Understanding changes in ecosystem services at regional scales is increasingly important for environmental scientists, managers, planners, and decision makers as climate and land-use change continue (de Groot et al. 2010). Management decisions require an improved understanding of the drivers and processes that influence ecosystem services. Change in major environmental drivers, such as climate and land cover, typically result in large changes in ecosystem service supply (Schroeter et al. 2005). These changes are interactive and complex across space and time (Chen et al. 2013), requiring the development of appropriate methods to elucidate functional tradeoffs between management strategies. In order for stakeholders and citizens to be able to assess the value of the ecosystem services being provided, they need them to be expressed in terms of indicators that clearly relate to environmental condition (Nelson et al. 2009, Carpenter et al. 2015, Qiu and Turner 2015). Indicators of ecosystem services should be quantifiable, scalable (Bagstad et al. 2013a, Carpenter et al. 2015), explicit in time and space, and sensitive to land-cover or management change (Burkhard et al. 2012, van Oudenhoven et al. 2012). Moreover, the appropriate indicator is dependent on the method by which the ecosystem service is being valued (de Groot et al. 2010). Commonly used indicators that are easily monetizable may be incomplete for more comprehensive sociocultural preference valuations (Mavrommati et al. 2017).
Aquatic ecosystems are strongly influenced by terrestrial environments in their watersheds. Indicators of environmental condition related to climate and land cover can be generated at regional scales from existing meteorological observations, downscaled projections, land-cover atlases, or land-cover change scenarios (Queiroz et al. 2015). However, these simple evaluations are difficult in freshwater systems because responses depend on terrestrial and climate conditions integrated over entire watersheds and that vary over time. Terrestrial environments, interacting with climate, determine flow, temperature, and nutrient regimes in regional drainage networks (Poff et al. 1997). Hydrology is the overriding control of many aquatic ecosystem services (Brauman et al. 2007) influencing water availability, instream habitats, water temperatures, and nutrient fluxes. Aquatic ecosystems are further influenced by internal processes such as temperature re-equilibration and nutrient removal as water flows from upstream to downstream (Hale et al. 2014). To understand changes in ecosystem services, linked terrestrial and aquatic ecosystem models are needed to capture the processes defining responses to changing land cover and climate, partition the influence of each, and permit estimation of responses beyond previously observed ranges.
A variety of spatially explicit models or tools have been used for watershed-scale studies of environmental indicators and ecosystem services. Examples include Artificial Intelligence for Ecosystem Services (ARIES) (Villa et al. 2009), Multiscale Integrated Models of Ecosystem Services (MIMES) (Boumans et al. 2015), and Integrated Valuation of Ecosystem Services and Tradeoffs (InVEST) (Tallis and Polasky 2009, Bagstad et al. 2013b, Tallis et al. 2013). These models do not capture seasonal or subseasonal climate variability, which is projected to change regionally (Wood et al. 2002, Hayhoe et al. 2007, Horton et al. 2014) and is important for capturing watershed functions related to flood attenuation, water provisioning, river temperature regulation, and other ecosystem services (Vigerstol and Aukema 2011). To fully account for changes in ecosystem function associated with altered precipitation, temperature, and land-use and land-cover patterns, process-based models that incorporate key space- and time-varying hydrological and ecological processes are critical (Bagstad et al. 2013b).
Several processed-based terrestrial and/or aquatic biogeophysical models have recently been used for ecosystem service valuation (Logsdon and Chaubey 2012a, Bagstad et al. 2013a, Carpenter et al. 2015). The Variable Infiltration Capacity (VIC) model is a large-scale, semidistributed hydrological model (Liang et al. 1994) that simulated provisioning hydrological ecosystem services (Vigerstol and Aukema 2011) and flood regulation (Lee et al. 2015). The Soil Water Assessment Tool (SWAT), a process-based, spatially distributed hydrological and water quality model (Notter et al. 2012) was used to evaluate aquatic environmental variables, including water yield (Karabulut et al. 2015) and water quality (Logsdon and Chaubey 2012b). To compute dynamic ecosystem services in the agricultural Yahara watershed, a linked terrestrial–aquatic model used a process-based agroecosystem model (Agro-IBIS) (Soylu et al. 2014, Carpenter et al. 2015), a terrestrial hydrology model (THMB) (Coe 2000), a three-dimensional groundwater flow model (MODFLOW) (Harbaugh 2005), and a hydrological routing model (HYDRA), which lacked instream biogeochemistry (Coe 2000). There are a few robust examples of coupled human and biogeophysical models that have quantified ecosystem services at global (Boumans et al. 2002) and watershed (Costanza et al. 2002) scales. The Global Unified Metamodel of the Biosphere (GUMBO) simplifies several existing dynamic global models of both natural and social systems at an intermediate level of complexity and annual time-step (Boumans et al. 2002). The Patuxent Landscape Model (PLM; Costanza et al. 2002) is a spatially explicit process-based model that addresses the effects of both the magnitude and spatial patterns of human settlements and agricultural practices on hydrology, plant productivity, and nutrient cycling in the landscape, also at an annual time-step. Finally, valuation of ecosystem service information for land-use decisions has been estimated directly through an agent-based modeling framework (Heckbert et al. 2010, 2014, Groeneveld et al. 2017).
Our study contributes to the evolution of ecosystem service models by emphasizing shorter-term temporal dynamics in a spatially distributed and process-based framework that links terrestrial and aquatic function. This framework provides a new perspective for understanding the impacts that climate and land-cover and land-use change have on terrestrial and aquatic resources.
We present an approach that links time-varying (daily time-step) terrestrial and aquatic ecosystem models at regional scales and apply this model into the future using scenarios of climate and land cover to project changes in ecosystem services. First, we describe an indicator framework that succinctly represents a comprehensive suite of environmental conditions relevant to important ecosystem services. Second, we describe the linkage and validation of the terrestrial and aquatic ecosystem models to simulate aquatic indicators through the 21st century. We integrated the Photosynthetic Evapotranspiration-Carbon and Nitrogen (PnET-CN) forest ecosystem model (Ollinger et al. 2002, 2008, Aber et al. 2005) and the Framework for the Aquatic Modeling of the Earth System (FrAMES) aquatic ecosystem model (Wollheim et al. 2008a, b, Wisser et al. 2010, Stewart et al. 2011, 2013, Mineau et al. 2015; Zuidema, Wollheim, Mineau, et al., unpublished manuscript). These models integrate the dynamics of terrestrial and aquatic processes and linkages at daily time-steps, making them ideal for studying aquatic ecosystem responses in forest-dominated watersheds. In coordination with a separate effort described elsewhere in this special issue (Mavrommati et al. 2017) to assess the value of ecosystem services provided by the Upper Merrimack River watershed (UMRW) of New Hampshire, we contrast two extremes of projected futures in climate and land-cover change. The outcome suggests that climate change influences most indicators of environmental condition in the UMRW more than changes in land cover, although land cover has important interactive capacity to dampen or exacerbate the effects of the changing supply of ecosystem services in the future.
The central goal of this study was to develop estimates of ecosystem function critical to current and future watershed residents of the UMRW. We present a description and rationale for studying this watershed and the development of environmental indicators relevant to ecosystem services of local residents. We next describe the integrated terrestrial–aquatic model that projects aquatic environmental indicators, including scenarios, parameterizations, and data sets needed to project aquatic indicators to 2100. Finally, to understand how aquatic and terrestrial processes control the projected environmental conditions, we evaluate model sensitivity to a suite of climate and land-cover change (Thorn et al. 2017).
The UMRW is located in south-central New Hampshire, USA. The UMRW comprises 8,000 km² of New Hampshire and drains through a point just south of the city of Manchester, New Hampshire (Latitude: +43.6575, Longitude: -71.5005; Fig. 1). The watershed is currently home to 410,000 people, with ongoing population growth and land-cover change through extensive residential development, and is also an important tourist destination. Annual precipitation currently averages 1,100 mm yr-1 and is evenly distributed throughout the year. Mean annual temperature is currently 8.2°C, with mean annual minimums and maximums of -8.3°C and 22.5°C, respectively. Land cover consists of a mix of deciduous and evergreen forest (82%), urban (4.2%), agriculture (3.8%), wetland (5.9%), and open water (4.2%). As the dominant land cover, forests are a critical influence on the aquatic environment, including water supply, water quality, and aquatic habitat. The watershed sits at the boundary between strong and weak winters, and is expected to transition to greater and more variable precipitation and warmer temperatures in coming decades (Hayhoe et al. 2007, Burakowski et al. 2008, Wake et al. 2014). The area is also experiencing rapid population growth and ongoing residential development, which is expected to continue into the future (U.S. Environmental Protection Agency (USEPA) 2010). The Merrimack watershed is ranked in the top five watersheds nationally in terms of projected changes in water quality due to increase housing density on private forest lands and ranks at the top of the list in terms of private forest land projected to experience increased housing density in the USA (Stein et al. 2009). These changes are leading to increased water use, nitrogen (N) fluxes, chloride concentrations, and other changes. Thus, this watershed is an ideal location to focus efforts in understanding ecosystem service change.
Identifying clear and tangible environmental indicators is a necessary step to communicate the response of complicated human–environmental systems to change (Müller and Burkhard 2012) and to track the performance of environmental programs, regulations, and agencies (Boyd 2004). A multidisciplinary group consisting of ecologists, hydrologists, engineers, economists, and decision scientists developed a set of potential environmental indicators important to the study region. We used an iterative and collaborative process to distill a set of climate, terrestrial, and aquatic ecosystem variables to environmental indicators. Changes in these indicators in response to different climate and land-cover scenarios were then assigned a relative value by participants in a set of multicriteria decision workshops described elsewhere (Mavrommati et al. 2017, Murphy et al. 2017).
The indicators represented multidecadal average conditions in the UMRW for the present (1980–2005) and projected for the end of the 21st century (2070–2099). Candidate indicators had relevance to climate, land, and water domains. Terrestrial indicators were based on land-cover scenarios (Thorn et al. 2017) and the Northern Research Station Climate Change Atlas (Iverson et al. 2008). Climate indicators were based the Geophysical Fluid Dynamics Laboratory (GFDL) CM2.1 global climate model (GCM) simulations using the SRES A1Fi (higher) and B1 (lower) emissions scenarios (Nakicenvoic and Swart 2000) that were statistically downscaled using the asynchronous regional regression model (Stoner et al. 2012, Wake et al. 2014). The downscaled GFDL simulations were used as they provide a reasonable representation of climate across the northeast USA (Hayhoe et al. 2007). Given the multiple land-cover and emissions scenarios used in this study, we only had the capacity to use output from one GCM. We used downscaled global climate model simulations for the Franklin, New Hampshire meteorological station, centrally located within the study domain. Water-related environmental factors were derived at the watershed scale using the coupled terrestrial–aquatic process-based model to predict ecosystem function consistent with the land-cover and climate scenarios. Generating indicators based on model simulations at daily time-steps permits the clear definition from key variables such as concentrations, temperature, or flow volumes compared with key thresholds relevant at the scale of days. Translation of environmental indicators to indicators of ecosystem services and valuation of these services is discussed later in the special issue (Mavrommati et al. 2017).
We coupled the existing forest (PnET-CN) and aquatic (FrAMES) models to simulate hydrological and water quality characteristics related to ecosystem services, consistent with forest responses to climate, at regional scales. The detailed description of the individual PnET-CN and FrAMES models are presented in Appendix 1, sections 1.2 and 1.3. The PnET-CN model is widely used to simulate forest water, carbon (C), and N dynamics (Aber et al. 2005, Ollinger et al. 2008). The FrAMES model (Wollheim et al. 2008a, b, Wisser et al. 2010, Stewart et al. 2011, 2013) is a spatially distributed gridded river network model that has been applied extensively at various spatial scales ranging from watershed to global domain for simulating N dynamics in rivers (Wollheim et al. 2008a, b, Stewart et al. 2011), runoff/discharge dynamics (Vörösmarty et al. 1998, Wisser et al. 2010), river water temperature (Stewart et al. 2013), and chloride concentration (Zuidema, Wollheim, Mineau, et al., unpublished manuscript). Typically, FrAMES has a land-surface hydrology component that operates independently of forest dynamics. Here, we substitute PnET-CN predictions of runoff and N to load material from forests to river networks.
The PnET-CN model accounts for the influence of photosynthesis on evapotranspiration and nutrient uptake, forest age, and plant physiological responses (stomatal conductance) to changing CO2 (Ollinger et al. 2008). Together, these factors control C sequestration, nutrient export, and runoff generation immediately relevant to several ecosystem services. In the coupled model (Fig. 2), PnET-CN calculates daily runoff and dissolved inorganic nitrogen (DIN) flux from forest rooting zones. These outputs are then partitioned into shallow groundwater or surface (quick flow) flow paths, with different characteristic travel times. In urban regions, precipitation and snowmelt on hydrologically connected impervious areas run directly to the stream network, with the remainder infiltrating to lawn areas. Chloride, a potential stressor of aquatic biota, from snowmelt on the road-salt-treated fraction of impervious areas is transported conservatively following the soil and groundwater flow paths (Zuidema, Wollheim, Mineau, et al., unpublished manuscript). To link with the aquatic network, we also incorporated the role of terrestrial flow paths and riparian zones in regulating DIN loads. The PnET-CN model predicts leaching from the forest rooting zone. To account for retention along terrestrial–riparian flow paths, we applied a constant retention factor of 70% to all leachate, consistent with reactivity of riparian zones (Green et al. 2009) or buffers that average about 25 m (Mayer et al. 2005). We account for DIN loading from urban and agricultural areas using an empirical relationship between DIN, land use, and flow found in other New England watersheds (Wollheim et al. 2008a, b, Stewart et al. 2011, Mineau et al. 2015). Water temperature in terrestrial runoff is modeled as described in Stewart et al. (2013). Water temperature and DIN inputs from land are further modified by instream processes as water flow through the river network.
FrAMES routes water, DIN, water temperature, and chloride originating from forests and urban and agricultural land through the river network, accounting for additional point sources of contaminants, such as N in waste water treatment plant (WWTP) effluent and thermal loads from power plant cooling water processes (Stewart et al. 2013, Miara and Vörösmarty 2013). It also accounts for instream transformations during routing, such as dilution, denitrification (Wollheim et al. 2008b, Stewart et al. 2011), and instream temperature re-equilibration (Stewart et al. 2013). River discharge, solute masses, and thermal fluxes are propagated through the gridded river network using a linear reservoir routing scheme. Thus, PnET–FrAMES accounts for terrestrial–aquatic and upstream–downstream linkages to provide a mutually consistent climate, land, and water response in future scenarios. The model was first applied to historical conditions (2000–2015) to test against field observations using downscaled reanalysis meteorological observations and current land cover. Meteorological forcing for future scenarios used statistically downscaled climate data from Hayhoe et al. (2007), and land cover from Thorn et al. (2017). A detailed summary of the model input data for contemporary and future scenarios is described in Appendix 2.
Model parameters were specified a priori. For future scenarios, parameters that are responsive to human management were altered to be consistent with each scenario narrative (Table 1). Key parameters include the fraction of hydrological connected imperviousness (affecting flow regime, total runoff), road salt application rates (affecting water quality), suburban and agricultural DIN nonpoint loading, and WWTP DIN point loading (affecting N exports). We are using the term “land cover” in describing future changes to the landscape. Management decisions that are consistent with the land-cover narratives (e.g., stormwater and wastewater infrastructure) are included in the parameterization of FrAMES–PnET. Specific parameterizations and how they vary by scenario are provided in Appendix 3.
To validate the linked suite of models, we compared model simulations with data from headwater and downstream gauge locations to determine how well the spatial and temporal distribution of inputs, dilution, and processing are simulated. Observed data represented aquatic conditions at a total of 106 stations located throughout the Merrimack River and the neighboring Piscataqua River watersheds (Fig. 1). The latter is included in contemporary simulations to increase the pool of observations. Mean daily discharge data (1979–2013) were available from the U.S. Geological Survey (USGS) (n = 41 stations). Additional observed variables, including water temperature and concentrations of DIN and chloride came from two networks of in situ sensors: the High Intensity Aquatic Network (HIAN) (n = 12 stations) and Lotic Volunteer Temperature Electrical Conductivity and Stage (LoVoTECS) network (n = 53 stations). Data from the HIAN (Mulukutla et al. 2015) and LoVoTECS network (LoVoTECS 2012) are available through the Data Discovery Center (http://ddc.unh.edu) over the timeframe of 2012–2014 and are also reported in Contosta et al. (2017). We compared predictions to observations of both mean annual and mean summer conditions using contemporary land-cover and climate drivers. We calculated model prediction errors (MPE) and root mean square errors (RMSE) for temperature, runoff, and nitrate and chloride concentrations. We compared predictions in headwater streams (to test predictions of inputs from land) and along the Merrimack River main stem to test the regionalization and aquatic process component of the model.
The iterative and collaborative process of environmental indicator selection identified ten indicators, from an initial set of 43 (Table A2.1), to represent a comprehensive suite of environmental conditions for the UMRW. The experts group used the criteria that indicators should hold perceived relevance to the general public, should equitably represent the three domains, and should be constrained by existing modeling capability. The selected indicators represented three indicators for each of land and climate domains, and four indicators for the water domain. The indicators are defined in Table 2 and explained in detail in the supplementary material (Appendix 2). The ten selected indicators are similar to several included in comprehensive lists generated previously (de Groot et al. 2010, Burkhard et al. 2012). However, our final indicators differed from previous lists to be more relevant to local residents. For instance, we used total forest cover as a percentage of the watershed instead of wood biomass stock in units of mass per area (de Groot e al. 2010) based on lack of relevance of wood biomass to the general public living in the watershed. The water indicators emphasized basin scale and subannual estimates of water conditions, requiring space- and time-varying aquatic modeling. The choice of indicators in representing specific ecosystem services is discussed more thoroughly elsewhere in this special issue (Mavrommati et al. 2017).
The component ecosystem models used in this analysis have been previously validated individually. The PnET-CN model was validated for water and nutrient balance, (Ollinger et al. 2002, 2008, Aber et al. 2005), whereas FrAMES was validated for river discharge using a simple water balance model (Vörösmarty et al. 1998, Wisser et al. 2010), water temperature (Stewart et al. 2013), chloride (Zuidema, Wollheim, Mineau, et al., unpublished manuscript), and riverine DIN (Wollheim et al. 2008a, b, Stewart et al. 2011). In this study, terrestrial vegetation processes in PnET–FrAMES drive regional runoff, nutrient, chloride, and water temperature loading that is then routed through the network.
Regional predictions of water variables using PnET–FrAMES generally compared favorably to observations. Modeled annual runoff matches observations, with little bias indicated by a low median model percent error (MPE) of 4.4% (IQR = −5.7–11.7%) at gauging stations throughout the project domain (Fig. 3a). Simulated runoff was somewhat higher than observations during summer months (median MPE = 15.9%, IQR = −2.5–24.7%) (Fig. 3b). Thus, whereas water indicators based on flow (Table 2) are reasonably well predicted, the Fish Habitat Loss and Water Shortfalls indicators (which are triggered by low-flow conditions) are likely conservative estimates.
Modeled water temperature across gauging stations is low for mean annual temperature (median MPE = -17.5%, IQR= −32.7–−9.2%) with little bias during summer (median MPE = -2.3%, IQR= −8.1–5.5%) (Fig. 3b). Because the model does well during summer when temperatures are high, the water temperature component of the Aquatic Habitat indicator, which is mostly driven by summer conditions, is robust. No differences in bias were seen across different river sizes.
Modeled chloride concentrations are reasonable (median MPE = -4.1%, IQR= -37.8–28.2%) for annual and biased low for summer periods (median MPE = -23.2%, IQR= -47.0–-3.8% for annual and summer, respectively) (Fig. 3). The low summer bias occurs because of a flashier response to summertime diluting rains than observations (Zuidema, Wollheim, Mineau, et al., unpublished manuscript). Rectifying this incompatibility is a subject of ongoing model improvement. Thus, the indicator that uses chloride (Fish Habitat Loss) is likely an underestimate, particularly during summers.
The PnET–FrAMES model captures the range of variability in mean annual DIN concentration loaded via nonpoint inputs in headwater streams of different land cover, which reflect catchment loading prior to river processing. Predicted DIN loading concentrations are higher than observed for forested catchments (n = 4), but show little bias for developed catchments (n = 6) (Fig. 3; Figs. A4.1c, A4.2c). The PnET-CN model is known to overestimate forest DIN (Zhou, Ollinger, Glidden, et. al., unpublished manuscript), and continues to do so here, even though we added a flow path/riparian retention term to PnET–FrAMES. As a result of high bias in forested sites, the percent errors for DIN in Fig. 3 are higher than other variables. However, absolute error at these sites is very low compared with the more developed sites (Figs. A4.1c, A4.2c). Overall bias is low in summer (median MPE = -3.9%, IQR = −53.0–111.4%) but high annually (median MPE = 17.5%, IQR = −41.8–186.4%). Dissolved inorganic N concentrations in headwater streams reflect loading because aquatic net removal has not had the opportunity to impact concentrations due to short surface water flow paths in small catchments (Wollheim et al. 2006). Observations along the river main stem (basin profile, Fig. A4.3a) indicate the model reasonably represents regional loading, dilution, and aquatic N removal throughout the river network (Fig. A4.4a).
We characterized the suite of environmental responses to two land-cover and two climate scenarios selected for the valuation workshop (Mavrommati et al. 2017) meant to represent extremes of potential change in future conditions. We drove the model with the Backyard and Small Community Food Land-Cover scenarios (Table 1), with higher (A1Fi) and lower (B1) climate scenarios, respectively (Backyard/High emissions and Small Community Food/Low emissions).
Changes in the land indicators (Table 2; Agricultural Cover per person and Forest Cover) directly reflect the land-cover scenarios (Thorn et al. 2017) and have important direct ecosystem service implications. Briefly, in the Backyard scenario, watershed forest cover declines from 80% in 2010 to 60% in 2100, to accommodate increased rural and suburban development (Fig. 4). In the Small Community Food/Low scenario, forest declines to 64% in 2100, replaced by agricultural lands, primary for pasture and hay. By 2100, population increases by 170% in the Backyard/High scenario and declines by 5% in the Small Community Food/Low scenario. The increase in agricultural land relative to population under the Small Community Food/Low scenario (~1.0 acres per person in 2100 compared with 0.2 in 2010) increases the potential for per capita local food production compared with the contemporary period (Thorn et al. 2017). The Maple Presence indicator, which primarily represents the geographic distribution of suitable habitat for Acer saccharum (Iverson et al. 2008), declines from 49% of total forest cover in the present day to 27% and 31% in the Backyard/High emission and Small Community Food/Low emission scenarios, respectively. The declines result from shifts in climate associated with each land-cover scenario.
Projected changes in climate indicators in the UMRW are strongly determined by emission scenario. Present-day climate trends (Hodgkins and Dudley 2006, Hayhoe et al. 2007, Burakowski et al. 2008) continue through 2050 regardless of emission scenario (Fig. 5). Around 2050, the rate of warming is projected to increase rapidly under the Backyard/High emission scenario. With the high emission scenario, the hot days indicator increases sharply, averaging >45 d above 32°C after 2075 (Fig. 5a). Under the lower emission scenario, the multiyear mean for hot days is considerably lower (<17 d yr-1) (Fig. 5a). The Snow Days indicator declines under the higher emissions (averaging 18 d yr-1) compared with contemporary mean (60 d yr-1) and relative to the lower emissions scenario (47 d yr-1) (Fig. 5c). Again, the change is accelerated after 2040. The Recreation Days indicator changes little under both the low and high emission scenarios (Fig. 5b). Minimal changes in recreation days results from loss of recreation days due to days becoming too hot during the summer being offset by additional days in the shoulder seasons (spring and fall) that were previously too cool falling in the recreation day range.
Environmental indicators in the water domain reflect the combined influence of projected climate and land-cover change, as well as terrestrial and aquatic processes. In general, all water indicators reflect increasing degradation of aquatic ecosystem services for the Backyard/High scenario, but remain relatively unchanged for the Small Community Food/Low scenario (Fig. 6). The Fish Habitat Loss indicator is projected to rise substantially in the Backyard/High scenario to average about 40% of total stream and river length after 2075. Only slight increases, and fewer extreme events, occur under the Small Community Food/Low scenario. The Nitrogen Export indicator increases steadily over the 100-yr period for the Backyard/High scenario to 995 tons N yr-1 up 3100% from current conditions. Under the Community/Low scenario, however, N export changes relatively less, increasing only 70% to an average of 52 tons N yr-1. The steady increase in the Backyard/High scenario is due to elevated N loading associated with widespread land-cover change and increased population. The pattern in Community/Low occurs because in this scenario, agricultural management activities were assumed to reflect the greater concern for the environment (optimal fertilizer application rates, improved waste water treatment plant N removal), and because it reflects a lower human population (Table 1).
The Water Shortfall indicator increases in the future in Backyard/High (averaging 6×106 person days [pd] yr-1 after the 2075) but remains unchanged in the Community/Low scenario. The Water Shortfall indicator increases in Backyard/High despite greater runoff that occurs due to increased precipitation (1320 compared with 1270 mm yr-1 between the higher and lower emission scenarios, respectively) and lower forest evapotranspiration (ET) due to trees responding to increased atmospheric CO2 by increased water-use efficiency (Ollinger et al. 2002). The increase occurs because the greater population (in Backyard) reflects a greater demand that at times would be unmet.
The number of people at Flood Risk is expected to increase for both scenarios, although there is considerable interannual variability, which is expected because extreme flows only occur periodically. Total population affected after 2075 by extreme flows leading to Flood Risk is higher in the Backyard/High scenario (1.2×105 pd) than in Small Community Food/Low scenario (4.4×104 pd) because the former has a larger population to potentially be impacted, has higher flows on average, and has greater storm runoff due to increased imperviousness.
Overall, the suite of indicators suggests that relevant environmental conditions remain relatively stable until 2050 regardless of climate and land-cover scenario. Beginning in 2050, there is a potential for climate and water indicators to significantly deteriorate in the high emission scenario. The timing of the change in water indicators in Backyard/High suggests primarily a climate control, as climate changes also show rapid increases at this time (Fig. 5).
Climate drivers govern the hydrological and biogeochemical response of catchments, but these changes are mediated by land use and land cover through impacts of impervious surfaces, plant growth, evapotranspiration, and water residence time (Ollinger et al. 2002, Raymond and Saiers 2010, Pellerin et al. 2011, Goodridge and Melack 2012). The two scenarios described above convolve both signals (i.e., both climate and land cover differ in the two scenarios). The relative importance of climate, land cover, and interactions on the water domain indicators can be determined by analyzing results from all combinations of the suite of land-cover change scenarios (Table 1) and both C emission (climate) scenarios (Fig. 7).
The relative importance of climate and land cover varies among the indicators. For Fish Habitat Loss, climate dominates the response as the high emission scenarios have greater habitat loss than the corresponding low emission scenario across all land-cover scenarios (Fig. 7a). Specific land-cover scenarios modify only moderately the degree of fish habitat loss. Fish habitat loss is primarily driven by temperature impairment in headwater streams (Fig. A4.4b), resulting in a 4- to 10-times increase in river kilometers impacted in the high compared with low emission scenarios (Fig. 5a, 7a). The Backyard scenario exacerbates fish habitat changes caused by climate, whereas the three other land-cover scenarios remain similar. The Fish Habitat indicator is composed of flow, water temperature, and chloride impairment (Table 2). The greater degradation in Backyard occurs because of the greater abundance of impervious surfaces, with increased runoff from the land leading to greater spatial and temporal probability of crossing chloride and water temperature thresholds. Residential development and associated impervious surfaces in the Backyard scenario result in greater summertime heat flux to streams through stormwater runoff relative to other build-out scenarios. In the Community (both Small Community Food and Large Community Wild) scenarios, reduced effective imperviousness from coupling low-impact design buildout with concentrated development results in negligible increases in heat flux impacts to rivers and no additional fish habitat loss relative to current land cover and projected climates.
For Nitrogen Export, climate is also the primary determinant of indicator response, but interactions with land cover are also evident (Fig. 7b). The Nitrogen Export indicator is higher in all higher emission scenarios relative to lower emission scenarios. Higher emissions scenarios result in greater coastal N flux than lower emissions scenarios due to greater loading to the river system. First, the greater precipitation in high relative to low emission scenario results in higher atmospheric N inputs as well as higher water runoff. Second, greater forest N leaching results from higher temperatures under future scenarios due to declining uptake by conifers. Third, the river system has reduced capacity to retain N under higher flow regimes (Wollheim et al. 2008b). Small Community Food shows little response under the low emission climate scenario (as discussed above) similar to if land cover remained unchanged. Both the Large Community Wildlands and Backyard scenarios show additional N export increases relative to other land-cover scenarios, indicating an interaction between land cover and climate. These scenarios have similar high populations compared with Small Community Food, and result in N loading into rivers via domestic waste, either through septic systems (Backyard) or WWTPs (Large Community), which lead to increased waste N inputs.
The Large Community Wildlands scenario projects greater N export than the Backyard scenario (Fig. 7b), despite the former conserving more forest, and upgrading WWTPs to higher treatment levels with lower per capita human waste N inputs. The logistic loading function that is used to parameterize nonpoint N inputs to rivers as a function of land cover (Appendix 3) assumes relatively low N increases for low to moderate density urban and agricultural development (average increases to about ~21% urban + Ag land cover for Backyard and 9% for Large Community Wildlands). The mechanistic explanation for this pattern is that N removal by terrestrial ecosystems remains high up to certain thresholds of natural ecosystem loss, consistent with a number of previous studies (Groffman et al. 2004, Wollheim et. 2005). The Backyard scenario assumes a low density of future land-cover change that is often below the threshold. Thus, human waste associated with population growth, which is managed through distributed septic systems and not WWTP, is mostly retained, and there is relatively little change in N inputs. In contrast, the Large Community Wildlands scenario assumes domestic waste N is transferred to WWTP with Total Nitrogen (TN) removal efficiencies of 90%. The assumption of low response to land-cover change embedded in the loading function is derived from one location with a certain social and biogeophysical context (e.g., rural/suburban septic infrastructure, wetland abundance, etc. Wollheim et al. 2008b). Although the generality of this loading function to the entire domain should be tested more thoroughly, the validation results suggest it is reasonable and therefore useful for exploring potential future responses.
An additional small driver of the difference in N export between Backyard and Large Community Wildlands was that the former also presents greater opportunity for instream removal than the latter because loading occurs to smaller streams (as nonpoint sources) compared with waste water effluent to larger rivers (as point sources). The resulting longer lotic travel time and greater exposure to benthic surfaces provide more opportunities for N removal through denitrification (Mineau et al. 2015). However, in our simulations, only an additional 1.6% of total network inorganic N inputs was removed (denitrified) in the aquatic system under the Backyard scenario compared with Large Community Wildlands.
Water shortfalls are controlled more by land-cover and population scenarios than by climate scenarios (Fig. 7c). The lack of a strong climate driver occurs because conditions are expected to become increasingly wet under both lower and higher emissions scenarios, alleviating water shortages. The highly dispersed development of the Backyard scenario places an increasing population throughout headwater catchments, dependent on well water. In contrast, the Community scenario places more people on public water supplies that obtain their water from larger catchments. The Water Shortfall indicator we used relates water demand accumulated through the river network (a function of population and per capita demand) with surface water discharge in the grid cell. In the Backyard scenario, people are presumed to depend primarily on wells, not urban water supply infrastructure, where small catchments have limited hydrological storage. Thus, demand more frequently outstrips supply in these local areas. Runoff increases in the higher emission scenario compared with the lower emission scenario, which reflects a combination of greater annual precipitation predicted under a warmer climate and greater forest water-use efficiency under a C-enriched atmosphere (Fig. A4.4c, d). As a result, the high emission scenario actually reduces water shortfalls relative to the low emission scenario (Fig. 7c).
Flood risk is projected to increase under both emission scenarios (Fig. 7d), resulting from greater precipitation (Fig. A4.4c) occurring during extreme events (Hayhoe et al. 2007, Trenberth 2011). Backyard and Large Community Wildlands scenarios project a larger flood-affected population due simply to the larger populations compared with other scenarios. Generally, the higher emission scenario has greater flood risk than the lower emission scenario, with the exception of the Backyard scenario. High emissions are expected to have slightly higher precipitation and runoff relative to low emissions. The Backyard scenario differs in that more people live further north in the watershed than in the Community scenario (concentrated on existing urban areas, Fig. 1), which may expose these populations to more extreme flows.
The spatially distributed modeling approach used here is able to identify changes in environmental indicators relative to human settlement. As an example, the Fish Habitat Loss indicator, as the fraction of years with unsuitable habitat, is greatest in the southern part of the study domain (Fig. 8). Greater impairment in the southern portion of the watershed occurs because warmer air temperatures increase water temperatures as rivers flow from north to south (Stewart et al. 2011) and because denser road networks relative to the northern portions of the watershed experience greater winter road-salt application.
The Contemporary period shows elevated fish habitat loss in the smaller streams in the southern half of the watershed, but not in the larger rivers (Fig. 8a). Meteorological droughts in 1981 and 1993 are responsible for the contemporary fish habitat loss, which was caused by the low flow component of the indicator. It is possible that future climate from the GFDL GCM used in these scenarios does not capture longer periods of summer dry conditions later in the 20th century (Hayhoe et al. 2007, Wake et al. 2014). Thus, the scenarios presented here may underestimate the effect of the low-flow criteria on future fish habitat loss (Fig. 8).
Future scenarios are likely to intensify fish habitat loss near population centers in the high emission scenarios (Fig. 8b, d, top), while spreading through more of the previously unimpaired part of the watershed. In contrast, the low emission scenarios tend to shrink the footprint of the Fish Habitat Loss indicator, although retaining the likelihood of impairment near the population centers (Fig. 8c, e, bottom). The Backyard scenario shows a stronger response than the Community scenario because greater impervious cover leads to both increased road-salt loading and increased temperatures from increased summertime impervious surface runoff.
The greater Fish Habitat Loss indicator in large rivers compared with the contemporary period is particularly evident across all future scenarios, resulting in large part because of warmer water temperatures. Large river water temperature responds disproportionately because atmospheric warming further warms water as it flows downstream during summers, whereas headwater streams remain cooler because of groundwater recharged during cooler times of year. The spatially distributed, time-varying modeling approach used here allows consideration of heterogenous impacts to environmental function that can be used in valuation studies.
Despite the predominant influence of climate on the environmental indicators in the UMRW, land-cover management can mitigate degradation to certain attributes of the ecosystem (e.g., Fig. 8). Build-out scenarios directly influence land-based environmental indicators such as Forest and Agricultural Cover. Scenarios that maintain greater forest land have slightly greater C sequestration (Fig. A4.4e), whereas scenarios with greater agriculture increase food supply (Thorn et al. 2017) and resiliency (Tilman et al. 2002). In addition, these land covers affect numerous sociocultural resources, including water-related ecosystem services. Depending on the relative value placed on different indicators (Murphy et al. 2017), the landscape can be managed to maximize ecosystem services that are most important.
Coupled terrestrial–aquatic models express the understanding of processes relevant to specific land-cover decisions, permit exploration of relevant tradeoffs (Antle and Capalbo 2001, Bennett et al. 2009), and are essential in understanding the response of ecosystem dynamics to changes in both climate and land cover. Results from PnET–FrAMES provide a measure of the potential impact of each land-cover scenario and each climate scenario on different ecosystem service indicators. As regional managers have little influence over climate change and because we are currently on a higher emission trajectory (Pachauri et al. 2014), managers should consider potential land cover in conjunction with the higher emissions climate scenario to guide future planning. Projections from PnET–FrAMES suggest that infilled development (Community) scenarios exhibit two types of paradigms with respect to specific aquatic environmental indicators under a high emission future (Fig. 7). The three land-cover scenarios with limited expansion of the residential footprint (Constant, Small Community Food, Large Community Wild) show smaller levels of Fish Habitat Loss and Water Shortfalls than the dispersed buildout (Backyard). The Large Community Wildlands scenario represents a doubling of population similar to Backyard scenario, but has much lower environmental detriment based on these two indicators. However, the Large Community Wildlands scenario does show increased impact on the Nitrogen Export indicator (more point sources) and Flood Risk indicator (more people along large river corridors) relative to Backyard, suggesting that there are tradeoffs that must be further managed under an infilled paradigm.
The scenarios tested are simple and do not encompass numerous potential approaches that managers may use to mitigate the adverse effects of the investigated environmental stressors. The scenarios do not account for potential nonpoint N runoff controls including managed riparian buffers (Mayer et al. 2005) or stormwater control (Collins et al. 2010) that would be consistent with Community scenarios. Nor do the scenarios account for novel flood control measures such as blue–green infrastructure (Thorne et al. 2015) or explicit development outside of flood zones, because flood zones are mapped at resolutions finer than our simulations. Therefore, the flood risk identified here may potentially be mitigated through engineered interventions. Finally, potential management interventions such as constructed stream refugia (Sedell et al. 1990, Isaak et al. 2015) are not considered but could mediate warming stream temperatures for instream biota, a threat under all build-out scenarios. The scenarios presented here identify those environmental stressors that are most likely to need more directed research and management to mitigate changing populations and expected warming.
Process-based terrestrial–aquatic models are essential for quantifying potential future environmental impacts due to spatially distributed changes in land cover and climate. Here, we simulated dynamic interactions across land and water domains at a daily temporal resolution and a spatial resolution that accounts for the heterogeneity and associated processes of the UMRW in New Hampshire. Our approach linked models of intermediate complexity to connect terrestrial and aquatic domains and allowed us to explore the impacts of changes in climate and land cover on aquatic ecosystems through the use of different scenario combinations. Process-based models enable understanding of macroscale ecosystem responses based on underlying ecosystem processes.
Water is arguably the most fundamental resource for society, and it is therefore critical to quantify how water quantity and quality will respond to projected changes in land cover and climate. Our results suggest that climate is expected to be the most influential driver of water-related impairment in the UMRW, although impacts to climate and aquatic environment conditions are likely to be modest until mid-century. After 2050, aquatic impacts are projected to increase rapidly in the high emission scenario, and degree of change is further influenced by land cover. Because recent analysis points to the likelihood of a high emissions scenario (Pachauri et al. 2014), the region should prepare for changes in ecosystem services. Land-cover management is the leverage point that local and state decision makers have to adapt to global climate change, and management interventions that mitigate changes in population and temperature will need to move beyond simple land-cover/land-use formulations analyzed here.
We gratefully acknowledge the technical assistance of S. Glidden and A. Proussevitch, the support of all Lotic Volunteer Temperature, Electrical Conductance, and Stage (LoVoTECS) volunteers and coordinators (M. Green, E. Volitis and A. Inserillo) as well as Bill McDowell, Lisle Snyder, and Jody Potter coordinating the High Intensity Aquatic Network (HIAN). LoVoTECS and HIAN data used herein is available at the Data Discovery Center (ddc.unh.edu). This research was funded by the NSF EPSCoR Ecosystems and Society (EPS: 1101245), and the Town of Durham and UNH facilities. Partial funding was provided by the National Sea Grant College Program of the U.S. Department of Commerce's National Oceanic and Atmospheric Administration, grant NA10OAR4170082 to the N.H. Sea Grant College Program, and by the New Hampshire Agricultural Experiment Station. This is NHAES Scientific Contribution Number: 2736. This work was supported by the USDA National Institute of Food and Agriculture (Hatch) Project (Project Number: 0225006).
Aber, J. D., S. V. Ollinger, C. T. Driscoll, C. A. Federer, and P. B. Reich. 2005. PnET models: carbon, nitrogen, water dynamics in forest ecosystems (vers. 4 and 5). Model product. Oak Ridge National Laboratory Distributed Active Archive Center, Oak Ridge, Tennessee, USA. http://dx.doi.org/10.3334/ORNLDAAC/817
Antle, J. M., and S. M. Capalbo. 2001. Econometric-process models for integrated assessment of agricultural production systems. Agricultural Economics 83(2):389–401. http://dx.doi.org/10.1111/0002-9092.00164
Bagstad, K. J., D. J. Semmens, S. Waage, and R. Winthrop. 2013a. A comparative assessment of decision-support tools for ecosystem services quantification and valuation. Ecosystem Services 5:e27–e39. http://dx.doi.org/https://doi.org/10.1016/j.ecoser.2013.07.004
Bagstad, K. J., D. J. Semmens, and R. Winthrop. 2013b. Comparing approaches to spatially explicit ecosystem service modeling: a case study from the San Pedro River, Arizona. Ecosystem Services 5:e40–e50. http://dx.doi.org/https://doi.org/10.1016/j.ecoser.2013.07.007
Bennett, E. M., G. D. Peterson, and L. J. Gordon. 2009. Understanding relationships among multiple ecosystem services. Ecology Letters 12:1394–1404. http://dx.doi.org/10.1111/j.1461-0248.2009.01387.x
Boumans, R., R. Costanza, J. Farley, A. M.Wilson, R. Portela, J. Rotmans, F. Villa, and M. Grasso. 2002. Modeling the dynamics of the integrated earth system and the value of global ecosystem services using the GUMBO model. Ecological Economics 41:529–560 http://dx.doi.org/10.1016/S0921-8009(02)00098-8
Boumans, R., J. Roman, I. Altman, and L. Kaufmanc. 2015. The multiscale integrated model of ecosystem services (MIMES): simulating the interactions of coupled human and natural systems. Ecosystem Services 12:30–41. http://dx.doi.org/10.1016/j.ecoser.2015.01.004
Boyd, J. 2004. What’s nature worth? Using indicators to open the black box of ecological valuation. Resources 154:18–22.
Brauman, K. A., G. C. Daily, T. K. Duarte, and H. A. Mooney. 2007. The nature and value of ecosystem services: an overview highlighting hydrologic services. Annual Review of Environmental Resources 32:6.1–6.32. http://dx.doi.org/10.1146/annurev.energy.32.031306.102758
Burakowski, E. A., P. Cameron, W. B. Braswell, and D. P. Brown. 2008. Trends in wintertime climate in the northeastern United States: 1965–2005. Geophysical Research 113: D20114. http://dx.doi.org/https://doi.org/10.1029/2008JD009870
Burkhard, B., F. Kroll, S. Nedkov, and F. Müller. 2012. Mapping ecosystem service supply, demand and budgets. Ecological Indicators 21:17–29. http://dx.doi.org/10.1016/j.ecolind.2011.06.019
Carpenter, S. R., E. G. Booth, S. Gillon, C. J. Kucharik, S. Loheide, A. S. Mase, M. Motew, J. Qiu, A. R. Rissman, J. Seifert, E. Soylu, M. Turner, and C. B. Wardropper. 2015. Plausible futures of a social–ecological system: Yahara watershed, Wisconsin, USA. Ecology and Society 20(2): 10. http://dx.doi.org/10.5751/ES-07433-200210
Chen, L. D., L. Yang, W. Wei, Z. T. Wang, B. R. Mo, and G. J. Cai. 2013. Towards sustainable integrated watershed ecosystem management: a case study in Dingxi on the Loess Plateau, China. Environmental Management 51:126–137. http://dx.doi.org/10.1007/s00267-011-9807-0
Coe, M. T. 2000. Modeling terrestrial hydrological systems at the continental scale: testing the accuracy of an atmospheric GCM. Climate 13:686–704. http://dx.doi.org/10.1175/1520-0442(2000)013<0686:MTHSAT>2.0.CO;2
Collins, K. A., T. J. Lawrence, E. K. Stander, R. J. Jontos, S. S. Kaushal, T. A. Newcomer, N. B. Grimm, and M. L. Cole Ekberg. 2010. Opportunities and challenges for managing nitrogen in urban stormwater: a review and synthesis. Ecological Engineering 36(11):1507–1519. http://dx.doi.org/10.1016/j.ecoleng.2010.03.015
Contosta, A. R., A. Adolph, D. Burchsted, E. Burakowski, M. Green, D. Guerra, M. Albert, J. Dibb, M. Martin, W. H. McDowell, M. Routhier, C. Wake, R. Whitaker, and W. Wollheim. 2017. A longer vernal window: the role of winter coldness and snowpack in driving spring transitions and lags. Global Change Biology 23:1610–1625. http://dx.doi.org/10.1111/gcb.13517
Costanza, R., A. Voinov, R. Boumans, T. Maxwell, F. Villa, L. Wainger, and H. Voinov. 2002. Integrated ecological economic modeling of the Patuxent River watershed, Maryland. Ecological Monographs 72:203–231. http://dx.doi.org/10.1890/0012-9615(2002)072[0203:IEEMOT]2.0.CO;2
de Groot, R. S., R. Alkemade, L. Braat, L. Hein, and L. Willemen. 2010. Challenges in integrating the concept of ecosystem services and values in landscape planning, management and decision making. Ecological Complexity 7:260–272. http://dx.doi.org/10.1016/j.ecocom.2009.10.006
Goodridge, B. M., and J. M. Melack. 2012. Land use control of stream nitrate concentrations in mountainous coastal California watersheds. Geophysical Research 117:G02005. http://dx.doi.org/10.1029/2011JG001833
Green, M. B., W. M. Wollheim, N. B. Basu, G. Gettel, P. S. Rao, N. Morse, and R. Stewart. 2009. Effective denitrification scales predictably with water residence time across diverse systems. Nature Proceedings 713.
Groeneveld, J., B. Müller, C. Buchmann, G. Dressler, C. Guo, N. Hase, F. Hoffmann, F. John, C. Klassert, T. Lauf, V. Liebelt, H. Nolzen, N. Pannicke, J. Schulze, H. Weise, and N. Schwarz. 2017. Theoretical foundations of human decision-making in agent-based land use models—a review. Environmental Modelling and Software 87:39–48. http://dx.doi.org/10.1016/j.envsoft.2016.10.008
Groffman, P. M., N. L. Law, K. T. Belt, L. E. Band, and G. T. Fisher. 2004. Nitrogen fluxes and retention in urban watershed ecosystems. Ecosystems 7:393–403. http://dx.doi.org/https://doi.org/10.1007/s10021-003-0039-x
Hale, R. L., L. Turnbull, S. Earl, N. Grimm, K. Riha, G. Michalski, K. A. Lohse, and D. Childers. 2014. Sources and transport of nitrogen in arid urban watersheds. Environmental Science and Technology 48(11):6211–6219. http://dx.doi.org/10.1021/es501039t
Harbaugh, A. W. 2005. MODFLOW-2005, the U.S. Geological Survey modular ground-water model—the ground-water flow process. U.S. Geological Survey techniques and methods 6-A16. U.S. Geological Survey, Reston, Virginia, USA. [online] URL: http://pubs.usgs.gov/tm/2005/tm6A16/
Hayhoe, K., C. P. Wake, T. G. Huntington, L. Luo, M. D. Schwartz, J. Sheffield, E. Wood, B. Anderson, J. Bradbury, A. DeGaetano, T. J. Troy, and D. Wolfe. 2007. Past and future changes in climate and hydrological indicators in the US Northeast. Climate Dynamics 28:381–407. http://dx.doi.org/10.1007/s00382-006-0187-8
Heckbert, S., T. Baynes, and A. Reeson. 2010. Agent-based modeling in ecological economics. Annals of the New York Academy of Sciences 1185:39–53. http://dx.doi.org/10.1111/j.1749-6632.2009.05286.x
Heckbert, S., R. Costanza, and L. Parrott. 2014. Achieving sustainable societies: lessons from modelling the Ancient Maya. The Solutions Journal 5(5):55–64.
Hodgkins, G. A., and R. W. Dudley. 2006. Changes in the timing of winter—spring high streamflows in eastern North America 1913–2002. Geophysical Research Letter 33(6):1–5 http://dx.doi.org/https://doi.org/10.1029/2005GL025593
Horton, R., G. Yohe, W. Easterling, R. Kates, M. Ruth, A. Whelchel, D. Wolfe, F. Lipschultz, and E. Sussman. 2014. Northeast. Chapter 16 in J. M. Melillo, T. C. Richmond, and G. W. Yohe, editors. Climate change impacts in the United States: the third national climate assessment. U.S. Global Change Research Program, U.S. Government, Washington, D.C., USA.
Isaak, D. J., M. K. Young, D. E. Nagel, D. L. Horan, and M. C. Groce. 2015. The cold-water climate shield: delineating refugia for preserving salmonid fishes through the 21st century. Global Change Biology 21(7):2540–2553. http://dx.doi.org/10.1111/gcb.12879
Iverson, L. R., A. M. Prasad, S. N. Matthews, and M. Peters. 2008. Estimating potential habitat for 134 eastern US tree species under six climate scenarios. Forest Ecology and Management 254:390–406. http://dx.doi.org/10.1016/j.foreco.2007.07.023
Karabulut, A. B. N. Egoh, D. Lanzanova, B. Grizzetti, G. Bidoglio, L. Pagliero, F. Bouraoui, A. Aloe, A. Reynaud, J. Maes, I. Vandecasteele, and S. Mubareka. 2015. Mapping water provisioning services to support the ecosystem–water–food–energy nexus in the Danube river basin. Ecosystem Services 17:278–292. http://dx.doi.org/10.1016/j.ecoser.2015.08.002
Lee, S. Y., M. E. Ryan, A. F. Hamlet, W. J. Palen, and J. J. Lawler. 2015. Correction: projecting the hydrologic impacts of climate change on montane wetlands. PLoS ONE 10(11): e0142960. http://dx.doi.org/10.1371/journal.pone.0142960
Liang, X., D. P. Lettenmaier, E. F. Wood, and S. J. Burges. 1994. A simple hydrologically based model of land surface water and energy fluxes for GSMs. Geophysical Research 99(D7):14,415–14,428.
Logsdon, R. A., and I. Chaubey. 2012a. Ecosystem services: a need to move from concept to application. Resource May/June: 31. http://dx.doi.org/https://doi.org/10.1016/j.ecolmodel.2013.02.009
Logsdon, R. A., and I. Chaubey. 2012b. A quantitative approach to evaluating ecosystem services: Wildcat Creek Watershed (WCW) in central Indiana. Ecological Modelling 257:57–65.
Lotic Volunter Temperature, Electrical Conductivity, and Stage Sensing Network (LoVOTECS). 2012. The Lotic Volunter Temperature, Electrical Conductivity, and Stage Sensing Network Newsletter. LoVoTECS Network, Plymouth State University and NH EPSCoR 1:1–4.
Mavrommati, G., M. Borsuk, and R. B. Howarth. 2017. A novel deliberative multicriteria evaluation approach to ecosystem service valuation. Ecology and Society 22(2): 39. http://dx.doi.org/10.5751/ES-09105-220239
Mayer, P., S. K. Reynolds, Jr., T. J. Canfield, and M. D. McCutchen. 2005. Riparian buffer width, vegetative cover, and nitrogen removal effectiveness: a review of current science and regulations. U.S. Environmental Protection Agency, National Risk Management Research Laboratory, Ada, Oklahoma, USA.
Miara, A., and C. J. Vörösmarty. 2013. A dynamic model to assess tradeoffs in power production and riverine ecosystem protection. Environmental Science Processes and Impacts 15:1113–1126. http://dx.doi.org/10.1039/c3em00196b
Mineau, M. M., W. M. Wollheim, and R. J. Stewart. 2015. An index to characterize the spatial distribution of land use within watersheds and implications for river network nutrient removal and export. Geophysical Research Letter 42:6688–6695. http://dx.doi.org/10.1002/2015GL064965
Müller, F., and B. Burkhard. 2012. The indicator side of ecosystem services. Ecosystem Services 1:26–30. http://dx.doi.org/10.1016/j.ecoser.2012.06.001
Mulukutla, G. K., B.T. Godbois, and F. Serita. 2015. Deployment of a large-scale soil moisture geosensor network- experience and lessons learnt. ACM SISGPATIAL Magazine 7(2):3–13.
Murphy, M., G. Mavrommati, V. R. Mallampalli, R. B. Howarth, and M. E. Borsuk. 2017. Comparing group deliberation to other forms of preference aggregation in valuing ecosystem services. Ecology and Society 22(4):17. http://dx.doi.org/10.5751/ES-09519-220417
Nakicenvoic, N., and R. Swart, editors. 2000. Special report on emissions scenarios. Working Group III, Intergovernmental Panel on Climate Change, Geneva, Switzerland and Cambridge University Press, Cambridge, UK.
Nelson, E., G. Mondoza, J. Regetz, S. Polasky, J. Tallis, D. R. Cameron, K. M. A. Chan, G. C. Daily, J. Goldstein, P. M. Kareiva, E. Londsdorf, R. Naidoo, T. H. Ricketts, and M. R. Shaw. 2009. Modelling multiple ecosystems services, biodiversity conservation, commodity production, and tradeoffs at landscape scale. Frontiers in Ecology and the Environment 7(1):4–11. http://dx.doi.org/https://doi.org/10.1890/080023
Notter, B. H. Hurni, U. Wiesmann, and K. C. Abbaspour. 2012. Modelling water provision as an ecosystem service in a large East African river basin. Hydrology and Earth System Science 16:69–86. http://dx.doi.org/10.5194/hess-16-69-2012
Ollinger, S. V., J. D. Aber, P. B. Reich, and R. J. Freuder. 2002. Interactive effects of nitrogen deposition, tropospheric ozone, elevated CO2 and land use history on the carbon dynamics of northern hardwood forests. Global Change Biology 8:545–562. http://dx.doi.org/10.1046/j.1365-2486.2002.00482.x
Ollinger, S. V., C. L. Goodale, K. Hayhoe, and J. P. Jenkins. 2008. Potential effects of climate change and rising CO2 on ecosystem processes in northeastern U.S. forests. Mitigation and Adaptation Strategies for Global Change 13:467–485. http://dx.doi.org/10.1007/s11027-007-9128-z
Pachauri, R. K., M. R. Allen, V. R. Barros, J. Broome, W. Cramer, R. Christ, J. A. Church, L. Clarke, Q. Dahe, P. Dasgupta, and others. 2014. Climate change 2014: synthesis report. Contribution of Working Groups I, II, and III to the Fifth Assessment Report, Intergovernmental Panel on Climate Change, Geneva, Switzerland.
Pellerin, B. A., B. A. Bergamaschi, P. S. Murdoch, B. D. Downing, J. F. Saraceno, G. R. Aiken, and R. G. Striegl. 2011. The aquatic real-time monitoring network; in-situ optical sensors for monitoring the nation’s water quality. Fact Sheet 2011-3061, U.S. Geological Survey, Washington, D.C., USA.
Poff, N. L., J. D. Allan, M. B. Bain, J. R. Karr, K. L. Prestegaard, B. D. Richter, R. E. Sparks, and J. C. Stromberg. 1997. The natural flow regime: a paradigm for river conservation and restoration. BioScience 47:769–784. http://dx.doi.org/10.2307/1313099
Qiu, J., and M. G. Turner. 2015. Importance of landscape heterogeneity in sustaining hydrologic ecosystem services in an agricultural watershed. Ecosphere 6(11):1–19. http://dx.doi.org/10.1890/ES15-00312.1
Queiroz, C., M. Meacham, K. Richter, A. V. Norström, E. Andersson, J. Norberg, and G. Peterson. 2015. Mapping bundles of ecosystem services reveals distinct types of multi-functionality within a Swedish landscape Ambio 44(Suppl. 1):S89–S101. http://dx.doi.org/https://doi.org/10.1007/s13280-014-0601-0
Raymond, P. A., and J. E. Saiers. 2010. Event controlled DOC export from forested watersheds. Biogeochemistry 100:197–209. http://dx.doi.org/10.1007/s10533-010-9416-7
Schroeter, D., W. Cramer, R. Leemans, I. C. Prentice, M. B. Araujo, N. W. Arnell, A. Bondeau, H. Bugmann, T. R. Carter, C. A. Gracia, A. C. de la Vega-Leinert, M. Erhard, F. Ewert, and others. 2005. Ecosystem service supply and vulnerability to global change in Europe. Science 310:1333–1337. http://dx.doi.org/10.1126/science.1115233
Sedell, J. R., G. H. Reeves, F. R. Hauer, J. A. Stanford, and C. P. Hawkins. 1990. Role of refugia in recovery from disturbances: modern fragmented and disconnected river systems. Environmental Management 14(5):711–724. http://dx.doi.org/10.1007/BF02394720
Soylu, M. E., C. J. Kucharik, and S. P. Loheide, II. 2014. Influence of groundwater on plant water use and productivity: development of an integrated ecosystem-variably saturated soil water flow model. Agricultural and Forest Meteorology 189–190(1):198–210. http://dx.doi.org/10.1016/j.agrformet.2014.01.019
Stein, S. M., R. E. McRoberts, L. G. Mahal, M. A. Carr, R. J. Alig, S. J. Comas, D. M. Theobald, and A. Cundiff. 2009. Private forests, public benefits: increased housing density and other pressures on private forests contributions. General Technical Report, U.S. Department of Agriculture, Forest Service, Pacific Northwest Research Station, Portland, Oregon, USA. http://dx.doi.org/10.2737/PNW-GTR-795
Stewart, R. J., W. M. Wollheim, M. N. Gooseff, M. A. Briggs, J. M. Jacobs, B. J. Peterson, and C. S. Hopkinson. 2011. Separation of river network-scale nitrogen removal among the main channel and two transient storage compartments. Water Resources Research 47(10):W00J10, http://dx.doi.org/10.1029/2010WR009896
Stewart, R. J., W. M. Wollheim, A. Miara, C. J. Vörösmarty, B. Fekete, R. B. Lammers, and B. Rosenzweig. 2013. Horizontal cooling towers: riverine ecosystem services and the fate of thermoelectric heat in the contemporary Northeast US. Environmental Research Letters 8(2):025010. http://dx.doi.org/10.1088/1748-9326/8/2/025010
Stoner, A. M. K., K. Hayhoe, X. Yang, and D. J. Wuebbles. 2012. An asynchronous regional regression model for statistical downscaling of daily climate variables. Journal of Climatology 33(11):2473–2494. http://dx.doi.org/10.1002/joc.3603
Tallis, H., and S. Polasky. 2009. Mapping and valuing ecosystem services as an approach for conservation and natural-resource management. New York Academy of Sciences 1162:265–283. http://dx.doi.org/10.1111/j.1749-6632.2009.04152.x
Tallis, H. T., T. Ricketts, A. Guerry, A. S. Wood, R. Sharp, E. Nelson, D. Ennaanay, S. Wolny, N. Olwero, K. Vigerstol, D. Pennington, G. Mendoza, J. Aukema, J. Foster, J. Forrest, D. Cameron, K. Arkema, E. Lonsdorf, C. Kennedy, G. Verutes, C. Kim, G. Guannel, M. Papenfus, J. Toft, M. Marsik, J. Bernhartdt, R. Griffin. 2013. InVEST2.6.0 user’s guide. The Natural Capital Project, Stanford, California, USA.
Thorn, A. M., C. P. Wake, C. Grimm, C. Mitchell, M. M. Mineau, and S. V. Ollinger. 2017. Development of scenarios for land cover, population density, impervious cover, and conservation in New Hampshire, 2010–2100. Ecology and Society 22(4):19. http://dx.doi.org/10.5751/ES-09733-220419
Thorne, C. R., E. C. Lawson, C. Ozawa, S. L. Hamlin, and L. A. Smith. 2015. Overcoming uncertainty and barriers to adoption of blue–green infrastructure for urban flood risk management: uncertainties and barriers to adoption of BGI. Journal of Flood Risk Management, online early. http://dx.doi.org/10.1111/jfr3.12218
Tilman, D., K. G. Cassman, P. A. Matson, R. Naylor, and S. Polasky. 2002. Agricultural sustainability and intensive production practices. Nature 418:671–677. http://dx.doi.org/10.1038/nature01014
Trenberth, K. E. 2011. Changes in precipitation with climate change. Climate Research 47:123–138. http://dx.doi.org/10.3354/cr00953
U.S. Environmental Protection Agency (USEPA). 2010. ICLUS tools and datasets (version 1.3.2). U.S. Environmental Protection Agency, Washington, D.C., USA.
van Oudenhoven, A. P. E., K. Petz, R. Alkemade, L. Hein, and R. S. de Groot. 2012. Framework for systematic indicator selection to assess effects of land management on ecosystem services. Ecological Indicators 21:110–122. http://dx.doi.org/10.1016/j.ecolind.2012.01.012
Vigerstol, K. L., and Aukema, J. E. 2011. A comparison of tools for modeling freshwater ecosystem services. Environmental Management 92:2403–2409. http://dx.doi.org/https://doi.org/10.1016/j.jenvman.2011.06.040
Villa, F., M. Ceroni, K. Bagstad, G. Johnson, and S. Krivovet. 2009. ARIES (ARtifcial Intelligence for Ecosystem Services): a new tool for ecosystem services assessment, planning, and valuation. Eleventh International BIOECON Conference on Economic Instruments to Enhance the Conservation and Sustainable Use of Biodiversity, Venice, Italy. [online] URL: http://www.ucl.ac.uk/bioecon/11th_2009/Villa.pdf
Vörösmarty, C. J., C. A. Federer, and A. L. Schloss. 1998. Potential evaporation functions compared on US watersheds: possible implications for global-scale water balance and terrestrial ecosystem modeling. Hydrology 207:147–169.
Wake, C., E. Burakowski, P. Wilkinson, K. Hayhoe, A. Stoner, C. Keeley, and J. LaBranche. 2014. Climate change in southern New Hampshire past, present, and future. Climate Solutions New England, Durham, New Hampshire, USA.
Wisser, D., B. M. Fekete, C. J. Vörösmarty, and A. H. Schumann. 2010. Reconstructing 20th century global hydrography: a contribution to the Global Terrestrial Network-Hydrology (GTN-H). Hydrology and Earth System Sciences 14:1–24. http://dx.doi.org/10.5194/hess-14-1-2010
Wollheim, W. M., B. A. Pellerin, C. J. Vörösmarty, and C. S. Hopkinson. 2005. N retention in urbanizing headwater catchments. Ecosystems 8:871–884. http://dx.doi.org/10.1007/s10021-005-0178-3
Wollheim, W. M., B. J. Peterson, C. J. Vörösmarty, C. Hopkinson, and S. A. Thomas. 2008a. Dynamics of N removal over annual time scales in a suburban river network. Geophysical Research 113: G03038.
Wollheim, W. M., C. J. Vörösmarty, A. F. Bouwman, P. A. Green, J. Harrison, E. Linder, B. J. Peterson, S. Seitzinger, and J. P. M. Syvitski. 2008b. Global N removal by freshwater aquatic systems: a spatially distributed, within-basin approach. Global Biogeochemical Cycle 22:GB2026.
Wollheim, W. M., C. J. Vörösmarty, B. J. Perterson, S. P. Seitzinger, and C. S. Hopkinson. 2006. Relationship between river size and nutrient removal. Geophysical Research Letters 33: L06410. http://dx.doi.org/10.1029/2006GL025845
Wood, A. W., E. P. Maurer, A. Kumar, and D. P. Lettenmaier. 2002. Long-range experimental hydrologic forecasting for the eastern United States, Geophysical Research 107(D20):ACL 6-1–ACL 6-15.