The next three sections describe the details of our research program and its relationship to our conceptual framework of resilience and vulnerability. Forest dynamics addresses the structure, diversity, and population dynamics of the boreal forest. Biogeochemistry addresses the underlying biogeochemical mechanisms that link population processes with physical drivers of change. Landscape dynamics integrates community and biogeochemical processes at regional scales. Within each section we address hypotheses relating to our three over-arching questions: (Question 1) resilience of current dynamics, (Question 2) vulnerability to change, and (Question 3) alternative future states (Fig. 19).

Fig. 19. Overview of research on forest dynamics. The linkages between external drivers, internal dynamics and the other themes (biogeochemistry and landscape dynamics) are shown together with the hypotheses that will examine these linkages.

Theme 1: Forest dynamics

Hypothesis 1: Physical environment exerts primary control over spatial patterns of ecosystem structure and species composition, whereas ecosystem feedbacks govern temporal change; legacies modify these expected patterns most strongly after disturbance and in late succession.We explore this hypothesis by studying spatial patterns (FD1-1) and temporal patterns within (FD1-2, FD1-3, FD1-4) and among successional trajectories (FD1-5).

FD1-1 (i.e., Forest Dynamics Hypothesis 1-1): Landscape distribution of vegetation is largely predictable from state factors, with the residual variation caused by measurement error and legacies of past events. Methods FD1-1

Vegetation patterns provide insight into the relative importance of current processes (physical controls and internal feedbacks) and historical legacies in governing ecosystem properties and distribution. By legacies, we mean those past events that are not predictable from state factors but which significantly affect the current structure and functioning of ecosystems. These include fire, land clearing, tree harvest, flooding, insect outbreaks, and snow-breakage events. We will modify and improve existing vegetation maps of our two intensive research areas (BCEF and CPCRW) based on high-resolution aerial photography. We also have a 1 km-resolution vegetation map of Alaska based on AVHRR (Calef et al. Submitted). The new databases will be developed hierarchically with the greatest detail in vegetation composition at the finest scales. We will ground truth the maps using data from existing LTER monitoring plots and randomly located new plots near the road network. For each vegetation polygon, we will estimate state factors (slope, aspect, equivalent latitude, January and July temperature, and summer and winter precipitation, and parent material). Maps of recent fires (1950-present), a key boreal legacy, are available (Kasischke et al. 2002)(Fig. 20), allowing us to test for interactions between state factors and this legacy. We use stand-age reconstructions to develop a longer-term history of fire legacies in selected locations (Yarie 1981, Mann and Plug 1999, Fastie et al. 2002) (See LD1-1). The hypothesis will be tested at each scale by determining the degree of relationship between vegetation type and state factors. Although we cannot quantify legacies with certainty, we will examine sites that appear to be outliers to the relationship between state factors and vegetation type for clues as to the controls over their composition. Products of this research will be maps of our intensive study areas and of Alaska with GIS layers for vegetation, climate, soil type, topography, and human population density. Derived variables will include permafrost distribution and fire probability. These are essential tools for regional modeling and geographic extrapolation of our site-based research. We can also compare the new vegetation map of BCEF with maps developed 20 years previously of BCEF (Yarie and Viereck unpubl.) and the floodplain of the entire Tanana River (Ott et al. 2001). [Walker]

Fig. 20. Map of large fires that have occurred in interior Alaska since 1950.

FD1-2: Resilience declines through succession because strong species interactions and feedbacks regulate species composition in mid-succession, whereas legacies are more likely to alter species composition in late successional communities. Methods FD1-2

This hypothesis of decreasing resilience through succession is a key assumption of resilience theory (Holling 1986, Holling and Gunderson 2002). From landscape patterns alone (FD1-1) we cannot distinguish the relative importance of direct physical controls over vegetation type that are related to slope, aspect, and climate from feedbacks that govern internal dynamics. We can use data from continued monitoring of the LTER permanent plots and long-term experiments to study species interactions and feedbacks because topographic position and parent material of these plots are constant through time. From our 15 years of data on species composition in permanent plots (see Research Design; Table 2), we will compare the timing and magnitude of change in species composition to the timing of potential legacies (ice-breakage event, snowshoe hare peak). We hypothesize that these events have greatest impact on species composition in late succession. [Walker, Chapin] We will also measure species composition in experiments that manipulate interactive controls (low-level nitrogen- and carbon-addition plots, rain-out shelters, and moose/hare exclosures: FD2-2, BG2-3; Table 4). We expect these experimental manipulations that alter internal feedbacks to have stronger effects in mid- than in late succession. Tests of these hypotheses will give insights into controls over both successional dynamics and system resilience.

FD1-3: The resilience of the landscape mosaic is controlled by disturbance regime and seed availability. Methods FD1-3

There is a spectrum of post-fire successional trajectories in interior Alaska that occur along gradients in soil temperature and moisture (Fig. 3) and which vary along latitudinal gradients, with black spruce becoming less frequent close to treeline (Lloyd, unpub.). This landscape mosaic would be resilient to climate warming, if warming-induced changes in fire regime or seed availability have negligible effects on post-fire successional trajectory (Fig. 21). Alternatively, if fire size, severity, or frequency, or the availability of off-site seed sources alters post-fire succession, we expect that climatically induced changes in these factors would change the upland vegetation mosaic. We currently do not know the frequency with which the development of each post-fire stand repeats the successional trajectory of the stand that preceded it, although field observations suggest that this is common (Mann and Plug 1999) but not universal (Johnstone et al. 2003). Most tree establishment in interior Alaska and adjacent Canada occurs within 1-2 decades after fire (except for black spruce, in which some individuals continue to establish but are less likely to reach the canopy) (Johnson 1992, Fastie et al. 2002, Johnstone and Chapin 2003, Yarie and Van Cleve In press).

Fig. 21. Triggers for change in successional trajectory (Johnstone 2003). Each successional trajectory tends to repeat itself but unusual events can trigger a shift in successional trajectory.

We will use three complementary approaches to examine the resilience of post-fire successional trajectories. (1) We will document tree recruitment and mortality in the LTER permanent plots and in plots established in earlier post-fire stands by Viereck and Foote (Foote 1983, Johnstone et al. In press). [Walker, Chapin] (2) We will establish permanent plots in several road-accessible burns (aged 10-50 years) in interior Alaska to compare pre-fire tree composition (based on densities of fire-killed trees) and post-fire tree composition (seedlings, saplings, and resprouts) (Johnstone and Chapin 2003). If a successional trajectory is resilient, we expect that the same tree species will be present both before (as adults) and after fire (as seedlings and resprouts) (Fig. 21). Failure of a species to regenerate or arrival of a new tree species that was absent before the fire suggests a shift to a new successional trajectory. We will permanently mark the positions of these plots, so they can be revisited in the future (perhaps decadally) to document subsequent vegetation change. We will document potential controlling factors (slope, aspect, parent material, organic-matter depth, thaw depth, and soil moisture). We will select these plots to span a wide range of environments, fire severities, and distance to seed source [Chapin, Lloyd]. (3) We will use stand demographic data to parameterize a series of matrix population models describing black and white spruce population dynamics along a transect from the interior to arctic treeline. The transect will include the permanent plots described previously as well as plots established north of Coldfoot, Alaska, by Lloyd (unpub.). New plots will be established in the area between the Yukon River and Coldfoot in order to complete the transect. Models will be based on data from tree cores and direct demographic observations. The resilience of populations to varying disturbance regimes can be tested with matrix models by parameterizing ‘non-fire’ matrices (which describe population dynamics in the absence of disturbance) and ‘fire’ matrices (which describe population dynamics in the years following a fire (Wilson, unpublished thesis). The models allow estimation of population stability under different disturbance regimes: resilient populations should be viable in a wide variety of disturbance regimes. [Lloyd]

These studies will collectively allow us to test two hypotheses about variation in the resilience of the landscape mosaic. First, within interior Alaska we hypothesize that the resilience of post-fire recruitment will be limited by the following factors: (1) fire return interval shorter than the age of first reproduction of black spruce (the only species that regenerates primarily from seeds produced within the burned area), (2) extremes of fire severity (either bare mineral soil or thick post-fire organic mat), and (3) stands distant from seed sources. Second, we hypothesize that resilience will change along latitudinal gradients in two ways. First, we expect that in the interior, succession to white spruce will be the most vulnerable to altered successional trajectory because of its dependence on seed inputs from living trees outside the burn. Seedling transplants will provide an experimental test of this hypothesis (FD3-1). Second, we expect that resilience of black spruce forests will decline with increasing latitude because of climatically determined limits on reproductive success (Black and Bliss 1980) (Wilson unpublished thesis). Results from these observations, when extrapolated regionally based on remote sensing (LD1-2), provide input to a landscape model of succession that projects changes in regional vegetation, based on scenarios of future climate, fire regime, and human activities (LD1-1). [Juday, Chapin]

FD1-4: Fluvial characteristics of the Tanana River control the resilience of the floodplain vegetation mosaic through effects on plant establishment and elimination of stands by erosion. Methods FD1-4

We will also examine resilience of successional trajectory in floodplains, where erosion is the disturbance that eliminates late successional stages and silt-bar formation provides opportunities for establishment of new stands. A recent satellite-derived map documents the stand types that were removed by erosion and the area of new silt bars formed for the Tanana River drainage (1980-2000) (Ott et al. 2001). We will continue measurements of seedling establishment and vegetation change in permanent LTER floodplain plots (FD1-2). We will assess the stability of the landscape mosaic, using a state-and-transition model of landscape change (LD1-4; Fig. 22) based on data for silt-bar formation, stand establishment, vegetation change, and stand loss by erosion. To explore the underlying mechanisms, we relate river flow regime (documented since 1920) to seedbed conditions for germination and establishment, survivorship in subsequent floods, and nutrient supply for growth through flooding and hyporrheic flux. To document the latter effects, we measure nutrient inputs from flooding, litterfall, and hyporrheic (i.e., subsurface groundwater) flow, using mass balance and stable-isotope tracers. [Kielland, Jones, Ruess]

Fig. 22. State and transition model of successional dynamics driven by interactions of herbivory with geomporphic processes (accretion and erosion) and vegetation processes (colonization and succession).

FD1-5: The population size of small mammals responds primarily to physical fluctuations in environment, whereas the population size of large animals responds primarily to species interactions (the interaction of food availability and predation). In other words, mammalian species differ predictably in their vulnerability to climatic and biotic change. Methods FD1-5

The relative importance of biotic and abiotic regulation of microtine population dynamics has been the subject of substantial research and controversy (Batzli 1992, Schweiger and Boutin 1995). At high northern latitudes in Europe, Fennoscandians conclude that biotic forces of predation regulate timing and amplitude of population cycles (Korpimäki et al. 2002). However, they argue that in North America, where microtine communities are dominated by species of Clethrionomys, there will be greater resilience and smaller fluctuations, and consequently abiotic factors may have the opportunity to exert influence. Because we cannot directly manipulate environmental fluctuations at large scales, we will use quasi-experimental designs (Cook and Campbell 1979) in which we will concurrently measure climatic factors Rexstad and Kielland In press) at LTER instrumented sites along with density of small mammals. This will be conducted in replicated floodplain and upland sites. Based on our past records, we expect differences between upland and lowland sites in temperature (particularly in winter) and snow depth. Differences in these environmental features among years and among sites will be used to assess our predicted linkage between environmental factors and small mammal population performance.

A population peak in snowshoe hares, an intermediate-sized herbivore, is projected to occur in 2008. Radiotelemetry will be used to track fates of snowshoe hares and measure predation rates. Concurrently, offtake rates will be merged with a rudimentary bioenergetics model to assess possible magnitude of nutritional stress on hares. Demonstrating the interactions of top-down vs. bottom-up regulation in snowshoe hare population dynamics is challenging Stenseth et al. 1997, Sinclair et al. 2000). However, we intend to determine whether biotic factors (incorporating both top-down and bottom-up processes) are important relative to abiotic influences in affecting the resilience of herbivores in boreal ecosystems. Resource agency data (e.g., Alaska Dept. of Fish and Game: ADF&G) will afford the opportunity to evaluate the response of moose dynamics to historical changes in food availability and levels of predation as altered by predator-control regimes instituted at various times in Alaska (NRC 1997) in a retrospective analysis. [Rexstad, Kielland]

Hypothesis 2: Changes in interactive controls trigger changes in ecosystem structure and function. We test this hypothesis by examining the response of forest dynamics to naturally occurring and manipulated changes in interactive controls: functional types of organisms (FD2-1, FD3-1, BG3-3) and climate/soil resources (FD2-2 and LD2-3) (Fig. 16).

Fig. 16. The relationship between state factors (outside the circle), interactive controls (inside the circle), and ecosystem processes (inside the box). The circle represents the boundary of the current ecosystem, whose structure and functioning respond to and affect interactive controls, which are ultimately governed by state factors. Ecosystem properties are also determined by legacies related to long-term ecosystem development (e.g., past migrations) and short-term succession change. Disturbance and other factors can push the system toward some new state or initiate a new cycle of succession (Chapin et al. In press-c).

FD2-1: Early successional vegetation types are vulnerable to invasion by exotic species because (1) there is less biotic control over establishment in early succession, and (2), once establishment occurs, early successional stands are strongly driven by feedbacks that can be modified by invaders, so invaders have a potentially strong effect on ecosystem function. Methods FD2-1

This research tests the hypothesis that ecosystem resilience is minimal immediately following disturbance (Holling and Gunderson 2002). Alaskan ecosystems have, until recently, been remarkably resistant to species invasions (Billings 1973). Recently, however, two invasive genera (Taraxacum and Rhinanthus) have been found in the BNZ LTER site, and larger-scale invasions by an N-fixing species (Melilotus) have occurred in the floodplains of southcentral and southeast Alaska (Shephard pers. comm.). Melilotus was intentionally sown along roadsides in Alaska by the Dept. of Transportation and has begun colonizing river floodplains where roads cross rivers. About 20 exotic species appear to be expanding their ranges in human disturbances in interior Alaska, but there has been no careful study of their potential establishment in natural ecosystems. This hypothesis is difficult to test, because establishment of exotics is still a rare event in interior Alaska. We visualize two approaches to the problem: (1) In our permanent plots, which represent all successional stages, we will document any establishment of exotics that occurs. (2) We will survey those locations in interior Alaska where we expect exotics to be most likely to colonize non-anthropogenic sites. These include places where roads or other human disturbances are adjacent to natural disturbances and places where exotic species are growing adjacent to natural stands of vegetation (e.g., roadsides or fields). We will collaborate with the State of Alaska in its effort to develop a spatial database of the movement of invasive species in Alaska. This combination of approaches enables us to study the process of species invasion when the process is just beginning. Once we know the geographic pattern and ecological controls over establishment of exotics, we can use our GIS database of vegetation and ecological factors to develop simple models of the spread of exotics, based on alternative projections of future climate, human disturbance regime, and control strategies. [Wurtz] [See also FD3-1 and BG2-2 for other tests of species effects.]

FD2-2: Intensification of drought stress associated with warming reduces the resilience of south-facing uplands to climatic warming. Methods FD2-2

Growth of white spruce correlates negatively with summer temperature because of its sensitivity to drought stress (Barber et al. 2000) (Fig. 2). Over the next phase of LTER research we will document the climate sensitivity of all other major tree species (black spruce, birch, aspen, larch) and determine how landscape position influences this sensitivity (upland vs. lowland black and white spruce; slope and aspect effects on all trees). Using these climate-growth relationships and the GIS database on vegetation distribution pattern and microclimate (FD1-1), we can describe regional patterns of climate sensitivity of forest growth and develop projections of regional vulnerability of each tree species and of the landscape mosaic to scenarios of future climate. [Juday]

We will test this hypothesis experimentally with our long-term rainout shelters, which eliminate summer rainfall, combined with snow-removal treatments (to be initiated in 2004) to eliminate inputs from snowmelt (Table 4) The rainout experiments have shown surprisingly little growth response of white spruce to elimination of summer rainfall (Yarie and Van Cleve 1996), and soil moisture measurements suggest that snowmelt may satisfy tree water requirements. Samples of plant water will be analyzed for ¹⁸O to determine whether the water comes primarily from winter or summer precipitation. We also alter soil N supply (both increases through low-level N fertilizer additions and reductions through sawdust addition; Table 4) to assess the relative importance of water and nutrients as limiting resources (FD1-2). [Yarie]

Hypothesis 3: Changes in key plant and animal taxa govern the relative frequency of alternative states. We test this by manipulating tree taxa that colonize after disturbance (FD3-1, BG2-2) and documenting food web structure and effects of predator control programs (FD3-2).

FD3-1: Changes in post-disturbance tree recruitment alter successional trajectories of species composition and ecosystem processes. Methods FD3-1

Research described above (FD1-4) will determine whether successional trajectories are changing in interior Alaska. Here we will test the consequences of these changes by altering experimentally the species composition of tree seedlings that establish after disturbance. In each of two recent burns (one wet, one dry), we will establish plots planted with seedlings of each major native tree species (black and white spruce, alder, aspen, birch, poplar, larch) and two exotic species that have been used extensively in forestry trials (lodgepole pine and Siberian Larch). Seedlings of each species will be grown in the greenhouse for 6 months, then planted out at 1 m spacing in 10 x 10 m plots (3 replicate plots per tree species per site). Seedling success, survival, and growth will be monitored annually for 6 years, then less frequently. Organic matter depth and soil moisture will be used as covariates to explain patterns of seedling success. Seedlings that colonize naturally will be weeded from all except the control (unplanted) plots. We intend to maintain these plots indefinitely, so we can follow the ecosystem consequences of alternative successional trajectories in terms of species composition, herbivory, and biogeochemistry. In addition, we will continue monitoring floodplain plots in which alder and white spruce were planted and maintained (or excluded) for 15 years [Chapin].

FD3-2: Changes in predation pressure associated with natural population fluctuations or predator control cascade down food webs. Methods FD3-2

In previous LTER research, we demonstrated substantial species and geographic variation in stable isotope signatures of soils, water, vegetation, and animals (small mammals, ungulates, and fur bearers) (Kielland et al. 1998, Kielland and Bryant 1998). In the research described here we will collaborate with ADF&G to examine predator-prey relationships across interior Alaska [Kielland]. We will use ¹³C and ¹⁵N to document geographic and temporal variation in species-specific isotopic signatures, as we have done previously for moose (Kielland and Finstad 1998, Kielland 2001b). This will be used in combination with earlier diet studies to construct food web relationships (Peterson and Fry 1987) and to document temporal and geographic variation in these food webs. Initial data will come from archived samples of fur bearers over the past 10+ years. Samples will also be analyzed from populations of microtines and snowshoe hares that are currently being monitored by the LTER (FD1-4) and by ADF&G and from populations where wolf populations have been substantially reduced in predator controls programs.

Theme 2: Biogeochemistry

Our research in this theme explores the biogeochemical mechanisms and consequences of Alaska’s changing boreal forest (Fig. 23).

Fig. 23. Overview of research on biogeochemistry. The linkages between external drivers, internal dynamics and the other themes (forest and landscape dynamics) are shown together with the hypotheses that will examine these linkages.

Hypothesis 1: Changes to factors influencing carbon and nutrient cycling determine the rate of transition between successional states. With this hypothesis, we examine successional changes in microbial (BG1-1), plant (BG1-2, BG1-3), and animal (BG1-4) controls over biogeochemical processes.

BG1-1 (i.e. Biogeochemistry hypothesis 1-1): Increased specialization of fungi through succession impacts the rate and pathways of N and P uptake by host plants through increased diversity of exoenzymes synthesized. Methods BG1-1

In collaboration with a recently funded fungal genomics project, we will document successional changes in fungal composition and their ecological consequences, using molecular fingerprinting techniques. This will enable us to test several predictions that have biogeochemical implications: (1) Fungi with significant proteolytic capabilities (“protein fungi”) will comprise a larger proportion of the fungal community than in temperate sites. (2) Ectomycorrhizal fungi will always dominate fungal biomass, but the ECM/saprophyte biomass ratio will decline in late succession. (3) Fungal diversity enhances enzymatic diversity, so both fungal diversity and substrate specialization will increase through succession. (4) Increased substrate specialization will be reflected by increased activity of fungal protease and acid phosphatase through succession. (5) Successional changes in fungal composition and diversity influence the growth, lifespan and decomposition of fine roots. If these hypothesized trends are confirmed, they would show increased connectedness through succession, as hypothesized by resilience theory (Holling and Gunderson 2002).

We will extract community DNAs from soil using the MoBio UltraClean Soil kit, then amplify the ribosomal ITS region, which is diagnostic of fungi at the species-level, using fungal-specific primers (Gardes and Bruns 1993). The diverse ITS amplicons will be separated using the new T-RFLP method (Dickie et al. 2002). Band sizes on the T-RFLP gels can be predicted from the sequences in our massive soil clone library and herbarium databases, derived from the funded genomics project. Hence, the majority of taxa will be identified and can be categorized as saprophytic, parasitic or mycorrhizal. Primers for RT-PCR will be designed based on conserved motifs in the extracellular alkaline proteases and acid phosphatases from the genome sequences of known arbuscular mycorrhizal, ectomycorrhizal, and saprophytic fungi. These studies will be among the first to characterize the relative abundances and gene expression of saprophytic and ectomycorrhizal fungi and to relate them to biogeochemical processes. [Taylor, Ruess, Kielland].

BG1-2: Proportional allocation to fine root production declines with increasing total NPP both within and among forest types. Methods BG1-2

Although we have considerable understanding of the controls over aboveground NPP (ANPP) (Yarie 1997, Yarie and Van Cleve In press), belowground NPP (BNPP) (Ruess et al. 1996, 2003, In press), and non-vascular NPP (NVNPP) (Ruess et al. 2003) (Fig. 6), we have never integrated these studies into a common conceptual framework, as we propose to do here. Annual low-level additions of nitrogen (doubling annual N cycling through vegetation) clearly demonstrate that aboveground tree production is N-limited and can be modeled based on a nitrogen-productivity relationship (Yarie and Van Cleve 1996, Yarie 1997). However, if root allocation increases with increasing nitrogen limitation, as expected from allocation theory (Garnier 1991), the relationship for total NPP may be quite different than for ANPP. We will focus initially on black spruce, which exhibits a near order-of-magnitude variation in ANPP in interior Alaska. We will expand our black spruce minirhizotron studies (measuring root production, longevity and decomposition) currently focusing on low-productivity floodplain stands to include 3 additional upland sites at mid to upper levels of ANPP. Aboveground N productivity equations (Yarie 1997) will be developed from site-specific allometric equations to evaluate the utility of foliar N as a surrogate for ANPP. In addition N-productivity relationships will be developed that include the belowground component and total NPP (trees, shrubs and moss) across this production gradient. Verification of the N-productivity relationships will entail establishing low-level N addition (100 kg N ha^-1 yr^-1) experiments in black spruce. [Yarie, Mack, Ruess]

We have documented % cover of all species in our permanent plots. We will develop regression relationships between %cover and ANPP for major understory species to extend our previous work (Yarie and Mead 1988). [Yarie, Chapin]. We will extend ongoing work on the productivity of mosses and lichens to incorporate nonvascular plants into our ecosystem-scale estimates of NPP [Sveinbjornsson, Mack].

BG1-3: Root respiration and root-derived heterotrophic respiration account for an increasing proportion of soil respiration as NPP declines because of increased root allocation and more rapid tissue turnover and decomposition of below- than aboveground tissues in unproductive stands. Methods BG1-3

Estimates of root respiratory contributions to total soil CO₂ efflux agree well between soil trenching methods and more direct estimates (minirhizotrons combined with root and soil respiration measures) (Ruess et al. 2003, Vogel et al. Submitted). However, these estimates are greater than those derived from ¹⁴CO₂ (Schuur, unpublished), and all methods have not been compared at the same sites. Across the expanded set of black spruce sites (BG1-2), we will estimate total belowground C allocation using litterfall and soil respiration (C balance method), minirhizotrons (intensive sites only), soil trenching experiments (Vogel et al. Submitted), and ¹⁴CO₂ studies. Finally, we will initiate a pilot study using whole-tree trace labeling with ¹⁴CO₂ (followed by sampling, fractionation and AMS analysis) to determine the fate of dead fine root C. Together these experiments and measurements will enable us to document ratios of aboveground to belowground NPP, tissue turnover above and belowground, and heterotrophic respiration, as a basis for regionalization and extrapolation. [Ruess, Mack, Schuur, Yarie]

In addition to intensive studies in black spruce, we will continue annual measurements of aboveground litterfall in upland and floodplain successional stages (continuing our 15-year record in permanent LTER plots) and in mature forest types in the Fairbanks region (continuing a 35-year record maintained by the Forest Soils Lab). Soil carbon and nitrogen stocks were estimated at the beginning of these studies. We will resample these plots to assess long-term changes in carbon storage. This provides us with information on the controls by climate, vegetation type, and successional stage on litter inputs and their relationship to carbon storage. We will be able to combine this information with the 10-year litterbag study that was recently completed across all the LTER sites and the log decomposition study that will start its tenth year in 2004. [Valentine, Yarie]

BG1-4: Vertebrate herbivory has strong effects on biogeochemical processes throughout succession. Methods BG1-4

Early in succession, herbivores control biogeochemistry through their direct effects of nutrient return and indirect effects on alder dominance (Kielland and Bryant 1998). During mid-succession, herbivore effects on biogeochemistry are mediated through control over vegetation composition and structure – both of which have short-term (soil water and nutrient balance) and long-term impacts on stand development and response to disturbance (Fig. 4). Late in succession (white to black spruce), herbivore effects are indirect, and primarily on the physical (thermal) environment through changes in canopy shading. We will expand our long-term vertebrate exclosure study by adding a permanent exclosure (15 X 15 m) to each mid- and late-successional site (n=3) in both the floodplain and upland forests within BCEF. These plots will be incorporated into our long-term sampling of soil climate, vegetation growth and litterfall. [Kielland, Ruess]

Hypothesis 2: Critical biogeochemical processes influential in ecosystem function are vulnerable to the interaction between disturbance and climatic change. We test this hypothesis by examining both direct (BG2-1, BG2-3) and vegetation-mediated effects (BG2-2).

BG2-1: The carbon balance of boreal forest soils is highly sensitive to disturbance and climate fluctuations because a large proportion of the soil carbon is above the mineral soil (the surface organic mat), much of the soil organic matter is “thermally protected” rather than recalcitrant, and mineral soil C is not highly stabilized by soil minerals. Methods BG2-1

We have collected samples to estimate the quantity of organic matter removed by recent fires in black spruce forests in wet and dry sites. We will assess the quality of these post-fire samples and unburned controls using long-term laboratory incubations under a factorial combination of cold (3˚C) vs. warm (10˚C) and wet vs. dry (120 and 50% of field capacity, respectively) conditions. Based on these measurements we can determine the relative proportions of labile and recalcitrant carbon before and after fire and whether this carbon is environmentally protected by cold and/or wet conditions. We can also determine whether the carbon in organic and mineral horizons differ in recalcitrance. [Valentine, Schuur]

We will continue ongoing measurements post-fire field measurements of soil respiration and its isotopic composition in the dry (FROSTFIRE) site and extend these to wetter sites in the Survey Line burn. Recent data from dry sites suggests that there is sometimes (but not always) an increase in heterotrophic respiration after fire (O'Neill et al. 2002, 2003), but there is not a large contribution from old (pre-1950) carbon to post fire respiration fluxes. This suggests that direct combustion of soil C by fire may play the largest role in determining the loss of old soil C stored over multiple fire cycles. By extending these measurements to the Survey Line site, we can determine whether these same conclusions apply to wet sites, which typically have the greatest soil carbon storage. [Valentine, Schuur]

BG2-2: Because key species and functional groups play a dominant role in biogeochemical processes and ecosystem response to disturbance, boreal forest C and N balances are vulnerable to changes in their growth characteristics and abundance. Methods BG2-2

Successional trends in biogeochemistry exhibit threshold changes at turning points that coincide with changes in dominant plant functional types (Fig. 5; FD1-2). We hypothesize that environmental changes that alter the abundance or activity of species responsible for successional turning points will dramatically alter biogeochemistry. Here we focus on alder and mosses, which precipitate key turning points in both upland and floodplain succession, and on lodgepole pine, a conifer that could become important in the future.

With alder, we focus on the interaction among alder, Frankia (its nitrogen fixing symbiont), and its ectomycorrhizal associates. N₂-fixation inputs appear to (a) decline through succession due to changes in the abundance of alder, nodulation rates and nitrogenase activity, (b) be greater in forests with A. tenuifolia compared with A. crispa, and (c) be strongly influenced by plant P status. It also appears that the genetic structure of the Frankia community differs between the 2 hosts and likely is influenced by ectomycorrhizal associations uniquely suited for P solubilization. What is less clear is how the Frankia/ectomycorrhizal associations respond following fire (where alder genets rapidly resprout), or as alder invades new environments. What is also poorly understood is the fate of fixed N, or the influence of fixed N in either accelerating or suppressing soil organic matter decomposition. We will expand work on the ecophysiology and genetics of the Alnus, Frankia, ectomycorrhizae triangle across successional stages and disturbance sites and set up experiments to study the effects of N inputs on SOM decomposition, i.e., a key mechanism linking nitrogen fixation to carbon cycling. [Ruess, Mack]

In the absence of N inputs from symbiotic fixation by alder, mosses regulate the rate at which N inputs become available to plants and soils because (1) atmospheric N is deposited on the moss canopy, (2) relatively high (1-2 kg ha^-1 yr^-1) amounts of asymbiotic fixation occur in the moss canopy, and (3) mosses accumulate N in recalcitrant tissues that have a long turnover time. We will test the role of mosses as regulators of N cycling using a combination of observations and experiments. In the LTER permanent plots, we will use estimates of moss biomass and production (FD1-2) and nitrogen content to estimate the stock and annual flux of nitrogen through mosses in each successional stage and the sensitivity of moss composition to fire. We will use the long-term low-level N-fertilization plots to estimate the sensitivity of moss biomass, production, and N sequestration to N inputs in the major boreal forest types. In white spruce forests, nitrogen, water, and hardwood-litter addition experiments provide additional opportunities to explore the controls over moss production and N sequestration, and in black-spruce forests ¹⁵N addition experiments allow estimates of N turnover time in mosses and the long-term fate of moss-sequestered N. Finally existing moss-removal experiments (Sphagnum and Hylocomium) provide an opportunity to document impacts of mosses on microenvironment, N supply, and soil C stocks. [Mack, Sveinbjornsson]

Lodgepole pine has been rapidly migrating northwestward. It reached Whitehorse, Yukon Territories within the last 300 years, and is actively expanding its range by colonizing natural wildfires (Johnstone and Chapin 2003) and through cultivation as plantations in Alaska. Plantations near Fairbanks are now over 20 years old, consist of well-established needle carpets that contrast starkly with the forest floors under aspen and birch canopies of similar ages. We will compare soil C and N stocks, N mineralization, soil respiration, soil properties (e.g., pH, CEC), and fungal community composition between lodgepole pine plantations and similarly aged postfire hardwood stands. Based on these initial surveys, we will develop more explicit hypotheses about the mechanisms by which lodgepole pine altars biogeochemistry and develop an appropriate measurement program. We will also reconstruct interannual variation in tree growth from the tree ring record in this stand and compare it with published records of the climate sensitivity of lodgepole ringwidth in naturally established stands. [Valentine, Taylor, Juday]

BG2-3: Climate warming will cause shoulder seasons to exert primary controls over C and N mineralization in boreal forests. Methods BG2-3

Growing season shoulders (i.e., the snow-covered periods with biologically active soils in fall and spring) are critical periods for biogeochemical processes because (1) their combined length exceeds growing season length, and (2) production and decomposition are uncoupled during a period favorable to soil microorganisms. Climate change will most likely extend the length of these shoulder seasons and invoke strong effects on C and N cycling during these periods. We find that microbial activity is substantial down to –3˚C and is sensitive to freeze-thaw events (Schimel and Clein 1996). Consequently, half of the annual N mineralization occurs between September and January (Fig. 24), but we know almost nothing about the controls over mineralization process during the non-growing season.

Fig. 24. The relative contribution of summer and winter to net N mineralization (from buried bags) in major floodplain ecosystems. The high rates in fall and early winter suggest that Nmin is controlled more by the pulse of autumn root and leaf litter inputs than by temperature.

We will use ongoing LTER monitoring of soil temperature profiles to document the annual pattern of soil temperature and soil moisture (including the unfrozen water content in winter) in our major LTER sites. With soil thermal models (LD2-1) we can estimate long-term trends in the length of the growing and shoulder seasons in interior Alaska and can estimate the volume and depth distribution of unfrozen water content, which we expect to be a critical control over winter microbial activity. We have already established that a large proportion of fine roots die and completely disappear sometime between September and June (Fig. 25). To characterize when this occurs, we will monitor fine root growth, survival and decomposition through the winter using minirhizotrons in black spruce stands. In addition, we will initiate a new long-term experiment examining the effects of snow cover on ecosystem function. Treatments mimicking early snow cover (early October using blue foam insulation covering the moss surface) and no snow cover (using plywood shelters to exclude snow; removing shelters and replacing snow in April) will be implemented in replicated black spruce stands (with 10 X 10 m plots) readily assessable during winter. The long-term constancy of these treatments will be juxtaposed to natural snowfall plots, the timing and depth of which is quite variable among years. Net N turnover (net DIN and amino acid production), C cycling (soil respiration, microbial biomass, filter paper + tongue depressor decay), and live fine root biomass will be measured throughout the fall, winter and spring. Fungal and bacterial communities and exoenzymes in soil will be sampled in summer, fall, winter and spring and characterized by T-RFLP, as described in BG1-1. [Kielland, Ruess, Taylor, Romanovsky]

Fig. 25. Short-lived fine roots decompose more rapidly than longer-lived larger roots. Roots that die in autumn largely decompose over winter, whereas roots that survive over winter decompose more slowly.

Hypothesis 3: Biogeochemical cycles will respond most dramatically to climate warming following disturbance. The magnitude of the response will scale to the level of the disturbance, and the trajectory of recovery will depend upon the site physiographic factors controlling surface energy balance and the life history traits of organisms.

BG3-1: Climate warming and disturbance that initiate thermokarst lead to (1) release of soil organic matter and N from frozen soils, (2) significant landscape evolution, and (3) inputs of sediments, old soil carbon, and nutrients into aquatic ecosystems, causing a net loss of these materials from watersheds. Methods BG3-1

Thermokarst topography forms as ice-rich permafrost thaws and the ground surface subsides, leading to high soil moisture in poorly drained areas and surface drying on slopes. Massive ice near the surface is the primary prerequisite for thermokarst. Almost any disturbance that disrupts the insulation by the organic mat can initiate thermokarst. Models of net radiation and permafrost distribution based on slope, aspect, and elevation provide an estimate of CPCRW locations where permafrost is no longer in equilibrium with climate (Fig. 10). Geophysical surveys (ground-penetrating radar) that document subsurface structure and material properties will provide information to validate and refine this model and to document the current status of permafrost distribution and depth within the watershed. These surveys will be repeated at least every 5 years. We have conducted high-resolution terrain survey transects in areas with high potential for thermokarst formation and ground surface subsidence.  These transects will be repeated and long-term monitoring will be initiated at sites displaying permafrost degradation. We will initiate sedimentation studies and maintain existing streamwater chemistry and discharge monitoring to document current and future transfers of materials from land to streams. These studies will yield critical information on the impacts of climate warming in areas dominated by permafrost. [Hinzman, Jones, Romanovsky]

One of our most unexpected recent results is that nitrate losses from N-limited watersheds at CPCRW are four to five-fold greater than deposition inputs, a pattern never-before reported for undisturbed watersheds (Fig. 26, 27). A major focus of our proposed research is to explain this imbalance in watershed nitrogen cycles. With the advent of a warming climate, large stores of organic matter frozen in permafrost are beginning to thaw. We will study the impact of permafrost and disturbance (fire) on watershed nitrogen losses. To determine the source of N in stream export, we will continue our monitoring of stream water fluxes to evaluate interannual inter-watershed variations in N budgets. Within watersheds, the sources of N to stream flow will be evaluated by quantifying N fixation by alders and feathermoss, estimating the sources of N in stream flow using ¹⁵N of nitrate (the dominant form of DIN) in stream water relative to ¹⁵N signatures of atmospheric N, organic soil N (under and away from alders), and mineral soil N; soil water ¹⁵N end members will be sampled using lysimeters. Stable isotope approaches will be confirmed using conservative solute tracers (e.g., Ca²⁺, Mg²⁺) in hydrologic end-member mixing models to determine the sources of ground water feeding stream flow. In addition to N inputs, we will also quantify the role of denitrification to evaluation potential losses of N within watersheds. [Jones, Hinzman, Kielland, Ruess]

BG3-2: Contraction of lakes and wetlands accompanied by a declining water table reduces methane emissions and increases CO₂ loss, whereas increases in wetlands and lakes have the opposite effect, especially where sedge is introduced. Spatial and long-term significance of changes to ecosystems and their C budgets depends on permafrost and landscape constraints and on vulnerabilities to disturbance. Methods BG3-2

In this research we will experimentally increase, reduce, and leave untouched the water table depths of a lowland ecosystem to test the effect of changing water table on vegetation composition and CO₂ and methane fluxes. In winter we will install water-impermeable dikes and either pump or flood the treatment plots to manipulate water tables. Before and after treatments, we will measure CO₂ and CH₄ flux and DOC concentration, collect cores to characterize the quantity and quality of organic matter in each plot, and install sensors to monitor water table, net radiation, soil temperature (5, 10, and 20 cm), air temperature, and relative humidity. Vegetation and soils will be sampled annually to monitor changes in NPP, substrate quality, and isotopic shifts in solid, liquid and gaseous C phases. Empirical models for ecosystem CO₂ uptake (i.e., gross primary production, GPP), ecosystem CO₂ release (ecosystem respiration, ER), and net CH₄ exchange (NME) will be developed based on gas exchange and environmental measurements. The results of these studies will be extrapolated regionally based on remote sensing that documents recent changes in wetland area (LD3-1). [McGuire]

Fig. 26. N fluxes in CPCRW subcatchments. Inputs calculated from 10 years of data are from an NADP site at CPCRW. Output in stream flow for 3 years of intensive sampling (1986, 2001, and 2002).

Fig. 27. Nitrate and DON concentrations in streamwater along a latitudinal gradient. Concentrations are highest in zones of discontinuous permafrost in interior Alaska. Using total dissolved N (TDN):Cl as an index of N retention, watersheds in discontinuous permafrost appear to be losing N; whereas more northern latitudes are closer to steady state. Toolik data (Peterson et al. 1992) and BNZ data collected along a transect (Jones and Finlay unpubl.) were used to construct this relationship.

Theme 3: Landscape Dynamics

Our research on landscape dynamics integrates the processes described under forest dynamics and biogeochemistry and incorporates additional processes, such as disturbance, that link patches within a landscape in order to scale processes to larger temporal and spatial scales (Fig. 28).

Fig. 28. Overview of research on landscape dynamics. The linkages between external drivers, internal dynamics and the other themes (forest dynamics and biogeochemistry) are shown together with the hypotheses that will examine these linkages.

Hypothesis 1: Landscape processes associated with disturbance increase the resilience of Alaska’s boreal forest. We test this hypothesis, using modeling and remote sensing, by examining fire effects on landscape heterogeneity (LD1-1), fire severity (LD1-2), and carbon balance (LD1-3) and effects of mammals on floodplain landscape structure (LD1-4).

LD1-1 (i.e., landscape dynamics hypothesis 1-1): Fire enhances regional resilience by maintaining a mosaic of flammable and non-flammable vegetation. Climatic or human changes in fire regime will therefore likely alter the landscape mosaic. Methods LD1-1

Fire is the dominant disturbance agent in interior Alaska and contributes to the complex mosaic that characterizes interior Alaska. Our goal is to explore the consequences of landscape vegetation patterns for fire regime and the resilience of this vegetation mosaic. The Bureau of Land Management, Alaska Fire Service (AFS) maintains records of the perimeter, size, and discovery date of all large fires (Fig. 20) and a tabular database that records ignition source (human vs. lightning), extinguishment date, and suppression costs (where applicable). Other GIS and historical databases provide information on vegetation type (FD1-1) and climate. From these databases we will determine the probability of burning of different vegetation types in high-fire and low-fire years and in continental vs. maritime portions of interior Alaska. Fieldwork funded elsewhere (Rupp and Mann) is testing whether the probability of burning depends only on vegetation type or changes with stand age. Together this information will enable us to characterize the vegetation and age dependence of fire probability in interior Alaska. This information will be used to improve parameterization of a spatially explicit landscape model (ALFRESCO) that simulates fire spread and seed dispersal on the landscape to assess the interactions among vegetation, fire, and climate (Rupp et al. 2000, 2002). The model is driven by stochastic variation around a mean climate state, so multiple runs of the model allow estimation of the frequency distribution of potential outcomes of a given scenario of average climate and initial vegetation composition (Fig. 29). We will use the model to explore the influence of changes in (1) the size distribution and spatial arrangement of vegetation patches, (2) assumptions about age- and vegetation-dependent probabilities of burning, and (3) the type, duration, and number of potential successional trajectories that can occur after fire (FD1-3, FD3-1) on future fire regime. These sensitivity analyses allow tests of factors governing the resilience of the landscape mosaic. [Rupp]

Fig. 29. The sensitivity of fire size distribution in interior Alaska simulated by ALFRESCO in response to potential climate change under a warmer drier scenario and under a warmer wetter scenario. The wetter scenario leads to more large fires because of fuel buildup (Turner et al. 2003).

LD1-2: Seasonality of fire combined with legacies from past fire events control fire severity, which in turn regulates the site thermal/moisture regime (LD2-2), soil carbon storage (LD1-3, BG2-1), and post-fire recruitment and succession (FD1-3, FD3-1). Methods LD1-2

Data collected over the past several years (Kasischke unpubl.) suggest that, in black spruce forests, the depth of the organic mat at the time of the fire combined with the timing and duration of the fire are more important than fire weather in determining fire severity (i.e., the extent of consumption of the soil organic mat; Fig. 30). We hypothesize that, in sites with permafrost, only after soils have thawed into mineral horizons can the surface organic mat dry sufficiently to be largely consumed by fire. This hypothesis suggests that the seasonality of fire and active layer depth strongly influence ecosystem properties that govern permafrost integrity, successional trajectory, and carbon storage, and therefore the resilience of the boreal forest. We will use satellite images to estimate spatial patterns of burn severity on road-accessible burns (Michalek et al. 2000) and test the validity of these satellite-derived estimates by field sampling (Kasischke et al. 2000a, Kasischke et al. 2000b). We will conduct field measures of depth of remaining organic soils in the different burn severity categories, and document the site conditions (mineral soil profile, and slope, aspect and elevation of site), and from fire records, determine the time of the fire during the growing season. Similar measures will be obtained from adjacent unburned stands to serve as controls. These data will be statistically analyzed to determine which factors have the greatest influence on surface fuel consumption. Finally, soil moisture and temperature measurements will be obtained as a function of burn severity and time after fire for use in understanding how fire severity influences talik and permafrost formation (LD2-2). [Kasischke, Verbyla]

Fig. 30. A comparison of unburned and burned organic matter depths from three different fires near Delta, Alaska. The reduction in organic matter depth is inversely correlated with pre-burn depth (Kasischke unpubl.).

LD1-3: The strength of the carbon sink in Alaska’s boreal forest is vulnerable to vegetation- and permafrost-induced changes in fire frequency and severity. Methods LD1-3

Carbon sequestration or emission from the boreal forest is sensitive to the frequency and severity of fires and the rate of successional changes in NPP and heterotrophic respiration, as these are integrated across the landscape mosaic (McGuire et al. In press) (Fig. 31). The Terrestrial Ecosystem Model (TEM), which incorporates the impact of permafrost and fire on carbon balance (Zhuang et al. 2001, 2002, McGuire et al. In press) currently simulates regional carbon balance of 1 x 1 km pixels for interior Alaska based on maps of historical climate, vegetation, and fire occurrence. Permafrost spatial distribution and its dynamics during the last millennium influence the fire conditions and especially the probability of a particular successional trajectory after the fire. Using permafrost temperature reanalysis (Romanovsky et al. 2002), we will establish the history of permafrost evolution for each major landscape type for the last millennium. These trajectories of permafrost change will take into account the effect of forest fire on permafrost dynamics (LD2-2). Using the same calibrated models for each landscape type and chosen scenarios of future climate change we will project permafrost dynamic in the 21^st century. We will also use the information on fire probability (LD1-1), fire severity (LD1-2), post-fire successional trajectory (FD1-3), and successional changes in carbon balance (BG2-1) to project regional carbon storage from 2000-2100 based on assumptions of current climate, and continued climate warming. We will conduct these simulations with a coupled version of TEM and ALFRESCO, which has been used to evaluate retrospective changes in carbon storage of the Alaska region in the NSF-funded Western Arctic Linkage Experiment (WALE). We will test the sensitivity of future carbon balance to integrity of permafrost by comparing the above simulations with permafrost present and permafrost absent. [McGuire, Rupp, Romanovsky]

Fig. 31. Cumulative changes in carbon storage of Alaska since 1950 simulated by the Terrestrial Ecosystem Model for assumptions that fire frequency prior to 1950 was more frequent, not different, or less frequent in comparisons with observations since 1950 (McGuire et al. In press).

LD1-4: Mammalian herbivores contribute to the resilience of the floodplain hardwood-conifer mosaic by accelerating the turnover of early successional stands (where establishment can occur) and increasing the landscape distribution of late-successional white spruce stands (which provide a conifer seed source). Methods LD1-4

Mammalian herbivores exert a strong influence on forest dynamics in interior Alaska. We will continue monitoring the impact of herbivore (hare and moose) exclosures (established in 1988 and in 2002) on forest dynamics and ecosystem processes (FD1-2, BG1-5; Fig. 4). In the research proposed here, we will use information collected on the relative abundance of early and late successional stands in areas with high vs. low moose density (Fig. 32) as input to a landscape model of riparian succession to explore the effects of moose on landscape dynamics (Butler 2003; Fig. 22). Our model simulates primary floodplain forest succession from initial colonization of bare silt bars to mature climax white spruce forest. The model will be modified to incorporate the primary life history stages of white spruce from seed production to seedling establishment, identifying the processes that control both the reproductive success and rate of transition from newly exposed silt bars to mature commercially valuable stands. We will use the fieldwork to document white spruce recruitment, growth, and seed production (FD1-3) to inform the model and the model to direct the fieldwork, in an iterative, adaptive fashion (Nicolson et al. 2002). The model will compare scenarios of high and low moose abundance, disturbance frequency (flooding, fire, and white-spruce harvest), and climate change to simulate the rate and frequency of development of commercially valuable white spruce stands. This modeling effort will be used to simulate the effects of future scenarios of these variables on the potential for commercial harvest of white spruce in the Tanana River Valley. [Kielland, Rupp, Wurtz]

Fig. 32. Landscape composition of deciduous trees and shrubs along the Tanana River near Fairbanks (high moose density) and Manley (low moose density) (pie chart). Bar chart shows number of stems m-2 (Kielland unpubl.).

Hypothesis 2: Changes in key landscape disturbances trigger changes in regional landscape structure and functioning. We use modeling and extensive observations to assess the potential impacts of changes in fire, thermokarst, and insect outbreaks.

LD2-1: Past changes in climate-vegetation-disturbance interactions triggered changes in landscape processes that caused threshold changes to new stable states. Methods LD2-1

We will synthesize information on the spatial pattern of Holocene vegetation changes that have occurred in Alaska and western Yukon (see past results) and use this information to parameterize ALFRESCO (LD1-1). We will model climate-vegetation-fire interactions at times of past vegetation transitions, using alternative scenarios of the role of fire in triggering these threshold changes in vegetation-fire interactions. The same permafrost reanalysis technique in LD1-3 will be used for defining Holocene permafrost evolution trajectories for the key landscape types. We propose to use paleorecords of vegetation, charcoal, and climate trends that have already been documented to develop historical model scenarios. This scenario planning approach (Peterson et al. 2003) to data-model comparisons of historical patterns is unique, and this methodology is currently being developed in a NSF sponsored project (Rupp and Brubaker). The key transitions we will focus on will be the white spruce-black spruce transition in the mid-Holocene and the recent white spruce-lodgepole pine transition in the Yukon Territories. [Rupp, Romanovsky]

LD2-2: Changes in fire regime are more important than climate warming in causing talik formation and loss of shallow permafrost. Methods LD2-2

Forest fires alter surface energy and water budget by drastically altering the surface albedo, roughness, infiltration rates, insulation effect of surface vegetation and organic layer, and moisture absorption capacity in organic soils. Although these processes change most radically immediately after fire, the resulting effects become part of a process of long-term (20-50 years) cumulative impacts associated with changes in thaw depth, drainage, and permafrost recovery (see past results). We will use a combination of long-term monitoring and simulation modeling to explore the consequences of changes in climate and disturbance regime for permafrost integrity. The simulation modeling activities will be based on permafrost temperature reanalysis technique (Romanovsky et al. 2002). For each major landscape type we will develop a calibrated model for undisturbed surface conditions. These models will be used then to investigate the effects of fire and post-fire vegetation change on permafrost evolution through their effects on thermal properties of snow, vegetation, and soils. We will continue monitoring surface and permafrost temperature and moisture in several arrays that document the influences of climate and vegetation at different spatial scales. These include (1) a topographic/vegetational gradient from poorly-drained muskeg to well-drained white spruce, (2) an elevational gradient from valley bottom to ridge top (200-900 m elevational range), (3) a latitudinal transect of boreholes from the Arctic Ocean to Gulkana (plus maritime sites on the Bering Sea coast), and (4) a new set of sites to be established in different-aged burns. At these sites we measure air temperature and vertical gradients in soil temperature and moisture from the soil surface into permafrost. At each site we will periodically measure snow depth and thaw depth above permafrost. Lateral profiles of permafrost extent, active layer thickness and total soil moisture content in the active layer will be developed using ground penetrating radar. These will enable us to document the distribution of massive ground ice from which we can predict the pattern of expected post-fire surface deformation. In areas with massive ground ice we will survey the area initially and at 5-year intervals to document the rate and patterns of thermokarst development. These sites will be used to follow the fate of permafrost, microenvironment, and successional changes in vegetation (FD1-2) and biogeochemistry (BG3-1). [Hinzman, Romanovsky]

LD2-3: Increases in summer temperature explain spatial and temporal patterns of change in insect outbreaks. Methods LD2-3

Temperature is a key regulator of insect dynamics in the Alaskan boreal forest. For example, outbreaks of the spear-marked black moth in birch, the la