The Open Collections website will be unavailable July 27 from 2100-2200 PST ahead of planned usability and performance enhancements on July 28. More information here.

Open Collections

UBC Undergraduate Research

Ecosystem Services of the British Columbia Coast: Modeling the Impacts of Agriculture on the Provision… Solomon, Cody; Thompson, Allison 2010-05-05

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Notice for Google Chrome users:
If you are having trouble viewing or searching the PDF with Google Chrome, please download it here instead.

Item Metadata


52966-Solomon_Cody_Thompson_Allison_UBC_2010_EOSC_Honours_Thesis.pdf [ 5.4MB ]
JSON: 52966-1.0053594.json
JSON-LD: 52966-1.0053594-ld.json
RDF/XML (Pretty): 52966-1.0053594-rdf.xml
RDF/JSON: 52966-1.0053594-rdf.json
Turtle: 52966-1.0053594-turtle.txt
N-Triples: 52966-1.0053594-rdf-ntriples.txt
Original Record: 52966-1.0053594-source.json
Full Text

Full Text

i  ECOSYSTEM SERVICES OF THE BRITISH COLUMBIA COAST: MODELING THE IMPACTS OF AGRICULTURE ON THE PROVISION OF SHELLFISH by CODY SOLOMON1 & ALLISON THOMPSON1 1B.Sc. Hons. University of British Columbia A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF BACHELOR OF SCIENCE (HONOURS) in THE FACULTY OF SCIENCE (Environmental Sciences)  This thesis conforms to the required standard  ......................................    ......................................     Kai M. A. Chan             Rebecca G. Martone Supervisors  THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver) APRIL 2010 ii ABSTRACT   The coastal ecosystems of British Columbia, Canada (BC) are highly impacted by human activities; this is due, in part, to a lack of understanding of the effects that multiple human activities may have on ecosystems and the services they provide for humans. Understanding the functioning of ecosystem services is important for both conservation efforts and natural resources management. We create a conceptual model of BC’s coastal ecosystem services and adopt a mechanistic approach to identifying and quantifying the linkages by which human activities may cumulatively impact ecosystem services. From this model we investigate the relationship between one human activity, agriculture, and one ecosystem service, the provision of shellfish, by identifying and quantifying the specific linkages between the two. We use a Geographic Information System (GIS) to model the impacts that agricultural runoff, in the form of nitrate, may have on the suitability of shellfish harvest sites (both wild and commercial) in BC. We use three different models to investigate the ways that nitrogenous runoff may cause changes to ecosystem functions, through eutrophication and the production of harmful algal blooms (HABs), HAB toxicity, and the subsequent closure of shellfish harvesting sites. From our results, we identify a trade-off between fertilizer use on agricultural land and potential production of shellfish in BC. A lack of spatially and temporally explicit data, along with uncertainty of the relationships in the literature, is the most important limitation to the applicability of our models to BC. We recommend further work investigating these relationships and an expansion of coastal monitoring systems in BC. iii ACKNOWLEDGEMENTS  Special thanks to our supervisors Rebecca G. Martone and Kai M. A. Chan for their invaluable support and contribution to our research. Also thanks to Brian Klinkenberg of the University of British Columbia (Vancouver) Geography department for his assistance with spatial modeling in GIS; as well as Douw Steyn and the entire Earth and Ocean Science department for all that they provided to help us along the way.         iv Table of Contents  TITLE PAGE….........................................................................................................................i ABSTRACT..............................................................................................................................ii ACKNOWLEDGEMENTS......................................................................................................iii TABLE OF CONTENTS..........................................................................................................iv LIST OF INTRODUCTION.....................................................................................................................1  Introduction to ecosystem services................................................................................1  Conceptual model..........................................................................................................5  Coastal ecosystem services of British Columbia...........................................................6 METHODS..............................................................................................................................12  Spatial mapping of agriculture and nitrate runoff........................................................12 The models...................................................................................................................14 RESULTS................................................................................................................................17  Agriculture and nitrate runoff in BC............................................................................17  Model 1........................................................................................................................18 Model 2........................................................................................................................18 Model 3........................................................................................................................24  DISCUSSION..........................................................................................................................26 v Spatial mapping...........................................................................................................26 HAB – Toxin – Shellfish Relationship........................................................................30 Limitations...................................................................................................................31 CONCLUSION........................................................................................................................35 BIBLIOGRAPHY....................................................................................................................37 APPENDICES .........................................................................................................................40 vi List of Figures  Figure 1.1 Four categories of ecosystem services.....................................................................2 Figure 1.2 How anthropogenic activities affect ecosystem services.........................................3 Figure 1.3 Conceptual model illustrating the processes by which human activities may cumulatively impact ecosystem services in British Columbia, Canada....................................9 Figure 2.1 Decision tree for algal bloom forecasting..............................................................15 Figure 3.1 Study area and location of agriculture, British Columbia, Canada....................... 17 Figure 3.2 Linear regression results of model 1......................................................................18 Figure 3.3 Model 2 results using 99.5% decay........................................................................20 Figure 3.4 Model 2 results using 66.7% decay........................................................................21 Figure 3.5 Model 2 results using 50% decay...........................................................................22 Figure 3.6 Model 2 results using 33.3% decay........................................................................23 Figure 3.7 Model 3 hydrodynamic stability results for least stratified conditions...................24 Figure 4.1 Shellfish culture capability in British Columbia ....................................................29  1 1. INTRODUCTION   Introduction to Ecosystem services Ecosystem services are the processes and conditions by which humans benefit from the ecosystems around us (Costanza et al. 1997; Millennium Ecosystem Assessment (MA) 2003, 2005). These benefits can be provided by ‘natural’ ecosystems, such as water filtration from pristine wetlands, or from engineered and managed systems such as the provisioning of food from agriculture. There are four generally accepted categories of ecosystem services: supporting, e.g. nutrient cycling and waste processing; provisioning e.g. the production of food; cultural, e.g. recreation and spiritual value; and regulating, e.g. carbon sequestration that mitigates climate change (Figure 1.1) (MA 2005; Chan et al. 2009). Obtaining the benefits provided by ecosystem services constitutes an integral part of the world’s socio- ecological and economic systems and in the absence of humans there are no services (Bennett et al. 2009). Clean drinking water, food production, climate regulation and aesthetic and spiritual values are all examples of ecosystem services and all of these can be negatively affected by misuse and degradation of the environment. The United Nations’ Millennium Ecosystem Assessment (2005) found that “60 per cent of the ecosystem services that support life on Earth... are being degraded or used unsustainably.” The causes of these negative impacts, which are most commonly the result of human activities, will be referred to as drivers, i.e. the factors that drive ecosystem change (Figure 1.1). It is important to note that while the majority of this paper is concerned with negative impacts to ecosystem services, both activities and drivers can force positive changes, as in the case of ecological restoration. Human activities can impact the functioning of ecosystem services in a multitude of ways. Ecosystem services, the benefits humans obtain from ecosystems, can change due to changes in ecosystem service providers – the ecosystems and ecosystem components that generate the services themselves. Often, societies modify their environments to maximize the  2 productivity of provisioning services, as in the case of agriculture (Kareiva et al. 2007; Foley et al. 2005). These modifications, however, can cause adverse reactions in the same, or other, ecosystem service providers, which in turn affects the benefits humans may gain from ecosystem services. Human activities may impact one or several ecosystem services, and a variety of human activities may impact a single ecosystem service provider, or service (Figure 1.2). The relationship between these impacts is largely unknown; they may be additive, multiplicative, offsetting, etc. Trade-offs occur when a human activity increases production of an ecosystem service while negatively affecting another. A common example of a trade-off is the intensive use of fertilizer on agriculture lands that can cause nutrients such as nitrogen and phosphorus to leach out of soil into nearby water bodies and  accelerate the process of eutrophication, resulting in harmful algal blooms that may reduce dissolved oxygen, negatively impacting water quality, fish and other aquatic wildlife populations. This was the case in the Gulf of Mexico, where the production of corn-based ethanol increased nutrient runoff into the Mississippi River. The resulting eutrophication caused the water Figure 1.1 There are four categories of ecosystem services as defined by the Millennium Ecosystem Assessment (2003, 2005).  Anthropogenic activities affect drivers, which can in turn affect ecosystem services (figure modified from MA 2005 and Chan et al. 2009).  3 along the shelf of the gulf to become hypoxic (Donner & Kucharik 2008). A synergy occurs when the effect of one or a combination of activities is greater than the sum of the activity or activities alone. A synergy may occur within one ecosystem service/provider, or across multiple. An example of a synergy is the improvement in water quality and filtration and flood control, both supporting services, that results from wetland restoration, a human activity (Hey et al. 2004; Zedler 2003). Cross-system linkages are those in which a human activity occurring in one ecosystem drives change in another, e.g. how land use affects coastal waters (Halpern et al. 2009). All of the previously described linkages are possible ways in which the cumulative impacts to ecosystem services may be manifested. Understanding the different types of linkages and how they operate is an essential step in evaluating the different types of relationships between human activities and ecosystem service providers and services and how they operate on a case-to-case basis; both of these are essential for successful ecosystem-based management. The concept of ecosystem services allows managers to explicitly evaluate the linkages between activities, their impacts and the changes to benefits that humans ultimately receive. Figure 1.2 Anthropogenic activities, whether intentional or not, drive change in ecosystems that then can directly affect the provisioning of services or cause interactions such as trade- offs which then change the benefits that society receives. Figure redrawn from Bennett et al. 2009). !  4  The ecosystem services concept is an increasingly important lens through which to perform both conservation and management of both resources and land-use (Nelson et al. 2009; Ban et al. in prep). Assessing decisions in terms of changes to ecosystem services allows for an integrated approach where trade-offs can be assessed for a variety of management objectives, all revolving around impacts to human well-being (Balvanera et al. 2001; Singh 2002; Kremen & Ostfeld 2005). This is in contrast to more traditional management approaches that focus on a single objective or single problem, often leading to mis-calculation or mis-identification of the benefits of various actions across sectors of society. Similarly, management practices that aim to maximize the production of one service often end up decreasing the productivity of a variety of other services (Bennett et al. 2009). An ecosystem services framework for management must include all ecosystem services for the relevant area in order to be effective, and the scale of assessment must be local (Bennett et al. 2009). Despite the increase in popularity of the ecosystem services framework and its integrated approach, there is little information available on the interactions between these services at a regional scale, which is the pertinent scale of much management (Chan et al. 2006). Mapping ecosystem services is integral for the ecological understanding of their function and spatial extent, and provides a beneficial lens through which to perform ecosystem-based management (Kremen and Ostfeld, 2005). Initiatives to map or catalogue ecosystem services at a global scale bring attention to the need for integrated management approaches, but do not give managers the detailed understanding that is necessary in order to assess trade-offs and make informed decisions (Naidoo et al. 2008, Bennett et al. 2009). Globally, cumulative impacts of human activities to marine ecosystems have been mapped and each region given an impact score; the framework provided by this global assessment has proven successful for more local scales as well (Halpern et al. 2008, Halpern et al. 2009). The 2005 Millennium Ecosystem Assessment evaluated threats to ecosystem services. More local approaches have set up a framework for evaluating the impact of land-use changes and conservation scenarios on ecosystem services, a result directly beneficial to managers (Chan et al 2006). Barbier et al. (2008), Egoh et al. (2008), and Naidoo & Rickets (2006) produced maps of ecosystem services to assess specific ecological questions with important management consequences. Others have attempted to  5 catalogue and map services in various regions, which can help land-use managers to prioritize various areas (Ban et al. in prep.; Chan et al. 2009). In this paper, we aim to build on the concept of evaluating cumulative human impacts to ecosystems, and extend this to the cumulative impacts of human activities on the ecosystem services themselves. Cumulative impacts are "spatially or temporally accumulated changes" caused by human activities that affect the functioning of ecosystem service providers and thus the production of ecosystem services (Forrex 2010). A quantitative evaluation of the impacts of human activities on a variety of ecosystem services would allow for effective management based on the ability to spatially prioritize conservation initiatives, and successfully assess the trade-offs of various land-use changes. To date, the most recent proposed method for assessing human impact to ecosystem services identifies two relationships: the relationships between drivers and ecosystem services and the interactions among multiple ecosystem services (Bennett et al. 2009). It is well-recognized that the relationship between ecosystem services and the way that human actions affect them is complex. Although drivers can directly impact the provisioning of ecosystem services in many cases, they also have indirect effects which may be more important and which require a more complex set of steps to identify and understand. For example, the relationship between nutrient runoff (a driver) and shellfish provision (a service) cannot be understood without the understanding of the relationships between nutrient runoff and harmful algal blooms, and the effect of harmful algal blooms on toxins and so on shellfish. We therefore expand on the driver-service methodology and arguing that the processes governing change to ecosystem services follow a pathway with multiple steps.  Conceptual Model We identify five aspects of the human activity – ecosystem service relationship: human activities, drivers, ecosystem change-agents, ecosystem service providers and ecosystem services (Figure 1.3). Activities include both the purposeful manipulation of ecosystems and the indirect effects that human societies can have on ecosystems. The component of the activity that enters an ecosystem is a driver, e.g. coastal development is an  6 activity and the pollution, sewage and runoff entering the marine environment are the drivers. Drivers cause changes in ecosystems through ecosystem change-agents: ecosystem changes, processes, phenomena, or elements that are frequent focal points of changes in ecosystem service providers. Ecosystem change-agents can either take the form of biological processes, e.g. oxygen consumption and harmful algal blooms, or abiotic factors, e.g. heavy metals. Ecosystem service providers are affected by threats; the providers are also what make ecosystem services possible, providing humans with tangible and intangible benefits that we call ecosystem services. The factor connecting all these together in a cyclical nature is society; humans require ecosystem services for well-being, but also change these services directly or indirectly through their actions. By adopting a mechanistic approach that evaluates impacts for each linkage, we aim to provide a more complete understanding of the dynamics between human activities and ecosystem services. Our conceptual model provides a comprehensive way of understanding both the relationships between human activities and ecosystem services, and the relationships between multiple ecosystem services. Coastal Ecosystem Services in British Columbia In this thesis, we investigated the cumulative impacts of multiple human activities and multiple ecosystem services of the coastal marine ecosystems in British Columbia (BC). We first identify a set of human activities and ecosystem services of the BC coast and the relationships between these. We then select a set of linkages that we quantify and model in GIS; the spatially explicit results will allow the assessment of various scenarios of human activity changes on the BC coast. There is very little understanding to date of ecosystem services in the marine environment, although marine environments are highly degraded globally (Ban et al. in prep). Coastal ecosystems are also much more vulnerable to human stressors than are open ocean ecosystems (Teck et al. 2009). The ecosystem stress caused by human activities integral to BC's economy and culture has taken its toll on coastal BC ecosystems. Approximately 98% of the BC continental shelf and slope are affected by anthropogenic activities, while only 6% of the coast is currently protected (Ban & Alder 2008). A global map of human impacts to ecosystems classifies the BC coast as ranging from medium to high impact (Halpern et al. 2008). The results of this global map and knowledge of local levels of degradation along with the reliance of BC's economy on ocean-related  7 activities highlight the importance of a local understanding of the impacts that human activities have on the functioning of BC's coastal ecosystem services. We have created a conceptual model that adapts work done by Ban et al. (in prep) cataloguing BC's coastal ecosystem services into our multiple-step format (Figure 1.3). Although the ecosystem services and human activities affecting them were compiled and described in terms of benefiting from or improving/degrading the production of ecosystem services, we aim to take a closer look at these relationships, quantify them and examine how multiple human activities can cause cumulative impacts to ecosystem services. It is beyond the scope of our analysis to research the entire set of linkages described in our conceptual diagram, so we selected a set of linkages to explore in full, by modeling and mapped in a spatially explicit setting, when possible.  The human activity we have chosen to investigate is agriculture, and the ecosystem service provider is shellfish production, a subset of the ecosystem service 'food provision'. Agriculture contributed 1% to the BC GDP in 1996, and employed 1.8% of the workforce (BC Ministry of Finance and Corporate Relations 1999). Although the agriculture sector is relatively small in BC, its impact on coastal ecosystems through runoff may be significant. Fertilizer travels from farms into aquatic ecosystems in the form of nutrient runoff, the driver of ecosystem change for this linkage. Nutrient runoff enriches ecosystems, accelerating the process of eutrophication and allowing phytoplankton to bloom if physical and chemical conditions are compliant (Heisler et al. 2008; Anderson et al. 2002; Sellner et al. 2003). Eutrophication leading to harmful algal blooms is the ecosystem change-agent we will be investigating. Finally, these harmful algal blooms release toxins into the water which may be taken up and concentrated by shellfish, affecting the viability of both commercial shellfish aquaculture and wild shellfish harvest. The linkages described here are demonstrated in figure 1.3 with arrows and bold lettering. For each linkage we are separately modeling and characterizing the impacts; this mechanistic approach will allow us to evaluate the impact of agriculture on shellfish production in the most comprehensive way possible. Shellfish production can also be considered a cultural service, as shellfish harvesting has traditionally been practiced both for subsistence and recreation historically in BC (Quayle 1969, etc.). We explore the impacts of agriculture on both shellfish aquaculture sites and potential wild  8 harvest sites, therefore addressing impacts to two different ecosystem services. Although it is difficult to quantify impacts to cultural ecosystem services, we are able to show potential area lost to recreational shellfish harvesting, as well as aquaculture site closures. Changes to the value of the service are indicated by mapping bloom probability; the option value of the service decreases as habitat suitability decreases, e.g. bloom event probability increases, and with it the potential toxicity of the water to shellfish. We will also evaluate the trade-offs between two types of agriculture and the provision of shellfish in coastal BC. We use three different models to generate three different impact scenarios, and discuss the outcome of each.  9              Figure 1.3 A conceptual model illustrating the processes by which human activities may cumulatively impact the provision of ecosystem services in coastal BC. The bold typeface with the arrows identifies the components that we will be evaluating in detail. Arrows represent direct relationships and may flow linearly from one "step" to the next, e.g. drivers directly impacting change-agents, or may directly impact a component that they are not right next to, e.g. an ecosystem driver directly impacting an ecosystem service provider. Figure adapted from research in Ban et al. (in prep.). !  10 As of 1990, the British Columbian shellfish aquaculture industry was losing $2 million per year due to harvesting site closures (Shumway 1990). As of 2008 there were 503 licenses issued for shellfish harvesting sites in BC (BC Shellfish Growers Association 2008). Shellfish site closures are due to two factors: toxin contamination from harmful algal blooms (HABs) or unhealthy fecal coliform levels from sewage inputs. The first recorded death due to harmful algal bloom toxicity in BC was in 1793, when one of Captain George Vancouver's crew died after eating Mussels collected along the BC coast (Quayle 1969). By 1927 it was recognized that the toxins produced by harmful algal blooms were a serious threat on the west coast of North America (Horner et al. 1997). At this time there was only one toxin recognized as harmful, Paralytic Shellfish Poison (PSP) caused by the Dinoflagellate genus Alexandrium, of which there are multiple toxin-causing species (Horner et al. 1997). Domoic acid poison (DAP) is the other major toxin on the west coast of North America that is detrimental to human health. DAP causes amnesiac shellfish poisoning (ASP), and the organism responsible is the diatom genus Pseudo-nitzschia, again with multiple toxin-producing species (CFIA 2008). DAP causing ASP is a relatively new problem; it was first recognized in 1991 on the west coast of California (Horner et al. 1997). Both of these toxins are harmful to human health and can sometimes be deadly (CFIA 2008). Currently, all shellfish harvesting sites on the BC coast are monitored on a daily basis for PSP and DAP. A third toxin-caused condition is diarrhetic shellfish poison (DSP), but this toxin is very rare and not monitored on a regular basis, so will not be considered in our analysis. Shellfish harvesting sites are closed when PSP values are 80 micrograms per 100 grams of shellfish, or ASP is over 14 micrograms per 100g of shellfish. Harmful algal blooms may be toxic (producing toxins), or noxious (causing anoxia and clogging the gills of filter feeding animals); we are considering only the toxic effects of HABs (Anderson et al. 2002). In addition to affecting the production of shellfish, HABs can also be responsible for beach closures, impacting the provisioning of cultural/recreational services in BC. Harmful algal blooms have become a worldwide issue, causing problems on every continent (Anderson et al. 2002). Globally, the incidence of harmful algal blooms has been increasing, although at the same time HAB-monitoring systems have also become much more advanced (Horner et al. 1997; Anderson et al. 2002; Heisler et al. 2008). A worldwide  11 increase in fertilizer use producing nutrient-rich runoff has occurred coincident with the increase in HABs. Although the rise in runoff may not be the root cause of increased HAB frequency, it is generally agreed that runoff is a factor in the development and intensity of HABs. Despite the increase in HABs over time, and their effect on human health through the vector of shellfish, current understanding of the HAB - toxin relationship is uncertain at best, although recent advances in the forecasting of blooms have been made (Anderson et al. 2002; Shumway et al. 1990; Hodgkiss & Ho 1997; Wong et al. 2009; Pickell et al. 2009). This complexity is due to the intricate nature of ocean dynamics; factors ranging from temperature, salinity, ocean current speed and direction, macro- and micro-nutrients, and nutrient ratios have all been implicated as having an influence on the formation of HABs (Heisler et al. 2008). Due to data limitations, we have assumed that the formation of HABs is a simple process that can be predicted by nitrate concentration. We have found two datasets allowing us to predict occurrence of HAB based on nitrate or fertilizer use, and one model that employs stability and nitrate concentration criteria. Using several models will allow us to evaluate or make predictions within a range of values developed from among different assumptions regarding the relationship between HABs and nutrient runoff. Given that the data necessary to validate our models either have not been collected or are unavailable, we are unable to test how accurate our modeling approaches are in the context of BC.  12 2. METHODS     We performed a literature search, using search engines Web of Science and Google Scholar, to investigate the current knowledge of relationships between agriculture and the production of shellfish. We broke the relationship down into the various components based on our framework: the relationship between agriculture and nutrient runoff, nutrient runoff and harmful algal blooms, harmful algal blooms and toxin production, and toxin production and shellfish production. We searched a variety of governmental websites (The Pacific Division of the Department of Fisheries and Oceans, Canadian Food Inspection Agency, Environment Canada, British Columbia Ministry of Land and Agriculture) for data on agriculture and shellfish harvesting in BC. Although we were able to find data on shellfish aquaculture in BC, we were unable find data on wild shellfish fisheries, which proves limiting for our results. Not only did we search the scientific literature for qualitative information on each linkage, but we searched for quantitative relationships, especially numerical data, and also for data in a format compatible with ArcGIS 9.3. We found that data for all aspects of the agriculture-shellfish relationship are scarce, especially Canadian data, and that there is no general scientific consensus on any of the linkages either. In light of the uncertainties in the science, we opted to use a variety of models to compare the cumulative impacts, and analyze the resulting scenarios.  Spatial mapping of agriculture and nitrate runoff  Environmental Systems Research Institute's geographical information systems (GIS) software suite ArcGIS 9.3 was used for the spatial modeling and data analysis component of our study. All data sources used for the GIS model can be found in Appendix 1. Geospatial data for land use, rivers and watersheds in BC were all used as GIS layers. Land use data was kept at its native hectare resolution but all modeling was performed at square kilometre resolution. A layer was created to identify each km2 cell containing a river’s entry to the ocean; we manually removed those rivers with no agriculture in their watershed from the  13 layer (as they would not contribute a significant source of fertilizer runoff). A cost-distance map was created to calculate the radius of runoff diffusion into the ocean from each river mouth in such a way that it would follow along coastlines and travel around islands. To quantify the relationship between agriculture and fertilizer use, we used agriculture, fertilizer use, nitrate concentration, and total volume of water flux data (all average yearly values) for the Fraser River watershed (Appendix 1). We used the average value for fertilizer use per hectare in Canada. Yearly fertilizer use for the Fraser River watershed was calculated; this number was compared with yearly average nitrogen concentrations for the Fraser River. These values were then extrapolated to create a relationship between fertilizer use and nitrogen loading in waters, which was applied to different watersheds throughout BC.  In order to determine yearly average concentrations of nitrate for all of BC’s rivers, a relationship was built between catchment area and flow rate (Appendix 1).  Average yearly figures for nitrate concentration were generated using total mass of nitrate and volume of water per river. In the case of watersheds with multiple major rivers, the nitrate loading and flow rates were equally distributed. Once nitrate concentration per river was determined, the concentration was diffused equally in each direction from a point source cell at the mouth of the river into the ocean. This was preformed individually for each river major river mouth, for a total of 65 times (there were 65 rivers used in this study).  Each operation produced an individual layer representing the diffusion of nitrate from that particular river.  These layers were then summed such that the value in cell is represented by:  , where [NO3-]i is nitrate concentration of the water in a cell, i represents the river and layer number (1-65), mi is the yearly mass flux of NO3- by river in micrograms, Vi is the yearly volume of water flux by river in litres (so that mi/Vi = initial concentration at mouth of river i), ! is the decay coefficient (one minus the rate of decay, between 0 and 1), and x is the distance from the mouth of river i.  The mass flux of NO3-, m, was determined by calculating the ratio of tonnes of nitrate per year that flows through the Fraser River (extracted from measured average concentration, 151 "g/L, and measured yearly flow rate, 112 km2/yr) to the total amount of fertilizer applied on agricultural land in the Fraser watershed (calculated  14 by multiplying the average value for fertilizer use in Canada, 56 kg/ha, with the total area of agriculture within the watershed, 500,000 ha).  This ratio of total mass of NO3- runoff to fertilizer application and agriculture area was then applied to all other watersheds in the study to determine the mass flux for each river given a known area of agriculture within the catchment.  The volume flux, V, was calculated in order to derive a statistical relationship between watershed area and flow rate, (flow rate = 0.0118*watershed area (km2) + 895) specific for British Columbia, using hydrometric data from Environment Canada's Water Survey program (Appendix 1). Halpern et al. (2008) use a similar method to model diffusion, employing a decay rate of 99.5% (i.e. !=0.005, so that the value in a given cell is one half of one percent the value in the previous cell). We used a range of decay coefficients for our model: 99.5, 66.7, 50 and 33.3 percent decay, and present results for each. Coastal currents and trapping by confined waterways and inlets were not included in our model due to a lack of spatially explicit data.  The Models We used three different models to evaluate the impact of nitrate runoff on the formation of harmful algal blooms. This was done for multiple reasons. In general, there is good reason to believe that nutrient runoff is a factor in the formation of HABs. Strong correlations have been demonstrated globally between total nitrogen input and phytoplankton production in marine and estuarine waters (Horner et al. 1997; Anderson et al. 2002; Heisler et al. 2008). In the coastal waters of China and Japan, increased nutrient loadings have coincided with the development and expansion of large biomass HABs, impacting fisheries and human health. These same regions have recorded reduced incidences of HABs when controls on nutrient input were put in place (Anderson et al. 2002). However, there is uncertainty around the relationship between nutrient runoff in HABs specific to BC, in part because there is very little available data for harmful algal blooms or nitrate loading and ambient concentration, as well as the other environmental conditions that may either prevent or facilitate blooms (Anderson et al. 2002; Barbier et al. 2008; Sellner et al. 2003; Robinson & Brown 1983). The four different models present a range of response variables, i.e. HABs per year, likelihood of a HAB. They also involve the input of a variety of parameters; from merely nitrogen concentration or fertilizer use, to those combined with wind speed, depth and  15 mixing conditions. For each model, we used the program Datathief to extract time series or spatial data, and fit statistical relationships between nutrient runoff and HAB formation. The first model is a statistical model relating incidence of HABs to fertilizer use for Chinese coastal waters. We extracted time series data of both fertilizer use (millions of tonnes per year) and HAB (events per year) located in Chinese coastal waters (Anderson et al. 2002, Zhang 1994). The natural log of HAB events were plotted against fertilizer use and a linear regression model was used to describe the relationship and to predict HAB events per year based on fertilizer use in BC, which was calculated by totalling the area of agriculture (Figure 3.1), multiplying that by the average fertilizer use per hectare per year. For the second model we extracted time series data of nitrate concentrations (mg/L) per month and monthly HAB events, from 1986 to 1989, also obtained from the Chinese coastal region (Hodgkiss & Ho 1997). We found that when nitrate concentration was over 50 "g/L there was a 0.76 probability of at least one HAB incident per month, and used this figure as our bloom “threshold”. Because this model involves nitrate concentrations from all sources, and we lacked data on background nitrate levels, our use of the model required an assumption of negligible background concentrations of nitrate levels (see discussion). The third model used involved methodology adapted from a forecasting model developed and applied to the coastal waters of China (Wong et al. 2009). This model uses a decision-making tree to forecast environmental conditions favourable for HABs to occur Figure 2.1 Decision tree for algal bloom forecasting involving hydrodynamic stability and nutrient threshold. Redrawn from Wong et al. (2009).  16 (Figure 2.1). The model's first step determines whether hydrodynamic conditions are suitable for bloom formation; if so, nutrient concentrations are considered. If both mixing and nutrient conditions are favourable, a bloom is predicted to form; otherwise it will not. Hydrodynamic stability (E in Figure 2.1) is a complex measure of how stable a water column is; in order for a bloom to occur this figure must be less than the critical turbulence threshold, represented as         (! refers to algal growth rate and l is depth of euphotic zone). The stability parameter, E, is a function of wind speed at 10 metres, W10 , water column depth, H, depth averaged tidal current, U, and the bulk Richardson number, Ri .  It is given by:  , where  when W10> 4.2 ms-1, and  when W10< 4.2 ms-1. The bulk Richardson number, Ri , is given by:                     , where # is water density and Us is the surface current (an empirical function of water column depth and wind speed obtained from three-dimensional hydrodynamic modeling), and is bounded by zero and 15 "above which there is no further reduction in diffusivity by stratification" (Wong et al.!2009).  Due to constraints in the availability of spatial data, we were only able to obtain the average wind speed and bathymetric (water column depth) data for BC (Appendix 1).  The remaining parameters were estimated from values used in the Wong et al. study (2009). We have adapted and simplified this model by making assumptions about all parameters excepting those requiring or related to wind speed and depth to accommodate for the limited amount of spatially explicit data we were able to obtain for the BC coast. For all three models, we assumed the same relationship between the occurrence of HABs and their effect on the viability of shellfish production sites. Due to data limitations and current lack of scientific understanding, we assumed that the occurrence of a HAB would always produce enough toxicity to prevent any suitable shellfish habitat from being harvested, be it aquaculture or recreational/wild harvest sites. There are relationships in the scientific literature linking HABs (in terms of cellular abundance) to toxin production but the connection between toxin production and its subsequent uptake and concentration by shellfish is unknown (Flynn 2002; Gedaria et al. 2007; Parkhill et al. 1999; Marchetti et  17 al. 2004). For this reason, we were unable to use toxin production to model potential shellfish closures.  18 3. RESULTS   Agriculture and Nitrate Runoff in BC                    Figure 3.1 A map of the coastal draining watersheds and locations of agriculture in British Columbia, Canada. !  19 Model 1 The first model is a statistical relationship between fertilizer use (millions of tonnes per year) and HABs (events per year), which we used to predict incidents of HABs based on fertilizer use. The equation obtained by linear regression is ln(HABs/yr) = 0.17*(fertilizer use in tons/yr) + 0.44, with an r2 value of 0.885. The total amount of agricultural area in BC (Figure 3.1) is 1367275 hectares of land, which was converted to 76.5 thousand tonnes of fertilizer for all of BC, using the average fertilizer use from Statistics Canada (2006). This amount of fertilizer produced a value of 1.45 HABs/year for all of BC. It was impossible to display this model in a spatially explicit format because fertilizer use for all of BC is significantly lower than the lowest data point extracted, and required extrapolating nearly to the y-axis.     Model 2 The probability of at least one HAB event per month when nitrate concentration is over 50 "g/L is 0.76. We assumed that at least one HAB per month was enough to cause a shellfish harvest closure and therefore did not consider the occurrence of multiple blooms for a given !"!#$" %"%#$" &"&#$" '"'#$" ("(#$" !" $" %!" %$" &!" &$" !" #$ % & '( )* +, -. /*,01!12*,.3'*.#41!!15"'.56.05""*'()*+,-. Figure 3.2 Linear regression (black line) of the relationship between fertilizer use and natural log of HABs per year (blue diamonds).  Original data from Zhang (1994). Appendix 3 contains additional information on suitability of linear regression.  20 area.  Due to the high variability in ambient nitrate concentrations in the waters of British Columbia's coast, an arbitrary value of 5 "g/L was chosen to represent background levels for visualization purposes. This value is slightly above the lowest detectable concentration of nitrate, 3.1 "g/L, but not high enough to influence HAB formation (Pickell et al. 2009). Figures 3.3, 3.4, 3.5, and 3.6 show the spatial distribution in concentrations of nitrate due to riverine plumes using decay coefficients of 0.005, 0.33, 0.5, and 0.67 respectively. Insets show two areas of interest; the Gulf Islands and Victoria region around south Vancouver Island where relatively high NO3- concentrations were produced from our model, and the Courtenay and Comox region on east Vancouver Island where there is a large density of shellfish harvesting sites (represented as purple dots).  Red pixels represent areas that have average concentrations over our nitrate threshold of 50 "g/L, and thus have a 76% probability of experiencing a bloom and thus a closure.  From Figure 3.3 it can be seen that using such a low decay coefficient (0.995; that used by Halpern et al.) produces limited results as the nitrate concentration in river plumes often becomes insignificant by the second cell (i.e. no effect is seen beyond a 2km radius of the river mouth), which is at odds with satellite images of river plumes.  Environment Canada states that in 2004 there were 161 square kilometres of shellfish harvesting closures as a direct result of agriculture runoff; we have listed the total area of cells with concentrations above our threshold in the description of each map.  Note that the areas surrounding the Fraser River and Victoria (top left and bottom middle of the insets on the right) have some of the largest affected areas, the implications of which can be seen from the lack of current shellfish tenures (as well as the areas left blank in Figure 4.1) these regions are permanently closed to shellfish harvesting due to a variety of reasons including agriculture runoff and sewage outflow (CFIA, DFO).  The much more sparsely populated north coast region, including the Haida Gwaii islands, were included in the analysis but due to the vast amount of land area and relatively small amount of agriculture, no substantial results were produced, so these areas are not shown on the maps.  21              Figure 3.3 Result of model 2 using a decay rate of 99.5% (! = 0.005). Although difficult to see using this high decay rate, there are a total of 34 square kilometers that are above our 76% HAB probability threshold, all located at the mouths of rivers.  22               Figure 3.4 Result of model 2 using a decay rate of 67% (! = 0.33, value in cell is one third the value in previous cell). There are a total of 83 kilometers that are above our 76% HAB probability threshold.  23              Figure 3.5 Result of model 2 using a decay rate of 50% (! = 0.50, value in cell is one half the value in previous cell). There are a total of 170 kilometers that are above our 76% HAB probability threshold.  24              Figure 3.7 Result of model 2 using a decay rate of 33% (! = 0.67, value in cell is two thirds the value in previous cell). There are a total of 495 kilometers that are above our 76% HAB probability threshold.  25 Model 3 The Wong et al. (2009) model is by far the most comprehensive and thus required more parameter to run than the other models.  Data availability for most of these parameters was a major constraint, so many of the parameters had to be estimated.  Figure 3.7 shows the hydrodynamic stability using extreme values for the bulk Richardson number and tidal current velocity (0 and 9.1 cm s-1 respectively, i.e. the lower bound for Ri and the maximum measured value for current) demonstrating an unrealistically unstratified environment for the purpose of showing that even when using extreme values for these parameters all of the shoreline remains stable with respect to the average critical turbulence threshold (         ) from Wong et al. (2009) of 2.82x10-4 m2 s-1 (conditions that favour the formation of blooms occur when E <         ).  Figure 3.7: Map showing hydrodynamic stability, E, using measured yearly wind speed and bathymetric data.  For this scenario depth averaged current was set to the maximum value from Wong et al. (2009) at 9.1 cm s-1 and the bulk Richardson number was set to the theoretical minimum of zero, in order to portray the least stratified conditions, demonstrating that even when these variables are set to extremes all of the coastline areas remain stable with respect to the critical turbulence threshold.!  26  Given that the entire BC coast meets the hydrodynamic stability condition, the next step in the decision tree (Figure 2.1) from this model is to determine areas in which the nitrate level exceeded the triggering level in order to conclude whether or not a bloom is likely to occur.  The triggering level used in the study was 533 !g/L.  Applying this triggering level to our modeled concentrations due to agriculture yields no locations where a bloom is likely to occur.  27 4. DISCUSSION    We implemented three models to predict HAB formation from agricultural fertilizer use because of the uncertainty of these relationships in the literature and general difficulty of obtaining the needed data. The first model provides a simple linear relationship between increasing fertilizer use over time and with it an increasing quantity of harmful algal bloom events. The second model uses a different approach by assigning a probability for a bloom given a concentration of nitrate. The third model, the forecasting model, is different from the previous two as it incorporates another variable based on hydrodynamic stability as a condition for bloom formation. We compared the results of this set of three models as an alternative to performing model validation techniques, which were impossible due to data limitation. The benefit of comparing three models is that we were able to generate a reasonable range of outcomes, and also examine the dynamic between fertilizer runoff and HAB production at a variety of scales. We were also able to identify aspects of the linkage that are least sound; this motivated our recommendations for further monitoring and data collection practices in British Columbia. The results of our three models varied greatly; this is due in part to differing scales and units of input parameters. The first model required extrapolating the linear regression line nearly to the y-axis, leading to unreliable outcomes that we were unable to display spatially. Though this model cannot accurately be applied to British Columbia due to a large difference in the magnitude of fertilizer use it does show with some degree of confidence that the two are in fact related. The second model was constrained by the fact that it could not be used to generate explicit predictions, but rather probabilities. Having said that, this model does demonstrate a strong relationship between elevated nitrate concentrations and HAB events, supporting the conclusions drawn from the first model. Another source of uncertainty to model 2 is the applicability of HAB event probabilities from the location of data collection to BC. The most important limitation of results generated by the third model is the number of parameters required that had to be estimated using data from China. This model has the highest potential to accurately model and predict HAB formation in BC, both spatially and  28 temporally and thus we base the majority of our recommendations around the factors currently impeding its application in BC. Our models may under predict the quantity of HAB events and thus closures due to the relatively small agriculture sector in BC and the uncertainty surrounding the magnitude of change that fertilizer runoff can cause in marine systems. Discrepancies between the results of the three models demonstrate a need for scientific research on the formation, frequency, and location of HABs, and their potential effect on shellfish and shellfish harvesting, specific to the BC coast. Despite the differences in each of the models, one common theme is that nitrate loading has a measurable effect on formation of harmful algal blooms, toxin production, and shellfish closures. Changes to potential shellfish harvesting areas represent potential changes to the ecosystem service provision, both cultural and provisioning. This information is important for managers in BC involved in policy creation and decision making, as management can benefit from effective consideration of tradeoffs and strategies for minimizing the impact of human activities, specifically agriculture, on marine systems and their functioning. All of the factors which acted to limit our application of these models in BC can be considered directions and objectives for further scientific research.  Applicability of the models to BC  The most limiting factor in the understanding of the ecological relationships between activities and services in BC is the lack of data. Not only is data lacking on specific relationships, but basic oceanic data is lacking for the BC coast, as well. A lack in both understanding of specific relationships and general data limited the design of our model, and our ability to verify evaluate the model projections. By constructing a variety of models demonstrating the effects of agriculture on potential shellfish production, we have demonstrated how the effect of a single activity can have a significant impact on an ecosystem service in BC. As more and more of the linkages become understood, and data becomes available, the results of a cumulative impact model will become much more  29 significant. Currently, shellfish site closures are determined by measuring shellfish for biotoxin contamination (PSP, ASP) on a regular basis. The employment of a HAB forecasting and monitoring program, similar to the one in place in China, along with shellfish biotoxin monitoring would allow for the development of a framework for discovering the specific causes of shellfish contamination, and managing those threats. The current method is a reactive approach; we recommend a proactive approach, as it would lead to better management of the ecological factors contributing to shellfish biotoxin contamination. The Pacific branch of the Department of Fisheries and Oceans Canada provides spatial data on shellfish culture capability for Manila clams, Pacific oysters, and Japanese scallops in British Columbia.  A map of suitable locations is shown (Figure 4.1) as a reference for, and as a complement to our results.  The parameters from which they determined this information are not explicit. Data are not included for this map around the cities of Vancouver and Victoria, because these locations are permanently closed to all shellfish harvesting due to high volumes of sewage outflow and other forms of runoff. Although agricultural runoff was not the only impact considered in the creation of this map, the regions designated as "poor" areas for shellfish harvesting generally coincide with the threshold exceeding areas generated by model 2 and suggests that all three of our models were conservative.  30 Figure 4.1  Map showing shellfish culture capability for Manila clams, Pacific oysters, and Japanese scallops in British Columbia. Redrawn from DFO Canada 2010. !  31 Spatial Mapping The GIS model we constructed is spatially explicit but lacks a temporal dimension; this is its most important limitation. All linkages defining the agriculture-shellfish relationship are sensitive to temporal variations. For example, fertilizer is applied to agricultural land seasonally, and rainfall, which influences the amount of fertilizer runoff entering waterways, varies both seasonally and annually. Nutrient inputs from fertilizer runoff can be both episodic and chronic in nature, both types of which influence the formation and persistence of HABs although the way in which they do so may differ (Heisler et al. 2008; Anderson et al. 2002). Another limitation of the GIS model is that it is unable to account for 3-dimensional dynamics, such as water column structure. HAB formation depends on mixing dynamics in the water column, which are determined by temperature, salinity, wind speed, bathymetry, upwelling conditions, and direction and speed of currents (Wong et al. 2009). Additionally, it is very difficult to model diffusion of nitrate concentration from river-based runoff inputs into the ocean without taking into account vertical structure of the estuary into which runoff is flowing. Direction and extent of diffusion depends currents, flushing time and estuary depth; all of these factors influence how long nutrients are available to phytoplankton for uptake (Anderson et al. 2002). Conversely, mapping our models spatially has allowed us to compare and define areas in BC where agriculture affects the production of shellfish most intensely.  HAB - Toxin - Shellfish Relationship There are many uncertainties in the HAB-toxin relationship; these uncertainties are expressed in the assumptions we made to model this linkage. Dinoflagellate species alone are able to produce a suite of twelve different toxins, each of which varies in its biological activity (Boyer et al. 1987; John & Flynn 2000). We assumed that all toxins produced by all species had the same effect on shellfish, and all toxins produced by a bloom of harmful algae were capable closing a site's availability to shellfish harvesting. Both Pseudo-nitzschia and Alexandrium, the most important toxin-producers in BC, exude more toxins under nutrient- and temperature-stressed conditions (Boyer et al. 1987; Flynn 2002). Toxicity of these  32 species is inversely related to growth rate, and increases as temperatures become less optimal and phosphorus limitation increases; Pseudo-nitzschia produce more toxins when their silica requirement is not met (Boyer et al. 1987; Flynn 2002). We were not able to include any of these additional variables in our model, which adds a source of uncertainty to our results.  The largest unknown in the relationship between agriculture and shellfish production is the relationship between toxin production by HAB species and shellfish toxin uptake. This relationship is very complex, both spatially and temporally, and can be influenced by a multitude of factors, the full range of which is unknown. Lab experiments have been able to measure toxin production by a selection of HAB species; the toxins are measured as concentration in water (Flynn 2002; Marchetti et al. 2004; Parkill & Cembella 1999; John & Flynn 2000). To date, the movement of these toxins in water and how the toxins are concentrated and taken up by shellfish has not been successfully measured. Closures of shellfish aquaculture sites are determined by measuring the toxicity in the shellfish; toxicity in water is not measured along with shellfish toxicity (CFIA 2008).   Limitations A limitation common to the application of models 1-3 to BC is the source of the data that inform them; all three models were based on datasets from the coastal waters of China. Chinese coastal waters are subtropical western Pacific, whereas BC coastal waters are temperate eastern Pacific. These two regions differ in both annual average temperature and precipitation patterns, upwelling dynamics, currents, nutrient conditions, wind speed, rate of nutrient turnover, and salinity (Meeuwig 1999). Despite these differences, both fertilizer use (leading to nutrient runoff) and HABs have increased globally over the same time scale as the data used for models 1 and 2 (Heisler et al. 2008; Parsons & Dortch 2002). Since mixing and nutrient ratios have both been recently described as important factors in defining favourable bloom conditions, these variables will create the largest uncertainty in using Chinese data on BC coastal waters (Heisler et al. 2008; Wong et al. 2009; ). Although the specific relationships may differ from China to BC (and will be discussed for each model separately), the general trend can be applied to different regions globally, as long as factors  33 potentially causing uncertainty are recognized. This conclusion applies for model 3 as well, since it is a model created for Chinese waters that we have applied to the BC coast. Another limitation common to all models is the variable ambient concentration of nitrate throughout the coastal ocean, and the variability in cellular concentrations that make up HABs and cause toxicity. Natural nitrate concentrations vary widely throughout BC's coast; because of this the effect of additional nitrate from agricultural runoff also varies widely. We were not able to account for the effect of natural nitrate concentrations, as spatially explicit nitrate data is not available for all of BC. The lack of this information implies that all of our models are conservative as we only considered nitrate from one source and were not adding this nitrate to an already-present baseline. Similarly, the concentration of cells that actually makes up a HAB varies by species and location. Generally, this concentration is considered to be 10^6 cells/L, and can increase to as high as 10^9 cells/L (Anderson et al. 2002). Despite this, concentrations as low as 100s of cells/L have been observed to cause enough toxicity in shellfish to harm humans due concentration of toxins by the shellfish feeding mechanism (Sellner et al. 2003). Although concentrations this low may not be explicitly considered HABs, they are still of concern for our purposes, as we are ultimately concerned about the change in the ecosystem service provider that is shellfish production. The r2 value for model 1 is very high, suggesting a strong relationship between fertilizer use and the formation of HABs. The causal relationship is likely not as strong as is suggested by the correlation coefficient, as both variables are increasing over time. It does imply, though, that eutrophication caused by fertilizer runoff is a contributing factor to the development and persistence of HABs (Trainer et al. 2003; Turner & Rabalais 1991; Parsons & Dortch 2002; Heisler et al. 2008). Another source of uncertainty is the scientists’ definition of a HAB. The amounts of fertilizer in the dataset are an order of magnitude larger than BC's annual fertilizer use and produce a relatively insignificant number of HABs per year. This suggests that the authors' methodology was centered on a conservative definition of what constitutes a HAB. The scale of the Chinese fertilizer data was the entire coastline; this may have led to the authors considering a number of smaller HABs occurring along the entire coastline at the same time to be "one" HAB event. Using data for all of the Chinese  34 coast and attempting to apply it to a Canadian province comes with a large source of uncertainty as the total amount of fertilizer used in China is many times larger than that for British Columbia, and generates a very conservative estimate of HABs. The data from model 2 were measured from China so the above discussion about uncertainty of its application to a different location applies. Another source of uncertainty is that we are considering anything equal to or over one HAB per month to be one HAB. Following from this, we are then assuming that one bloom per month can generate enough toxins to close a shellfish site from possible harvesting.  While our model did account for flow rate in terms of catchment area, it did not account for a decay coefficient that varies with river output.  In other words, a large river will most likely diffuse more slowly over distance than a smaller river due to total volume of water entering the ocean.  Another caveat to our method of diffusion is that we were unable to account for direction-specific advection and transport of nitrate by ocean currents and mixing; we assumed that the difference in densities between river/estuarine water and ocean water was great enough to have 100% stratification (i.e. no loss of nitrate due to vertical mixing).  It is important to note that our modeled values for [NO3-] only represent concentrations as a direct result of agriculture runoff; other sources of nitrate (e.g. coastal upwelling, waste from aquaculture) were not included in this analysis.  In determining concentrations within rivers, using data from the Fraser, we assumed that 100% of the nitrate loading was from agricultural runoff.  There are, of course, other sources of nitrogen within the natural and manmade environments, such as plant litter and sewage outflow, but we could not account for their representative fractions.  Although census data is available for quantity of livestock, whose waste is a source of nitrate and thus an inherent contribution to [NO3-] in rivers, we could not break down the percentages by land use that livestock agriculture occupies because the census data was not aerial.  Thus every catchment was assumed to have the same composition of types of agriculture as within the Fraser watershed, though one can be fairly certain there is a high degree of spatial variability in agriculture types (e.g. fruit orchards, grain fields, fodder crops) and fertilization methods (e.g. organic, animal manure, chemical) that can contribute to sources of variation in our result  35 Model 3 was the most comprehensive model applied to our study. Due to data constraints, we were unable to include all of the variables that the creators of this forecasting model suggested. Our simplifications decrease the accuracy of the predictions, although this model is the most comprehensive of those that we employed.  Although the coastal waters of British Columbia were shown to be hydrodynamically stable even under unrealistic conditions, the nitrate triggering parameter used by the authors was higher than our maximum modeled NO3- concentration, which resulted in zero predicted HABs.  A HAB- triggering level tailored to British Columbia's coast using empirical data would increase the applicability of this model to BC.  In terms of forecasting, this model would be better for real world application if actual measured oceanic nitrate concentrations were used rather than their theoretical component from agriculture alone, as there are many sources of nitrogen to the coastal region of BC missing from our model (Mackas & Harrison 1997).  Another constraint to the application of this forecasting model in BC is that to keep consistent with the temporal scale of nitrate data, we had to use yearly averaged wind speed values rather than using real-time data. This decision tree forecasting process implemented into a geographic information system does however provide a solid framework from which to actively monitor and predict where blooms are likely to occur in real time given adequate data availability. Despite the variety of results produced by our models and the lack of their current applicability to BC, our models show that there definitely exists a trade-off between fertilizer use on agricultural land and shellfish production in BC. Better management of agricultural runoff can be implemented without necessarily reducing agricultural production. With lower amounts of nutrient runoff our models all predict an increase in the number of available shellfish harvesting sites, and therefore the theoretical production of shellfish increase. With effective management of agriculture in BC, both these food provision services can increase at the same time.  36 5. CONCLUSION   Our approach provides a framework for identifying the ecological impacts of human activities on ecosystem services. The methods provided can be expanded on and used to identify locations in BC where management should target in order to assess and minimize cumulative impacts to ecosystem services, and also maximize the productivity of a variety of ecosystem services. Highly impacted services can be targets for sustainable management and/or conservation, and areas where better management practices for high-impact activities may be identified. Using a mechanistic approach allows managers to identify which aspect(s) of the activity-service relationship is causing the highest impact and target that specific linkage. The results of our spatial mapping identify regions of BC where better management regulations of a human activity can benefit not only the specific service (shellfish production) that we investigated, but a variety of ecosystem services, including food provision, regulating services, and cultural/recreational services. Shellfish harvesting is an important part of both the BC culture and economy and on top of this shellfish can act to purify water by filtering out toxins and bacteria. If fertilizer use in BC were better managed in order to avoid runoff, the provision of shellfish in BC could theoretically increase due to less site closures, without a decrease in agricultural production. Using the GIS as an analysis tool, different scenarios for runoff management can be assessed in terms of the gain or loss of potential shellfish harvesting areas, and the change in the provision of the service can be quantified. If fertilizer runoff was properly managed in BC, more shellfish habitat would become available for harvesting, increasing revenue and generating employment. Additionally, the cultural service that the harvest of wild shellfish provides would be increased, and the water filtration capability of shellfish would expand. In creating a conceptual model of the linkages between human activities and ecosystem services in BC, and modeling one of these pathways, we have demonstrated a framework for the evaluation of cumulative impacts to ecosystem services in coastal BC. We  37 have identified an important trade-off between an ecosystem driver caused by agriculture on shellfish production, as well as the gaps in knowledge of this relationship. We hope that our research into this complex cross-system trade-off will contribute to the understanding of how human actions can both benefit from and impact ecosystem services, the amount of work that needs to be done in order for successful ecosystem-based management to be practiced in BC, and why an ecosystem services framework is important factor in successful management.  38 7. BIBLIOGRAPHY   Anderson, D. M., P. M. Gilbert, and J. M. Burkholder. 2002. Estuaries 25(4b): 704-726. Ban, N. C., and J. Alder. 2008. How wild is the ocean? Assessing the intensity of anthropogenic marine activities in British Columbia, Canada. Aquatic Conservation – Marine and Freshwater Ecosystems 18(1): 55-85. Ban, N. C., G. Layton-Cartier, K. M. A. Chan. 2010. Marine ecosystem services and human activities: Benefits and impacts. In prep. Barbier, E. B., E. W. Koch, B. R. Silliman, S. D. Hacker, E. Wolanski, J, Primavera, E. F. Granek, S. Polasky, S. Aswani, L. A. Cramer, D. M. Stoms, C. J. Kennedy, D. Bael, C. V. Kappel, G. M. E. Perillo, D. J. Reed. 2008. Coastal ecosystem-based management with nonlinear ecological functions and values. Science 319(321): 321- 323. Beaumont N. J., M. C. Austen, J. P. Atkins, D. Burdon, S. Degraer, T. P. Dentinho, S. Derous, P. Holm, T. Horton, E. van Ierland, A. H. Marboe, D. J. Starkey, M. Townsend, and T. Zarzycki. 2007. Identification, definition and quantification of goods and services provided by marine biodiversity: implications for the ecosystem approach. Marine Pollution Bulletin 54(3): 253-265. Bennett, E. M., G. D. Peterson, and L. J. Gordon. 2009. Understanding relationships among multiple ecosystem services. Ecology Letters 12: 1394-1404. Boyer, G. L., J. J. Sullivan, R. J. Andersen, P. J. Harrison, and F. J. R. Taylor. 1987. Effects of nutrient limitation on toxin production and composition in the marine dinoflagellate Protogonyaulax tamarensis. Marine Biology 96(1): 1987. BC Ministry of Finance and Corporate Relations. 1999. The status of agriculture in British Columbia. Government of British Columbia. BC Shellfish Growers Association. 2008. BC shellfish industry strategic plan. Canadian Food Inspection Agency. 2008. Shellfish sanitation manual of operations. CFIA. Chan, K. M. A., M. R. Shaw, D. R. Cameron, E. C. Underwood, G. C. Daily. 2006. Conservation planning for ecosystem services. PLoS Biology 4(11): 2138-2152. Chan, K. M. A., P. White and N. C. Ban (2009). Catalogue of Key Ecosystem Services in the Great Bear Rainforest and Preliminary Assessment of Associated Revenue  39 Generation and Enhanced Value Opportunities, For Rainforest Solutions Project and Coastal First Nations: 50. Costanza, R., R. d'Arge, R. De Groot, S. Farberk, M. Grasso, B. Hannon, K. Limburg, S. Naeem, R. V. O’Neill, J. Paruelo,R. G. Raskin, P. Sutton, and M. van den Belt. 1997. The Value of the world's ecosystem services and natural capital. Nature 387: 253- 260. Donner, S.D., and C. J. Kucharik. 2008. Corn-based ethanol production compromises goal of reducing nitrogen export by the Mississippi river. Proceedings of the National Academy of Sciences 105: 4513-4518. Egoh, B., and B. Reyers. 2008. Mapping ecosystem services for planning and management: a biogeographical approach. Paper presented at the International Congress for Conservation Biology, Convention centre, Chattanooga, TN. Flynn, K. J. 2002. Toxin production in migrating dinoflagellates: a modelling study of PSP producing Alexandrium. Harmful Algae 1: 147-153. Foley, J. A., R. DeFries, G. P. Asner, C. Barford, G. Bonan, S. R. Carpenter, F. S. Chapin, M. T. Coe, G. C. Daily, H. K. Gibbs, J. H. Helkowski, T. Holloway, E. A. Howard, C. J. Kucharik, C. Monfreda, J. A. Patz, I. C. Prentice, N. Ramankutty, and P. K. Snyder. 2005. Global consequences of land use. Science 309(5734): 570-574. Krzyzanowski, J. and P. L. Almuedo (2010). Cumulative impacts of natural resource development on ecosystems and wildlife: An annotated bibliography for British Columbia. Forrex Forum for Research and Extension in Natural Resources, FORREX Series. Kamloops, BC: 43. Halpern, B. S., S. Walbridge, K. A. Selkoe, C. V. Kappel, F. Micheli, C. D’Agrosa, J. F. Bruno, K. S. Casey, C. Ebert, H. E. Fox, R. Fujita, D. Heinemann, H. S. Lenihan, E. M. P. Maldin, M. T. Perry, E. R. Selig, M. Spalding, R. Steneck, and R. Watson. 2008. A global map of human impact on marine ecosystems. Science 319(5865): 948- 952. Halpern, B.S., C. V. Kappel, K. A. Selkoe, F. Micheli, C. M. Ebert, C. Kontgis, C. M. Crain, R. G. Martone, C. Shearer, and S. J. Teck. 2009. Mapping cumulative human impacts to California Current marine ecosystems. Conservation Letters 2(3): 138-148. Heisler, J., P. M. Gilbert, J. M. Burkholder, D. M. Anderson, W. Cochlan, W. C. Dennison, Q. Dortch, C. J. Gobler, C. A. Heil, E. Humphries, A. Lewitus, R. Magnien, H. G. Marshall, K. Sellner, D. A. Stockwell, D. K. Stoecker, and M. Suddleson. 2008. Eutrophication and harmful algal blooms: A scientific consensus. Harmful Algae 8: 3- 13  40 Hey, D. L., D. L. Montgomery, L.S. Urban, T. Prato, A. Forbes, M. Martell, J. Pollack, Y. Steele, and R. Zarwell. 2004. Flood damage reduction in the upper Mississippi basin: an ecological alternative. The Wetlands Initiative. Chicago, IL. Hodgkiss, I. J., and Ho, K. C. 1997. Are changes in N:P ratios in coastal waters the key to increased red tide blooms? Hydrogiologia 352: 141-147. Horner, R. A., D. L. Garrison, and F. G. Plumley. 1997. Harmful algal blooms and red tide problems on the U.S. west coast. Limnology and Oceanography 42(5): 1076-1088. John, E. H. and K. J. Flynn. 2000. Growth dynamics and toxicity of Alexandrium fundyense (Dinophycae): the effect of changing N:P supply ratios on internal toxin and nutrient levels. Eur. J. Phycol. 35: 11-23. Kareiva, P., S. Watts, R. McDonald, and T. Boucher. 2007. Domesticated nature: shaping landscapes and ecosystems for human welfare. Science 316: 1866-1869. Kremen, C. and R. S. Ostfeld. 2005. A call to ecologists: Measuring, analyzing, and managing ecosystems. Ecology and the Environment 3(10): 540-548. Mackas, D. L. and P. J. Harrison. 1997. Nitrogenous Nutrient Sources and Sinks in the Juan de Fuca Strait/Strait of Georgia/Puget Sound Estuarine System: Assessing the Potential for Eutrophication.  Estuarine, Coastal and Shelf Science 44: 1-21. Marchetti, A., V.  L. Trainer, and P. J. Harrison. 2004. Environmental conditions and phytoplankton dynamics associated with Pseudo-nitzschia abundance and domoic acid in the Juan de Fuca eddy. Mar. Ecol. Prog. Ser. 281: 1-12 Meeuwig, J. J. 1999. Predicting coastal eutrophication from land use: an empirical approach to small non-stratified estuaries. Mar. Ecol. Prog. Ser. 176: 231-241. Millenium Ecosystem Assessment. 2003. Ecosystems and human well-being: a framework for assessment. Island Press, Washington, DC. Millennium Ecosystem Assessment. 2005. Ecosystems and human well-being: synthesis. Island Press, Washington, DC. Naidoo, R. and T. H. Ricketts. 2006. Mapping the economic costs and benefits of conservation. PLoS Biology 4(11): e360. Nelson, E., G. Mendoza, J. Regetz, S. Polasky, H. Tallis, D. R. Cameron, K. M. A. Chan, G. C. Daily, J. Goldstein, P. M. Kareiva, E. Lonsdorf, R. Naidoo, Taylor H. Ricketts, and M. Rebecca. Shaw. 2009. Ecol Environ 7(1): 4-11. Parkhill, J-P. and A. D. Cembella. 1999. Effects of salinity, light and inorganic nitrogen on growth and toxigenity of the marine dinoflagellate Alexandium tamarense from northeastern Canada. J. Plankt. Res. 21(5): 939-955.  41 Parsons, M. L., Q. Dortch, and R. E. Turner. 2002. Sedimentological evidence of an increase in Pseudo-nitzschia (Bacillariophycae) abundance in response to coastal eutrophication. Limnology and Oceanography 51: 448-462. Pickell, L. D., M. L. Wells, C. G. Trick, W. P. Cochlan. 2009. A sea-going continuous culture system for investigating phytoplankton community response to macro- and micro-nutrient manipulations. Limnol. Oceanogr. Methods 7: 21-32. Quayle, D. B. 1969. Paralytic shellfish poisoning in British Columbia. Ottawa, Canada, Fish. Res. Bd. Can. Bull. 168. Robinson, M. G., and L. N. Brown. 1983. A recurrent red tide in a British Columbia coastal lagoon. Can. J. Fish. Aquat. Sci. 40: 2135-2143. Rodríguez, J. P., T. D. Beard, Jr., E. M. Bennett, G. S. Cumming, S. Cork, J. Agard, A. P. Dobson, and G. D. Peterson. 2006. Trade-offs across space, time, and ecosystem services. Ecology and Society 11(1): 28. Sellner, K. G., G. J. Doucette, and G. J. Kirkpatrick. 2003. Harmful algal blooms: causes, impacts and detection. J. Ind. Microbiol. Biotechnol. 30: 383-406. Smil, V. 2001. Enriching the earth: Fritz Haber, Carl Bosch, and the transformation of world food. The MIT Press, Cambridge, Uk. Shumway, S. E. 1990.  A review of the effects of algal blooms on shellfish and aquaculture. J. World Aquaculture Soc. 21(2): 65-104. Taylor, F. J. R., and P. J. Harrison. 2002. Harmful algal blooms in western Canadian coastal waters. In: Taylor, F. J. R., and V. L. Trainer (Eds.) Harmful algal blooms in the PICES region of the North Pacific. PICES Scientific Report 23: 77-89. Teck, S., B. Halpern, C.V. Kappel, F. Micheli, K. Selkoe, C. Crain, R. G. Martone, C. Shearer, J. Arvai, B. Fischhoff, G. Murray, R. Neslo, and R. Cooke. 2009. Using expert judgement to assess the vulnerability of marine ecosystems in the California Current. Ecological Applications e-view. doi: 10.1890/09-1173. Turner, R. E., and N. Rabalais. 1991. Changes in Mississippi river quality this century and implications for coastal food webs. BioScience 41: 140-147. Wong, K. T. M., J. H. W. Lee, and P. J. Harrison. 2009. Forecasting of environmental risk maps of coastal algal blooms. Harmful Algae 8: 407-420. Zedler, J. B. 2003. Wetlands at your service: reducing impacts of agriculture at the watershed scale. Frontiers in Ecology and the Environment 1(2): 65-72. Zhang, J. 1994. Atmospheric wet depositions of nutrient elements: Correlations with harmful algal blooms in the Northwest Pacific coastal zones. Ambio 23: 464-468.  42  8. APPENDICES Appendix 1 Spatial (GIS) data sources:  • British Columbia Watershed Atlas (Ministry of Environment) <> • Canadian Wind Energy Atlas (Environment Canada) <> • Water Survey (Environment Canada) <> • Water Quality – Fraser River Monitoring Program (Environment Canada) <> • Integrated Land Management Bureau (Province of British Columbia) <>   Other data sources: • Statistics Canada 2006 Census of Agriculture o <>   43 Appendix 2 Map of British Columbia for reference to coastal study area.  Redrawn from information available from Google Maps.                       44 Appendix 3 Additional information for model 1 linear regression.  Original data extracted from Anderson et al. (2002) (redrawn from Smil 2001 and Zhang 1994) using the program Datathief. The data did not appear to be linear, so y-values were ln- transformed (Figure 3.2).  A histogram displaying the approximately normal distribution of residuals of the linear regression performed on ln(HABs per year) vs. fertilizer use (millions of tonnes per year) (Figure 3.2). The far left value (-0.8 to -0.6) looks like an outlier at a glance, but is not, as it is only one unit greater than its neighbour to the right. !"#!" $!"%!" &!"'!" (!" !" '" #!" #'" $!" $'" ! " # $% &' () %* (+ ), % -()./0/1()%2$(%&3/00/45$%46%.45$%'()%*(+),% !"#" $"%" &"'" (")" "*!+,"-."*!+(" "*!+("-."*!+&" "*!+&"-."*!+$" "*!+$"-."!""!"-."!+$" "!+$"-."!+&" "!+&"-."!+(" "!+("-."!+," "!+,"/01"/2.34"  45  X-axis values (fertilizer use) plotted versus residuals of the linear regression (Figure 3.2) showing approximate homoscedasticity of the residuals about the x-axis.  *!+,"*!+(" *!+&"*!+$" !"!+$" !+&"!+(" !+," !" '" #!" #'" $!" $'"7($/89 +0 $% -()./0/1()%2$(%&3/00/45$%46%.455($:*(+),%


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



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"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items