Assessment of freshwater ecosystem services in the Tualatin and Yamhill basins under climate change and urbanization

Assessment of freshwater ecosystem services in the Tualatin and Yamhill basins under climate change and urbanization

Applied Geography 53 (2014) 402e416 Contents lists available at ScienceDirect Applied Geography journal homepage: www.elsevier.com/locate/apgeog As...

4MB Sizes 1 Downloads 19 Views

Applied Geography 53 (2014) 402e416

Contents lists available at ScienceDirect

Applied Geography journal homepage: www.elsevier.com/locate/apgeog

Assessment of freshwater ecosystem services in the Tualatin and Yamhill basins under climate change and urbanization Robert Hoyer a, *, Heejun Chang a, 1 a

Department of Geography, PO Box 751 e GEOG, Portland State University, Portland, OR 97207 e 0751, USA

a b s t r a c t Keywords: Freshwater ecosystem services InVEST Scenario analysis Climate change Urbanization Riparian restoration

We estimate and map the provision of freshwater ecosystem services (ES) for the Tualatin and Yamhill basins of northwestern Oregon under a series of urbanization and climate change scenarios centered on the year 2050 using the Integrated Valuation of Environmental Services and Tradeoffs (InVEST) modeling toolset. Results for the study area suggest that water yield estimates are highly sensitive to climate, especially in the lowlands, while nutrient export and retention estimates are overwhelmingly driven by land cover. Sediment exports and retention are projected to increase throughout the study area due to higher erosivity from increasing winter rainfall. When the ES estimates are summarized as bundles, the spatial patterns of the levels of ES provision are largely consistent regardless of climate and urbanization scenarios. InVEST has utility as a landscape planning tool, and the presented analysis supports a conclusion that restoration efforts in the Yamhill basin would have a greater effect than those in the Tualatin in improving water quality downstream. The estimated relative changes in ES provision under different climate and urbanization scenarios are valuable for land management decisions because they show potential tradeoffs between provisioning and regulating ES. © 2014 Elsevier Ltd. All rights reserved.

Introduction A number of earth's ecosystem functions can be categorized as ecosystem services (ES) that are necessary to support life and provide benefits to humanity (MA, 2005). It is often argued they are not fully considered or altogether missing from decision-making on natural resource use (Costanza et al., 1997; Daily, 1997). Solving real-world conservation problems necessitates the formulation of standardized ES assessment methods (de Groot, Wilson, & Boumans, 2002). Standardization is important because it provides the transparency, repeatability, and ultimately the credibility for integration in the institutional decision-making structures for conservation and natural resource management (Crossman et al., 2013; Martínez-Harms & Balvanera, 2012). So even as the theoretical underpinnings of ES are still debated, a key challenge remains in the development of assessment tools based on sound interdisciplinary scientific knowledge (Daily et al., 2009; Portman, 2013; Postschin & Haines-Young, 2011). Institutional trust in

* Corresponding author. Permanent address: 5329 NE, 15th Ave., Portland, OR 97211, USA. Tel.: þ1 937 475 3354. E-mail addresses: [email protected], [email protected] (R. Hoyer), [email protected] (H. Chang). 1 Tel.: þ1 503 725 3162. http://dx.doi.org/10.1016/j.apgeog.2014.06.023 0143-6228/© 2014 Elsevier Ltd. All rights reserved.

assessment tools will facilitate addressing the problem of establishing a relationship between incentivized land uses and biophysical outputs that produce the desired ES. This is essential for defining the conditionality necessary for participants in marketbased instruments like ES credit exchanges or payments for ecosystem services (PES) schemes (Engel, Pagiola, & Wunder, 2008; Jack, Kousky, & Sims, 2008; Zheng et al., 2013). The costs and benefits from ecosystems are not distributed evenly and must be quantified in a spatially explicit manner (Eade & Moran, 1996; Naidoo & Ricketts, 2006; Troy & Wilson, 2006). Additionally, scientifically sound techniques for quantification and mapping of ES are essential components of an ES assessment (Burkhard et al., 2012; Burkhard, Crossman, Nedkov, Petz, & Alkemade, 2013). Researchers have presented several methods to do this (Chan, Shaw, Cameron, Underwood, & Daily, 2006; Egoh et al., 2008; Estoque & Murayama, 2012; Raudsepp-Hearne, €gner, Brander, Mae, & Hartje, Peterson, & Bennet, 2010; Scha 2013; Su, Xiao, Jiang, & Zhang, 2012), but they can vary widely. The spatially explicit ES modeling tool, Integrated Valuation of Environmental Services and Tradeoffs (InVEST), developed by the Natural Capital Project (www.naturalcapitalproject.org) (Tallis et al., 2013), offers a standardized approach to evaluate scenarios based on simple ecological production functions parameterized on LULC. InVEST has been used to assess ES in a variety of conservation settings around the globe (Ruckelshaus et al., 2013).

R. Hoyer, H. Chang / Applied Geography 53 (2014) 402e416

403

LULC is a dominant factor in driving the heterogeneity of ES on the landscape and tradeoffs among different types of ES (Bennett, Peterson, & Gordon, 2009; Qiu & Turner, 2013). Shifts in the pattern and makeup of LULC over time have consequences to the future amount and location of ES on the landscape (Lautenbach, Kegel, Lausch, & Seppelt, 2011). Future urban growth in the face of increased population will likely influence the amount and the spatial patterns of supplies and demand of various ES types (Eigenbrod et al., 2011). Relevant ES assessments need to consider future changes in climate as well as LULC change especially when considering water-related ES. Climate change will shift the amount and timing of water movement through the landscape, which alter the transport dynamics of nutrients and sediments. Identifying projected changes in freshwater ES is essential for new adaptive management strategies. Scenario analysis is seen as a method of evaluating possible futures, and is now advocated as an important research and planning tool in environmental studies (Peterson, Cumming, & Carpenter, 2003; Thompson et al., 2012). Involving stakeholders in the scenario development is often perceived as an important component, adding legitimacy to the process (Patel, Kok, & Rothman, 2007). Thus far, scenario analyses with InVEST have used LULC as the ES change driver (Bagstad, Semmens, & Winthrop, 2013; Goldstein et al., 2012; Nelson et al., 2009, 2010; Polasky, Nelson, Pennington, & Johnson, 2011). Fewer studies assessing future ES have incorporated climate (Bateman et al., 2013), and only one so far investigating climate impacts with InVEST (Terrado, ~ a, Ennaanay, Tallis, & Sabater, 2014). Acun This research investigates the potential freshwater ecosystem services response to the impacts of both climate change and land cover change in the form of increased urbanization and riparian buffer restoration. Our approach simplifies a complex system by implicitly assuming that changes to the landscape and to climate are independent for the study period we investigated. The study area is two river basins in northwest Oregon with a legacy of water quality issues and a precedent of active management aimed at ES enhancement. Using InVEST along with urbanization scenarios developed with stakeholder involvement (Hoyer & Chang, 2014), we investigate the following questions relevant to our study area.

from 2001 to 2010 (USGS, 2013b). The adjacent Chehalem Creek basin is also included due to its similarities to the larger basins. The study area falls into three broad categories of upland forests (~53%), valley floors dominated by agriculture (~30%), and urban land (~14%) based on USGS National Land Cover Dataset (NLCD) 2006 (Fry et al., 2011). Both river systems have reaches defined as impaired and subject to total maximum daily loads (TMDLs) for several water quality indicators under the Clean Water Act (ODEQ, 2012). Water quality issues in these basins contribute to the problems facing the greater Willamette River Basin (ODEQ, 2006). Upgrades to treatment plants successfully reduced loading over time, but the TMDL was recently amended to allow all facilities in the Tualatin to contribute loadings during low-flow summer months because of anticipated population growth (ODEQ, 2012). More developed land cover from increased population coupled with climate change has the potential for deleterious effects on the biological integrity of these river systems, as increasing temperature and nutrient loads further degrade dissolved oxygen levels necessary for a healthy aquatic system (Chang & Lawler, 2011; Praskievicz & Chang, 2011). Though facing similar problems, these basins have key differences affecting their provision of water-related ES. The Tualatin basin contains a much larger urban land cover base whereas the Yamhill basin contains more agricultural lands and contributes more water to the Willamette River (Fig. 1). However, in both basins, ES like water purification and sediment retention have the potential to mitigate some of the land uses and activities negatively impacting water quality as well as reduce the burden of existing and new infrastructure. There is already a precedent in the Tualatin River basin where a riparian restoration incentive program enhanced the ES of thermal shading (Cochran & Logue, 2011) and water purification (Singh & Chang 2014). In addition to shade, natural vegetation has the capacity to filter contaminants, excess nutrients, and mobilized sediments (Brauman, Daily, Duarte, & Mooney, 2007), and increase the sale price of near stream properties (Netusil, Kincaid, & Chang, 2014).

(1) What are the estimated changes in freshwater ES in the future relative to current estimates under the two main change drivers, namely climate and LULC? Which driver is more influential in determining future shifts in each ES? (2) What is the spatial distribution of freshwater ES currently and how might it change under the above mentioned drivers? (3) Finally, what are the usefulness and limitations of InVEST's freshwater components? Does it provide useful information to the ecologically-based management of the study area?

We manipulated and processed spatially explicit land cover and climate data (Table 1) as well as ran most InVEST models in ArcGIS 10.1 (ESRI, 2012). Each model requires its own set of variables, but several are common among the three. We chose the sub-watershed spatial unit to provide a level of detail allowing for differentiation of ES characteristics in the study area. The scenario LULC data were derived from NLCD 2006 (Fry et al., 2011) using a simple land change model based on the criteria identified at a stakeholder workshop. We interviewed several stakeholders who represented federal, state, and county perspectives regarding land use. For our study area, urban conversion of agricultural and forest lands was identified as the major change factor in the coming decades with agricultural conversion of natural vegetation being negligible under the assumption of all suitable lands already exploited. Specifics of the process are presented in Hoyer & Chang (2014). This model only focused on urban growth, and produced a low and a high scenario. The difference between them is the areal amount of urban growth occurring along the current urban/rural fringe. In order to test InVEST's response to a landscape scale change in vegetation management, we simulated a thirty meter total buffer strip of riparian vegetation directly adjacent to both sides of streams in the study area that meet two criteria: They are on privately held land and they are not in already developed areas. We chose a 30 m width to represent the smallest vegetation change possible with NLCD as a project partner and stakeholders communicated anything larger for the study area would be unrealistic (Cochran, personal

Methods Study area The Tualatin, Yamhill and adjacent Chehalem river basins in northwestern Oregon serve as our study area (Fig. 1). These basins were chosen since they are one of the fastest growing metropolitan areas in Oregon (Hoyer & Chang 2014), and future climates are projected to be hotter and drier in summer (Chang & Jung 2010), which will have large implications for the provision of freshwater ES. The Yamhill basin is historically wetter than the Tualatin. The southern fork of the Yamhill, at approximately three-quarters (1350 km2) the area of the whole Tualatin (1844 km2), yielded 982 mm of water annually on average than the Tualatin's 659 mm

Spatial data

404

R. Hoyer, H. Chang / Applied Geography 53 (2014) 402e416

Fig. 1. Study area e The Tualatin and Yamhill River basins and the interstitial Chehalem Creek basin.

communication). Although we do anticipate restoration in current urban lands, we assume that these will be targeted and will not always register relative to NLCD's resolution (30 m). Climate data were produced by combining historic data of varying spatial and temporal resolutions (Abatzoglou, 2013). Future scenarios used three global climate models (GCMs) from the Climate Model Intercomparison Project five (CMIP5) (Taylor, Stouffer, & Meehl, 2012), and were downscaled with the multivariate adapted constructed analogs (MACA) method (Abatzoglou & Brown, 2012). They were chosen to provide a low, medium, and high range of potential climate paths for our study area (Fig. 2). With no readily available future erosivity data, spatially explicit estimates were derived from climate model precipitation estimates (Nearing, 2001). The annual erosivity estimate is a function of a modified Fournier coefficient. An index was calculated using average monthly and average annual precipitation (Renard & Freimund, 1994). The future period was set at 2036e2065 in order to center the scenario at 2050. InVEST modeling took place at both historical and future periods with each model in order to calculate a percent change in ES provision. InVEST freshwater models Integrated Valuation of Environmental Service and Tradeoffs (InVEST) is an ES modeling toolset implemented in a GIS

environment developed by the Natural Capital Project (Kareiva, Tallis, Ricketts, Daily, & Polasky, 2011; Tallis et al., 2013). We provide short descriptions of the three freshwater models used. We chose the freshwater set for their relevancy to the current conservation issues facing the study area and to keep the scope of the assessment manageable. Although valuation functions are available, we chose to keep the assessment in biophysical production units. We felt they either did not apply (i.e. value of hydropower) or required accurate treatment cost data that are not currently available to the researchers. InVEST water yield InVEST water yield estimates an annual average water yield over a long term (>10 years) based on equations developed by Zhang, Dawes, and Walker (2001).

 Yxj ¼

1

AETxj Px



AETxj 1 þ ux Rxj ¼ Px 1 þ ux Rxj þ R1xj ux ¼ Z

AWCx Px

(1)

(2)

(3)

R. Hoyer, H. Chang / Applied Geography 53 (2014) 402e416

405

Table 1 Data requirements and sources for the InVEST freshwater models. A e Water yield, B e Water purification, C e Sediment retention. Data type

InVEST model

Source

Description

Annual average precipitation

A

University of Idaho Gridded Surface Meteorological Data (METDATA)

Potential or reference evapotranspiration

A

University of Idaho METDATA

Soil depth

A

Plant available water content Land use/land cover

A

USDA Natural Resource Conservation Service (NRCS) State Soil Geographic Database (STATSGO) USDA NRCS STATSGO

Thirty years of daily downscaled data summed to the annual scale and then averaged over the time periods 1981e2010 and 2036e2065. Resolution is 0.041667 decimal degrees (4 km  4 km). See above. Daily data for minimum/maximum temperature and solar radiation were averaged to the monthly scale for calculation of reference evapotranspiration using the Hargreaves method (Beguería & Vicente-Serrano, 2013; R Core Team, 2013). Maximum soil depth was set to 7000 mm. Polygons converted to 500 m  500 m raster dataset.

Watershed polygons

A, B, C

USGS National Land Cover Database (NLCD) 2006 Derived from DEM

Water yield Digital elevation model (DEM) Rainfall erosivity

B B, C

InVEST Water Yield Model NED 2004 DEM

C

Soil erodibility

C

USDA Isoerodent Map (Renard, Foster, Weesies, McCool, & Yoder, 1997) of the US (digitized by Natural Capital) USDA NRCS STATSGO

Rxj ¼

kxj $ETox Px

A, B, C

(4)

where Yxj is water yield at pixel x of LULC j, AETxj is actual evapotranspiration at pixel x of LULC j, Px is annual average precipitation at pixel x, ux is a ratio of a soil's plant available water capacity to Px, Rxj is the Budyko dryness index (Budyko, 1974) at pixel x of LULC j, Z is the Zhang coefficient, an integer from 1 to 10 that summarizes the area's annual precipitation distribution, AWCx is volume of water held in soil available to plants at pixel x, kxj is evapotranspiration coefficient at pixel x of LULC (vegetation) j, and ETox is reference evapotranspiration at pixel x. Soil and/or root depth of vegetation also help determine AWCx. The model assumes no change in groundwater storage over the long term. InVEST water purification The InVEST water purification model produces an annual export (kg ha1 yr1) by streamflow estimate for total phosphorus (TP) e total nitrogen (TN) can also be estimated e using the land cover type export coefficient method (Reckhow, Beaulac, & Simpson, 1980). The model adjusts the export values based on the hydrologic sensitivity score, the relative dryness or wetness of a pixel compared to the watershed's average water yield.

ALVx ¼ HSSx $polx

(5)

lx lW

(6)

HSSx ¼

lx ¼ log

X

! YU

(7)

U

where ALVx is the adjusted loading value at pixel x, HSSx is the hydrologic sensitivity score at pixel x, polx is the export coefficient

The fraction of water in soil that is available to plants. Polygons converted to 500 m  500 m raster dataset. Standard national land cover product for the contiguous United States. Study areas contain 15 land cover categories. Calibration polygons delineated using Arc Hydro (ESRI, 2012) from National Elevation Dataset (NED) 30 m DEM. Scenario and calibration for sediment retention polygons delineated in Soil and Water Assessment Tool (SWAT) (Arnold et al., 2012) Non-aggregated raster of water yield (mm). Hydrologically conditioned using National Hydrography Dataset (NHD) Plus version 1 stream layer. 30 m resolution. Erosion potential due to kinetic energy of rainfall. (megajoules mm/ha hour year) K-factor is soil's susceptibility to detachment and transport by rainfall. (metric tons ha hour/ha megajoules mm)

at pixel x, lx is runoff index at pixel x, lW is the average runoff index P in the watershed of interest, and U YU is the water yield above and including pixel x. It then employs a percent retention parameter for a land cover type to calculate the ES of nutrient retention (kg ha1 yr1). It then tracks the nutrient load as it moves downslope while accounting for each subsequent pixel's loading and retention until it reaches a stream where both are aggregated to the outlet (Tallis et al., 2013). A stream is set at the user-defined flow threshold, which we set to match closely to the USGS National Hydrography Dataset stream layer (USGS, 2013a). InVEST does not account for uptake limits or in-stream processes that can affect nutrient loadings. InVEST sediment retention InVEST sediment retention, like water purification, estimates streamflow export (metric ton ha1 yr1) and the ES of sediment retention (metric ton ha1 yr1). It scales up the well-known fieldlevel developed Universal Soil Loss Equation (USLE) (Wischmeier & Smith, 1978). Users can adjust the USLE terms C (Cover) and P (Practice) factors that account for vegetation (land cover) type and management practices' effects on soil mobilization. InVEST applies the equation to each map pixel. As with water purification, it adds a retention rate (Tallis et al., 2013). It is limited in the same way the water purification model is with no uptake limit or in-stream processes. InVEST calibration InVEST models were calibrated to empirically derived estimates of water yield along with nutrient and sediment loads. Calibrations were performed at nine locations at different time intervals (10, 15, 30 years) depending on data availability within the full historical study period of 1981e2010. Using the USGS estimator program LOADEST (Runkel, Crawford, & Cohn, 2004), streamflow and concentration samples of total nitrogen, total phosphorus, and total

406

R. Hoyer, H. Chang / Applied Geography 53 (2014) 402e416

Fig. 2. Plots of mean differences for the whole study area from historical (1981e2010) to future (2036e2065) periods for daily maximum temperature averaged by month and average monthly precipitation for the climate datasets in the scenario analysis for assessing freshwater ecosystem services in the Tualatin and Yamhill basins. The downscaled GFDLESM2M, MIROC5, and HadGEM2-ES general climate models represent the low, medium, and high climate change scenarios, respectively.

suspended sediments were used to estimate annual average exports. Samples were taken at gauges on a weekly schedule with some gaps. With available data, we subtracted wastewater treatment plant contribution estimates from the final load estimate for nutrients in order to focus on landscape effects since InVEST cannot account for point sources. Sensitivity analysis We tested the sensitivity of InVEST freshwater tools using the Tualatin River West Linn gauge near its mouth. There is no recommended parameter range associated with InVEST tools, so the procedure used the calibrated values (see Supplemental information) as a starting point. A single parameter type underwent a series of positive and negative adjustments. In some cases, a land cover's parameter reached the maximum or minimum allowable value before the full adjustment. For proportional parameters, we used 0.001 as the minimum, and 1.0 as the maximum. Effective retention in both water purification and sediment retention were set from 0 to 100%. Root depth was set to a minimum of 10 cm to the dataset maximum of 7000 cm. A minimum annual export coefficient was set at 1 kg per hectare and no maximum was set.

Mapping and statistical analysis For the scenario analysis we summarize results per subwatershed normalized by area. It is important to understand if stream exports and retention of nutrient and sediments vary across the landscape in a similar way. We use Spearman's rho to assess the relationships among each scenario output. For sediment results, we also tested for spatial autocorrelation with Moran's I. Rook's continuity was used for creating spatial weight metrics. We summarize all results by presenting them as an ES bundle in a single map. First, the sub-watershed estimations are normalized from 0.1 to 0.9. A weighted average is then taken where water yield is given forty percent of the weight and TN retention, TP retention, and sediment retention each represent twenty percent, based on initial stakeholder suggestions. Results Calibration InVEST freshwater models produced estimates comparable to the empirical observations. For water yield, these ranged from approximately ±10% for water yield, 8% to 18% for nutrient exports,

R. Hoyer, H. Chang / Applied Geography 53 (2014) 402e416

and 1% to 6% for sediments exports. The LOADEST results used for calibration suggest that both flow and time have a relationship with load. They additionally revealed spatial variation in the distribution of nutrient and sediment sources. The Yamhill basin showed higher loading estimates than the Tualatin for both nutrients and sediments. Accordingly, we assigned higher export coefficients to agricultural land in the Yamhill basin. For the water purification model, our results show the hydrologic sensitivity score does not lead to a response from the export estimate in our study area (Table 2). Scenario analysis The changes in InVEST outputs for each scenario are summarized for the study area in Table 3. The results concisely show whether climate or urbanization is the main change driver in each freshwater model. Water yield InVEST water yield model suggests that long term changes in water supply are more sensitive to climate change than the modeled increase in urban land cover or the simulated increase in riparian vegetation. The land cover change effect relative to the whole study area is small (from 2.5% to 7% depending on the scenario). This leads to stable sub-watershed patterns of water yield among the three climate change scenarios (Fig. 3). For the low and medium climate scenarios, wetter climates are projected in the future. The projected increases in precipitation in these scenarios more than compensate for increased temperature, resulting in an increase in annual water yield. The high climate scenarios show more seasonal variability in precipitation with increases during early winter (November and December) offset by losses during mid-late winter (January and February) (Fig. 2). Couple this to higher mean annual temperature leads to an increase in evapotranspiration and thus a small decrease in annual water yield. The spatial pattern suggests sub-watersheds lower in the basins will experience larger positive (under the medium climate scenario) or negative (under the high climate scenario) changes in water yield. The maps reveal subtle differences in yield change for some lowland sub-watersheds among the scenarios. These changes are attributed to either the new urbanization or riparian buffer installation effect on evapotranspiration. Water purification Higher TP exports occur only in the urbanizing watersheds of the Tualatin basin and in some scenarios, the interstitial Chehalem Creek basin as well. The scenarios support the evidence from

Table 2 Output of InVEST water purification for total phosphorus at two locations on the Tualatin River at two different time periods. Parameter values are the same in all cases with only the average water yield differing. The drier 2000s period is over predicted in both cases compared to the wetter 1990s, the stream loadings of which are more in line with the annual average for the entire study period (1981e2010) used in the analysis. Location: gauge and station

Time period

14207500 West Linn 14203500 Dilley

1991e2000 2001e2010 1991e2000 2001e2010

Empirical estimate of ave. ann. stream load (kg/yr.)

InVEST ave. ann. stream load (kg/yr.)

Percent difference

874 (838) 707 (659)

173,541 98,719

166,958 166,919

3.79% 69.08%

1103 (1172) 869 (967)

31,804 20,517

33,154 33,240

4.24% 62.01%

InVEST water yield estimate (empirical observation)

407

Table 3 Average sub-watershed percent change for each scenario's InVEST freshwater model output for the whole study area Scenario Medium historic climate Low urban growth High urban growth High urban growth with riparian buffers Low future climate Historic urban growth Low urban growth High urban growth High urban growth with riparian buffers Medium future climate Historic urban growth Low urban growth High urban growth High urban growth with riparian buffers High future climate Historic urban growth Low urban growth High urban growth High urban growth with riparian buffers

Water yield

TP export

TP retention

Sediment export

Sediment retention

0% 0% 0%

1% 3% 7%

0% 1% 5%

0% 2% 17%

0% 0% 0%

8% 8% 8% 8%

0% 1% 3% 7%

0% 0% 1% 5%

12% 12% 14% 7%

12% 12% 13% 13%

15% 15% 15% 15%

0% 1% 3% 7%

0% 0% 1% 5%

40% 41% 44% 17%

41% 41% 41% 41%

6% 6% 6% 6%

0% 1% 3% 7%

0% 0% 1% 6%

18% 18% 21% 1%

18% 18% 18% 18%

calibration that TP exports are land cover driven. Maps of the percent differences from baseline to future scenarios reveal the patterns of change (Fig. 4). There are clear differences in relative changes in the Yamhill and the Tualatin basins. Yamhill exports much more TP than the Tualatin. Most of this is attributed to agriculture thus a change to urban land cover types has a reduction effect. Riparian buffers decrease stream exports or mitigate the effects of increased urbanization. Nutrient retention results show a similar pattern to nutrient exports (Fig. 5). Correlations between retention and export support this especially for TP (r ¼ 0.92  0.94). Nutrient retention estimates are tied to the percent effective retention parameter in InVEST. For instance, decreases in TP exports and retention estimates are observed in the Yamhill basin the sub-watershed where we expect the most urbanization. There is no uptake limit so where there is more upslope nutrient mobilization, there is also more retention in the downslope pixels and vice versa. Widespread installation of riparian buffer strips has the potential to increase retention of nutrients in the agriculture dominated sub-watersheds. This is based on the assumption that it can retain a high percentage of nutrients that were exported upslope of them. Their placement near streams results in highest effectiveness since all mobilized nutrients must pass through a riparian pixel prior to export to a stream. Sediment retention Sediment export maps suggest that there are influences from both land cover and climate (Fig. 6). Spatial autocorrelation varies greatly for projected differences among the scenarios (I ¼ 0.07  0.83, p < 0.05). High urbanization with riparian buffers (managed high in the map) produces the same values (I ¼ 0.34) regardless of climate scenario. Climate scenarios without new urban land cover show variable influence depending on the climate scenario (low: I ¼ 0.46, medium: I ¼ 0.67, high: I ¼ 0.83). These examples show how the interplay between the two variables can lead to either localized or more global effects. Projected increases in soil mobilization are largest in the medium climate scenarios. The Miroc5 climate model displays as much as a 60 mm average

408

R. Hoyer, H. Chang / Applied Geography 53 (2014) 402e416

Fig. 3. Scenario maps depicting percent change in water yield estimates modeled by InVEST's water yield model from a historic baseline (1981e2012) to potential future (2036e2065) conditions. The left column depicts the baseline in absolute terms the current land cover and modeled historic thirty year average climate. The middle two columns show change under the two urbanization scenarios and modeled future climates. The right column displays a “managed high” scenario where the 30 m riparian buffer strip is also present along with high urbanization and modeled future climates.

increase in monthly winter precipitation from the historic to future time period (Fig. 2), leading to the highest projected erosivity rate. The same explanation can be applied to the other climate scenarios but to a lesser extent as there is less increase or even decreases in precipitation in some months. As with the nutrient modeling, riparian buffer strips reduce sediment exports. For the medium climate scenario, soil loss potential through mobilization is great enough that the buffers can only mitigate the effects of a wetter winter in the future. Sediment retention shows a different pattern than exports (Fig. 7, upper left corner). This is supported by a weak negative correlation between retention and exports (r ¼ 0.50  0.41, p < 0.001). Spatial autocorrelation in projected differences are more clearly driven by the future climate scenario (low and medium: I ¼ 0.54  0.66, high: I ¼ 0.79  0.82, p < 0.001, polygon contiguity). Those using historic climate show weaker autocorrelation (I ¼ 0.13  0.22, p < 0.001). The maps for export and retention suggest a broad trend of increases for both except for exports when riparian buffers are simulated. The autocorrelation evidence suggests that underlying this is a small but significant spatial divergence based on climate, which can be related to the spatial patterns of water yield maps under different scenarios (Fig. 3). So as seen in the export results, the temporal distribution of rainfall under each climate scenario affects erosive potential, and the spatial

distribution of erosion potential influences where those changes in retention are slightly more pronounced. Ecosystem bundling When the sub-watershed estimates are normalized and bundled, there are small changes in the ranking of each per scenario (Fig. 8). The Yamhill basin contains the majority of subwatersheds providing the highest levels of bundled services under the weights used. Slight shifts do occur where a few of the watersheds in Yamhill increase in bundled ES with installation of riparian buffers. The upland portions of the study area have a higher bundled value particularly the two sub-watersheds in the southwest of the Yamhill basin due to high water yield estimates that are also weighted at double the other freshwater ES. InVEST sensitivity All available parameters in InVEST display some degree of sensitivity. For water yield, most parameters are set to relatively high values. This is supported by the sensitivity result showing further reduction in all three parameters e the evapotranspiration coefficient, root depth, and the Zhang seasonality coefficient e all increase water yield (Table 4). In a system where peak rainfall and peak potential evapotranspiration are largely out of phase like in

R. Hoyer, H. Chang / Applied Geography 53 (2014) 402e416

409

Fig. 4. Scenario maps depicting change in total phosphorus (TP) export from runoff estimates modeled by InVEST's water purification model from a historic baseline (1981e2010) to potential future (2035e2065) conditions. The left column depicts the baseline in absolute terms the current land cover and modeled historic thirty year average climate. The middle two columns show change under the two urbanization scenarios and modeled future average climate. The right column displays a “managed high” scenario where the 30 m riparian buffer strip is also present along with high urbanization and modeled future average climate.

our study area, it suggests water storage in soil and vegetation with a relatively high transpiration rates help explain long term annual average water yield. The water purification model sensitivity tests support the observed scenario map patterns. As exports coefficients are reduced both estimated exports by streamflow and retention are reduced and vice versa (Table 5). An export coefficient needs to be increased a substantial amount in order to see a much smaller response in export estimates. The effective retention percentage has a symmetrical effect on exports and retention rates. Manipulating the flow threshold effectively works the same way as the effective retention parameter. This is inadvisable as it should reflect the stream network in the study area. The sediment retention estimates are more sensitive than the export estimates (Table 6). C and P factor show a great degree of sensitivity. Their behavior depends on the starting calibration parameter value. C factor displays the largest fluctuation because of this. The effective retention parameter exhibits similar behavior to the water purification model. Its relatively small effect on retention rates suggests that it is in fact less important than the parameters that are terms in USLE. The length slope factor determines when one of two equations in the model is used for determining slope factor in USLE (Tallis et al, 2013). Tests suggest that it is a major determinant in retention estimates.

Discussion Mapping freshwater ES with InVEST Our scenario maps offer a potential tool for determining what areas in the study area are most sensitive to potential LULC and climate change. It is clear climate drives the annual average supply of water, a result echoing those of studies using process-based hydrologic models in our study area and elsewhere (Castillo, Güneralp, & Güneralp, 2014; Franzcyk & Chang, 2009; Praskievicz & Chang, 2011; Tong, Sun, Ranatunga, He, & Yang, 2012). Nutrient export and water purification service changes are driven by LULC change (Chang, 2004). Another InVEST freshwater study did display some small adjustments to nutrient exports and retention for dry and wet years (Terrado et al., 2014). We hypothesize that differences among the characteristics of each climate scenario were not sufficient to alter the hydrologic sensitivity score for each scenario run of the water purification tool. New urban land cover is typically within sub-watersheds that are already urbanized to a large extent. This explains the little to no locational shift in the sub-watersheds providing the most freshwater ES embodied by the stability observed in the normalized and bundled scenario maps (Fig. 8). The change maps also suggest a relatively stable landscape in terms of ES provision with deviations typically less than a third of current estimates. An exception occurs at the mouth of the Tualatin basin.

410

R. Hoyer, H. Chang / Applied Geography 53 (2014) 402e416

Fig. 5. Scenario maps depicting change in total phosphorus (TP) retention by land cover estimates modeled by InVEST's water purification model from a historic baseline (1981e2010) to potential future (2035e2065) conditions. The left column depicts the baseline in absolute terms the current land cover and modeled historic thirty year average climate. The middle two columns show change under the two urbanization scenarios and modeled future average climate. The right column displays a “managed high” scenario where the 30 m riparian buffer strip is also present along with high urbanization and modeled future average climate.

However, when projected rainfall substantially increased over a single season like in the Miroc5 medium scenario, large changes do occur for sediment export estimates where certain sub-watersheds are projected to see a doubling or more. In InVEST, the actual ecosystem service result of nutrient or sediment retention also appears to be highly contingent on how much exports are coming from upslope. For our study area, this relationship between export and retention estimates leads to the result where sub-watersheds with the highest exports also contain the highest amounts of ecosystem services most of which are located in the Yamhill basin (Fig. 8). InVEST characterizes these services as retention of previously mobilized nutrients and sediments. It does not explicitly consider a land cover type's ability to hold them in situ as a service. It is therefore recommended to use both export and retention maps in tandem since they represent total nutrient/sediment production when interpreting results. This suggestion is also furnished by Terrado et al. (2014). InVEST utility and implications for land management The spatially-explicit outputs produced by InVEST offer many useful applications. The modeling framework allows for a level of validation in terms of ES mapping. More akin to calibration unlike process-based hydrologic modeling, the values created in the InVEST modeling effort are adjusted to reflect field observations.

This is a critique in the majority of ES mapping projects that lack field observations (Martínez-Harms & Balvanera, 2012; Sch€ agner et al., 2013; Seppelt, Dormann, Eppink, Lautenbach, & Schmidt, 2011). Although fine scale location of management priorities is outside InVEST's scope, overall patterns of ES on the landscape can be revealed, and thus aid in allocation of conservation resources. Using the bundle maps as evidence, the Yamhill basin provides a greater level of freshwater ES than the Tualatin (Fig. 8). This would increase with riparian buffer installation. Therefore, InVEST can serve as a scoping tool and first assessment of an area's ES profile. Sub-watersheds can be prioritized for further study or restoration. In this case, we suggest more examination of the potential for restoration in the Yamhill basin as it may have more impact on Willamette River downstream water quality issues. The large export increase projected at the Tualatin's mouth also warrants further study considering its potential for increased urban land cover (Hoyer & Chang, 2014). InVEST is a landscape planning tool and decision support system. It allows for the assessment of trade-offs among ES, and offers methods to capture a portion of their monetary value. From this perspective, it can offer much to those faced with land use decisions (Goldstein et al., 2012). Our findings suggest that riparian vegetation improves water quality or mitigates potential declines caused by increased urbanization. Although not necessarily surprising, we now have a spatially explicit estimate of the areal increase needed

R. Hoyer, H. Chang / Applied Geography 53 (2014) 402e416

411

Fig. 6. Scenario maps depicting percent change in sediment export from runoff estimates modeled by InVEST's sediment retention model from a historic baseline (1981e2010) to potential future (2036e2065) conditions. The left column depicts the baseline in absolute terms the current land cover and modeled historic thirty year average climate. The middle two columns show change under the two urbanization scenarios and modeled future average climate. The right column displays a “managed high” scenario where the 30 m riparian buffer strip is also present along with high urbanization and modeled future average climate.

to achieve the corresponding social benefit of improved or maintained water quality. Future research can proceed with spatially explicit estimation of the value of the ES benefits versus the costs associated with giving up that land's agricultural and residential value. Johnson, Polasky, Nelson, and Pennington (2012) showed that the benefits from increases in riparian vegetation in the Minnesota River basin outweighed the losses incurred to returns from agricultural land. They highlighted that uncertainties in the value of ES and of commodity prices can lead to a change in the balance among tradeoffs. This is an analytical nuance where InVEST's ability to produce spatially explicit provision of ES estimates is extremely useful. Economic returns from land are also spatially variable. This will allow for a spatial targeting of where investment in enhancing ES provision is most cost effective. In our case, the result of the individual ES maps along with all three bundled suggest targeting sub-watersheds in the Yamhill basin for riparian restoration and ES enhancement would lead to the most gains in water quality in relation to the entire Willamette Basin. Will targeting these areas also be the most cost effective areas as well? An assessment tool like InVEST can potentially help address the problem of determining if the conditions of a PES scheme are being met, and transition from being input-based to outcome-based, a common characteristic of current schemes focused on freshwater ES (Martin-Ortega, Ojea, & Roux, 2013). It can also aid in locating where benefits to sellers most outweigh opportunity costs.

Potential sources of uncertainty in ES mapping and modeling Spatially explicit ES assessments contain uncertainty. The fact that ES are a product of a complex system evaluated with imperfect data and imperfect tools must be acknowledged and clearly communicated by the producer of an assessment to the user of the information. Hou, Burkhard, and Müller (2013) point to original input data as the major source of uncertainty in ES assessments. Additionally, error is present in the assessment model because of its explicit assumptions and incomplete knowledge of the system it simulates. In light of the concerns associated with ES assessment, we discuss the three potential sources of ES assessment error e input data used, model parameters, and model structure. InVEST requires two main input data e climate and land cover, which are subject to error. First, land use and land cover (LULC) data are not perfectly accurate. Classification of landscapes constitutes a major source of uncertainty in any ecological assessment (Hou et al., 2013). While the general accuracy of NLCD is high, it contains error especially in grass dominated categories, and accuracy assessment procedures are still developing (Wickham et al., 2013). Second, the calibration climate dataset exhibits high correlation with field observations and minimal bias, but local scale effects still lead to error in certain climatic variables (Abatzoglou, 2013), which could affect yield estimates for some sub-watersheds. While InVEST water models have only

412

R. Hoyer, H. Chang / Applied Geography 53 (2014) 402e416

Fig. 7. Scenario maps depicting percent change in sediment retention by land cover estimates modeled by InVEST's sediment retention model from a historic baseline (1981e2010) to potential future (2036e2065) conditions. The left column depicts the baseline in absolute terms the current land cover and modeled historic thirty year average climate. The middle two columns show change under the two urbanization scenarios and modeled future average climate. The right column displays a “managed high” scenario where the 30 m riparian buffer strip is also present along with high urbanization and modeled future average climate.

a few parameters, as shown in sensitivity analysis, the model outputs are sensitive to all of them. Considering the lack of knowledge or a support system to access parameter values, parameterization is necessarily user-defined and a potential source of error (Bagstad, Semmens, Waage, & Winthrop, 2013). For example, in the sediment retention model, the erosivity variable does allow for climatic influence, which in our analysis led to increases in exports and retention. The sensitivity analysis reveals very large responses in the retention estimate to small changes in several of the parameters, making its final estimate uncertain. This illustrates the problem of transferring the USLE method at the landscape scale. Currently, there is no InVEST output such as an uncertainty bound for estimates, which would be a useful feature in future versions. A quantitative evaluation of uncertainty would aid in a check of the reliability of the results to potential users of the assessment. One study developed a quantitative sensitivity procedure for the water yield model. Its approach shows that the yield estimate's precision will be highly contingent on the accuracy of the precipitation and potential nchez-Canales et al., 2012). Future evapotranspiration data (Sa climate uncertainty is one reason for our using several scenarios. There is yet to be published a procedure for assessing InVEST output uncertainty caused by the user-defined parameters. An ES assessment with InVEST is geared toward long term average conditions. The simplicity of its structure makes inter-

annual variability difficult to model, and is not equipped to provide estimates at the seasonal, monthly, or daily time-step. More time scale sensitive ES like flood regulation are not addressed. Low flow water quality is already an issue in our study area (Boeder & Chang, 2008; Chang & Lawler, 2011; Kelly, Lynch, & Rounds, 1999), and InVEST cannot provide a seasonal or monthly distribution of nutrient loads. This leads to a potential scale mismatch between management decisions and InVEST outputs. A more temporally disaggregated modeling framework would necessitate incorporating the effects of the amount of runoff on water quality. In our case, results suggest a lack of sensitivity to climate input in the water purification tool (Table 2), and portions of the study area with no land cover changes shows little to no change in exports or loads (Figs. 4e7), so LULC being the major determinant of loadings is a critical assumption. Although the coefficient modeling approach proved valid previously, it required a high level of data collection in specific basins for several LULC sub-categories (Johnes, 1996). Doing similar work for an InVEST analysis would be a timeconsuming process. So on one hand, the modeling framework itself has simplifications that make it difficult to claim the simulation of important natural processes within the model captures the full complexity inherent in the system. On the other hand, these simplifications make it easier to approach, and can be viewed as a strength when data availability is limited (Vigorstol & Aukema, 2011).

R. Hoyer, H. Chang / Applied Geography 53 (2014) 402e416

413

Fig. 8. Scenario maps depicting bundled freshwater ecosystem services produced by InVEST's freshwater models from a historic baseline (1981e2010) to potential future (2036e2065) conditions representing combinations of varying climate change and low and high urbanization as well as high urbanization with a 30 m riparian buffer in large portions of the study area. Each estimate is normalized from 0.1 to 0.9, and then a weighted average is taken. The bundled services include water yield (40%), phosphorus retention (20%), nitrogen retention (20%), and sediment retention (20%).

Conclusions Using the modeling toolset InVEST offers several insights into the possible response of water-related ecosystem services in urbanizing basins. Our scenario analysis is based on a few simple assumptions, but it incorporates both land cover and climate change into the assessment albeit as independent variables rather than being dynamically interrelated. Using both of these variables is not common in the ES mapping literature. We were able to map the freshwater ES in the study area at the landscape scale and gain insights into ES assessment research, and could inform land management decisions in the study area. Water yields are projected to modestly increase in the low and medium climate scenarios and slightly in the high climate scenario. This is driven mainly by climate change with urbanization and riparian buffer installation playing a small role. Nutrient exports and retention respond almost exclusively to land cover with high sensitivity to the export coefficients explaining changes in the scenarios. Loss of natural vegetation increases exports moderately and reduces retention. Response to loss of agricultural land varies by location. Sediment exports and retention are influenced by land cover and climate. The projected increase in winter rainfall leads to higher erosivity rates and is the main driver in a near across the board increase for both. The simulated management strategy of riparian buffer construction reduced exports and increased retention rates in sub-watersheds where they are placed.

The spatial distribution of freshwater ES remains relatively stable at the sub-watershed scale. The lowland areas are projected to have more change in water yield than the upland portions of the study area. Nutrient exports are projected to increase in response to increased urban development in the Tualatin Basin. This is reversed in the Yamhill since urban land cover replaces agricultural lands with much higher nutrient exports. When the estimates are

Table 4 Sensitivity analysis of parameters in the InVEST water yield model. Calibrated model for the Tualatin River at the mouth was the test basin. Parameters were adjusted to amount shown until the minimum or maximum was reached. Parameter

Adjustment

Response (mm)

Evapotranspiration coefficient (etk) Min. 0.001eMax. 1.0

Down 0.2 Down 0.1 Down 0.05 Up 0.1 Up 0.05 Down 2000 Down 1000 Down 500 Up 2000 Up 1000 Up 500 9 5 3

95.8 47.4 23.6 45 22.7 32 15.8 0 0.8 0.8 0.8 3.2 22.5 40.5

Root Depth in mm Min. 10eMax. 7000

Zhang coefficient (set at 10 in calibration) Min. 1eMax. 10

414

R. Hoyer, H. Chang / Applied Geography 53 (2014) 402e416

Table 5 Sensitivity analysis of parameters in the InVEST water purification model. Calibrated model for the Tualatin River at the mouth was the test basin. Parameters were adjusted to amount shown until the minimum or maximum was reached. Parameter

Adjustment

Export response (kg/ha)

Retention response (kg/ha)

Land cover export coefficient (kg/ha) Min. 0eMax. None

Down 5000 Down 2000 Down 1000 Up 5000 Up 2000 Up 1000 Down 25 Down 10 Down 5 Up 25 Up 10 Up 5 Down 800 Up 800

0.77 0.52 0.28 1.43 0.57 0.29 1.45 0.67 0.50 0.53 0.35 0.23 0.28 0.13

1.65 1.26 0.71 3.56 1.43 0.71 1.45 0.67 0.50 0.54 0.35 0.23 0.28 0.13

Effective retention (%) Min. 0eMax. 100

Flow threshold (calibrated model set at 1300) Min. 1eMax. None

bundled into a weighted average, there is very little change in spatial pattern of freshwater ES amongst the scenarios. It does reveal the Yamhill basin providing more ES overall than the Tualatin basin. InVEST is useful for ES assessments but with some limitations. The ecosystem service retention estimates go up with the increase in exports but do not account for the limits on the system's uptake potential. It contains a small number of parameters all of which display sensitivity, and have a large impact on the final ES estimate. With no recommended values, calibration can prove challenging and final estimates of ES contain uncertainty. However, it allows researchers to calibrate to data from the local area, a step uncommon in ES mapping projects. We argue this offers some credibility

Table 6 Sensitivity analysis of parameters in the InVEST sediment retention model. Calibrated model for the Tualatin River at the mouth was the test basin. Parameters were adjusted to amount shown until the minimum or maximum was reached. Parameter

Adjustment

Export response (MT/ha)

Retention response (MT/ha)

Universal soil loss equation (USLE) Crop (C) factor and management (P) factor Min. 0.001eMax. 1

C and P down 0.01 C and P up 0.01 C and P up 0.05 C down 0.01 C down 0.02 C up 0.01 C up 0.02 C up 0.05 P down 0.01 P down 0.02 P up 0.01 P up 0.02 P up 0.05 Down 25 Down 10 Down 5 Up 25 Up 10 Up 5 Down 20 Up 20

0.086 0.122 0.839 0.047 0.053 0.038 0.163 0.402 0.038 0.073 0.037 0.075 0.187 0.317 0.080 0.036 0.061 0.035 0.021 0.012 0.018

4.29 8.80 42.33 52.39 53.62 38.29 28.88 0.384 0.238 0.458 0.240 0.480 1.198 0.324 0.081 0.037 0.063 0.036 0.021 45.99 55.46

Down 800 Up 800

0.056 0.025

12.76 7.79

Effective retention (%) Min. 0eMax. 100

Length slope factor (calibrated model set at default of 75) Min. 1eMax. 100 Flow threshold (calibrated model set at 1300) Min. 1eMax. None

to the outputs especially in terms of the relative change exhibited in freshwater ES. Also being spatially explicit, InVEST can help managers gain a landscape scale picture of where in a management area provides the most ecosystem service benefit. Our analysis suggests a further study of the less managed Yamhill basin since targeting it for ES enhancement could potentially lead to the greatest gains in downstream water quality. This finding is a starting point toward elucidating the tradeoffs among regulating and provisioning ES in watersheds that have mixed land uses. Even with some assumptions, this analysis still can provide decision-relevant information and assist managers in understanding the potential patterns of freshwater ES of urbanizing basins under the dual pressures of climate change and land development. Acknowledgments This research was supported by a US National Science Foundation Grant #1226629 and by the Institute for Sustainable Solutions at Portland State University. We appreciate Natural Capital Project team members for providing some biophysical data and technical support for InVEST modeling. We would like to thank Clean Water Service Staff members for providing water quality data and stakeholder workshop participants for their invaluable comment. An anonymous reviewer provided comments that greatly improved the paper. Views expressed are our own and do not necessarily reflect those of sponsoring agencies. Appendix A. Supplemental information Supplemental information related to this article can be found online at http://dx.doi.org/10.1016/j.apgeog.2014.06.023. References Abatzoglou, J. T. (2013). Development of gridded surface meteorological data for ecological applications and modelling. International Journal of Climatology, 33, 121e131. http://dx.doi.org/10.1002/joc.3413. Abatzoglou, J. T., & Brown, T. J. (2012). A comparison of statistical downscaling methods suited for wildfire applications. International Journal of Climatology, 32, 772e780. http://dx.doi.org/10.1002/joc.2312. Arnold, J. G., Moriasi, D. N., Gassman, P. W., Abbaspour, K. C., White, M. J., Srinivasan, R., et al. (2012). SWAT: model use, calibration, and validation. Transactions of the American Society of Agricultural and Biological Engineers, 55(4), 1491e1508. Bagstad, K. J., Semmens, D. J., Waage, S., & Winthrop, R. (2013). A comparative assessment of decision-support tools for ecosystem services quantification and valuation. Ecosystem Services, 5, 27e39. http://dx.doi.org/10.1016/ j.ecoser.2013.07.004. Bagstad, K. J., Semmens, D. J., & Winthrop, R. (2013). Comparing approaches to spatially explicit ecosystem service modeling: a case study from the San Pedro River, Arizona. Ecosystem Services, 5, 40e50. http://dx.doi.org/10.1016/ j.ecoser.07.007. Bateman, I. J., Harwood, A. R., Mace, G. M., Watson, R. T., Abson, D. J., Andrew, B., et al. (2013). Bringing ecosystem services into economic decision-making: land use in the United Kingdom. Science, 341(6141), 45e50. http://dx.doi.org/ 10.1126/science.1234379. Beguería, S., & Vicente-Serrano, S. M. (2013). SPEI: Calculation of the standardized precipitation-evapotranspiration index. R package version 1.3 http://CRAN.Rproject.org/package¼SPEI. Bennett, E. M., Peterson, G. D., & Gordon, L. J. (2009). Understanding relationships among multiple ecosystem services. Ecology Letters, 12, 1394e1404. http:// dx.doi.org/10.1111/j.1461-0248.2009.01387.x. Boeder, M., & Chang, H. (2008). Multi-scale analysis of oxygen demand trends in an urbanizing Oregon watershed. Journal of Environmental Management, 87, 567e581. http://dx.doi.org/10.1016/j.jenvman.2007.12.009. Brauman, K. A., Daily, G. C., Duarte, T. K., & Mooney, H. A. (2007). The nature and value of ecosystem services: an overview highlighting hydrologic services. Annual Review of Environment and Resources, 32, 67e98. http://dx.doi.org/ 10.1146/annurev.energy.32.031306.102758. Budyko, M. I. (1974). Climate and life. New York: Academic Press. Burkhard, B., de Groot, R., Costanza, R., Seppelt, R., Jørgensen, S. E., & Potschin, M. (2012). Solutions for sustaining natural capital and ecosystem services. Ecological Indicators, 21, 1e6. http://dx.doi.org/10.1016/j.ecolind.2012.03.008.

R. Hoyer, H. Chang / Applied Geography 53 (2014) 402e416 Burkhard, B., Crossman, N., Nedkov, S., Petz, K., & Alkemade, R. (2013). Mapping and modelling ecosystem services for science, policy and practice. Ecosystem Services, 4, 1e3. http://dx.doi.org/10.1016/j.ecoser.2013.04.005. _ & Güneralp, B. (2014). Influence of changes in developed Castillo, C. R., Güneralp, I., land and precipitation on hydrology of a coastal Texas watershed. Applied Geography, 47, 154e167. http://dx.doi.org/10.1016/j.apgeog.2013.12.009. Chan, K. M. A., Shaw, M. R., Cameron, D. R., Underwood, E. C., & Daily, G. C. (2006). Conservation planning for ecosystem services. Plos Biology, 4, 2138e2152. http://dx.doi.org/10.1371/journal.pbio.0040379. Chang, H. (2004). Water quality impacts of climate and land use changes in southeastern Pennsylvania. The Professional Geographer, 56(2), 240e257. http:// dx.doi.org/10.1111/j.0033-0124.2004.05602008.x. Chang, H., & Jung, I.-W. (2010). Spatial and temporal changes in runoff caused by climate change in a complex large river basin in Oregon. Journal of Hydrology, 388(3e4), 186e207. http://dx.doi.org/10.1016/j.jhydrol.2010.04.040. Chang, H., & Lawler, K. (2011). Impacts of climate variability and change on water temperature in an urbanizing Oregon basin. In N. E. Peters, V. Krysanova, A. Lepisto, R. Prasad, M. C. Thoms, R. Wilby, & S. Zandaryaa (Eds.), Water quality: Current trends and expected climate impacts (pp. 123e128). Oxfordshire: IAHS Publication 348. Cochran, B., & Logue, C. (2011). A watershed approach to improve water quality: a case study of Clean Water Services' Tualatin River program. Journal of the American Water Resources Association, 47(1), 29e38. http://dx.doi.org/10.1111/ j.1752-1688.2010.0049.x. Costanza, R., D'Arge, R., de Groot, R., Farber, S., Grasso, M., Hannon, B., et al. (1997). The value of the world's ecosystem services and natural capital. Nature, 387, 253e260. Crossman, N. D., Burkhard, B., Nedkov, S., Willemen, L., Petz, K., Palomo, I., et al. (2013). A blueprint for mapping and modelling ecosystem services. Ecosystem Services, 4, 4e14. http://dx.doi.org/10.1016/j.ecoser.2013.02.001. Daily, G. C. (Ed.). (1997). Nature's services: Societal dependence on natural ecosystems. Washington D.C.: Island Press. Daily, G. C., Polasky, S., Goldstein, J., Kareiva, P. M., Mooney, H. A., Pejchar, L., et al. (2009). Ecosystem services in decision making: time to deliver. Frontiers in Ecology and the Environment, 7(1), 21e28. http://dx.doi.org/10.1890/080025. Eade, J. D. O., & Moran, D. (1996). Spatial economic valuation: benefits transfer using geographic information systems. Journal of Environmental Management, 4, 91e110. http://dx.doi.org/10.1006/jema.1996.9990. Eigenbrod, F., Bell, V. A., Davies, H. N., Heinemeyer, A., Armsworth, P. R., & Gaston, K. J. (2011). The impact of projected increases in urbanization on ecosystem services. Proceedings of the Royal Society B, 278(1722), 3201e3208. http://dx.doi.org/10.1098/rspb.2010.2754. Egoh, B., Reyers, B., Rouget, M., Richardson, D., Lemaitre, D., & van Jaarsveld, A. S. (2008). Mapping ecosystem services for planning and management. Agriculture, Ecosystems & Environment, 127, 135e140. http://dx.doi.org/10.1016/ j.agee.2008.03.013. Engel, S., Pagiola, S., & Wunder, S. (2008). Designing payments for environmental services in theory and practice: an overview of the issues. Ecological Economics, 65, 663e674. http://dx.doi.org/10.1016/j.ecolecon.2008.03.011. ESRI (Environmental Science Research Institute). (2012). ArcGIS 10.1. Redlands, CA, U.S.A. Estoque, R. C., & Murayama, Y. (2012). Examining the potential impact of land use/ cover changes on the ecosystem services of Baguio city, the Philippines: a scenario-based analysis. Applied Geography, 35(1e2), 316e326. http:// dx.doi.org/10.1016/j.apgeog.2012.08.006. Franzcyk, J., & Chang, H. (2009). The effects of climate change and urbanization on the runoff of the Rock Creek basin in the Portland metropolitan area, Oregon, USA. Hydrological Processes, 23, 805e815. http://dx.doi.org/10.1002/hyp.7176. Fry, J., Xian, G., Jin, S., Dewitz, J., Homer, C., Yang, L., et al. (2011). Completion of the 2006 National Land Cover Database for the conterminous United States. Photogrammetric Engineering & Remote Sensing, 77(9), 858e864. Goldstein, J. H., Caldarone, G., Duarte, T. K., Ennaanay, D., Hannahs, N., Mendoza, G., et al. (2012). Integrating ecosystem-service tradeoffs into land-use decisions. Proceedings of the National Academy of Sciences of the United States of America, 109(19), 7565e7570. http://dx.doi.org/10.1073/pnas.1201040109. de Groot, R. S., Wilson, M. A., & Boumans, R. M. (2002). A typology for the classification, description and valuation of ecosystem functions, goods and services. Ecological Economics, 41(3), 393e408. http://dx.doi.org/10.1016/S0921-8009(02) 00089-7. Hou, Y., Burkhard, B., & Müller, F. (2013). Uncertainties in landscape analysis and ecosystem service assessment. Journal of Environmental Management, 127, S117eS131. http://dx.doi.org/10.1016/j.envman.2012.12.002. Hoyer, R. W., & Chang, H. (2014). Development of Future Land Cover Change Scenarios in the Metropolitan Fringe, Oregon, U.S., with Stakeholder Involvement. Land, 3(1), 322e342. doi:10.3390yland3010322. Jack, B. K., Kousky, C., & Sims, K. R. E. (2008). Designing payments for ecosystem services: lessons from previous experience with incentive-based mechanisms. Proceedings of the National Academy of Sciences of the United States of America, 105(28), 9465e9470. http://dx.doi.org/10.1073/pnas.0705503104. Johnes, P. J. (1996). Evaluation and management of the impact of land use change on the nitrogen and phosphorus load delivered to surface waters: the export coefficient modelling approach. Journal of Hydrology, 183, 323e349. http:// dx.doi.org/10.1016/0022-1694(95)02951-6. Johnson, K. A., Polasky, S., Nelson, E., & Pennington, D. (2012). Uncertainty in ecosystem services valuation and implication for assessing land use tradeoffs:

415

an agricultural case study in the Minnesota River basin. Ecological Economics, 79, 71e79. doi:10.106/j.ecolecon.2012.04.020. Kareiva, P., Tallis, H., Ricketts, T. H., Daily, G. C., & Polasky, S. (Eds.). (2011). Natural capital: Theory & practice of mapping ecosystem services. Oxford: Oxford University Press. Kelly, V. J., Lynch, D. D., & Rounds, S. A. (1999). Sources and transport of phosphorus and nitrogen during low-flow conditions in the Tualatin River, Oregon, 1991e1993 (United States Geological Survey Water-Supply Paper 2465-C) http://pubs.er. usgs.gov/publication/wsp2465C. Lautenbach, S., Kegel, C., Lausch, A., & Seppelt, R. (2011). Analysis of historic changes in regional ecosystem service provisioning using land use data. Ecological Indicators, 11, 676e687. http://dx.doi.org/10.1016/j.ecolind.2010.09.007. MA (Millennium Ecosystem Assessment). (2005). Ecosystems and human well-being. Washington, DC: Island Press. Martin-Ortega, J., Ojea, E., & Roux, C. (2013). Payments for water ecosystem services in Latin America: a literature review and conceptual model. Ecosystem Services, 6, 122e132. http://dx.doi.org/10.1016/j.ecoser.2013.09.008. Martínez-Harms, M. J., & Balvanera, P. (2012). Methods for mapping ecosystem service supply: a review. International Journal of Biodiversity Science, Ecosystem Services & Management, 8(1e2), 17e25. http://dx.doi.org/10.1080/ 21513732.2012.663792. Naidoo, R., & Ricketts, T. H. (2006). Mapping the economic costs and benefits of conservation. PLoS Biology, 4(11), e360. http://dx.doi.org/10.1371/ journal.pbio.0040360. Nearing, M. A. (2001). Potential changes in rainfall erosivity in the U.S. with climate change during the 21st century. Journal of Soil and Water Conservation, 56(3), 229e232. Nelson, E., Mendoza, G., Regetz, J., Polasky, S., Tallis, H., Richard, D., et al. (2009). Modeling ecosystems services, biodiversity conservation, commodity production, and tradeoffs at landscape scales. Frontiers in Ecology and the Environment, 7(1), 4e11. http://dx.doi.org/10.1890/080023. Nelson, E., Sander, H., Hawthorne, P., Conte, M., Ennaanay, D., Wolny, S., et al. (2010). Projecting global land-use change and its effect on ecosystem service provision and biodiversity with simple models. PLoS ONE, 5(12), e14327. http://dx.doi.org/ 10.1371/journal.pone0014327. Netusil, N., Kincaid, M., & Chang, H. (2014). Valuing water quality in urban watersheds: a comparative analysis of Johnson Creek, Oregon and Burnt Bridge Creek, Washington. Water Resources Research. http://dx.doi.org/10.1002/ 2013WR014546. ODEQ (Oregon Department of Environmental Quality). (2006). Willamette Basin TMDL: Chapter 1: Overview. Retrieved from http://www.deq.state.or.us/wq/ tmdls/docs/willamettebasin/willamette/chpt1overview.pdf Accessed 04.11.13. ODEQ (Oregon Department of Environmental Quality). (2012). Tualatin subbasin total maximum daily load and water quality management plan: Chapter 1 e Overview and background. Retrieved from http://www.deq.state.or.us/wq/ tmdls/docs/willamettebasin/tualatin/revision/Ch0CoverExecSummary.pdf Accessed 03.03.13. Patel, M., Kok, K., & Rothman, D. S. (2007). Participatory scenario construction in land use analysis: an insight in the experiences created by stakeholder involvement in the northern Mediterranean. Land Use Policy, 24, 546e561. http://dx.doi.org/10.1016/j.landusepol.2006.02.005. Peterson, G. D., Cumming, G. S., & Carpenter, S. R. (2003). Scenario planning: a tool for conservation in an uncertain world. Conservation Biology, 17(2), 366e385. http://dx.doi.org/10.1046/j.1523-1739.2003.01491.x. Polasky, S., Nelson, E., Pennington, D., & Johnson, K. A. (2011). The impact of landuse change on ecosystem services, biodiversity and returns to landowners: a case study in the state of Minnesota. Environmental Resource Economics, 48, 219e242. http://dx.doi.org/10.1007/s10640-010-9407-0. Portman, M. E. (2013). Ecosystem services in practice: challenges to real world implementation of ecosystem services across multiple landscapes e a critical review. Applied Geography, 45, 185e192. http://dx.doi.org/10.1016/j.apgeog.2013.09.011. Postschin, M. B., & Haines-Young, R. (2011). Ecosystem services: exploring a geographical perspective. Progress in Physical Geography, 35(5), 575e594. http:// dx.doi.org/10.1177/0309133311423172. Praskievicz, S., & Chang, H. (2011). Impacts of climate change and urban development on water resources in the Tualatin River basin, Oregon. Annals of the Association of American Geographers, 101(2), 249e271. http://dx.doi.org/10.1080/ 00045608.2010.544934. Qiu, J., & Turner, M. G. (2013). Spatial interactions among ecosystem services in an urbanizing agricultural watershed. Proceedings of the National Academy of Sciences of the United States of America, 110(29), 12149e12154. http://dx.doi.org/ 10.1073/pnas.1310539110. R Core Team. (2013). R: A language and environment for statistical computing. Vienna, Austria: R Foundation for Statistical Computing. http ewww.R-project.org/. Raudsepp-Hearne, C., Peterson, G. D., & Bennet, E. M. (2010). Ecosystem service bundles for analyzing tradeoffs in diverse landscapes. Proceedings of the National Academy of Sciences of the United States of America, 107(11), 5242e5247. http://dx.doi.org/10.1073/pnas.9097284107. Reckhow, K. H., Beaulac, M. N., & Simpson, J. T. (1980). Modeling phosphorus loading and lake response under uncertainty: A manual and compilation of export coefficients. Washington, D. C.: U. S. Environmental Protection Agency. http:// nepis.epa.gov/Exe/ZyPDF.cgi?Dockey¼00001KWE.PDF. Renard, K. G., & Freimund, J. R. (1994). Using monthly precipitation data to estimate the R factor in the revised USLE. Journal of Hydrology, 157, 287e306. http:// dx.doi.org/10.1016/0022-1694(94)90110-4.

416

R. Hoyer, H. Chang / Applied Geography 53 (2014) 402e416

Renard, K. G., Foster, G. R., Weesies, G. A., McCool, D. K., & Yoder, D. C. (1997). Predicting soil erosion by water: A guide to conservation planning with the Revised Soil Loss Equation (RUSLE) (U.S. Department of Agriculture, Agriculture Handbook No. 703) http://www.epa.gov/npdes/pubs/ruslech2.pdf. Ruckelshaus, M., McKenzie, E., Tallis, H., Guerry, A., Daily, G., Kareiva, P., et al. (2013). Notes from the field: lessons learned from using ecosystem service approaches to inform real-world decisions. Ecological Economics,. http://dx.doi.org/10.1016/ j.ecolecon.2013.07.009. Runkel, R. L., Crawford, C. G., & Cohn, T. A. (2004). Load Estimator (LOADEST): A Fortran program for estimating constituent loads in streams and rivers (US Geological Survey Techniques and Methods: 4-A.) http://pubs.usgs.gov/tm/ 2005/tm4A5/. ~ a, V., et al. S anchez-Canales, M., Benito, A. L., Passuello, A., Terrado, M., Ziv, G., Acun (2012). Sensitivity analysis of ecosystem service valuation in a Mediterranean watershed. Science of the Total Environment, 440, 140e153. http://dx.doi.org/ 10.1016/j.scitotenv.2012/07.071. €gner, J. P., Brander, L., Maes, J., & Hartje, V. (2013). Mapping ecosystem services' Scha values: current practice and future prospects. Ecosystem Services, 4, 33e46. http://dx.doi.org/10.1016/j.ecoser.2013.02.003. Seppelt, R., Dormann, C. F., Eppink, F. V., Lautenbach, S., & Schmidt, S. (2011). A quantitative review of ecosystem service studies: approaches, shortcomings and the road ahead. Journal of Applied Ecology, 48, 630e636. http://dx.doi.org/ 10.1111/j.1365-2664.2010.01952.x. Singh, S., & Chang, H. (2014). Effects of land cover change on water quality in urban streams at two spatial scales. International Journal of Geospatial and Environmental Research, 1. Su, S., Xiao, R., Jiang, Z., & Zhang, Y. (2012). Characterizing landscape pattern and ecosystem service value changes for urbanization impacts at an eco-regional scale. Applied Geography, 34, 295e305. http://dx.doi.org/10.1016/ j.apgeog.2011.12.001. Tallis, H. T., Ricketts, T., Guerry, A. D., Wood, S. A., Sharp, R., Nelson, E., et al. (2013). InVEST 2.5.3 user's guide. The Natural Capital Project, Stanford http://www. naturalcapitalproject.org. Taylor, K. E., Stouffer, R. J., & Meehl, G. A. (2012). An overview of the CMIP5 and the experiment design. Bulletin of the American Meteorological Society, 93(4), 485e498. http://dx.doi.org/10.1175/BAMS-D-11-00094.1.

~ a, V., Ennaanay, D., Tallis, H., & Sabater, S. (2014). Impact of climate Terrado, M., Acun extremes on hydrological ecosystem services in a heavily humanized Mediterranean basin. Ecological Indicators, 37(A), 199e209. http://dx.doi.org/10.1016/ j.ecolind.2013.01016. Thompson, J. R., Wiek, A., Swanson, F. J., Carpenter, S. R., Hollingsworth, T., Spies, T. A., et al. (2012). Scenario studies as a synthetic and integrative research activity for long-term ecological research. BioScience, 62(4), 367e376. http:// dx.doi.org/10.1525/bio.2012.62.4.8. Tong, S. T. Y., Sun, Y., Ranatunga, T., He, J., & Yang, Y. J. (2012). Predicting plausible impacts of sets of climate and land use change scenarios on water resources. Applied Geography, 32(2), 477e489. http://dx.doi.org/10.1016/j.apgeog.2011.06.014. Troy, A., & Wilson, M. (2006). Mapping ecosystem services: practical challenges and opportunities in linking GIS and value transfer. Ecological Economics, 60(2), 435e449. http://dx.doi.org/10.1016/j.ecolecon.2006.04.007. USGS (US Geological Survey). (2013a). National Hydrography Dataset Plus Version 2 [Data]. Pacific Northwest. Retrieved from http://www.horizon-systems.com/ NHDPlus/NHDPlusV2_17.php. USGS (US Geological Survey). (2013b). National water information system [data]. USGS surface-water daily data for the nation. Retrieved from http://waterdata. usgs.gov/nwis/dv/?referred_module¼sw Accessed 23.09.13. Vigorstol, K. L., & Aukema, J. E. (2011). A comparison of tools for modeling freshwater ecosystem services. Journal of Environmental Management, 92, 2403e2409. http://dx.doi.org/10.1016/j.jenvman.2011.06.040. Wickham, J. D., Stehman, S. V., Gass, L., Dewitz, J., Fry, J. A., & Wade, T. G. (2013). Accuracy assessment of NLCD 2006 land cover and impervious surface. Remote Sensing of Environment, 130, 294e304. http://dx.doi.org/10.1016/ j.rse.2012.12.001. Wischmeier, W. H., & Smith, D. (1978). Predicting rainfall erosion losses: A guide to conservation planning (USDA e ARS Agriculture Handbook, Washington D.C). Zhang, L., Dawes, W. R., & Walker, G. R. (2001). Response of mean annual evapotranspiration to vegetation change at a catchment scale. Water Resources Research, 37(3), 701e708. http://dx.doi.org/10.1029/2000WR900325. Zheng, H., Robinson, B. E., Liang, Y., Polasky, S., Wang, F.-C., Ruckelshaus, M., et al. (2013). Benefits, costs, and livelihood implications of a regional payment for ecosystem service program. Proceedings of the National Academy of Sciences of the United States of America. http://dx.doi.org/10.1073/pnas.1312324110.