Summary of Findings from Prior Research

Climate Sensitivity
Successional Processes
Thresholds and State Changes

Climate Sensitivity

Alaska warmed rapidly at the end of the last glacial period. After the thermal maximum in the early Holocene (ca 11,000-9000 years ago), climate became wetter and gradually cooler (Lloyd et al. 2006). Deciduous woodland and shrubland dominated in the early Holocene (Edwards et al. 2005), followed subsequently by white spruce, then rather suddenly by black spruce. The expansion of black spruce coincided with a threshold increase in fire frequency 6,000 years ago, despite cooler, moister climate, suggesting that vegetation rather than climate drove long-term trends in fire regime (Lynch et al. 2002, Lloyd et al. 2006) (Fig. 1). Thus the boreal forest has shown both gradual and abrupt climate responses through the Holocene.

Warming during the 20th century has decreased the growth of white spruce (Fig. 2), due to warming-induced drought stress (Jacoby et al. 1999, Barber et al. 2000), with projections of zero net annual growth and perhaps the loss of white spruce and birch from uplands before the end of the 21st century (Lloyd and Fastie 2002, Wilmking 2003, Juday et al. In press). At the southern limit of Alaska's boreal forest, spruce bark beetle outbreaks eliminated extensive areas of forest, because warmer temperatures reduced tree resistance to bark beetles and shortened the life cycle of the beetle from two years to one, shifting the tree-beetle interaction in favor of the insect (Werner and Holsten 1985, Werner and Illman 1994, Wallin and Raffa 2004, Werner et al. 2006, Werner et al. In press). At its altitudinal and latitudinal limits, the boreal forest is expanding into tundra because of high rates of tree recruitment beyond treeline during recent warm decades (Silapaswan et al. 2001, Lloyd and Fastie 2002). At arctic treeline, spruce establishment in tundra depends at least partially on thawing permafrost (Lloyd et al. 2003b). In contrast to the expansion of predominantly white spruce forests at treeline, which occurs independent of disturbance by fire, northward expansion of black spruce may depend on fire (Lloyd et al. 2006). In summary, current trends show gradual expansion of forest into tundra in the north, abrupt decline in the south, and impending major compositional changes in central portions of Alaska's boreal forest, suggesting that the boreal forest is on the cusp of major structural and functional changes.

Many boreal animals exhibit large population fluctuations. Densities of small mammals (Microtus and Clethrionomys) correlate most strongly with climate (Rexstad and Kielland 2006), whereas larger mammals (moose and hares) appear to be more sensitive to food availability and predation (Flora 2002). Two native insects have changed from decadal outbreaks to consistently low populations (large aspen tortrix since 1985; spear-marked blackmoth since 1975), whereas other species that had negligible populations before 1990 have shown large outbreaks (eastern spruce budworm, spruce coneworm, larch sawfly, and aspen leaf miner) (Werner 1994, 1996)(Table 1). Thus, several factors, including climate, influence animal population densities.

Ecosystem processes in Interior Alaska are sensitive both to topographic variation in environment and to successional age. Aboveground production, for example, varies by more than an order of magnitude among forest types (Van Cleve et al. 1983, Yarie and Van Cleve 2006) (Fig. 3). It is greatest in midsuccessional stands on floodplains, where soil temperature and moisture are relatively high, and is constrained on south-facing slopes by drought and on north-facing slopes by soil temperature. On temperature-limited sites, mosses account for 45% of aboveground production (Fig. 4). Temperature constraints on tree production appear mediated largely by nitrogen supply (Yarie 1997, Yarie and Van Cleve 2006).

Low temperature and nitrogen supply promote belowground allocation. Thus, carbon and nutrient cycling rates in fine roots are several orders of magnitude faster than in aboveground tissues (Ruess et al. 1998, Ruess et al. 2003, Vogel et al. 2005). Fine root production is concentrated close to the soil surface (Fig. 5), and there is a progressive increase in fine root production into deeper soil layers as the soil warms through the season. Fine root life span and the associated physiological and morphological traits of roots vary across sites in parallel with patterns observed for aboveground tissues (Ruess et al. 2006). Cross-site studies demonstrate that boreal trees are similar to most woody plants in the morphological, phenological, and physiological traits of first-order roots and differ primarily in root allocation, size distribution, and lifespan (Burton et al. 2002, Pregitzer et al. 2002).

Ratios of aboveground litterfall to soil respiration in interior Alaskan forests are among the lowest recorded in North America (Raich and Nadelhoffer 1989, Ruess et al. 1996), suggesting that a large proportion of boreal soil respiration originates from root-derived C. Fine-root respiration constitutes approximately 60% of soil respiration in black spruce forests (Ruess et al. 2003, Vogel et al. 2005). Trenching to eliminate root production caused a 12% loss of total soil C within 2 years (Fig. 6), suggesting that much of the root-derived soil carbon is labile and can decline rapidly in the absence of root inputs. In summary, biogeochemical processes are quite sensitive to temperature, but many of these effects are mediated by variations in species composition, allocation, and nitrogen supply.

Successional Processes

The major physical disturbances in Interior Alaska are flooding in floodplains, fire in uplands, and water-table changes in wetlands. River discharge and flooding are climatically sensitive, but glacial rivers have maximum discharge in midsummer glacier melt is maximal, whereas clearwater rivers have maximum discharge with spring snowmelt (Fig. 7).

Fire return time varies regionally from <50 years to > 100 years (Yarie 1981, Fastie et al. 2002). Area burned correlates positively with temperature (r = 0.63) and vegetation cover (r = 0.65) and negatively with precipitation (r = -0.61) (Kasischke et al. 2002, Duffy et al. 2005). In association with recent warming, Interior Alaska has experienced a sharp increase in wildfire. Seven of the 11 largest fire years since 1950 occurred since 1988, accounting for half of the cumulative area burned (Fig. 8). In the last two years alone, 4.6 million ha (10% of Interior Alaskan forests) have burned. During large fire years, 36% of the area burns after 1 August, when soils are deeply thawed and well drained, leading to unusually severe fires. Lightning, which accounts for 90% of the area burned (Kasischke et al. 2006), is controlled by both synoptic processes related to El Niño and by local factors such as topography and presence of forest vegetation (Dissing and Verbyla 2003). Human ignitions, which account for 60% of the fires in Alaska, generally produce small fires that occur at times and places where fire does not readily spread (Kasischke et al. 2002, Chapin et al. 2003). Human activities reduce area burned because suppression has greater impact than human ignitions (DeWilde 2003) (Fig. 9).

The trajectory of succession is sensitive to environment, propagule availability, legacies associated with prefire vegetation, and disturbance severity. Primary succession predominates in the active floodplain. After initial establishment, competition, facilitation, and herbivory interact to drive successional change (Walker et al. 1986, Walker and Chapin 1987, Adams 1999). Ecosystem controls change at key turning points (thresholds), where a shift in dominance of plant functional types radically alters the physical and chemical environment that govern ecosystem processes and disturbance probability (Van Cleve et al. 1991). In the floodplain, intense herbivory by moose initially constrains canopy development, creating an ecosystem dominated by physical controls over soil water movement, surface evaporation and gypsum accumulation at the soil surface (Dyrness and Van Cleve 1993, Marion et al. 1993, Van Cleve et al. 1993, Kielland and Bryant 1998). Colonization by alder shifts the system from physical to biological control (Van Cleve et al. 1991, Viereck et al. 1993), adds 60-70% of the nitrogen that accumulates during succession (Van Cleve et al. 1971, Van Cleve et al. 1983, Van Cleve et al. 1993, Uliassi et al. 2000, Uliassi and Ruess 2002), and causes herbivory to change from a deterrent to an accelerator of succession by eliminating palatable early successional species (Bryant and Chapin 1986, Bryant et al. 1991, Kielland et al. 1997, Kielland and Bryant 1998) (Fig. 10). Other key turning points include (1) a shift to balsam poplar dominance, where changes in productive potential and litter chemistry enhance NPP and nitrogen cycling rates (Van Cleve et al. 1983, Schimel et al. 1996) and (2) the shift to white spruce dominance, where mosses grow rapidly in the absence of smothering broadleaved litter (Oechel and Van Cleve 1986), reduce nutrient cycling rates by sequestering nutrients in low-quality litter (Yarie 1997), and increase fire probability by producing resinous fuels that dry quickly (Chapin et al. 2003). In late-successional black spruce stands, root turnover governs nutrient supply.

Secondary succession is the rule in the uplands. Self-replacement, in which the prefire tree species returns to dominance shortly after fire, generally occurs in extreme environments, whereas succession with multiple stages is more common in intermediate sites (Mann and Plug 1999, Fastie et al. 2002, Chapin et al. 2004a). Late-successional conifers establish during the initial 1-2 decades after fire, but their establishment success is sensitive to the depth of the organic mat remaining after fire (Johnstone and Chapin In Press-a), understory species composition (Cater and Chapin 2000), and seed availability from on-site serotinous cones (black spruce) or off-site seed sources (white spruce) (Zasada et al. 1992, Mann and Plug 1999, Cater and Chapin 2000, Johnstone and Chapin 2003). Variations in fire frequency (Johnstone and Chapin In press-b) or severity (Mann and Plug 1999, Johnstone and Chapin In Press-a, Harden et al. Submitted) can alter plant regeneration feedbacks that stabilize community composition and cause rapid shifts in forest cover types (Johnstone and Kasischke 2005). Changes in any of these processes could alter vegetation composition and successional trajectory.

Species diversity is low in the boreal forest (Waide et al. 1999) and varies through succession with peaks in early succession (e.g., fire-specialist plants, herbivorous insects, neotropical migrant birds, and mammals) (Rees and Juday 2002, Werner 2002, Rexstad and Kielland 2006) and late succession (non-vascular plants and saprophagous insects) (Chapin and Danell 2001). Logging reduces plant diversity by 30% by eliminating fire specialists (Rees and Juday 2002). We are beginning to document patterns of microbial diversity using genomics.

N2-fixation inputs by Alnus crispa (uplands) and A. tenuifolia (floodplain) account for the largest percentage of total N accumulated during succession and appear to be strongly limited by soil P availability (Uliassi and Ruess 2002, Anderson et al. 2004) or periodic insect or pathogen attacks (Ruess et al. Submitted). Fixation inputs appear to exceed plant N demand, and significant amounts of fixed N may be lost via leaching or denitrification, particularly in mid-successional stages (Uliassi and Ruess 2002), where nitrification potential is high and soil microbial biomass is more C- than N-limited (Brenner et al. Submitted).

Mammalian herbivores play a key role in the biogeochemistry of the boreal forest. In the floodplain willow communities, they consume 40% of aboveground NPP (Kielland and Bryant 1998). When they are excluded, biogeochemistry changes more quickly from a system dominated by inorganic C cycling and solubility equilibria, to a biologically controlled pattern of cycling dominated by NPP and decomposition (Kielland and Bryant 1998, Ruess et al. 1998).

Despite the large concentrations and rapid cycling of organic N and high proteolytic activity in boreal soils (Fig. 11) (Kielland et al. Submitted-a), the vegetation is strongly N-limited (Yarie and Van Cleve 1996, 2006). However, this organic N is quite dynamic. Amino acids turn over more rapidly than inorganic N (Kielland 2001, Jones and Kielland 2002) and are a major source of N absorbed by both plants and microbes (McFarland et al. 2002) (Fig. 12, 13).

Long-term forest harvest studies permit an assessment of potential future human impacts on Alaska's boreal forest. Low intensity forest harvest (no scarification) reduces initial seedling establishment but maximizes long-term growth of tree seedlings (Wurtz et al. 2006) (Fig. 14). Overstory retention treatments had no long-term effect on tree recruitment and growth (Wurtz and Zasada 2001). These studies suggest that low-intensity management after clear-cutting, an approach that mimics certain aspects of natural fire cycles, may maximize ecological recovery

Thresholds and State Changes

Presence or absence of permafrost is probably the most important threshold regulating the structure and functioning of Alaska's boreal forest. Permafrost is generally present on north-facing slopes and valley bottoms, where it leads to cold water-logged soils, and absent on south-facing slopes, where soils drain freely. In flat upland areas, the presence or absence of permafrost in black spruce forests correlates strongly with the depth of organic layer lying on top of mineral soil (Kasischke and Johnstone 2005). Permafrost temperatures are now typically warmer than 2˚C, and have warmed about 0.7˚C per decade since 1970 (Osterkamp and Romanovsky 1999) in response to regional warming and changes in insulation by snow and vegetation. Currently 38% of our research watershed (CPCRW) has unstable or thawing permafrost (Yoshikawa et al. 2002, Hinzman et al. 2006) (Fig. 15), and continued warming would likely lead to extensive permafrost degradation within 10-25 years (Romanovsky et al. 2001).

Permafrost response to climate warming depends on changes in insulation by snow, moss, and the surface organic mat (Osterkamp and Romanovsky 1999, Sazonova and Romanosky 2003, Kasischke and Johnstone 2005). Insulation declines dramatically after fire, increasing the layer of thawed soil from about 50 cm to 2-4 m. As permafrost recovers during post-fire succession, an unfrozen layer (talik) forms between a seasonally frozen layer and original permafrost. In sloping terrain, water drains laterally through the talik, drying surface soils. Thawing of ice-rich permafrost may cause subsidence of the ground surface (thermokarst), leading to impoundment or drainage depending on topography (Myers-Smith 2005). Thus the impact of climate warming on soil moisture in permafrost terrain depends strongly on factors controlling talik formation and drainage conditions (Yoshikawa et al. 2003).

Low-permafrost watersheds or watersheds with well-developed taliks have greater base flow (80% of discharge) and are less flashy (i.e., less likely to cause floods) than high-permafrost watersheds, in which base flow increases from 50-60% of discharge in early summer to values similar to those of low-permafrost watersheds in late summer when mineral soils have thawed (Ishikawa et al. 2001, Hinzman et al. 2002). Thick aufeis in areas with abundant winter groundwater flow kill most woody vegetation. Groundwater flow also generates higher concentrations of base cations, inorganic nitrogen, and dissolved CO2 and less dissolved organic carbon and nitrogen than in permafrost-dominated watersheds, where most water flows through the organic mat (MacLean et al. 1999, Petrone et al. 2000). Thus permafrost and talik distribution strongly influence soil moisture, land-water interactions, and stream discharge and chemistry. In contrast to temperate ecosystems, nitrate losses in streams are 4 to 5-fold greater than deposition inputs (Fig. 16, 17), a result that we cannot currently explain in light of the strong nitrogen limitation of watershed vegetation.

Fires are a second major cause of threshold changes in the boreal forest. Extensive large fires in the 1860’s set the stage for the current age distribution and C dynamics in Alaska and Canada (McGuire et al. 2004). Differences in fire severity strongly influence patterns of vegetation C storage across the circumboreal north; western and central Siberia are dominated by ground fires that tend not to kill trees, whereas far eastern Siberia and boreal North America are dominated by crown fires that do tend to kill trees (McGuire et al. 2002).

The interaction between fire and permafrost thaw is a particularly important determinant of climate feedbacks (Myers-Smith 2005). The lower albedo and greater sensible heat flux of spruce compared to deciduous forests or non-forested wetlands (Chapin et al. 2000, Chambers and Chapin 2002) (Fig. 18) suggest that northward forest expansion could be a positive feedback to regional warming, but that loss of forests to the south or net conversion from conifer to deciduous forests resulting from fire could have a net cooling effect, one of the few negative feedbacks to high-latitude warming that has been identified (Chapin et al. 2000, McGuire and Chapin 2006).

Boreal forests contain approximately 27% of the world's vegetation carbon inventory and 28% of the world's soil carbon inventory (equivalent to 75% of the total atmospheric carbon) (McGuire et al. 1997), so warming effects on net ecosystem production (NPP – decomposition) or on fire regime could substantially alter the global climate system (Potter et al. 2001, Clein et al. 2002). Warming appears to enhance carbon release in dry areas, enhance uptake in wet areas, and enhance methane release in wet areas (Thompson et al. In press, Zhuang et al. In press). The net effect of fire depends on fire severity and on changes in fire frequency (Kasischke et al. 1995, Zhuang et al. 2002). All of these effects on trace-gas feedbacks hinge on permafrost and hydrologic changes, which are poorly known (Chapin et al. 2000, Harden et al. 2003, McGuire and Chapin 2006) (Fig. 19). The recent shrinkage of lakes and wetlands in interior Alaska suggests, however, that the CO2 efflux is increasing and methane efflux is decreasing (McGuire et al. 2004).

Fig. 19. The response of carbon storage in interior Alaska to climatic change depends on fire severity and frequency, which are influenced by changes in the hydrologic state of the system (Harden et al. 2000).

Boreal forest dynamics thus reflect a complex interplay between disturbance regime, climate, and species interactions. Our prior research leads to the prediction that resilience to external perturbation (e.g., climate warming) will depend on the traits of the dominant species (e.g., those species' ability to affect ecosystem properties) and the degree of coupling between climate and disturbance.