Printer Friendly

Landscape controls over major nutrients and primary productivity of Arctic lakes.


Changing climatic factors are altering Arctic landscapes, affecting the dynamics between the landscapes and the lakes located within them (Schindler and Smol 2006). One of the major climatic factors researchers have documented over the last three decades is an increase of 3.5[degrees]C in surface temperature (IPCC, AR4, Synthesis report: 32), a trend expected to continue in the near future. Warming temperatures are causing permafrost to melt, which induces more frequent thermokarst activities (Serreze et al. 2000). Scientists expect the cumulative effect of these activities will be the release of nutrients, including compounds of nitrogen, phosphorus and various ions, into streams and lakes (Keller et al. 2007). With changes in the nature of the influx, oligotrophic Arctic lakes may increase their primary productivity.

Recent research, particularly in the disciplines of biology and ecology, has provided a foundation for studying nutrient modeling as it relates to ecosystems in the Arctic. What is largely missing, however, is a spatial perspective, which we argue could be provided by employing satellite imagery and landscape metrics. Marion et al. (1989), for example, conducted a study of the tundra ecosystem and observed variations within major nutrients (nitrogen and phosphorus), along with potassium, calcium and magnesium in soils supporting different vegetation communities. In a similar study, Chapin and Shaver (1988) inferred that water in the soil played a significant role in cycling nutrients within vegetation communities. Moreover, water tracks within the watershed aided nutrient transfer. Besides the water content of soil, they also found temperature and active soil depth to be important factors in the mineralization process of nutrients within forbs and graminiods.

Focusing on biogeochemical diversity in the Arctic, Giblin et al. (1991) demonstrated that the uptake rate for nutrients by vegetation species partially controlled extractable nutrient levels in the surrounding soil. These researchers also studied vegetation land cover to understand how the characteristic root structures of different vegetation species interacted with soil horizons. They found that certain dominant species, such as Eriophorum vaginatum, could intercept more nutrients due to deep root structure; other species, however, such as Ledum decumbuns, relied on the upper organic layer of the soil for their nitrogen and phosphorus supply. Recently, a more comprehensive study by Keller et al. (2007) examined geochemical weathering and its effects on soils and streams in the Arctic. They found that the weathering rate of parent material caused a higher concentration of nutrients to reach surface waters. The rate of weathering was related to both glacial age and active soil depth. Nutrient modeling studies of the Arctic may largely fail to address spatial context issues, but there are studies outside the region that have used spatial techniques to analyze the contribution of land surface processes towards surface water chemistry (King et al. 2005; Allan and Johnson 1997; Griffith et al. 2002). Percentages (Gergel et al. 2002), proportion metrics, and fragmentation metrics of prominent land cover types, such as agriculture or riparian vegetation (Jones, et al. 2001) were all found useful in predicting water quality in this body of work. These studies also highlighted that the structural arrangement of land cover in the immediate vicinity of water channels played a significant role in allowing nutrients to, or restraining nutrients from, entering surface waters.

The purpose of this study was to incorporate a spatial perspective, via landscape metrics, to the study of nutrient modeling in the Arctic. Using landscape metrics as a means of describing the spatial and structural properties of land cover, we assessed how well the metrics predicted lake primary productivity and major nutrient concentrations in Arctic lakes. The literature suggests, for instance, that shape complexity, both cumulative and of individual patches, might indicate the dominance of a particular land cover category, as well as be connected to nutrient cycling activities (Ostendorf and Reynolds 1993; Medley et al. 1995). If this is true in Arctic landscapes, simpler shapes should be the rule of thumb for dominant land cover types, with minor land cover types taking more complex forms.

To explore this idea further, we calculated patch density, shape complexity, and fragmentation landscape metrics for each land cover in forty watersheds, located in the foothills of the Brooks range in Arctic Alaska, at two levels; one at the entire watershed level and the other restricted to a 20-meter buffer distance from water channels. The data were analyzed using stepwise regression models, where lake chemistry parameters were used as dependent variables and landscape metrics served as the independent variables.


Study Area

This research was undertaken in the Toolik Lake region (68[degrees] 38' N, 149[degrees] 36' W), situated in foothills of the Brooks Range on the North Slope of Alaska (Figure 1). Hillocks, exposed barren areas, and moraines characterize the entire landscape. Water tracks, streams, and rivers, along with different types of lakes, dissect the area. The lakes were formed by ice scouring, thermokarst activity, and the melting of ice blocks. The Sagavanirktok, Toolik, Itkillik, and Kuparuk rivers comprise the major rivers in the study area.

The Toolik Lake region experienced multiple glaciations in the mid and late Pleistocene eras. Lakes included in the current research reside in both the Sagavanirktok and Kuparuk River watersheds, which represent different till ages. The Itkillik I glacial advance dates to approximately 53,000 B.P. and the Itkillik II advances range from 25 to 11.5 ka B.P. (Before Present). The Sagvanirktok glaciation ranges from 780,000 to 125,000 B.P. (Hamilton, 1986).


Non-Spatial Data

Lake Chemistry Parameters

Volumetric estimates of chlorophyll a (Chla_V, [micro]g [L.sup.-1]), total nitrogen (TN, [micro]mol [L.sup.-1]), and total phosphorus (TP, [micro]mol [L.sup.-1]) for each of the lakes in the study area (Table 1) were collected over three thawing seasons (summer of 2001, 2002, and 2003). Chlorophyll a served as a surrogate measure for direct productivity readings because productivity measurement in this case is affected by sunlight exposure at the time of estimation and may not represent the true productivity level of the lakes.

Each lake was sampled once during the study. Due to the remoteness of the area, lakes could only be reached via helicopter. This added expense limited sampling frequency and timing, which, in turn, excluded the possibility of addressing seasonal variation for this particular study.

Spatial Data

Watersheds and Land Covers

A digital elevation model was downloaded from the Toolik Lake web site to serve as the base for delineating the study watersheds. With a spatial resolution of 5m by 5m, it was acquired using Interferometric Synthetic Aperture Radar (IFSAR) by Intermap Technologies Corporation STAR-3i systems (Intermap 2010). ArcGIS hydrology tools were used to derive the watersheds from the DEM. These watersheds were then used to subdivide a SPOT color infrared image with a 5m spatial resolution to create a land cover classification for each watershed. This image, acquired from SPOT Corporation, was preprocessed by SPOT and arrived ready for classification (FAQ, SPOT Systems).

This classification process used an adopted land cover scheme, formed by combining the classification systems used by Walker et al. (1994) with field modifications conducted by Stine and Ray during the summers of 2006-2008 (Table 2). Apart from the Dalton highway, there were no man-made structures present in the given watersheds; therefore, only land cover categories and their spatial properties were considered crucial to the study.

Using ERDAS Imagine, the data from the SPOT image, along with its normalized difference vegetation index (NDVI) layer, were combined and classified using the ISODATA clustering method with unsupervised classification. Unsupervised classification was chosen because the land cover was highly heterogonous, making it difficult to identify uniform land cover patches using visual methods to feed a supervised classification. The initial classification of the vegetation complexes was not able to identify the snowbed complex because of its small spatial extent. For this vegetation complex, the authors refined the classification process by incorporating an expert system utilizing slope and aspect layers (Stine et al. 2010) (Figure 2). The accuracy of the final classification for the vegetation complexes was 86%.

Ground truth information obtained for six watersheds during the summer of 2008 was used to assess the accuracy of this classification. Although the ground truth data were collected more recently than the data sampled during the study, any change in land cover for the area should have been slight enough not to significantly affect classification accuracy The region's low temperatures, coupled with the limited days of thawing and vegetation growth, all combine to minimize such short-term changes. Furthermore, the major land cover types in the region are significantly influenced by the age of the glacial till, where relatively younger tills are associated with moist non-acidic tussock tundra (MNT) and older tills support moist acidic tusock tundra (MAT) (Hamilton 1986).


Landscape Metrics

Landscape metrics are indices formed from algorithms designed to measure various spatial characteristics of a landscape or landscape patch. Five different metrics were used in this study: Percentage of Land Cover Class, Patch Density, Mean Shape Complexity, Fractal Dimension, and the Landscape Shape Index. Each attempted to quantify how different land covers were spatially configured, and to relate that to lake primary productivity and the major nutrient concentrations of chlorophyll a, total nitrogen, and total phosphorus. Calculated using Fragstats 3.3 and ESRI's ArcGIS, each metric was derived for both the entire watershed level as well as for 20-meter buffer distances from water channels within each watershed.

Percentage of Land Cover Class represents the area of a land cover divided by the area of the watershed in which it resides. Patch Density is a fragmentation index, depicted by the number of patches of a land cover per unit area to create a density measure. Lower densities indicate a smaller number of contiguous patches; higher numbers suggest larger numbers of contiguous patches.

Mean Shape Index and Fractal Dimension are cursors of shape complexity. Mean Shape Index is an index that represents the general nature of a shape in the landscape: is it more compact or more irregular in shape? This idea is typically quantified by using a perimeter-to-area ratio, which is standardized to a simple shape such as a square (given a value of 1). The higher the number associated with a landscape patch, the more complex or irregular it is. Fractal Dimension is a shape index that makes use of perimeter-to-area ratios within the context of fractal analysis by describing the power relationship between a patch's area and its perimeter. If a patch is simpler, or more geometric, in shape, its perimeter will increase more slowly as the corresponding area increases, resulting in a lower index value. In patches with more complex shapes, the perimeter increases more rapidly in relation to the patch's area, resulting in larger index values (McGarigal et al. 2002).

The Landscape Shape Index presents a measure of dispersion/interspersion for landscape patches of a specific land cover. Here, the perimeter-to-area ratio is calculated for all patches within a specific land cover, normalized and compared to a standard geometric shape of the same size, such as a square. Values less than one indicate increasing aggregation of land cover patches, whereas values greater than one suggest increasing interspersion (McGarigal et al. 2002).

Other Landscape Factors

Other parameters derived from the classified watersheds were Latitude and Longitude (i.e., lake centroids), Lake Area, Watershed Area and their ratios. Lake Order (Riera et al. 2000), where lakes are ranked on the basis of their connectivity to other neighboring lakes (Strahler 1952) was derived from topographic maps of the area.

Data Analyses

The relationships between the landscape metrics, other landscape factors and lake chemistry parameters collected during fieldwork were analyzed using a series of backward stepwise regression models. Prior to running the models, regression diagnostics were computed and evaluated to ensure none of the standard assumptions had been violated. These diagnostics included tests for multi-collinearity, a potential issue with landscape metrics. As with the standard regression assumptions, no problems were noted. Landscape metrics and the other landscape factors served as the independent variables in the regression models; lake chemistry parameters were the dependent variables in the models.

Issues that may influence model powers include nutrient availability, data collection timing, the satellite image data, and the number of lakes sampled. The amounts of nutrients available in Arctic ecosystems are very low, for example, with those flowing into lakes being utilized immediately by algae for chlorophyll production. Hence, there was less variance within the observed concentrations of the lake nutrients measured (Levine and Whalen 2001). The data were also collected over three different thawing seasons and the satellite image available for the area was from another thawing season. Although seasonality varied in the data collection process, it was over such a short period it any changes due to timing would have been minimal; thus date of collection was not considered as a variable in the regression analyses for this study. Finally, there was a limit to the number of lakes visited. Conducting fieldwork in such a remote area was challenging; the lakes were not accessible by road requiring the use of a helicopter to facilitate ground-truthing. The expenses of engaging a helicopter further restricted the times and numbers of watersheds that could be visited.


Early in the backward stepwise regressions for each of the lake chemistry measures, Latitude and Longitude Locations, Lake Area, Watershed Area, Lake-to-Watershed Area Ratio, and Lake Order each dropped out as they were not statistically significant. Chlorophyll a, total nitrogen, and total phosphorus are volumetric measurements. The landscape parameters derived from the satellite imagery were based on two-dimensional fixed spatial resolutions; therefore, lake chemistry measures were not significantly affected by the size of lakes or watersheds as measured in this study. Landscape metrics that were significant in predicting lake chemistry measures are listed in Table 3.

Chlorophyll a Volumetric (Chla_V) Regression Model

The relationship between the volumetric estimates of chlorophyll a [micro]g [L.sup.-1]) and the landscape metrics in the regression model (Table 4) explained 52% of the variation in this lake chemistry parameter among the watersheds in the study ([r.sup.2] = 0.52, p < 0.001). The strongest predictors of chlorophyll a were the Landscape Shape Index for riparian patches (LSI_Rip, std. coeff. = 0.649) and the Patch Density for heath complex at buffer level (PD_B_Heath, std. coeff. = 0.345). The effect of these two metrics suggests that increasing interspersions of riparian patches and larger numbers of contiguous patches of heath complex within water channel buffers contribute the most to increasing measures of chlorophyll a in these watersheds, where the increase in interspersions of riparian patches has the most positive influence.

Total Nitrogen Regression Model

The regression for total nitrogen estimates (TN, mol [L.sup.-1]) explained 48% of the variation in TN estimates among the watersheds in the study ([r.sup.2] =0.48, p < 0.001). Two landscape metrics (Table 4) were the strongest predictors of these estimates: Percentage of riparian complex at buffer level (Prct_B_Rip, std. coeff. = 0.552) and Patch Density for fen complex at buffer level (PD_B_ Fen, std. coeff. = 0.196). The combined effect of these metrics suggest that larger numbers of contiguous patches of fen complex and higher percentages of riparian complex within the water channel buffers contribute the most to increasing measures of total nitrogen in the watersheds, where changes in the riparian complex have the most positive influence.

Total Phosphorus Regression Model

Total phosphorus (TP, mol [L.sup.-1]) represents dissolved as well as particulate phosphorus content in the water. In this model (Table 4), the relationship between volumetric estimates of TP and the landscape metrics explained 52% of the variation in total phosphorus among the watersheds in the study ([r.sup.2] = 0.52, p < 0.001). The strongest positive predictors of TP were the Patch Density for shrub complex (PD_Shrub. std. coeff = 0.52) and the Mean Shape Index for moist acidic tundra complex (MS_MAT, std. coeff. = 0.459). In this case, both variables have almost equally strong positive effects, where larger numbers of contiguous patches of shrub complex within water channel buffers and increasingly complex patches of MS_M_AT complex with the watersheds themselves contribute the most to increasing measures of total phosphorus in the watersheds. Patch Density for aquatic vegetation (PD_AV, std. coeff. = -0.334) contributed negatively to this model, suggesting that decreasing contiguous patches of aquatic vegetation also contributed to decreasing measures of TP.


The landscape metrics used in this study explained a significant amount of the variance of chlorophyll a, total nitrogen, and total phosphorus in the Arctic lakes sampled. Among the broad leaf vegetation category, the riparian complex was a significant predicting variable for both volumetric chlorophyll a estimates and total nitrogen estimates. The regression model for volumetric chlorophyll a, for example, suggested that with an increase in the complexity of riparian patches, the amount of chlorophyll a in lakes would also increase. As noted by Giblin et al. (1991), riparian complexes, which occur in low-lying areas and experience higher water flux than upland soils, have deep root structures and riparian soil rich in nitrogen content. With the observed positive relationship between the shape complexity of riparian patches and the Chlorophyll a content of lakes, we can propose that when more complex landscape patches of riparian vegetation are present, higher nitrogen content is likely to be released into the lakes, leading to higher primary production (i.e., higher Chlorophyll a concentration).


This supposition is illustrated in Figure 3. The riparian patches of watershed GTH 153 (a) are contiguous and represent only a 16% increase from an ideal value 1 for the LSI metric. In contrast, there was 43% increase in LSI value from the value of 1 for the riparian patches observed in watershed GTH 144 (b), which exhibited relatively higher chlorophyll a concentration content than GTH 153.


Similarly, the percentage of riparian complexes occurring at buffer level was a good indicator of total nitrogen concentration in the lakes. In this case, watershed GTH 143 had approximately 5% of its area occupied by riparian complex patches and the observed concentration of total nitrogen was 17.5 [micro]mol [L.sup.-1]. When the percent area occupied by riparian complex was 9.5%, on the other hand, as in case of watershed GTH 154, the total nitrogen concentration measured in the lake was 24.6 [micro]mol [L.sup.-1].

Another significant predictor for total nitrogen concentration was the Patch Density for fen complex at buffer level. Here, we observed that fragmented fen complex was associated with increasing nitrogen content of the lakes. This correlates well with the findings of Giblin et al. (1991), who showed that soils supporting fen complexes were saturated with nitrogen compounds. In our study, Watersheds GTH 115 and GTH 147 exemplify this phenomenon. The Patch Density for fen complex at buffer level for GTH 115 was 67 patches per 100 hectares and the total nitrogen concentration measured was 12 mol [L.sup.-1]. In contrast, a higher Patch Density for fen complex at buffer level (i.e., 114 patches per 100 hectares in GTH 147), was observed with 21 mol [L.sup.-1] concentration of total nitrogen in the lake.

The primary predictor from the broad leaf vegetation community for total phosphorus content in Arctic lakes was the shrub complex. The regression model for total phosphorus indicated that highly fragmented patches of shrub complex, depicted by higher patch densities, were correlated with an increase in phosphorus concentration in the lakes. Previous research has documented that deciduous and evergreen shrub species in the Arctic exhibit higher uptake rates for phosphorus (Chapin and Shaver 1988). Marion et al. (1989), in fact, specifically mentioned that Salix and Betula species showed higher phosphorus contents in their leaves than other species. These studies suggest that shrub communities are actually sinks for phosphorus; however, as fragmentation increases, more phosphorus content reaches the lakes.

This finding is supported by observations for watersheds GTH 129 and GTH 148 (Figure 4). There are patches of shrub complex in the GTH 129 watershed (a), although they are few and contiguous. In contrast, the GTH 148 watershed (b) displays very fragmented and widely distributed shrub patches. These conditions are reflected in the Patch Density for shrub complex, where watershed GTH 129 had a lower Patch Density (86 patches per 100 hectares) than watershed GTH 148 (439 patches per 100 hectares). Moreover, the GTH 129 watershed experienced lower total phosphorus concentrations than GTH 148, which supports our proposed relationship between Patch Density for shrub complex and phosphorus release into lakes.

Although Giblin et al. (1991) noticed that heath complexes, which occur on rocky surfaces, were also sources of loosely bound phosphorus, this complex did not emerge as a predictor in the regression model for total phosphorus estimates. Interestingly enough, however, it was one of the predictors for volumetric estimates of Chlorophyll a. According to the observed relationship in this research, the more fragmented the patches of heath complex are, the higher the Chlorophyll a concentration. As phosphorus is one of the major nutrients for phytoplankton production (Wetzel 2001), however, it also seems likely that fragmented heath patches at buffer level acted as a source of phosphorus content for the nearby lakes, enhancing their Chlorophyll a concentration. If multiple seasonal readings for total phosphorus could be obtained, it is possible the relationship between heath complex and total phosphorus could be verified.


Another significant indicator for total phosphorus content in Arctic lakes was the Mean Shape Complexity index for the Moist Acidic Tundra (MAT) complex. MAT is one of the dominant land covers in the Arctic. The available literature, however, did not indicate any direct relationship between this vegetation community and soil phosphorus content. We do know that MAT complexes are composed of deep root structure species such as Eriophorum vaginatum. We also know that this species is associated, overall, with more nutrients (including phosphorus) in its surrounding soil than others. We surmise from this that more fragmentation, reflected by a higher Patch Density Index, might act as a source of phosphorus for the lakes. If the patches were more contiguous, on the other hand, they would act as sinks instead. As shown in Figure 5, an increase in shape complexity as slight as 3% between watershed GTH 129 (36%) and watershed GTH 149 (39.85%) was associated with an increase in total phosphorus.

A third significant predictor in the total phosphorus model was the Patch Density for Aquatic Vegetation (PD_AV). This predictor refers to the vegetation growing within near-shore zones of lakes. The near-shore zone intercepts higher allochthonous loads of nutrients (Giblin et al. 1991). Based on this and the observed negative correlation of its Patch Density with phosphorus concentration in lakes, we believe that the PD_AV complex played the role of nutrient sink, provided that it would be present in contiguous non-fragmented patches. If there are a higher number of PD_AV patches represented by higher patch density, however, then we suspect they would likely reverse their role by allowing phosphorus compounds to reach the lakes. Such a scenario is illustrated by the example of watersheds GTH 131 and GTH 151. In GTH 131, the PD_AV complex was 10.27 and the concentration of total phosphorus was 0.12 mol [L.sup.-1]. Compare that to watershed GTH 151, where the PD_ AV complex was 29.5 and the total phosphorus concentration was 0.33 mol [L.sup.-1].


This research was successful in implementing landscape metrics to assess probable relationships between landscape factors, major nutrients and Chlorophyll a. Overall, shape complexity and patch density indices for broad leaf vegetation were found to be the most influential predicting factors, especially for total nitrogen and chlorophyll a. In addition, the MAT complex was a significant predicting variable for total phosphorus. Landscape metrics also helped us predict the probable root zone interactions in support of the observed relationships between lakes and landscapes. Future research could restrict statistical modeling to landscape metrics of broad leaf vegetation communities along with metrics for the MAT complex and MNT complex. Sampling more lakes at multiple times during a thaw season, as well as acquiring and using multi-temporal satellite imagery, would further enhance our understanding of temporal changes within nutrient levels and their relationships to land cover phenology.

This study has laid a foundation for future research by identifying significant land cover categories that are correlated with the nutrient conditions of Arctic lakes. Scientists have predicted that with warming temperatures more broad leaf vegetation categories will prevail in the Arctic (Tape et al. 2006). Similarly, an increasing dominance of the MAT complex is expected in the near future (Walker et al. 2001). Therefore, the presented landscape factors should be useful in predicting the nutrient conditions of lakes as land covers are modified in the future.


This research was funded by National Science Foundation grant 051604. We would also like to thank the anonymous reviewers for their helpful comments and suggestions, which led to substantial improvements of the manuscript.


Allan, J.D. and L.B. Johnson. 1997. Catchment-scale analysis of aquatic ecosystems. Freshwater Biology 37: 107-112.

Chapin, ES. III and G.R. Shaver. 1988. Differences in carbon and nutrient fractions among arctic growth forms. Oecologia 77: 506-514.

Gergel, S.E., M.G. Turner, J.R. Miller, J.M. Melack and E.H. Stanley 2002. Landscape indicators of human impacts to riverine systems. Aquatic Science 64: 118-28.

Giblin, A.E., KJ. Nadelhoffer, G.R. Shaver, J.A. Laundre and AJ. McKerrow. 1991. Biogeochemical diversity along a riverside toposequence in arctic Alaska. Ecological Monographs 61: 415-436.

Griffith, J.A., E.A. Martinko, J.L. Whistlern and K.R Price. 2002. Interrelationships among landscapes, NDVI, and stream water quality in the US central plains. Ecological Applications 12: 1702-18.

Hamilton, T.D. 1986. Late Cenozoic glaciation of the Central Brooks Range. In T.D. Hamilton, K.M. Reed and R.M. Thorson (Eds.), Glaciation in Alaska: the Geologic Record. Alaska Geological Society, pp. 9-49.

Hershey A.E., S. Beaty, K. Fortino, M. Keyse, P.P. Mou, W.J. O'Brien, A.J. Ulseth, G.A. Gettel, P.W. Lienesch, C. Luecke, M.E. McDonald, C.H. Mayer, M.C. Miller, C. Richards, J.A. Schuldt and S.C. Whalen. 2006. Effects of landscape factors on fish distribution in arctic Alaskan lakes. Freshwater Biology 51: 39-55.

Hess, K. R., M.C. Abbruzzese, R. Lenzi, M.N. Raber and J.L. Abbruzzese. 1999. Classification and regression tree analysis of 1000 consecutive patients with unknown carcinoma. Clinical Cancer Research 5(11): 3103-3410.

Intermap. 2010. Product Handbook and Quick Start Guide, Standard Edition. Intermap, p. v 4.4.

International Panel on Climate Change. 2007. Climate Change 2007: Synthesis Report. Contribution of Working Groups I, II and III to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change. IPCC, Geneva, Switzerland, pp. 104.

Johnson, L.B. 1990. Analyzing spatial and temporal phenomena using geographic information systems: a review of ecological applications. Landscape Ecology 4: 31-43.

Jones, K.B., A.C. Neale, M.S. Nash, R.D. Van Remortel and J.D. Wickham. 2001. Predicting nutrient and sediment loadings to streams from landscape metrics: a multiple watershed study from the United States Mid-Atlantic Region. Landscape Ecology 16: 301-312.

Keller, K., J. Blum and G. Kling. 2007. Geochemistry of soils and streams on surfaces of varying ages in arctic Alaska. Arctic, Antarctic, and Alpine Research 39: 84-98.

King R.S., M.E. Baker, D.F. Whigham, D.E. Weller, T.E. Jordan, P.F. Kazyak and M.K. Hurd. 2005. Spatial considerations for linking watershed land cover to ecological indicators in streams. Ecological Applications 15: 137-153.

Levine, M.A. and S.C. Whalen. 2001. Nutrient limitation of phytoplankton production in Alaskan Arctic Foothill Lakes. Hydrobiologia 455: 189-201.

Liu, Z., G.R. Hallberg and G.P. Malanson. 1997. Structural equation modeling of dynamics of nitrate contamination in ground water. Journal of the American Water Resources Association 33(6): 1219-125.

Marion, G.M., SJ. Hastings, S.F. Oberbauer and WC. Oechel. 1989. Soil-plant element relationships in a tundra ecosystem. Holarctic Ecology 12: 296-303.

McGarigal, K., S.A. Cushman, M.C. Neel and E. Ene. 2002. FRAGSTATS: Spatial Pattern Analysis Program for Categorical Maps. Retrieved April 11, 2011, from research/fragstats/fragstats.html.

Medley, K.E., B.W. Okey. G.W. Barrette, M.F. Lucas and W.H. Renwick. 1995. Landscape change with agricultural intensification in a rural watershed, southwestern Ohio, U.S.A. Landscape Ecology 10(3): 161-176.

Ostendorf, B. and J.E Reynolds. 1993. Relationships between a terrain-based hydrologic model and patch-scale vegetation patterns in an arctic tundra landscape. Landscape Ecology 8(4): 229-237.

Riera, J.L., JJ. Magnuson, T.K. Kratz and K.E. Webster. 2000. A geomorphic template for the analysis of lake districts applied to the Northern Highland Lake District, Wisconsin, USA. Freshwater Biology 43: 301-318.

Serreze, M.C.,J.E. Walsh, ES. Chapin, T. Osterkamp, M. Dyurgerov, V. Romanvosky, W.C. Oechel, J. Morison, T. Zhang and R.G. Barry. 2000. Observation evidence of recent change in the northern high-latitude environment. Climatic Change 46: 159-207.

Schindler, D.W. and J.P. Smol. 2006. Cumulative effects of climate warming and other human activities on freshwaters of arctic and subarctic North America. Ambio 35: 160-168.

Strahler, A.N. 1952. Dynamic basis of geomorphology. Geological Society of America Bulletin 63: 923-938.

Stine R., D. Chaudhuri, P. Ray, P. Pathak and M. Hall-Brown. 2010. Comparing Image Classification Techniques for Arctic Tundra Land Cover, Toolik Lake, Alaska. GIScience and Remote Sensing 47(1): 78-98.

Tape, K., M. Sturm and C. Racine. 2006. The evidence for shrub expansion in Northern Alaska and the Pan-Arctic. Global Change Biology 12: 686-702.

Walker, M.D., D.A. Walker and N.A. Auerbach. 1994. Plant communities of a tussock tundra landscape in the Brooks Range Foothills, Alaska. Journal of Vegetation Science 5: 843-866.

Walker, D.A., J.G. Bockheim, F.S. Chaopn III, W. Eugster, F.E. Nelson and C.L. Ping. 2001. Calcium-rich tundra, wildlife, and the 'Mammoth Steppe'. Quaternary Science Reviews 20: 149-163.

Wetzel, R.G. 2001. Limnology: Lake and River Ecosystems. Third Edition. Academic Press.

Prasad Pathak, GIS-Remote Sensing, NIIT University, Rajasthan, India, Ernail: <>; Roy Stine, Elisabeth Nelson and Zhi-Jun Liu, Department of Geography, University of North Carotina at Greensboro, Greensboro, NC, USA, Email: <rstine>, <esnelso2>, <z_liu>@<>; Anne Hershey, Department of Biotogy, University of North Carolina at Greensboro, Greensboron, NC, USA, Email: <>; Stephen Whalen, Department of Environmental Sciences and Engineering, University of North Carolina, Chapel Hill, NC, USA, Email: <Steve_Whalen@>.

DOI: 10.1559/15230406394187
Table 1. Lakes and Sampling Dates.

Lake #               Sampling Date

115, 116, 117, 118   29 June 2001
119, 120, 121, 122    7 July 2001
127, 128, 129, 130    8 July 2002
131, 132, 133, 134   10 July 2002
135, 136, 137, 138   12 July 2002
139, 140, 141, 142   28 June 2003
143, 144, 145, 146    3 July 2003
147, 148, 149, 150   14 July 2003
151, 152, 153, 154   24 July 2003
E4                   28 July 2003
123, 124, 125, 126    4 Aug 2003

Table 2. Vegetation community categories and species composition
(Stine et al. 2010; Walker et al. 1994).

Class                       Description

Barren complex              Barren surfaces, sparsely vegetated,
                            rocks covered with lichens

Moist Acidic Tussock        Eriophorum vaginatum, Carex bigelowii,
Tundra (MAT) complex        Betula nana, Salix pulchra, Sphagnum spp.

Moist Non-acidic Tussock    Salix reticulate, Saxifraga
Tundra complex              oppositifolia, Cares bigelowii, Carex
                            membranacea, Dryas integrifolia, Ledum
                            decumbens, Equisetum ravens

Shrub complex               Betula spp., Salix spp., Sphagnum spp.

Riparian complex            Eriophorum anguistifolium, Salix pulchra,
                            Salix alaxensis, Salix richardisonii

Fen Complex                 Carex rariflora, Carex rotundata, and
                            mosses like Sphagnum spp., Carex
                            chordorrhiza, Carex aquitilis, and
                            Tomentypnum nitens

Heath complex               Festuca altaica, Empetrum hermaphroditum,
                            Loiseleuria procumbens, Dryas octopetala,
                            Cassiope tetragona, Salix phlebophylla

Snowbed complex             Cassiope tetragona, Salix rotundifolia,
                            Arnica frigida

Mountain Meadow complex     Carex biglowii, Cassiope tetragona, Salix

Aquatic vegetation          Similar to fen but on lake-fringes

Water                       Lakes, streams, and rivers



Table 3. Description of landscape metric acronyms that
were significant in predicting lake chemistry measures.

Acronym      Description

LSI_Rip      Landscape shape index Riparian com

PD_Shrub     Patch density of Shrub complex

PD_AV        Patch density of Aquatic vegetation

MS-MAT       Mean shape index Moist Acidic Tundra

PD_B_Heath   Patch density of Heath complex at
             buffer level

PD_B_Fen     Patch density of Fen complex at buffer

PD_B_Rip     Patch density index Riparian complex at
             buffer level

Prct_B_Rip   Percentage of Riparian complex at
             buffer level

Table 4. Regression Model Summary.

      Dependent                                Standardized
No.   Variable        Regression Model         Coefficients      R2

1      Chla_V     1.130 + 0.043(LSI_Rip) +   LSI_Rip    0.649   0.52
                     0.003(PD_B_Heath)       PD_B_Heath 0.345

2        TN       14.65 + 0.58(Prct_B_Rip)   Prct_B_Rip 0.552   0.48
                     + 0.011(PD_B_Fen)       PD_B_Fen   0.196

3        TP        0.094 - 0.001(PD_AV) +    PD_AV     -0.334
                     0.001(PD_Shrub) +       PD_Shrub    0.52   0.52
                       0.004(MS_MAT)         MS_MAT     0.459
COPYRIGHT 2012 Cartography and Geographic Information Society, Inc.
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2012 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Pathak, Prasad; Stine, Roy; Hershey, Anne; Whalen, Stephen; Nelson, Elisabeth; Liu, Zhi-Jun
Publication:Cartography and Geographic Information Science
Article Type:Report
Geographic Code:1USA
Date:Oct 1, 2012
Previous Article:Integrating sketch maps with GIS to explore fear of crime in the urban environment: a review of the past and prospects for the future.
Next Article:An automated system for image-to-vector georeferencing.

Terms of use | Copyright © 2017 Farlex, Inc. | Feedback | For webmasters