The Response of Arctic Lakes to Environmental Change
W. John O’Brien
Neil Bettez
Greta Burkart
MaryAnn Evans
Gretchen Gettel
Anne Giblin
George Kling
Chris Luecke
Sally MacIntyre
Parke Rublee
Sampling and analyses of Toolik lake and lakes and ponds near the Toolik field station began in the late 1970’s. The initial focus of this research was to describe limnological conditions and processes that characterized these arctic environments.
Physical, chemical and biological components of these lakes all respond to the seasonal changes in sunlight, temperature, and precipitation typical of the Arctic. The glacial history of the Pleistocene epoch created a diverse suite of landscape characteristics that influences these lake environments. Recent and future warming of this region brings with it the potential for large changes in the processes that define aquatic ecosystems.
We begin by describing current physical, chemical, and biological conditions of lakes of this region. Time-series measurements of limnological conditions provide a basis for assessing effects of environmental change associated with a warming climate.
Then we describe results of an extensive series of experiments conducted to better understand processes important in the biogeochemistry of these systems under present and future conditions. Finally we summarize results from integrative analyses and modeling exercises to assess likely conditions of future lake ecosystems.
Landscape settings of lakes in the Toolik Lake region.
The landscape near the Toolik region is underlain by permafrost. The relatively unfractured nature of the permafrost in this region minimizes deep ground water flows and provides increased connectivity between water courses on the terrestrial hillslopes and downstream water bodies. This permafrost boundary limits transient storage and reduces residence time of water within the watersheds. Research highlighted in previous chapters documents the impact of permafrost on water and solute budgets in these arctic landscapes.
Lakes and ponds near the Toolik Lake Field Station (68 o 38’N, 149 o 38’W) in arctic Alaska originated during glacial activity during the past 300,000 years. Extensive ice-deposits from five glacial advances resulted in the formation of glacial kettle lakes embedded in a complex landscape (Figure 1). Lakes in this region lie on surfaces
1

resulting from glaciations of approximately 12,000 years , 60,000, and greater than
300,000 years before the present (YBP) (Hamilton 2003). Effects of erosion and solifluction are particularly apparent on surfaces of older glaciations (> 60,000 YBP) where Hamilton suggests erosion and solilfluction may have led to the disappearance of some stream channels and a decreased the depth and number of glacial kettle lakes. In contrast, landscapes with more recent glacial advances (< 12,000 YBP) are characterized by well-defined stream channels connecting kettle lakes (Hamilton 2003).
Physical and chemical characteristics of the lakes are largely determined by the glacial history of their watersheds and by morphometery. Most of the larger lakes, including
Toolik Lake, are kettle basins formed approximately 12,000 YBP (Hamilton and Porter
1975). Toolik Lake is a multiple kettle basin formed from several ice blocks left behind as the glaciers retreated. Consequently the lake has a series of shoals and a resulting complex morphometry (Figure 2). Thaw ponds occur in the Toolik region but are not as common as those on the northern coastal plane (Milner and Oswood 1997). These ponds form in depressions in ice-rich permafrost. Water in these depressions warms in summer and melts into the permafrost. We have observed several of these ponds that rapidly drain through fractures in the permafrost and refill the following spring. These shallow ponds are generally not part of a drainage network, often freeze solid during the winter, and thus tend to not support fish populations. Their large ratio of sediment area to water volume, their lack of extended thermal stratification during summer, and the lack of fish result in distinct chemical and biological characteristics in these ponds.
The glacial history affects the distribution and the size of lakes in this region. On the older glacial surfaces, lakes are less abundant, have smaller surface areas, and are less deep than lakes on more recently glaciated landscapes. Surfaces that have not been for >300,000 YBP have less than 0.2% of the landscape area covered by lakes compared to greater than 3% of the surface area covered by water on the newly glaciated regions (Burkart 2007). Mean surface area of individual lakes is much greater on surfaces glaciated 10,000 YBP (4.2 Ha) compared to a mean surface area of lakes on the oldest landscapes of 0.9 Ha. The morphological differences in lakes on landscapes of different glacial histories have implications for their community ecology. The presence of overwintering fish populations is strongly related to lake size (O’Brien 2004; Burkart
2007). An analysis of over 446 lakes in the region indicated that the proportion of lakes suitable to hold permanent fish populations decreased with time since the last glaciation
(Figure 3.)
The differences in glacial history and lake morphometry affect chemical and biological characteristics of the lakes and ponds near the Toolik Field Station. An ordination of xxx morphemetric, chemical, and biological characteristics of these lakes indicated that two major gradients explained much of the differences among bodies of water (Figure 4). This ordination used 18 lakes near the field station where we have similar data collected in 5 different years during the past 15 years. The primary dimension related to characteristics associated with lake size. The larger lakes were deeper, had fewer solutes and contained resident populations of fish. The second
2

dominant axis was related to variables associated with time since the last glaciation.
Lakes on older glacial landscapes had low conductivities resulting from low concentrations of major cations and anions. We use these two dominant axes to group lakes by size class and by glacial history in the subsequent discussions of arctic lake function.
Physical properties of lakes.
The seasonality of solar radiation is extreme in the Arctic. Lack of direct sunlight during winter months results in air temperatures averaging xxx and little liquid water present at depths of less than 1.5 m in lakes and streams. The insulating properties of overlaying snow and ice allow water temperatures to remain above 2 C near the bottom of most lakes. Although the bottoms of streams and lake bottoms freeze when at depths of less than 1.5 m, temperatures rarely are reduced to more than -5 C. Geothermal springs are present in lakes and along some stream courses in the mountains further to the north and provide over wintering habitat for fish, birds and invertebrates but are uncommon in the Toolik Lake region (Huryn 2006).
Lakes near the Toolik Field Staiton are frozen until mid-May. During May, continuous sunlight and warming air temperatures begin to melt snow and cause some runoff into the lakes. Candle ice forms on the lakes and allows sufficient light to enter to support photosynthesis of phytoplankton and benthic algae. Snowmelt on the terrestrial landscape occurs mostly before the lakes thaw. Water budgets for Toolik Lake indicate that peak runoff usually occurs before ice out. (Table 1or 2 on ice out dates and water budgets.)
Arctic lakes stratify quickly after ice off which occurs near the time of the summer solstice and solar radiation is near maximal values (O’Brien et al. 1997). The pattern of thermal stratification of any lake or pond varies considerably from year to year
Figure 5 depicts the stratification patterns in 2003 and 2004 in Toolik Lake. Maximal water temperatures were near 16 o C in Toolik Lake in 2003, a cold summer and metalimnetic thickness ranged from 1 to 4 m. In contrast, in the warm summer of 2004, epilimnetic temperatures were over 18 o
C for extended periods and the metalimnion was 4 to 8 m thick. Internal wave amplitudes were higher in the cooler year. In cool summers, the mixing induced by internal waves in the lower water column has the capacity to transport regenerated nutrients vertically from the sediment-water interface.
The stability of the water column resulting from thermal stratification affects the degree to which wind energy mixes epilimnetic and metalimnetic waters (Imberger and
Patterson 1990). When the degree of water column thermal stratification is low, resistance to mixing is low and the upwelling of hypolimnetic water will occur frequently. These mixing processes have consequences for the supply of nutrients to surface waters from the hypolimnion, from groundwaters, and from incoming streams.
3

The degree of thermal stratification also affects how cold fronts facilitate transfer of water and chemical components between the upper and lower water colum (MacIntyre et al. 2001). Metalimnetic intrusions of inflow water during storm events in the cool summer of 1999 were dispersed over much of the lake within a few days. In contrast during the warmer summer of 2004, three inflow events occurred but the bulk of the incoming water in the first event did not mix into the upper water column until flushed by the third stream inflow event over a month later (Rueda et al. 2006). The water column quickly restratified after this event (day 193, 2004) and remained stratified until autumn cooling began in August (Figure 5).
Our understanding of the physics of water column mixing comes largely from detailed measurements on the larger lakes in our area. Stratification patterns vary consistently with the depth and morphometry of the lake basin. Lakes with maximum depths deeper than 5 m stratify thermally shortly after ice-out and generally develop distinct epilimnia. Shallow lakes become polymictic, stratifying during the mid-day hours and sometimes remaining stratified for several days at a time before wind events mix the water column. In small shallow lakes, the likelihood of wind forcing energizing an internal wave field becomes smaller. Most of the mixing dynamics in these smaller lakes occurs by convection only.
Sally’s story on shallow and deep lakes keep box available
Chemical Properties of Lakes
Precipitation in the Toolik region is very dilute and not heavily influenced by oceanic influences (Kling et al 1992). In many regions of the world, the relative proportions of direct precipitation to soil water inputs are the major determinate of local variations in lake chemistry. However, in the Toolik region, local geological changes are so large that the major factor controlling differences in the conductivity and major ion chemistry of lakes is the variation in the glacial history of the different watersheds.
Unlike the physical and biological factors of lakes in the region, chemistry is only weakly affected by lake size, lake residence time, or lake to watershed area (Table 2).
Results from the multiple dimensional scaling analysis indicated that variation in lake characteristics derives from two dominant axes, both of which are related to the glacial history of the region. Older glacial landscapes tend to have more weathered soils and lower solute yields than younger landscapes. More recently glaciated landscapes tended to contain larger and deeper lakes with well-defined drainage patterns. Biological communities tend to be more species rich in the larger lakes due to the inclusion of larger organisms like fish and multivoltine invertebrates that need to over winter in a liquid water. We grouped our most frequently sampled lakes according to the glacial history of their watersheds as depicted in Figure 1 and according to lake size where small lakes are
4

those lakes less than 5 m deep or with surface areas of less than 1 ha. Intermediate lakes are those with maximum depths from 5-10 m, and larger lakes are those with depths greater than 10 m or with surface areas exceeding 10 ha (Table 2?).
Lakes surrounded entirely by the most recently glaciated landscape (<12,000 bp) are characterized by relatively high conductivities (> 90 uS cm -1 ), and have the greatest concentrations of Ca (>500 uM), Mg (>100uM), and bicarbonate (>1000uM) (Table or fig AG1). Sulfate concentrations are highly variable (15-250uM) but are usually greater than in lakes found on older surfaces, and the highest sulfate concentrations are only found in lakes on young surfaces. Lake chemistry is consistent with the soil water influenced by the weathering of limestone and dolomite found in these younger surfaces but suggests that gypsum and/or pyritic shale is present more sporadically (Brown &
Berg 1980; Brown and Kreig 1983). Concentrations of Cl, Na, and K are all quite low and show little predictable variation with landscape age consistent with the low input of recent sea salts to the region (Kling 1992).
Lakes on the oldest surfaces (>150,000) are very dilute, and have the lowest conductivities (<35 uS cm
-1
) and major ion concentrations. Sulfate concentrations in the lakes for which data is available, are extremely low, less than 5 uM. Lakes on the landscapes of intermediate glacial ages have chemistries that are intermediate between the youngest and oldest lakes but overlap in the ranges of some solutes. Many lakes, including Toolik Lake and the lakes in the Toolik inlet area, drain watersheds of mixed aged landscapes. These lakes too have chemistries that reflect the mix of the different geological surfaces.
Biogeochemical budgets and cycles
A major effort to develop water and nutrient budgets for Toolik Lake was conducted in the late 1970’s and early 1980’s (Whalen and Cornwell 1985). The two major inlets to Toolik Lake accounted for approximately 70% and 10% of the total discharge into the lake with the remaining 20% coming from seepage and smaller inlets.
Up to 30% of the N and P inputs occurred during the first 10 days of stream flow when the lake was still ice covered. Nitrogen inputs from the two inlet streams were dominated by organic forms of dissolved N, which made up over 80% of the total N inputs.
Dissolved P concentrations were usually higher than particulate P and dominated by organic forms, although not to as great as an extent as for N (at the moment your table
2). The chemistry of seepage water was not directly measured but assumed to be similar to the volume weighted inputs of the surface streams. Nitrogen and phosphorus from direct precipitationwere extremely small in comparison to stream inputs making up only
2 and 5% of the total respectively. There were a few early measurements of N
2
fixation but the data suggested that the rates were low and fixation was not included in the budget
(Alexander et al. 1989).
Mass balance calculations using surface water input and outputs for Na, a conservative tracer, were within 3% suggesting that the water budget was well constrained. Using the same surface water input/output data, it was calculated that about
18% of the N and 30% of the P was retained in the basin.
5

More recent research in lakes in the vicinity of Toolik Lake now shows that both benthic and pelagic habitats support rates of N
2
fixation that are significant to whole-lake function and to N budgets, especially in small lakes with low stream inputs (Gettel 2006).
Summer-time epilimnetic rates of benthic and pelagic N
2 fixation ranged from 0.12 – 1.5 mg N m
-2
day
-1 (x¯= 0.51; n=12 lakes) and 0 – 2.56 mg N m -2 day
-1 (x¯=1.61; n=12 lakes) respectively. Pelagic N
2
fixation rates integrated over the entire water column are higher than reported elsewhere (x¯ = 4.03 ± 0.83 mg N m -2 day
-1
; ranging 0 – 10.4 mg N m
-2 day
-1
) and comprised up to 75% of N input to Fog 2. Benthic N
2
fixation is lower than has been reported for other oligotrophic systems, but is roughly equivalent to annual
N inputs from atmospheric deposition when scaled over an entire year (~25 mg N m
-2 year
-1
). When benthic and pelagic rates are integrated to provide areal estimates on an annual basis, results range from 22 to 696 mg N m
-2
year
-1
, and exceed or equal N inputs from fixation measured in the terrestrial landscape (19 – 255 mg N m -2 year -1 ; Chapin and
Bledsoe 1992).
A model of nutrient cycling in Toolik lake (Whalen et al. 1988; Whalen and
Cornwell 1985) suggested that primary production was driven largely by pelagic recycling of nutrients. The reduced importance of sediment-supplied nutrients was not due to the lake depth, but rather the very low release rate of nutrients from the sediments.
Nitrogen release from the sediments was estimated through the use of benthic chamber measurements. Results of chamber incubations indicated shallow sediments only released 0.005-0.01 mmol N m
-2
d
-1
. This nitrogen release rate was similar to that needed to explain the small accumulation of dissolved nitrogen present under the ice at the end of the winter season. Annual rates of primary production in Toolik Lake indicated that
80%-90% of the N requirements of the phytoplankton needed to come from water column recycling. Benthic release of phosphorus from sediments of Toolik lake were extremely low. Any phophorus released during diagensis was adsorbed onto iron oxides which are abundant near the sediment surface (Cornwell 1987). Conclusions from these early studies of nutrient fluxes indicated that water column water column processes supplied virtually all of the phosphorus and most of the nitrogen required by primary producers in these lakes.
More recent studies in a wide variety of lakes have confirmed that nutrient release from most sediments is extremely low (Kipphut, Giblin unpublished). Most lakes show essentially no N or P release during the summer under ambient conditions. The exceptions are in the few lakes which experience low bottom water dissolved oxygen conditions or prolonged anoxia due to meromixis. In these cases some N release may be observed, and even more rarely, some P release. Lakes undergoing fertilization do begin to release nutrients but it may take several years (see box below; Obrien et al. 2005).
The pelagic recycling of nutrients is dominated by zooplankton excretion.
Johnson (2009) estimated that excretion by macrozooplankton could account for 10-80% of demand for nitrogen by photosynthetic phytoplankton. Rates of nutrient excretion were highest in lakes that were dominated by small-bodied copepods. Excretion by fish amounted to less than 5% of nutrient recycling rates estimated for macrozooplankton.
Little information has been collected on excretion rates of microzooplankton, but these rates are comparable to macrozooplankton rates in temperature lakes (Vanni et al. 2003).
6

The biological processing of nutrients in lakes is responsible for the spatial patterns of dissolved and particulate nutrients present in lakes within a direct series of surface drainage (Kling et al. 2000). In the lake-stream series along the main inlet to
Toolik Lake, in-lake processing reduced dissolved concentrations of carbon and nitrogen; resulting in higher concentrations of particulate carbon and nitrogen leaving the lakes.
Accumulation of carbon and nitrogen by phytoplankton was responsible for this transformation. The temporal synchrony of changes in dissolved and particulate nutrients in this series of lakes highlights the importance of viewing lakes processes within a landscape context.
Measurements of carbon cycling in lakes near the Toolik Field Station demonstrate the degree to which arctic lakes function as a source of carbon dioxide for the atmosphere (Kling et al. 2001; Cole et al. 1994: Rouse et al. 1997). High concentrations of organic matter in tundra soils and the input and respiration of the organic material in the lakes is responsible for the super saturation of CO
2
in surface waters, and subsequent transfer to the atmosphere. The transport of carbon from terrestrial systems to the lakes contributes to the high concentrations of CO
2
in surface waters and is discuss in Chapter xxx.
Anne’s story on sediments and nutrient fluxes?
Biological Properties of Arctic Lakes
Food webs of arctic lakes are relatively simple with low levels of species diversity for all trophic levels. Figure 5a depicts a simplified description of the food web for
Toolik Lake. This food web is similar to other lakes near the Toolik Field Station. Some of the smaller lakes do not support the fish species. The overall structure of the food web in these arctic lakes is similar to those of temperature lakes, except that the potential fish species present are greatly reduced in the Arctic. The relatively high percentage of lakes that cannot support permanent fish populations (Figure 3) allow for greater occupation of the highest trophic levels by invertebrate predators.
Bacterial Processing of Terrestrial Material in Lakes
Organic matter in particulate and dissolved form dominates the nutrient and carbon budgets of arctic surface waters. While the response of aquatic organisms to dissolved nutrient input from land is well understood, the response to particulate material washed in from land is less clear. Decomposition through bacterial activity appears to be limited by the supply of organic substrate rather than by grazing of bacteria by protozoans in all but the most eutrophic systems. Some of this organic substrate originates in terrestrial environments and some is produced within aquatic environments as exudates from algae and zooplankton. The relative importance of these two sources to bacterial activity varies, in part as a result of the amounts of dissolved organic matter (DOM) available from each source and in part as a result of the quality of the DOM.
7

Contrary to these expectations, results of an enclosure study in Toolik Lake
(Rublee and Bettez 1991) demonstrated that, contrary to our expectation, some fraction of terrestrial DOM leached from surface plant litter contains labile compounds and strongly stimulates bacterial activity in the lake. Because most arctic lakes are oligotrophic and in-lake sources of DOM are limited, bacterial activity and decomposition appear to be regulated by the inputs of terrestrial DOM in these systems. In addition we replicated the results of this experiment by measuring DOC inputs to Toolik Lake and the corresponding bacterial activity. During spring runoff DOC concentrations increase and strongly stimulate bacterial activity in the lake. However, during storm events later in the summer when DOC inputs to the lake also increase, there is no corresponding increase in bacterial activity. The reasons for this pattern are unclear, although they may involve the changing quality of DOC leached from land at different times of the year, or the changing bacterial composition in the lake as the season progresses. Finally, there is also evidence that in arctic lakes this terrestrial material initially incorporated by bacterial secondary production is made available to higher trophic levels through trophic interactions in the microbial food web (Rublee 1992; Kling 1994), as has been found in other aquatic systems (Azam et al. 1983; Stockner and Porter 1988).
Primary Producers
A lake’s overall primary productivity depends on the supply of both nutrients and light. Supply and availability of these resources is governed by several factors, including geological setting, lake morphometry, biological interactions and hydrodynamic processes that are driven by climatic forcing. Primary production in pelagic region of these lakes has been studied extensively since the 1970’s. Recent studies on benthic primary production indicate that benthic algae are responsible for most of the primary production that occurs in shallow lakes, and a considerable portion of production in the larger lakes.
Pelagic primary production – Rates of primary production in surface waters in lakes near the Toolik Field Station range from approximately 5-50 mgC m
-3
d
-1
(Kling et al. 1992). These daily rates were estimated using 14 C techniques (Peterson et al. 1989) and are similar to rates measured in oligotrophic lakes in the temperate zone. Annual rates of primary production in arctic lakes are low because of the shortened season of open water. Rates of primary production are highest shortly after ice-out and decrease as thermal stratification becomes more pronounced in July of most years (MacIntyre et al
2006). Depth profiles of primary production indicate that substantially more than half of this production occurs in the epilimnion (Fig. 8). Although many of the study lakes support chlorophyll maxima at metalimnetic depths, the high light extinction in these stained lakes reduces metalimnetic rates of photosynthesis.
Integration of measured rates of 14 C uptake leads to areal estimates of primary production in our study lakes that range from 20-200 mg C m
-2
d
-1
. This variation is derived mostly from seasonal changes in production within lakes rather than from differences among lakes. Shallow lakes tend to have slightly greater rates of primary
8

production measured on a per liter basis, but the integration of deeper waters in the larger lakes makes areal rates of primary production similar across the landscape.
Mary Ann's story on internal waves
The effect of glacial history of the watersheds in the region had little effect on the algal biomass of the lakes (Fig. 9). Observed differences in chlorophyll concentrations among lakes were more influenced by lake depth than by glacial history. A multiple regression of lake surface area, maximum depth, glacial history, and specific conductivity on mean summer epilimnetic concentration of chlorophyll a indicated that lake depth was the only variable that was significantly correlated with chlorophyll concentrations. The relationship was negative but weak, with depth explaining only 22% of the variation in chlorophyll. Interestingly, neither glacial history nor specific conductivity was significantly correlated with chlorophyll a concentration of epilimnetic water in these lakes. That being said, the only deep lake present on the oldest glacial surface (Lake E5) exhibited low specific conductivities and had some of the lowest concentrations of epilimnetic chlorophyll a. Shallow lakes on the older glacial surfaces had moderate chlorophyll concentrations and were similar to lakes in recently glaciated watersheds.
Chlorophyll concentrations were higher in shallow compared to deeper lakes in our study area. Surveys of small, medium and large lakes from 1999 to 2003 indicated that small and medium lakes had mean summer chlorophyll a concentrations in epilimnetic waters of more than 1.5 mg m -3 (Fig. 9). These values were approximately
50% higher than epilimnetic chlorophyll concentrations of the larger lake in the survey.
The lakes deeper than 5 m often developed deep chlorophyll maxima during the summer months. Metalimnetic and hypolimnetic chlorophyll concentrations were frequently more than twice values measured in epilimnetic water (Fig. 8). Due to light limitation, the high chlorophyll concentrations in the deeper strata did not contribute substantially to the primary production of the lake.
Phytoplankton species composition has not been routinely analyzed from lakes in this region. Samples from Toolik Lake collected in 1991 and 1992 indicated that the biomass of phytoplankton during summer was dominated by chrysophytes (50-70%) and dinoflagellates (10-25%) (O’Brien et al. 1997). Cryptomonads occasionally comprised up to 28% of the phytoplankton biomss. Diatoms, chlorophytes and cyanobacteria were rare in Toolik comprising less than 5% of the biomass.
Comparison of the phytoplankton in shallow and deeper lakes was examined in
2002. The abundance and biomass of these pelagic primary producers differed in lakes of different morphometries. The shallow lakes (E6 and Fog4) had greater phytoplankton diversity compared to two deeper lakes (E5 and Fog2). The addition of Dinobryon sp., several dinoflagellates, and increased cyanobacteria was responsible for the greater diversity (Fig. 10). These additional taxa are generally resistant to grazing and are likely selected for in the shallow lakes with higher densities of cladoceran zooplankton.
Burkart (2007) reported that grazing rates of macrozooplankton were 20-50% greater in
9

deep compared to shallow lakes. The observed differences in phytoplankton composition were likely responsible for the differences in measured grazing rates.
Benthic primary production – Low chlorophyll concentrations in surface waters generally lead to high water clarity and sufficient light penetration to support extensive benthic primary production (Ramlal et al. 1994; Vadeboncoerur et al. 2003). In lakes near the Toolik Field Station, benthic production comprises the majority of lake primary productivity. The extent of benthic primary production depends on lake morphometry and water clarity (Fig. 11).
Chlorophyll concentrations in shallow (<4m) soft sediments range from 100 to over 600 mg m
-2
. These values typically exceed the total pelagic chlorophyll in the overlying water by 2-20 times. Chlorophyll concentrations are much lower on cobble and gravel substrates but these substrates tend to make up a small portion of the sediments in the (Cuker et al. 1983). Sediment chlorophyll concentrations within lakes can show significant inter-annual variation (Gettel in review) and are not correlated well with the age of the surrounding landscape.
Benthic primary production in shallow sediments ranges from 25 to 250 mg C m
-
2 d
-1
(Fig. 12). When integrated over the growing season these rates are similar to those measured in the overlying water column. Rates of benthic primary production drop off rapidly with depth and there is little net GPP below 6-7 meters except in the clearest lakes. At all but the very shallowest depths, GPP appears to be light limited, even though benthic primary producers reach maximal photosynthetic rates at relatively low light intensities (Fig on hand out or similar). Low average surface irradiance at these high latitudes and the high rates of light attenuation due to high concentrations of DOC likely contribute to the high light gathering efficiencies of benthic algae in these lakes (average growing season PAR 323-500 uE s -1 in Gettel submitted). In the larger lakes, low light conditions limit rates of benthic primary production on sediments deeper than 5 m.
Benthic N
2
fixation is controlled most strongly by the availability of phosphorus
(P) and to a lesser extent light availability and grazing by the snail Lymnaea elodes
(Gettel et al. 2007; Gettel 2006). Results from nutrient and grazer manipulations suggest that P stimulates benthic N
2
fixation while snail grazing slightly suppresses it. These controls play out at the landscape scale in the Toolik Lake area across patterns of glacial history. Benthic N
2
fixation is generally higher on younger-surface lakes (de-glaciated
12,000 years ago) where P concentrations are higher than on old-surface lakes (deglaciated >25,000 years ago). This pattern is evident despite the fact that Lymnaea elodes, which slightly suppresses benthic N
2
fixation, is absent on old-surface lakes due to a lack of calcium availability (Gettel 2006).
Other controls on benthic production are beginning to be examined. Production on soft sediments does not seem to be significantly affected by grazing by snails (Gettel et al. 2007) but snails may play a bigger role on cobbles where benthic producer biomass is much less (Cuker et al. 1983). While nutrients could stimulate benthic production, studies in other lakes suggest that the stimulatory affect of nutrients on benthic
10

production is countered by the decreased light reaching the bottom when phytoplankton on stimulated.
Microzooplankton
The microplankton community in the lakes around Toolik is composed primarily of ciliated protozoans, small particle feeding rotifers, and zooplankton naupli. The ciliates, mainly oligotichs < 50 µm in size, are dominated by the three genera
Halteria,
Srombidium, and Strobilidium while the rotifers are dominated by fours species Keratella cochlearis, Kellicottia longispina, Polyarthra vulgaris, and Conochilus unicornis (Rublee
1992).
In general, the biomass of the microplankton in these oligotrophic lakes is < 10 µg
C l
-1
with the mean biomass of the protozoans usually < 0.5µg C l
-1
(~150 ind l
-1
), the rotifers ~ 2-3 µg C l -1 (100 ind l -1 ) and the remainder, 7-8 µg C l -1 (~7 ind l -1 ) being naupli. However, the biomass can be higher in more productive lakes (up to 30 µg C l
-1
)
(Rublee 1992, Rublee and Bettez 1995), and can vary within a lake i.e. from 1989 – 1999 in Toolik it varied from 3-9 µg C l -1 (Rublee and Bettez 2001) (Fig. 12).
The spatial and temporal patterns also vary by the different members of the microplankton community. The protozoan numbers are highest (up to ~1000 individual’s l
-1
and 1-2 µg C l
-1
biomass) in the epilimnion early in the summer and generally decrease with depth and over time. The rotifer population however, doesn’t correlate with depth and is low in the early summer (10-300 indiv l
-1
) and may increase 5 to 10-fold by mid
July or early August. The naupli biomass equaled or exceeded rotifer biomass and showed a positive correlation with depth and time (Rublee 1992).
Macrozooplankton
Species of crustacean zooplankton are low in this region with only seven common pelagic species present in over 100 lakes that have been sampled (O’Brien et al. 2004).
The glacial history of the watersheds of this region has an impact on a few of these populations with Holopedium giberum being present more often in lakes on the older glacial surfaces. O’brien et al. (2004) report negative correlations between
Holopedium abundance and conductivity of surface waters in the lakes of the Toolik region. This correlation has been reported for Holopedium from other regions of the world (Tessier
1980). In contrast to H. gibberum, Diaptomus pribilofensis occurs more frequently in lakes of higher conductivity. In a comparison of crustacean zooplankton biomass in seven lakes on recently glaciated landscapes compared to five lakes on older glacial surfaces, biomass of H. gibberum was 10-fold greater in lakes on older glacial surfaces
(Burkart 2007).
Lake size, depth, and the presence of fish has a pronounced effect on the abundance and distribution of crustacean zooplankton in the study lakes. Daphnia middendorffianna and Heteorcope septentrionalis are more abundant in small compared to medium and large lakes (Fig.14). These two large bodied zooplankters are likely eliminated by fish predation in the medium and large lakes. O’brien et al. (2004) report
11

that densities of D. middendorffiana were twice as high in fishless lakes compared to lakes with fish. In a more detailed study of 12 lakes, Burkart (2007) indicated that total crustacean zooplankton biomass was 5-fold higher in lakes without fish compared to lakes with grayling, lake trout, or sculpin. Surveys of lakes near the Toolik Field Station indicate that densities of Daphnia middendorffiana, Heterocope septentrionalis, and
Diaptomus pribilofensis were higher in small lakes and that Daphnia longerimis and
Bosmina longerostris were more abundant in medium and larger lakes (Figure 8).
INSERT BOX ON PHOTOPROTECTION AND FISH PREDATION
Invertebrates predators are important in structuring zooplantkon assemblages, particularly in fishless lakes. Heteorcope septentrionalis , Cyclops scutifer , Chaoborus and Hyalella azteca at times feed extensively on herbivorous zooplankton. Laboratory feeding trials and field enclosure experiments indicated that the presence of H. septentrionalis determines the composition of herbivorous zooplankton in the ponds and lakes of the region (O’Brien et al. 1978; Luecke and O’Brien 1981). An assessment of isotopic composition of these copepod predators indicated that their feeding habits ranged from mostly carnivory to almost complete herbivory (Kling et al. 1994). Differences in prey availability among lakes likely resulted in the observed variation.
Fish distributions
Only six species of fish inhabit the lakes and streams near the Toolik Field
Station. Lake trout ( Salvelinus namaycush ), burbot ( Lota lota ) and Arctic char
( Salvelinus arcticus ) occupy the highest trophic levels feeding extensively on benthic invertebrates, but occasionally becoming piscivores. Arctic grayling ( Thymallus arcticus ) and round whitefish ( Prosopium cylindraceum ) feed on benthic invertebrates and occasionally zooplankton, and slimy sculpin ( Cottus cognatus ) live on substrates of most lakes feeding on benthic invertebrates. Extensive analyses of the distributions of these fish species have been performed by Anne Hershey and many collaborators. Their analyses of lake morphemetry and landscape position indicated that lake trout and arctic char occupied only deep lakes that were part of well-established drainage systems
(Hershey et al. 2001, Fortino et al. 2005). Lake trout and round whitefish were most often found in lakes of the Itkilik drainages and Arctic char were most often present in lakes with drainage connections to the Sagavanirkok River. The distribution of grayling and sculpin indicated high dispersal abilities in that these two species were present in all watersheds sampled and were the only fish taxa found in lakes with weak drainage connections to other lakes (Hershey et al. 2006).
Lake trout function as a top predator in these lakes, feeding on a mix of invertebrate and vertebrate prey. Investigations of spatial distributions and experimental manipulation indicated that the presence of lake trout have large impacts on the abundance and distribution of their prey, but also exhibit significant effects on lower trophic level organisms. In a detailed set of investigations researchers demonstrated that lake trout reduce numbers of slimy sculpin and restrict remaining sculpin to near-shore habitats (Hanson et al. 1992; Goyke et al. 1992). A variety of benthic invertebrates preyed on by sculpin respond positively to the presence of lake trout. This result was one
12

of the first where a trophic cascade occurred because of the presence of a salmonid predator (McDonald et al. 1992; Hershey et al. 2001).
Section 2. Experimental results and Analyses of Lake Function
Initial experiments on the limnological processes that define arctic lakes were set up to assess effects of land use changes on lake productivity. A series of experiments to assess the effects of increased nutrient loadings into arctic lakes was intiated. These experiments spanned spatial scales ranging from batch bioassays to assess short-term nutrient limitation to large mesocosm experiments to whole lake manipulation of nutrient income. The goals of these experiments were designed to assess the degree of nutrient limitation of phytoplankton to specific nutrients to the response of the lake ecosystem to increases in nutrient income.
The focus of these experiments was especially relevant as climate warming became recognized as the major environmental change occurring in the Arctic (ACIA
2004). The deepening thaw of the permafrost near the Toolik Field Station (as inferred from changes in strontium isotopic ratios, Chapter xxx) likely resulted in the observed increase in alkalinity of surface waters. Our ability to detect changes in concentrations of dissolved nutrients in surface waters is low because of analytical detection levels
(O’Brien et al. 1997), but the increase in alkalinity is very likely associated with an enrichment of nitrogen, phosphorus and DOM into these lakes. Results from the following experiments provide insight into the effects of these chemical changes on the productivity of arctic lakes.
Studies on nutrient limitation
Initial studies of nutrient limitation addressed the question of which nutrients limited phytoplankton growth rates in Toolik Lake (Whalen and Alexander 1986). Batch bioassays indicated that the degree of nutrient limitation varied seasonally but most often indicated that phytoplankton were limited by both nitrogen and phosphorus.
Occasionally nitrogen was indicated as the limiting nutrient, but rarely did additions of phophorus alone stimulate phytoplankton growth.
Comparison of nutrient limitation in 39 lakes near the Toolik Field Station indicated that the growth of lake phytoplankton was limited by both nitrogen and phosphorus. Primary productivity in short term batch bioassays increased most when both N and P were added to containers, with average increases in production of over 60% in all 54 experiments (Levine and Whalen 2001). Nitrogen limitation was more common than phosphorus limitation in these experiments. Results of a typical bioassay are shown in Figure 14 for lake Fog 2. In this example, additions of nitrogen stimulated phytoplankton growth more than did additions of phosphorus. The greatest response was seen in treatments with both N and P additions. Results from these batch bioassays reveal two other interesting observations. The increase in chlorophyll concentrations in many of the control treatments compared to chlorophyll concentrations present at the
13

beginning of the experiments suggest that grazing by zooplankton has a measurable impact on phytoplankton biomass. Also, the decline in chlorophyll concentration in treatments where phosphorus was added alone suggests that bacterial populations may be
P-limited and respond numerically and out-compete the photosynthetic producers in these treatments.
Levine and Whalen (2001) found several correlations between landscape attributes of watersheds and results from the nutrient enrichment assays. Only 6 of the 54 bioassays failed to show significant effects of additions of nitrogen and/or phosphorus, but 5 of these six experiments occurred in headwater lakes, suggesting that lake phytoplankton rapidly take up dissolved nutrients from the watershed and export particulate nutrients to downstream lakes (Kling et al. 1987). Shallow lakes and lakes with higher alkalinities demonstrated a greater degree of P-limitation.
Mesocosm experiments
The predominance of co-limitation of phytoplankton growth by nitrogen and phosphorus shaped the design of larger scaled experiments to assess the impact of increased nutrient loading to the lake ecosystems. In 1983-85 a limnocorral experiment was established in one of the deep bays of Toolik Lake (O’brien et al. 1992). Up to six circular limnocorrals, open to the atmosphere, isolated approximately 70 m3 of water, and enclosed 20 m
2
area of benthic sediments. A floating collar on each limnocorral kept waves from entering the structures and an anchor of chain buried in the sediments kept the walls of the enclosures from shifting (photo xxx). Three nutrient treatments (no addition, low level fertilization and a high level of fertilization) were coupled with two levels of planktivorous fish (grayling). The low level nutrient addition was equivalent to
1 times the ambient nutrient loading of nitrogen and phosphorus to Toolik Lake (Whalen and Cornwell 1986) and the high level of nutrient addition either 5 or 10 times that amount depending on the year. Fish were added to some limnocorrals as either free ranging individuals or as caged individuals to reduce their consumption of zooplankton.
In either event, the fish addition treatments likely exceeded natural abundance of fish in these lakes. These mesocosms were sampled for physical, chemical and biological parameter (O’Brien et al. 1992).
The major result of the limnocorral experiments was that phytoplankton responded dramatically to additions of nitrogen and phosphorus. Measurements of pelagic primary production were 10-20 fold higher in limnocorrals that received nutrients. These results indicated that not only were phytoplankton in these arctic lakes limited by N and P, but that sustained additions of these two nutrients had the effect of greatly increasing the biomass and primary productivity of Toolik lake water.
The increase in phytoplankton had dramatic effects on physical and chemical parameters in the limnocorral. The increased biomass of phytoplankton decreased light penetration into the corrals receiving nutrients as Secchi transparency declined from five m in reference limnocorrals to less than 2 m nutrient addition treatments. Both pH and oxygen concentrations increased greatly in the limnocorrals receiving nutrient additions due to the increases in rates of photosynthesis.
14

Response of heterotrophic organisms to the dramatic increases in phytoplankton abundance was mixed. Bacteria and their microzooplankton grazers increased greatly in limnocorrals receiving nutrient additions (Fig. 13) (Rublee 1992). The response of macrozooplantkon grazers however was mixed, likely due to predation by fish in the fish treatments. When fish predation was absent present at low levels, larger cladocerans increased in abundance in response to higher biomass of phytoplankton in the fertilized treatments. The long and complex life histories of the pelagic copepods and many of the benthic invertebrates likely reduced their response to higher abundances of phytoplankton
(Hershey 1992).
Whole-Lake Experiments
In order to more fully understand the effects of nutrient additions on arctic lake ecosystems, a series of whole-lake fertilization experiments were conducted near Toolik
Lake. Results of these experiments allow inferences concerning response of fish and benthic invertebrates that could not be incorporated in the limnocorral experiments. In
1985 Lake N2 was divided in half by a polyethylene curtain and the downstream end of the lake was fertilized with phosphoric acid and ammonium nitrate such that annual loading of these two nutrients was 5 times ambient values (O’Brien et al. 2005). This fertilization experiment ran for 6 years and then was examined in subsequent years to assess the time scale of recovery from eutrophication.
In all six years of nutrient addition, phytoplankton biomass and productivity were greater in the treated sector than the reference sector (Fig. 15). In the first four years of nutrient addition there was no flux of phosphorus from the mineral-rich sediments. This changed in the last two years of nutrient addition as phosphorus was released to the lake
(Fig. 16).
The response of the animal community to increased plant production was mixed.
One of the four macro-zooplankton species ( Daphnia longiremis ) increased in number by about two-fold in the first five years. However, the copepod Cyclops scutifer showed no response during the treatment phase of the study (O’Brien et al. 2005). Benthic snails responded significantly to increased nutrient loadings in the fertilization experiment of
Lake N2 (Hershey 1992). Both Valvata lewisi and Lymnaea elodes increased in abundance on the fertilized side of this divided lake. The lack of response of chironomid larvae suggests that chironomids may be limited by predatory sculpins rather than food in these arctic lakes. Because sculpins fed primarily on chironomid larvae in this lake, it was not surprising that sculpin growth rates were similar in the fertilized and control sections of the lake. This lack of response of chironomid larvae and sculpin growth was similar in both the limnocorral and divided lake fertilization experiments and suggests a potential decoupling of pelagic and benthic components of food webs in arctic lakes.
A second whole-lake fertilization experiment was conducted on Lake N1 from
1990-94. Nitrogen and phosphorus were added to this lake during summer months at a rate of approximately 4 times ambient loadings. Lake N1 is deeper than Lake N2 and
15

contains four of the six fish species present in the region: lake trout ( Salvelinus namaycush ), Arctic grayling ( Thymallus arcticus ), burbot ( Lota lota ), and slimy sculpin
( Cottus cognatus ). Physical, chemical and biological parameters were examined each summer to assess the effects of the fertilization treatment.
As in previous nutrient addition experiments, phytoplankton biomass and primary production increased dramatically in response to additions of nitrogen and phosphorus
(Fig. 17, from Fig. 1 of Lienesch et al.2005) Chlorophyll concentrations and rates of primary production increased more than 10 fold compared to prefertilization measurements. These increased rates of primary production were cumulative during the five years of the experiment and eventually resulted in the development of an anoxic zone in the bottom of Lake N1 in 1994 (Lienesch et al. 2005).
As in the fertilization experiment in Lake N2, snail abundances in Lake N1 increased substantially during fertilization years. Densities of both Valvata lewisi and
Lymnaea elodes were four times greater after the experimental fertilization. Lienesch et al. (2005) report that growth rate of lake trout responded positively to this increase in snail abundance (Fig. 18). The lack of a numeric response of lake trout to the fertilization was attributed to the effects of low oxygen conditions on the survival of lake trout eggs.
Most limnological parameters returned to pre-fertilization values within 1-2 years after the nitrogen and phosphorus additions ceased a the end of 1994. The exception to this statement was hypolimnetic oxygen concentrations which remained low through the summer of 1998 (Lienesch et al. 2005). The accumulation of sedimented phytoplankton during the fertilization years continued to provide a labile substrate for bacterial respiration long after the fertilization was ended. These results demonstrate the importance of maintaining long-term assessment of experiments to gain an understanding of processes that operate under very different time scales.
The response of microzooplankton to the various fertilization experiments was mixed. In the mesocosm (O'Brien et al. 1992) and whole lake experiments (O'Brien et al.
2005) changes in microplankton abundance was related to the degree of fertilization, although the response and recovery of the microplankton differed in the various experiments (Fig. 13). In the Limnocorral experiment in which nitrate or ammonium were added at 10x ambient loading there was up to a 10-fold increase in microplankton biomass while in lake N2 which was fertilized with N and P at 5x ambient, the biomass doubled (20 µg C l -1
) and in lake N1 which was only fertilized with N and P at 3-4x ambient the biomass increased 5-fold (50 µg C l
-1
) (Rublee 1992, Rublee and Bettez
1995, Rublee and Bettez 2001).
In Lake N1, the only lake in which we have data before, during, and after the experiment, there were shifts in both the number of individuals and types of species present throughout the experiment. The microplankton community shifted from one dominated by small particle feeding rotifers (mainly Conochilus unicornins) during the first year of fertilization to bacteriovorus peritrich protozoans ( Epistylus rotans, and
Vorticella sp .) during the second year, and finally to the predatory rotifers Synchaeta,
16

Polyarthra, and Trichocerca and zooplankton naupli during the third fourth and fifth years of the fertilization (Rublee and Bettez 1995, Rublee and Bettez 2001, Bettez et al.
2002).
While there was no significant change in the biomass of the microplankton during the first year, in the second and third years of the fertilization (1991 &1992) significant increases in the number and biomass of the protozoans were observed. A change in the temporal trend of this response also occurred. Instead of the highest values being found early in the summer and decreasing over time, the protozoan biomass increased by 3 orders of magnitude (up to 150 µg C l -1
) during a late summer bloom of a previously seldom seen species Epistylus rotan s (Rublee and Bettez 1995). In the 4th and
5th years of fertilization (1994 & 1995) the number and biomass of rotifer species found prior to fertilization decreased while the abundance of predatory rotifers increased to as high as 5.7 x 10
3 indivdiuals l
-1
, accounting for 98% of total rotifer biomass.
In contrast to Lake N1, microplankton in Lake N2 did not undergo similar species shifts. During the last two years of fertilization in Lake N2, total microplankton biomass increased in the fertilized treatment but community composition was similar to that found on the unfertilized reference side of the lake (Rublee and Bettez 2001).
With the end of fertilization the biomass of the microplankton returned to pretreatment levels after 1 year in Lake N1 and two years in Lake N2. In Lake N1 in the first year of recovery, the predatory rotifers decreased to their pre-fertilization levels and the composition of the microplankton returned to being dominated oligotrichs and small particle feeding rotifers. In Lake N2, the microplankton biomass did not return to pre fertilization levels for 2 years, likely due slower release of acquired nutrients from the more organic sediments of Lake N2, which maintained elevated chlorophyll and primary productivity in the first year of recovery (Rublee and Bettez 2001, O'Brien et al. 2005).
Thus, the differences between the response and recovery times in the two lakes are likely due to differences in their morphology and sediment composition.
Mesocosm experiments and whole-lake fertilization of Lakes N1 and N2 indicated that phytoplankton and bacteria responded dramatically to nutrient increases of
4-10 times ambient loading rates. The mixed response of higher trophic level organisms was potentially confounded by the large differences in light and oxygen conditions present in fertilized treatments. In 2001 a low level fertilization experiment of a deep and a shallow lake was initiatied to assess the degree to which likely increases in nutrient loading due to climate warming and associated environmental changes. Fertilization at 2 times ambient loading was begun in Lake E6 (maximum depth of 3m , fishless) and Lake
E5 (maximum depth 12 m, with Arctic char) to assess the degree to which hypothesized increases in primary production would result in increased biomass and productivity of zooplankton and fish. Lakes Fog2 (deep) and Fog4 (shallow) were chosen as reference systems. In these whole-lake fertilization experiments, small amounts of 15N were added along with the nitrogen and phosphorus to serve as a tracer for this additional nutrient.
This tracer allowed us to assess the time lags inherent in how different food web components are able to utilize the added nutrients.
17

This low level fertilization of Lakes E5 and E6 is ongoing. Preliminary results indicate that the fertilization treatments had immediate effects on phytoplankton populations. Chlorophyll concentration and rates of primary production of phytoplankton increased after the initiation of the fertilization treatments. These increased levels have been enhanced each year indicating that the effects of fertilization during one summer carry over to subsequent years. This carry over occurs even though residence time of the water in these lakes is much less than 1 year (George Kling, unpublished data).
As in the other whole-lake fertilization experiments, the effects on higher trophic levels have been less obvious. Some taxa of crustacean zooplankton have increased in abundance in Lakes E5 and E6 relative to the reference lakes, but only in years with higher than average epilimnetic temperatures. Temperature effects on the egg development of cladocerans appears to limit their population response to added food during cooler summers.
In the fertilized lakes, the phytoplankton and zooplankton took up the
15
N tracer during the first summer. The concentration of
15
N in phytoplankton became saturated within 10 days of the initiation of the fertilizer dripper. Different zooplankton taxa accrued the
15
N label at different rates with the faster uptake occurring in cladocerans and slowest uptake in cyclopoid copepods. Detectable increases in
15
N was not observed in benthic chironomids or snails until the second summer of the experiment. Increases in
15 N concentrations of Arctic char were not observed until the third summer of the nutrient additions. This lagged response of the benthic components of the lakes food web is providing an opportunity to assess linkages between pelagic and benthic systems in these arctic lakes.
Consumer effects on lake processes
The effect of predation on structuring arctic lake communities was initially examined by John O’Brien and members of his research laboratory (O’Brien 198x).
Investigations of the feeding rates and prey selectivity of individual predators were conducted under laboratory conditions. These investigations reveal that both vertebrate and invertebrate planktivores had dramatic impacts on zooplankton assemblages.
Invertebrate predators, particularly the large calanoid copepod Heterocope septentrionalis
, fed selectively on smaller cladocerans with weak and carapaces (O’Brien et al. 1978; Kettle and O’brien 1980, Luecke and O’brien 1983). In lakes where H. septrionalis was abundant, the predominant zooplankton possessed a variety of defenses such as gelatinous coverings ( Holopedium gibberum ), extended helmets ( Daphnia longeremis ), or behaviors that prevented effective consumption ( Bosmina longirostris and Diaptomus pribilofensis ). In water bodies where H. septentrionalis was not present, the dominant herbivorous zooplankton was Daphnia pulex , a highly vulnerable prey item.
Rates of H. septentrionalis feeding on D. pulex in the laboratory indicated that predation
18

rates could exceed reproduction of D. pulex under field conditions (Luecke and O’brien
1983).
Studies on the feeding selectivity of planktivorous fishes indicated that large and heavily pigmented zooplankton were fed upon preferentially by both juvenile grayling
(Schmidt and O’brien 1982) and juvenile lake trout (Kettle and O’brien 1979). A simulation model of Arctic grayling feeding (O’Brien and Evans 1992) indicated that predation rates on large-bodied zooplantkon like D. middendorffiana would be sufficient to explain the differences in zooplankton assemblages present in fish and fishless lakes near the Toolik Field Station (O’Brien et al. 2004). Results of the laboratory experiments and modeling simulations also explained why the limnocorral treatments that contained fish (O’Brien et al. 1992) had decreased abundances of
D. middendorffiana and H. septentrionalis . These results help explain why the presence of fish has a dramatic impact on zooplankton assemblages in these arctic lakes, but that diet composition and analyses of stable isotopes of carbon and nitrogen indicate that zooplankton comprise a very minor part of both grayling and lake trout diets. The high selectivity of the juvenile stages of these two fish species for large-bodied zooplankton greatly reduces these prey in lakes with high fish densities. After the large-bodied zooplankton are nearly eliminated, fish begin feeding on benthic invertebrates.
This conceptual model of the impact of fish predation is further supported by results of several whole lake food web manipulations. In 1988-89 most of the lake trout greater than 300 mm were removed from Lake NE12 through intensive gillnetting. These larger lake trout are potential piscivores and feed to some degree on small sculpin and juvenile grayling. After the lake trout removal, juveniles of both grayling and lake trout began consuming more pelagic zooplankton prey resulting in several very successful recruitment years (Keyse et al. 2007). As these juvenile fishes became more plentiful, their feeding rates on zooplankton resulted in a decrease in the abundance of Daphnia
(xxx).
These complex interactions in arctic lake food webs were further demonstrated by experimental reductions in lake trout density in a series of lakes. In addition to sculpin effects, the removal of lake trout was accompanied by a large increase in recruitment of both juvenile lake trout and arctic grayling. These successful year classes provided forage for burbot, a formerly low density piscivore, whose population increased as an indirect function of lake trout removal. These compensatory changes demonstrate the complexities inherent in food webs in arctic lakes.
John’s Story on whole lake manipulations of zoops
Evidence of recent climate change
Toolik Lake serves as our primary long term monitoring site for lentic systems.
Thermal structure, nutrient, and cation/anion measurements have been made in the water column for the past 30 years. The average July temperature of Toolik Lake at 2 meters
19

varies greatly among years, but shows no overall significant warming trend during the
1975-2005 sampling period (Figure 19). During most of our study period, the Arctic
Oscillation and related Pacific Decadal Oscillation (Mantua et al. 1997) have been in their high phase with occasional shifts to a low phase. These atmospheric pressure oscillations exert a small influence on inter-annual thermal conditions of the lakes in the
Toolik region (Fig. 19). Observations indicate that mid-summer lake temperatures are influenced by local and regional climatic events as opposed to long-term warming trends or decadal oscillations. These differences between years are important for vertical exchanges and the fate of incoming stream waters which introduce phytoplankton, bacteria, and nutrients from higher in the watershed (MacIntyre et al. 2006).
Although mid-summer surface temperatures in Toolik have not increased in recent years, the alkalinity of lake water has increased substantially (Chapter xxx). The early season warming of soil and permafrost along with possible increases in thermokarst activity is likely responsible for the higher alkalinity measured in Toolik Lake. The increase in lake alkalinity is concurrent with changes in strontium isotope ratios , indicating a deepening of the permafrost layers in soils of the region. Both increases in permafrost depth and thermokarst activity create the potential for increased DOC loadings in the lakes. These increases in DOC will enhance lake heterotrophy and decrease autotrophy through light limitation. This pattern will almost certainly result in increased CO
2
saturation in surface waters and enhance evacuation of CO
2 to the atmosphere (Cole et al. 1994).
Future Predictions of Ecosystem Processes in Arctic Lakes
The potential for climate warming in the Toolik region is high. Although indices of summer lake temperatures and lake productivity have not shown dramatic changes at
Toolik Lake since the mid-1970’s increases in air and soil temperatures have been recorded in nearby regions (Chapin et al. 2007). The location of Toolik in the foothills region of the Brooks Mountains increases local meteorological conditions and makes identification of climatic trends more difficult. That being said, several very warm and dry summers have occurred in the region since 2001, resulting in warm lake surface waters and intense thermal stratification of the lakes. In 2007 a 100,000 ha tundra fire occurred 35 km north of Toolik Lake. The effects of this tundra fire on aquatic ecosystems are currently being investigated.
The warming of tundra soils and deepening of permafrost has been shown to increase alkalinity of lakes in the Toolik region (Chapter xxx). This increase in solutes in lake water could contribute to increased loading of nutrients limiting primary production in the lakes and ponds. Increased thermokarst failures due to climate warming has also been shown to increase solute transport to streams and lakes (Gooseff et al. 2009).
Increased solute loading to lakes coupled with warmer summer temperatures will likely influence biotic production in the lakes of the region. Results from whole-lake fertilization studies indicate that phytoplankton abundance and productivity increases
20

with nutrient loading even when relatively small amounts of nitrogen and phosphorus are added to the lakes (Evan 2008). Unlike phytoplankton, zooplankton responded positively to lake fertilization only when temperature conditions allow for sufficient egg development. Zooplankton showed no increase to the fertilization of lakes E5 and E6 when epilimnetic water temperatures remained below 15 C (C. Luecke, unpublished data). Under climate warming predictions outline in the ACIA (2004) report, increased water temperatures will likely be sufficient for zooplankton to respond to increases in primary production.
The differences in thermal stratification and mixing in warm versus cold years affect the pathways of incoming stream waters and the rate at which incoming water reaches the upper mixed layer. Storm events can lead to loading of inorganic nitrogen to the euphotic zone of lakes in excess of phytoplankton needs (MacIntyre et al. 2006;
Evans 2007). Increases in primary productivity are comparable to those caused by wind mixing (Evans, 2007) and can be high. In addition to the impacts on primary productivity, inflows from storms create the potential for vertical structure in biological communities. Nutrient and particulate delivery is time dependent in arctic lakes because the incoming stream waters decrease in temperature during storms due to their interaction with the permafrost. Consequently, water with different concentrations of solutes and particulates are injected at different depths (MacIntyre et al. 2001). Hence, variations in surface heating and cooling as well as mixing potential in the metalimnion influence the percentage of nutrients immediately available for growth in the upper water column as well as the potential for spatial variations in community structure.
Between-year differences in the frequency of these storm events may explain the high variance in previous measurements of primary productivity at Toolik Lake during the ice free period (O’Brien et al. 1997). Summer rainfall at Toolik from 1988 through
1996 was significantly less ( p = 0.0029) than from 1997 though 2004 (Arctic LTER data;
MacIntyre et al. 2006). Discharge > 15 m
3
s
-1 and > 8 m
3
s
-1 occurred in 40% and 67%, respectively, of years from 1992 to 2004. In consequence, increased stream discharge may be associated with global change in the Arctic as observed in some temperate regions (Kunkel et al. 1999). In large lakes like Toolik, the effect of stream inflows will likely be greater than in small lakes with smaller watersheds. Increased melting of permafrost will likely lead to increased nutrient flux into the upper water column of lakes. The fate of these solutes will depend on whether the combination of cooling events and wind forcing of the internal wave field occurs as in Toolik Lake or whether downward mixing depends upon fall cooling. Changes in the prevalence of warm and cold fronts will thus have considerable impact on lake hydrodynamics with concomitant implications for nutrient fluxes and changes in productivity across arctic landscapes.
Future increases in summer lake temperatures and increased thermal stratification of the lakes would greatly influence the structure of biological communities and ecosystem processes of these lakes. A simulation model coupling physical and biological components of arctic lakes (John O'Brien, unpublished data) indicated that increased lake temperatures would greatly reduce the ability of lake trout to thrive in these lakes. The lack of abundant deep-water habitat in most of the lakes of the region means that usable
21

summer habitat for lake trout would decrease by 30% if epilimnetic temperatures rise by
2 degrees C. This reduction of habitat is exasperated further if nutrient loading to the lakes increases phytoplankton productivity resulting in decreases in hypolimnetic oxygen concentrations.
The effect of climate warming on fish populations in the lakes near the Toolik
Field Station may already be happening. Analyses of the length-weight relationships of the common fishes of the Toolik area indicated that summers when epilimnetic temperatures exceed 15 C for extended periods, a decrease in the slope of the lengthweight relationship was apparent for lake trout, arctic grayling, and arctic char (Fig. 20).
Additional increases in epilimnetic temperature would likely result in dramatic declines in fitness of the fishes in these lakes. The shallow nature of many of the lakes coupled with the lack of food resources below the epilimnion may make it difficult for these fish populations to persist.
The net effect of additional climate warming may enhance rates of primary production through a combination of increased nutrient supply and increased water temperatures, but also increases the likelihood that the organisms at higher trophic levels will have difficulties utilizing this extra food resource. The complex life histories of organisms that have evolved to thrive in arctic conditions may fare poorly under a warmer Arctic.
22

Figure 1. Glacial surfaces near Toolik Lake showing three different ages and typical study sites.
Fig. 1 Time before present when landscapes near the Toolik Field Station were last glaciated.
23

Fig.2. Bathymetry of Toolik Lake showing distinct kettle basins derived from ice blocks remaining from the most recent glaciation.
24

10
5
20
15
30
25
0
<12,000 12,000-60,000
Years Since Glaciation
>300,000
Fig. 3. The percentage of lakes with surface areas greater than 4.7 Ha.
Correlation analyses indicated that lakes with surface areas less than 4.7 Ha typically were not inhabited by fish. This surface area criteria correctly predicted the presence of fish in 83% of 46 lakes in the Toolik Lake Research Natural Area where frequent sampling for fish has occurred (Burkart 2007).
25

1
0.75
0.5
NE9B
S7
S11
S6
N2
N1
NE12
0.25
Fog4
0 Fog2
I6
E1
I8
-0.25
I7
I3 I5
-0.5
TLK
-0.75
E6
E5
-1
-1 -0.75
-0.5
-0.25
0 0.25
MDS 1 - Lake Size
0.5
0.75
1
Fig. 4 Multiple dimensional scaling of lakes near the Toolik Field Station based on 24 limnological characteristics.
26

Fig. 5 Thermal stratification of Toolik Lake during the cool summer of 2003 (top) and the warm summer of 2004 (bottom). Julian date is plotted on the x-axis from ice out in late June to ice cover in fall.
27

200
150
100
50
0
12,000 60,000 300,000 I series Toolik
2.0
1.5
1.0
0.5
0.0
12,000 60,000 300,000 I series Toolik
1500
1000
500
0
12,000 60,000 300,000 I series Toolik
125
100
75
50
25
0
12,000 60,000 300,000 I series Toolik
Fig 6. Measurements of specific conductivity, alkalinity, and measurements of calcium and sulfate in surface waters of lakes near the Toolik Field Station. I-series and Toolik
Lake have watersheds of mixed glacial ages as indicated in Table 1.
28

29

30

Fig. 7 Food web of Toolik Lake from O'brien et al. 1997.
31

Fig. 8. Depth profile of temp, do, chl, and primary production for Toolik July 9,1999
0 2 4 6 8 10 12 14 16 18
32

5
4
3
2
1
0
12,000 60,000 300,000 I series Toolik
5
4
3
2
1
0
<2 2-<5
>5 Toolik
Fig. 9 Mean summer chlorophyll concentration in the epilimnion of lakes with waterhseds of different glacial ages (top panel) and in the epilimnion of lakes of different maximum depths (bottom panel). Values for Toolik Lake are shown for comparison.
Means + SE are shown.
33

Figure 10. Biovolume of major divisions of phytoplankton present in Lakes E5,
E6, Fog2, and Fog 4 on July 11, 2002. Lakes E6 and Fog4 are small, shallow lakes.
Lakes E5 and Fog2 are medium sized lakes with depths of 12 and 20 m respectively.
Low level fertilization of lakes E5 and E6 had begun on July 1 of that year.
80000
60000
40000
20000
0
180000
160000
140000
120000
100000
E5
Phytoplankton Abundance July 11, 2002
Dinobryon
Pyrrophyta
Cyanobacteria
Chlorophyta
Chysohyta
Cryptophyta
E6
Lake
Fog2 Fog4
34

GP
60
50
40
30
20
10
0
0 50 100 150 200
Light Level (

E)
250 300 350
Fig. 11 - Gross primary production estimated from oxygen production of sediment cores collected from Toolik Lake at 3m and incubated under a range of light intensities.
Fig. 12 - Gross primary production estimated from benthic chambers placed at 3 m in a suite of lakes near the Toolik Field Station.
250
200
150
GPP
100
50
0
E5
G
TH
8
6
F2
To ol ik
S6
N
E
9B E6 F4 S7
35

36

2
1.5
1
0.5
Small
Medium
Large
T oolik
0
D.midd
D.long
Bosmina
Species
Holo Hetero
10
8
6
4
2
20
18
16
14
12
Small
Medium
Large
T oolik
0
Diaptomus Cyclops
Species
Fig. 14. Density of cladocerans (top) and copepods (bottom) in upper portion of the water column in small, medium, and large lakes compared to densities in Toolik
Lake. Mean + SE for lakes listed in Table 1 for the years 2000-2005.
37

Fig. 15. Mean summer concentration of pelagic chlorophyll a (Reference dark circles,
Treated open circles) and annual summer total of primary productivity (Reference dark squares, Treated open squares) in Lake N2. The treated section of this lake was fertilized with dissolved nitrogen and dissolved phosphorus during the summers of 1985-1990.
From O’Brien et al. 2005.
100
90
80
70
60
50
40
30
20
10
0
PP-reference
PP-treatment
Chl-reference
Chl-treatment
20
15
10
5
85 86 87 88 89 90 91 92 93 94 95 96
Year
0
38

2
1.5
1
0.5
0
85 86 87 88 89 90 91 92 93 94 95 96
Year
Fig. 16. The average daily fluxes of inorganic nitrogen (open circles) and phosphorus
(solid squares) from the sediments of the fertilized sector of Lake N2. The treatment fertilization was conducted each summer during 1985-1990.
0
0.1
0.08
0.06
0.04
0.02
39

16
14
12
10
8
6
4
2
Fertilization Post-fertilization
Chlorophyll a
Primary Production
180
160
140
120
100
80
60
40
20
0 0
1989 1990 1991 1992 1993
Year
1994 1995 1996 1997
Fig. 17 Response of chlorophyll a and rates of primary production in water column of
Lake N1 to fertilization. Mean summer values + 1 SD are shown. Addition of phosphoric acid and ammonium nitrate occurred during summer months of 1990-1994.
40

80
70
60
50
40
30
20
Pre-fertilization Fertilization
Post-fertilization
Relative Growth
Snail Density
3.5
3
2.5
2
1.5
10
0
1
0.5
-10
-20 0
1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998
Year
Fig. 18. Relative growth of individually tagged lake trout in Lake N1 before during and after fertiliztion (filled circles are mean values + 1 SE). Open squares represent density of snails observed by divers. (After Lienesch et al. 2005).
41

Fig. 19. Top. Mean temperature in epilimnion of Toolik Lake each July for
1975-2007. The regression line is not significantly different from zero. Bottom.
Relationship between epilimnetic water temperature in toolik Lake and the Pacific
Decadal Oscillation from 1975-2007. p<0.05.
16
15
14
18
17
13
12
11
10
-6
18
17
16
15
14
13
12
11
10
9
8
1970
Toolik Surface Waters in July
1975 1980 y = 0.045x - 76
R 2
1985
= 0.082
1990
Year
1995
-4 -2 0 2 4 6 8
Pacific Decadal Oscillation
2000 y = 0.14 x +
12.7
R
2
= 0.16
2005 2010
10 12 14
42

3.5
3.4
3.3
3.2
3.1
3
2.9
2.8
Lake Trout y = -0.077x + 4.14
R 2 = 0.61
Grayling
Y = -0.0497x + 3.77
R 2 = 0.681
Char y = -0.0687x + 4.02
R 2 = 0.837
2.7
10 11 12 13 14 15
Epilimnetic Temperature
16 17 18
Fig. 20. Slope of the log length-weight regression versus epilimnetic temperature in lakes NE-12 (grayling and lake trout) and lake Fog-2 (char).
43

Box xxx. Mixing Dynamics in deep and shallow Arctic Lakes
Hydrodynamics affects ecological processes in arctic lakes in numerous ways. These include determining temperatures which affect metabolic rate processes, degree of stratification, and depth and frequency of mixing events. In arctic Alaska, lakes which are deeper than 5 m stratify at temperatures above 4 o C during summer. Shallower lakes are polymictic.
The extent of stratification and mixing are determined by incoming solar radiation, air temperatures, relative humidity, and wind (Figure 1). During summer, days with warm air temperatures (> 15 o
C), sunny skies (maximum short wave in > 700 W m
-2
) , and low relative humidity (RH < 60%) are interspersed with days when air temperatures drop by 5 to 15 o C, skies are overcast, and relative humidity is higher. In the first case, a warm air mass is over the region; in the second the polar front has moved south. During the transitions, wind speeds exceed 5 m s
-1
for periods of a day or longer. At ice off, surface temperatures in the lakes are between 4 and 8 o C, and the rate of temperature increase depends on whether warm or cold air masses are present at ice off and whether winds are low and the heat accumulates in the upper water column or is mixed downwards into the lake. Maximum summer temperatures in surface waters of Toolik Lake are generally between 15 and 20 o
C.
The rate of heating and cooling of the lakes is determined by the air – water temperature differences which affect the rate of sensible heat exchange, by the vapor pressure gradient and wind speed which affect latent heat exchange, and by net long and short wave radiation. The balance between the surface energy fluxes and the net short wave radiation determines whether or not the upper mixed layer of a lake gains heat. During sunny periods, the effective heat flux is positive in the day but goes negative at night (Figure 2, lower panel). In such cases, a diurnal mixed layer typically forms. However, during cold fronts, the sum of the effective heat flux over several days is negative and the lake loses heat. On particularly cloudy days, the lake does not gain heat even during the day.
The consequences of the alternation between warm and cold air masses and the resulting heat fluxes can be seen in the thermal stratification of Toolik Lake (Figure 3, upper panel). Because the waters of Toolik and near by lakes are highly stained with humic acids, the water is tea colored and the diffuse attenuation coefficient k d
is relatively high
(> 0.5 m
-1
). Consequently, the heat from the sun is trapped in a relatively shallow upper mixed layer. Aspects of the surface energy budget of arctic lakes also contribute to the shallow upper mixed layer. Heat losses at night are relatively low compared to temperate and tropical lakes. Hence, once the water column has stratified (e.g., days 193 to 218 in summer 2002), the mixed layer only deepens by a few meters at night when warm air masses are present. Even during cold fronts in mid-summer, the mixed layer does not deepen to more than about 5 m unless winds are strong.
The consequences of the similarities and differences in the surface energy budgets can be seen in the time series temperature data (Figure 3). The warming and cooling periods are all at the same time. However, temperatures in surface waters are lowest in Toolik Lake
44

and highest in Lake E6. Temperatures in the upper water column in Toolik and E5 are also cooler than those in E6 because mixing events entrain cooler water from depth. For the first half of the summer, 170 until day 195, little heat was gained in any of the lakes.
In fact, mean temperatures in Toolik were the coldest of the last 10 years during that period. Stratification occurred in the second half of the summer, day 195 until day 218.
During this time, a shallow mixed layer formed in all lakes in the upper 2 m. The seasonal thermocline formed below that depth in Toolik and E5. It was only 2 m thick in
E5 and up to 6 m thick in Toolik.
5
10
15
170 175 180 185 190 195 200 205 210 215 220
2
4
6
8
170 175 180 185 190 195 200 205 210 215 220
1
1.2
1.4
1.6
1.8
2
2.2
170 175 180 185 190 195 200 205 210 215 220
20
15
10
5
10
5
20
15
20
15
10
5
Box Figure 1. Time series temperatures in Toolik Lake (upper panel), Lake E5 (middle panel) and Lake E6 (lower panel) for days of year 170 through 224 in 2002. * indicate depths of thermistors. Note difference in vertical scale in each panel. Data are 10 minute averages and isotherms are 1 o
C. Colorbar shows temperatures in degrees Celcius.
45

Box 2. Changes in Sediment Chemistry Induced by Fertilization
The addition of nutrients to arctic lakes causes large changes in sediment chemistry due to increased deposition of organic matter and changes in bottom water oxygenation. Long-term sediment data from lake N2, which was fertilized for 5 years, provide insights into the biogeochemical changes that occurred during fertilization and suggest that the sediments may not fully return to pre-fertilization conditions.
Sediment cores have been taken from lake N2 from a variety of depths several times since fertilization ceased in 1990. Sediments have been analyzed for total sulfur, chromium reducible sulfur
(primarily pyrite), total C and N, total and iron oxides, total and inorganic phosphate, and other components
(the methods used are the same as those in Giblin et al. 1990 and are similar to those described in Cornwell and Kipphut 1992).
Below a depth of about 3meters there are dramatic differences between sediments from the control and fertilized side of the lake. Sediments taken from the fertilized side at 6m, 9 years after fertilization had stopped, had higher %carbon, nitrogen and % sulfur in the top several cm of the sediments than those taken from the control side (Fig. 1, a,b,c). These sediment samples also indicated that total sulfur had increased in sediments on both the fertilized and the control side since the experiment began
(Fig. 1c). The change on the control side may have been a result of; the hypoxia at 6 m which reached into both sides of the lake (O’Brien et al. 2005), the slight increase in productivity on the control side during fertilization, or both. No increase in total S was seen in sediments taken from the control side at 5 m where the sediments did not experience hypoxia (data not shown), which suggests that oxygen changes in the bottom water were primarily responsible for the increase in S burial.
The changes in the total sulfur profiles in the sediments are due in large part to the incorporation of chromium reducible sulfur (CRS) (Fig 1, d). Although CRS is an operational definition that includes a number of sulfur forms, in lake sediments the fraction is usually dominated by iron sulfide minerals (Giblin et al 1990). These iron sulfide mineral form when sulfide produced via sulfate reduction reacts with iron in the sediments. The consequence of the formation on these iron sulfide minerals is to decrease the pool of iron oxides in the sediments (Fig. 1, e). Sediments from the fertilized side of N2 contain substantially lower pools of iron oxides over the top 5 cm than do sediments from the control side. This is also true of the samples taken from depths of 5 m (data not shown).
What is the impact of this switch from iron oxides to iron sulfides in the sediments? Iron plays a major role in controlling phosphorus cycling in arctic lake sediments. Organic matter decomposition within the sediments releases P that diffuses toward the surface of the sediments. However, in most lakes in the Toolik region, this P is trapped by the adsorption onto iron oxides in surface sediments which limits the release of P back to the water column (Cornwell 1987). Fertilization has evidently depleted the iron oxide crust by trapping iron as sulfide minerals within the sediments. Additional iron may have also been solublized to Fe +2 and lost from the lake during anoxia, however, since the total iron pools (oxides + Fe in
CRS) are similar on both sides of the lakes we believe that iron sulfide formation was the most important process leading to the loss of iron oxides.
It took several years for P to be released from the sediments of N2 even after bottom waters had become anoxic (O’Brien et al. 2005). This was probably because even when active iron dissolution and iron sulfide formation was taking place there was a sufficient iron oxide pool to adsorb the P being released. Sediments can retain P as long as the ratio of iron oxides to P is great enough. This ratio
(moles/mole) has been estimate to range between 7 (Williams et al. 1997) to 20 ( Lijklema 1980). Once there had been sufficient formation of iron sulfides to deplete the oxide layer P was released.
Why did the P release stop after fertilization? The fertilized sediments still contain more P than the sediments from the control side and much of this P is in the form of inorganic P (Fig. 1f). However, currently all of the sediments from the fertilized side have molar (Fe oxide)/(inorganic P) ratios above 20 suggesting that the sediments have come to a new equilibrium with inorganic P. Ratios on the control side are greater.
These findings have implications for the ability of the lakes to recover from nutrient additions.
Iron sulfides can be oxidized back to iron oxides but with persistent hypoxia and low rates of bioturbation it seems likely that the iron sulfides will be buried and the oxide pool on the fertilized side of the lake will remain lower than on the control side. The loss of this pool of iron oxides would accelerate the response time of sediment P fluxes to future nutrient additions.
46

5
% Carbon
10 15
-20.0
-25.0
-30.0
0
0.0
-5.0
-10.0
0
-2
-4
-6
-8
-10
-12
0
-15.0
1999 Fert 1999 CT
0.2
% Sulfur
0.4
6M Depth
0.6
6M Depth
20
0.8
1999 CT Pre-Fert 1999 Fert
-6
-8
-10
-12
-2
-4
0
0.0
% Nitrogen
0.5
1.0
1.5
6M Depth
1999 Fert 1999 CT
-6
-8
-10
-12
-2
-4
0
0
Cr reducible Sulfur
30 60 90 120
1999 Fert 1999 CT
6M Depth
-2
-4
-6
-8
0
0
-10
-12
40 80
Fe oxides
120 160 200 240
6M Depth
1999 Fert 1999 CT
47
-6.0
-8.0
-10.0
-12.0
0.0
0
Inorganic P (umoles/gdw)
40 80 120 160 200
-2.0
-4.0
6M Depth
1999 CT 1999 FT

Box 3. Effect of internal waves on phytoplankton primary production
Just as surface waves form on lakes at the air water interface, waves can form within a lake at boundaries between water layers of different densities. These internal waves can be much larger in amplitude than the surface waves on a given lake because the difference in density between the water layers in much smaller than that between water and air. Internal waves move water, chemicals, and any nonmotile or slow moving organisms vertically through the water column. Because of this movement they can have important impacts on nutrient availability in the upper water column, spatial or temporal aggregation or patchiness of plankton, and light availability to deep phytoplankton.
Phytoplankton riding on internal waves are periodically moved closer to and further from the lake surface. Because light decreases exponentially with distance from the lake surface, this vertical movement can increase the average light exposure of the phytoplankton. The effects of this motion on the light climate for phytoplankton, and its implications for photosynthesis, have been previously described based on mathematical models for the situation where surface light is constant or varies very slowly compared to the period of the internal waves (see references in “for more information” list).
Work at Toolik Lake has shown that internal waves near the bottom of the euphotic zone can cause a large enhancement of photosynthesis and that this effect is modified by variations in surface light due to cloud cover. In situ experiments were conducted at Toolik Lake during the summers of 2003 and 2004 to test the prior theory on internal wave effects. In each experiment a set of bottles containing lake water was incubated at depth in Toolik Lake for 24 hours and the photosynthesis in each bottle was measured. During the 24 hours some of the bottles were held at a constant depth (5 m) and some were moved up and down between 4 and 6 m using a motor mounted at the lake surface. The moving bottles followed a smooth sin wave with a 4 hour period, similar to the dynamics of natural internal waves in the lake. The results of this experiment were surprising when compared to the earlier theory. As predicted, enhancement of photosynthesis was observed in the moving bottles. However, on some days this enhancement was greater than expected and on other days photosynthesis was unaffected by the vertical movement (Figure).
To explore the mechanism causing these unexpected results, a computer simulation model was constructed and tested using the surface light, light attenuation, and phytoplankton physiology measured on the day of each experiment. This model showed that variation around the photosynthetic enhancement predicted by earlier theory can be caused by changes in the amount of light entering the lake during the incubation. When surface light varies due to the passage of clouds with approximately the same period or time scale as the internal wave motion, and the time of brightest light aligns with the crest of a wave, the enhancement of photosynthesis is strengthened. Conversely, when the time of brightest light aligns with a wave trough, the enhancement of photosynthesis is reduced or eliminated. Even though internal wave depth and surface light were poorly correlated over 24-hour time series (generally abs(R) < 0.3) at Toolik
Lake, large interactive effects of waves and surface light on photosynthesis were still generated. This suggest that the interaction is robust and likely important in other systems with internal waves and moderate levels of cloud cover.
48

Box 3 Figure: Phytoplankton primary production (
 g C L -1 d -1 ) from a series of in situ internal wave experiments in 2003 and 2004. Stationary treatments (depth = 5 m) had generally lower primary production than treatments in which phytoplankton were moved sinusoidally through the water column with average depth of 5 m, period of 4 h, and amplitude in the vertical of 0.5 m or 1 m. Statistically significant differences were determined between treatments within date at the p =0.05 level and indicated by the letters a, b, and c, where treatments with the same letter are not significantly different and where ‘ns’ indicates that there were no significant differences on that date.
Box. 4 Zooplankton pigmentation patterns results from extreme arctic environments.
The extreme climate regime in the arctic enhances the difference in zooplankton assemblages in deep and shallow arctic lakes. One example of these differences is the pigmentation of zooplankton in arctic lakes. The short growing season and relatively low lake temperatures enhance the value of inhabiting the warm, food-rich surface waters for zooplankton. Arctic lakes thaw each year close to the summer solstice when solar radiation is most intense and harmful UV rays damage animal cells. Dark red, blue and black pigments in the tissues of some zooplankton species reduce the damage that occurs from sunlight in shallow water. Although these pigments provide protection from the harmful effects of the sun’s rays, they make the zooplankton more vulnerable to visually feeding fish.
Extreme winter conditions allow for the variation in zooplankton pigmentation to exist. In shallow lakes where high light conditions demand photo-protective pigments of the zooplankton, the lakes freeze solid in winter eliminating fish populations. In deeper lakes the fish are present and preferentially consume pigmented zooplankton. In these
49

lakes the zooplankton are unpigmented, but can escape the intense sunlight by occupying deeper waters. The presence of these habitats provided the opportunity for researchers to investigate the impacts of fish on lake communities and ecosystem function. These initial studies conducted in arctic systems provided the impetus for other researchers to examine the more subtle effects of fish predation and photodamage on zooplankton in temperate regions.
For more details on these studies see O’Brien et al 1976; Luecke and O’Brien
1981; Luecke and O’Brien 1983; O’Brien et al. 2004; Burkart, unpublished data; Johnson unpublished data.
Box 5. Importance of invertebrate predators in structuring arctic zooplankton communities
Early work on the importance of the invertebrate predator Heterocope septentrionalis on herbivorous zooplankton focused on cage experiments. These cage experiments, where H. septentrionalis was paired up with different zooplankton species, showed that there was a 50 to 80 fold difference in vulnerability between the most vulnerable to the least vulnerable species. Intriguingly D. pulex proved to be the most vulnerable while D. middendorffiana was the least vulnerable. Bosmina lonirostris was the next most vulnerable species while a copepod, Diaptomus pribilofensis, was only slightly more vulnerable than D. middendorffiana . This accorded well with field observations where D. pulex never coexisted with H. septentrionalis while D. middendorffiana commonly did coexist with the predator. Likewise B. longirostris rarely coexisted with H. septentrionalis but D. pribilofensis did coexist.
To better demonstrate that H. septentrionalis can structure arctic zooplankton communities we performed a series of mesocosms experiments. The general pattern of these experiments was to add around 20 H. septentrionalis and 200 D. pulex and 200 B. longirostris to a 75 L mesocosm and follow the population densities of the herbivorous zooplankton over a month’s time. We also paired D. pulex and D. middendorffiana in mesocosm experiments. The general pattern was that D. pulex and B. longirostris declined while D. middendorffiana either increased or did not change in density over the month. However, such experiments have inherent problems. For example the experimenter chooses the density of the predator and clearly high density would lead to false conclusions about the potential impact of the predator. Plus these experiments don’t allow a numerical response to the prey and there are surely fewer possible refuges in a plastic mesocosm than in a natural pond.
To alleviate these and other problems we added H. septentrionalis to a small pond that naturally had D. pulex, B. longirostris and D. pribilofensis . We added H. septentrionalis at a density of 1 per liter such that it would have to make a numerical response to effect the herbivorous zooplankton. There was a reference pond which had
B. longirostris and D. pribilofensis but no D. pulex. In three years D. pulex declined 4 orders of magnitude and in four years B. longirostris declined almost as much, however,
D. pribilofensis increased 2 orders of magnitude. We feel this last experiment clearly shows the awesome power of the invertebrate predator, Heterocope septentrionalis, to structure arctic zooplankton communities.
50

References (still a mess):
Brown, J. S., S. L. Dingman, and R. I. Lewellen. 1968. Hydrology of a drainage basin on the Alaskan Coastal Plain. Res. Rep. 240. U. S. Army Cold Regions Res. Eng.
Lab., Hanover, NH. 18 p.
Cornwell, J. C. 1985. Sediment accumulation rates in an Alaskan arctic lake using a modified 210 Pb technique. Can. J. Fish Aquat. Sci. 42:809-914.
Cornwell, J. C. 1986. Diagenetic trace-metal profiles in arctic lake sediment.
Environmental Science & Technology 20:299-302.
Cornwell, J. C. 1987. Phosphorus cycling in arctic lake sediments: adsorption and authigenic minerals. Arch. Hydrobiol. 109:161-179.
Cornwell, J. C. and G. W. Kipphut. 1992. Biogeochemistry of manganese and iron rich sediments in Toolik Lake, Alaska. Hydrobiology 240:45-59. de March, L. 1978. Permanent sedimentation of nitrogen, phosphorus and organic carbon in a high arctic lake. J. Fish. Res. Bd. Can. 35:1089-1094.
Dingman, S. L., R. G. Barry, G. Weller, C. Benson, E. F. LeDrew, and C. W.
Goodman. 1980. Climate, snow fover, microclimate and hydrology, p. 30-65 In J.
Brown, P. C. Miller, L. L. Tieszen, and F. l. Bunnell (ed.) An arctic ecosystem: the coastal tundra at Barrow. Alaska. Dowden, Hutchinson and Ross, Stroudsburg, PA.
Gooseff, M.N., A. Balser, W.B. Bowden, and J.B. Jones. 2009. Effects of
Hisllsope thermokarst in northern Alaska. EOS 90:29-30.
Hamilton, T. D. and S. C. Porter. 1975. Itkillik glaciation in the Brook’s Range,
Northern Alaska. Quat. Res. 5:471-497.
Hobbie, J. E. 1962. Limnological cycles and primary productivity of two lakes in the Alaskan Arctic. Thesis, Indiana University, Bloomington, Ind. 124 pp.
Hobbie, J. E. 1984. Polar limnology, pp. 63-105 In F. B. Taub (ed.), Lakes and
Reservoirs. Elsevier, Amsterdam.
Hutchinson, G. E. 1957. A treatise on limnology. Vol. I. Geography, physics, and chemistry. Wiley and Sons. N.Y.
Kling, G. W., G. W. Kipphut, and M. C. Miller. 1991. Arctic lakes and streams as gas conduits to the atmosphere: Implications for tundra carbon budgets. Science
251:298-301.
51

Kling, G. W., W. J. O’Brien, M. C. Miller, and A. E. Hershey. 1992. The biogeochemistry and zoogeography of lakes and rivers in arctic Alaska. Hydrobiology
240:1-14.
Kling, G. W., G. W. Kipphut, M. C. Miller, and W. J. O’Brien. 2000. Integration of lakes and streams in a landscape perspective: the importance of material procession on spatial patterns and temporal coherence. Freshwater Biology 43:477-497.
Kipphut, G. W. 1988. Sediments and organic carbon cycling in an arctic lake.
SCOPE/UNEP Sonderband 66:129-135.
Miller, M. C., R. T. Prentki, and R. J. Barsdate. 1980. Physics, p. 51-75. In J.
E.Hobbie (ed.) Limnology of tundra ponds. Dowden, Hutchinson and Ross, Stroudsburg,
PA.
Miller, M. C., G. R. Hatter, P. Spatt, P. Westlake, and D. Yeakel. 1986. Primary production and its control in Toolik Lake, Alaska. Arch. Hydrobiol./Suppl. 74:97-131.
Morris, D. P., H. Zagarese, G. E. Williamson, E. G. Balseiro, B. R. Hargeaves, B.
Modenutti, R. Moeller, and C. Queimalinos. 1995. The attenuation of solar UV radiation in lakes and the role of dissolved organic carbon. 40:1381-1391.
Schindler, D. W., G. J. Brunskill, S. Emerson, W. S. Broecker, and T. H. Peng.
1972. Atomspheric carbon dioxide: Its role in maintaining phytoplankton standing crops.
Science 177:1192-1194.
Schindler, D. W., R. W. Newbury, K. G. Beatty, and P. Campbell. 1976. Natural and chemical budgets for a small Precambrian lake basin in central Canada. J. Fish. Res.
Bd. Canada 33:2526-2543.
Whalen, S. C. and J. C. Cornwell. 1985. Nitrogen, phosphorus, and organic carbon cycling in an arctic lake. Can. J. Fish. Aquat. Sci. 42:797-808.
References
Banerjee, S, and S. MacIntyre. 2004. The air-water interface: turbulence and scalar exchange. pp. 181-237 in John Grue, Philip L. F. Liu, and Geir K. Pedersen eds.
Advances in Coastal and Ocean Engineering. World Scientific.
Evans, M.A. 2007. The Phytoplankton Ecology of Arctic Lakes. Ph.D. Thesis. University of Michigan, Ann Arbor, MI.
Kunkel, K.E., K. Andsager, and D.R. Easterling. 1999. Long-term trends in heavy precipitation events over North America. J. Climate 12, 2513-2525.
MacIntyre, S., J.O. Sickman, S.A. Goldthwait, and G.W. Kling. 2006. Physical pathways of nutrient supply in a small, ultra-oligotrophic lake during summer stratification.
Limnol. Oceanogr. 51: 1107-1124.
52

Miller, M.C., P. Spatt, P. Westlake, D. Yeakel and G.R. Hater. 1986. Primary production and its control in Toolik Lake, Alaska. Arch. Hydrobiol. Suppl. 74: 97-131.
O’Brien, W. J., M. Bahr, A.E. Hershey, J.E. Hobbie, G.W. Kipphut, G.W. Kling, H.
Kling, M. McDonald, M.C. Miller, P. Rublee, and J. R. Vestal. 1997. The limnology of Toolik Lake. In A.M. Milner and M.W. Oswood, [eds.], Freshwaters of Alaska. Ecological Studies volume 119. Springer-Verlag.
Rueda, F., S. MacIntyre, and G.W. Kling. 2006. Fate and transport of storm-river-water in a small, ultra-oligotrophic lake: Toolik Lake, Alaska. Proceedings Volume:
Physical Processes in Natural Waters: 10 th
European Physical Limnology
Workshop. Granada, Spain.
Whalen, S.C., and V. Alexander. 1984. Influence of temperature and light on rates of inorganic nitrogen transport by algae in an arctic lake. Can. J. Fish. Aquat. Sci.
41: 1310-1318.
Whalen, S.C., and V. Alexander. 1986. Seasonal inorganic carbon and nitrogen transport by phytoplankton in an arctic lake. Can. J. Fish. Aquat. Sci.: 43: 1177-1186.
Bettez, N. D., P. A. Rublee, J. O'Brien, and M. C. Miller. 2002. Changes in abundance, composition and controls within the plankton of a fertilised arctic lake.
Freshwater Biology 47 :303-311.
O'Brien, W. J., M. Barfield, N. Bettez, A. E. Hershey, J. E. Hobbie, G. Kipphut, G.
Kling, and M. C. Miller. 2005. Long-term response and recovery to nutrient addition of a partitioned arctic lake. Freshwater Biology 50 :731-741.
O'Brien, W. J., A. Hershey, J. Hobbie, M. A. Hullar, G. W. Kipphut, M. C. Miller, B.
Moller, and J. R. Vestal. 1992. Control Mechanisms of arctic lake ecosystems: a limnocorral experiment. Hydrobiologia 240 :143-188.
Rublee, P. A. 1992. Community Structure and Bottom-up Regulation of Heterotrophic
Microplankton in Arctic Lter Lakes. Hydrobiologia 240 :133-141.
Rublee, P. A., and N. Bettez. 1995. Change of microplankton community structure in response to fertilization of an arctic lake. Hydrobiologia 312 :183-190.
Rublee, P. A., and N. D. Bettez. 2001. Lake characteristics influence recovery of microplankton in arctic LTER lakes following experimental fertilization.
Hydrobiologia 446 :229-232.
Azam F, Fenchel T, Field JG, Grey JS, Meyer-Reil LA, Thingstad F (1983) The ecological role of water-column microbes in the sea. Mar Ecol Prog Ser 10:257-
263
Kling GW (1994) Ecosystem-scale experiments in freshwaters: the use of stable isotopes.
In: Baker LA (ed) Environmental Chemistry of Lakes and Reservoirs, Advances in Chemistry Series 237, American Chemical Society, Washington D.C., pp 91-
120
Kling GW, Kipphut GW, Miller MC (1991) Arctic lakes and rivers as gas conduits to the atmosphere: implications for tundra carbon budgets. Science 251:298-301.
Palenik B, Morel FMM (1990) Amino acid utilization by marine phytoplankton: A novel mechanism. Limnol Oceanogr 35:260-269
53

Rublee P (1992) Community structure and bottom-up regulation of heterotrophic microplankton in arctic lakes. Hydrobiologia 240:133-141
Stockner JG, Porter KG (1988) Microbial food webs in freshwater planktonic ecosystems. In: Carpenter SR (ed), Complex Interactions in Lake Communities,
Springer-Verlag, NY, pp 69-84.

Alexander, V., S.C. Whalen and K.M. Klingensmith. 1988. Nitrogen cycling in arctic lakes and ponds. Hydrobiologia 172:165-172.

Cornwell, J.C. 1987. Phosphorus cycling in arctic lake sediments: adsorption and au thigenic minerals. Archiv für Hydrobiologie 109:161-179.

Cuker, B.E. 1983. Grazing and nutrient interactions in controlling the activity and composition of the epilithic algal community of an arctic lake. Limnology and
Oceanography 28:133-141.
Evans, M. A., S. M. MacIntyre, and G. W. Kling. 2008. Internal wave effects on photosynthesis: experiments, theory, and modeling. Limnology and Oceanography 53:339-353.
Haury, L. R., P.H. Wiebe, M. H. Orr., and M. G. Brisco 1983. Tidally generated high-frequency internal wave packets and their effects on plankton in Massachusetts Bay. J. Mar. Res. 41: 65-112.
Holloway, G. 1984. Effects of velocity fluctuations on vertical distributions of phytoplankton. J. Mar. Res.
42: 559-571.
Holloway, G., and K. Denman. 1989. Influence of internal waves on primary production. J. Plankton. Res.
11: 409-413.
Kamykowski, D. 1974. Possible interactions between plankton and semidiurnal internal tides. J. Mar. Res.
32: 67-89.
54

Table 1. Morphometric conditions of LTER lakes near the Toolik Field Station.
The lakes were grouped into small, medium and large based on clustering analysis of 14 physical, chemical and biological characteristics. Landscape age refers to time since last glaciation. Composite refers to lakes draining watersheds with glacial ages of both 12,000 and 60,000.
Small lakes
E6
Fog4
S7
NE9b
S11
Surface
Area (Ha)
1.9
1.9
0.8
0.4
0.3
Maximum
Depth (m)
3.2
4.4
2.9
6.0
9.5
Landscape
Age (YBP)
300,000
composite
12,000
12,000
12,000
Medium lakes
E1
N2
I8
I3
S6
Large Lakes
Fog2
E5
I5
I6
I7
N1
NE12
Toolik Lake
2.6
1.6
16.7
3.1
1.1
5.9
11.3
18.7
16.9
2.9
4.3
8.2
149
11.0
9.7
9.5
5.1
5.2
20.3
12.7
8.5
15.0
14.0
14.2
17.1
25
60,000
12,000
composite
12,000
12,000
composite
300,000
composite
composite
composite
12,000
12,000
composite
55

Table 2 will have a list of lakes and chemical parameters.
56