted. In addition to studying to what extent the
heavy metals affected the microbial community, we wanted to
compare the changes in PLFA patterns at the two sites, to see
if changes in the community structure (as revealed by the
PLFA analyses) were correlated to changes in community tol-
erance. We also wanted to see if increased bacterial commu-
nity tolerance to different heavy metals could be found.

MATERIALS AND METHODS

Study areas and sampling. Two heavy metal-contaminated gradients were
sampled, one in southwest Finland at the Harjavalta smelter and one in northern
Sweden around the Rönnskär smelter. Humus (the F/H-layer) samples from 12
plots were collected from Scots pine (Pinus sylvestris L.) forest stands in Harja-
valta. Each sample consisted of 20 soil cores (diameter, 7.2 cm). The plots were
situated in groups of three at four sampling locations 0.5, 2, 4, and 8 km southeast
of the smelter and represented the controls of a larger fertilization experiment
along the pollution gradient. The samples were sieved (4-mm mesh) and stored
at 48C for 3 to 5 weeks before analysis. The soil organic matter content varied
between 64 and 98% of the dry matter (d.m.), and the H2O pH was about 4.2
along the gradient. For a more detailed description of the humus chemistry, see
the work of Fritze et al. (22).
The sampling sites around Rönnskär were also coniferous forest stands, con- FIG. 1. Soil humus Cu concentrations at different distances from the smelters
sisting mainly of Norway spruce (Picea abies L.). Thirty-nine samples were at Harjavalta (a) and Rönnskär (b).
collected from the humus (F/H) layer. The sites were situated from 2 to 55 km
both south and north of the smelter. The soil samples were frozen for 1 year, and
they were then thawed and stored at 48C for 4 weeks before further processing.
Soil characteristics of Rönnskär were similar to those of Harjavalta with respect
to their organic matter content (about 87%) and pH (around 4). The area has is suggested to be of fungal origin (20) and is known to correlate with the amount
been previously described by Nordgren et al. (35). of ergosterol (24), a sterol found only in fungi. The ratio fungal PLFA/bacterial
For the last 50 years Cu has been the dominant source of pollution in the PLFA was used as an index of the ratio of fungal/bacterial biomass in soil.
Harjavalta area. Only small quantities of Ni, Zn, Pb, and Cd have been deposited Bacterial growth rates and community tolerance to metals. Bacterial growth
(22). The Rönnskär area has been exposed since the 1930s to a mixture of heavy rates were estimated by the thymidine incorporation technique using the bacte-
metal deposition consisting mainly of Pb, Cu, Zn, As, Cd, and Hg. The Cu rial community extracted by homogenization-centrifugation as described by
concentration of the humus layer was chosen to represent the pollution gradient Bååth (4). Briefly, soil (2.5 g [fresh weight]) was homogenized with 100 ml of
for both areas. Variation in humus Cu concentrations (determined after acid distilled water and centrifuged for 10 min at 750 3 g. The supernatant, contain-
digestion) with distance from the smelters is shown in Fig. 1. The concentrations ing about 20% of the bacteria in the soil, was subsequently used. Two-milliliter
of other heavy metals in Harjavalta and Rönnskär are given by Fritze et al. (22) portions of this bacterial suspension were incubated for 2 h with 100 nM [methyl-
and Nordgren et al. (35), respectively. 3
H]thymidine (925 GBq mmol21; Amersham) at 208C. Thymidine incorporation
PLFA analysis. The phospholipid extraction and analysis of PLFAs was as was then measured in cold acid-insoluble material after filtration and washing
previously described by Frostegård et al. (26). Briefly, 0.5 g (fresh weight) of (4).
humus was extracted with a chloroform-methanol-citrate buffer mixture (1:2:0.8), Heavy metal tolerance of the bacterial community was measured by the thy-
and the lipids were separated into neutral lipids, glycolipids, and phospholipids midine incorporation technique (5, 12). The following metal salts were used:
on a silicic acid column. The phospholipids were subjected to mild alkaline CuSO4, ZnSO4, CdSO4, and Ni(NO3)2. Each metal was added at 10 concentra-
methanolysis, and the fatty acid methyl esters were separated by gas chromatog- tions, ranging between 1022 and 1027 M (final concentration), to the bacterial
raphy (flame ionization detector) using a 50-m HP-5 (phenylmethyl silicone) solution before thymidine was added. A control without any added metals was
capillary column. Hydrogen was used as a carrier gas. Peak areas were quantified always included. Thymidine incorporation was then measured as described above
by adding methyl nonadecanoate fatty acid (19:0) as an internal standard. PLFA and expressed as a percentage of this control value. The heavy metal tolerance
determinations were carried out twice on the Harjavalta samples, while the of the different bacterial communities was then estimated by calculating the
Rönnskär samples were analyzed only once. concentration of added metals which results in 50% incorporation into the
Fatty acids are designated in terms of total number of carbon atoms:number bacterial suspension compared with the control (no metal added) (IC50). Percent
of double bonds, followed by the position of the double bond from the methyl inhibition was plotted against log added metal concentration, and the IC50s were
end of the molecule. The prefixes a and i indicate anteiso and iso branching, br determined from the slope of the decreasing linear part of the curve. The
indicates unknown branching, and cy indicates a cyclopropane fatty acid. Methyl changes in the level of tolerance to different metals were expressed as DIC50s,
branching (Me) is indicated as the position of the methyl group from the carboxyl which were calculated by subtracting the IC50s of unpolluted control plots from
end of the chain. those of the three most polluted ones.
The sum of PLFAs considered to be predominantly of bacterial origin (i15:0, In Rönnskär the tolerance of the bacterial community to Cu was measured in
a15:0, 15:0, i16:0, 16:1v9, 16:v7t, i17:0, a17:0, 17:0, cy17:0, 18:1v7, and cy19:0) soil from 24 plots. Tolerance to the three other metals was measured in only six
was chosen to represent bacterial biomass (bacterial PLFA) (24). The quantity of plots, three of the most polluted (2, 2.5, and 3 km from the smelter) and three
18:2v6 was used as an indicator of fungal biomass (fungal PLFA), since 18:2v6 control plots (55, 50, and 40 km from the smelter). From the Harjavalta area the
422 PENNANEN ET AL. APPL. ENVIRON. MICROBIOL.

FIG. 2. Score plot of PCA showing the separation of the Harjavalta plots FIG. 3. Scores of the first principal component (PC 1) from the analysis in
along the first two principal components (PC 1 and PC 2) using PLFA data. Fig. 2 in relation to the soil Cu concentration showing the pollution effect on the
Black and grey symbols indicate two different laboratory measurement occasions. PLFA pattern.
Distances from the smelter are indicated as follows: squares, 0.5 km; triangles, 2
km; circles, 4 km; and diamonds, 8 km.

To investigate if the metal pollution at the two areas induced

tolerance of the bacterial community to Cu, Zn, Ni, and Cd was analyzed with all the same changes in the PLFA patterns, the loadings for the
the 12 study plots. individual PLFAs for the first PCA axis (indicating the pollu-
The number of total bacterial cells in the suspension was determined by
counting acridine orange-stained cells in the same bacterial solution used for tion) were plotted against each other (Fig. 5). The metal pol-
thymidine incorporation measurements (6). lution caused a similar change in most of the PLFAs in both
Statistical analyses. The moles percents of the PLFA values from Harjavalta study areas, and thus, a good correlation between loadings was
and Rönnskär were log10 transformed before being subjected to separate prin- found (r 5 0.76; P , 0.01). The relative moles percent of
cipal component analyses (PCA) using the Sirius program (33). The scores of the
first principal component were then used to compare changes in the PLFA several branched PLFAs, especially br18:0, br17:0, i16:0, and
pattern with the changes in bacterial community tolerance to Cu, while the i16:1, increased in both areas because of pollution (indicated
loadings from the two sites for the individual PLFAs were regressed against each by a high positive loading value for these PLFAs). On the other
other to detect if the changes in the PLFA pattern were similar for the two areas hand, the PLFAs 18:2v6 and 20:4 decreased the most because
studied.
of the pollution in both areas (Fig. 5). To give an indication of
how the proportions of PLFAs changed because of the pollu-
RESULTS tion, the mean moles percents of some PLFAs of the most
polluted (0.5 km) and the remotest (8 km) plots at Harjavalta
PLFA patterns. PCA of all PLFAs at Harjavalta resulted in can be compared. The values for the most polluted plots were
a separation of the different plots along the first axis, which 6.15, 0.60, 1.55, 3.28, and 0.16 mol% for the PLFAs i16:0,
explained 63% of the variation in the data (Fig. 2). The most br17:0, br18:0, 18:2v6, and 20:4, respectively, while the corre-
polluted plots having a high score were found to the right, and sponding values in the least polluted plots were 3.70, 0.21, 0.44,
the least polluted sites having a low score were found to the left 14.7, and 2.39 mol%.
in the graph. The two measurements of the same soil samples At both sites increasing humus Cu concentrations resulted in
separated along the second axis (explaining 11% of the total
variation). The results from the different measurement occa-
sions were reproducible in relation to the first axis. This indi-
cated that the scores for the different plots along the first axis
showed the effect of the pollution. Plotting the scores for the
first axis against the Cu concentrations of the humus (Fig. 3)
stressed this further. The pollution effect levelled off with de-
creasing soil Cu content and thus with increasing distance from
the smelter. The plots 4 km away did not differ much from the
remotest plots (8 km), indicating that a Cu concentration
around 1,000 mg g of d.m.21 was a threshold value for a
pollution effect on the PLFA pattern at Harjavalta.
The PCA performed on all PLFAs at Rönnskär extracted
two components, explaining totally 52% of the variation (data
not shown). As with the Harjavalta data, the scores of the first
component (explaining 38% of the total variation) appeared to
reflect the pollution level of the humus. However, the variation
between plots with similar Cu concentrations was greater than
at Harjavalta, and thus a less good relationship between the
scores for the different plots along the first axis and the Cu
content of the soil was found at Rönnskär (Fig. 4). Still, the FIG. 4. Scores of the first component (PC 1) from a PCA of PLFA data from
highest scores were found in the most polluted plots, and the Rönnskär in relation to the soil Cu concentration showing the pollution effect on
lowest scores were usually found in less polluted plots. the PLFA pattern.
VOL. 62, 1996 EFFECTS OF HEAVY METALS ON SOIL MICROBIAL COMMUNITIES 423

FIG. 5. Loading values for the individual PLFAs of the first principal com-
ponent (PC 1) from Harjavalta plotted against those from Rönnskär. PLFAs that
decreased in proportional abundance at higher pollution levels are found in the
lower left corner of the plot, while those that increased are found at the upper
right corner.

a reduced fungal/bacterial biomass ratio as estimated from the

fungal PLFA 18:2v6 and the sum of bacterial PLFAs. The
reduction of the ratio due to increasing metal load was clearly
seen in Harjavalta (Fig. 6a). At Rönnskär low ratios were
always found for the most polluted plots, while in the control
plots both high and low values were found (Fig. 6b).
Bacterial community tolerance to heavy metals. The bacte-
rial community tolerance to Cu was estimated by comparing
the measured IC50s (log metal concentration) by the thymidine
incorporation method. Plotting the IC50s from Harjavalta
against the Cu concentration of the humus (Fig. 7a) showed a FIG. 6. The effect of the soil Cu concentration on fungal/bacterial biomass
higher tolerance of the bacterial community to Cu with in- ratio estimated from PLFAs at Harjavalta (a) and Rönnskär (b).
creasing humus Cu concentrations. Cu tolerance of the bacte-
rial communities clearly increased in the plots situated 2 and
0.5 km from the smelter. At a distance of 4 km from the pollution level than tolerance to the other metals (Table 1). It
smelter, where the Cu concentration of the humus was around must be emphasized that the threshold levels indicated by the
1,000 mg of Cu g of d.m.21, IC50s were similar to those at the linear extrapolation will be higher than that found by using the
remote plots (8 km). No or little increased tolerance to Zn, Ni, exponential relationship, but since data for Cd, Ni, and Zn
or Cd was observed, even in the most polluted plots. This is were from only the three least and three most polluted plots,
shown in Table 1, where the increases in bacterial community only linear approximations were possible. However, even if the
tolerance (DIC50 [the difference between IC50s for the three actual threshold level was too high, it was still possible to
most and three least polluted plots]) to all of the tested metals compare levels for tolerance to different metals.
are presented. Relation between bacterial community tolerance to copper
A strong increase in the IC50s of Cu with increasing metal and the PLFA pattern. The relationship between the bacterial
concentrations of the humus was found at Rönnskär (Fig. 7b), community heavy metal tolerance and the metal pollution-
indicating an increased bacterial community tolerance to Cu. induced change in the microbial community structure is shown
There appeared to be an exponential relationship between in Fig. 8 as a correlation between the tolerance of bacterial
IC50s and log Cu content of the soil, where the IC50s levelled communities to Cu (IC50s) and the scores for the first axis of
off at Cu concentrations between 100 and 200 mg of Cu g of the PCA of the PLFA data. In Harjavalta the tolerance of the
d.m.21, corresponding to distances between 10 and 15 km from bacterial community and the PCA scores appeared to have a
the pollution source. Tolerance of the bacterial communities to strong linear correlation (r 5 0.93; P , 0.01) (Fig. 8a). How-
other metals was also found, but the increases in level of ever, in Rönnskär only the study plots with concentrations of
tolerance, measured as DIC50, were smaller than with Cu (Ta- more than 200 mg of Cu g of d.m.21, which were situated no
ble 1). This could also be indicated by estimating threshold more than 10 km from the smelter, showed a correlation be-
pollution levels for changes in IC50s for the different metals by tween tolerance of the bacterial communities to Cu and the
a linear extrapolation. Thus, the toxicities of different metals to PCA scores (r 5 0.86; P , 0.01) (Fig. 8b). In the more remote
the bacterial communities at Rönnskär were estimated by cal- plots changes in the PLFA pattern were not reflected in a
culating the crossing points of the slopes of the IC50 curves for concomitant change in the bacterial community tolerance to
the different heavy metals from the three most polluted plots Cu.
and three control plots. Cu appeared to be the most toxic Microbial biomass, bacterial number, and activity. The spe-
metal, since an increase in tolerance was found at a lower cific thymidine incorporation (expressed per bacterial cell) de-
424 PENNANEN ET AL. APPL. ENVIRON. MICROBIOL.

FIG. 7. Bacterial community tolerance to Cu (expressed as IC50s) along the FIG. 8. The relationship between the scores of the first component (PC 1)
pollution gradients at Harjavalta (a) and Rönnskär (b). from the PCA (indicating pollution effects on the microbial community struc-
ture) and bacterial community tolerance to Cu (expressed as IC50s) at Harjavalta
(a) and Rönnskär (b).

creased in Harjavalta with increasing Cu concentration (Fig.

9a), indicating a decreased growth rate of bacteria in the vi-
cinity of the smelter. A reduction in specific thymidine incor- plots, 0.5 km from the smelter (Fig. 10a). In Rönnskär bacte-
poration was found only in the most polluted plots in Rönnskär rial numbers were affected only in the two plots with the high-
(between 2 and 3 km from the smelter), but in most of the plots est Cu concentrations (Fig. 10b).
the growth rate seemed to be unaffected by the pollution (Fig. The total amount of bacterial PLFAs that were chosen to
9b). represent the bacterial part of the biomass was also little af-
The total number of acridine orange-stainable bacteria ex- fected by the metal pollution. The amount of the bacterial
tracted from the humus of Harjavalta by the homogenization- PLFAs was 390 nmol g of d.m.21 in the least polluted plots of
centrifugation method was not affected by the increasing cop- Harjavalta, which decreased to 360 nmol g of d.m.21 in the
per concentration of the humus, except at the most polluted plots nearest to the smelter. Corresponding values for the
Rönnskär area were 530 and 470 nmol g of d.m.21.
The total microbial biomass, measured as the total amount
of PLFAs, showed a reduction in both study areas because of
TABLE 1. Increase in bacterial community tolerance (DIC50)
between the three most polluted plots and the three least polluted metal pollution. The total amount of PLFAs was 1,380 nmol g
plots at Harjavalta and Rönnskära of d.m.21 in the least polluted plots of Harjavalta, which de-
creased to 910 nmol g of d.m.21 in the plots situated nearest to
Rönnskär result the smelter. Corresponding values for the Rönnskär area were
Pollutant
Harjavalta Threshold pollution level 1,550 and 1,250 nmol g of d.m.21.
DIC50 DIC50 for tolerance (mg of
Cu g of d.m.21)
DISCUSSION
Cu 2.33 2.52 0.9
Cd 0.18 1.62 1.3 The effect of heavy metal pollution on the PLFA pattern was
Ni 0.18 1.63 1.2 most clearly seen at the Harjavalta smelter, where 63% of the
Zn 0.04 1.32 1.3 variation in the PLFA data was explained by the first principal
a
By linear extrapolation the pollution level (expressed as soil Cu concentra- component. The first principal component for the Rönnskär
tions) eliciting no tolerance to the different metals was estimated at Rönnskär. data explained only 38% of the variation, indicating that much
VOL. 62, 1996 EFFECTS OF HEAVY METALS ON SOIL MICROBIAL COMMUNITIES 425

FIG. 9. The specific thymidine (TdR) incorporation of the extracted bacte- FIG. 10. The total number of the acridine orange (AO)-stainable bacteria
rial community along the pollution gradients at Harjavalta (a) and Rönnskär (b). extracted from the soil by the homogenization-centrifugation technique along
the pollution gradients at Harjavalta (a) and Rönnskär (b).

of the variation in the data was due to factors other than the
heavy metal pollution at this site. This was expected, since (29, 32). In agreement with this, a laboratory study showed
more control samples and few polluted samples were taken at increased proportions of the fungal 18:2v6 in metal-polluted
Rönnskär compared with the Harjavalta site. Still, the scores soil, at least in an arable soil (26). However, in the present field
for the first component were correlated with the pollution level study, a decrease in 18:2v6 could clearly be seen in Harjavalta,
at both sites (Fig. 3 and 4). This could therefore be considered and a similar tendency was also found for the Rönnskär site
a pollution component, and the changes in the PLFA loadings (Fig. 5). One explanation could be that the main part of 18:2v6
along this axis could be considered to be due to metal pollution. in the soils studied here originated from ectomycorrhizal fungi,
The majority of the fatty acids which exhibited changes in since these are known to constitute a great part of the fungal
relative concentrations in relation to heavy metal pollution biomass in forest soils (21). The decrease in the amount of
reacted in the same way, either increasing or decreasing, at 18:2v6 would then be due to a decrease in ectomycorrhizal
both sites. This was seen as a correlation between PLFA load- fungi, which in turn could be due to damage to the fine roots
ings along the first principal component of the different sites of trees because of pollution, as found in Harjavalta by Helm-
(Fig. 5). The changes in the PLFA patterns of the Harjavalta isaari et al. (28). The trees were also visually affected near the
and Rönnskär coniferous forest soils were therefore similar, Rönnskär smelter. In the present study the strong reduction in
suggesting that the heavy metal pollution had resulted in sim- the PLFA 18:2v6 was responsible for the decrease in the ratio
ilar changes in the microbial community structure at both sites, of fungal to bacterial biomass (Fig. 6). There was only a slight
despite the fact that different pollution situations (different decrease in the amount of bacterial PLFAs and number of the
mixtures of heavy metals) were present. This is thus an indi- acridine orange-stained bacteria in the gradient (Fig. 10). The
cation that the microbial community structure will change in assumption that the reduction in 18:2v6 was due to fungi was
similar ways because of the heavy metal pollution in different supported by microscopic measurements of fungal lengths (22)
coniferous forest soils. If similar changes in the PLFA pattern and ergosterol (23) in Harjavalta showing a decrease in total
are found at other sites, this could thus be taken as an indica- length of hyphae near the smelter. The decrease in fungal/
tion of a heavy metal effect. A different pattern would indicate bacterial biomass ratio was not as clearly seen in Rönnskär as
that other environmental factors are of greater importance. in Harjavalta (Fig. 6). This might explain why Nordgren et al.
According to previous investigations, fungi appear to be (35) did not find changes in the length of fungal hyphae in the
more tolerant to heavy metals than bacteria and actinomycetes Rönnskär area.
426 PENNANEN ET AL. APPL. ENVIRON. MICROBIOL.

Many branched PLFAs, like br17:0 and br18:0 (branching this, at least partly, was the reason for the increased commu-
site unknown), or iso- and anteiso-branched PLFAs, like i15:0, nity tolerance of bacteria in the present study. Consequently,
i16:0, i16:1, and i17:0, increased in the high-metal-content soils the heavy metal tolerance of the bacterial community should
(Fig. 5). They are commonly found in gram-positive bacteria be reflected in a change in microbial community structure,
(36). It is usually thought that gram-negative bacteria domi- which would be detected as a shift in the PLFA pattern. This
nate in metal-contaminated soils compared with gram-positive appeared to be the case, since the bacterial community toler-
bacteria (13, 14, 18, 29). Reports indicating a dominance of ance to Cu and PLFA scores of the first component were
gram-positive bacteria in metal-polluted soils are rare (41). correlated linearly in the Harjavalta samples and in the most
One explanation of these contradictory results could be that in polluted plots of Rönnskär (Fig. 8). Thus, both the Cu toler-
soil, gram-negative bacteria are typically found in the rhizo- ance and the shift in community structure due to metals were
sphere environment (27, 40). A reduction of rhizosphere mi- found at the same level of pollution. This was also found in
crohabitats could therefore lead to a decrease in the abun- laboratory studies using these two techniques (12, 26). The
dance of gram-negative bacteria. There was a reduction in the changes in PLFAs of the remote plots of Rönnskär were not
forest floor vegetation due to the metal pollution in the vicinity correlated with the bacterial community tolerance to Cu (Fig.
of the Harjavalta smelter (22, 28), and a subsequent reduction 8b). This indicated variation in the microbial community struc-
in plant root biomass would be expected. ture (as revealed by the PLFA analyses), which was not directly
The PLFA 16:1v5 is an example of a fatty acid which was not related to the heavy metal pollution. The measurement of
affected by metal contamination. Recently, Olsson et al. (39) bacterial community tolerance is a more direct way of detect-
demonstrated that 16:1v5 can be used as a signature fatty acid ing toxic effects in soil than less specific measurements like the
for arbuscular mycorrhizal fungi, but it is also found in bacteria PLFA analysis. The same would probably also be true for
(31, 34, 45). In the present study the latter is more probable, biomass and activity measurements.
since the forest floor vegetation consisted of only a few species The tolerance of the bacterial community to Cu was higher
of dwarf shrubs up to 4 km from the Harjavalta smelter (22) with increasing amounts of heavy metals in the humus layer
and the presence of arbuscular mycorrhizal plants was unlikely. (Fig. 7). The Rönnskär data indicated that bacterial tolerance
In a laboratory study in which a forest soil was contaminated was unaltered up to a humus Cu concentration of between 100
artificially with different metals and then incubated for 6 and 200 mg g of d.m.21 (equivalent to a distance of 10 to 15 km
months, the PLFA pattern was drastically affected by the metal from the smelter). Closer to the smelter, the community tol-
addition (26). The changes in the PLFA pattern in the present erance increased exponentially with the pollution level (Fig.
field study were different from those results. Although several 7b). Dı́az-Raviña et al. (12) used only three different metal
PLFAs, e.g., 20:4, br17:0, i16:0, and i16:1, reacted to the heavy levels in their laboratory study. This provided too few datum
metals similarly in both studies, several exceptions were seen. points to detect an exponential function. Instead, they used a
This discrepancy between field and laboratory studies could be linear approximation, although they stated that this linearity
due to roots, since in the laboratory experiments with incu- between IC50s and added metals did not necessarily hold at low
bated soils no plants were present. Although laboratory studies pollution levels. This was also shown later, when lower pollu-
can give indications of the direct effects of metals and may be tion levels were studied in the laboratory system and a clear
better to indicate the sensitivity of a new technique because of exponential relationship between IC50s and added amount of
reduced variation of other factors, they can never simulate metals was found (11). Any exponential relationship between
field conditions. However, PLFAs that were similarly affected IC50s and soil Cu content was more difficult to detect in the
both in the field and in the laboratory studies should be indic- Harjavalta area because of too few datum points (Fig. 7a). An
ative of organisms that are good indicators of metal pollution. increased bacterial community tolerance was found around a
This was the case with 20:4, which was the PLFA most sensitive humus Cu concentration of 1,000 mg g of d.m.21. It was, how-
to heavy metals in both studies. This PLFA is typical for eu- ever, impossible to determine the lowest level of deposition to
caryotic organisms and has, for example, been found in special induce increased Cu tolerance in Harjavalta, since the average
fungi (Mortierella subgenus Mortierella [1]) which are common of the Cu concentrations in the remotest plots (8 km from the
in forest soils (16), and it is known to be sensitive to heavy smelter) was 294 mg g of d.m.21.
metals (2). The heavy metal deposition in Harjavalta induced a bacte-
Heavy metal pollution increased the Cu tolerance of the rial community tolerance only to Cu, while at the Rönnskär
bacterial communities in both Harjavalta and Rönnskär (Fig. smelter, tolerance to all the metals tested in this study (Cu, Zn,
7), and bacterial community tolerance was the most sensitive Cd, and Ni; Table 1) was found. Even though the concentra-
technique to detect heavy metal pollution of the methods used tions of Zn and Ni near the Harjavalta smelter were about
in the present study when both areas were considered. Our 10-fold higher than those in the control plots (22), no or little
results showed that it was possible to use the new thymidine tolerance to Zn or Ni was found at this site (Table 1). The time
incorporation technique to measure community tolerance for of exposure to metals could not be the limiting factor for the
field samples with high organic matter content and not only in tolerance development in this field study, since in laboratory
laboratory studies with soils low in organic matter (12). In the studies bacterial community tolerance to Cd, Cu, Ni, and Zn
case of Rönnskär it was also possible to use soil samples that was found after 6 to 8 months (12).
had been frozen for a year with good results. This is a further An increased tolerance to metals other than the principal
advantage, as it is not always possible to process samples im- contaminant (multiple tolerance) is a phenomenon reported
mediately. several times (reviewed by Doelman [13]). Using artificially
The increased heavy metal tolerance of the bacterial com- polluted arable soil, Dı́az-Raviña et al. (12) revealed tolerance
munity could be due to either an acquired tolerance by adap- couplings for the bacterial community between Zn and Cd, Cd
tation, a genetically altered tolerance, or a shift in species and Pb, and Zn and Pb. In contrast, they did not find that Cu
composition in which organisms already tolerant become more pollution, inducing Cu tolerance, automatically led to commu-
competitive and thus more numerous (3). The latter explana- nity tolerance to other metals. This was also found at Harja-
tion was suggested to be the major reason for the high level of valta, where only Cu tolerance was found, with no evidence of
tolerant fungi in metal-polluted soils (2), and it is probable that multiple tolerance to Cd, Ni, and Zn. The interpretations of
VOL. 62, 1996 EFFECTS OF HEAVY METALS ON SOIL MICROBIAL COMMUNITIES 427

multiple tolerance are difficult in field experiments, since sev- ronment Protection Agency and the Swedish Natural Science Re-
eral pollutants are present at the same time, as was the case in search Council (E.B.).
Rönnskär. However, since the Cu pollution resulted in almost REFERENCES
the same increase in bacterial community tolerance at both 1. Amano, N., Y. Shinmen, K. Akimoto, H. Kawashima, and T. Amichi. 1992.
sites (Table 1), the increased tolerance to Cd and/or Zn at Chemotaxonomic significance of fatty acid composition in the genus Mor-
Rönnskär probably was due to these metals exerting a toxic tierella (Zygomycetes, Mortierellaceae). Mycotaxon 44:257–265.
influence and not to multiple tolerance due to Cu. However, 2. Arnebrandt, K., E. Bååth, and A. Nordgren. 1987. Copper tolerance of
microfungi isolated from polluted and unpolluted forest soil. Mycologia
some indications of multiple tolerance in Rönnskär were 79:890–895.
found. The bacterial community was tolerant to Ni, even 3. Bååth, E. 1989. Effects of heavy metals in soil on microbial processes and
though Ni emissions were minor and the amount of Ni in the populations (a review). Water Air Soil Pollut. 47:335–379.
humus was similar to that at Harjavalta. 4. Bååth, E. 1992. Thymidine incorporation into macromolecules of bacteria
extracted from soil by homogenization-centrifugation. Soil Biol. Biochem.
It is difficult to elucidate in a complex pollution situation, as 24:1157–1165.
in Rönnskär, which metal has the most toxic influence. Step- 5. Bååth, E. 1992. Measurement of heavy metal tolerance of soil bacteria using
wise multiple regression analysis (e.g., see references 19 and thymidine incorporation into bacteria extracted after homogenization-cen-
43) is not possible, as there will be a strong correlation between trifugation. Soil Biol. Biochem. 24:1167–1172.
6. Bååth, E., and K. Arnebrant. 1994. Growth rate and response of bacterial
the different metals emitted. The use of bacterial community communities to pH in limed and ash treated forest soils. Soil Biol. Biochem.
tolerance measurements will give some indications, although 26:995–1001.
multiple tolerance will complicate the interpretation. The in- 7. Bååth, E., Å. Frostegård, and H. Fritze. 1992. Soil bacterial biomass, activity,
creased community tolerance to Cu was larger than tolerance phospholipid fatty acid pattern, and pH tolerance in an area polluted with
alkaline dust deposition. Appl. Environ. Microbiol. 58:4026–4031.
to the other metals at Rönnskär (Table 1), indicating that of 8. Bååth, E., Å. Frostegård, T. Pennanen, and H. Fritze. 1995. Microbial com-
these four metals Cu was the most important one. However, we munity structure and pH response in relation to soil organic matter quality
do not know if we can compare DIC50s for different metals in wood-ash fertilized, clear-cut or burned coniferous forest soils. Soil Biol.
directly. Dı́az-Raviña et al. (12) compared threshold levels for Biochem. 27:229–240.
9. Blanck, H., and S.-Å. Wängberg. 1988. Induced community tolerance in
bacterial tolerance of different metals using a linear extrapo- marine periphyton established under arsenate stress. Can. J. Fish. Aquat.
lation and found that often the lowest threshold level was Sci. 45:1816–1819.
found for the metal exerting the toxic influence. In Rönnskär, 10. Blanck, H., and S.-Å. Wängberg. 1988. Validity of an ecotoxicological test
using the same linear extrapolation, Cu gave the lowest thresh- system: systems. Can. J. Fish. Aquat. Sci. 45:1807–1815.
11. Dı́az-Raviña, M., and E. Bååth. Unpublished data.
old level compared with the other three metals (Table 1), 12. Dı́az-Raviña, M., E. Bååth, and Å. Frostegård. 1994. Multiple heavy metal
indicating that Cu was the most toxic of these metals in the tolerance of soil bacterial communities and its measurement by a thymidine
Rönnskär area. incorporation technique. Appl. Environ. Microbiol. 60:2238–2247.
Decreased microbial activity, measured as the ratio of mi- 13. Doelman, P. 1985. Resistance of soil microbial communities to heavy metals,
p. 369–384. In V. Jensen, A. Kjøller, and L. H. Sørensen (ed.), Microbial
crobial respiration to microbial biomass, and microbial bio- communities in soil. Elsevier Science Publishing, Barking, United Kingdom.
mass in the vicinity of the Harjavalta smelter were found by 14. Doelman, P., and L. Haanstra. 1979. Effects of lead on the soil bacterial
Fritze et al. (22, 23). In our study the continuous presence of microflora. Soil Biol. Biochem. 11:487–491.
heavy metals also resulted in decreased activity and biomass of 15. Doelman, P., E. Jansen, M. Michels, and M. van Til. 1994. Effects of heavy
metals in soil on microbial diversity and activity as shown by the sensitivity-
bacteria, since a general reduction in the growth rate of bac- resistance index, an ecologically relevant parameter. Biol. Fertil. Soils 17:
teria (indicated by the specific thymidine incorporation) was 177–184.
observed in the vicinity of both smelters (Fig. 9), and the 16. Domsch, K. H., W. Gams, and T.-H. Anderson. 1980. Compendium of soil
number of bacteria was reduced in the most polluted plots in fungi. Academic Press, London.
17. Duxbury, T. 1985. Ecological aspects of heavy metal response in microor-
Harjavalta and in the two sites nearest the smelter at Rönnskär ganisms. Adv. Microb. Ecol. 8:185–235.
(Fig. 10). Changes in the microbial community structure (es- 18. Duxbury, T., and B. Bicknell. 1983. Metal-tolerant bacterial populations
timated as both the PLFA pattern and bacterial community from natural and metal-polluted soils. Soil Biol. Biochem. 15:243–250.
tolerance) were found at lower soil metal concentrations than 19. Ebregt, A., and J. M. A. M. Boldewijn. 1977. Influence of heavy metals in
spruce forest soil on amylase activity, CO2 evolution from starch, and soil
the changes in bacterial biomass and activity. This suggested respiration. Plant Soil 47:137–148.
that the altered microbial community with increased tolerance 20. Federle, T. W. 1986. Microbial distribution in soil—new techniques, p. 493–
to metals had at least partly counteracted the toxic effects of 498. In F. Megusar and M. Gantar (ed.), Perspectives in microbial ecology.
the metals, as seen by biomass and activity. Slovene Society for Microbiology, Ljubljana, Slovenia.
21. Finlay, R. D., and B. Söderström. 1989. Mycorrhizal mycelia and their role
The present study has shown the potential for using PLFA in soil and plant communities, p. 139–148. In M. Clarholm and L. Bergström
analysis and bacterial community tolerance to study heavy (ed.), Ecology of arable land. Kluwer Academic Publishers, London.
metal pollution in field sites. Both techniques appeared to be 22. Fritze, H., S. Niini, K. Mikkola, and A. Mäkinen. 1989. Soil microbial effects
well suited to detect effects of heavy metals, although the use of a Cu-Ni smelter in southwestern Finland. Biol. Fertil. Soils 8:87–94.
23. Fritze, H., P. Vanhala, J. Pietikäinen, and E. Mälkönen. Vitality fertilization
of community tolerance was more directly related to the heavy of Scots pine stands growing along a gradient of heavy metal pollution:
metal pollution. Thus, measurements of community tolerance short-term effects on microbial biomass and respiration rate of the humus
can be used to deduce which changes in the PLFA pattern are layer. Fresenius’ J. Anal. Chem., in press.
due to the metal toxicity in field studies, while changes in the 24. Frostegård, Å., and E. Bååth. The use of phospholipid fatty acid analysis to
estimate bacterial and fungal biomass in soil. Biol. Fertil. Soils, in press.
PLFA pattern can be used to some extent to elucidate which 25. Frostegård, Å., E. Bååth, and A. Tunlid. 1993. Shifts in the structure of soil
groups of organisms are affected. microbial communities in limed forest as revealed by phospholipid fatty acid
analysis. Soil Biol. Biochem. 25:723–730.
26. Frostegård, Å., A. Tunlid, and E. Bååth. 1993. Phospholipid fatty acid com-
ACKNOWLEDGMENTS
position, biomass, and activity of microbial communities from two soil types
We thank Cecilia Palmborg for sampling and measuring soil Cu in experimentally exposed to different heavy metals. Appl. Environ. Microbiol.
the Rönnskär area, Janna Pietikäinen for technical assistance in the 59:3605–3617.
27. Gilbert, G. S., J. Handelsman, and J. L. Parke. 1994. Root camouflage and
Harjavalta area, and Jonathan Steer for revising the English in this
disease control. Phytopathology 84:222–225.
paper. 28. Helmisaari, H.-S., J. Derome, H. Fritze, T. Nieminen, K. Palmgren, M.
This study was supported by a travelling grant from the Finnish Salemaa, and I. Vanha-Majamaa. Copper in Scots pine forests around a
Cultural Foundation of Northern Savo (T.P.), a grant from the Acad- heavy-metal smelter in south-western Finland. Water Air Soil Pollut., in
emy of Finland (H.F.), and grants from the Swedish National Envi- press.
428 PENNANEN ET AL. APPL. ENVIRON. MICROBIOL.

29. Hiroki, M. 1992. Effects of heavy metal contamination on soil microbial counts within different size groups. Microb. Ecol. 13:103–114.
population. Soil Sci. Plant Nutr. 38:141–147. 38. Olson, B. H., and I. Thornton. 1982. The resistance patterns to metals of
30. Huysman, F., W. Verstrate, and P. C. Brookes. 1994. Effect of manuring bacterial populations in contaminated land. J. Soil Sci. 33:271–277.
practices in increased copper concentrations on soil microbial populations. 39. Olsson, P. A., E. Bååth, I. Jakobsen, and B. Söderström. 1995. The use of
Soil Biol. Biochem. 26:103–110. phospholipid and neutral lipid fatty acids to estimate biomass of arbuscular
31. Intriago, P. 1992. The regulation of fatty acid biosynthesis in some estuarine mycorrhizal fungi in soil. Mycol. Res. 99:623–629.
strains of Flexibacter. J. Gen. Microbiol. 138:109–114. 40. Paul, E. A., and F. E. Clark. 1989. Soil microbiology and biochemistry.
32. Jordan, M. J., and M. P. Lechevalier. 1975. Effects of zinc-smelter emissions Academic Press, London.
on forest soil microflora. Can. J. Microbiol. 21:1855–1865. 41. Timoney, J. F., J. Port, J. Giles, and J. Spanier. 1978. Heavy-metal and
33. Kvalheim, O. M., and T. V. Karstang. 1987. A general purpose program for antibiotic resistance in the bacterial flora of sediments of New York Bight.
multivariate data analysis. Chemometrics Intelligent Lab. Syst. 2:235–237. Appl. Environ. Microbiol. 36:465–472.
34. Nichols, P. D., B. K. Stulp, J. G. Jones, and D. C. White. 1986. Comparison 42. Tunlid, A., and D. C. White. 1992. Biochemical analysis of biomass, commu-
of fatty acid content and DNA homology of the filamentous gliding bacteria nity structure, nutritional status, and metabolic activity of microbial commu-
Vitreoscilla, Flexibacter, and Filibacter. Arch. Microbiol. 146:1–6. nities in soil. Soil Biochem. 7:229–262.
35. Nordgren, A., T. Kauri, E. Bååth, and B. Söderström. 1986. Soil microbial 43. Tyler, G. 1976. Heavy metal pollution, phosphatase activity, and mineraliza-
activity, myceliar lengths and physiological groups of bacteria in heavy metal tion of organic phosphorus in forest soil. Soil Biol. Biochem. 8:327–332.
polluted area. Environ. Pollut. (Ser. A) 41:89–100. 44. Tyler, G. 1981. Heavy metals in soil biology and biochemistry. Soil Biochem.
36. O’Leary, W. M., and S. G. Wilkinson. 1988. Gram-positive bacteria, p. 5:372–414.
117–185. In C. Ratledge and S. G. Wilkinson (ed.), Microbial lipids, vol. 1. 45. Walker, R. W. 1969. Cis-11-hexadecenoic acid from Cytophaga hutchinsonii
Academic Press Ltd., London. lipids. Lipids 4:15–18.
37. Olsen, R. A., and L. R. Bakken. 1987. Viability of soil bacteria: optimization 46. Yamamoto, H., K. Tatsyama, and T. Uchiwa. 1985. Fungal flora of soil
of plate-counting technique and comparison between total counts and plate polluted with copper. Soil Biol. Biochem. 17:785–790.