across managed small streams in western Washington. J. Dongâ , J. Chenâ¡â, K. D. Brosofskeâ¡ and R. J. Naiman§. Leaving riparian strips on both sides of a ...
Journal of Environmental Management (1998) 53, 309–321 Article No. ev980217
Modelling air temperature gradients across managed small streams in western Washington J. Dong†, J. Chen‡∗, K. D. Brosofske‡ and R. J. Naiman§ Leaving riparian strips on both sides of a stream is widely accepted to be an effective management approach in sustaining the valuable functions of stream and riparian ecosystems. The authors’ overall objective is to provide microclimatic information for assessing the effectiveness of these strips. During the summer of 1993 and 1994, air temperatures were collected across 20 small, buffered streams in western Washington, USA, including five streams sampled before and after harvesting of the forest. These data were statistically analysed to examine the effects of adjacent harvesting with preservation of 16–72 m riparian forest strips. Regression models were developed to predict air temperatures at the stream and buffer edges, the difference between two locations, and seasonal changes. The authors found: (1) clearcutting in winter 1993/94 increased air temperature on the stream by up to 4°C, and changes in temperature variability from the stream to the upland, measured by coefficient of variation (CV), were significantly higher after harvesting; (2) forest buffers provided minimal protection for stream air temperature during the middle of summer (July) but were more effective early and late in the season; (3) buffer width was not a significant variable in predicting stream air temperature, suggesting that even a 72 m buffer was not sufficient to maintain a stream environment because of greater depth of edge influences. 1998 Academic Press
Keywords: air temperature, stream, riparian, prediction, regression, stream buffers.
Introduction Managing stream ecosystems by partially preserving riparian vegetation has become an important practice in many parts of the world. Riparian vegetation provides many critical functions at both stand and landscape levels (Gregory et al., 1991; Naiman et al., 1998). It is generally believed that water quality, wildlife habitat, biological diversity, erosion, landscape connectivity and other ecological values can be sustained while adjacent harvesting or land conversion processes continue (Warner and Hendrix, 1984; FEMAT, 1993; Berg, 1995). While some functions provided by riparian buffers are obviously positive (e.g. shading and reducing water temperature, filtering surface erosional inputs into the stream), scientists and managers suspect that narrow riparian buffers may not provide sufficient protection for overall ecosystem functioning in the long term. Current scientific investigations are focused 0301–4797/98/040309+13 $30.00/0
on various aspects of ecosystem functioning provided by these buffers (Naiman, 1992). Stream and riparian ecosystems are highly connected landscape elements, forming a network across the landscape. However, more than 80% of the riparian corridor area of North America and Europe has disappeared in the last 200 years (Naiman and De´camp, 1990, 1997). In the west coastal watersheds of the Pacific North-west, there exists more than 650 km/km2 of streams; 7·5% of the landscape lies within 60 m from the streams (Chen et al., 1993). These regions, although minor in terms of area, are critical habitats supporting about 67% of plant species and 70% of vertebrates in the region (Kelsey and West, 1998). Understanding the physical environment of steam and riparian ecosystems in both intact and managed conditions is a high priority for research, simply because it drives or affects most ecological processes. For example, through shading, riparian forests largely control primary productivity in
†Department of Mathematical Sciences ‡School of Forestry and Wood Products, Michigan Technological University, Houghton, MI 49931, USA §School of Fisheries, Box 357980, University of Washington, Seattle, WA 98195, USA ∗ Corresponding author Received 14 May 1997; accepted 25 April 1998
1998 Academic Press
310
J. Dong et al.
the stream. At higher latitudes, shading effects are also important in determining water temperature and its variability over time, which is crucial for fish survival because of specific requirements for thermal ranges (Magnuson et al., 1979). In addition, the abundant amphibians in terrestrial portions of stream ecosystems (i.e. stream sides, including riparian areas) may largely depend on a cool and humid environment to keep their skins wet and protect them from loss of water through evaporation (see Kelsey and West, 1998). For plants, a cool and relatively stable environment has been found to be highly correlated with species richness and diversity at the landscape level (Xu et al., 1997). Previous studies of riparian microclimate and its responses to harvesting are mostly limited to monitoring stream water temperature and the amount of light penetrating through the open strips of forest canopies (Brown, 1969; Brown and Krygier, 1970; Barton et al., 1985; Beschta and Taylor, 1988; Larson and Larson, 1996). The only study of which the authors are aware that describes microclimatic conditions near small streams was completed in western Washington, USA, by Brosofske et al. (1997). While this descriptive work provided synthetic summaries of microclimate (including air, soil and water temperature, light, moisture and wind speed) and its responses to adjacent harvesting, predictive models were not developed. Nevertheless, such models are needed by managers and ecological investigators for understanding the mechanisms of stream ecosystems and development of proper management strategies.
Stream
The objective of this paper is to develop empirical models and quantitatively describe air temperature responses to harvesting using the data reported by Brosofske et al. (1997). More specifically, the authors will: (1) quantitatively describe the changes in pre- and post-harvest air temperature with distance from the stream; (2) quantify the variability of air temperature at the stream and riparian buffer edges as affected by adjacent harvesting; (3) model the seasonal dynamics of stream and riparian temperatures and their relationships with buffer width; and (4) identify the most important variables influencing air temperature at the stream.
Methods Study sites Sampling sites were located in the foothills of the Cascade Mountain Range in western Washington, with elevations ranging from about 150–600 m. Overstory vegetation consisted primarily of Douglas-fir [Pseudotsuga menziesii (Mirb.) Franco], western hemlock [Tsuga heterophylla (Raf.) Sarg.], red alder [Alnus rubra Bong.], western red cedar [Thuja plicata Donn] and grand fir [Abies grandis (Dougl.) Lindl.]. Sampled streams were of similar size (2–4 m width) and had overstory canopy coverage of about 70–80%. Twenty transects were sampled across 15 stream sites (Table 1). In the summer of 1993, five transects were sampled across unharvested (forested) streams, and four were
Distance from the buffer (m) 0
15
30
S1 S0 S2 S3 S4
60 S5
180 S6
Buffer edge
Figure 1. Experimental design for sampling air temperatures along gradients from small streams to the uplands in western Washington. Three temperature measurements were repeated at each location (S0–S6). One station was installed in an adjacent upland forest (S7) to provide references for comparison.
Air temperatures across buffered streams Table 1. Characteristics of sites examined in this paper, where B and A indicates pre-harvest (before) and post-harvest (after) conditions, respectively Stream
Sampling period M/D/Y (Julian day)
Structure
1B 1A 2B 2A 3B 3A 4B 4A 5B 5A 6 7 8 9 10 11 12 13 14 15
7/28–8/5/93 (209–217) 6/23–6/30/94 (174–181) 8/5–8/16/93 (217–228) 6/30–7/7/94 (181–188) 8/17–8/27/93 (229–239) 7/7–7/14/94 (188–195) 8/27–9/6/93 (239–249) 7/14–7/21/94 (195–202) 9/8–9/17/93 (251–260) 6/7–6/16/94 (158–167) 6/12–6/28/93 (163–179) 6/28–7/7/93 (179–188) 7/7–7/15/93 (188–196) 7/16–7/27/93 (197–208) 6/16–6/23/94 (167–174) 7/22–7/28/94 (203–209) 7/28–8/4/94 (209–216) 8/4–8/11/94 (216–223) 8/11–8/18/94 (223–230) 8/18–8/25/94 (230–237)
Forested cc (93/94) Forested cc (93/94) Forested cc (93/94) Forested cc (93/94) Forested cc (93/94) cc (90) cc (90) cc (90) cc (90) cc (93/94) cc (90/91) cc (91/92) cc (93/94) cc (93/94) cc (91/92)
Buffer width (m) 20 23 20 17 16 25 16 60 32 72 12 14 14 22 44 0 (25) 10 13 14 12
Opposite B. width (m) 20 17 20 23 16 60 16 25 44 44 12 14 14 23 72 0 (60) 22 7 11 55
Dominant overstory species
Avg. dom. tree height (m)
DF
32
DF
32
DF
48
DF
48
DF, RA, WH
26
WH DF, RA, WH DF, RA, WH WH, GF DF, RA, WH WH DF, WH DF, RA, WH DF, RA, WH WH
33 33 33 41 26 60 39 41 41 38
Buffer widths for pre-harvest sites were obtained from harvest plans and do not indicate an actual buffer edge at that point; those for post-harvest gradients were actual buffer widths. The number in parentheses is the distance from the stream at which the weather station was placed. Opposite B. width=the width of the buffer on the slope opposite that used for the gradients transect; cc=clearcut (year site was harvested). DF, Douglas-fir (Pseudotsuga menziesii); RA, red alder (Aluns rubra); WH, western hemlock (Tsuga heterophylla); GF, grand fir (Abies grandis).
sampled across sites which had been harvested and planted in 1990–91 where variouswidth buffers were left intact along the streams. The following winter, the five forested sites were harvested, leaving buffers intact. In 1994, these five sites were resampled and six additional sites representing 1–4 year old clearcuts were also monitored. Widths of buffers ranged from 16–72 m. One site (transect 11) was accidentally burned and consequently resulted in an almost unbuffered stream at the time of sampling.
Data collection A transect was established perpendicular to the stream at each sampling site (Figure 1). Microclimate monitoring stations were placed in the center of the stream (S0), at the edge of the clearcut/riparian buffer on the opposite side of the stream (S1), at the clearcut/buffer edge on the slope being sampled (S2) and at 15, 30 and 60 m from the buffer
edge (S3–S5). The locations of the buffer stations at pre-harvest sites were obtained from harvest plans and do not indicate an actual clearcut edge at that point. At harvested sites, buffer stations were placed at the physical clearcut/buffer edge. In addition, one reference station was established in upland interior forest (S6). Three replications of air temperature were measured at each weather station using custom-built 24 gauge E-type thermocouples (two replications) and a 207 Phys-Chem Temperature and RH probe (Campbell Scientific Inc., Logan, UT, USA). All measurements were taken 2 m above the ground surface. One replication of air temperature was taken in the middle of the transect, and the other two were taken 15 m on each side of the transect, parallel to the stream (Figure 1). Temperatures were sampled every 15 s and averaged every 30 min for final storage with 21X and CR21 Microloggers (Campbell Scientific, Inc., Logan, UT, USA). Measurements were taken for 6–15 days at each site, then
311
312
J. Dong et al.
were moved to another location to repeat the data collection process.
Statistical analysis Air temperature gradients and changes over time across the above 20 transects cannot be compared directly because data collection was independently completed during different periods of the season from June to August in 2 years. Since the major interest in this study is to quantify the effects of harvesting on the stream environment, air temperature at the stream (S0) was used as the reference temperature (T0) for standardizing the temperatures collected along each transect (Ti ) by subtracting T0 from Ti . The standardized temperatures (DTi=Ti−T0) from different streams were further analysed to evaluate the importance of seasonal changes and buffer width. The changes in daily average, maximum and minimum of Ti were calculated for each station to describe the gradients from the stream to the upland. The least squares method was used to estimate the slope (b1) and intercept (b0) of a simple linear regression model (DTi=b0+b1∗ distance) for changes in DTi with distance from the stream, separated for pre- and post-harvest conditions. A positive b1 value suggests temperature increases with distance from the stream. Changes in b0 and b1 after harvesting indicate the direction and levels of impact of adjacent harvesting on temperature gradients. Reduced b0 and b1 after harvest suggests that stream air temperature has increased or temperatures along the transect have decreased. Daily maximum, minimum and average temperatures were independently calculated as: min (Ti ), max(Ti ) and (Ti /48) for each station and day, where Ti represents the 30 min averages throughout the 24 h period (i.e. i=1, . . ., 48) (note: these calculations are different from numbers reported in weather forecasts; see Linacre, 1992). Five streams were sampled prior to and after harvesting in 1993 and 1994 to directly assess the effects of harvesting on stream temperatures as a function of various buffer widths. A major goal of leaving wider buffers is to provide a stream environment similar to the pre-harvest condition. Therefore, a reference measurement is required to gauge any changes that occur due to harvesting (Chen
et al., 1993). The temperature collected in a nearby upland forest (i.e. S6) was used as the base to calculate the relative temperatures at the stream and buffer edges for each of the five streams. Daily averages were used to examine effects of harvesting on stream (S0) and riparian (S1 and S2) temperatures. The effects of management on temperature variabilities across riparian systems is critical information. To assess these effects for temperatures collected at different periods of the season over 2 years, the daily coefficient of variance (CV) was calculated for each station as: SD Ti 48
CV=
]
(1)
where SD is the standard deviation of diurnal temperature, and RTi /48 is the daily average temperature. The coefficient of variance provides a measurement of daily temperature variation relative to the daily means. Average CV values of each sampling period were independently calculated by location at each of five streams sampled under pre-harvest and post-harvest conditions. It was hypothesized that harvesting would significantly increase temperature variation from the stream to the upland (also see Brosofske et al., 1997). Post-harvest temperatures at the stream and buffer edges, and the differences between them, were calculated to examine the seasonal effect. Three sinusoidal models: E(yi )=a∗cos
A
B
2p(x−c) +d i=1, 2, 3 (2) 365
are used, where yi, i=1, 2, 3, represents postharvest temperatures at the stream, the buffer edges, and the difference between them, respectively, x is the day of the year, and a, c, d are parameters to be estimated from regression analysis. These models can be further expressed as:
A B A B
E(yi )=a∗cos
+a∗sin
2px 2pc ∗cos 365 365
A B A B
2px 2pc ∗sin +d 365 365
(3)
Air temperatures across buffered streams
which can be rewritten in the form of a linear multiple regression as: E(yi )=b0+b1∗x1+b2∗x2
(4)
where b0=d, b1=a∗cos(2pc/365), b2= a∗sin(2pc/356), x1=cos(2px/365), and x2= sin(2px/365). Using estimated b0, b1, and b2, the coefficients a, c and d are: b1
a= cos
c=
A B 2pc 365
AB
365 b ∗tan−1 2 2p b1
d=b0.
(5)
Information on air temperature at the stream is frequently demanded for many reasons, including predicting water temperature, which is correlated to organism activities. However, air temperature at a stream is seldom known because weather stations are not usually installed at the site. Temperatures collected along the transect in both 1993 and 1994 (i.e. T1–T6) were used as independent variables in a multiple regression analysis to predict the temperature of a stream. A stepwise regression procedure selected the most important independent variables. The coefficient of determination (R2) was used to judge how well the model fit the data.
Results The mean, maximum and minimum standardized temperature gradients were clearly different in pre-harvest and post-harvest conditions (Figure 2). Prior to harvest, the average and minimum temperatures increased with distance from the stream to upland, but decreased after the harvest [Figure 2(a–b)]. However, the stream had the lowest maximum temperature both before and after harvesting [Figure 2(c)]. These changes in temperature patterns were further supported by the estimated regression coefficients (Table 2): both the intercepts and the slopes of the regression functions (for mean, minimum and maximum temperature) for post-harvest
conditions were smaller than those for preharvest conditions. In addition, the daily minimum temperature variation in post-harvest conditions was higher than that in preharvest conditions. This relationship was not observed for daily average and maximum temperatures. Since the temperature at the stream was used in the data transformation, the above results can be interpreted to mean that havesting had either raised stream air temperature or that upland temperature had decreased, or a combination of both effects. Using temperature recorded in an adjacent upland forest interior as a reference, it was determined that temperatures at and near the stream increased greatly after harvesting at all five streams (Figure 3). Temperatures at the stream in pre-harvest conditions were about 2–4°C lower than in the interior forest; temperatures in the riparian areas (i.e. predefined buffer edges) were higher than at the stream but still lower than in the forest. With harvesting, temperatures at all five streams became greater than that in the forest interior. Riparian temperatures were also generally increased by harvesting but remained similar to pre-harvest levels on some days during the sampling periods. Temperature variability from the stream to the upland also was modified by harvesting, as evidenced by comparing the daily coefficient of variation between pre- and postharvest conditions (Figure 4). Temperature variability was much higher after the harvest, increasing with distance from the stream, but stabilizing at about 30–40 m from the buffer for four of five sites. Topography appeared to influence the pattern at Stream 5, with a unique flat terrace occurring just after the buffer edge, unlike the other four streams. In contrast, changes with distance for pre-harvest conditions were small and consistent along the transect. These results suggest that both stream and riparian air temperatures and variabilities were increased by harvesting the adjacent forest, even though riparian forests were partially preserved. The seasonal patterns of air temperature at the stream and buffer edges after harvest were not pronounced [Figure 5(a–b)], but the difference between them did show a seasonal pattern [Figure 5(c)]. During the middle of the growing season, air temperatures at the buffer edges were similar and fluctuated
313
314
J. Dong et al. (a) 4
2
0
–2
–4 0 6
50
100
50
100
(b)
4 2 0 –2 –4 –6 0 (c) 6
2
–2
–6 0
50 100 Distance from the stream (m)
Figure 2. Changes in: (a) average, (b) minimum and (c) maximum temperature differences (°C) between the stream and other sampling locations of pre-harvest (solid circles) and post-harvest (open circles) streams.
around an average of 11°C, suggesting that riparian buffers provided little protection for air temperature at the stream. Early and late in the season, however, mean air temperatures at the stream were about 3·8°C lower than at the buffer edges, suggesting
effects of streams on terrestrial air temperature. Regression models based on these data are presented in Table 3. A value of c= 18·6 indicates that the seasonal pattern of temperature differences peaked at 18·6 days after Julian day 182, or on July 19 [Figure
Air temperatures across buffered streams Table 2. Least squares estimates of the coefficients of the simple linear regression functions for the minimum, maximum and average standardized by stream air temperatures from both pre-harvest and post-harvest measurements Estimated regression coefficients
b0 b1
Minimum
Maximum
Average
Pre-
Post-
Pre-
Post-
Pre-
Post-
0·9724 0·0145
−0·9299 −0·0039
2·2659 0·0142
−1·1964 0·0133
1·4422 0·0141
−1·9161 0·0035
The distance from the stream (m) was used as an independent variable in the analysis. All of the estimates are significant at a=0·05 level.
5(c)], when temperature differences reached their minimum (i.e. differences were >0). Overall, riparian buffers reduced air temperature at the stream by about 1°C. However, there was very little difference between streams buffered with different riparian forest widths (Figure 6). Although wider buffers (i.e. 44 and 60 m) began to increase temperature differences, larger temperature differences (>4°C) were more related to the time of season (i.e. in June and late August). Regression models were developed to predict the air temperature at the stream (Table 4). The better predictors were those temperatures measured near the stream (T1–T2) or in the nearby interior forest (T6). Predictability seemed to be higher for pre-harvest streams (R2>98%) than for harvested streams. Using a simple linear regression model (i.e. only one independent variable), air temperatures on 18 streams could be predicted with R2>95%; the temperature at the two remaining streams could be predicted with R2>81%. With two independent variables, temperatures on 18 streams could be predicted with R2≅99%, and the other two with R2≅87–89%. In terms of the amount of effort involved, a simple regression model is probably best for predicting air temperatures at the stream.
Discussion With recognition of the importance of riparian areas, protecting riparian vegetation during harvesting has become a widely applied practice in forest management (Warner and Hendrix, 1984). Buffer widths of about 7–90 m are arbitrarily applied without solid scientific
evidence showing the efficiency and sufficiency of these buffers (FEMAT, 1993). As pointed out by Brosofske et al. (1997), a minimum 45 m buffer on both sides of the five streams is necessary to reduce the effects of harvesting on soil temperature and moisture. However, these results indicate that even a 70 m forest buffer did not protect against an increase in air temperature (Figure 6). In a study on microclimatic changes caused by clearcutting in the uplands of this region, Chen et al. (1995) found that edge effects on air temperature can be detected at >180 m inside the forest. Although the depth of edge influences may be smaller near riparian areas because of protection from surrounding slopes, edge effects across riparian buffers are likely to extend beyond 70 m from the clearcut edges (Figure 5, Brosofske et al., 1997). In addition, riparian buffers suffer edge effects from both sides, suggesting that the degree of edge influence is higher than that across forest-clearcut boundaries. Air temperature at the stream was raised by 2–4°C after harvesting (Figure 3). These increases are almost the same magnitude as predicted for global temperature increases in the next 50 years, except that the change happens within a much shorter period of time (i.e. usually one winter). These changes may directly affect organism activities, such as the use of cooler riparian areas by wildlife to escape from summer heat, or the diurnal and seasonal movement of herpetofauna between water and land, which is permitted by the lower and more stable temperatures (Kelsey and West, 1998). For example, Hartman et al. (1984) found that small temperature increases in the stream can have significant effects on the life history of young salmon. In
315
J. Dong et al. 2
(a) Stream 1
0
–2
–4
2
(b) Stream 2
0
–2 Air temperature difference from the forest (° C)
316
–4
2
(c) Stream 3
1 0 –1 –2 –3 –4 2
(d) Stream 4
0
–2
–4
2
(e) Stream 5
0
–2
–4
S1
S0
S2
Figure 3. Pre-harvest (solid lines) and post-harvest (dash lines) air temperature differences between the interior forest reference (S6) and the stream (S0) and buffer edges (S1, S2) at five sampled streams in western Washington in 1993 and 1994.
addition, temperature increases will cause significant changes in other physical variables, such as evaporation from the stream,
transpiration of riparian plants, relative humidity, vapour pressure deficit and the energy budget.
Air temperatures across buffered streams 30 (a) Stream 1 28 26 24 22 20 22
(b) Stream 2
20 18 16 14 12 10 (c) Stream 3
Average CV
35 30 25 20 15 35
(d) Stream 4
30 25 20 15 32
(e) Stream 5
30 28 26 24 22 20 0
20
40
60
80
100 120 140
Distance from the buffer (m)
Figure 4. Changes in pre-harvest and post-harvest temperature variation, measured by average coefficient of variation (CV), from small streams to the uplands in western Washington.
Empirical equations provided by Campbell (1977) can be used to demonstrate the importance of these changes. Brosofske et al. (1997) reported that average air temperature on the stream ranged between 10·8 and 14·4°C in pre-harvest conditions, and relative humidity
ranged from 93·8 to 97·2%. To simulate the effects of temperature on relative humidity and vapour pressure deficit, an average 94% relative humidity was used and temperatures of 10, 12, 14 and 16°C to compute the changes in relative humidity and vapour pressure deficit caused by temperature increases from 0 to 4°C. As illustrated in Figure 7, relative humidity exponentially decreased from 94 to about 77% (i.e. 18% reduction) with a 4°C temperature increase regardless of initial temperature [Figure 7(a)]. Conversely, vapour pressure deficit exponentially increased from 0·64 to >5 g m−3, depending on the initial temperature. While Brosofske et al. (1997) did not quantify the vapour pressure deficit related to riparian management, their data indicate that relative humidities after harvesting were 2·5–13·8% lower than in preharvest conditions. The differences between our theoretical prediction [Figure 7(a)] and field results of Brosofske et al. (1997) further suggest that absolute vapour density at the stream was probably increased through evaporation of stream water due to increased air temperature, suggesting that more water is lost from streams to the air after harvesting. These losses, along with increased transpiration rates from riparian vegetation due to high temperature, may be significant in affecting the hydrological budget of the watershed through cumulative effects across the landscape. From a biological point of view, increased air temperatures and vapour deficit may damage the abundant herpetofauna, bryophytes and lichens in riparian areas, since these organisms rely largely on vapourized water during their life cycles. In addition, timber harvesting also increased the temperature variability at and near the streams (Figures 2–4). While a heterogeneous environment is believed by conservation biologists to maintain multiple habitats, its fluctuation may exceed certain organisms’ adaptive range and cause mortality and reductions in population size or abundance. Xu et al. (1997) found these negative effects to be especially true at intermediate spatial scales (≈900 m). Clearly, process-based research (e.g. physiological explorations of organism responses to a variable environment near streams) is needed to mechanically explain the above hypotheses.
317
J. Dong et al.
30
(a)
25 20 15 10 5 160
25 Air temperature (° C)
318
180
200
220
240
180
200
220
240
220
240
(b)
20
10
0 160
4
(c)
2 0 –2 –4 160
180
200 Julian Day
Figure 5. Seasonal temperature changes in: (a) the stream, (b) the buffer edge and (c) the difference between stream and buffer edge. Data were collected in the summers of 1993 and 1994. A sinusoidal model, estimated by the least squares method, was developed for each change (see Table 3 for estimated parameters and interpretations).
Table 3. Least squares estimates of a, c and d in the model T=a(COS[(p(day−c)/365)+d] for quantifying the seasonal dynamics of stream and buffer air temperatures (see Figure 5) Location Buffer edge T2 Stream T0 Difference (T2−T0)
a
c
d
−2·49 −17·05 −15·05
59·3 23·8 18·6
11·37 −3·32 −14·69
Several cautions should be noted regarding the authors’ statistical results. First, empirical models developed through this study
are based on data collected in western Washington where summers are dominated by warm and dry days (i.e. typical Mediterranean climate). The models are not appropriate for other times of the year or for different geological regions, even though forest harvesting may cause increases in temperature and its variability in other landscapes as well (also see Brown and Krygier, 1970; Bescha et al., 1987). Long-term monitoring programmes on stream environments subjected to various experimental buffers are desirable in the future. Further, selection of buffer width should not
Air temperatures across buffered streams 0
Mean difference (C)
–1
–2
–3
–4 Day 158–167
Day 223–237 –5
5
15
25
35 45 Buffer width (m)
55
65
75
Figure 6. Changes in daily average temperature differences between the buffer edge and the stream with buffer width. The differences were significantly higher in the early and late parts of the sampling season, 7–16 June and 18–25 August (open circles), in 1993 and 1994.
Table 4. Coefficients of multiple determinations (R2) from stepwise regression analysis in predicting stream air temperatures (T0) using an automatic search method for selecting the first two potential regressors, i.e. air temperatures sampled along the transects from the stream to the upland Transect 1B 1A 2B 2A 3B 3A 4B 4A 5B 5A 6 7 8 9 10 11 12 13 14 15
First regressor
R2
Second regressor
R2
T2 T6 T2 T2 T3 T2 T2 T6 T6 T6 T2 T1 T1 T1 T6 T2 T2 T1 T1 T1
0·9972 0·8147 0·9982 0·9824 0·9883 0·9684 0·9971 0·9870 0·9841 0·9857 0·9951 0·9901 0·9923 0·9825 0·8810 0·9907 0·9888 0·9935 0·9955 0·9940
T7 T2 T6 T6 T4 T2 T3 T1 T5 T2 T1 T2 T2 T2 T1 T7 T1 T2 T2 T6
0·9976 0·8663 0·9983 0·9902 0·9901 0·9733 0·9986 0·9926 0·9851 0·9931 0·9965 0·9939 0·9960 0·9864 0·8899 0·9942 0·9947 0·9983 0·9996 0·9970
See Table 1 for stream characteristics and Figure 1 for sampling locations.
be solely based on temperature gradients. Because healthy ecosystem functioning relies on interactions between multiple factors, information on other characteristics of ecosystems, especially biological responses, are
needed before a final conclusion can be reached. The value of riparian buffers of arbitrary widths such as are often used in current management practices needs to be re-
319
320
J. Dong et al. (a) Relative humidity (%)
costly because of economic competition for the high quantity and quality of merchantable trees next to the stream. An integrated economic, ecological and social assessment is necessary when designing and maintaining buffer strips (Warner and Hendrix, 1984; Franklin, 1992; Berg, 1995).
100 95 90 85 80
Acknowledgements
75 70 10
12
14
16
18
20
14 16 18 Air temperature (C)
20
(b) Vapour deficit (g m–3) 6 5 4 3 2 1 0 10
12
Figure 7. Effects of increased air temperature (0–4°C) on (a) relative humidity and (b) vapour deficit, with initial relative humidity at 94% and four different temperatures of 10, 12, 14 and 16°C. The vapor pressure density was held as constant during the calculations.
examined. Early studies found that species richness, productivity, biomass, water quality and other ecosystem measurements were significantly higher at buffered streams (Warner and Hendrix, 1984; Kelsey and West, 1998). Less is known about the degree of protection achieved when wider buffers are used. If a 70 m wide buffer cannot protect the stream environment from adjacent harvesting, how wide do buffers need to be? While a final answer may require long-term studies across different types of landscapes, relevant management measures can be readily taken in the interim, such as protecting understory vegetation near the buffer edges to reduce effects from adjacent clearcutting, practising green-tree retention techniques to minimize the extremes in the open area, promoting regeneration near the edges and creating feathered edges (Chen et al., 1995). Finally, efforts to preserve riparian vegetation are
This study was supported by the USDA Forest Service Pacific Northwest Research Station through a co-operative agreement with Jiquan Chen at the Michigan Technological University (PNW-94-0520). Jerry Franklin participated in the original experimental design; Steve West assisted in site selection; Lucy Krakowiak, Kara Otto and Jon Fosgitt helped in field data collection. The authors thank Brian Palik and Doris Gerdes for their comments on the earlier drafts of the manuscript. The authors would also like to thank the two anonymous reviewers for their helpful remarks which improved the presentation of the manuscript.
References Barton, D. R., Taylor, W. D. and Biette, R. M. (1985). Dimensions of riparian buffer strips required to maintain trout habitat in southern Ontario streams. North American Journal of Fisheries Management 5, 364–378. Berg, D. R. (1995). Riparian silvicultural system design and assessment in the Pacific Northwest Cascade Mountains, USA. Ecological Applications 5, 87–96. Beschta, R. L. and Taylor, R. L. (1988). Stream temperature increases and land use in a forested Oregon watershed. Water Resources Bulletin 24, 19–25. Beschta, R. L., Bilby, R. E., Brown, G. W., Holtby, L. B. and Hofstra, T. D. (1987). Stream temperature and aquatic habitat: fisheries and forestry interactions. In Streamside Management: Fisheries and Forestry Interactions (E. O. Salo and T. W. Cundy, eds), pp. 191–232. University of Washington, Seattle, WA: Contribution No. 57, Institute of Forest Resources. Brosofske, K. D., Chen, J., Naiman, R. J. and Franklin, J. F. (1997). Effects of harvesting on microclimatic gradients from small streams to uplands in western Washington. Ecological Applications (in press). Brown, G. W. (1969). Predicting temperatures of small streams. Water Resources Research 5, 68– 75. Brown, G. W. and Krygier, J. T. (1970). Effects of clear-cutting on stream temperature. Water Resources Research 6, 1133–1139.
Air temperatures across buffered streams Campbell, G. S. (1977). An Introduction to Environmental Biophysics. New York, USA: Springer-Verlag. Chen, J., Franklin, J. F. and Spies, T. A. (1993). An empirical model for predicting diurnal air temperature gradients from edge into oldgrowth Douglas-fir forest. Ecological Modeling 67, 179–198. Chen, J., Franklin, J. F. and Spies, T. A. (1995). Growing season microclimatic gradients from clearcut edges into old-growth Douglas-fir forests. Ecological Applications 5, 74–86. Forest Ecosystem Management Team (FEMAT) (1993). Aquatic ecosystem assessment. In Forest Ecosystem Management: An Ecological, Economic, and Social Assessment. Washington, DC, USA: Report of the Forest Ecosystem Management Assessment Team, USDA Forest Service and others. Franklin, J. F. (1992). Forest stewardship in an ecological age. The 9th Annual C. E. Fransworth Memorial Lecture. Syracuse, New York: Faculty of Forestry, Miscellaneous Publication No. 27. SUNY. Gregory, S. V., Swanson, F. J., McKee, W. A. and Cummins, K. W. (1991). An ecosystem perspective of riparian zones. BioScience 41, 540– 551. Hartman, G. F., Holtby, L. B. and Scrivener, J. C. (1984). Some effects on natural and loggingrelated winter stream temperature changes on the early life history of Coho salmon (Oncohynchus kisuch) in Carnation Creek, British Columbia. In Fish and Wildlife Relationships in Old-growth Forests (W. R. Meehan et al., eds), pp. 141–149. Juneau, AL: American Institute of Fishery Research.
Kelsey, K. and West, S. (1998). Wildlife. In River Ecology and Management: Lessons from the Pacific Coastal Ecoregion (R. J. Naiman and R. E. Bilby, eds), pp. 231–254. New York, USA: Springer-Verlag. Larson, L. L. and Larson, S. L. (1996). Riparian shade and stream temperature: a perspective. Rangelands 18, 149–152. Linacre, E. (1992). Climate Data and Resources: a Reference and Guide. New York, USA: Routledge. Magnuson, J. J., Crowder, L. B. and Medvick, P. A. (1979). Temperature as an ecological resource. American Zoologist 19, 331–343. Naiman, R. J. (ed.) (1992). Watershed Management: Balancing Sustainability and Environmental Change. New York, USA: SpringerVerlag. Naiman, R. J. and De´camp, H. (eds) (1990). The Ecology and Management of Aquatic-terrestrial Ecotones. Paris: UNESCO and Carnforth, UK: Parthenon Publishing Group. Naiman, R. J. and De´camp, H. (1997). The ecology of interfaces—riparian zones. Annual Reviews of Ecology and Systematics 28, 621–658. Naiman, R. J., Fetherston, K. L., McKay, S. and Chen, J. (1998). Riparian forests. In River Ecology and Management: Lessons from the Pacific Coastal Ecoregion (R. J. Naiman and R. E. Bilby, eds), pp. 289–323. New York, USA: Springer-Verlag. Warner, R. E. and Hendrix, K. M. (1984). California Riparian Systems: Ecology, Conservation, and Productive Management. Berkeley, CA: University of California Press. Xu, M., Chen, J. and Brookshire, B. L. (1997). Temperature and its variability in the oak forests in southeast Missouri Ozarks. Climate Research 8, 209–223.
321
