Introduction
Streams connect the landscape in the mountainous Pacific Northwest from upstream to down, among plant and animal communities and through geology and time. The character of these streams is impacted by the use of the landscape surrounding them. Among stream quality parameters, temperature has been singled out as one of the most important drivers in stream health due to its biological impacts.
This study evaluates the watershed scale factors that influence the temperature of 11 first- and second-order streams in northwest Washington (Horton 1945). The purposes are to examine whether the watershed landscape condition, especially forest harvest, has an impact on stream temperature and to provide information on the physical environment and heat budget of the study streams.
Water temperature is a determining factor in the distribution, migration, growth and survival of aquatic organisms. In this region, fish, particularly salmon, are of greatest concern to society, due to their role in ecosystems, cultures and economies. Salmon are poikilotherms and are heavily influenced by thermal regimes at all points in their life cycle, including upstream migration, spawning, incubation and freshwater rearing (Bjorn and Reiser 1991). Sub-lethal effects are a more common problem than direct mortality and can ultimately lead to increased mortality. All species of Pacific salmon prefer temperatures between 12 and 14oC with an avoidance of temperatures above 15oC (Brett 1952). Increased water temperatures affect salmonid growth and survival.
Warm waters have been linked to disease outbreaks - notably kidney disease, columnaris, furunculosis and vibriosis - due to metabolic changes and physiological stress (Lantz 1970; Beschta et al. 1987; Reeves et al. 1987; Bjorn and Reiser 1991). Metabolism increases at higher temperatures, resulting in decreased proportion of food ration available for growth (Lantz, 1970). Temperature rises lead to energy intensive physiological changes, such as altered respiratory rates, fluid-electrolyte imbalances and changes in blood alkalinity (Crawshaw 1977). In addition, unsuitably high temperatures result in decreased swimming performance and altered maturation rates depending on the stage of the life cycle (Bjorn and Reiser 1991).
Stream temperature can alter timing of life history events, behavior and structure of the fish community. Warm waters contribute to delays in upstream migration (Bjorn and Reiser 1991), earlier fry emergence and earlier downstream migration of smolts (Holtby 1988). Both competition between steelhead trout (Salmo gairdneri) and redside shiner (Ricardsonius balteatus) and species productivity were impacted by water temperature, with trout dominating cool water and shiner warm (Reeves et al. 1987). Preventing excessive warming is of critical importance to salmon.
Forest management can alter stream temperature regimes; therefore, a major goal of forest practice regulations is to minimize these effects. How this goal can be accomplished has been the focus of many studies since the 1960’s (Brown 1969; Anderson 1973; Beschta et al. 1987; Sullivan et al. 1990; Hatten and Conrad 1995; Brosofske et al. 1997; Macdonald et al. 2003). Standards for water temperature have been set in place by the Washington Department of Ecology with maximum water temperatures of 12, 16 and 17.5oC for aquatic life classes (i) char, (ii) core salmon/trout and (iii) non-core salmon/trout, respectively. These standards are established by section 200 of Chapter 173-201A WAC (Surface Water Quality Standards for the State of Washington 2003). In addition, the United States Federal Water Pollution Control Act Amendments of 1972 require each state to develop Best Management Practices to limit changes in stream temperature resulting from forest harvest. However, questions remain regarding how to best minimize the effects of forest management on stream temperature.
Many studies have evaluated impacts of forest harvest on stream temperatures (Brown and Krygier 1970; Brown 1971; Andersen 1973; Lynch et al. 1984; Holtby 1988; Sinokrot and Stefan 1993; Beschta et al. 1997; Johnson and Jones 2000; Macdonald et al. 2003). In general, these studies emphasize a link between the loss of riparian vegetation and stream temperature increases. Many heat budget studies single out solar radiation as the input of greatest importance to stream temperature (Brown 1969; Brown 1970; Theurer et al. 1985; Sinokrot and Stefan 1993; Webb and Zhang 1997).
The direct shade provided by riparian buffers is considered to be critical for preventing stream warming in areas of forest harvest (Brown 1971; Andersen 1973; Burton and Likens 1973; Theurer et al. 1985; Beschta et al. 1987). Clearcuts in riparian areas have elevated stream temperatures in comparison with unharvested and shaded streams (Brown 1969; Brown and Krygier 1970; Brown 1971; Swift and Messer 1971; Burton and Likens 1973; Feller 1981; Swift 1982; Lynch et al. 1984; Johnson and Jones 2000). These temperature increases can persist for fifteen years after harvest (Beschta and Taylor 1988; Johnson and Jones 2000) and can occur during both summer and winter (Feller 1981). Diel fluctuation in stream temperature can also increase significantly with forest harvest (Swift and Messer 1971; Burton and Likens 1973; Lynch et al. 1984; Garman and Moring 1991; Johnson and Jones 2000). The importance of shading in maintaining stream temperature has resulted in the establishment of standards for riparian buffers at stream edges (WFPB 1996). Castelle et al. (1992) provide a literature review of the efforts to quantify the buffer size necessary to maintain water quality.
In addition to solar radiation, other major factors influencing stream temperature are air temperature, evaporation, convection, advection, including groundwater and precipitation, friction, conduction and outgoing radiation (Figure 1). These inputs and outputs can vary greatly in importance depending on the stream’s location in a watershed, aspect, relief, elevation, watershed size, degree of forest cover, type of bedrock, soil depth and texture, climate and other factors.
Sensible heat exchange with overlying air is a significant component of the stream heat budget (Webb and Zhang 1997). Sensible heat is the heat absorbed or transmitted during a change of temperature without a change of state, i.e. evaporation and condensation. There is no established consensus concerning the degree of importance of air temperature on the stream heat budget (Johnson 2003). As previously mentioned, the major body of research supports solar radiation as the dominant component of the stream heat budget; however, Adams and Sullivan (1989) found air temperature to be more important. Issues of scale may be important in determining the significance of air temperature. Streams tend toward equilibrium with air temperature downstream (Edinger et al. 1968; Sullivan et al. 1990; Mohseni and Stefan 1993). In the western Cascades, equilibrium is reached in 5th order streams when average stream depth is 0.6 meters (Sullivan and Adams 1991). It follows that smaller, high order streams are influenced to a greater degree by environmental variables besides air temperature.
Figure 1: Factors affecting stream temperature in a small watershed. These inputs and outputs can vary greatly in importance depending on the stream’s shading, location in a watershed, aspect, relief, elevation, watershed size, degree of forest cover, type of bedrock, soil depth and texture, climate and other factors. Adapted from Johnson and Jones (2000).
Air temperature records are more readily available than stream temperature measurements and can be used to estimate or predict stream temperature using models based on modified regression analysis (Crisp and Howson 1982; Jeppesen and Iversen 1987; Erickson and Stefan 1996; Webb and Nobilis 1997). Some studies have found the relationship to be non-linear and have incorporated correlation analysis or shading and sheltering to increase accuracy (Stefan and Preud’homme 1993; Bogan et al. 2003).
Crisp and Howson (1982) regressed 5- to 7-day mean stream temperature with air temperature and found that air temperature was responsible for 87 to 98% of the stream temperature variance. Webb and Nobilis (1997) found that air temperature was responsible for over 95% of the monthly mean stream temperature variance, yet concluded that the link between air and water temperatures is unstable, especially in summer, complex and not driven by cause and effect. Instead, they proposed that air and stream temperatures may respond to climate and heat fluxes in a similar way. Johnson (2003) proposes that stream and air temperatures are responding to solar radiation.
Evaporation and convection are driven by the heat exchange between the atmosphere and the stream. In forested streams, evaporation and convection rates are low and contribute little to the energy balance (Brown 1969; Beschta et al. 1987). In addition, the two processes may balance each other (Beschta et al. 1987).
The strength of the relationship between air and water temperatures weakens with increasing proportion of incoming groundwater (Smith and Lavis 1975; Jeppesen and Iversen 1987). Both groundwater and rainfall are advective heat inputs. Heat gains from precipitation, however, are not significant in the stream heat budget (Webb and Zhang 1997). The degree to which groundwater affects a stream’s temperature depends on the amount of groundwater in proportion to the amount of stream water and the degree of difference between the two. Regionally, groundwater approximately 15 m deep is one to two degrees Celsius higher than the mean annual air temperature year round, although local deviations are possible (Todd 1980). Groundwater is therefore a moderating influence on stream temperature by having a warming effect in winter and a cooling effect during summer (Smith and Lavis 1975). Shallow groundwater is more variable than deeper groundwater and will fluctuate in temperature (Todd 1980). The amount of groundwater entering a stream varies depending on the location in the watershed, type of bedrock and other factors. Smith and Lavis (1975) found that groundwater lowered stream temperature by 4 to 5oC over a 300 m stretch during reduced summer flows.
Both advective and conductive processes can comprise heat transfer within a streambed. The two-way exchange of water, and with it, advective heat, between the stream channel and the hyporheic zone can appreciably impact stream temperature, especially in low flow streams (Evans et al.1995; Johnson and Jones 2000; Storey et al. 2003). In addition, it plays a stronger role at forested than cleared sites because the impact of solar radiative fluxes is curtailed (Storey et al. 2003).
The importance of conduction is in part due to a higher rate of heat transfer between water and substrate than between water and air (Johnson and Jones 2000). Conductive heat transfer, like advective, is especially significant in shallow streams (Sinokrot and Stefan 1993). When subsurface and hyporheic flow are present, the stream substrate contacts with more water (Johnson and Jones 2000). The type of substrate and degree of permeability will affect the dynamics of conduction (Evans et al. 1995; Johnson and Jones 2000). Johnson and Jones (2000) found that conduction may be the second most important factor influencing stream temperature after solar radiation when groundwater is absent.
Soil temperature affects the temperature of sub-surface waters and is, in turn, modified by the degree of forest cover (Monteith and Unsworth 1993). Due to more solar radiation reaching the forest floor, soils and alluvial substrates are warmer in clearcuts than in forests. Johnson and Jones (2000) found that surface soil temperatures were 5oC lower and had less diel flux in an old growth stand than in a forest gap. Brosofske et al. (1997) found that soil temperatures up to 180 m distant from the stream have a more significant impact on water temperatures than soil temperature in the riparian buffer. They theorized that a large proportion of their streams’ water was coming from the surrounding landscape. Soil temperatures impact the subsurface water that flows through the soil. In addition, these subsurface waters move more slowly through the upland areas, provided greater exposure to upland soil temperatures, then flow quickly down through the steeper riparian buffer into the channel (Brosofske et al. 1997).
Channel morphology relates to temperature dynamics. The depth and width of a stream are important because the surface area to volume ratio affects the degree of influence of solar radiation, convection and evaporation. Wide and shallow streams will have more temperature variation than narrow and deep streams (Moore and Miner 1997). Stream temperature change is positively related to surface area and inversely related to flow volume (Brown 1970). In addition, riparian buffers shade a larger proportion of narrower streams.
Channel reach pattern can vary from cascade to braided, possessing varying degrees of roughness - resistance to the flow of water - and occurring at different percent slopes (Montgomery and Buffington 1993). A steep channel can result in significant heat inputs due to the friction of water over rock. Webb and Zhang (1997) found that friction contributed 22% on average of the non-advective energy gains in the river heat budget. At the same time, slope will affect the time water spends in a reach, with a steep slope reducing the residence time of water and the responsiveness of stream temperature to heat inputs and outputs. Streamflow is important in a similar way. The higher the discharge, the greater the capacity for heat storage and the less responsive the stream becomes to environmental factors (Smith and Lavis 1975).
Despite the complex factors affecting stream temperature, Pacific Northwest forest management guidelines depend largely on riparian buffers to preserve water quality, but are buffer strips effective in precluding temperature increases in streams? Land use occurs at watershed scale; therefore, its impact on streams cannot be fully understood by studying only the stream and riparian zone (Wang et al. 1997). Landscape processes can override or inhibit local processes in watersheds (Naiman 1992). For example, the microclimate of the riparian zone can be affected by changes in the landscape. Brosofske et al. (1997) found that the microclimate of the riparian zone can be altered by clearcuts even when buffers from 30 to 72 m are retained. When forest harvest occurs away from streams, the buffered riparian zone microclimate becomes more similar to clearcut conditions rather than forest interior values, with increased air temperatures and decreased relative humidity (Brosofske et al. 1997). Recent clearcut edge influence on microclimates can stretch for four to six tree heights from a clearcut, which can extend beyond 400 meters in the Pacific Northwest (Chen et al. 1999).
Richards et al. (1996) states, “the influence of landscape cover characteristics throughout a drainage basin may be as important as riparian vegetation in understanding stream ecosystems.” However, streams are complex and dynamic, complicating the question of how to separate the footprint of the landscape on stream temperature from other important variables such as solar radiation. Logged watersheds with high levels of riparian shade have been found to have significantly warmer water temperatures than unlogged watersheds with similar levels of riparian shade (Hatten and Conrad 1995). Beschta and Taylor (1988), in a 30-year study in the Western Cascades, Oregon, found a significant relationship between maximum stream temperatures and the cumulative effect of forest harvest.