- Top of page
Riparian zones are among the most diverse and functionally important ecotones on earth (Naiman et al. 1998). Riparian vegetation modifies light and temperature regimes, provides food for aquatic and terrestrial consumers and is the source of woody debris to streams (Pollock 1998). The removal of riparian trees along streams or lakes can affect aquatic ecosystem structure and function (Platts & Megahan 1975; Vouri & Joensuu 1996). For example, forest harvest can increase sediment delivery to streams (Chamberlin, Harr & Everest 1991), covering stream substrates (Davies & Nelson 1994) and negatively affecting some stream organisms (Osmundson et al. 2002). In small, headwater streams, one of the major microclimatic changes resulting from logging is an increase in solar energy reaching the stream surface (Brosofske et al. 1997). Increased solar energy can affect a host of factors such as water temperature (Beschta 1997), primary production (Hill, Ryon & Schilling 1995) and insect abundance (Fuller, Roelofs & Fry 1986). As far as we know, however, there have been no experimental studies linking changes in microclimatic gradients, as a result of different forest management strategies, to process-based studies on stream organisms.
A management practice designed to minimize the impacts of forest harvest on aquatic systems, especially for resources such as water quality and fish habitat, is to leave a strip of trees (riparian buffers or reserves) adjacent to the water body. The required width of this buffer depends on many factors, including the management objective. If the objective is to protect terrestrial or aquatic vertebrates, buffer zones need to be wider than if the objective is to protect water quality (Castelle, Johnson & Conolly 1994). For instance, Semlitsch (1998) recommended a 164-m wide buffer around wetlands for an assemblage of pond-breeding amphibians, whereas a 30-m wide forest buffer may be sufficient to remove excess nitrate from groundwater (Pinay & Dècamps 1988).
Despite the widespread use of riparian buffers as a management tool to maintain a variety of ecological functions (FEMAT 1993), there has been little experimental evaluation of how effective these buffers are. Watershed-scale experiments are critical for evaluating the functional significance of riparian buffers of different widths, but replicated experiments at this scale are logistically difficult (except see Darveau et al. 1995). As a result, the effects of riparian buffers on streams is known mostly from short-term (one or two seasons of sampling), post-hoc observational studies (Newbold, Erman & Roby 1980; Murphy et al. 1986; Davies & Nelson 1994) and unreplicated experimental watershed studies (Hall, Brown & Lantz 1987; Hartman et al. 1987). Therefore, there is a need for replicated experiments to infer causation in different riparian management approaches and to separate site-specific differences from treatment effects.
Small, headwater (first- and second-order channels; Strahler 1957) streams can account for 70–80% of a total watershed area (Leopold, Wolman & Miller 1964; Gomi, Sidle & Richardson 2002) and they supply water, organic matter, sediment and nutrients to downstream fish-bearing channels (Kiffney, Richardson & Feller 2000; Wipfli & Gregovich 2002; Volk, Kiffney & Edmonds 2003). Not only do these channels provide ecological services to the downstream network, such as the sequestration of nitrogen (Peterson et al. 2001), but they are also inhabited by fauna not found in other portions of the river network. For example, the tailed frog Ascaphus truei Stejneger and the coastal giant salamander Dicamptodon tenebrous Good are specifically adapted to the physically demanding conditions of coastal headwater streams of the Pacific north-west. Despite the high density of headwater streams, their importance to downstream channels, unique fauna and their potential importance as fish habitat (Brown & Hartman 1988), they receive little protection in the form of riparian reserves in the Pacific north-west (Young 2000) or elsewhere (Meyer & Wallace 2001).
In 1996, a riparian management experiment was initiated at the University of British Columbia's Malcolm Knapp Research Forest (MKRF), Canada. The purpose was to evaluate the effects of riparian buffer width on headwater streams using a large-scale, replicated experiment. These riparian manipulations included clear-cutting to the stream edge, 10-m and 30-m wide riparian buffers and uncut controls. In this paper, we present data from two concurrent studies that examined the response of periphyton and primary consumers to this riparian gradient. We measured surface-water nutrient concentrations, periphyton biomass and inorganic mass and insect consumer abundance monthly for 1 year before and 1 year after logging in 13 streams. We also present data on surface-water temperature and light regime in all 13 streams after logging. An intensive, short-term colonization study was conducted in the first year after logging, where periphyton biomass, periphyton inorganic mass and insect abundance were measured weekly for 6 weeks, four times a year, in one stream in each treatment.
Abiotic characteristics (e.g. light) that affect stream food webs can change dramatically following clear-cutting of riparian trees (Brosofske et al. 1997). Less well known is how riparian buffer width mediates change in these abiotic factors to regulate the distribution and abundance of periphyton and insect consumers, and whether these effects differ by season. With this in mind, the main questions in our study were as follows. How does riparian buffer width affect periphyton biomass, periphyton composition and insect consumer abundance, as mediated by changes in light, nutrients, and water temperature? Was there a buffer width that had no detectable effect on these response variables? Were the effects of logging on stream communities consistent among seasons?
- Top of page
This study is unique because we examined the effects of a replicated, experimentally created gradient of riparian buffer width on a variety of abiotic and biotic response variables across all seasons. This approach allowed us to develop potential causal links between changes in PAR and water temperature resulting from riparian manipulations with changes in biological communities. Our experimental manipulation of riparian forests showed that these headwater streams were highly sensitive to forest harvest. We found that mean and maximum water temperature, PAR, periphyton AFDM, chlorophyll a, periphyton inorganic mass and chironomid abundance increased as buffer width narrowed. The greatest differences were observed at the clear-cut and 10-m buffer treatments; however, our results also showed that abiotic and biotic attributes were even higher in the 30-m buffer treatment compared with controls during some seasons.
After logging, water temperature and PAR reaching the stream increased as buffer width narrowed. Increased solar flux as riparian buffer width narrowed was a direct result of removal of riparian vegetation, because solar flux is a function of tree height and overstorey canopy cover (Lee 1978). Our observations suggest that additional light penetration comes through the sides of the buffer (Brosofske et al. 1997). Brosofske et al. (1997) also observed a significant relationship between light level and buffer width along small streams of western Washington. Changes in water temperature were probably a result of increased solar radiation, as direct-beam solar radiation is the main driver influencing water temperature (Beschta et al. 1987; Beschta 1997). Support for this relationship was provided by the strong correlation between water temperature and PAR (r = 0·92, n= 13). Other studies have also shown a strong relationship between riparian vegetation cover and water temperature (Holtby 1988). Skelly, Freidenburg & Kiesecker (2002) observed that mean water temperature in open-canopied ponds was 5 °C warmer than close-canopied ponds. Vegetation in the riparian zone helps regulate the microclimate of aquatic–riparian ecosystems, and increased solar energy input can lead to higher maximum water temperature, especially in small streams (Sullivan et al. 1990). Small streams may be particularly sensitive to removal of riparian vegetation and therefore at increased risk from temperature problems relative to large streams, because small streams have low flow rates and high width-to-depth ratios (Welch, Jacoby & May 1998).
In our study, differences among treatments for water temperature and PAR are based on the assumption that these variables were similar among treatments before logging. We assert that this assumption is valid for our study for the following reasons. First, solar flux has been measured in upstream, forested sections of streams that are part of our riparian experiment, and these values were similar to those measured in controls (P. M. Kiffney, unpublished data). Secondly, elevation, aspect, forest age and composition, geology and geographical location were similar among streams, with the major difference being the riparian buffer treatment left along the water's edge. Thirdly, buffers were experimentally manipulated, and there were at least three replicates within each treatment. Because of these considerations, we suggest that after logging differences in PAR and water temperature can be attributed to riparian buffer width.
Although the actual manipulation of riparian buffer width constituted a disturbance, we suggest impacts associated with these activities were minimal and transient compared with changes in microclimate created by the gradient in riparian buffer width (see the Results, Replicated, whole-watershed experiment). First, to minimize any direct impact on stream habitat, trees were felled away from the stream, which prevents impacts on stream habitat. Secondly, logging equipment (e.g. skidders) was never driven through our stream reaches in the process of removing cut logs. Thirdly, no new roads were built to access watersheds for logging; road construction during logging can have a major impact on stream habitat (Ziemer 1981). These actions therefore minimized any direct impacts of forest removal on our study reaches.
We observed biological gradients that mirrored microclimatic gradients, which were related to buffer width. Periphyton biomass increased as light levels and water temperature increased and buffer width narrowed. We found similar patterns in two large-scale (215–650-m long reaches) and long-term (1–2 years) experimental studies, whereas most of the research examining controls on periphyton communities has been conducted at relatively small spatial and short temporal scales (Hillebrand 2002) and primarily under summer, base flow conditions (Feminella & Hawkins 1995). Periphyton biomass was higher in studies comparing clear-cut with unlogged streams (Gregory 1980; Lowe, Golladay & Webster 1986; Kiffney & Bull 2000; but see Shortreed & Stockner 1983) and in open vs. closed canopy reaches within a stream (Hill & Knight 1988; Feminella, Power & Resh 1989). In general, stream periphyton increases as a non-linear function of light due to increases in photosynthetic rate (Hill 1996). Consumers (Hill, Ryon & Schilling 1995), trophic structure (Wootton & Power 1993) and nutrients (Hillebrand 2002) can also be important in controlling algal accrual. We did not manipulate consumer abundance or nutrients, but our data strongly support the hypothesis that light was the primary constraint on accrual of periphyton biomass. Specifically, our analyses showed that PAR was the single best predictor variable for periphyton biomass.
Not only did periphyton biomass increase, so did the abundance of common primary consumers. Although our ability to detect statistical differences among treatments for insects was low compared with other measures, the consistent trends among seasons and between studies suggest that these animals were responding to changes in light and water temperature, mediated by riparian buffer width. For example, summer chironomid abundance was approximately 100–150% higher in the clear-cut, 10-m and 30-m buffer treatments compared with controls in the replicated experiment, and 300–600% higher during the summer colonization study. Mayflies showed similar patterns. Consumer abundance may have been limited by food resources, as periphyton biomass increased with increased light level associated with narrow buffers. The abundance or biomass of primary consumers has been shown to increase (Quinn et al. 1997) as light levels increase, possibly due to increased primary production (Hill, Ryon & Schilling 1995). Others have found that primary consumers did not increase along a light gradient in experimental channels, because predators cropped surplus secondary production (Wootton & Power 1993).
Water temperature can also constrain insect populations (Rempel & Carter 1986) and we found that chironomid abundance was positively related to maximum water temperature. Water temperature has complex effects on life cycles of stream biota (Hogg & Williams 1996). For example, water temperature influences the rate at which eggs develop and juveniles grow, which, in turn, determines voltinism, rates of growth and productivity (Allan 1995). The strong relationship between water temperature and Chironomidae abundance, however, was confounded by the correlation between water temperature and PAR. One of the difficulties with large-scale experiments such as this one is that multiple factors (e.g. light and water temperature) change in response to the manipulation. We addressed this issue with a small-scale channel study (P.M. Kiffney, J.S. Richardson & J.P. Bull, unpublished data). Results from this study showed that chironomids and other primary consumers were indirectly constrained by light level, probably mediated through increased primary production because water temperature was held constant.
Although periphyton and consumer abundance increased as buffer width narrowed, so did periphyton inorganic mass. Forest harvest can influence both upland erosional processes and the way that streams process sediment in channels (Chamberlin, Harr & Everest 1991). We propose an additional mechanism that may help explain the increased sediment levels in periphyton. High light environments primarily support filamentous algal growth forms, while low light environments primarily support diatoms (Hansmann & Phinney 1973; Duncan & Blinn 1989; Wellnitz, Radar & Ward 1996). We observed similar qualitative differences in algal community composition during spring in the four streams of the colonization study (Fig. 8). We hypothesize that filamentous growth forms at the clear-cut and 10-m streams were more efficient at trapping suspended sediment from the water column than the diatom-dominated biofilm of the control and 30-m sites. Evidence for this can be found in the ratio of organic mass to inorganic mass averaged across seasons: a ratio less than one indicates a periphyton community dominated by organic matter, whereas a ratio greater than one indicates a community dominated by inorganic matter. The ratio was less than one in the control and 30-m treatments but increased to 1·7–1·9 in the 10-m and clear-cut treatments. Suren & Jowett (2001) showed that the periphyton mat in untreated stream channels contained less fine sediment and more organic material than channels where fine sediment was added. It is possible that the increase in periphyton inorganic mass was a result of increasing algal biomass. In other words, increases in periphyton inorganic mass may reflect an increase in ash content due to greater algal biomass. We suggest that this was not the case in our study, because algal communities dominated by diatoms will have a higher ash content due to the presence of silica in their cell walls than other algal forms (Nalewajko 1966). The relative abundance of diatoms was higher at control and 30-m buffer sites, while the high light sites were dominated by filamentous algae and diatoms were rare. Therefore, we conclude that the high inorganic content of periphyton in the clear-cut and narrow buffer sites was due to deposition of fine sediment on the stream bottom.
We also speculate that algal growth form and higher periphyton inorganic content accounted for some of the non-linear responses of consumers to riparian buffer width. For example, there was a trend for mayfly and chironomid abundance in spring to be lower in the clear-cut treatment than in the 10-m treatment. Kiffney & Bull (2000) found that insect abundance on tiles was negatively correlated with periphyton inorganic mass on the same tiles, and periphyton inorganic mass was higher in logged streams compared with controls. Results from the long-term replicated and short-term colonization studies provided additional evidence to suggest that some invertebrates (i.e. mayflies) are negatively correlated with periphyton inorganic mass. Stream invertebrates are negatively affected by sediment deposited on the streambed (Suren & Jowett 2001; Zweig & Rabeni 2001) and high suspended sediment loads (Shaw & Richardson 2001). Insect drift was more than double from stream channels treated with fine sediment compared with untreated channels (Suren & Jowett 2001). Large amounts of inorganic sediment in the periphyton mat and the filamentous nature of the algal community may inhibit attachment by some grazers (Kiffney & Bull 2000) or it may decrease the nutritional quality of periphyton (Hawkins & Sedell 1981). Increased sediment delivery to streams due to forest harvest also has negative consequences for fish (Scrivener & Brownlee 1989; Osmundson et al. 2002) and amphibians (Corn & Bury 1989).
management issues and applications
The most surprising results from our study were significant changes in some abiotic and biotic attributes at the widest buffer (30 m), which points out how sensitive these headwater streams are to forest harvest. Forest clearing has been shown to affect a wide variety of taxa, such as bryophytes (Hylander, Jonsson & Nilsson 2002), terrestrial (Watt, Stork & Nigel 2002; Hamer et al. 2003) and aquatic invertebrates (Newbold, Erman & Roby 1980), salamanders (Vesely & McComb 2002), fish (Murphy et al. 1986; Rowe, Smith, Quinn & Boothroyd 2002), birds (Pearson & Manuwal 2001; Williams et al. 2001) and mammals (Law & Chidel 2002; Cockle & Richardson 2003). Some of these studies specifically addressed the relationship between buffer width and biotic abundance and diversity, with most finding that wide buffers (> 30 m) minimized the ecological effects of clear-cut logging on aquatic and terrestrial ecosystems. Results from our replicated, watershed-scale experiment support this conclusion. Other factors besides riparian buffers are important in maintaining ecosystem structure and function when harvesting trees. These include environmental heterogeneity (Hamer et al. 2003), successional trajectories of regenerated forest (Summerville & Crist 2002), corridors between watersheds (Law & Chidel 2002) and partial harvest techniques (Sullivan & Sullivan 2001).
It is important to note that we observed these differences by logging a relatively small proportion of the watershed (c. 20–25% of total area logged). Although we observed higher dissolved nitrate concentrations in water as the buffer width narrowed, these differences were not statistically significant. If more of the watershed was logged, we may have observed larger differences in nitrate levels (Likens et al. 1970). Lewis et al. (2001) found that the average increase in annual storm runoff was higher when more of the watershed was logged. We also emphasize that this portion of the study was concerned with elements of the aquatic food web, and other components of the ecosystem such as fish, mammals and amphibians must be considered when determining the most appropriate strategy for managing forested landscapes. These issues will be addressed in future papers.