UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Temporal coherence of nutrients and implications for understanding British Columbia lake water quality.. Jensen, Ernest Victor 2010-10-18

You don't seem to have a PDF reader installed, try download the pdf

Item Metadata

Download

Media
24-ubc_2010_spring_jensen_ernest.pdf [ 1.24MB ]
Metadata
JSON: 24-1.0069368.json
JSON-LD: 24-1.0069368-ld.json
RDF/XML (Pretty): 24-1.0069368-rdf.xml
RDF/JSON: 24-1.0069368-rdf.json
Turtle: 24-1.0069368-turtle.txt
N-Triples: 24-1.0069368-rdf-ntriples.txt
Original Record: 24-1.0069368-source.json
Full Text
24-1.0069368-fulltext.txt
Citation
24-1.0069368.ris

Full Text

  TEMPORAL COHERENCE OF NUTRIENTS AND IMPLICATIONS FOR UNDERSTANDING BRITISH COLUMBIA LAKE WATER QUALITY VARIATION  by ERNEST VICTOR JENSEN  A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF  MASTER OF SCIENCE in  The College of Graduate Studies (Environmental Sciences)   THE UNIVERSITY OF BRITISH COLUMBIA (Okanagan)  February 2010 © Ernest Victor Jensen, 2010  ii  ABSTRACT  Temporal coherence, or the degree to which lakes behave similarly through time, provides new insights into the relationship between extrinsic drivers such as climate, and synchronous variation in lake variables at scales beyond the individual lake and catchment. A review of relevant literature suggests coherence of nutrients is highest among proximate pairs of short water residence time lakes within a common drainage path. High connectivity and short water residence times reduce the potential for catchment and lake specific alteration of the discharge signal. Few studies to date have explored coherence of total nitrogen and total phosphorus among lakes. To evaluate nutrient coherence among British Columbia (BC) lakes, I assembled spring estimates of total nitrogen and total phosphorus values over the period 1977 to 2007 for twenty- six lakes. To optimize coherence, I utilized a depth composite mean from a single deep site on lakes with at least 10 years of spring data. All lakes occur in catchments with varying degrees of anthropogenic disturbance and reflect a broad range of lake morphology and climatic conditions in both coastal and interior areas of BC. I explored the potential for climate to cause nutrient coherence by determining whether discharge was synchronous among nearby drainages, and whether lake nutrient variability was dependent on antecedent catchment discharge. Subsequently, I tested whether nutrient coherence was dependent on lake-pair proximity, similarity in lake water residence time, or lake fertility. Temporal coherence was calculated as the average Pearson Product Moment Correlation (r). Discharge was found to be highly coherent, particularly among streams in southern BC, over distances of several hundred kilometers. Temporal coherence of total nitrogen among lakes at all spatial scales was very weak but positive. Coherence was greater and positive for total phosphorus. Phosphorus coherence was not dependent on lake pair proximity, and dependence on water residence time was limited to coastal lakes. Temporal coherence of phosphorus was significantly greater for oligo-mesotrophic lakes than eutrophic lakes across significant spatial scales. These findings should be considered when interpreting lake response to local stressors and setting lake management targets.    iii  TABLE OF CONTENTS  TEMPORAL COHERENCE OF NUTRIENTS AND IMPLICATIONS FOR UNDERSTANDING BRITISH COLUMBIA LAKE WATER QUALITY VARIATION .......................................................... ii ABSTRACT  ................................................................................................................... ii TABLE OF CONTENTS ................................................................................................. iii LIST OF FIGURES ........................................................................................................ vii LIST OF TABLES ......................................................................................................... viii LIST OF ABBREVIATIONS .......................................................................................... ix ACKNOWLEDGEMENTS ...............................................................................................x CHAPTER 1  TEMPORAL COHERENCE OF NUTRIENTS: A REVIEW OF RECENT FINDINGS AND IMPLICATIONS FOR UNDERSTANDING BC LAKE WATER QUALITY VARIATION. ............................................................................................1 INTRODUCTION ...................................................................................................................1 TEMPORAL COHERENCE ANALYSIS .....................................................................................3 LIMITED PREVIOUS STUDY OF NUTRIENT COHERENCE .........................................................4 THE CLIMATE SIGNAL AND COHERENCE ..............................................................................5 HYDROLOGIC VARIABILITY AS A CLIMATE SIGNAL .............................................................6 CONTRASTS IN COHERENCE OF CONSERVATIVE AND REACTIVE IONS ..................................8 FLOW PATH AND COHERENCE .............................................................................................9 WATER RESIDENCE TIME AND COHERENCE OF NUTRIENTS ................................................10 TROPHIC STATUS AND COHERENCE ...................................................................................10 SUMMARY  ..................................................................................................................12 THE NEED FOR FURTHER STUDY ........................................................................................12 THE BC CONTEXT .............................................................................................................14 HYPOTHESES  ..................................................................................................................16 SIGNIFICANCE  ..................................................................................................................16 REFERENCES  ..................................................................................................................18  CHAPTER 2 ESTABLISHING DATA SELECTION CRITERIA FOR TEMPORAL COHERENCE ANALYSIS OF NUTRIENTS IN BC LAKES .........................................................................................23 INTRODUCTION .................................................................................................................23 PURPOSE  ..................................................................................................................24  iv  METHODS  ..................................................................................................................25 Nutrient data .................................................................................................................25 Data analysis .................................................................................................................26 RESULTS  ..................................................................................................................26 Influence of detection limit ..........................................................................................26 Influence of sample date ...............................................................................................28 Influence of time series length .....................................................................................30 Influence of time series fragmentation .........................................................................31 Influence of sampling depth .........................................................................................32 Influence of numbers of sites .......................................................................................32 CONCLUSIONS  ..................................................................................................................33 REFERENCES  ..................................................................................................................35  CHAPTER 3  TEMPORAL COHERENCE OF TOTAL NITROGEN AND TOTAL PHOSPHORUS AMONG BC LAKES .....................................38 INTRODUCTION .................................................................................................................38 METHODS  ..................................................................................................................40 Lake selection ...............................................................................................................40 Water quality database .................................................................................................43 Stream hydrograph data ................................................................................................44 Temporal coherence analysis .......................................................................................45 Regressions linking nutrients to hydrology ..................................................................45 RESULTS  ..................................................................................................................46 Interannual variation in stream discharge ....................................................................46 Spatial aspects of nutrient concentrations and dependence on discharge ....................49 Temporal coherence of nutrients ..................................................................................50 Dependence of nutrient coherence on lake pair proximity ...........................................52 Dependence of nutrient coherence on lake water retention time .................................53 Dependence of nutrient coherence on nutrient concentration ......................................53 DISCUSSION  ..................................................................................................................55 The influence of hydrology ..........................................................................................55 Coherence of nutrients ..................................................................................................56 REFERENCES  ..................................................................................................................60    v  CHAPTER 4  SIGNIFICANCE OF FINDINGS AND NEXT STEPS ..........................65 INTRODUCTION .................................................................................................................65 SIGNIFICANCE OF FINDINGS ..............................................................................................66 Developing a broader understanding of nutrient coherence .........................................66 Application of findings to BC lake management .........................................................67 NEXT STEPS  ..................................................................................................................68 Developing a broader understanding of nutrient coherence .........................................68 Application of findings to BC lake management .........................................................70 REFERENCES  ..................................................................................................................73  APPENDICES .................................................................................................................76 Appendix 1A  Variation in coherence of a hypothetical data set following substitution of non-detect values with the minimum detectable concentration (mdc), substitution of half the mdc, or pairwise deletion. ...................................76 Appendix 1B  Variation in coherence for spring total dissolved phosphorus (TDP) in Christina and Sugar lakes following substitution of non-detect values with the minimum detectable concentration (mdc), substitution of half the mdc, or pairwise deletion (underlined values are below the mdc value of 3 µg/L). .......................................................................................................77 Appendix 2  Map of hydrometric and lake sites. .............................................................78 Appendix 3A  Spring total nitrogen concentrations (µg/L) for selected coastal and interior lakes of BC .................................................79 Appendix 3B  Spring total phosphorus concentrations (µg/L) for selected coastal and interior lakes of BC .................................................80 Appendix 4  One way analysis of long term spring nutrient concentration by area ...............................................................................81 Appendix 5A  Correlation (r) between lake annual spring TN concentration (µg /L) and antecedent (yr-1) annual mean discharge (m3/s) at proximate river sites. .......................................82 Appendix 5B  Correlation (r) between lake annual spring TP concentration (µg /L) and antecedent (yr-1) annual mean discharge (m3/s) at proximate river sites. .......................................83 Appendix 6A  Pearson product moment correlations (r) for spring total nitrogen among BC lakes. ................................................................84 Appendix 6B  Pearson product moment correlations (r) for spring total phosphorus among BC lakes. ...........................................................86  vi  Appendix 7A  Testing null hypothesis: lake pair correlations for spring TN not different from zero. ...........................................................90 Appendix 7B  Testing null hypothesis: lake pair correlations for spring TP not different from zero. ...........................................................91 Appendix 8A  Spring total nitrogen coherence (r) grouped by three ranges of lake water residence time (<1 yr, >1 yr to 5 yr, >5 yrs). .............................................................................................92 Appendix 8B  Spring total phosphorus coherence (r) grouped by approximate lake pair quartiles with water residence times of < 0.75 years, 0.75-1.2 years, 1.2 -5 years, and > 5 years. ...........................................................................................93 Appendix 9A  Spring total nitrogen coherence (r) among lake pairs grouped by 3 ranges of TN concentration: < 200 µg /L, 200-400 µg /L, and >400 µg /L. .........................................................94 Appendix 9B  Spring total phosphorus coherence (r) among lake pairs grouped by 5 ranges of TP concentration: <10 µg /L, 10-20 µg /L, and >25 µg /L. ..........................................................95   vii  LIST OF FIGURES  Figure 2.1  Variation in coherence of three hypothetical data sets (high coherence r=0.84; medium coherence r=0.58; low coherence r=0.24) following substitution of non-detect values with the minimum detectable concentration (mdc), substitution of half the mdc, or pairwise deletion.   .............................27  Figure 2.2 Sensitivity of temporal coherence estimates (r) to the number of years of record for nitrogen and phosphorus in Mabel and Mara lakes. All consecutive sets of years from 3 to 29 were used (e.g.: 29 values for 3-year sets; 1 value for final 29 year group).   ...................................................................................31  Figure 3.1   Standardized average annual discharge (Water Survey Canada data, 1977 to 2005) for large and small drainages in various districts: coastal (Shawnigan, Cowichan), Columbia (Okanagan River, Mission Creek), Shuswap (South Thompson River, Criss Creek), Fraser (Chilcotin River, San Jose River), Skeena (Bulkley River, Goathorn Creek), and Peace (Alces Creek, Beatton River).   .....................................................................................................................47  Figure 3.2  Dependence of correlations among standardized annual average discharge (Water Survey Canada data, 1977 to 2005) on the distance within and between drainages. Drainages are: Coastal (Shawnigan, Cowichan), Columbia (Okanagan River, Mission Creek), Shuswap (South Thompson River, Criss Creek), Fraser (Chilcotin River, San Jose River), Skeena (Bulkley River, Goathorn Creek), and Peace (Alces Creek, Beatton River).   .....................................48  Figure 3.3  Significant relationships between lake nutrient concentration (µg/L) and regional average annual discharge (m3/s) in the antecedent year.   .....................50  Figure 3.4  Relationship between distance (km) between lake pairs, and TP (left) and TN (right) correlations (r) of all lake pairs with more than 10 years concurrent data.   ....................................................................................................................53  Figure 3.5  Coherence (r) and standard errors bars for all lake pairs, grouped by spring nutrient concentration ranges TP (<10 ug/L, 10- 20 ug/L and >25 ug/L) and TN (<200 ug/L, 200-400 ug/L, >400 ug/L).   ........................................................55  Figure 4.1  Dependence of coherence (r) on phosphorus concentration from multiple coherence studies and lake districts.   ......................................................................70     viii   LIST OF TABLES  Table 1.1  Lake characteristics and average nutrient coherence for various lake groups.   ......................................................................................................................... 5  Table 2.1  Relationship between spring sampling date and nutrient concentration in Skaha Lake in 1985, and Kalamalka Lake in 1988.   ............................... 29  Table 2.2  Stepwise multiple regression for total phosphorus, total dissolved phosphorus, total nitrogen and nitrate nitrogen concentrations in Skaha and Kalamalka lakes   ............................................................................................... 29  Table 3.1  Lake metrics, and average nutrient values (µg/L) between 1977 and 2007. Number of years varies by lake and parameter from a minimum of 10 to a maximum of 31, and averages 23 years for total phosphorus (TP), and 19 for total nitrogen (TN).   ................................................................................. 43  Table 3.2  Area, location and Water Survey of Canada reference number for large and small drainages proximate to study lake areas.   ............................................ 44  Table 3.3  Average temporal coherence (r) for total phosphorus (TP) and total nitrogen (TN) between all lake pairs at provincial, coastal and interior locations, and at sub-group scales.   .................................................................................... 52  Table 3.4  Average temporal coherence (r) for total phosphorus (TP) and total nitrogen (TN) between all lake pairs grouped by nutrient concentration (µg/L).   ......................................................................................................... 54  Table 4.1  Average coherence and concentrations for TP, TDP and SRP among various lake groupings.   .......................................................................................... 69   ix  LIST OF ABBREVIATIONS  BC  British Columbia ELA  Experimental Lakes Area ENSO  El Nino Southern Oscillation LTER  Long term ecological research Km  kilometres µg/L  microgram per litre mdc  minimum detectable concentration NAO  North Atlantic Oscillation NH3-N ammonia nitrogen NO3-N nitrate nitrogen SD  standard deviation SE  standard error SO4  sulphate TKN  total kjeldhal nitrogen TN  total nitrogen TP  total phosphorus TDP  total dissolved phosphorus WRT  water residence time     x     ACKNOWLEDGEMENTS I wish to thank Dr. Jeff Curtis for his guidance during this research. I also thank my mentors over the course of my career, Drs. Jim Bryan and Rick Nordin; thank you for your belief in the value of long term ecological monitoring. I also gratefully acknowledge the assistance of Ministry of Environment for providing financial support, and the opportunity to undertake this project. In particular, I thank the following Ministry staff for helping with indentifying candidate lakes and data access: Kevin Rieberger, Les Swain, Deb Epps, and Rosie Barlak for coastal lakes; Bruce Carmichael and James Jacklin for Charlie and Tabor lakes; John Love and Greg Tamblyn for Kathlyn Lake data; Norm Zirnhelt, Tracy Chapman, Kym Keogh, Chris Swan and Jennifer Raifteiri-McArdle for Cariboo lakes data. I take full responsibility for the final selection of lakes, sites, variables and annual estimates of nutrient concentration. Without the sample collection efforts over many years by many Ministry of Environment staff, this research would not have been possible; to them, I owe much thanks, and hope this work will inspire continued interest in long term monitoring of BC lakes. Finally, very special thanks to my wife Linda and daughter Sierra, for inspiration, and understanding, as I spent many hours at the computer.  1   CHAPTER 1   TEMPORAL COHERENCE OF NUTRIENTS: A REVIEW OF RECENT FINDINGS AND IMPLICATIONS FOR UNDERSTANDING BC LAKE WATER QUALITY VARIATION   INTRODUCTION  Fundamental to the management of freshwater resources is an understanding of lake nutrient dynamics through time, and the ability to apportion the influence of extrinsic forces such as climate variability on lake water quality, relative to that which may be intrinsic to the lake and catchment or be driven locally due to land-use or environmental pollution (Scarsbrook, 2003). Without this broad long-term view of lake dynamics, our understanding of individual lake ecosystems can be constrained. Magnuson et al., (2004) refers to this metaphorically as being in the “invisible present” and “invisible place”; in other words being unaware of broader processes and longer term dynamics and antecedent conditions affecting short term observations on individual lakes and variables of interest. This concern is particularly acute for resource management agencies engaged in lake monitoring and management programs which typically lack long-term reference lakes within integrated monitoring networks. Lacking this broader perspective, each lake is considered largely unique in terms of catchment and climate interaction, and resulting water quality variability.   Climate is a strong driver of lake dynamics on interannual time scales, and climate variation and change are expected to profoundly affect lake ecosystems through altered hydrologic flowpaths, biogeochemical fluxes, and water residence times of surface waters (Carpenter et al., 1992; Melack et al., 1997; Schindler, 2001). Although much of our understanding of lake ecosystems arises from studies occurring at the lake and catchment scale, it is reasonable to expect that lakes and their catchments within a region should experience similar interannual variation in climate forces, and limnological variables strongly affected by climate might co-vary similarly through time in neighbouring lakes. In contrast, as the interannual variation of a variable is  2  increasingly governed by intrinsic factors such as internal nutrient recycling or complex biotic interactions, or local disturbance and inputs, the climate signal would be modified or attenuated (Magnuson et al., 1990). The degree to which lakes within a region behave similarly through time has recently been described as temporal coherence, or lake synchrony (Magnuson et al., 1990). Temporal coherence of parameters among diverse lakes, and across various spatial scales, offers a new means of assessing the role of extrinsic drivers such as climate or other regional drivers such as acid rain, on lake dynamics. Moreover, temporal coherence has practical importance, as it potentially enables extrapolation of results from a few well studied lakes to other lakes, and aid prediction of lake dynamics in the future (Magnuson et al., 2006).   Research on lake coherence in general, and nutrient coherence in particular, is still exploratory. While physical variables such as surface temperature, are shown to be highly coherent within a region (Baines et al., 2000), chemical variables, especially nutrients, appear to be only weakly to moderately coherent among nearby lakes. Coherence of reactive ions and nutrients among lakes, is attributed to a variety of factors including variability in material processing within catchments as a function of flow path complexity (Kratz et al., 1998; Webster et al., 2000) and proximity of lake pairs (Baron & Caine, 2000; Kling et al., 2000), or material processing as a function of lake water residence time (Sorrano et al., 1999). Coherence of nutrients among lakes also occurs where simultaneous anthropogenic nutrient load changes are common among catchments (Anneville et al., 2005; George et al., 2000; Kratz et al., 1998). Regardless of cause, temporal coherence of nutrients among lakes continues to be of much interest as it provides a pattern of response which establishes the importance of external drivers to lake ecosystems across larger spatial scales.    At the present time it is not known to what extent long term nutrient variation in BC lakes might be governed by climatic forces acting over larger spatial and temporal scales. I suggest that without this understanding, interpretation of long term trends in limnological variables related to eutrophication control, catchment protection initiatives, or fisheries and hydrologic manipulation in BC lakes, could be incomplete and  3  compromise water management decisions. Nutrients are important water quality measures linked to trophic classification (Carlson, 1977), trend assessment and water quality protection initiatives in BC (Nordin, 2005) and elsewhere (Anderson et al., 2005). Moreover, nitrogen and phosphorus concentrations are considered important sentinels of climate warming on freshwater ecosystems (Schindler, 2009). Surprisingly, these nutrients, in particular total nitrogen and total phosphorus, are poorly incorporated in coherence studies to date.  In this chapter, I further introduce analysis of temporal coherence as a measure of patterned variation among lakes. Given the limited number of North American and European studies which have explored coherence of nutrients, it is reasonable to first briefly chronicle this work. My intention is not to provide an exhaustive review of temporal coherence across all variables and landscapes, but rather to explain the basic organizing concepts, and review recent work describing coherence of nutrients. I will first however, review and illustrate temporal coherence using physical parameters and a general conceptual model put forward by Magnuson et al. (1990), that of climate acting as a signal and the catchment and lake serving as a receptor. I then consider ion reactivity and material processing, within catchment and lake environs, as the second conceptual model to organize a synthetic understanding of the coherence work to date for nutrients. Finally, I describe the relevance of these findings to lake management in BC, and establish the rationale to examine the extent to which nitrogen and phosphorous may vary synchronously among BC lakes. Based on previous work, and the BC context, I propose a number of hypotheses to guide further analysis, and discuss the significance of any observed coherence.  TEMPORAL COHERENCE ANALYSIS Temporal coherence for a parameter of interest is typically reported as the average statistical correlation among all possible pairs of lakes across the spatial scale of interest (Magnuson et al., 2006). Studies to date often report variable coherence as the average Pearson product-moment correlation (r) among lakes within a drainage or region (Baron & Caine, 2000; George et al., 2000; Kling et al., 2000; Soranno et al., 1999; Webster et  4  al., 2000). Correlation coefficients are typically calculated using annual or seasonal mean values for the same parameter from a lake-by-year matrix. Concerns and assumptions over use of time series data for coherence analysis (null hypothesis equal to zero for all lake pairs; equal variances; influence of autocorrelation in the time series; effect of maximum values; seasonal or long term trends) have been explored and generally not found to seriously affect assessment of coherence using correlations (Kling et al., 2000; Kratz et al., 1998; Magnuson et al., 1990). As temporal coherence aggregates multiple lake-pair correlations, and coherence estimates are not statistically independent, some studies employ a Bonferroni, or similar correction, to determine what proportion of the correlation coefficients are significantly different from zero (Anneville et al., 2005; Baron & Caine, 2000). The Bonferroni correction is a safeguard against multiple tests of statistical significance on the same data falsely giving the appearance of significance, as 1 out of every 20 hypothesis-tests is expected to be significant at the α = 0.05 level purely by chance.  LIMITED PREVIOUS STUDY OF NUTRIENT COHERENCE  Only six North American studies and one European study, have directly examined and reported interannual coherence of nutrients among lakes for 11 lake districts. The seminal work of Magnuson et al., (1990) examined a 7 year data set for 7 groundwater dominated lakes in the Wisconsin Long Term Ecological Research (LTER) lake district. This work is expanded by Kratz et al., (1998) to include 4 additional lakes connected by surface drainage in southern Wisconsin over a 13 year period. Subsequent studies include a 7 year study of a chain of 9 lakes in the Alaska Arctic LTER area (Kling et al., 2000), and a 10 year data set for 3 lakes in the Loch Vale and  4 lakes in the Green Valley lake chains of the Rocky Mountain front range in Colorado (Baron & Caine, 2000). A synthesis of the studies, along with new data for Adirondack and Qu’appelle lake systems and Tennessee reservoirs, is provided by Soranno et al. (1999). Shortly thereafter temporal coherence of chemical variables over a 14 year period was also examined for 4 lake districts in the upper Great Lakes region (Webster et al., 2000) and  5  among lakes within the English Lake District (George et al., 2000). Average nutrient coherence values reported in these studies are summarized in Table 1.1.   Table 1.1  Lake characteristics and average nutrient coherence for various lake groups.    THE CLIMATE SIGNAL AND COHERENCE  Climate variation on an interannual scale can be considered conceptually as a signal that is modified in complex ways by lake and catchment attributes to attenuate, transform, or delay the climate signal (Magnuson et al., 2006). Coherence is expected to be most evident when the climate signal has sufficient interannual variation and limited spatial variation, so as to cause easily measured variation, in a variable of interest (George et al., 2000; Kling et al., 2000). Perfect coherence is only conceivable among identical lakes for a variable which responds directly and faithfully to external drivers such as climate. In reality, various aspects of catchment local weather, soils, vegetation or lake morphometrics, trophic status, and hydrology can combine to alter and degrade the climate signal and reduce coherence among lakes. Thus the potential for high coherence should be greater for variables such as water temperature with direct mechanistic links to meteorological forcing, than chemical or biotic measures which are more often the product of complex interactions (Magnuson et al., 2006). Indeed, coherence of surface  6  water temperatures, resulting from direct signal transmission of solar radiant energy to the lake surface, is high among proximate lakes in North America (Benson et al., 2000; Magnuson et al., 2004) and Europe (Livingstone, 2008). For example, epilimnetic water temperatures, are highly coherent (r>0.8) within and among lakes spanning tens of kilometres to a few hundred kilometres in the upper Great Lakes Region of North America (Benson et al., 2000). Similarly, ice-off dates are highly coherent among lakes across significant spatial scales in both North America (Kratz et al., 1998; Magnuson et al., 2004) and northern Europe (Livingston et al., 2008). However, as lake specific exposure to the thermal signal is altered by variation in lake morphology or other factors governing the vertical transfer of heat (e.g. surface area to mean depth ratios, fetch, wind speed), the climate signal is modified or attenuated, and temporal coherence diminishes. For example, coherence of thermocline depth, is moderate (r ~ 0.5) within lake districts, and low (r < 0.26) between lake districts in the upper Great Lakes Region of North America (Benson et al., 2000). Thus solar radiation with low spatial variation enables high coherence of a physical variable, when lake specific modification of the climate signal is low.   Although precipitation and run-off are more spatially variable than temperature, these same concepts of the catchment and lake functioning in various ways to attenuate and modify the climate signal, also apply to coherence of variables affected by precipitation and consequent hydrologic variability.  HYDROLOGIC VARIABILITY AS A CLIMATE SIGNAL Hydrology is an important climate driver governing the dynamics of freshwater ecosystems (Poff & Allen, 1995; Schindler et al., 1996). Similar to the examples given above for temperature, temporal coherence should be high among neighbouring lakes for variables linked mechanistically to variation in precipitation and run-off. Furthermore, it is reasonable to expect neighbouring lakes to experience similar year to year variation in precipitation including anomalous periods of multiple year drought, or wetter years with greater annual precipitation. These interannual patterns in  7  precipitation should potentially produce corresponding variation in run-off and coherence of catchment discharge among neighbouring lakes. Although few lake studies to date have examined physical limnological measures with direct mechanistic links to discharge variation, in one case, lake water levels are shown to be highly coherent  (r=0.87-0.94) among a set of groundwater fed lakes in the Northern Wisconsin Long Term Ecological Research (LTER) area (Magnuson et al., 1990). However, when the analysis also included data for surface drainage lakes of the southern Wisconsin area, coherence of lake levels was reduced (r=0.55) (Kratz et al., 1998). Thus as diversity and complexity of water flow path increases, and catchments differ in the ratio of infiltration to run-off and evaporative losses, the climate signal is increasingly modified or filtered, and coherence diminishes among nearby lakes. Hydrology is a strong determinant of lake water chemistry. Within catchments, periods of higher flow contribute to greater export of sediments, solutes and nutrients, to downstream waters (Schindler et al., 1996). Conversely, during periods of drought and lower flow, water movement and flux of materials from catchments to lakes diminishes, and evapoconcentration of conservative ions, and in-lake scavenging of reactive solutes increase (Schindler et al., 1996; Webster et al., 2000). Conservative ions in lakes vary largely as a function of mechanistic processes of soil weathering, hydrologic transport along surface and groundwater flow paths, and evapoconcentration. Moreover, because biological and chemical reactions within lakes are relatively minor for conservative ions, outputs from a lake closely follow inputs (Soranno et al, 1999). Conversely, lakes retain or process a significant portion of their nutrient load as a function of loading and water residence time, and as water residence time increases, lake specific reactions are increasingly important to nutrient variability. Thus, contrasts in hydrology among catchments and among lakes may provide insights into the coherent tendencies of chemical variables.   8  CONTRASTS IN COHERENCE OF CONSERVATIVE AND REACTIVE IONS Studies of individual lake districts, in almost all cases, show greater coherence for conservative ions and measures, than for nutrients and reactive variables (George et al., 2000; Kratz et al., 1998; Soranno et al., 1999; Webster et al., 2000). For example, calcium, a conservative ion largely unaffected by chemical or biotic reactions along ground or surface water flow paths is highly synchronous (r=0.81) among groundwater fed lakes in the Northern Wisconsin LTER lake district, as well as in Southern Wisconsin surface water dominated lake district (r=0.76) (Soranno et al., 1999). In strong contrast, the average overall synchrony of total nitrogen and total phosphorus, is uniformly lower among both the Northern (r=0.36) and Southern (r=0.49) Wisconsin lake districts (Soranno et al., 1999). Similar to calcium, chloride is another conservative ion and among lakes within 3 surface water dominated lake districts on Canadian Shield bedrock, chloride is consistently more coherent (r=0.62-0.92) than nutrient variables such as total phosphorus (r=0.27-0.48) or total nitrogen (r=0.49-0.69) (Webster et al., 2000). And finally, within the English Lake District, alkalinity coherence among five lakes is high (r=0.94) relative to nitrate nitrogen (r=0.73) or dissolved reactive phosphorus (r=0.65). Thus in these diverse hydrogeological settings, coherence of nutrient variables is consistently lower relative to conservative ions.   It seems logical then, that among lakes where contrasts in flow path complexity (e.g. groundwater versus surface water; position within a series of lakes) and water residence time (WRT) are evident, coherence of nutrients should be reduced over landscapes and lakes which are similar in these processes. Clearly, mass balance steady state models, which incorporate water residence time, have been useful in developing a predictive understanding of the concentrations of various chemicals in waterbodies, most notably that of phosphorus (Dillon & Molot, 1996), nitrogen (Windolf et al., 1996), and dissolved organic carbon (Curtis, 1998). Conceptually, longer water residence times enable longer flow path and longer water contact with catchment soils as well as longer within lake processing. Combined these should contribute to greater individuality of chemical signals in downstream waters and thus lower coherence.    9  FLOW PATH AND COHERENCE  Hydrologic flow path and lake pair proximity are important to coherence expression. Contrasts in flow path are most pronounced between surface-water fed lakes and seepage lakes fed largely by groundwater. Groundwater flow is orders of magnitude slower than that of surface water. Where lake districts are largely dominated by groundwater flow paths, or incorporate both seepage and drainage lakes, coherence is reduced, particularly for TP. For example, among Northern Wisconsin LTER seepage lakes governed by groundwater flow paths, coherence is moderate for TN (r=0.573) but very low for TP (r=0.139) (Webster et al., 2000).   Direct surface water flow paths between catchment and lake, and from lake to lake, also contribute to higher coherence, and proximate lakes are generally more coherent than distant pairs. For example, among Arctic LTER lakes, where flow path dynamics are governed by permafrost melt-water flow along surface water drainages and shallow soil horizons, coherence is high (NO3-N r=0.691; SRP r=0.7) but diminishes with distance between lakes (Kling et al., 2000). In this setting, Kling et al. (2000) conclude coherence is primarily related to proximity through the consistent ion processing along the drainage, and secondarily to water residence time. Similarly among Colorado lakes, coherence of NO3-N is higher among proximate pairs of lakes within the drainage, than distant pairs within or between the Vale Loch and Green Lake drainages (Baron & Caine, 2000). Finally, in a much different setting, that of the Ontario ELA, Dorset and Red Lake areas, moderate coherence of both total nitrogen (r=0.416-0.689) and total phosphorus (r=0.267-0.467) occurs among lakes at high and low positions along the hydrologic flow path (Webster et al., 2000). The Ontario lakes (ELA, Red Lake and Dorset) lie on bed rock with thin soils, and surface drainage flow paths enable annual changes in precipitation, particularly multiple years of drought, to be conveyed synchronously to downstream lakes.      10  WATER RESIDENCE TIME AND COHERENCE OF NUTRIENTS  Support for water residence time (WRT) as an organizing concept, comes from a synthesis of data from a number of lake district studies (Sorrano et al., 1999). Using aggregated data for lake chains in Wisconsin, Arctic, Colorado Front Range, and Adirondack lake districts, Sorrano et al. (1999) shows the collective coherence of dissolved reactive variables (sulphate, dissolved silica, nitrate, ammonia) decrease with increasing lake district median WRT (r2=0.31, P=0.003). Colorado and North Wisconsin lakes serve as end members in this relationship. Snowmelt over thin soils of the Colorado Front Range, result in very short water residence times (0.02-0.08 yr). In these drainages, water moves quickly from catchment to lake and from lake to lake, and average coherence of dissolved reactive ions are high (r=0.79) (Sorrano et al., 1999). Conversely, average coherence of dissolved reactive variables is low (r=0.234) among the Northern Wisconsin LTER groundwater lakes with relatively long residence times (median ~9 yrs) (Sorrano et al., 1999). Interestingly, the average WRT of the Ontario Lakes reported by Webster et al., (2000) is greater than that of the Northern Wisconsin lakes. Thus low nutrient coherence among lakes with long water residence times is not universal. Webster et al. (2000) suggest the low coherence of Wisconsin groundwater lakes is related more to hydrologic setting, than to differences in water residence time among lakes.   TROPHIC STATUS AND COHERENCE  Trophic status as defined by nutrient concentration ranges, notably that of phosphorus, is a common descriptor of lake dynamics, and enables categorization of lakes based on similarities and differences in productivity (Carlson, 1977). Coherence studies reported to date are largely of oligo-mesotrophic systems (Arctic, Colorado, Adirondack, and Ontario lake districts). Only Wisconsin and the English lake districts provide lakes of higher and diverse trophic status. Nonetheless, three patterns are suggested from these limited data. First, coherence of nitrogen species is often greater than that of phosphorus. This occurs among Arctic LTER, all three Ontario lake districts, the English Lake district, and the Northern Wisconsin LTER lakes (Table 1.1).  11   Second, coherence of phosphorus tends to be greater when lake groupings incorporate lakes of lower and similar trophic status. For example Arctic LTER lakes are ultra- oligotrophic (e.g. SRP 3.8 µg/L; NO3-N 2 µg/L), and coherence of SRP (r=0.7) is high. Similarly, for Ontario lake districts average nutrient concentrations are low for TP (ELA 8.3 µg/L; Dorset 6µg/L; Red Lake 9.5 µg/L), and TN (ELA 350 µg/L; Dorset  237µg/L; Red Lake 329 µg/L), and coherence is moderate for TP (r=0.267-0.467) among lakes of each district (Webster et al., 2000). Finally among the combined Wisconsin lakes of higher TP status (47 µg/L) coherence of TP is low (r=0.15) (Kratz et al., 1998). In this study lowest coherence occurs among lake pairs including a dystrophic lake.  And finally, simultaneous changes in lake productivity related to land use can serve as a common driver within catchments. For example, among Southern Wisconsin lakes, higher coherence (TN r=0.311; TP r=0.674) is attributed to agriculture and urban landuse common along the drainage path (Sorrano et al., 1999). High coherence is driven by strong hydrologic connections among the three lower lakes where internal loads may not vary much year to year (consistently high), but large and similar interannual differences in external loads are common among the lakes (Sorrano, pers. comm.). Similar to the Southern Wisconsin lakes, high coherence of NO3-N (r=0.73) and SRP (r=0.65) occurs among productive English Lake District lakes where changes in fertilizer application and land-use are simultaneous and common among the catchments (George et al., 2000). Thus, in disturbed catchments, simultaneous landuse or nutrient load can contribute to higher coherence of both nitrogen and phosphorus.     12   SUMMARY  In summary, coherence of nutrients is higher among proximate lakes with direct surface water flow paths connecting lakes to catchments, and lake to lake. However, in most cases studied to date, the catchments and drainages with direct flow paths also have short water residence times. From the work to date the dependence of nutrient coherence on water residence time is unclear.   Nitrogen is generally more coherent than phosphorus. However, phosphorus coherence is greater among lakes of low trophic status and high connectivity and short flow paths.  In each study to date flow path complexity, water residence time and trophic status may act in concert to produce varying coherence among lakes. Although lake coherence is strongly dependent on external drivers, the importance of individual processes cannot be inferred from the current literature. Thus the limited work to date constrains establishing the fundamental processes governing patterns of nutrient coherence among neighbouring lakes. I suggest further study of nutrient coherence among lakes is warranted.  THE NEED FOR FURTHER STUDY  Further study, to more clearly define factors contributing to coherence of nutrient variables among lakes, should incorporate three important variables: spatial proximity, water residence time, and trophic status.  First, future studies of nutrient coherence should examine spatial scales beyond that of the individual district or drainage. Coherence studies cited above, are largely confined to that of lake pairs within drainages or districts of modest size. For example, the Arctic LTER lakes, Colorado Vale and Green Valley lakes, and Northern Wisconsin and Southern Wisconsin lakes occur within chains or districts of 15km or less. Similarly, for Dorset, ELA and Red Lake study areas of modest size, coherence estimates are constrained to among lakes within districts, rather than among districts.  13  Although limited regionalization of coherence findings is suggested (Kratz et al., 1998; Webster et al., 2000), the broader spatial application of these findings is unclear. Climate drivers such as snow accumulation and freshet, acting uniformly over large spatial scales, and beyond that of lake districts (Cayan et al., 1999), could contribute to coherence on spatial scales beyond that of the lake district. Similarly, multiple years of drought can act as a driver of lake ecosystems on significant spatial scales  (Webster et al., 2000). Certainly, examining coherence among districts over large spatial scales has been beneficial to understanding coherence of physical variables such as surface water temperature (Livingstone, 2008). Thus, determining whether common dynamics of chemical variables occur among lakes across spatial scales beyond that of neighbouring lakes, could be beneficial to understanding the spatial extent of chemical coherence and sensitivity to climate variation.   Second, further coherence work should incorporate lakes with diverse water residence times. The limited coherence work to date has focused largely on short residence time systems. Five of the seven lakes districts have residence times less than 1 year; only two districts have residence times greater than 1 year (Arctic and Northern Wisconsin LTER), and these two might well be considered atypical of drainage waterbodies common to lake management districts in North America. In the Arctic case, the range of water residence times is uniformly low (mean 1.7 + 0.4yrs), and coherence is attributed to consistent material processing along the lake chain flow path (Kling et al., 2000). In the Northern Wisconsin case, hydrologic setting and groundwater flow paths produce both longer soil water contact times as well as longer water residence times. Combined these contribute to individuality among lakes and lower coherence values for dissolved reactive ions. Whether groundwater lakes serve as a logical end member in the Sorrano et al. (1999) analysis for all reactive variables (both total and dissolved), including those such as TP, which is strongly associated with overland flow and erosion (Sorrano et al., 1996), is unclear.   Finally, future studies of coherence should include TN and TP and evaluate the potential role of lake trophic status in governing coherence. Much of the work to date  14  involves dissolved ions among lower productivity lakes (Arctic, Colorado, Ontario), and lake districts of relatively similar status, thus determining whether contrasts in coherence occur in relation to trophic status among nearby lakes is not possible. However, the Wisconsin lakes illustrate how catchment specific nutrient drivers and dissimilar trophic status can contribute to low coherence among lakes. In this case, coherence was low among dystrophic lakes, and lake pairs including a dystrophic lake (Kratz et al., 1998). Differences in trophic status can also arise from internal cycling of nutrients. For example, seasonal anoxia and return of phosphorus from sediments back into the water column is important to productive lake status (Nurnberg, 1984). It is conceivable that among lakes with varying degrees of within-lake cycling of nutrients, coherence would be reduced, over lakes where this intrinsic driver is absent. I suggest examination of additional lakes across the entire trophic gradient may provide useful insights into coherent tendencies of nutrients.    THE BC CONTEXT  Coherence is expected to be more readily detected among lakes where significant year to year variation occurs in a key driving variable (George et al., 2000; Magnuson et al., 1990). Climate variation across BC is strongly influenced by proximity to the Pacific Ocean. Strong westerly air flow from the Pacific collides and rises up and over coastal mountain ranges dropping rain and snow in an enhanced orographic effect.  Hydrographs of coastal streams are dominated by winter rains. Leeward of the coastal mountains precipitation decreases markedly and continental climate enables storage of precipitation as snow with increasing latitude and elevation. Discharge on interior streams is therefore snow melt dominated.  In addition to spatial variation in precipitation, interannual climate variation in the Pacific Northwest and BC is linked to Pacific Ocean circulation and climate anomalies associated with the El Nino-Southern Oscillation (ENSO) index (Melack et al., 1997). Similar to NAO, ENSO influences run- off during El Nino (dry) and La Nina (wet) phases over a large area of southern Canada from BC, through the prairies and into the Great Lakes region (Shabbar et al., 1997).  15  Southern British Columbia for example, tends to receive more snow during La Nina years and flows in the Columbia River system, are approximately 20% higher in La Nina years than El Nino and neutral years (Mote et al., 1999). Thus strong spatial and temporal variation in BC climate could combine to contribute to contrasts in coherence of variables among BC lakes.  In only a few cases has nutrient variation in BC lakes been considered as a function of varying climate (Nordin et al., 1985; Jensen and Epp, 2001; Regnier, 1998). Nevertheless, spring nutrient status and trend information is considered essential to development of lake specific management targets or objectives (Cavanagh et al., 1985; McKean et al., 1987; Nordin, 2005; Rieberger, 2007), guiding nutrient management programs (Allen & LeFloch, 2009; Nordin, 2005), and the enabling state of environment reporting for BC lakes.   Given the importance of nutrients to BC lake management, and the strong climate variation across this landscape, I propose to determine whether coherence occurs among representative hydrometric signals proximate to the study lakes. The degree of coherence among discharge sites could provide insights into the spatial nature and potential for interannual variation to contribute to coherent dynamics of nearby lakes. Subsequently, I propose to examine data from representative BC lakes for temporal coherence of nutrient variables. However, because, regional lake monitoring programs in BC do not stem from a unified network, I question whether aspects of sampling program variation over time, within and among regions, could impose significant data integration challenges and introduce data artefacts which could obscure coherence. Therefore, I propose to initially assemble nutrient data from a subset of representative lakes to examine whether data gaps, time series length, variations in the methods and timing of sampling, could individually or in combination compromise estimates of coherence of nitrogen and phosphorus. On the basis of those findings, I propose to assemble data from lakes across BC and assess nutrient coherence among lakes grouped at various spatial scales. To guide this work I propose a number of hypotheses.   16  HYPOTHESES Rather than explore the coherent tendencies among lakes at the lake district scale, where some commonality in geomorphic and hydrometric factors is expected, I initially ask whether coherence occurs among lakes at a larger spatial scale. Thus, my first expectation is that both nitrogen and phosphorus nutrients would demonstrate low but positive synchronous interannual variation among BC lakes. This expectation is premised on the presence of strong climate drivers acting across the BC landscape, as well as limited observations elsewhere that nutrient coherence is generally positive but low.  Thus, my hypothesis is that both nitrogen and phosphorus nutrients will demonstrate low but positive synchronous interannual variation among BC lakes. Related to this is my expectation that a spatially structured response may also occur as a result of hydrologic differences between coastal and interior run-off regimes, or individual drainage groups, similar to that of a lake district. Thus, I will test whether coherence is dependent on distances among lakes grouped at provincial, coastal, interior, and drainage scales. My second hypothesis is that parameter coherence among lakes will decrease with increasing water retention times (WRT). This follows on the work to date which suggests that longer and variable WRT’s within a lake district, increase lake-specific material processing of nutrient inputs, and contribute to lower coherence among lakes.  Finally, I propose to explore whether the strength of the coherence might be related to lake fertility as estimated by long term average nutrient concentrations. Therefore, my hypothesis is that coherence of nutrients among lakes will be dependent on lake status as described by nutrient concentration.  SIGNIFICANCE  Lakes are important to ecology and economies at local and provincial scales in BC (Nordin, 2005) and considerable investment occurs to protect surface water quality and  17  monitor its status and trends through time (Allen & LeFloch, 2009). Predicting and recognizing trends and patterns in lake ecosystems is fundamental to water resource management. Because climate variation and change are expected to increasingly stress freshwater ecosystems (Schindler, 2009), and because water protection initiatives are costly, coherence could provide a broader understanding of lake quality variation and ensure protection measures are adequate and appropriate. For example, understanding when lake nutrient variation was largely attributable to climate variation would ensure protection measures were not applied prematurely. Conversely, reduced coherence or changes in coherence of a lake relative to some norm, would suggest lake specific behaviour linked to local drivers such as changing landuse and anthropogenic nutrient load change, or intrinsic processes such as internal loading.   Resource allocation to lake monitoring, and the design of lake monitoring programs are common challenges among lake management and research agencies (Schindler, 2001). Therefore, I suggest that identifying lakes which respond similarly to climate or other external drivers, will enable enhanced monitoring of climate variability, selection of appropriate indicator or reference ecosystems within synoptic surveys or networks, and aid lake managers to recognize lakes which deviate from the regional norm.    18   REFERENCES Allen P.A. & LeFloch C. (2009) Okanagan Basin Master Management Plan Update 2009. Report to the Okanagan Basin Water Board. CLPA Consulting Ltd., 149p  Anderson J.N., Jeppersen E. & Sondergaard M. (2005) Ecological effects of reduced nutrient loading (oligotrophication) on lakes: an introduction. Freshwater Biology, 50, 1589-1593. Anneville O., Gammeter S.A. & Straile D. (2005) Phosphorus decrease and climate variability: mediators of synchrony in phytoplankton changes among European peri- alpine lakes. Freshwater Biology, 50, 1731-1746. Baines S.B., Webster K.E., Kratz T.K., Carpenter S.R. & Magnuson J.J. (2000) Synchronous behavior of temperature, calcium, and chlorophyll in lakes of northern Wisconsin. Ecology, 81, 815-825. Baron J.S. & Caine N. (2000) Temporal coherence of two alpine lake basins of the Colorado front range, USA. Freshwater Biology, 43, 463-476. Benson B.J., Lenters J.D., Magnuson J.J., Stubbs M., Kratz T.K., Dillon P.J., Hecky R.E. & Lathrop R.C. (2000) Regional coherence of climatic and lake thermal variables of four lake districts in the Upper Great Lakes Region of North America. Freshwater Biology, 43, 517-527.  Carlson R.E. (1977) A trophic state index for lakes. Limnology and Oceanography, 22, 361-369.  Carpenter S.R., Fisher S.G., Grimm N.B. & Kitchell J.F. (1992) Global change and freshwater ecosystems. Annual Review of Ecological Systems, 23, 119-139.   19  Cavanagh N., Nordin R. & Bryan J.E. (1994) Christina Lake Water quality assessment and objectives. Technical Appendix. Ministry of Environment Lands and Parks, BC. 106p.  Cayan D.R., Redmond K.T. & Riddle L.G. (1999) ENSO and hydrological extremes in the western United States. Journal of Climate, 12, 2881-2893.  Curtis P.J. (1998) Climatic and hydrologic control of DOM concentration and quality in lakes. In: Ecological Studies, 133, 93-103. Aquatic Humic Substances Ecology and Biogeochemistry. (eds) D. Hessin & L.J. Tranvik. Springer Press.  Dillon P.J. & Molot L.A. (1996) Long-term phosphorus budgets and an examination of a steady-state  mass-balance model for central Ontario. Water Resources, 30, 2273- 2280.  George D.G., Talling J.F. & Rigg E. (2000) Factors influencing the temporal coherence of five lakes in the English Lake District. Freshwater Biology, 43, 449-461.  Jensen E.V. & Epp P.F. (2001) Water quality trends in Okanagan, Skaha and Osoyoos lakes in response to nutrient reductions and hydrologic variation. BC Ministry of Environment, Penticton, BC. 17p. Kling G.W., Kipphut G.W., Miller M.M. & O’Briens W.J. (2000) Integration of lakes and streams in a landscape perspective: The importance of material processing on spatial patterns and temporal coherence. Freshwater Biology, 43, 477-497.  Kratz T. K., Sorrano P.A., Baines S.B., Benson B.J., Magnuson J.J., Frost T.M. & Lathrop R.C. (1998) Interannual synchronous dynamics in northern Wisconsin lakes in Wisconsin, USA. In: Management of Lakes and Reservoirs during Global Climate Change. (eds. D.G. George et al.) pp 273-287. Kluwer Academic Publishers, Netherlands.   20  Livingstone D.M. (2008) A change of climate provokes a change of paradigm: taking leave of two tacit assumptions about physical lake forcing. International Review of Hydrobiology, 93, 404-414.  Magnuson J.J., Benson B.J. & Kratz T. K. (1990) Temporal coherence in the limnology of a suite of lakes in Wisconsin, U.S.A. Freshwater Biology, 23, 145-159.  Magnuson J.J., Benson B.J. & Kratz T. K. (2004) Patterns of coherent dynamics within and between lake districts at local to intercontinental scales. Boreal Environment Research, 9, 359-369.  Magnuson J.J., Kratz T.K., Benson B.J. & Webster K.E. (2006) Coherent dynamics among lakes. In: Long Term Dynamics of Lakes in the Landscape. Long Term Ecological Research on North Temperate Lakes. Eds. J.J. Magnuson, T.K. Kratz and K.E. Benson. Oxford University Press. p 89-106.   McKean C.J.P., Nagpal, N.K. &. Zirhnelt, N.A. (1987) Williams Lake water quality assessment and objectives. BC Ministry of Environment. 75p.   Melack J.M., Dozier J., Golman C.R., Greeland D., Milner A.M. & Naiman R.J. (1997) Effects of climate change on inland waters of the Pacific coastal mountains and Western Great Basin of North America. Hydrological Processes, 11, 971-992. Mote P.W., Fluharty D., Francis R., Franklin J., Hamlet A., Hersham M., Holmberg M., Ideker K.G., Keeton W., Lettenmaier D., Leung R., Mantua N., Miles E., Noble B., Parandvash H., Peterson D.W., Snover A. & Willard S. (1999) Impacts of climate variablility and change: Pacific Northwest. A report of the Pacific Northwest Regional Assessment Group for the US Global Change Research  Program. JISAO/SMA Climate Impacts Group, University of Washington, Seattle, USA. 73p. Nordin R. (1985) Phosphorus in the Okanagan valley lakes: sources, water quality objectives and control possibilities. Ministry of Environment, Victoria, B.C. 103p.  21  Nordin R. (2005) Water quality objectives for Okanagan Lake: A first update and   overview report. Ministry of Environment, Victoria BC. 135p. Nurnberg G. (1984) The prediction of internal P loading in lakes with anoxic hypolimnia. Limnology and Oceanography, 29, 111-124. Poff N.L. & Allen J.D. (1995) Functional organization of stream fish assemblages in relation to hydrological variability. Ecology, 76, 606-627. Regnier R. (1998) Trend analysis of annual spring-overturn total phosphorus in 8 small lakes in southern Vancouver Island , British Columbia. Report to Ministry of Environment.   Rieberger K. (2007) Water quality assessment and objectives for Shawnigan Lake. Technical Appendix. Ministry of Environment, Victoria, B.C. 85p.  Scarsbrook M.R., McBride C.G., McBride G.B. & Bryers G.G. (2003) Effects of climate variability for long term water quality analyses. Journal of the American Water Resource Association, 39, 1435-1447.  Schindler D.W. ( 2001) The cumulative effects of climate warming and other human stresses on Canadian freshwaters in the new millennium. Canadian Journal of Fisheries and Aquatic Science, 58, 18-29.  Schindler D.W. (2009) Lakes as sentinels and integrators for the effects of climate change on watersheds, airsheds, and landscapes. Limnology and Oceanography, 54, 2349-2358.  Schindler D.W., Bayley S.E., Parker B.R., Beaty K.G., Cruikshank D.R., Fee E.J., Schindler E.U. & Stainton M.P. (1996) The effects of climatic warming on the properties of boreal lakes and streams at the Experimental Lakes Area, northwestern Ontario. Limnology and Oceanography, 41, 1004-1017.  22   Shabbar A., Bonsal B. & Khandekar M. (1997) Canadian precipitation patterns associated with the Southern Oscillation. Journal of Climate, 10, 3016-3027.  Sorrano P.A., Hubler S.L., Carpenter S.R. & Lathrop R.C. (1996) Phosphorus loads to surface waters: a simple model to account for spatial pattern of land use. Ecological Applications, 6, 865-878.  Sorrano P.A., Webster K.E., Riera J.L., Kratz T.K., Baron J.S., Bukaveckas P.A., Kling G.W., Whiter D.S., Caine N., Lathrop R.C. & Leavitt P.R. (1999) Spatial variation among lakes within landscapes: ecological organization along lake chains. Ecosystems, 2, 395-410.  Webster K.E., Sorrano P.A., Baines S.B., Kratz T.K., Bowser C.J., Dillon P.J., Campbell P., Everett J.F. & Hecky R.E. (2000) Structuring features of lake districts: Landscape controls on lake chemical responses to drought. Freshwater Biology, 43, 499-515 Windolf J., Jepperson E., Jensen J.P. & Kristensen P. (1996) Modeling of seasonal variation in nitrogen retention and in-lake concentration: a four-year mass balance study in 16 shallow Danish lakes. Biogeochemistry, 33, 25-44.   23  CHAPTER 2  ESTABLISHING DATA SELECTION CRITERIA FOR TEMPORAL COHERENCE ANALYSIS OF NUTRIENTS IN BC LAKES     INTRODUCTION  Recent analysis of long time series data for lakes, attribute temporal coherence, or shared interannual variation of nutrients among lakes, to climate forces, or landuse changes affecting neighbouring lakes similarly over time (Baron & Caine, 2000; Kling et al., 2000; Kratz et al. 1998; Magnuson et al., 1990; Sorrano et al., 1999; Webster et al., 2000). Temporal coherence, estimated as the average correlation among all possible lake pairs for a variable of interest, provides a means of understanding which variables and drivers lead to coherent dynamics among lakes over various spatial scales (Magnuson et al., 2004).  To date, no studies of lake temporal coherence are reported west of the Rocky Mountains. Given the importance of British Columbia (BC) freshwaters to the economy and environment, it is useful to ask whether various long term water quality data sets for BC lakes are suitable for analysis of temporal coherence. The majority of BC lakes have been studied to address concerns related to eutrophication, sewage or septic seepage inputs, influence of changing landuse patterns, or other non-point source nutrient or contaminant loading concerns (e.g. Boyd et al., 1985; Cavanagh et al., 1994; French & Booth, 2004; Holmes, 1996a,b,c,d; Jensen & Epp, 2001; McKean et al., 1987; Nordin, 2005; Rieberger, 2007; Zirnhelt et al., 1997). In all cases, nitrogen and phosphorus provide important water quality status and trend indicators linked to nutrient management efforts. Therefore, determining whether nutrient variation in BC lakes exhibits temporal coherence has practical significance.    24   PURPOSE  To facilitate selecting lakes, sites, and variables, for temporal coherence analysis among BC lakes, I initially questioned how variation in laboratory detection limits, and changes in sampling frequencies and protocols used by the regional offices over this multi-decadal time span, might affect coherence estimates. For example, minimum detectable concentration (mdc) limits for total phosphorus and nitrate decreased over the data collection period, and non-detectable concentration records were evident in the time series. Data points below the detection limit present significant challenges to data analysis (Helsel, 2005). Therefore, I explored how left censored data records might affect coherence estimates.   Although temporal coherence studies elsewhere use seasonal or annual means of either discrete or composite epilimnetic samples (Kratz et al., 1998; Sorrano et al., 1999) or volume weighted averages (Magnuson et al., 1990), seasonal data collection is uncommon for BC lakes, particularly for consecutive years. Spring data collection, when primary production is low and thermal stratification is absent or minimal, has been a common strategy among monitoring networks in BC over the period of 1977 to 2007. However, because for many BC lakes spring sampling dates can vary by weeks from one year to the next, and increasing solar radiation through the spring period could affect nutrient uptake and scavenging from surface waters (Barica, 1990), I questioned whether variation in spring sampling date could compromise coherence estimates. As well, given that coherence of a broad range of limnological variables is shown to be higher in decadal scale time series (Kratz et al., 1998; Magnuson et al., 1990; Webster et al., 2000), I questioned whether the short (i.e. < 10yrs) or fragmented time series for many BC lakes could confound coherence analysis. Other factors such as lake morphometry could also be important to temporal coherence (Magnuson et al. 1990), thus I questioned whether spatial integration of data using concentrations averaged over multiple depths or multiple sites within a lake, could potentially improve estimation of spring nutrient concentrations and temporal coherence.    25  To evaluate these issues, I employed a sequential and iterative process, beginning with the data censor issue, and then progressing from smaller to larger temporal and spatial scales.  I first evaluated the change in temporal coherence due to increasing proportions of non-detect values using both a hypothetical dataset, and a BC dataset. Next, I examined the seasonal effect of sampling date variation, and subsequently I examined the effect of time series length and fragmentation at the inter-annual level. Similarly for spatial considerations, I first examined the influence of sampling depth on temporal coherence, and then subsequently examined whether temporal coherence was improved by using data from multiple sites on a lake, versus using data from a single site. In each case I evaluated data sets which were expected to provide optimal coherence. From this work I established data selection criteria to optimize subsequent temporal coherence analysis of spring nutrients among BC lakes.  METHODS  Nutrient data  For temporal coherence analysis, I used total phosphorus (TP), total dissolved phosphorus (TDP), total nitrogen (TN), and nitrate nitrogen (NO3-N), as these are important and commonly reported nutrient variables in BC lake monitoring programs. Nutrients for all lakes were analyzed using standard analytical methods at a common series of laboratories (BC Environmental Laboratory 1977 to 1989; Zenon Laboratories 1989 to 1997; Environment Canada Pacific Environmental Science Centre 1997-2002; and Maxxam Analytical 2002-present). As such, at the provincial scale, the data are of high and comparable quality. Of note is that in 1997 and subsequent years, laboratory minimum detectable concentrations (mdc) decreased for TP, from 3 µg/L to 2µg/L, and for NO3-N from 20 µg/L to 2 µg/L. For some lakes, total Kjeldhal nitrogen (TKN) is reported in the provincial database rather than TN.  Where necessary, I report TN as the sum of TKN and NO3-N values, otherwise all values were reported as TKN for the lake. The majority of the data reported in this work is archived in the provincial Environment Management System (EMS) or in files at the Ministry of Environment regional offices.   26   Data analysis I assembled the data for each variable into a year by lake matrix. The influence of outliers and distributions were inspected using scatterplot matrices.  Normal distributions were expected given the nature of the comparisons, and for all possible lake pairs, Spearman rank correlations were similar to Pearson product-moment correlations (e.g. for TP, Spearmans = 0.0417 + 0.884 Pearson, (Prob >F, <0.001; r2=0.81). Therefore, to be consistent with the majority of previous studies, I report temporal coherence using the Pearson product-moment correlation (r) (Kratz et al., 1998; Magnuson et al., 2004; Webster et al., 2000). I use an average of correlations among multiple lake pairs to estimate temporal coherence at the spatial scale of interest, either provincial, regional, or by drainage.  RESULTS  Influence of detection limit  I tested two hypothetical data series of 20 paired values to determine how three different treatments of non-detectable (i.e. < mdc) data values might affect coherence estimates. The treatments were: retention of the mdc value, substitution of half the mdc, and pair-wise deletion of records below the mdc. Values of one series were selected to initially provide high (r=0.84), medium (r=0.58) and low (r=0.24) correlation. Values of the second series were uniformly and progressively decreased so the proportion below the mdc increased from 0% to 50%. Substitution of either the mdc or half the mdc, gave very similar correlations, with no change in coherence up to 50% affect levels (Figure 2.1; Appendix 1a).    27   Figure 2.1  Variation in coherence of three hypothetical data sets (high coherence r=0.84; medium coherence r=0.58; low coherence r=0.24) following substitution of non-detect values with the minimum detectable concentration (mdc), substitution of half the mdc, or pairwise deletion.  Coherence estimates based on pair-wise deletion of non-detect values were significantly different between 20% and 30% affect level groups.  Thus, for these hypothetical data, as the proportion of non-detect data increased beyond 20%,  28  coherence estimates based on pair-wise deletion were significantly different than those retaining the mdc or substituting half the mdc.  I examined this further using TDP data for Christina and Sugar lakes for which 22% and 32% of the records respectively, are below the mdc. In this case coherence using pair-wise deletion (r=0.327) was equivalent to retention of the mdc (r=0.337) whereas using half the mdc provides a similar but greater coherence (r=0.404) (Appendix 1b).  On the basis of this, I concluded that retention of mdc values, or pair-wise deletion of the mdc value may make little difference to coherence estimates when  non-detectable values make up to less than 20% of the data. However, as the proportion of non-detect data increased beyond 20%, pair-wise deletion could significantly decrease coherence estimates. Over estimation of coherence by substitution with the mdc however cannot be ruled out, as it may produce internally consistent sets of results. Nevertheless, as records below the mdc have informational value, I conclude that retention of the mdc value for the non-detect records is a reasonable compromise to a difficult data management issue. However, I recommend against inclusion of datasets where non- detect records exceed 20%.  Influence of sample date  I tested the effect of variation in spring sampling date on coherence using data from Skaha and Kalamalka lakes over the period 1985 to 2007.  Over this period, these lakes have a high number of multiple spring sampling dates, and for individual years, significant relationships were evident between spring date and nutrient concentration for TDP in Kalamalka Lake and NO3-N in both lakes (Table 2.1). The difference in sampling date between lakes was also high, ranging from 1-29 days and averaged 12 days. Thus, the large range in spring sampling dates, and the observed seasonal variation in these nutrient forms, could confound coherence estimates.    29  Table 2.1  Relationship between spring sampling date and nutrient concentration in Skaha Lake in 1985, and Kalamalka Lake in 1988.   Coefficients of determination for first spring sampling date nutrient concentrations between Skaha and Kalamalka lakes ranged from r2=0.258 for TP to r2=0.111 for TDP (Table 2.2). Residuals were then regressed against the difference in annual sampling dates between the two lakes. I found that the difference in sampling date between these two lakes, over 21 years, had no significant additional explanatory power at the p=0.05 level (Table 2.2).   Table 2.2  Stepwise multiple regression for total phosphorus, total dissolved phosphorus, total nitrogen and nitrate nitrogen concentrations in Skaha and Kalamalka lakes.    I conclude, variation in sampling date between lakes within the period of February 1 and May 1, may not significantly confound temporal coherence estimates.   Variable Regression r2 p >F TP Skaha x Kalamalka 1 0.258 0.019 residuals x days  2 0.156 0.076 TDP Skaha x Kalamalka 0.111 0.14 residuals x days  0.005 0.759 TN Skaha x Kalamalka 0.184 0.052 residuals x days  0.046 0.348 NO3 Skaha x Kalamalka 0.183 0.053 residuals x days  0.03 0.454 1  regression of spring nutrient concentration between lakes 2  regression of residuals from 1st regression against sampling date difference between lakes r 2  p > F  r 2  p > F  Skaha* TP 0.227 0.118 Kalamalka ** TP 0.071 0.358 TDP 0.308 0.061 TDP 0.34 0.029 NO3 0.861 <0.0001 NO3 0.726 0.015 TN 0.247 0.101 TN 0.535 0.062 * 1985, five dates March 26- April 24 ** 1988, four dates January 21 and April 27  30   Influence of time series length   I examined the influence of time series length on coherence using data for Mabel and Mara lakes in the Shuswap drainage. Direct flow paths between short water residence time lakes in drainages with low spatial variation geomorphology are considered to be important determinants of coherence (Kling et al., 2000; Kratz, et al. 1998). Thus, higher coherence might be expected between Mabel and Mara lakes, given they are hydrologically connected and proximate, have similar landuse, and are exposed to the same climate. Data for all consecutive sets of years from 3 to 29 (e.g. 29 values for 3- year sets; 1 value for final 29 year set) were used to calculate correlations for each nutrient variable. For this lake pair, both total and total dissolved phosphorus coherence increased as the length of record increased from 3 years (TP: r=0.48; TDP: r=0.32) to 11 years (TP r=0.78; TDP r=0.76) then stabilized with little increase in correlation to 29 years (mean coherence: TP r=0.73; TDP r=0.66)  (Figure 2.2). Much more consistent, but reduced temporal coherence, was apparent for nitrogen species (TN: r=0.4; NO3-N r=0.23), with little variation over the 29 year period. Thus, my initial concern that lower estimates of coherence might arise from concurrent data sets of less than a decade was supported, particularly for phosphorus variables. I conclude a minimum time series length of 10 years will optimize coherence estimates. However, time series longer than a decade may not necessarily yield appreciably greater coherence values.    31     Figure 2.2 Sensitivity of temporal coherence estimates (r) to the number of years of record for nitrogen and phosphorus in Mabel and Mara lakes. All consecutive sets of years from 3 to 29 were used (e.g.: 29 values for 3-year sets; 1 value for final 29 year group).  Influence of time series fragmentation   Given the relationship between TP coherence and time series length, I further examined the Mabel and Mara data to determine the influence of discontinuity or fragmentation in the time series, on estimates of temporal coherence. Correlations remained high after removal of every 5th (r=0.790), 3rd (r=0.782) and 2nd (r=0.887) value in the time series. Random exclusion of years to achieve a sample size of 10, provided an estimate of coherence slightly lower (mean r=0.696, n=10), but similar to the entire data series (mean r=0.76, n=29).  Random sub-sets of 7 years of TP data however, provided lower estimates of coherence (mean r=0.33, n=10) and were significantly different than groups of 10 (p<0.0001). Thus, I conclude time series of 10 years or more, whether contiguous or fragmented, should yield optimal estimates of temporal coherence of 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 3 5 7 9 11 13 15 17 19 21 23 25 27 29 Number of consecutive years in calculation C o h er en ce  ( r) TP r TDP r TN  r NO3  r 32  phosphorus, whereas inclusion of lake pairs for which there are less than 10 years in common, contiguous or otherwise, is likely to yield lower estimates of coherence.  Influence of sampling depth  I tested whether temporal coherence estimates for total forms of nutrients at a central site on the first sampling date, were improved by inclusion of all water column nutrient analyses, as compared to estimates based on only a single surface value. Because southern interior lakes are sampled using surface (1-10m) and deep (>20m) composites, and coastal, central, and northern interior lakes have often been sampled with discrete depths, I evaluated the effect of sampling depth for both sampling method groups.  I selected Shawnigan, St Mary, Prospect, and Lac La Hache lakes with discrete depth collection, and Skaha, Kalamalka, Mabel and Mara lakes with composite sample collection due to similar maximum time series length, and the range of trophic states these lakes provided. The average correlation or coherence among all possible lake pairs within the discrete depth group was marginally greater for composite depth data (TP r=0.292; TN r=0.157) than surface only (TP r=0.242; TN r=0.148). For the southern interior lakes however, combining surface and deep water composites provided higher correlations for TP (TP r=0.546 ; TN r=0.003) than using only the surface composite (TP r=0.387; TN r=-0.069), but little difference for TN. Given the consistently greater coherence obtained by depth composite data, particularly for TP, I conclude that temporal coherence estimates are improved through spatial averaging of all data for the water column.    Influence of numbers of sites  I questioned whether aggregation of data from multiple sites on a lake, would significantly improve temporal coherence estimates among pairs of lakes. Although many BC lakes had multiple sites with varying data records, in most cases one site for a lake has the longest record of collection. For smaller lakes of simple morphology, correlations in concurrent time series between sites were expected to be high. For example temporal coherence of TP and TN for proximate sites on lakes of simple basin  33  morphology, such as Wood Lake are high (TP r=0.929; TN r=0.860). However, for morphologically complex lakes, estimates of annual nutrient concentration may vary among sites due to spatial variation in nutrient inputs and incomplete mixing. I explored this concern using data for Okanagan Lake, a long narrow lake with previously noted spatial variation in spring TP (Nordin, 2005). In this case, average coherence of TP among the three basins of the lake is only moderate (r=0.548). Okanagan Lake spring TP data averaged across the three sites, provided only slightly greater estimates of average temporal coherence (r=0.487) with other oligotrophic lakes nearby (Christina, Mabel, Mara, Sugar lakes), than using data from just one central site on Okanagan Lake (r=0.467). However, correlations using an Okanagan Lake average value, were always significant (p < 0.05) with the other oligotrophic lakes, whereas individual Okanagan Lake site correlations were not significant in 54% of the comparisons. Similarly, for TN, temporal coherence estimates based on multiple sites (r=0.132) were similar to individual correlations with other southern interior oligotrophic lakes (r=0.138), however, few correlations in either case were significant. Thus, at least for TP, lake pair correlations based on an average Okanagan Lake spring concentration yielded greater coherence estimates. Amongst other lake pairs however, I conclude the slight improvement in temporal coherence estimates afforded by using data from all sites for a lake, are unlikely to justify the significant data collation effort.  CONCLUSIONS  Variation in analytical and sampling regimes over time, within and among sampling programs, are important concerns when broader spatial and temporal interests require pooling or merging of data sets.  Based on my analysis of both hypothetical and representative nutrient data from BC lakes, I conclude that temporal coherence estimates for BC lakes may be optimized by implementing the following data selection criteria:  1. Restrict temporal coherence analysis to data series with fewer than 20% non- detect values in the data series. For data series with non-detect data records of 20% or less, I advocate retention of the mdc value, given that the temporal  34  coherence estimates based on substitution of half the mdc or pair-wise deletion are not significantly different.  2. To reduce the potential for variation in sampling date to influence temporal coherences estimates among BC lakes, restrict temporal coherence analysis to data collected on the first sampling date, between February 1 and May 1.   3. As continuous or fragmented time series of less than ten years are shown here to yield lower estimates of coherence particularly for phosphorus, restrict the analysis to those lake pairs with at least 10 years of concurrent data,  4. Use an average of all water column data to estimate the spring nutrient concentration.  Temporal coherence estimates for phosphorus are incrementally greater among low nutrient lakes when all water column data are employed in the analysis as opposed to a single surface sample.   5. Use spring nutrient data from one central site. Spring nutrient estimates based on data from a number of sites are unlikely to provide higher estimates of temporal coherence unless significant spatial variability is apparent for the lake.  In summary the analyses employed here to screen and clean the data have been used to provide defensible and optimal estimates of temporal coherence of nutrients among BC lakes, and may serve as a guide for similar studies elsewhere. Further assessment of nutrient coherence at the provincial scale is clearly warranted. The greater temporal coherence of phosphorus relative to nitrogen in these exploratory analyses suggests a contrast in coherence may occur between phosphorus and nitrogen. Thus climate may more directly affect phosphorus variation in BC lakes, than nitrogen. Further analysis is warranted to determine the strength and spatial extent of this relationship, and from this, better inform lake nutrient status and trend interpretations by BC lake managers. These findings also suggest that lake phosphorus trend interpretation, using less than 10 years of data, may be compromised by variation related to climate. Thus 10 years may serve as a minimum lake data collection target for nutrients or other water quality measures, where the influence of climate variation is of interest.     35  REFERENCES Barica J. (1990) Seasonal variability of N:P ratios in eutrophic lakes. Hydrobiologia, 191, 97-103. Baron J.S. & Caine, N. (2000) Temporal coherence of two alpine lake basins of the Colorado front range, USA. Freshwater Biology, 43, 463-476.  Boyd I.T., McKean C.J.P., Nordin, R.N. & Wilkes, B. D. (1985) Skeena, Nass area, Kathlyn, Seymour, Round and Tyhee lakes: Water quality assessment and objectives. Technical Appendix. Ministry of Environment, 110p.  Cavanagh N., Nordin R. & Bryan J.E. (1994) Christina Lake Water quality assessment and objectives. Technical Appendix. Ministry of Environment Lands and Parks, BC. 106p.  French T.D. & Booth B.P.  (2004)  A Long-Term Strategic Plan for the Improvement of Water Quality in the Charlie Lake Watershed. Prepared for the Charlie Lake Conservation Society, c/o Box 720, Charlie Lake, BC, Canada, V0C 1H0. 183pp. (+Appendices).  Helsel D.R. (2005)  More than obvious: better methods for interpreting nondetect data. Env. Sci. & Technology, 39, 419-423.  Holmes G. B. (1996a)  State of water quality of Prospect Lake, 1980-1995. Ministry of Environment Lands and Parks, Victoria BC. http://www.env.gov.bc.ca/wat/wq/quality/wqpl/wqpl.html  Holmes G.B. (1996b) State of water quality of St. Mary Lake 1975-1995. Ministry of Environment Lands and Parks, Victoria BC. http://www.env.gov.bc.ca/wat/wq/quality/wqsml/wqsml.html#f14   36  Holmes G.B. (1996c) State of water quality of Quamichan Lake 1988-1995. Ministry of Environment Lands and Parks, Victoria BC. http://www.env.gov.bc.ca/wat/wq/quality/wqql/wqql.html  Holmes G.B. (1996d) State of water quality of Prospect Lake 1980-1995. Ministry of Environment Lands and Parks, Victoria BC. http://www.env.gov.bc.ca/wat/wq/quality/wqpl/wqpl.html#f3  Jensen E.V. & Epp P.F. (2001) Water quality trends in Okanagan, Skaha and Osoyoos lakes in response to nutrient reductions and hydrologic variation. BC Ministry of Environment, Penticton, BC. 17p. Kling G.W., Kipphut G.W., Miller M.M. & O’Briens W.J. (2000) Integration of lakes and streams in a landscape perspective: The importance of material processing on spatial patterns and temporal coherence. Freshwater Biology, 43, 477-497.  Kratz T. K., Sorrano P.A., Baines S.B., Benson B.J., Magnuson J.J., Frost T. M. & Lathrop R.C. (1998) Interannual synchronous dynamics in northern Wisconsin lakes in Wisconsin, USA. In: Management of Lakes and Reservoirs During Global Climate Change. (eds. D.G. George et al.) pp 273-287. Kluwer Academic Publishers, Netherlands.  Magnuson J.J., Benson B.J. & Kratz T. K. (1990) Temporal coherence in the limnology of a suite of lakes in Wisconsin, U.S.A. Freshwater Biology, 23, 145-159.  Magnuson J.J., Benson B.J. & Kratz T. K. (2004) Patterns of coherent dynamics within and between lake districts at local to intercontinental scales. Boreal Environment Research, 9, 359-369.    37  McKean C.J.P., Nagpal N.K. &. Zirhnelt N.A. (1987) Williams Lake water quality assessment and objectives. BC Ministry of Environment. Victoria, B.C. 75p.   Nordin R. (2005) Water quality objectives for Okanagan Lake: A first update and   overview report. Ministry of Environment, Victoria BC. 135p. Rieberger K. (2007) Water quality assessment and objectives for Shawnigan Lake. Ministry of Environment, Victoria, B.C. 18p. http://www.env.gov.bc.ca/wat/wq/objectives/shawnigan/shawn_lk_overview.pdf  Sorrano P.A., Webster K.E., Riera J.L., Kratz T.K., Baron J.S., Bukaveckas P.A., Kling G.W., Whiter D.S., Caine N., Lathrop R.C. & Leavitt P.R. (1999) Spatial variation among lakes within landscapes: ecological organization along lake chains. Ecosystems, 2, 395-410.  Webster K.E., Sorrano P.A., Baines, S.B., Kratz T.K., Bowser C.J., Dillon P.J., Campbell P., Everett J.F. & Hecky R.E. (2000) Structuring features of lake districts: Landscape controls on lake chemical responses to drought. Freshwater Biology, 43, 499-515  Zirnhelt N., Simpson J., Nordin R., Andrews K. & Pommen L.W. (1997) Horse Lake – Bridge Creek water quality assessment. BC Ministry of Environment Lands and Parks, Williams Lake, BC. 212 p http://www.env.gov.bc.ca/wat/wq/studies/horse97.pdf     38  CHAPTER 3  TEMPORAL COHERENCE OF TOTAL NITROGEN AND TOTAL PHOSPHORUS AMONG BC LAKES      INTRODUCTION  Limnologists have recently been interested in understanding the degree to which variables demonstrate temporal coherence or vary synchronously through time in neighbouring lakes (Magnuson et al., 2006). Climate is an important driver of lake dynamics (Carpenter et al., 1992; Schindler, 2001) and is expected to be a primary contributor to synchronous variation or temporal coherence among lakes at various spatial scales (Kratz et al., 1998; Magnuson et al., 1990; Sorrano et al., 1999; Webster et al., 2000). Because neighbouring lakes are exposed to the same climate, the extent to which temporal coherence occurs provides useful in-sights into the sensitivity of parameters and lakes, to climate variation (Magnuson et al., 1990) and furthers our understanding of the long term dynamics of important variables at the regional level (Webster et al., 2000; Zhang et al. 2010). Importantly, the predictability of variables to climate forcing governs extrapolation or regionalization of results and development of new conceptual models to explain lake dynamics (Livingstone, 2008; Magnuson et al., 2004).  Indeed, for some physical properties such as surface water temperature, ice off dates, and lake level, which are directly and mechanistically linked to climate variation, coherence is high among lakes over significant spatial scales (Benson et al., 2000; Magnuson et al., 2004). However, as variation in a lake property is increasingly the product of interaction between local climate and intrinsic aspects of catchment and lake interaction, the climate signal is altered (attenuated, delayed, extended) and coherence among lakes diminishes (Magnuson et al., 2006). Predicting nutrient variation and concentrations in lakes continues to be central to limnology and lake management (Reckhow and Chapra, 1983; Schindler et al., 1996). Although nutrient concentrations  39  are the product of complex interactions between external drivers such as hydrology, and catchment and lake specific attributes (soils and land use, aquatic food webs), coherent interannual variation in reactive ions and nutrients have been reported among North American (Magnuson et al., 1990; Kling et al., 2000; Kratz et al., 1998; Webster et al., 2000) and European lakes (Anneville et al., 2005; Bleckner et al., 2007; George et al. 2000).  Among neighbouring lakes, in uniform hydrologic and climatic settings, coherence of reactive ions and nutrients is found to be positive. Proximity or high connectivity, and short water residence times promote coherence of reactive variables among lakes with streams linking catchment to lake, and lake to lake (Magnuson et al., 2006). Conversely, long catchment flow paths and long lake water residence times may lead to increasing contrasts among lakes within a drainage, and thus lower overall coherence of reactive variables (Sorrano et al., 1999). Other factors such as simultaneous changes in nutrient loading (George et al., 2000) or landuse (Kratz et al., 1998) also contribute to coherence of nutrients among lakes. In the few landscapes studied to date, some, or all of these factors may simultaneously influence nutrient coherence among lakes (Kling et al., 2000; Kratz et al., 1998). While assessments of coherence among neighbouring lakes within catchments of relatively homogeneous geomorphology (i.e. a lake district), provides insights into spatial and temporal patterns of conservative ions, neither concentration (Quinlan et al., 2003) nor coherence of TN or TP are clearly associated with lake order or location along hydrologic flow paths within the landscape (Sorrano et al., 1999; Webster et al., 2000). Thus, establishing fundamental patterns, and clarifying the importance of flow path, water residence time or trophic state, to TN and TP coherence among lakes, requires further study. No coherence studies have focused on nutrients and fewer still have incorporated total nitrogen (TN) and total phosphorus (TP) data. Thus, the spatial extent of TN and TP coherence, as well as the explanatory power of lake pair proximity and catchment flow path, lake water residence time (WRT), or lake trophic state, to these important lake status indicators, is unclear. Examination of large scale regional data sets to assess the response of lakes to climate variation is a worthy challenge (Quinlan et al., 2003).  40   Temporal coherence has not been reported for lakes in the Pacific Northwest. Here, I assembled hydrometric, and spring nutrient data provided by BC regional lake monitoring programs over multi-decadal time scales. I expect there is a continuum of coherence across various spatial scales. Therefore, I considered the following questions: (1) Over what spatial scale are interannual patterns in catchment discharge hydrographs coherent and potentially acting as synchronous drivers of lake dynamics? (2) Do lake nutrient concentrations vary with regional hydrology, thus linking this climate driver to nutrient variation? (3) Is coherence of TN or TP observable among British Columbia (BC) lakes, and if so over what spatial scales? For example does coherence decrease with distance between lake pairs within and among drainages? (4) Does lake water residence time (WRT) influence coherence of either total nitrogen or total phosphorus? Based on the findings of previous work, I expected positive coherence for both TN and TP among lakes grouped at provincial and smaller spatial scales. However, differences in hydrologic regimes and timing of run-off between coastal lakes and interior lakes, could contribute to a spatially structured response. Given the importance of WRT to lake dynamics, I expected nutrient coherence should diminish among lakes with longer water residence times. However, as lake trophic status increases, the influence of climate variation on lake dynamics should be increasingly obscured by variation in nutrient loading and processing intrinsic to either catchment or lake. Therefore, (5) I also questioned whether nutrient coherence might demonstrate some dependence on trophic state. To the extent that temporal coherence occurs among BC lakes, this work should contribute to a broader understanding of the sensitivity of various forms of nutrients, and lakes in general, to climate variation and improve our ability to predict lake dynamics in the future.   METHODS  Lake selection  Lake selection was governed by data availability. Twenty-six lakes, with at least 10 years of nutrient data collection between 1977 and 2007, were chosen to reflect a broad range of lake and climatic conditions in both coastal and interior areas of BC (Table  41  3.1). These lakes express considerable morphologic and trophic diversity but are all drainage lakes within catchments of varying degrees of anthropogenic disturbance. Across this scale, precipitation and run-off vary considerably with latitude and proximity to the Pacific Ocean. I grouped lakes based on run-off regimes, as either Coastal with winter rain dominated hydrographs, or Interior with snow-melt dominated hydrographs. I used lake pair proximity to estimate spatial aspects of flow path. Proximate lakes are more likely to be within common surface water flow paths, and exposed to similar variation in hydrology, than would be the case for distant lake pairs. To further understand the importance of flow path and lake pair proximity to coherence, I also examined coherence among lakes grouped by major drainage membership (Columbia, Shuswap, Fraser, Skeena, Peace).  Nine of the lakes are from the Coastal zone of southern BC. Lizard, Shawnigan, Prospect, Quamichan, Glen, Fork and Stocking lakes occur on southern Vancouver Island, and St. Mary and Maxwell lakes are nearby on Saltspring Island. These relatively small and shallow lakes have relatively short water residence times (range 0.1-14 yr; av. 3yrs) and although these lakes are located within a 35 km radius of each other, none are interconnected by surface flows. Coastal lakes experience a mild maritime climate with limited incidence and duration of ice cover. Stream hydrographs, are dominated by winter precipitation, and little if any precipitation is stored as snow. An orographic effect of the Vancouver Island Range and Olympic Mountains produces a strong precipitation gradient across the Coastal area, with annual precipitation ranging from a high of 3612 mm at Lizard Lake on the west coast near Port Renfrew, to lows of 977 mm on Saltspring Island.   Fifteen lakes represent the Interior zone. Ellison, Wood, Kalamalka, Okanagan, Skaha, and Osoyoos, occur 300 km east of the coast, and form the Okanagan chain of lakes. Combined with Christina Lake 100 km to the east, these 7 lakes drain to the Columbia River. Sugar, Mabel and Mara lakes, 50 km northeast of the Okanagan basin are interconnected along the Shuswap drainage. Chimney, Horse, Lac La Hache and Williams lakes in the Cariboo area and Tabor Lake, farther north, occur some 400- 42  600km north of the Coastal zone, and flow to the Fraser River. Kathlyn Lake is located in the Skeena drainage and Charlie Lake, is located east of the continental divide in the Peace River drainage approximately 880km from the Coastal lakes. Annual precipitation ranges from 318 mm at Osoyoos in the south, to 600 mm at Prince George in central BC. In all cases a significant portion of the winter precipitation is stored at higher elevations as snow, and interior stream hydrographs are dominated by spring snow melt and freshet in late May and June. The majority of the interior lakes are considerably larger than the coastal lakes, and on average have longer water residence times (0.13-60 yrs; av. 20 yrs) (Table 3.1). Aspects of water quality status and trends for some but not all lakes have been previously reported (Boerger, 2001; Boyd et al. 1985; Cavanagh, et al. 1994; French & Petticrew, 2007; French & Carmichael, 1999; Bryan & Jensen, 1999; Jensen & Bryan, 2001; Jensen & Epp, 2001; Holmes, 1996a,b; McKean et al., 1987; Nordin, 2005; Pommen, 1996; Rieberger, 2003; Zirnhelt et al. 1997).    43   Table 3.1  Lake metrics, and average nutrient values (µg/L) between 1977 and 2007. Number of years varies by lake and parameter from a minimum of 10 to a maximum of 31, and averages 23 years for total phosphorus (TP), and 19 for total nitrogen (TN).    Water quality database   I focused my analyses on spring estimates for total nitrogen and total phosphorus. For these lakes, over the period of study, no records for TN were below the minimum detectable concentration (mdc). For the few TP records below the detection limits, I retained the mdc as the spring estimate. Spring is defined here as the first spring sampling date of the calendar year occurring between February 1 and May 1. During this period, thermal stratification is absent or minimal, and precedes spring freshet for interior lakes, and follows the winter storm period on the coast. Variation in sampling date among lakes, within this period has little or no affect on coherence estimates (see Chapter 2). As time series with 10 years of data, either contiguous or fragmented, are known to optimize coherence (see Chapter 2), I included only lake pairs with at least 10 Area Drainage Lake TP TDP TN NO3-N WRT SA V Zm Zx Latitude Longitude Interior Columbia Ellison 35 13 407 12 1.2 2.1 5 2.5 5 49.99074 119.40172 Wood 47 35 433 92 22 9.3 200 22 34 50.0794 119.38997 Kalamalka 10 5 287 132 51 25.9 1520 59 142 50.16681 119.34785 Okanagan 9 5 200 73 53 351 24640 76 230 49.87724 119.51258 Skaha 13 6 256 11 1.2 20.1 560 26 57 49.4116 19.58474 Osoyoos 21 9 315 63 0.7 23 400 14 63 49.05694 119.48074 Christina 7 6 104 2 4.5 25.1 930 37 54 49.05367 118.22703 Shuswap Sugar 6 4 181 103 0.6 20.8 730 35 83 50.39979 118.52247 Mabel 6 4 145 89 2.8 59.9 7180 120 200 50.54064 118.73679 Mara 11 6 189 104 0.13 19.4 357 18 46 50.78966 119.00611 Fraser Williams 65 42 717 114 0.62 7.2 88.2 12 24 52.11858 122.07287 Lac le Hache 13 7 451 11 10 23 336 14.6 ND 51.83853 121.58548 Horse 19 12 350 3 3.5 11.6 175 15.2 34 51.58925 121.11721 Chimney 19 10 819 9 16.8 4.3 37.5 9.5 20.9 51.91556 121.96111 Tabor 25 13 406 34 0.7 4.1 220 5.4 9.2 53.91701 122.54293 Skeena Kathlyn 23 13 314 52 1.2 1.7 8 4.6 9.5 54.82372 127.20611 Peace Charlie 43 26 630 107 5 19 127 7 15 56.31991 120.9757 Coastal Shawnigan 7 4 230 105 1.2 5.4 64 12 52 48.63725 123.64064 St Mary 23 9 423 65 14 1.82 15.9 8.8 16.7 48.8899 123.5427 Prospect 14 9 366 74 0.8 0.7 4.1 6.9 13.5 48.51341 123.44248 Stocking 6 4 136 9 1 0.23 2.1 9 19 48.96008 123.82704 Fork 10 6 227 52 0.1 0.002 0.092 2.3 10 48.51941 123.48492 Quamichan 19 10 403 33 1.3 0.313 13.7 4.7 8.2 48.79908 123.66489 Glen 21 11 292 403 0.4 0.17 1.2 6.4 14 48.43752 123.52244 Maxwell 10 5 249 20 7 0.3 2.3 7.7 19.2 48.82221 123.54146 Lizard 6 3 81 ND 0.75 0.009 0.66 7.5 15.5 48.6062 124.22345 TP=total phosphorus (µg/L); TDP=total dissolved phosphorus (µg/L);TN=total nitrogen (µg/L); NO3-N=nitrate nitrogen as nitrogen (µg/L) WRT= water residence time (yrs); SA=lake surface area (km2); V=lake volume in Mm3; Zm=mean depth (m); Zx=max depth (m); ND=no data 44  years of concurrent data in coherence estimates. As well, because depth integrated averages can provide higher coherence estimates than surface samples (see Chapter 2),  I used a depth integrated average concentration from a single main site on each lake, typically a central deep site, to represent spring nutrient conditions and accommodate variation in sampling protocols. All nutrient samples were analyzed using accepted standard analytical methods at a common series of laboratories.    Stream hydrograph data  To estimate hydrographic variation I used average annual discharge records for 12 drainages proximate to the lakes of interest, as regional climate proxies of discharge. My intention here was to develop a general understanding of the spatial nature of any synchronous interannual hydrologic variation within and among lake districts. To aid this spatial analysis, discharge records for 2 drainages of contrasting size, within each drainage area were obtained from Water Survey of Canada records for the period 1976 to 2005 (Table 3.2). Distances between hydrometric sites within and among lake drainage areas ranged from 15 to 100km and 141 to 861km respectively.  Table 3.2  Area, location and Water Survey Canada reference number for large and small drainages proximate to study lake areas.     Area Drainage name Area (km2) Latitude Longitude WSC No. Skeena Goathorn Creek near Telkwa 126 54 38' 50" 127 07' 20" 08EE008 Bulkley River at Quick 7350 54 37' 05" 126 53' 55" 08EE004 Fraser San Jose River above Borland 1990 52 04' 37" 121 59' 27" 08MC040 Chilcotin River below Big Creek 19300 51 50' 52" 122 39' 11" 08MB005 Shuswap Criss Creek near Savona 490 50 53' 04" 120 57' 54" 08LF007 South Thompson R at Chase 16200 50 45' 54" 119 44' 25" 08LE031 Columbia Mission Creek near E Kelowna 811 49 52' 44" 119 24' 47" 08NM116 Okanagan River at Penticton 6090 49 29' 44" 119 36' 55" 08NM050 Coastal Shawnigan Creek near Mill Bay 92 49 39' 29" 123 34' 08" 08HA033 Cowichan River near Duncan 826 48 46' 22" 123 42' 44" 08HA011 Peace Alces River at 22nd Base Line 303 56 10' 01" 120 09' 17" 07FD004 Beatton River near Ft. St. John 15600 56 16' 48" 120 42' 20" 07FC001  45  Temporal coherence analysis I assembled annual stream discharge (not shown) and spring lake nutrient values (Appendix 3) into year-by-site matrices and used the average Pearson product-moment correlation coefficient (r) to estimate temporal coherence of stream discharge and spring nutrient concentrations within and among coastal and interior lake areas. Similar to the work of others, average correlations were considered as a measure of temporal coherence, not as an inferential statistic (Magnuson et al., 1990). Temporal coherence aggregates multiple lake-pair correlations, therefore, coherence estimates are not statistically independent.  I employed a Bonferroni correction, to determine what proportion of the correlation coefficients, were significantly different from zero. The Bonferroni correction is a safeguard against multiple tests of statistical significance on the same data falsely giving the appearance of significance, as 1 out of every 20 hypothesis-tests is expected to be significant at the α=0.05 level purely due to chance.   Regressions linking nutrients to hydrology  I used linear regression to assess the dependence of temporal coherence of annual average discharge on proximity or distance between drainages. Similarly, I used linear regression to broadly determine whether lake nutrient concentrations were responding to local variation in mean annual discharge over the preceding year, or whether coherence among lake pairs could be explained by proximity or distance between lake pairs. To test whether nutrient coherence varied with lake water residence time (WRT) I estimated coherence of both TN and TP among multiple lake pairs grouped by WRT. For TN, equal number of lakes were assigned to WRT groups of  < 1 year, 1-5 years, and > 5years.  Because of the larger number of lake pair combinations available for TP, lake pair combinations were divided into approximate quartiles with 2 groups at or below WRT of approximately 1 year (<0.75 yr; 0.75-1.2 yrs) and 2 groups having water residence time greater than a year (1.2-5 yrs; >5 yrs).  Similarly, to test whether a relationship occurred between nutrient coherence and nutrient concentration, I calculated coherence among lakes grouped over three  46  concentration ranges for both TN (<200 µg/L; 200-400 µg/L; >400 µg/L) and TP (<10 µg/L; 10-20 µg/L; >20 µg/L). For TP, the concentration ranges reflect approximate boundaries between oligotrophic, mesotrophic and eutrophic states. Finally, I used analysis of variance to determine if differences among lake groups based on WRT or concentration groups were significant.   RESULTS   Interannual variation in stream discharge  Standardized average annual discharge plots show similar interannual patterns for the hydrograph pairs within each of the drainage areas (Figure 3.1). Moreover, cyclical interannual patterns of high and low discharge were visually evident in the standardized discharge plots, particularly for Coastal, Okanagan and Shuswap drainages. For these systems, multi-year periods of higher than average run-off were evident for years 1981- 84, 1990-1991, and 1996-2000. Periods of lower discharge were common among these same districts 1985-1989, 1992-1995, and 2001-2007.   47     Figure 3.1   Standardized average annual discharge (Water Survey Canada data, 1977 to 2005) for large and small drainages in various districts: coastal (Shawnigan, Cowichan), Columbia (Okanagan River, Mission Creek), Shuswap (South Thompson River, Criss Creek), Fraser (Chilcotin River, San Jose River), Skeena (Bulkley River, Goathorn Creek), and Peace (Alces Creek, Beatton River).   -3.000 -2.000 -1.000 0.000 1.000 2.000 3.000 19 77 19 79 19 81 19 83 19 85 19 87 19 89 19 91 19 93 19 95 19 97 19 99 20 01 20 03 20 05 Skeena Bulkley Goathorn -2.000 -1.000 0.000 1.000 2.000 3.000 4.000 5.000 19 77 19 79 19 81 19 83 19 85 19 87 19 89 19 91 19 93 19 95 19 97 19 99 20 01 20 03 20 05 Cariboo Chilcotin San Jose -3.000 -2.000 -1.000 0.000 1.000 2.000 3.000 19 77 19 79 19 81 19 83 19 85 19 87 19 89 19 91 19 93 19 95 19 97 19 99 20 01 20 03 20 05 Coastal Cowichan Shawnigan -3.000 -2.000 -1.000 0.000 1.000 2.000 3.000 4.000 19 77 19 79 19 81 19 83 19 85 19 87 19 89 19 91 19 93 19 95 19 97 19 99 20 01 20 03 20 05 Columbia Okanagan Mission -2.000 -1.000 0.000 1.000 2.000 3.000 4.000 19 77 19 79 19 81 19 83 19 85 19 87 19 89 19 91 19 93 19 95 19 97 19 99 20 01 20 03 20 05 Shuswap S Thompson Criss -2.000 -1.000 0.000 1.000 2.000 3.000 4.000 19 77 19 79 19 81 19 83 19 85 19 87 19 89 19 91 19 93 19 95 19 97 19 99 20 01 20 03 20 05 Peace Beatton Alces 48  Within each area, the hydrograph pairs were highly correlated:  Coastal (r=0.933), Columbia (r=0.902), Peace (r=0.883), (Skeena (r=0.837), Shuswap (r=0.769), Cariboo (r=0.713). Hydrograph coherence was also moderately strong between Okanagan River and Cowichan River (r=0.556), Okanagan and Chilcotin (r=0.529) and Shuswap and Chilcotin (r=0.776) drainage areas. Coherence was weakest among pairs with the Bulkley River (r=0.226) and strongest among those matched with Okanagan River (r=0.493). Using all sites, a strong relationship was evident between hydrograph coherence and distance (r2=0.748, p<0.0001) (Figure 3.2). Thus, to the degree that regional hydrology acts as a driver of lake dynamics, the potential for coherence to manifest itself among lake pairs should be greater among proximate lakes, particularly those in southern interior and coastal zones of BC, than those separated by more than a few hundred kilometers.     Figure 3.2  Dependence of correlations among standardized annual average discharge (Water Survey Canada data, 1977 to 2005) on the distance within and between drainages. Drainages are: Coastal (Shawnigan, Cowichan), Columbia (Okanagan River, Mission Creek), Shuswap (South Thompson River, Criss Creek), Fraser (Chilcotin River, San Jose River), Skeena (Bulkley River, Goathorn Creek), and Peace (Alces Creek, Beatton River).   R² = 0.748 0.000 0.200 0.400 0.600 0.800 1.000 0 200 400 600 800 1000 r km 49  Spatial aspects of nutrient concentrations and dependence on discharge  Spring total nitrogen and total phosphorus concentrations vary considerably among lakes at all spatial scales (Appendix 3A,B). However, both nitrogen and phosphorus were significantly greater in lakes of the Fraser, Skeena and Peace (TN, p<0.0001; TP p< 0.05) than Coastal, Columbia and Shuswap drainages, reflecting the prevalence of eutrophic and mesotrophic lakes sampled in those geographic areas (Appendix 4).   Correlation between lake spring TN concentration and antecedent (lag 1 year) annual average discharge (m3/s) at nearby hydrographic sites were negative for 6 of 9 Coastal lakes, 5 of 7 Columbia lakes, 1 of 3 Shuswap lakes, 3 of 5 Fraser lakes, and Kathryn Lake in the Skeena (Appendix 5A). None of the distributions (positive versus negative correlations) could be considered significantly different from zero at α=0.05 level. However for Kathlyn Lake a significant (r2=0.725, p=0.03) relationship (negative) occurred with antecedent Bulkley River discharge. In contrast, significant relationships (positive) occurred between TN in Charlie (r2=0.978, p=0.011) and Chimney (r2=0.293, p=0.056) lakes, and antecedent discharge on the Beatton and Chilcotin rivers respectively (Figure 3.3).  In contrast to the prevailing inverse relationship between TN concentration and discharge, correlations between lake spring TP concentration and antecedent discharge at proximate discharge sites were positive for 7 of 9 Coastal lakes, 4 of 7 Columbia lakes, all 3 Shuswap lakes, but only 2 of 5 Fraser lakes (Appendix 5B). However, none of the distributions were significantly different from zero at the α=0.05 level, and significant (positive) relationships between TP concentration and antecedent discharge were only evident for Christina Lake (r2=0.311, p <0.0007) and Okanagan Lake (r2=0.324, p <0.001) with antecedent discharge on Okanagan River (Figure 3.3).     50     Figure 3.3  Significant relationships between lake nutrient concentration (µg/L) and regional average annual discharge (m3/s) in the antecedent year.   Temporal coherence of nutrients Although a maximum of 322 pair-wise correlations are possible from 26 lakes, after screening for missing data (see Chapter 2) the number of lake pairs with 10 years of concurrent data was reduced to 192 for TN and 296 for TP. Correlations among lake pair combinations at the provincial scale, ranged considerably (TN r = -0.696 to 0.825;TP r = -0.996 to 0.980) (Appendix 6A,B).  However the average of all R² = 0.311 0 2 4 6 8 10 12 14 16 18 0 10 20 30 40 50 TP (ug/L) Annual average discharge (m3/s) on Okanagan River Christina Lake p =0.007 R² = 0.293 0 200 400 600 800 1000 1200 1400 60 80 100 120 140 TN (ug/L) Annual average discharge (m3/s) on Chilcoten River Chimney Lake R² = 0.978 0 200 400 600 800 1000 1200 0 20 40 60 80 TN (ug/L) Annual average discharge (m3/s) on Beatton River p=0.011 Charlie Lake R² = 0.725 0 100 200 300 400 500 600 0 50 100 150 200 TN (ug/L) Annual average discharge (m3/s) Bulkley River p=0.032 Kathlyn Lake R² = 0.324 0 2 4 6 8 10 12 14 0 10 20 30 40 50 TP (ug/L) Annual average discharge (m3/s) on Okanagan River Okanagan Lake p=0.001 51  correlations, or temporal coherence for both TN and TP were positive (TN r = 0.058; TP r = 0.284) and statistically different from zero for both TN (p<0.01) and TP (p<0.0001) (Table 3.3; Appendix 7A,B). TN coherence however, was very low at the full study scale, and improved only modestly at smaller spatial scales (Shuswap lakes r = 0.18; Fraser lakes r = 0.17).  Importantly, temporal coherence was consistently greater for TP than TN at all spatial scales examined. Low to moderate TP coherence occurred among all lakes at the full study scale (TP  r = 0.284), and among interior lakes grouped at various smaller spatial scales (all r = 0.314; Cariboo r = 0.397; Columbia r = 0.380; Shuswap r = 0.645) (Table 3.3). TP coherence was low however among Coastal lakes (r = 0.166). Interestingly, moderate TP coherence occurred between certain individual lakes from both interior and coastal areas, and all other lakes at the full study scale: Mara (r =0.501), Fork (r = 0.468), Kalamalka (r = 0.452), Shawnigan (r = 0.415), and Lac La Hache (r = 0.411). Among all sites, and at coastal and interior drainage (Columbia, Shuswap, Fraser) scales, strong correlations (after Bonferroni correction) were absent. However, at the drainage scale, strong correlations were more frequently observed for TP within the Shuswap (100%), and the Columbia lakes (38%). Percent strong correlations were absent for TN at all scales.  Interestingly, two lakes demonstrated negative TP coherence tendencies with many other lakes.  For example, TP correlations were negative in 55% of lake pairs involving Lizard Lake, and coherence was negative between Lizard Lake and other Coastal lakes (r = -0.194), and between Lizard Lake and all lakes (r = -0.03). Similarly, TP correlations were negative in 85% of lake pairs involving Charlie Lake. Temporal coherence was negative between Charlie Lake and other interior lakes (r = -0.166) and with all lakes at the full study scale (r = -0.09). Nonetheless, my expectation that nutrient coherence would be positive among all lakes, and at smaller drainage scales, was supported for both TN and TP.    52  Table 3.3  Average temporal coherence (r) for total phosphorus (TP) and total nitrogen (TN) between all lake pairs at provincial, coastal and interior locations, and at sub-group scales.      TP TN Area Drainage no. lake pairs r no. lake pairs r Coastal all 32 0.166 11 0.045             Interior all 124 0.314 93 0.039   Columbia 21 0.380 21 0.050   Shuswap 3 0.645 3 0.180   Fraser 11 0.416 5 0.170             Provincial all lakes 296 0.284 192 0.058         Dependence of nutrient coherence on lake pair proximity  I examined whether nutrient coherence could be explained by distance between lake pairs.  For all lakes, the relationship between lake pair TN coherence and lake pair distance was not significant (r2=0.0008, p=0.689) (Figure 3.4). At smaller scales however, a significant (negative) relationship was only evident between TN coherence and lake proximity (r2=0.463, p=0.09) within the Coastal area; in this case the relationship relied heavily on a relationship between TN coherence and distance for Fork and St. Mary lakes. A weak but significant (negative) relationship (r2=0.09, p<0.0001) was also evident between TP coherence and distance among all sites at the provincial scale (Figure 3.4). Removal of distant Charlie Lake, which exhibited negative correlations with most other lakes, further reduced the dependence of TP coherence on proximity of lake pairs (r2=0.01, p=0.1). Only for Coastal lakes was a significant relationship (negative) evident, with TP coherence diminishing with distance (r2=0.266, p=0.003) largely due to negative correlations between Lizard Lake and other Coastal lakes. For Fraser drainage lakes, a relationship (positive) (r2=0.187, p=0.2) was evident due to a high correlation between Horse Lake and distant Tabor Lake. No relationship was evident between TP coherence and distance for combined Columbia and Shuswap lake pairs (r2=0.001, p=0.8). I concluded that TN appears to have low coherence, even when lakes are in close proximity. Similarly, TP coherence  53  among lakes is not strongly associated with proximity, particularly when Charlie Lake is removed from the data set.    Figure 3.4  Relationship between distance (km) between lake pairs, and TP (left) and TN (right) correlations (r) of all lake pairs with more than 10 years concurrent data.   Dependence of nutrient coherence on lake water retention time  At the full scale, coherence of TN among lakes was uniformly low for lakes with WRT <1year (r=0.083, SD 0.315, n=20), between 1 and 5 years (r=0.116, SD 0.262, n=18) and those with WRT greater than 5 years (r=0.061, SD 0.302, n=14) (Appendix 8A). Similarly, TP coherence among lake pairs with WRT <0.75years (r=0.334, SD=0.262, n=15) was not different from lake pairs with WRT between 0.75 and 1.2 years (r=0.138, SD=0.288, n=15), those with WRT greater than 1.2 and 5 years (r=0.212, SD=0.392, n=13) or those with WRT greater than 5 years (r=0.319, SD=0.285, n=17) (Appendix 8B). Thus, for the lakes examined, coherence of TN nor TP, could be explained on the basis of lake water residence time.   Dependence of nutrient coherence on nutrient concentration  Among all lakes, temporal coherence of TN was uniformly low. However, to determine whether coherence might be greater among lakes with similar spring TN concentrations, I calculated coherence for lakes divided relatively evenly into 3 broad concentration ranges: <200 µg/L, 200-400 µg/L and >400 µg/L.  Coherence among lakes with <200 µg/L was greater than the 200-400 µg/L group, but not significantly so R² = 0.07 -0.600 -0.400 -0.200 0.000 0.200 0.400 0.600 0.800 1.000 0 200 400 600 800 1000 r TP km R² = 0.0008 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1 0 100 200 300 400 500 r TN km 54  (p=0.08), and not different than the >400 µg/L lake group (Figure 3.5, Table 3.4, Appendix 9a). Overall there was no significant difference (p=0.2) among the TN concentrations groups. (Appendix 9A).  Table 3.4  Average temporal coherence (r) for total phosphorus (TP) and total nitrogen (TN) between all lake pairs grouped by nutrient concentration (µg/L).      In contrast, TP coherence was significantly lower for the >20 µg/L lake group, than for <10µg/L and 10-20 µg/L lake groups (p<0.0001) (Table 3.4; Figure 3.5; Appendix 9B). To evaluate the influence of the anomalous lakes, I retested the TP lake groups after exclusion of Lizard Lake from the <10µg/L group, and Charlie Lake from the >25 µg/L group. Removal of these lakes increased coherence of each group (Table 3.4), however, coherence of the >25 µg/L group remained significantly lower (p<0.0001) than that of <10 µg/L and 10-20 µg/L lake groups (Appendix 9B). Thus, although the average distance between site pairs was significantly greater (p=0.0001) for the >20µg/L group due to Charlie Lake, TP coherence for this group, with Charlie Lake removed, was still significantly different (lower) (p <0.0001) from the lower concentration groups. I assessed whether TP groups were also significantly different in WRT. The three TP concentration groups were not significantly different in terms of WRT (p=0.63). I concluded that TN coherence shows no clear dependence on concentration. However for TP, coherence is significantly greater among lakes below 20 µg/L than among lakes above this value.    Variable Lake group # lake pairs average r SE TP <10 µg/L 20 0.380 0.071 Lizard removed 15 0.561 0.045 10-20 µg/L 26 0.536 0.042 >20 µg/L 42 0.113 0.047 Charlie removed 32 0.147 0.052 TN <200 µg/L 14 0.167 0.064 200-400 µg/L 28 -0.040 0.058 >400 µg/L 16 0.103 0.082 55    Figure 3.5  Coherence (r) and standard errors bars for all lake pairs, grouped by spring nutrient concentration ranges TP (<10 µg/L, 10- 20 µg/L and >25 µg/L) and TN (<200 µg/L, 200-400 µg/L, >400 µg/L).  DISCUSSION  I assessed to what extent interannual variation in catchment discharge was synchronous within and among lake districts, and helped explain TN and TP concentrations or coherence among representative BC lakes. Furthermore, I tested whether nutrient coherence could be explained by lake pair proximity, similarity in water residence time, or nutrient concentration.   The influence of hydrology  Coherent behavior of variables among lakes requires that an external driver impose a common dynamic to lakes across the spatial scale of interest (Magnuson et al., 2006). In this study I demonstrated that coherence of stream hydrographs, as estimated by average annual stream discharge, was strongly related to proximity. Coherence was high among sites separated by 10’s of kilometers but was weak among sites separated by distances of more than a few hundred kilometers. In addition to this direct relationship between discharge and scale, similar interannual periods of alternating high and low discharge were evident for coastal and southern interior stream hydrographs, but not Skeena and Peace hydrographs. Despite discharge coherence, strong relationships between lake nutrient concentration and local catchment discharge variation were limited. -0.2 -0.1 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 <10 ug/L 10-20 ug/L >20 ug/L r TP concentration group -0.2 -0.1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 <200 ug/L 200-400 ug/L >400 ug/L r TN concentration group 56  Catchment export of nitrogen and phosphorus are known to be linked to annual water yield (Jaworski et al., 1992; Schindler et al., 1996; Sorrano et al., 2000). However, lag effects between timing of input and complete mixing, or internal cycling of nutrients, may partially account for the lack of clear and direct links between run-off and lake concentrations. Lake concentrations may also be strongly influenced by magnitude and duration of extreme discharge events, antecedent catchment conditions (Vanni et al. 2001) or linked to discharge averaging periods other than the annual average discharge estimate employed here. For example, within lake districts in the Upper Great Lakes region, relationships between precipitation and lake nutrient concentrations were not consistent among districts, and varied in strength and direction (negative or positive) depending on averaging period (Webster et al., 2000).   Coherence of nutrients  Over decadal time scales, and among a diversity of lakes, I observed four important spatial patterns of TN and TP coherence among BC lakes. First, TN was consistently less coherent than TP. Direct comparison of these findings with work elsewhere is constrained by the limited reporting of TN and TP coherence to date. Nevertheless, coherence of TN among BC lakes at all spatial scales was weak, and generally less than reported elsewhere for nitrate nitrogen (Baron & Caine, 2000; George et al., 2000; Sorrano et al., 1999), TN (Kratz et al., 1998; Webster et al., 2000) or particulate nitrogen (Kling et al., 2000).   In contrast, the TP coherence reported here, particularly among the freshet dominated interior lakes, was moderate and similar to that reported for ELA, Red Lake and Dorset study areas (Webster et al., 2000), but less than for SRP among Arctic LTER lakes (Kling et al., 2000) or for TP among Southern Wisconsin or English Lake District lakes (George et al., 2000; Sorrano et al., 1999). Strength of coherence between lakes is expected to be greater for limnological variables directly affected by climatic factors, than for those influenced by a complexity of factors (Magnuson et al., 1990). Low coherence could result from some combination of little variation over time in the  57  variable of interest, weak linkages between variation in the driver and the variable of interest, or strong spatial patterns (Kling et al., 2000). That coherence for TP is consistently greater than TN among BC lakes suggests TN demonstrates greater intrinsic variation than TP, and climate more clearly affects TP across these landscapes. That coherence of TN is similar to or greater than TP in other settings such as ELA, Dorset and Red Lake study areas (Webster et al., 2000) suggests the influence of regional drivers such as atmospheric deposition or other regional drivers not influencing TN dynamics of BC lakes.  Second, coherence of neither TN nor TP were clearly dependent on BC lake pair proximity, even at the drainage scale. This is somewhat contradictory with previous findings which show coherence of reactive variables is greatest among proximate lake pairs within drainages (Baron & Caine, 2000; George et al., 2000; Kling et al., 2000). For example, among Arctic LTER lakes along a common drainage path of approximately 10 kilometers, nutrient coherence was greater among proximate pairs than distant pairs (Kling et al., 2000). Webster reports coherence for TN (r=0.426- 0.689) and TP (r=0.267-0.476) within ELA, Red Lake and Dorset lake districts over distances of 10’s of kilometers. Here I report that TP coherence was moderate and positive (r>0.4) among lake pairs over distances of a few hundred kilometers. Negative correlations among lake pairs containing either Lizard or Charlie lakes located at the boundaries of the study area, suggest spatial limits to coherence or the influence of drivers not explored in this study.  Third, while WRT is central to phosphorus mass balance models, at the large spatial scale examined here, no relationship could be established between nutrient coherence and water residence time other than a weak and inverse relationship among coastal lakes. This is inconsistent with the conceptual model put forward by Sorrano et al. (1999) showing coherence of dissolved reactive ions inversely related to WRT. Water residence time is strongly linked to nutrient dynamics in lakes (Reckhow & Chapra, 1983). However, lake TP is also known to be strongly linked to inflow concentration (Brett et al., 2008). Thus, within a series of proximate lakes, high coherence might  58  logically occur where riverine conditions and very short WRT (days-weeks) constrain the potential of within-lake dynamics. Although I was not able to explore this condition fully in this study, highest TP coherence was observed for the connected Shuswap lakes, having relatively short water residence times.  Finally, in this study, coherence of TP was strongly and inversely related to lake pair spring TP concentrations. TP coherence was significantly greater among lakes with < 20 µg/L TP, than among those having >20 µg/L TP at spring overturn. Importantly, coherence of TP was moderate among low TP lakes across distances of several hundred kilometers in southern BC including coastal, Columbia and Fraser drainage areas. Moreover, a number of low TP lakes from Coastal, Columbia and Fraser drainages were moderately coherent with all other lakes. These findings are unique, as the few previous studies to include TP, have either found no relationship between TP coherence and trophic state (Bleckner et al., 2007) or have reported greater coherence among lakes in disturbed catchments with higher nutrient loading and lake concentrations (George et al., 2000; Kratz et al., 1998).   In this study, all the BC lakes with > 20µg/L spring TP, have some combination of intrinsic drivers such as high levels of catchment disturbance (Jensen & Epp, 2001; Zirnhelt et al., 1997), internal loading (French & Petticrew, 2007; Rieberger, 2003) or poor stratification (McPherson, 2006). This study suggests that as lake TP increases through internal cycling or catchment specific loading, sensitivity to interannual variation of hydrology and external load diminishes, and coherence of TP among lakes decreases. Conversely, low to moderate TP lakes with little catchment specific or internal nutrient load, respond more directly to variability in catchment TP flux and demonstrate coherence across significant spatial scales in southern BC. I can not exclude the possibility that coherence may also be a function of other factors not examined here.  In summary, this examination of temporal coherence of a restricted variable group, across a diversity of lakes and spatial scales, has provided new insights into nutrient  59  coherence in general and that of nutrient dynamics among BC lakes in relation to extrinsic forces. Dependence of TP coherence on nutrient concentration potentially provides a new organizing concept to explore TP dynamics among lakes. Importantly, the temporal coherence of TP among low to moderate productivity BC lakes, should be considered when interpreting trends to local stressors and setting lake management objectives.     60   REFERENCES Anneville O., Gammeter S.A. & Straile D. (2005) Phosphorus decrease and climate variability: mediators of synchrony in phytoplankton changes among European peri- alpine lakes. Freshwater Biology, 50, 1731-1746. Baron J.S. & Caine N. (2000) Temporal coherence of two alpine lake basins of the Colorado Front Range, U.S.A. Freshwater Biology, 43, 463-476. Benson B.J., Lenters J.D., Magnuson J.J., Stubbs M., Kratz T.K., Dillon P.J., Hecky R.E. & Lathrop R.C. (2000) Regional coherence of climatic and lake thermal variables of four lake districts in the Upper Great Lakes Region of North America. Freshwater Biology, 43, 517-527. Bleckner T., Adrian R., Livinstone D.M., Jennings E., Weyhenmeyer G.A., George D.G., Jankowski T., Jarvinen M., Aonghusa C.N., Noges T., Straile D. & Teubner, K. (2007) Large-scale climatic signatures in lakes across Europe: a meta-analysis. Global Change Biology, 13, 1314-1326.  Boerger H. (2001) Prospect Lake water quality and lakeshed assessment. Habitat Acquisition Trust, Victoria, BC, 97p.  Boyd I.T., McKean C.J.P., Nordin R.N. & Wilkes B. D. (1985) Skeena, Nass area, Kathlyn, Seymour, Round and Tyhee lakes: Water quality assessment and objectives. Technical Appendix. Ministry of Environment, Victoria, BC,110p.  Brett M.T. & Benjamin M.T. (2008) A review and reassessment of lake phosphorus retention and the nutrient loading concept. Freshwater Biology, 53, 194-211.  Bryan J.E. & Jensen E.V. (1999) Water quality trends in Mara, Mabel and Sugar lakes 1971-1998. BC Ministry of Environment, Penticton, BC, 50p.   61  Carpenter S.R., Fisher S.G., Grimm N.B. & Kitchell J.F. (1992) Global change and freshwater ecosystems. Annual Review of Ecological Systems, 23, 119-139.  Cavanagh N., Nordin R. & Bryan J.E. (1994) Christina Lake Water quality assessment and objectives. Technical Appendix. Ministry of Environment Lands and Parks, BC. 106p.  French T.D. & Carmichael N. B. (1999) Limnological aspects of Charlie Lake (Peace River Drainage, British Columbia): A Summary of Data Collected between 1974 and 1995). BC Ministry of Environment, Lands and Parks. Prince George, 115p.   French T.D. & Petticrew E.L. (2007) Chlorophyll a seasonality in four shallow eutrophic lakes (northern British Columbia, Canada) and the critical roles of internal phosphorus loading and temperature. Hydrobiologia, 575, 285-299.  George D.G., Talling J.F. & Rigg E. ( 2000) Factors influencing the temporal coherence of five lakes in the English Lake District. Freshwater Biology, 43, 449- 461.  Holmes G.B. (1996a) State of water quality of St. Mary Lake 1975-1995. Ministry of Environment Lands and Parks, Victoria, BC.   Holmes G.B. (1996b) State of water quality of Quamichan Lake 1988-1995. Ministry of Environment Lands and Parks, Victoria, BC.  Jaworski N.A., Groffman P.M., Keller A.A. & Prager J.C. (1992) A watershed nitrogen and phosphorus balance: the upper Potomac River basin. Estuaries, 15, 83-95.  Jensen E.V. & Bryan J. E. (2001) Water quality trends in Kalamalka, Wood, and Ellison lakes 1969 to 1999. BC Ministry of Environment. 60p   62  Jensen E.V. & Epp P.F. (2001) Water quality trends in Okanagan, Skaha and Osoyoos lakes in response to nutrient reductions and hydrologic variation. BC Ministry of Environment, Penticton, BC. 17p. Kling G.W., Kipphut G.W., Miller M.M. & O’Briens W.J. (2000) Integration of lakes and streams in a landscape perspective: The importance of material processing on spatial patterns and temporal coherence. Freshwater Biology, 43, 477-497.  Kratz T. K., Sorrano P.A., Baines S.B., Benson B.J., Magnuson J.J., Frost T. M. & Lathrop R.C. (1998) Interannual synchronous dynamics in northern Wisconsin lakes in Wisconsin, USA. In: Management of Lakes and Reservoirs During Global Climate Change. (eds. D.G. George et al., ) p 273-287. Kluwer Academic Publishers, Netherlands.  Livingstone D.M. (2008) A change of climate provokes a change of paradigm: taking leave of two tacit assumptions about physical lake forcing. International Review Hydrobiologia, 93, 404-414.  Magnuson J.J., Benson B.J. & Kratz T. K. (1990) Temporal coherence in the limnology of a suite of lakes in Wisconsin, U.S.A. Freshwater Biology, 23, 145-159.  Magnuson J.J., Kratz T.K., Benson B.J. & Webster K.E. (2006) Coherent dynamics among lakes. Chapter 5 in: Long Term Dynamics of Lakes in the Landscape. Long Term Ecological Research on North Temperate Lakes. Eds. J.J. Magnuson, T.K. Kratz and K.E. Benson. Oxford University Press. p 89-106.  Magnuson J.J., Benson B.J. & Kratz T. K. (2004) Patterns of coherent dynamics within and between lake districts at local to intercontinental scales. Boreal Environment Research, 9, 359-369.    63  McKean C.J.P., Nagpal N.K. & Zirnhelt N.A. (1987) Williams Lake water quality assessment and objectives. BC Ministry of Environment., Victoria, BC,75p.   McPherson S. (2006) A water quality assessment of Quamichan Lake (Vancouver Island, British Columba): a summary of water quality data collected between 1951 and 2006. Ministry of Environment, Nanaimo, BC, 84p.  Nordin R. (2005) Water quality objectives for Okanagan Lake: A first update and   overview report. Ministry of Environment, Victoria BC. 135p. Pommen L.W. (1996) Ambient water quality objectives for Holland Creek and Stocking Lake watersheds. BC Ministry of Environment Lands and Parks. 50p.  Quinlan R., Paterson M., Hall R.I., Dillon P.J., Wilkinson A.N., Cumming B.F. & Douglas M.S. (2003). A landscape approach to examining spatial patterns of limnological variables and long-term environmental change in a southern Canadian lake district. Freshwater Biology, 48, 1676-1697.  Rechow K.H. & Chapra S.C. (1983) Engineering approaches for lake management. Volume 1. Data analysis and empirical modeling. Boston: Butterworth Publishers. 340p.  Rieberger, K. (2003) Highlands lakes water quality assessment,1997 – 2003. Ministry of Water, Land and Air Protection, 43p.  Schindler D.W. ( 2001) The cumulative effects of climate warming and other human stresses on Canadian freshwaters in the new millennium. Canadian Journal of Fisheries and Aquatic Science, 58, 18-29.  Schindler D.W., Bayley S.E., Parker B.R., Beaty K.G., Cruikshank D.R., Fee E.J., Schindler E.U. & Stainton M.P. (1996) The effects of climatic warming on the  64  properties of boreal lakes and streams at the Experimental Lakes Area, northwestern Ontario. Limnology and Oceanography, 41, 1004-1017.  Sorrano P.A., Webster K.E., Riera J.L., Kratz T.K., Baron J.S., Bukaveckas P.A., Kling G.W., Whiter D.S., Caine N., Lathrop R.C. & Leavitt P.R. (1999) Spatial variation among lakes within landscapes: ecological organization along lake chains. Ecosystems, 2, 395-410.  Vanni M.J., Renwick W.H., Headworth J.L., Auch J.D. and Schaus M.H. (2001) Dissolved and particulate nutrient flux from three adjacent agricultural watersheds: a five-year study. Biogeochemistry, 1023, 85-114.  Water Survey of Canada: on-line data source: http://www.wsc.ec.gc.ca/hydat/H2O/index_e.cfm?cname=main_e.cfm Webster K.E., Sorrano P.A., Baines S.B., Kratz T.K., Bowser C.J., Dillon P.J., Campbell P., Everett J.F. & Hecky R.E. (2000) Structuring features of lake districts: Landscape controls on lake chemical responses to drought. Freshwater Biology, 43, 499-515  Zhang J., Hudson J., Neal R., Sereda J., Clair T., Turner M., Jefferies D., Dillon P., Molot L., Somers K. & Hesslein R. (2010) Long-term patterns of dissolved organic carbon in lakes across eastern Canada: evidence of a pronounced climate effect. Limnology and Oceanography, 55, 30-42.  Zirnhelt N., Simpson J., Nordin R., Andrews K. & Pommen L.W. (1997) Horse Lake – Bridge Creek water quality assessment. BC Ministry of Environment Lands and Parks, Williams Lake, BC. 212 p       65  CHAPTER 4 SIGNIFICANCE OF FINDINGS AND NEXT STEPS   INTRODUCTION  Temporal coherence of parameters among lakes across significant spatial scales, offers a means of assessing the role of climate, on lake dynamics, and development and testing of organizing concepts to explain nutrient variation. Importantly, temporal coherence, where demonstrated has practical value, as it enables extrapolation of results from one lake to other lakes, and aids prediction of lake dynamics in the future (Magnuson et al., 2004). I assessed to what spatial extent interannual variation in hydrology was synchronous among 6 drainage areas and could explain TN and TP concentrations and coherence among 26 BC lakes over multi-decadal time frames.   My results showed that an important extrinsic driver, catchment discharge, was highly coherent among coastal, and southern interior drainages. Strong dependence of nutrient concentrations on antecedent average annual discharge was limited to a few interior lakes. However relationships between TP and TN and antecedent discharge tended to be positive and negative respectively. TN coherence was very low among lakes and could not be explained by proximity, or similarity in flushing or trophic state. Coherence of TP however, demonstrated low to moderate coherence over a wide area, and varied inversely and significantly with lake TP concentration. Oligotrophic and mesotrophic lakes (<20 µg/L) were significantly more coherent than eutrophic lakes (>20 µg/L). These finds have importance in terms of providing a broader understanding of coherence for nutrients, as well as application to lake management in BC. I will discuss these separately and then consider next steps.    66   SIGNIFICANCE OF FINDINGS   Developing a broader understanding of nutrient coherence   This work expands the spatial exploration of temporal coherence, as no other work to date has been reported for lakes in western Canada or the Pacific Northwest of North America. Previous coherence studies have examined long term data collections from Arctic, Colorado, and Wisconsin LTER lake networks, or Upper Great Lakes research areas such as ELA, Dorset, and Red Lake areas. Thus this study fills an important spatial gap in the studies of nutrient coherence among lakes.   This study also significantly adds to the understanding of coherence of TN and TP among lakes. Few other coherence studies to date have focused on these important variables. TN coherence reported here, is consistently less than reported elsewhere for nitrate nitrogen (Baron & Caine, 2000; George et al., 2000; Sorrano et al., 1999), TN (Kratz et al., 1998; Webster et al., 2000), or particulate nitrogen (Kling et al., 2000). The weak coherence of TN among BC lakes, suggests weak linkages between climate and TN variability either through intrinsic variation, or strong spatial patterns, which are perhaps less apparent in other areas due to regional stressors such as atmospheric deposition of nitrogen (Baron & Caine, 2000) or fertilizer application to catchments (George et al., 2000).  My findings do not support the hypothesis that coherence of nutrients among these study lakes is linked to WRT. Only a weak and negative relationship between TP coherence and water residence time could be found among coastal lakes. Thus the conceptual model put forward by Sorrano et al. (1999) showing coherence of dissolved reactive ions inversely related to WRT may not be applicable to TP or to the lakes examined here. My findings are also unique in that they show TP coherence among BC lakes was strongly and inversely related to lake pair spring TP concentrations. Previous studies have either found no dependence between nutrient coherence and trophic state (Bleckner et al., 2007) or have reported greater coherence among lakes in disturbed  67  catchments with higher nutrient loading and lake concentrations (George et al., 2000; Kratz et al., 1998). Results of my study suggest that as lake TP increases through internal cycling or catchment specific loading, sensitivity to interannual variation of external load diminishes and the potential for coherence of phosphorus is reduced among high TP lakes. Conversely, low to moderate TP lakes (<20 µg/L) with limited internal nutrient cycling, and lower external loading, respond more directly to significant variation in climate driven external load and demonstrate coherence across significant spatial scales.   Application of findings to BC lake management These findings can be readily incorporated into two key aspects of the BC lake monitoring and assessment network. First, because decadal scale time series incorporate more climate variability, and have been shown to provide higher estimates of coherence (see Chapter 2) I suggest that trend assessments using less than 10 years of data may be compromised due to climate variation, particularly among lakes with < 20 µg/L TP. And second, as a significant portion of the variability of spring TP among low TP lakes is shown here to be linked to climate variation, this variable and these lakes can now provide important trend assessment potential, and should be central to a provincial lake monitoring network. In conceptual terms, individual lakes which shift toward becoming more coherent with reference lakes would suggest decreasing intrinsic control. For example, success of external phosphorus load reductions to Okanagan lakes (Jensen & Epp, 2001), or internal load reductions through hypolimnetic aeration of lakes such as St. Mary Lake (Rieberger,1992) could be evaluated in part by comparing lake specific trends against the regional norm. Conversely, individual lakes which become less coherent with a population of interest, would suggest increasing intrinsic or catchment specific control over TP variability.   Finally, given that spring TP is a common water quality objective established as a BC lake and nutrient management target, temporal coherence of this variable should be considered during the objectives setting and attainment monitoring processes.     68   NEXT STEPS   Developing a broader understanding of nutrient coherence  Clearly the relationship between TP coherence and lake status requires further exploration to determine whether this relationship contributes to a broader conceptual understanding of factors governing coherence. Phosphorus is not well represented in coherence studies to date. Therefore, I suggest that combining the BC data with that of other lake drainages previously reported, and testing a hypothesis that TP coherence is dependent on lake status would be a logical next step in developing an organizing concept for TP coherence among lakes.   As a preliminary step to further this ambition, I assembled readily available TP coherence and phosphorus concentration estimates for all forms of phosphorus and lake groupings reported to date (Table 4.1).    69   Table 4.1  Average coherence and concentrations for TP, TDP and SRP among various lake groupings.    As various Wisconsin lakes groupings are reported in the literature, multiple coherence and concentration estimates are possible. Adirondack, Colorado and the English Lake district were not included, as phosphorus concentrations were not readily available or phosphorus was not included in the coherence study. Nevertheless, using all data in Table 4.1 the relationship between lake status and coherence is insignificant (r2=0.003; p=0.8). However, removal of the single data value (r=0.674) for the South Wisconsin lakes, due to the strong anthropogenic influence on nutrients, yields a significant relationship (r2=0.253; p=0.095) for the remainder of the data (Figure 4.1).   Lake grouping P form µg/L r Northern Wisconsin LTER TP 22 0.139 Northern Wisconsin LTER a TDP 22 0.39 Southern Wisconsin TP 63 0.674* Northern and Southern Wisconsin b TP 47 0.15 Northern and Southern Wisconsin b TDP 47 0.22 Arctic LTER c SRP 3.8 0.7 Arctic LTER c PP 3.8 0.18 ELA d TP 8.3 0.267 Dorset d TP 6 0.385 Red Lake d TP 9.5 0.467 BC e TP 10 0.561 BC e TP 15 0.536 BC e TP 20 0.113 all data from Sorrano  et al ., (1999) or as indicated; a-Magnuson  et al ., (1990); b- Kratz  et al ., (1998); c-Kling  et al ., (2000); Webster  et al ., (2000); e-Jensen (see  Chapter 3); * anomalous value excluded from Figure 4.1  70   Figure 4.1  Dependence of coherence (r) on phosphorus concentration from multiple coherence studies and lake districts.  These results are intriguing and warrant further evaluation by considering lake pair TP concentration and coherence values over concurrent time series and at various spatial groupings.   Application of findings to BC lake management  My findings indicate that a measurable portion of the interannual TP variation in oligotrophic and mesotrophic lakes in this study is driven by climate variation over significant spatial scales. However, the boundaries of this relationship are poorly defined. Lizard and Charlie lakes may be responding to spatial boundaries of climate expression or to other intrinsic drivers. Therefore, I recommend incorporating new or existing data for low nutrient lakes from central and northern areas of BC to further define the spatial scale of TP coherence. I also recommend continued spring TP data collection from low TP lakes to verify and track coherence among these lakes.   I suggest it is important to incorporate these findings when interpreting lake specific TP trends in relation to landuse, nutrient management or other lake management considerations. Lake managers are interested in the underlying water quality signals and trends, as opposed to year to year variability or noise. Therefore, regionalization or R² = 0.2553, p=0.095 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0 10 20 30 40 50 60 70 r concentration in ug/L 71  the extrapolation of results from a few intensively studied sites to other lakes is a logical outcome where predictable coherence occurs. Certainly for BC lakes with < 20 µg/L TP, comparison of TP trends relative to that of neighbouring lakes of similar status should provide a useful reference for trend assessment purposes. To facilitate this and to provide a more spatially robust reference signal indicative of TP and climate interaction, I suggest a single standarized TP coherence value or index, joining the TP dynamics common to a number of key lakes, over a longer averaging period such as a 2-3 year moving averaging, could prove beneficial to regionalization of results from a few lakes to other lakes within the spatial scale of interest. As with other indices, the TP index would integrate spring measurements over appropriate spatial and temporal scales, to demonstrate a common signal. Further evaluation of this concept is necessary to establish and validate optimal spatial and temporal aspects of the index.  At the regional and provincial scale it appears that TP coherence among low TP lakes is governed by extrinsic controls related to climate. What is not entirely clear from my study is how climate and catchments interact to enable coherence. I recommend other discharge and nutrient averaging periods be explored to clarify the relationships between TP coherence and hydrology. Moving averages of more than one antecedent year might better reflect underlying lake response to climate variation.  As well, my study did not consider whether variation in regional drivers such as temperature could be contributing to coherence. Air and surface water temperature are highly coherent over large distances and are strong drivers of ecosystems over large spatial scales (Livingstone, 2008). Systematic or episodic variation in atmospheric temperature alone, or combined with precipitation, could contribute to synchronized patterns of snow accumulation, timing of snow melt, or nutrient uptake in terrestrial (George et al., 2000) or aquatic ecosystems (Anneville et al., 2005; Schindler, 2001). Climate indices such as NAO and ENSO reduce complex spatial and temporal variability into measures depicting underlying climate patterns (Stenseth et al., 2003), and in the case of NAO, is useful in understanding TP variability among European lakes (Anneville et al., 2005). Considerable information exists for ENSO, and its index is known to reflect timing of snow melt and volume of catchment discharge (Cayan et al., 1999; Moore &  72  McKendry, 1996) and material transport (Kiffney et al., 2002) in drainages of the Pacific Northwest. Therefore, I suggest examining whether TP coherence among BC lakes is dependent on ENSO might prove beneficial to predicting TP variability in the future.       73   REFERENCES Anneville O., Gammeter S.A. & Straile D. (2005) Phosphorus decrease and climate variability: mediators of synchrony in phytoplankton changes among European peri- alpine lakes. Freshwater Biology, 50, 1731-1746. Baron J.S. & Caine N. (2000) Temporal coherence of two alpine lake basins of the Colorado Front Range, U.S.A. Freshwater Biology, 43, 463-476.  Bleckner T., Adrian R., Livinstone D.M., Jennings E., Weyhenmeyer G.A., George D.G., Jankowski T., Jarvinen M., Aonghusa C.N., Noges T., Straile D. & Teubner, K. (2007) Large-scale climatic signatures in lakes across Europe: a meta-analysis. Global Change Biology, 13, 1314-1326.  Cayan D.R., Redmond K.T. & Riddle L.G. (1999) ENSO and hydrological extremes in the western United States. Journal of Climate, 12 , 2881-2893.  George D.G., Talling J.F. & Rigg E. (2000) Factors influencing the temporal coherence of five lakes in the English Lake District. Freshwater Biology, 43, 449-461.  Jensen E.V. & Epp P.F. (2001) Water quality trends in Okanagan, Skaha and Osoyoos lakes in response to nutrient reductions and hydrologic variation. BC Ministry of Environment, Penticton, BC. 17p.  Kiffney P.M., Bull J.P. & Feller, M.C. (2002) Climatic and hydrologic variability in a coastal watershed of southwestern British Columbia. Journal of the American Water Resources, 38, 1437-1451. Kling G.W., Kipphut G.W., Miller M.M. & O’Briens W.J. (2000) Integration of lakes and streams in a landscape perspective: The importance of material processing on spatial patterns and temporal coherence. Freshwater Biology, 43, 477-497.  74   Kratz T. K., Sorrano P.A., Baines S.B., Benson B.J., Magnuson J.J., Frost T. M. & Lathrop R.C. (1998) Interannual synchronous dynamics in northern Wisconsin lakes in Wisconsin, USA. In Managament of Lakes and Reservoirs During Global Climate Change. (eds. D.G. George et al.) pp 273-287. Kluwer Academic Publishers, Netherlands.  Livingstone D.M. (2008) A change of climate provokes a change of paradigm: taking leave of two tacit assumptions about physical lake forcing. International Review Hydrobiologia, 93, 404-414.  Magnuson J.J., Benson B.J. & Kratz T. K. (1990) Temporal coherence in the limnology of a suite of lakes in Wisconsin, U.S.A. Freshwater Biology, 23, 145-159. Magnuson J.J., Benson B.J. & Kratz T. K. (2004) Patterns of coherent dynamics within and between lake districts at local to intercontinental scales. Boreal Environment Research, 9, 359-369.  Moore R.D. & McKendry I.G. (1996). Spring snowpack anomaly patterns and winter climatic variability, British Columbia, Canada. Water Resources Research, 32, 623- 632.  Rieberger, K. (1992) Effects of hypolimnetic aeration on fisheries habitat of St. Mary Lake. Ministry of Environment. pp 62  Sorrano P.A., Webster K.E., Riera J.L., Kratz T.K., Baron J.S., Bukaveckas P.A., Kling G.W., Whiter D.S., Caine N., Lathrop R.C. & Leavitt P.R. (1999) Spatial variation among lakes within landscapes: ecological organization along lake chains. Ecosystems, 2, 395-410.   Stenseth N.C., Otterson G., Hurrell, J.W., Mysterud A., Lima M., Chan K., Yoccoz N.G. & Adlandsvik B. (2003). Studying climate indices: the North Atlantic  75  Oscillation, El Nino Southern Oscillation and beyond. Proceedings, Royal Society of London, 270, 2087-2096.  Webster K.E., Sorrano P.A., Baines S.B., Kratz T.K., Bowser C.J., Dillon P.J., Campbell P., Everett J.F. & Hecky R.E. (2000) Structuring features of lake districts: Landscape controls on lake chemical responses to drought. Freshwater Biology, 43, 499-515   76   APPENDICES  APPENDIX 1A  Variation in coherence of a hypothetical data set following substitution of non-detect values with the minimum detectable concentration (mdc), substitution of half the mdc, or pairwise deletion.       % data < mdc high r mdc high r mdc/2 high r delete mdc med r mdc med r mdc/2 med r delete mdc low r mdc low r mdc/2 low r delete mdc 0 0.843 0.843 0.843 0.580 0.580 0.580 0.237 0.237 0.237 20 0.843 0.858 0.758 0.580 0.584 0.514 0.237 0.253 0.136 30 0.798 0.827 0.611 0.569 0.580 0.386 0.205 0.227 -0.023 35 0.756 0.767 0.541 0.550 0.556 0.411 0.176 0.195 -0.040 50 0.753 0.756 0.929 0.552 0.550 0.373 0.149 0.176 -0.225 77   APPENDIX 1B  Variation in coherence for spring total dissolved phosphorus (TDP) in Christina and Sugar lakes following substitution of non-detect values with the minimum detectable concentration (mdc), substitution of half the mdc, or pairwise deletion (underlined values are below the mdc value of 3 µg/L).  TDP in µg/L  Correlation matrix substitute mdc  Christina  Sugar    Christina  Sugar      Christina  1    3  Sugar 0.337269 1           5     6 5  substitute mdc/2         Christina  Sugar 5 4  Christina  1  5 6  Sugar 0.404238 1   4     3 3       3  pair-wise deletion of < mdc 4 3    Christina  Sugar   3  Christina  1  3 3  Sugar 0.327166 1 3 3     3 3     3 3     4 5     3 5     17 3     14 7     9 2     11 4     13 6     7 2     4 2     6 3     3 2     2 3     4 4     6 4        78  Lakes 1 Christina 2 Ellison 3 Kalamalka 4 Okanagan 5 Osoyoos 6 Skaha 7 Wood 8 Mabel 9 Mara 10 Sugar 11 Fork 12 Glen 13 Lizard 14 Maxwell 15 Prospect 16 Quamichan 17 Shawnigan 18 St Mary 19 Stocking 20 Chimney 21 Horse 22 Lac La Hache 23 Tabor 24 Williams 25 Charlie 26 Kathlyn Hydrographs A Alces B Beatton C Bulkley D Chilcotin E Cowichan F Criss G Goathorn H Mission I Okanagan J San Jose K Shawnigan L South Thompson APPENDIX 2  Map of hydrometric and lake sites.     i ii iii iv  v  79   APPENDIX 3A  Spring total nitrogen concentrations (µg/L) for selected coastal and interior lakes of BC    Lake district Lake 19 77 19 78 19 79 19 80 19 81 19 82 19 83 19 84 19 85 19 86 19 87 19 88 19 89 19 90 19 91 19 92 19 93 19 94 19 95 19 96 19 97 19 98 19 99 20 00 20 01 20 02 20 03 20 04 20 05 20 06 20 07 C ou nt M ax im um M in im um M ea n S td  D ev . Coastal Shawnigan 340 257 260 227 223 290 295 170 230 203 163 223 240 300 210 225 230 235 260 263 300 250 313 233 277 200 26 340 26 230 72 St Mary 413 750 670 503 385 370 266 385 267 285 687 460 495 290 400 410 365 340 355 470 470 683 483 23 750 23 423 177 Prospect 497 267 383 400 365 353 313 485 560 430 405 380 325 370 390 285 405 420 420 363 20 560 20 366 129 Quamichan 385 365 410 345 485 390 405 485 495 730 10 730 10 403 206 Lizard 30 150 80 40 80 90 70 120 110 60 133 11 150 11 81 46 Stocking 80 170 110 80 350 90 120 130 8 350 8 136 111 Maxwell 190 260 300 320 250 200 360 230 240 190 330 325 280 300 250 270 260 260 290 230 20 360 20 249 85 Glen 293 267 400 323 233 290 440 320 380 360 280 330 12 440 12 292 124 Fork 267 260 215 240 290 300 323 290 220 200 10 323 10 227 100 Interior Ellison 510 640 390 460 680 530 370 450 390 440 560 370 320 330 320 600 370 360 340 340 390 360 440 530 500 500 540 400 28 680 28 422 145 Osoyoos 125 115 115 440 280 357 355 400 365 365 358 340 335 365 320 460 385 360 355 178 230 265 325 290 305 305 305 260 295 310 350 31 460 31 315 95 Skaha 217 275 310 343 280 380 340 290 303 315 340 320 320 345 285 330 240 265 155 190 215 225 230 250 215 225 210 230 215 220 225 31 380 31 256 82 Okanagan 295 235 238 257 200 209 243 215 198 183 207 173 160 180 182 150 207 205 242 190 208 235 220 218 218 212 222 228 217 249 242 31 257 31 200 51 Kalamalka 282 260 215 253 289 248 295 272 284 265 274 283 248 273 222 263 232 268 228 357 322 345 353 377 357 377 357 327 333 388 347 31 388 31 287 85 Wood 840 560 660 528 581 572 583 508 565 452 454 440 440 320 460 310 380 340 460 450 450 460 450 440 430 510 450 380 480 29 840 29 448 149 Christina 160 55 140 120 90 135 110 115 190 120 110 110 90 90 85 130 75 110 80 95 85 90 125 120 24 190 24 104 38 Sugar 193 200 205 207 175 180 210 165 135 145 185 120 315 245 240 230 250 165 185 155 165 155 155 140 170 190 195 165 28 315 28 181 62 Mabel 215 150 195 125 130 195 185 150 135 150 100 120 155 140 125 140 160 150 140 145 175 140 155 120 165 205 140 180 170 130 30 215 30 145 40 Mara 155 230 180 235 195 245 280 200 185 200 180 170 130 190 205 150 195 245 160 245 260 220 190 185 150 185 170 170 195 170 220 31 280 31 189 55 Lac La Hache 650 610 490 410 470 463 515 490 430 404 450 556 468 550 540 537 491 508 557 19 650 19 451 158 Chimney 770 910 810 780 770 780 627 1072 1100 1275 1022 1003 833 13 1275 13 819 376 Horse 370 350 370 350 458 430 398 390 403 456 398 378 359 366 377 321 16 458 16 350 124 Williams 606 850 659 631 1010 755 733 892 674 651 619 858 880 815 783 673 894 826 907 19 1010 19 717 250 Charlie 703 713 760 1007 840 5 1007 5 630 377 Tabor 490 395 600 480 500 413 483 490 8 600 8 406 197 Kathlyn 410 505 395 390 240 363 320 7 505 7 314 171 80   APPENDIX 3B  Spring total phosphorus concentrations (µg/L) for selected coastal and interior lakes of BC       Area Lake 19 77 19 78 19 79 19 80 19 81 19 82 19 83 19 84 19 85 19 86 19 87 19 88 19 89 19 90 19 91 19 92 19 93 19 94 19 95 19 96 19 97 19 98 19 99 20 00 20 01 20 02 20 03 20 04 20 05 20 06 20 07 C ou nt M ax im um M in im um M ea n S td  D ev . Coastal Shawnigan 6 4 9 9 7 7 8 6 5 5 4 4 4 4 4 3 3 6 7 6 7 9 5 4 2 2 3 6 28 28 2 6.7 6.3 St Mary 14 48 41 33 25 18 16 20 33 28 17 17 22 14 18 26 12 9 30 14 17 18 16 14 12 16 26 46 33 29 48 9 23.4 10.2 Prospect 19 20 12 14 16 12 9 12 11 8 4 12 7 9 24 28 17 13 16 17 7 12 10 16 18 25 28 4 14.0 6.5 Quamichan 20 24 7 13 12 27 6 21 20 21 12 39 16 19 54 15 54 6 21.3 13.9 Stocking 7 7 8 5 7 5 6 3 5 3 11 3.5 5 5.5 5 15 15 3 6.6 3.5 Fork 9 11 8 15 8 14 13 10 8 8 7 7 8 13 15 7 10.1 2.9 Glen 25 30 22 28 31 15 15 12 20 18 32 12 6 12 48 20 24 30 31 10 19 12 19 12 9 25 48 6 21.1 10.8 Maxwell 9 12 12 11 7 9 12 9 9 8 13 5 10 10 12 14 7 7 8 5 8 10 22 22 5 10.2 4.2 Lizard 7 5 5 3 4 5 4 5 4 5 8 5 4 5 3 12 3 2 18 18 2 6.0 4.4 Interior Ellison 35 57 47 52 40 41 20 20 27 38 36 38 43 37 49 11 14 35 37 29 45 34 42 36 27 38 23 30 31 29 57 11 34.6 11.3 Osoyoos 25 22 20 22 17 39 26 33 26 25 24 38 31 24 24 24 23 18 10 14 16 13 17 16 16 14 19 11 6 11 17 31 39 6 20.8 9.1 Skaha 24 30 25 26 18 16 24 27 20 16 15 12 12 11 5 9 10 6 10 10 12 3 9 10 12 3 9 6 3 8 7 31 31 3 12.6 7.7 Okanagan 8 8 11 10 10 12 11 10 9 7 6 7 6 5 6 7 4 4 7 7 9 11 8 11 7 5 5 4 5 3 5 31 31 3 8.6 6.5 Wood 70 93 54 82 72 82 76 79 53 48 40 53 44 38 28 35 16 38 34 27 46 29 34 42 54 45 50 41 38 34 66 31 93 16 47.2 18.8 Kalamalka 5 7 12 10 10 12 10 12 11 11 8 7 7 6 7 6 6 7 4 22 13 6 8 10 12 8 6 5 3 4 7 31 31 3 9.8 6.8 Christina 6 8 10 11 8 8 3 4 4 4 5 3 16 14 11 13 16 9 5 6 4 2 4 9 24 24 2 8.6 6.0 Sugar 4 9 7 5 8 5 3 3 4 3 3 3 3 5 5 4 4 11 4 6 7 3 2 3 6 2 3 4 28 28 2 5.9 6.3 Mabel 6 7 6 5 7 5 9 8 5 3 3 3 3 4 3 9 4 3 10 8 5 5 8 5 3 3 4 2 2 5 30 30 2 6.4 6.7 Mara 7 10 10 12 12 15 16 13 12 10 6 6 8 6 9 4 9 8 5 25 22 12 11 11 9 7 5 6 5 2 5 31 31 2 10.6 7.5 Williams 78 76 68 50 68 86 85 71 83 75 56 53 68 72 51 48 54 83 87 87 44 54 70 71 56 60 65 77 28 87 28 65.0 16.7 Chimney 12 17 17 20 21 22 24 35 30 23 19 29 18 8 7 7 16 35 7 19.6 8.9 Tabor 27 30 16 23 27 18 22 30 26 22 67 32 28 38 35 27 27 20 21 22 9 22 22 67 9 27.5 13.4 Lac La Hache 14 15 13 19 16 16 22 14 16 13 13 11 12 9 11 14 9 13 18 17 2 8 4 9 24 24 2 13.2 5.9 Horse 15 15 16 19 14 17 35 32 14 24 17 21 22 17 20 15 35 14 20.3 6.8 Kathyln 28 25 25 19 40 17 20 23 20 22 10 40 10 22.8 8.7 Charlie 33 35 20 47 38 74 38 48 50 38 38 59 61 44 14 74 14 42.5 17.3 81  APPENDIX 4  One way analysis of long term spring nutrient concentration by area  Oneway Analysis of Long term mean spring TN By CNI  CNI- central and northern interior; SI-southern interior   Analysis of Variance Source DF Sum of Squares Mean Square F Ratio Prob > F CNI 2 0.6264266 0.313213 13.8414 0.0001 Error 24 0.5430881 0.022629   C. Total 26 1.1695147       Oneway Analysis of Long term mean spring TP By CNI  CNI- central and northern interior; SI-southern interior  Analysis of Variance Source DF Sum of Squares Mean Square F Ratio Prob > F CNI 2 1326.6778 663.339 3.3227 0.0532 Error 24 4791.3222 199.638   C. Total 26 6118.0000      0 0.2 0.4 0.6 0.8 1 Lo ng  t er m m ea n sp rin g T N CNI Coastal SI CNI 0 10 20 30 40 50 60 70 Lo ng  t er m m ea n sp rin g T P CNI Coastal SI CNI 82  APPENDIX 5A  Correlation (r) between lake annual spring TN concentration (µg /L) and antecedent (yr-1) annual mean discharge (m3/s) at proximate river sites.    Area Drainage River Lake r     Coastal Coastal Cowichan Shawnigan -0.275 St Mary 0.046 Prospect -0.080 Quamichan 0.450 Lizard -0.547 Stocking -0.292 Maxwell -0.030 Glen 0.219 Fork -0.559 Interior Columbia Okanagan Ellison -0.198 Osoyoos -0.051 Skaha -0.098 Okanagan 0.201 Kalamalka 0.215 Wood -0.082 Christina -0.009 Shuswap South Thompson Sugar -0.185 Mabel 0.070 Mara 0.079 Fraser Chilcoten Lac La Hache -0.227 Chimney 0.541 Horse 0.445 Williams -0.028 Tabor -0.127 Skeena Bulkley Kathlyn -0.851 Peace Beatton Charlie 0.989 83  APPENDIX 5B  Correlation (r) between lake annual spring TP concentration (µg /L) and antecedent (yr-1) annual mean discharge (m3/s) at proximate river sites.      Area Drainage River Lake r     Coastal Coastal Cowichan Shawnigan 0.111 St Mary 0.060 Prospect 0.308 Quamichan 0.110 Stocking -0.081 Fork 0.020 Glen 0.443 Maxwell -0.049 Lizard 0.020 Interior Columbia Okanagan Ellison -0.083 Osoyoos -0.003 Skaha 0.021 Okanagan 0.569 Wood -0.034 Kalamalka 0.206 Christina 0.557 Shuswap South Thompson Mara 0.228 Sugar 0.109 Mabel  0.215 Fraser Chilcoten Williams -0.280 Chimney 0.390 Lac la Hache -0.165 Horse 0.329 Tabor -0.004 Skeena Bulkley Kathyln -0.024 Peace Beatton Charlie 0.245 84   APPENDIX 6A  Pearson product moment correlations (r) for spring total nitrogen among BC lakes.      Pearson r Years p Pearson r Years p Chimney Lac La Hache 0.205 10 0.570 Kalamalka Osoyoos -0.171 30 0.365 Chimney Christina -0.243 11 0.472 Kalamalka Skaha -0.495 30 0.005 Chimney Ellison -0.696 12 0.012 Kalamalka Okanagan 0.182 30 0.336 Chimney Kalamalka 0.682 12 0.015 Lac La Hache Christina 0.099 14 0.737 Chimney Sugar -0.332 12 0.291 Lac La Hache Sugar 0.066 17 0.800 Chimney Osoyoos -0.307 13 0.307 Lac La Hache Ellison 0.171 18 0.497 Chimney Skaha -0.622 13 0.023 Lac La Hache Kalamalka 0.276 18 0.268 Chimney Okanagan 0.079 13 0.797 Lac La Hache Wood 0.341 18 0.166 Chimney Skaha -0.272 13 0.369 Lac La Hache Osoyoos -0.630 19 0.004 Chimney Mabel -0.038 13 0.902 Lac La Hache Skaha -0.359 19 0.131 Chimney Mara 0.097 13 0.751 Lac La Hache Okanagan 0.409 19 0.082 Christina Ellison 0.035 22 0.877 Lac La Hache Mabel 0.595 19 0.007 Christina Wood 0.219 23 0.314 Lac La Hache Mara 0.317 19 0.186 Christina Osoyoos 0.314 24 0.135 Lizard Wood 0.294 10 0.410 Christina Skaha 0.403 24 0.051 Lizard Shaw nigan 0.601 10 0.066 Christina Okanagan -0.159 24 0.457 Lizard Ellison 0.306 11 0.359 Christina Kalamalka -0.382 24 0.065 Lizard Osoyoos 0.152 11 0.655 Fork Osoyoos -0.417 10 0.230 Lizard Skaha 0.004 11 0.991 Fork Skaha -0.411 10 0.238 Lizard Okanagan -0.106 11 0.757 Fork Okanagan -0.529 10 0.116 Lizard Kalamalka 0.258 11 0.444 Fork Kalamalka 0.236 10 0.512 Lizard Sugar 0.490 11 0.126 Fork Wood 0.263 10 0.462 Lizard Mabel 0.319 11 0.339 Fork Christina -0.594 10 0.070 Lizard Mara 0.269 11 0.424 Fork Sugar -0.494 10 0.147 Mabel Christina 0.055 24 0.799 Fork Mabel 0.387 10 0.269 Mabel Ellison 0.533 28 0.004 Fork Mara -0.308 10 0.387 Mabel Sugar 0.010 28 0.960 Fork St Mary -0.492 10 0.149 Mabel Kalamalka 0.190 29 0.322 Fork Prospect 0.008 10 0.982 Mabel Wood 0.490 29 0.007 Glen Christina 0.786 10 0.007 Mabel Osoyoos -0.157 30 0.406 Glen Wood -0.297 11 0.374 Mabel Skaha -0.091 30 0.632 Glen Ellison -0.405 12 0.191 Mabel Okanagan 0.588 30 0.001 Glen Osoyoos 0.552 12 0.063 Mara Christina -0.060 24 0.779 Glen Skaha -0.012 12 0.970 Mara Ellison 0.210 28 0.283 Glen Okanagan 0.046 12 0.887 Mara Sugar 0.160 28 0.416 Glen Kalamalka -0.166 12 0.605 Mara Wood 0.344 29 0.067 Glen Sugar 0.593 12 0.042 Mara Kalamalka -0.104 30 0.584 Glen Mabel -0.074 12 0.818 Mara Mabel 0.371 30 0.043 Glen Mara 0.061 12 0.851 Mara Osoyoos -0.047 31 0.804 Glen Shaw nigan 0.173 12 0.590 Mara Skaha 0.146 31 0.434 Horse Chimney 0.141 10 0.698 Mara Okanagan 0.157 31 0.399 Horse Lac La Hache 0.335 11 0.314 Maxw ell Lac La Hache 0.825 10 0.003 Horse Ellison -0.051 15 0.857 Maxw ell Horse 0.604 10 0.065 Horse Kalamalka -0.088 15 0.756 Maxw ell Prospect -0.269 12 0.398 Horse Christina -0.127 15 0.651 Maxw ell Williams -0.109 15 0.699 Horse Sugar 0.394 15 0.147 Maxw ell St Mary -0.289 15 0.297 Horse Osoyoos 0.006 16 0.984 Maxw ell Christina -0.240 17 0.353 Horse Skaha -0.234 16 0.383 Maxw ell Shaw nigan 0.239 18 0.340 Horse Okanagan -0.205 16 0.447 Maxw ell Ellison 0.220 19 0.365 Horse Wood -0.606 16 0.013 Maxw ell Wood 0.021 19 0.933 Horse Mabel -0.026 16 0.925 Maxw ell Osoyoos -0.267 20 0.254 Horse Mara 0.031 16 0.910 Maxw ell Skaha -0.240 20 0.309 Kalamalka Ellison 0.039 27 0.848 Maxw ell Okanagan 0.204 20 0.388 Lake pair Lake pair 85   Appendix 6A continued. Pearson product moment correlations (r) for spring total nitrogen among BC lakes.     Pearson r Years p Pearson r Years p Maxw ell Kalamalka 0.227 20 0.335 St Mary Christina -0.251 19 0.299 Maxw ell Sugar -0.285 20 0.224 St Mary Sugar -0.010 21 0.967 Maxw ell Mabel 0.184 20 0.438 St Mary Shaw nigan 0.191 21 0.407 Maxw ell Mara -0.107 20 0.654 St Mary Ellison 0.460 22 0.031 Okanagan Ellison 0.378 28 0.047 St Mary Kalamalka -0.135 22 0.548 Okanagan Osoyoos -0.400 31 0.026 St Mary Wood 0.128 22 0.570 Okanagan Skaha -0.377 31 0.036 St Mary Osoyoos -0.007 23 0.974 Osoyoos Ellison 0.105 28 0.594 St Mary Skaha 0.045 23 0.838 Prospect Horse 0.067 11 0.845 St Mary Okanagan 0.409 23 0.053 Prospect Williams -0.103 11 0.762 St Mary Mabel 0.125 23 0.569 Prospect Ellison -0.086 14 0.771 St Mary Mara 0.291 23 0.177 Prospect Wood 0.164 14 0.576 Sugar Christina 0.335 24 0.110 Prospect Christina 0.265 14 0.360 Sugar Ellison -0.064 26 0.757 Prospect Shaw nigan -0.375 14 0.187 Sugar Wood 0.003 27 0.987 Prospect Osoyoos 0.377 15 0.166 Sugar Osoyoos 0.228 28 0.244 Prospect Skaha 0.190 15 0.498 Sugar Skaha 0.026 28 0.894 Prospect Okanagan -0.208 15 0.456 Sugar Okanagan -0.113 28 0.567 Prospect Kalamalka -0.493 15 0.062 Sugar Kalamalka -0.326 28 0.091 Prospect Sugar 0.468 15 0.079 Williams Lac La Hache 0.170 11 0.617 Prospect Mabel -0.032 15 0.910 Williams Horse -0.002 11 0.996 Prospect Mara -0.287 15 0.299 Williams Christina 0.016 17 0.953 Prospect St Mary 0.399 15 0.141 Williams Ellison 0.238 18 0.341 Quamichan Osoyoos 0.103 10 0.776 Williams Wood -0.292 18 0.240 Quamichan Skaha -0.145 10 0.689 Williams Osoyoos 0.033 19 0.894 Quamichan Okanagan 0.478 10 0.162 Williams Skaha -0.082 19 0.737 Quamichan Kalamalka 0.022 10 0.953 Williams Okanagan 0.084 19 0.732 Quamichan Wood 0.189 10 0.600 Williams Kalamalka 0.107 19 0.663 Quamichan Christina -0.054 10 0.883 Williams Sugar -0.176 19 0.472 Quamichan Sugar -0.140 10 0.699 Williams Mabel 0.056 19 0.821 Quamichan Mabel -0.086 10 0.812 Williams Mara 0.051 19 0.836 Quamichan Mara 0.135 10 0.710 Wood Ellison 0.494 27 0.009 Quamichan St Mary 0.399 10 0.254 Wood Kalamalka -0.144 28 0.464 Quamichan Prospect -0.049 10 0.893 Wood Osoyoos -0.272 29 0.153 Shaw nigan Chimney -0.465 10 0.175 Wood Skaha 0.402 29 0.030 Shaw nigan Horse 0.097 13 0.753 Wood Okanagan 0.266 29 0.164 Shaw nigan Lac La Hache 0.658 15 0.008 Shaw nigan Williams 0.067 17 0.799 Shaw nigan Christina 0.321 21 0.156 Shaw nigan Ellison 0.611 24 0.002 Shaw nigan Sugar 0.094 24 0.661 Shaw nigan Wood 0.699 25 0.000 Shaw nigan Kalamalka -0.036 26 0.861 Shaw nigan Mabel 0.524 26 0.006 Shaw nigan Osoyoos -0.422 27 0.028 Shaw nigan Skaha 0.049 27 0.807 Shaw nigan Okanagan 0.163 27 0.417 Shaw nigan Mara 0.314 27 0.111 Skaha Ellison 0.278 28 0.152 Skaha Osoyoos 0.383 31 0.034 St Mary Chimney -0.314 12 0.321 St Mary Lac La Hache 0.115 15 0.683 St Mary Horse 0.096 15 0.734 St Mary Williams 0.313 15 0.256 Lake pair Lake pair 86  APPENDIX 6B  Pearson product moment correlations (r) for spring total phosphorus among BC lakes.      Pearson r Years p Pearson r Years p Charlie Christina -0.230 11 0.496 Fork St Mary -0.087 13 0.778 Charlie Williams -0.375 13 0.207 Glen Kathyln -0.116 10 0.749 Charlie Tabor 0.213 13 0.486 Glen Fork 0.586 12 0.045 Charlie Lac La Hache -0.212 13 0.487 Glen Horse 0.569 13 0.042 Charlie Ellison 0.430 14 0.125 Glen Chimney 0.531 14 0.051 Charlie Osoyoos -0.132 14 0.652 Glen Charlie -0.006 14 0.982 Charlie Skaha -0.362 14 0.203 Glen Quamichan -0.027 14 0.926 Charlie Okanagan -0.462 14 0.096 Glen Stocking 0.046 15 0.871 Charlie Wood -0.076 14 0.796 Glen Lac La Hache 0.410 20 0.073 Charlie Kalamalka -0.439 14 0.116 Glen Tabor 0.747 21 0.000 Charlie Sugar 0.105 14 0.722 Glen Christina 0.640 22 0.001 Charlie Mabel -0.111 14 0.706 Glen Shaw nigan 0.440 22 0.040 Charlie Mara -0.371 14 0.192 Glen Prospect 0.320 22 0.146 Chimney Christina 0.814 14 0.000 Glen Williams 0.304 23 0.158 Chimney Williams 0.232 14 0.425 Glen Ellison 0.205 25 0.325 Chimney Sugar 0.326 15 0.236 Glen Osoyoos 0.059 25 0.780 Chimney Ellison -0.144 16 0.594 Glen Skaha 0.271 25 0.190 Chimney Osoyoos 0.041 16 0.880 Glen Okanagan 0.555 25 0.004 Chimney Skaha 0.004 16 0.987 Glen Wood 0.083 25 0.694 Chimney Okanagan 0.343 16 0.193 Glen Kalamalka 0.669 25 0.000 Chimney Wood -0.252 16 0.346 Glen Sugar 0.210 25 0.313 Chimney Kalamalka 0.643 16 0.007 Glen Mabel 0.529 25 0.007 Chimney Mabel 0.724 16 0.002 Glen Mara 0.638 25 0.001 Chimney Mara 0.789 16 0.000 Glen St Mary -0.311 25 0.130 Christina Ellison 0.237 23 0.276 Horse Tabor 0.723 11 0.012 Christina Osoyoos 0.082 24 0.702 Horse Lac La Hache 0.362 11 0.274 Christina Skaha 0.225 24 0.289 Horse Chimney 0.540 12 0.070 Christina Okanagan 0.678 24 0.000 Horse Christina 0.639 14 0.014 Christina Wood 0.152 24 0.478 Horse Sugar 0.379 14 0.181 Christina Kalamalka 0.758 24 0.000 Horse Williams 0.454 14 0.103 Fork Williams 0.334 11 0.316 Horse Ellison 0.216 15 0.439 Fork Prospect 0.515 11 0.105 Horse Osoyoos -0.136 15 0.630 Fork Tabor 0.809 12 0.001 Horse Skaha -0.110 15 0.697 Fork Ellison 0.412 13 0.162 Horse Okanagan -0.034 15 0.906 Fork Osoyoos 0.062 13 0.840 Horse Wood -0.266 15 0.339 Fork Skaha 0.585 13 0.036 Horse Kalamalka 0.668 15 0.007 Fork Okanagan 0.598 13 0.031 Horse Mabel 0.616 15 0.014 Fork Wood -0.033 13 0.915 Horse Mara 0.652 15 0.008 Fork Kalamalka 0.716 13 0.006 Kalamalka Ellison 0.245 29 0.200 Fork Christina 0.739 13 0.004 Kalamalka Osoyoos 0.179 31 0.335 Fork Sugar 0.731 13 0.005 Kalamalka Skaha 0.313 31 0.086 Fork Mabel 0.725 13 0.005 Kalamalka Okanagan 0.468 31 0.008 Fork Mara 0.748 13 0.003 Kalamalka Wood 0.179 31 0.334 Fork Shaw nigan 0.456 13 0.118 Lake pair Lake pair 87  Appendix 6B continued. Pearson product moment correlations (r) for spring total phosphorus among BC lakes.     Pearson r Years p Pearson r Years p Kathyln Ellison -0.362 10 0.304 Mabel Christina 0.702 24 0.000 Kathyln Osoyoos 0.217 10 0.547 Mabel Sugar 0.600 28 0.001 Kathyln Skaha 0.365 10 0.299 Mabel Ellison 0.309 29 0.103 Kathyln Okanagan 0.219 10 0.543 Mabel Osoyoos 0.187 30 0.321 Kathyln Wood 0.344 10 0.330 Mabel Skaha 0.433 30 0.017 Kathyln Kalamalka 0.285 10 0.424 Mabel Okanagan 0.517 30 0.003 Kathyln Sugar -0.153 10 0.674 Mabel Wood 0.191 30 0.313 Kathyln Mabel -0.064 10 0.860 Mabel Kalamalka 0.739 30 0.000 Kathyln Mara 0.200 10 0.580 Mara Christina 0.767 24 0.000 Lac La Hache Chimney 0.232 11 0.492 Mara Sugar 0.566 28 0.002 Lac La Hache Tabor 0.481 16 0.059 Mara Ellison 0.294 29 0.122 Lac La Hache Christina 0.500 18 0.034 Mara Mabel 0.768 30 0.000 Lac La Hache Sugar 0.482 22 0.023 Mara Osoyoos 0.112 31 0.547 Lac La Hache Ellison 0.322 23 0.134 Mara Skaha 0.286 31 0.118 Lac La Hache Williams 0.563 23 0.005 Mara Okanagan 0.576 31 0.001 Lac La Hache Osoyoos 0.443 24 0.030 Mara Wood 0.171 31 0.359 Lac La Hache Skaha 0.681 24 0.000 Mara Kalamalka 0.849 31 0.000 Lac La Hache Okanagan 0.406 24 0.049 Maxw ell Chimney 0.772 11 0.005 Lac La Hache Wood 0.496 24 0.014 Maxw ell Horse -0.106 11 0.757 Lac La Hache Kalamalka 0.677 24 0.000 Maxw ell Quamichan 0.034 12 0.916 Lac La Hache Mabel 0.611 24 0.002 Maxw ell Fork 0.523 12 0.081 Lac La Hache Mara 0.524 24 0.009 Maxw ell Charlie -0.166 13 0.589 Lizard Kathyln 0.183 10 0.612 Maxw ell Stocking 0.279 14 0.334 Lizard Quamichan -0.377 10 0.283 Maxw ell Lac La Hache 0.232 16 0.388 Lizard Lac La Hache 0.372 13 0.211 Maxw ell Christina 0.623 19 0.004 Lizard Charlie -0.289 13 0.338 Maxw ell Tabor 0.468 19 0.043 Lizard Stocking -0.120 14 0.682 Maxw ell Prospect 0.307 19 0.201 Lizard Christina -0.113 15 0.688 Maxw ell Williams -0.052 20 0.827 Lizard Prospect -0.495 16 0.051 Maxw ell Shaw nigan 0.635 20 0.003 Lizard Williams 0.382 17 0.130 Maxw ell Glen 0.210 21 0.362 Lizard Tabor 0.068 17 0.794 Maxw ell Ellison 0.132 22 0.559 Lizard Shaw nigan -0.024 17 0.927 Maxw ell Osoyoos 0.376 22 0.085 Lizard Maxw ell -0.180 17 0.489 Maxw ell Skaha 0.385 22 0.077 Lizard Ellison -0.217 18 0.388 Maxw ell Okanagan 0.368 22 0.092 Lizard Osoyoos -0.198 18 0.430 Maxw ell Wood 0.323 22 0.142 Lizard Skaha -0.058 18 0.820 Maxw ell Kalamalka 0.723 22 0.000 Lizard Okanagan 0.046 18 0.855 Maxw ell Sugar 0.353 22 0.107 Lizard Wood 0.034 18 0.894 Maxw ell Mabel 0.645 22 0.001 Lizard Kalamalka 0.114 18 0.654 Maxw ell Mara 0.608 22 0.003 Lizard Sugar -0.062 18 0.807 Maxw ell St Mary -0.057 22 0.802 Lizard Mabel 0.018 18 0.944 Okanagan Ellison 0.335 29 0.075 Lizard Mara 0.040 18 0.875 Okanagan Osoyoos 0.328 31 0.072 Lizard St Mary -0.349 18 0.155 Okanagan Skaha 0.569 31 0.001 Lizard Glen 0.190 18 0.449 Osoyoos Ellison 0.348 29 0.064 Lake pair Lake pair 88   Appendix 6B continued. Pearson product moment correlations (r) for spring total phosphorus among BC lakes.     Pearson r Years p Pearson r Years p Prospect Charlie -0.034 12 0.918 Shaw nigan Mabel 0.594 27 0.001 Prospect Horse 0.769 13 0.002 Shaw nigan Osoyoos 0.307 28 0.112 Prospect Chimney 0.654 15 0.008 Shaw nigan Skaha 0.533 28 0.004 Prospect Lac La Hache 0.255 18 0.307 Shaw nigan Okanagan 0.764 28 0.000 Prospect Tabor 0.547 19 0.015 Shaw nigan Wood 0.492 28 0.008 Prospect Christina 0.725 21 0.000 Shaw nigan Kalamalka 0.799 28 0.000 Prospect Williams 0.390 22 0.073 Shaw nigan Mara 0.567 28 0.002 Prospect Shaw nigan 0.559 23 0.006 Skaha Ellison 0.318 29 0.093 Prospect Ellison 0.265 24 0.211 Skaha Osoyoos 0.485 31 0.006 Prospect Sugar 0.554 24 0.005 St Mary Kathyln -0.165 10 0.649 Prospect Mabel 0.652 24 0.001 St Mary Charlie -0.480 14 0.083 Prospect St Mary 0.233 24 0.273 St Mary Horse -0.196 15 0.485 Prospect Osoyoos -0.152 25 0.469 St Mary Chimney 0.056 16 0.836 Prospect Skaha 0.260 25 0.209 St Mary Tabor -0.499 22 0.018 Prospect Okanagan 0.443 25 0.027 St Mary Christina -0.095 23 0.666 Prospect Wood 0.255 25 0.219 St Mary Lac La Hache 0.079 23 0.720 Prospect Kalamalka 0.562 25 0.003 St Mary Shaw nigan 0.095 26 0.644 Prospect Mara 0.716 25 0.000 St Mary Sugar 0.323 27 0.101 Quamichan Lac La Hache 0.219 10 0.544 St Mary Williams -0.035 27 0.861 Quamichan Horse 0.202 10 0.576 St Mary Ellison 0.280 29 0.141 Quamichan Williams 0.279 13 0.355 St Mary Osoyoos 0.089 29 0.646 Quamichan Shaw nigan 0.044 14 0.883 St Mary Skaha 0.239 29 0.213 Quamichan Prospect 0.182 14 0.534 St Mary Okanagan 0.049 29 0.802 Quamichan Ellison 0.344 15 0.209 St Mary Wood 0.453 29 0.014 Quamichan Osoyoos 0.139 15 0.620 St Mary Kalamalka -0.110 29 0.571 Quamichan Skaha 0.181 15 0.520 St Mary Mabel -0.146 29 0.450 Quamichan Okanagan -0.243 15 0.383 St Mary Mara -0.059 29 0.761 Quamichan Wood 0.580 15 0.023 Stocking Christina 0.580 12 0.048 Quamichan Kalamalka -0.096 15 0.733 Stocking Charlie -0.469 12 0.124 Quamichan Christina 0.013 15 0.965 Stocking Lac La Hache 0.561 13 0.046 Quamichan Sugar 0.246 15 0.377 Stocking Prospect 0.572 13 0.041 Quamichan Mabel 0.080 15 0.777 Stocking Tabor 0.220 14 0.450 Quamichan Mara -0.175 15 0.532 Stocking Shaw nigan 0.197 14 0.500 Quamichan Tabor -0.224 15 0.422 Stocking Ellison 0.134 15 0.634 Quamichan St Mary 0.331 15 0.228 Stocking Osoyoos 0.220 15 0.431 Shaw nigan Horse -0.010 13 0.973 Stocking Skaha 0.494 15 0.061 Shaw nigan Charlie -0.310 13 0.303 Stocking Okanagan 0.094 15 0.739 Shaw nigan Chimney 0.277 14 0.337 Stocking Wood 0.396 15 0.144 Shaw nigan Tabor 0.527 20 0.017 Stocking Kalamalka 0.678 15 0.006 Shaw nigan Lac La Hache 0.504 21 0.020 Stocking Sugar 0.588 15 0.021 Shaw nigan Christina 0.788 22 0.000 Stocking Mabel 0.548 15 0.034 Shaw nigan Sugar 0.442 25 0.027 Stocking Mara 0.680 15 0.005 Shaw nigan Williams 0.280 25 0.175 Stocking Williams 0.532 15 0.041 Shaw nigan Ellison 0.470 26 0.015 Stocking St Mary 0.591 15 0.020 Lake pair Lake pair 89  Appendix 6B continued. Pearson product moment correlations (r) for spring total phosphorus among BC lakes.     Pearson r Years p Sugar Christina 0.573 24 0.003 Sugar Ellison 0.388 27 0.045 Sugar Osoyoos 0.060 28 0.761 Sugar Skaha 0.415 28 0.028 Sugar Okanagan 0.552 28 0.002 Sugar Wood 0.345 28 0.073 Sugar Kalamalka 0.360 28 0.060 Tabor Chimney 0.712 11 0.014 Tabor Christina 0.715 19 0.001 Tabor Williams 0.278 20 0.235 Tabor Ellison 0.187 22 0.406 Tabor Osoyoos -0.070 22 0.758 Tabor Skaha 0.116 22 0.608 Tabor Okanagan 0.389 22 0.073 Tabor Wood -0.215 22 0.336 Tabor Kalamalka 0.826 22 0.000 Tabor Sugar 0.285 22 0.199 Tabor Mabel 0.733 22 0.000 Tabor Mara 0.827 22 0.000 Williams Christina 0.264 22 0.236 Williams Sugar 0.172 26 0.402 Williams Ellison -0.076 27 0.705 Williams Osoyoos 0.017 28 0.930 Williams Skaha 0.418 28 0.027 Williams Okanagan 0.154 28 0.433 Williams Wood 0.310 28 0.108 Williams Kalamalka 0.487 28 0.009 Williams Mabel 0.290 28 0.135 Williams Mara 0.417 28 0.027 Wood Ellison 0.362 29 0.054 Wood Osoyoos 0.428 31 0.016 Wood Skaha 0.774 31 0.000 Wood Okanagan 0.524 31 0.002 Lake pair 90  APPENDIX 7A  Testing null hypothesis: lake pair correlations for spring TN not different from zero.  o H0: correlations not significantly different from zero using all years and lakes for those lake pairs with more than 9 yrs of concurrent data. o HA: correlations different from zero.  Distributions TN  Test Mean=value     Hypothesized Value 0 Actual Estimate 0.05797 df 191 Std Dev 0.30783    t Test Test Statistic 2.6092 Prob > |t| 0.0098 Prob > t 0.0049 Prob < t 0.9951      o Result: t=2.609, p<0.01, df=191 o Conclusion: reject H0; mean of correlation significantly different than zero   0 -0.08 -0.06-0.04 -0.02 .00 .02 .04 .06 .08 91  APPENDIX 7B  Testing null hypothesis: lake pair correlations for spring TP not different from zero.  o H0: correlations not significantly different from zero using all years and lakes for those lake pairs with more than 9 yrs of concurrent data. o HA: correlations different from zero.  Distributions of Pearson Correlations (r)   Test Mean=value     Hypothesized Value 0 Actual Estimate 0.28435 df 295 Std Dev 0.31924    t Test Test Statistic 15.3244 Prob > |t| <.0001 Prob > t <.0001 Prob < t 1.0000      o Result: t=15.32, p<0.0001, DF=295 o Conclusion: reject H0; mean of correlation significantly different than zero     -0.6 -0.5 -0.4 -0.3 -0.2 -0.1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 -0.2 -0.1 .0 .1 .2 .3 92  APPENDIX 8A  Spring total nitrogen coherence (r) grouped by three ranges of lake water residence time (<1 yr, >1 yr to 5 yr, >5 yrs).     Water residence time <1 year Sugar Mara Williams Tabor Prospect Fork Glen Lizard Stocking Osoyoos 0.228 -0.047 0.033 id 0.377 -0.417 0.552 0.152 id Sugar 0.16 -0.176 id 0.468 -0.4494 0.593 0.49 id Mara 0.051 id -0.268 -0.308 0.061 0.269 id Williams id -0.103 id id id id Tabor id 0.008 id id id Prospect id id id id Fork id id id Glen id id Lizard id Stocking mean r = 0.083; standard deviation=0.315; n=20;  ; id=insufficient data (< 10 pairs) Water residence time >1 year to < 5 years Skaha Christina Mabel Horse Kathlyn Charlie Quamichan Shawnigan Ellison 0.278 0.035 0.533 -0.051 id id id 0.616 Skaha 0.403 -0.091 -0.234 id id -0.154 0.049 Christina 0.055 -0.127 id id -0.045 0.32 Mabel -0.026 id id -0.086 0.524 Horse id id id 0.097 Kathlyn id id id Charlie id id Quamichan id mean r = 0.116; standard deviation=0.262; n=18 ; id=insufficient data (< 10 pairs) Water residence time > 5 years Kalamalka Okanagan Chimney St Mary Maxwell Wood -0.144 0.266 -0.272 0.128 0.021 Kalamalka 0.182 0.682 -0.326 0.227 Okanagan 0.079 0.409 0.204 Chimney -0.314 id St Mary -0.289 Maxwell mean r =0.061; standard deviation=0.302; n=14;  ; id=insufficient data (< 10 pairs) 93  APPENDIX 8B  Spring total phosphorus coherence (r) grouped by approximate lake pair quartiles with water residence times of < 0.75 years, 0.75-1.2 years, 1.2 -5 years, and > 5 years.     Water residence time <0.75 year Mara Glen Sugar Williams Osoyoos Fork 0.748 0.586 0.731 0.334 0.062 Mara 0.638 0.566 0.417 0.112 Glen 0.21 0.304 0.059 Sugar 0.172 0.06 Williams 0.017 mean r = 0.334; standard deviation=0.262; n=15; id=insufficient data (< 10 pairs) Water residence time 0.75-1.2 years Lizard Prospect Stocking Ellison Skaha Tabor 0.068 0.547 0.22 0.187 0.116 Lizard -0.495 -0.12 -0.217 -0.058 Prospect 0.572 0.265 0.26 Stocking 0.134 0.494 Ellison 0.092 mean r = 0.138; standard deviation=0.288; n=15; id=insufficient data (< 10 pairs) Water residence time 1.2 - 5 years Shawnigan Quamichan Mabel Horse Christina Charlie Kathlyn id id -0.064 id id id Shawnigan 0.044 0.594 -0.01 0.788 -0.31 Quamichan 0.08 id 0.013 id Mabel 0.616 0.702 -0.111 Horse 0.639 id Christina -0.23 mean r = 0.212; standard deviation=0.392; n=13; id=insufficient data (< 10 pairs) Water residence time >5 years Lac La Hache St Mary Chimney Wood Kalamalka Okanagan Maxwell id id id 0.323 0.723 0.368 Lac La Hache id 0.232 0.495 0.677 0.406 St Mary -0.095 0.453 -0.11 0.049 Chimney -0.252 0.643 0.343 Wood 0.179 0.524 Kalamalka 0.468 mean r = 0.319; standard deviation=0.285; n=17; id=insufficient data (< 10 pairs) 94  APPENDIX 9A  Spring total nitrogen coherence (r) among lake pairs grouped by 3 ranges of TN concentration: < 200 µg /L, 200-400 µg /L, and >400 µg /L.       Spring mean total nitrogen concentration < 200 ug/L Christina Stocking Mabel Sugar Mara Okanagan Lizard id id 0.319 0.49 0.29 -0.106 Christina id 0.055 0.335 -0.06 -0.159 Stocking id id id id Mabel 0.01 0.371 0.588 Sugar 0.16 -0.113 Mara 0.157 mean r=0.167; standard deviation=0.238; n=14;  ; id=insufficient data (< 10 pairs) Spring mean total nitrogen concentration range: 200-400 ug/L Fork Shawnigan Maxwell Skaha Kalamalka Glen Kathlyn Osoyoos Horse Prospect Fork id id -0.411 id id id -0.417 0.008 Shawnigan 0.239 0.049 -0.036 0.173 id -0.422 0.097 -0.375 Maxwell -0.24 0.227 id id -0.267 0.604 -0.269 Skaha -0.495 -0.012 id 0.387 -0.234 0.19 Kalamalka -0.166 id -0.171 -0.088 -0.493 Glen id 0.552 id id Kathlyn id id id Osoyoos 0.006 0.377 Horse 0.067 mean r=-0.04; standard deviation=0.0.309; n=28;  ; id=insufficient data (< 10 pairs) Spring mean total nitrogen concentration range: > 400 ug/L Quamichan Tabor Ellison St Mary Wood Lac La Hache Charlie Williams Chimney Quamichan id id 0.399 0.189 id id id id Tabor id id id id id id id Ellison 0.46 0.494 0.171 id 0.238 -0.696 St Mary 0.128 0.115 id 0.3123 -0.314 Wood 0.341 id -0.292 -0.272 Lac le Hache id 0.17 0.205 Charlie id id Williams id mean r=0.103; standard deviation=0.329; n=16; id=insufficient data (< 10 pairs) Anova: Single Factor SUMMARY Groups Count Sum Average Variance <200 ug/L 14 2.337 0.166929 0.056748 200-400 ug/L 26 -0.11 -0.00423 0.092673 >400 ug/L 16 1.6483 0.103019 0.108216 ANOVA Source of Variation SS df MS F P-value F crit Between Groups 0.292206 2 0.146103 1.655361 0.200746 3.171626 Within Groups 4.677799 53 0.08826 Total 4.970005 55 95  APPENDIX 9B  Spring total phosphorus coherence (r) among lake pairs grouped by 5 ranges of TP concentration: <10 µg /L, 10-20 µg /L, and >25 µg /L.     Spring TP < 10 ug/L Sugar Lizard Mabel Stocking Okanagan Christina Shawnigan 0.442 -0.024 0.594 0.194 0.764 0.788 Sugar -0.062 0.600 0.588 0.552 0.573 Lizard 0.018 -0.120 0.046 -0.113 Mabel 0.279 0.517 0.702 Stocking id 0.580 Okanagan 0.678 all sites: mean r= 0.380; standard deviation 0.317; n=20; id = insufficient data (lake pairs < 10) Lizard data removed: mean r =0.561, standard deviation 0.0.167; n=14 Spring TP 10-20 ug/L Fork Maxwell Mara Skaha Lac la Hache Prospect Chimney Kalamalka 0.716 0.723 0.849 0.313 0.677 0.562 0.643 Fork 0.527 0.748 0.585 id 0.515 id Maxwell 0.608 0.385 0.611 0.307 0.772 Mara 0.286 0.524 0.716 0.789 Skaha 0.681 0.260 0.004 Lac la Hache 0.255 0.232 Prospect 0.654 all sites: mean r= 0.536; standard deviation 0.216; n=26 Spring TP > 20 ug/L Osoyoos Glen Quamichan Kathyln St Mary Tabor Ellison Charlie Wood Williams Horse -0.136 id id id id 0.723 0.216 id -0.266 0.454 Osoyoos 0.059 0.139 0.217 0.089 -0.070 id -0.132 0.428 0.017 Glen id -0.116 -0.311 0.747 0.205 -0.006 0.083 0.304 Quamichan id 0.331 -0.224 0.344 id 0.580 0.279 Kathyln id id -0.362 id 0.344 id St Mary -0.499 0.280 -0.480 0.453 -0.035 Tabor 0.187 0.213 -0.215 0.278 Ellison 0.430 0.362 -0.076 Charlie -0.076 -0.315 Wood 0.310 all sites: mean r= 0.110; standard deviation 0.0.307; n=42 Charlie data removed: mean r =0.147, standard deviation 0.294; n=32 Anova: Single Factor with all data SUMMARY Groups Count Sum Average Variance <10 ug/L 20 7.596 0.3798 0.100649 10-20 ug/L 26 13.942 0.536231 0.046451 > 20 ug/L 42 4.6279 0.110188 0.094182 ANOVA Source of Variation SS df MS F P-value F crit Between Groups 3.09086691 2 1.545433 18.94174 1.57337E-07 3.103839 Within Groups 6.93504754 85 0.081589 Total 10.0259145 87

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.24.1-0069368/manifest

Comment

Related Items