Microb. Ecol. 7:323-330(1981)
MICROBI,4LECOLOGY
Seasonal Selection in a Freshwater Heterotrophic Bacterial Community Jimmy N. Trentham and Ted R. James Departmentof BiologicalSciences, Universityof Tennesseeat Martin, Martin, Tennessee38238, USA
Abstract. The objective of this study was to determine if a seasonal selection could be demonstrated in the heterotrophic component of a freshwater bacterial community. Surface samples were taken at approximately monthly intervals covering an annual seasonal cycle, and counts were made of the numbers of bacteria capable of growing at each of 10 incubation temperatures from 0 ~ to 45~ at 5~ intervals. Evidence for seasonal selection was provided by a 6~ shift in the mean temperature of the counts from the summer sample to the winter sample. The selection was even more evident when the number of organisms capable of growing at 10~ and those capable of growing at 35~ were compared over the seasonal cycle. The counts at these two incubation temperatures varied inversely to each other. Although a negligible number of organisms from a representative summer sample grew at 10~ 18% of the organisms from a representative winter sample grew at this temperature. The data of this study indicate that, although seasonal selection does occur, the magnitude of that selection is not great enough to permit the growth of bacteria during the coldest month to approach the levels of growth observed during the summer months. However, the selection appears to be adequate to permit significant activity during the spring and fall transition months.
Introduction The numbers and types of bacterial populations in an aquatic community are determined in part by the physical and chemical properties of the body of water, with temperature playing a principal role. Temperature particularly influences the dynamics of a bacterial community in a temperate climate. Such a community is exposed annually to a wider range of temperature fluctuations than the span over which most individual species of bacteria can grow and function effectively. Mesophilic and psychrotrophic bacteria are the two thermal types whose ranges of growth fall within the span of fluctuations of a temperate aquatic habitat. The relative seasonal contribution of a thermal type is determined both by the temperature range over which it can grow and perform significant metabolic activity and by the temperature range over which it can survive and remain available for recolonization when the temperature of the body of water again is conducive for its growth and metabolism. 0095-3628/81/0007-0323 $01.60 9 1981 Springer-VerlagNew YorkInc.
324
J.N. Trentham and T. R. James
Mesophilic bacteria, which are capable of growing at the high temperatures characteristic of the summer months, usually cannot grow during the winter months but may survive the low water temperatures of this season. In contrast, psychrophilic bacteria, which grow at the low temperatures of winter, seldom are able to grow or even to survive the high temperatures of summer. Both thermal types probably grow and function well during the transition months of spring and fall when the water temperature is moderate. One possible response of the aquatic bacterial community to the seasonal temperature fluctuations could be for the species composition to remain relatively constant throughout the year. Under this static seasonal community hypothesis, the community would be populated by mesophilic organisms capable of activity throughout all but the coldest temperatures of the annual cycle. The bacterial activity of the community would correspond to the composite temperature curves of the many mesophilic species components of the community. The overall rate of bacterial metabolism within the habitat would be at its maximum during the summer months and at its minimum during the winter season. Another plausible community response to seasonal temperature fluctuations might involve the selection of an altered composition within the bacterial community, with the components being selected for their ability to grow at the seasonal ambient temperature. This more dynamic hypothesis could result in a greater overall metabolic activity within a body of water throughout the year due to a constantly adapting bacterial community. A choice between the dynamic and static seasonal community hypotheses has been complicated by conflicting conclusions from several pertinent investigations on a diversity of aquatic environments. Boylen and Brock [4] did not detect seasonal selection among heterotrophic bacteria in Lake Wingra sediments by either culture or radioisotope techniques. Similarly, there was no seasonal trend of optimum temperatures for sulfate reduction in a salt marsh sediment detected by Abdollahi and Nedwell [ 1]. In contrast, seasonal selection of heterotrophic bacteria has been reported in estuarine water [9], in intertidal sediments [7], and in lake sediments [6]. Evidence for seasonal temperature selection among populations of denitrifiers was found in salt marsh sediments [5]. It has been reported that selection for or adaptation by a psychrotrophic bacterial population occurred in the sediment of Lake George, New York, as detected by the temperature optima for heterotrophic activity [10]. The objective of our study was to determine if seasonal selection could be demonstrated in the heterotrophic component of a freshwater bacterial community, thus contributing to the choice between the static or dynamic hypotheses of seasonal bacterial community composition. We adapted the estuarine methods used by Sieburth [9] to a very different aquatic habitat, a large shallow lake that is highly eutrophic and does not stratify. Sieburth [9] demonstrated selection by constructing a temperature spectrum, based on the composite growth response to a wide range of incubation temperatures of the many populations comprising the heterotrophic community, for each of several water samples collected throughout the year. His approach was attractive to us because it provided a means to demonstrate the presence of seasonal selection and permitted some analysis of the dynamics of the process. Our results confirmed Sieburth's [9] findings with some exceptions attributable to differences in the environments studied and in the methodology employed.
Seasonal Selection
325
Materials and Methods
Sampling Site and Procedure The water samples for this study were taken from the surface water of Reelfoot Lake, a shallow, eutrophic lake located in northwest Tennessee. This lake was formed in 181 I-I 812 by a series of earthquakes along the New Madrid fault. The lake is approximately 18 miles long by zl miles wide and is divided into two major open water basins by encroaching peninsulas of riparian swamp forest. It does not thermally stratify during the summer months. A more detailed description of Reelfoot Lake can be found in Baker [2]. Water samples were collected aseptically (from the end of a 30-foot pier located on the northwestern basin of the lake at Gray's Camp Resort) at approximately monthly intervals. The samples were packed in ice during transportation to the laboratory and were processed within 2 hours after collection.
Cultivation and Counting The growth medium was Trypticase Soy Agar supplemented with yeast extract at the rate of 2 g/I. Plates containing the prepared medium were dried at 37~ for 24 hours and then tempered overnight at the incubation temperature prior to inoculation. Samples were diluted when necessary to obtain countable plates at each incubation temperature. The dilutant was sterile Trypticase Soy Broth, which was tempered for 3 hours at the appropriate incubation temperature before it was used. Plates were inoculated by spreading 0.1 ml of the sample or a dilution of it over the surface of the dried and tempered medium in Petri plates. The inoculated plates were incubated at the following temperatures and for the following times: at 45 ~ 40 ~ and 35~ for 3 days; at 30 ~ 25 ~ and 20~ for 5 days; at 15~ and 10*(2 for 7 days; and at 5 ~ and 0~ for 14 days. The estimated number of heterotrophic bacteria in the sample capable of growing at each of the incubation temperatures was established from these counts.
Construction of Temperature Spectra Sieburth [9] used the term temperature spectrum to describe a plot of the logarithm of the viable plate counts at several incubation temperatures. Environmental temperature selection can be demonstrated by seasonal shifts of these temperature spectra. We used a slight modification of this type of analysis in our study. Temperature spectra were constructed for the 11 samples in our study by calculating what percentage the counts at each incubation temperature represented of the sum of the counts at all of the incubation temperatures. These percentages were then plotted against their respective incubation temperatures. A growth temperature spectrum for a bacterial community provides some of the same information about the community that the temperature range of growth and cardinal temperatures provide about a species.
Results Temperature spectra for the 11 samples are presented in Fig. 1. These graphs permit the visual identification of a seasonal shift of the spectra to lower temperatures during the winter. This shift is most evident when the spectrum of February 21, representative of the winter community, is compared with that of August 14, typical of the summer community. This analysis of the actual counts at the 10 growth temperatures consists of calculating (a) the mean temperature of each spectrum, depicted in Fig. 2 by the horizontal line connecting the means of the samples, (b) the standard deviation from each mean shown as a thick vertical line for each sample, (c) the 95% confidence limits for the growth
326
J.N. Trentham and T. R. James
IMor25 r
-
20 - - April 30 10 - . . e . . . . = . .. ~ - - ' l
w ~
20 10
May 11
20
June 3
~--
~1. " ~ =
-"~'-:. ~
: - ~ . . ~
;
lo 20
July 7
g ~ 20" u 1(1 Q.
,
20 10
July 15
-
Aug 14
-
~
---
9
~-
=
"-
9
~--
a
i
20 10
- ~
-
Sept 4
~
~
-
2o-oc,22
lO20
-
_ ~ Jan 20
0
5
10
15
20
25
30
35
40
45
Incubation temperature
Fig. 1. Temperature spectra for heterotrophic bacteria from each of I 1 Reelfoot Lake, Tennessee, samples spanning an annual seasonal cycle. A spectrum consists of the percentages which the counts at each incubation temperature are of the total counts at all incubation temperatures plotted against their respective incubation temperatures.
temperature range of each sample presented as thin vertical lines, and (d) the 95% confidence limits of each mean (-t- 1.96 SE) indicated numerically above each sample. The 95% confidence limit for the community is used to define the temperature growth range of 95% of the bacterial community. The 95% confidence limit of the mean ranged from _+0.05~ to-+ 0.27~ with 9 samples included between---0.10~ and---+0.20~ The trend toward a seasonal shift seen in Fig. 1 is verified by these data. The mean temperatures of the spectra ranged from a lo~v of 19.4~ on February 21 to a high of 25.4~ on August 14. Although this 6~ shift is not dramatic, the changes in the mean during the year resulted in a smooth curve directly related to the seasons. A regression analysis between mean growth temperatures of the spectra and the lake water temperatures at the times of sampling was used to evaluate further the effect of seasonal temperature fluctuation on the mean temperatures of the spectra. The results of this analysis are presented in Fig. 3. The correlation was 0.90, which is statistically significant at the 0.001 level. This confirms the direct relationship between environmental temperature and mean temperatures of the spectra.
Seasonal Selection
327
45--
0.190 0.27
40--
+ 1.96SE = 0.06 35-3O 0 P 22 25 E 2O
0.13
4 ~.16
0.10
0.19
0.10
Oi 0.10
I
15
10 5 0
I
2
I
3
I
I
I
I
I
I
4
5
6
7
8
9
I
10
I
11
I
12
I
I
1
2
Months Fig. 2. Analysis of the I I hetemtrophic bacterial temperature spectra from Reelfoot Lake, Tennessee, including mean temperature, standard deviation, 95% confidence limits of the mean, and 95% confidence limits for the growth range. The mean temperatures of the spectra are connected by the horizontal line. The standard deviation from each mean is depicted as thick vertical lines, while the 95% confidence limits of the growth range are shown as thin vertical lines. The 95% confidence interval is represented numerically above each sample as + 1.96 SE.
The seasonal effects of temperature on the bacterial community can be demonstrated more dramatically and probably more meaningfully by comparing the proportion of the community which can grow at temperatures near the upper and lower extremes of the growth range. Such a comparison, presented in Fig. 4, was constructed by plotting the percentages of the total counts which grew at 35~ and the percentages which grew at 10~ for each of the 11 samples. The portion of the bacterial community which grew at 10~ varied from a high of 18.2% of the total growth in March to a low of 0.5% during August. Conversely, that portion of the community which grew at 35~ rose from a low of 3.8% in February to reach a high of 17.8% in June. It is readily evident from these data that significant seasonal changes occurred in the percentages of the community capable of growing at these two temperatures, and that growth responses at 10~ and at 35~ were asynchronous.
Discussion Our study provides evidence for a seasonal selection or adaptation of thermal types in an aquatic, heterotrophic bacterial community (Figs. 1, 2, and 4). This evidence is consistent with the findings of several investigations [3, 5-7, 9, 10] for seasonal selection or adaptation in a variety of environments but is contrary to other studies [ I, 4].
J.N. Trentham and T. R. James
328
26-7_15~8-14 _o/ , ==-~/ 9
25 24
9 10-22
/
/
i 23
~1 - / x~)Y~5 11
L E 22--
q'~q)f~J"
/
9 6-3
e4-30
o 21
/
9 3-25
20 - / f 9 1-20 9 2-21
19
18
I
10
I
I
1
15 20 25 Lakewater temperature C
I
30
Fig. 3. A regression analysis between mean growth temperatures of the I 1 heterotrophic bacterial spectra and
Reel foot Lake, Tennessee, lake water temperatures at the time of sampling.
The investigations cited are dissimilar in design and were conducted on several types of environments, so conflicts in conclusions are expected. The study presented here was similar enough in design to that of Sieburth [9] for comparisons to be made. In spite of the differences between the freshwater environment of our study and the estuarine environment of Sieburth's investigation, our results conf'm-ned his f'mdings with some specific differences, which probably can be attributed to differences in the environments and in the methodology. Although evidence for a seasonal shift in thermal types is provided by both Sieburth [9] and us, it is not possible to compare quantitatively the extent of the shifts in the'two studies because there were significant differences in data analysis. We estimated statistically that the maximum seasonal shift in the mean temperature was 6~ (Fig. 2). Sieburth [9] estimated graphically that the maximum seasonal shift was 12-15~ In addition, Sieburth [9] limited his study to the dominant flora, whereas we calculated the means o f spectra based on the counts of all bacteria in the samples. We did not detect a lag between the mean spectral temperatures and the water temperatures, in contrast to an estimated 2-month lag reported by Sieburth [9]. The
Seasonal Selection
329
20-
10C
15 -
"6
10:-
35 C
x
I
I,
I
I
i
1
I
I
I
I
I
I
I
2
3
4
5
6
7
8
9
10
11
12
1
2
Months
Fig. 4. Comparison of the percentages of the total heterotrophic bacterial counts from Reelfoot Lake, Tennessee, which grew at 35~ and the percentages which grew at 10~ for each sample.
absence of a detectable lag in a lake as compared to an estuary may be explained by a greater availability of microorganisms for recolonization of the former as the seasonal temperature shifts create an inhospitable environment for the existing organisms within the community. Although both an estuary and a lake have a constant influx of new bacteria into them from the surrounding soil and stream communities, Rheinheimer [8] suggests that freshwater probably is more suitable than brackish water for the survival and growth of these exogenous microorganisms. Consequently, there should be a greater variety of thermal types present in freshwater throughout the year, reducing the time required for recolonization as the temperature changes. The 6oC maximum mean temperature shift between summer and winter spectra probably underestimated the seasonal selection which occurred at the lower and higher temperatures of the ecologically significant range. The underestimation probably occurred because both the bacteria adapted to the summer temperatures and those adapted to winter temperatures grew in the mid-range of a temperature spectrum. This large number of organisms which grew at the mid-range temperatures during all seasons concealed any sizable changes at the two extremes of a spectrum. The data presented in Fig. 4 support this interpretation by demonstrating that the seasonal changes in counts at the extreme temperatures of the spectra are much larger than would be expected from the 6~ maximum mean temperature shift. Although the percent of the counts capable of growing at 10~ in the summer sample was negligible, it was as high as 18% during the winter. The counts at 10~ and at 35~ varied asynchronously. Although the mean
330
J.N. Trentham and T. R. James
temperatures of the spectra are useful variables in our analysis, a study of the counts at the extreme temperatures of the spectrum is more ecologically significant for a comparison of the changes in the summer and winter communities. Although our study provided evidence for a seasonal shift, the shift does not appear to be great enough to permit a significant number of bacteria to grow when the water is at its coldest, often as low as 4~ or less. Only 3.6 % and 9.7% of the total counts from the January sample would grow at 0~ and 5~ respectively (Fig. 1). In contrast, 18.2% of the total counts from the March sample grew at 10~ a temperature in the range that prevails during the fall and spring transitional months (Fig. 1). These observations suggest that the seasonal shift was not adequate to permit growth during the coldest winter period, but that it provided the potential for signficant growth during the seasonal transitional months of the fall and spring. The data from this study support the dynamic seasonal community hypothesis for determining the seasonal bacterial community composition. There was a discernible shift in the mean temperature of each spectrum, with the mean being highly correlated with the water temperature at the time of sampling. The seasonal shift is most evident at the two extremes of the ecological range.
References 1. Abdollahi, H., and D. B. Nedwell: Seasonal temperature as a factor influencing bacterial sulfate reduction in a saltmarsh sediment. Microb. Ecol. 5, 73-79 (1979) 2. Baker, C. L.: A preliminary study of Reelfoot Lake with suggestions for possible improvements. J. Tenn. Acad. Sci. 4, 4-21 (1940) 3. Bott, T. L.: Bacterial growth rates and temperature optima in a stream with a fluctuating thermal regime. Limnol. Oceanogr. 20, 191-197 (1975) 4. Boylen, C. W., and T. D. Brock: Bacterial decomposition processes in Lake Wingra sediments during winter. Linmol. Oceanogr. lg, 628-634 (1973) 5. Kaplan, W. A., J. M. Teal, and 1. Valiela: Dentrification in salt marsh sediments: evidence for seasonal temperature selection among populations of denitrifiers. Microb. Ecol. 3, 193-204 (! 977) 6. Larkin, J. M.: Seasonal incidence of bacterial temperature types in Louisiana soil and water. Appl. Micobriol. 20, 286-288 (1970) 7. Nedwell, D. B., and G. D. Floodgate: The seasonal selection by temperature of heterotrophic bacteria in an intertidal sediment. Mar. Biol. 11,306-310 (1971) 8. Rheinheimer, G.: Introduction. In G. Rheinheimer (ed.): Microbial Ecology of Brackish Water Environment. Springer-Verlag, New York (1977) 9. Sieburth, J. McN.: Seasonal selection of estuary bacteria by water temperature. J. Exp. Mar. Biol. Ecol. 1, 98-121 (1967) 10. Tison, D. L., D. H. Pope, and C. W. Boylen: Influence of seasonal temperature on the temperature optima of bacteria in sediments of Lake George, New York. Appl.Environ. Microbiol. 39,675---677 (1980)
MICROBI,4LECOLOGY
Seasonal Selection in a Freshwater Heterotrophic Bacterial Community Jimmy N. Trentham and Ted R. James Departmentof BiologicalSciences, Universityof Tennesseeat Martin, Martin, Tennessee38238, USA
Abstract. The objective of this study was to determine if a seasonal selection could be demonstrated in the heterotrophic component of a freshwater bacterial community. Surface samples were taken at approximately monthly intervals covering an annual seasonal cycle, and counts were made of the numbers of bacteria capable of growing at each of 10 incubation temperatures from 0 ~ to 45~ at 5~ intervals. Evidence for seasonal selection was provided by a 6~ shift in the mean temperature of the counts from the summer sample to the winter sample. The selection was even more evident when the number of organisms capable of growing at 10~ and those capable of growing at 35~ were compared over the seasonal cycle. The counts at these two incubation temperatures varied inversely to each other. Although a negligible number of organisms from a representative summer sample grew at 10~ 18% of the organisms from a representative winter sample grew at this temperature. The data of this study indicate that, although seasonal selection does occur, the magnitude of that selection is not great enough to permit the growth of bacteria during the coldest month to approach the levels of growth observed during the summer months. However, the selection appears to be adequate to permit significant activity during the spring and fall transition months.
Introduction The numbers and types of bacterial populations in an aquatic community are determined in part by the physical and chemical properties of the body of water, with temperature playing a principal role. Temperature particularly influences the dynamics of a bacterial community in a temperate climate. Such a community is exposed annually to a wider range of temperature fluctuations than the span over which most individual species of bacteria can grow and function effectively. Mesophilic and psychrotrophic bacteria are the two thermal types whose ranges of growth fall within the span of fluctuations of a temperate aquatic habitat. The relative seasonal contribution of a thermal type is determined both by the temperature range over which it can grow and perform significant metabolic activity and by the temperature range over which it can survive and remain available for recolonization when the temperature of the body of water again is conducive for its growth and metabolism. 0095-3628/81/0007-0323 $01.60 9 1981 Springer-VerlagNew YorkInc.
324
J.N. Trentham and T. R. James
Mesophilic bacteria, which are capable of growing at the high temperatures characteristic of the summer months, usually cannot grow during the winter months but may survive the low water temperatures of this season. In contrast, psychrophilic bacteria, which grow at the low temperatures of winter, seldom are able to grow or even to survive the high temperatures of summer. Both thermal types probably grow and function well during the transition months of spring and fall when the water temperature is moderate. One possible response of the aquatic bacterial community to the seasonal temperature fluctuations could be for the species composition to remain relatively constant throughout the year. Under this static seasonal community hypothesis, the community would be populated by mesophilic organisms capable of activity throughout all but the coldest temperatures of the annual cycle. The bacterial activity of the community would correspond to the composite temperature curves of the many mesophilic species components of the community. The overall rate of bacterial metabolism within the habitat would be at its maximum during the summer months and at its minimum during the winter season. Another plausible community response to seasonal temperature fluctuations might involve the selection of an altered composition within the bacterial community, with the components being selected for their ability to grow at the seasonal ambient temperature. This more dynamic hypothesis could result in a greater overall metabolic activity within a body of water throughout the year due to a constantly adapting bacterial community. A choice between the dynamic and static seasonal community hypotheses has been complicated by conflicting conclusions from several pertinent investigations on a diversity of aquatic environments. Boylen and Brock [4] did not detect seasonal selection among heterotrophic bacteria in Lake Wingra sediments by either culture or radioisotope techniques. Similarly, there was no seasonal trend of optimum temperatures for sulfate reduction in a salt marsh sediment detected by Abdollahi and Nedwell [ 1]. In contrast, seasonal selection of heterotrophic bacteria has been reported in estuarine water [9], in intertidal sediments [7], and in lake sediments [6]. Evidence for seasonal temperature selection among populations of denitrifiers was found in salt marsh sediments [5]. It has been reported that selection for or adaptation by a psychrotrophic bacterial population occurred in the sediment of Lake George, New York, as detected by the temperature optima for heterotrophic activity [10]. The objective of our study was to determine if seasonal selection could be demonstrated in the heterotrophic component of a freshwater bacterial community, thus contributing to the choice between the static or dynamic hypotheses of seasonal bacterial community composition. We adapted the estuarine methods used by Sieburth [9] to a very different aquatic habitat, a large shallow lake that is highly eutrophic and does not stratify. Sieburth [9] demonstrated selection by constructing a temperature spectrum, based on the composite growth response to a wide range of incubation temperatures of the many populations comprising the heterotrophic community, for each of several water samples collected throughout the year. His approach was attractive to us because it provided a means to demonstrate the presence of seasonal selection and permitted some analysis of the dynamics of the process. Our results confirmed Sieburth's [9] findings with some exceptions attributable to differences in the environments studied and in the methodology employed.
Seasonal Selection
325
Materials and Methods
Sampling Site and Procedure The water samples for this study were taken from the surface water of Reelfoot Lake, a shallow, eutrophic lake located in northwest Tennessee. This lake was formed in 181 I-I 812 by a series of earthquakes along the New Madrid fault. The lake is approximately 18 miles long by zl miles wide and is divided into two major open water basins by encroaching peninsulas of riparian swamp forest. It does not thermally stratify during the summer months. A more detailed description of Reelfoot Lake can be found in Baker [2]. Water samples were collected aseptically (from the end of a 30-foot pier located on the northwestern basin of the lake at Gray's Camp Resort) at approximately monthly intervals. The samples were packed in ice during transportation to the laboratory and were processed within 2 hours after collection.
Cultivation and Counting The growth medium was Trypticase Soy Agar supplemented with yeast extract at the rate of 2 g/I. Plates containing the prepared medium were dried at 37~ for 24 hours and then tempered overnight at the incubation temperature prior to inoculation. Samples were diluted when necessary to obtain countable plates at each incubation temperature. The dilutant was sterile Trypticase Soy Broth, which was tempered for 3 hours at the appropriate incubation temperature before it was used. Plates were inoculated by spreading 0.1 ml of the sample or a dilution of it over the surface of the dried and tempered medium in Petri plates. The inoculated plates were incubated at the following temperatures and for the following times: at 45 ~ 40 ~ and 35~ for 3 days; at 30 ~ 25 ~ and 20~ for 5 days; at 15~ and 10*(2 for 7 days; and at 5 ~ and 0~ for 14 days. The estimated number of heterotrophic bacteria in the sample capable of growing at each of the incubation temperatures was established from these counts.
Construction of Temperature Spectra Sieburth [9] used the term temperature spectrum to describe a plot of the logarithm of the viable plate counts at several incubation temperatures. Environmental temperature selection can be demonstrated by seasonal shifts of these temperature spectra. We used a slight modification of this type of analysis in our study. Temperature spectra were constructed for the 11 samples in our study by calculating what percentage the counts at each incubation temperature represented of the sum of the counts at all of the incubation temperatures. These percentages were then plotted against their respective incubation temperatures. A growth temperature spectrum for a bacterial community provides some of the same information about the community that the temperature range of growth and cardinal temperatures provide about a species.
Results Temperature spectra for the 11 samples are presented in Fig. 1. These graphs permit the visual identification of a seasonal shift of the spectra to lower temperatures during the winter. This shift is most evident when the spectrum of February 21, representative of the winter community, is compared with that of August 14, typical of the summer community. This analysis of the actual counts at the 10 growth temperatures consists of calculating (a) the mean temperature of each spectrum, depicted in Fig. 2 by the horizontal line connecting the means of the samples, (b) the standard deviation from each mean shown as a thick vertical line for each sample, (c) the 95% confidence limits for the growth
326
J.N. Trentham and T. R. James
IMor25 r
-
20 - - April 30 10 - . . e . . . . = . .. ~ - - ' l
w ~
20 10
May 11
20
June 3
~--
~1. " ~ =
-"~'-:. ~
: - ~ . . ~
;
lo 20
July 7
g ~ 20" u 1(1 Q.
,
20 10
July 15
-
Aug 14
-
~
---
9
~-
=
"-
9
~--
a
i
20 10
- ~
-
Sept 4
~
~
-
2o-oc,22
lO20
-
_ ~ Jan 20
0
5
10
15
20
25
30
35
40
45
Incubation temperature
Fig. 1. Temperature spectra for heterotrophic bacteria from each of I 1 Reelfoot Lake, Tennessee, samples spanning an annual seasonal cycle. A spectrum consists of the percentages which the counts at each incubation temperature are of the total counts at all incubation temperatures plotted against their respective incubation temperatures.
temperature range of each sample presented as thin vertical lines, and (d) the 95% confidence limits of each mean (-t- 1.96 SE) indicated numerically above each sample. The 95% confidence limit for the community is used to define the temperature growth range of 95% of the bacterial community. The 95% confidence limit of the mean ranged from _+0.05~ to-+ 0.27~ with 9 samples included between---0.10~ and---+0.20~ The trend toward a seasonal shift seen in Fig. 1 is verified by these data. The mean temperatures of the spectra ranged from a lo~v of 19.4~ on February 21 to a high of 25.4~ on August 14. Although this 6~ shift is not dramatic, the changes in the mean during the year resulted in a smooth curve directly related to the seasons. A regression analysis between mean growth temperatures of the spectra and the lake water temperatures at the times of sampling was used to evaluate further the effect of seasonal temperature fluctuation on the mean temperatures of the spectra. The results of this analysis are presented in Fig. 3. The correlation was 0.90, which is statistically significant at the 0.001 level. This confirms the direct relationship between environmental temperature and mean temperatures of the spectra.
Seasonal Selection
327
45--
0.190 0.27
40--
+ 1.96SE = 0.06 35-3O 0 P 22 25 E 2O
0.13
4 ~.16
0.10
0.19
0.10
Oi 0.10
I
15
10 5 0
I
2
I
3
I
I
I
I
I
I
4
5
6
7
8
9
I
10
I
11
I
12
I
I
1
2
Months Fig. 2. Analysis of the I I hetemtrophic bacterial temperature spectra from Reelfoot Lake, Tennessee, including mean temperature, standard deviation, 95% confidence limits of the mean, and 95% confidence limits for the growth range. The mean temperatures of the spectra are connected by the horizontal line. The standard deviation from each mean is depicted as thick vertical lines, while the 95% confidence limits of the growth range are shown as thin vertical lines. The 95% confidence interval is represented numerically above each sample as + 1.96 SE.
The seasonal effects of temperature on the bacterial community can be demonstrated more dramatically and probably more meaningfully by comparing the proportion of the community which can grow at temperatures near the upper and lower extremes of the growth range. Such a comparison, presented in Fig. 4, was constructed by plotting the percentages of the total counts which grew at 35~ and the percentages which grew at 10~ for each of the 11 samples. The portion of the bacterial community which grew at 10~ varied from a high of 18.2% of the total growth in March to a low of 0.5% during August. Conversely, that portion of the community which grew at 35~ rose from a low of 3.8% in February to reach a high of 17.8% in June. It is readily evident from these data that significant seasonal changes occurred in the percentages of the community capable of growing at these two temperatures, and that growth responses at 10~ and at 35~ were asynchronous.
Discussion Our study provides evidence for a seasonal selection or adaptation of thermal types in an aquatic, heterotrophic bacterial community (Figs. 1, 2, and 4). This evidence is consistent with the findings of several investigations [3, 5-7, 9, 10] for seasonal selection or adaptation in a variety of environments but is contrary to other studies [ I, 4].
J.N. Trentham and T. R. James
328
26-7_15~8-14 _o/ , ==-~/ 9
25 24
9 10-22
/
/
i 23
~1 - / x~)Y~5 11
L E 22--
q'~q)f~J"
/
9 6-3
e4-30
o 21
/
9 3-25
20 - / f 9 1-20 9 2-21
19
18
I
10
I
I
1
15 20 25 Lakewater temperature C
I
30
Fig. 3. A regression analysis between mean growth temperatures of the I 1 heterotrophic bacterial spectra and
Reel foot Lake, Tennessee, lake water temperatures at the time of sampling.
The investigations cited are dissimilar in design and were conducted on several types of environments, so conflicts in conclusions are expected. The study presented here was similar enough in design to that of Sieburth [9] for comparisons to be made. In spite of the differences between the freshwater environment of our study and the estuarine environment of Sieburth's investigation, our results conf'm-ned his f'mdings with some specific differences, which probably can be attributed to differences in the environments and in the methodology. Although evidence for a seasonal shift in thermal types is provided by both Sieburth [9] and us, it is not possible to compare quantitatively the extent of the shifts in the'two studies because there were significant differences in data analysis. We estimated statistically that the maximum seasonal shift in the mean temperature was 6~ (Fig. 2). Sieburth [9] estimated graphically that the maximum seasonal shift was 12-15~ In addition, Sieburth [9] limited his study to the dominant flora, whereas we calculated the means o f spectra based on the counts of all bacteria in the samples. We did not detect a lag between the mean spectral temperatures and the water temperatures, in contrast to an estimated 2-month lag reported by Sieburth [9]. The
Seasonal Selection
329
20-
10C
15 -
"6
10:-
35 C
x
I
I,
I
I
i
1
I
I
I
I
I
I
I
2
3
4
5
6
7
8
9
10
11
12
1
2
Months
Fig. 4. Comparison of the percentages of the total heterotrophic bacterial counts from Reelfoot Lake, Tennessee, which grew at 35~ and the percentages which grew at 10~ for each sample.
absence of a detectable lag in a lake as compared to an estuary may be explained by a greater availability of microorganisms for recolonization of the former as the seasonal temperature shifts create an inhospitable environment for the existing organisms within the community. Although both an estuary and a lake have a constant influx of new bacteria into them from the surrounding soil and stream communities, Rheinheimer [8] suggests that freshwater probably is more suitable than brackish water for the survival and growth of these exogenous microorganisms. Consequently, there should be a greater variety of thermal types present in freshwater throughout the year, reducing the time required for recolonization as the temperature changes. The 6oC maximum mean temperature shift between summer and winter spectra probably underestimated the seasonal selection which occurred at the lower and higher temperatures of the ecologically significant range. The underestimation probably occurred because both the bacteria adapted to the summer temperatures and those adapted to winter temperatures grew in the mid-range of a temperature spectrum. This large number of organisms which grew at the mid-range temperatures during all seasons concealed any sizable changes at the two extremes of a spectrum. The data presented in Fig. 4 support this interpretation by demonstrating that the seasonal changes in counts at the extreme temperatures of the spectra are much larger than would be expected from the 6~ maximum mean temperature shift. Although the percent of the counts capable of growing at 10~ in the summer sample was negligible, it was as high as 18% during the winter. The counts at 10~ and at 35~ varied asynchronously. Although the mean
330
J.N. Trentham and T. R. James
temperatures of the spectra are useful variables in our analysis, a study of the counts at the extreme temperatures of the spectrum is more ecologically significant for a comparison of the changes in the summer and winter communities. Although our study provided evidence for a seasonal shift, the shift does not appear to be great enough to permit a significant number of bacteria to grow when the water is at its coldest, often as low as 4~ or less. Only 3.6 % and 9.7% of the total counts from the January sample would grow at 0~ and 5~ respectively (Fig. 1). In contrast, 18.2% of the total counts from the March sample grew at 10~ a temperature in the range that prevails during the fall and spring transitional months (Fig. 1). These observations suggest that the seasonal shift was not adequate to permit growth during the coldest winter period, but that it provided the potential for signficant growth during the seasonal transitional months of the fall and spring. The data from this study support the dynamic seasonal community hypothesis for determining the seasonal bacterial community composition. There was a discernible shift in the mean temperature of each spectrum, with the mean being highly correlated with the water temperature at the time of sampling. The seasonal shift is most evident at the two extremes of the ecological range.
References 1. Abdollahi, H., and D. B. Nedwell: Seasonal temperature as a factor influencing bacterial sulfate reduction in a saltmarsh sediment. Microb. Ecol. 5, 73-79 (1979) 2. Baker, C. L.: A preliminary study of Reelfoot Lake with suggestions for possible improvements. J. Tenn. Acad. Sci. 4, 4-21 (1940) 3. Bott, T. L.: Bacterial growth rates and temperature optima in a stream with a fluctuating thermal regime. Limnol. Oceanogr. 20, 191-197 (1975) 4. Boylen, C. W., and T. D. Brock: Bacterial decomposition processes in Lake Wingra sediments during winter. Linmol. Oceanogr. lg, 628-634 (1973) 5. Kaplan, W. A., J. M. Teal, and 1. Valiela: Dentrification in salt marsh sediments: evidence for seasonal temperature selection among populations of denitrifiers. Microb. Ecol. 3, 193-204 (! 977) 6. Larkin, J. M.: Seasonal incidence of bacterial temperature types in Louisiana soil and water. Appl. Micobriol. 20, 286-288 (1970) 7. Nedwell, D. B., and G. D. Floodgate: The seasonal selection by temperature of heterotrophic bacteria in an intertidal sediment. Mar. Biol. 11,306-310 (1971) 8. Rheinheimer, G.: Introduction. In G. Rheinheimer (ed.): Microbial Ecology of Brackish Water Environment. Springer-Verlag, New York (1977) 9. Sieburth, J. McN.: Seasonal selection of estuary bacteria by water temperature. J. Exp. Mar. Biol. Ecol. 1, 98-121 (1967) 10. Tison, D. L., D. H. Pope, and C. W. Boylen: Influence of seasonal temperature on the temperature optima of bacteria in sediments of Lake George, New York. Appl.Environ. Microbiol. 39,675---677 (1980)
