Economic Evaluation of Water Management Alternatives in the Upper Green River Basin of Wyoming

Abstract

Water use efficiency measures are generally recommended to reduce water use. Yet, flood irrigation practices in high-elevation mountain valleys of the Colorado River Basin headwaters generate return flows, which support late-season streamflow and groundwater recharge. Return flows support the ecosystem and provide recreational benefits. This study provides a framework for quantifying how land-use changes and associated return flow patterns affect the economic value of water across uses in a hydrologically connected, shallow alluvial aquifer system. This study first investigates how return flow patterns could change under three alternatives to flood irrigation: an increased use of center pivots, increased residential development, and conversion to pasture. The brown trout was used as an indicator species to track eco-hydrology, return flow, and capacity for recreational activities under each alternative. Estimates from the non-market valuation literature coupled with predicted changes in brown trout productivity approximate associated changes to recreational angler value. Recreational angler values are highest under the flood irrigation alternative. The inclusion of recreational angler values with agricultural values alters the magnitude of returns but not the rankings. These results highlight the potential heterogeneity of conclusions to be drawn regarding water use efficiency, depending on the economic value of water in different uses and the degree of hydrologic connectivity. This study also highlights data gaps and modeling needs for conducting similar future analyses.

1. Introduction

Policymakers, government agencies, and researchers often focus on improved irrigation efficiencies to reduce water usage by agriculture so that it may be transferred to other uses [1,2,3,4]. However, net water gains from adopting irrigation efficiency technologies may not always be as beneficial as expected, and widespread adoption could cause unintended consequences related to ecosystem service provision from return flows in rivers [5,6,7]. The current literature proposes that policies targeting reductions in agricultural water usage should consider the economic value provided by the water resource, including ecosystem service provision, as well as returns from production or other uses to create the greatest social value [8,9,10].

Unfortunately, the economic value of water associated with non-market goods such as ecosystem services is often not readily available when policy decisions are made regarding changes in agricultural practices or allocation of this scarce resource away from agriculture [11]. Literature exists describing the potential for using existing market-derived estimates as well as non-market estimates to determine the economic value of ecosystem services. Both types of estimates have advantages and disadvantages [12]. However, failing to account for the value of ecosystem services from water may lead to suboptimal outcomes.

The research objective of this study is to evaluate the potential importance of including the ecosystem service value of water in consideration of alternatives affecting water allocation amongst various uses. In the study area, the Upper Green River Basin of Southwestern Wyoming, USA, agricultural producers increasingly have financial incentives to transition away from traditional flood irrigation practices in three primary ways. First, subsidies are available in some parts of the region to encourage the adoption of more efficient sprinkler irrigation systems [13]. Second, producers have a financial incentive to subdivide their operations for real estate development due to the breathtaking scenery and proximity to Yellowstone and Grand Teton National Parks [14]. Third, water managers of the Colorado River Basin (of which the Upper Green is a headwaters) are exploring ways to make more water available for downstream users; one such option is full- or partial-season fallow by Upper Green producers [15,16]. Each of these changes would reduce artificial wetland provision and late-season return flows from current flood irrigation practices [17,18,19].

A producer’s decision to continue flood irrigation or adopt some other alternative consequently affects the local ecosystem and the presence of natural amenities due to the high level of hydrologic connectivity in the region. Flood irrigation in this context is a classic case of a positive economic externality; the decision-maker’s activities affect others, yet the decision-maker is not compensated for the marginal benefits realized by others [20]. A producer’s financial return from flood irrigating is lower than the full benefits these services provide for the ecosystem in the form of groundwater recharge that supports wetlands and late-season stream flows that improve riparian habitat and benefit downstream irrigators.

The value of angling benefits is used as an economic measure of ecosystem service benefits related to the positive externality of water available from different alternatives (additional ecosystem benefits likely exist but are harder to measure). Estimates of consumer willingness to pay for fishing in the basin and similar locations in the Intermountain West are significant [21,22,23]. If these willingness-to-pay estimates are applied to the differences in recreational opportunities that might result in the basin from altered water availability for ecosystem services, would they be sufficient to compensate producers for maintaining current flood irrigation practices in view of the various pressures described above?

This study estimates changes in the direct economic value—both agricultural and recreational—of water resources in the Upper Green River Basin that arise from changes in land use. First, the hydrologic response is estimated for the following alternatives: center-pivot adoption, fallow, and residential development. Then, the response of a key indicator species to the altered return flow patterns is quantified using existing research. Willingness-to-pay estimates from elsewhere in the region are then used to estimate non-agricultural values of water associated with recreational angling. Finally, the implications of incorporating ecosystem service values into the evaluation are discussed.

2. Study Area and Framework

2.1. Study Area

The New Fork River is located on the western slopes of the Wind River Mountain range in West–Central Wyoming (see Figure 1). There is a decrease in elevation of approximately 300 m (2400 to 2100 m) over its 110 km length from New Fork Lake to its confluence with the Green River. The Green River joins the Colorado River in Utah, approximately 1100 km downstream of the New Fork. The Green River contributes an average of 2210 million cubic meters to the Colorado River Basin per year, which amounts to 34% of the Colorado River System’s total water [24]. The New Fork is a major tributary to the Upper Green River.

The New Fork watershed has a semi-arid climate with an annual average precipitation of 305 mm [25]. Most of this precipitation is in the form of snowfall that accumulates in the Wind River Range during the winter months and melts in the spring. Streamflow in the New Fork depends on snowpack in the mountains and the quantity of snowmelt water stored in New Fork Lake. Water from the New Fork supports irrigation and recreation as well as aquatic life throughout the region.

The New Fork Irrigation District “NFID” comprises 5900 hectares (ha) northeast of Pinedale, Wyoming. Its 94 members are a mix of agricultural and residential water users. The short growing season limits commercial agricultural production to native grasses (primarily timothy grass) for livestock forage. Producers on the New Fork flood irrigate their fields for hay production using gravity-fed flood irrigation. In late May/early June, they divert water across fields until the surface is covered and the soil is saturated. In average and above-water years, they maintain this saturation period until late July, at which point they cease delivery and dry the soil out for harvest. Harvest occurs in August to early September. NFID producers harvest just one cutting of hay per year due to the short growing season. After harvest, producers re-apply water to the fields for stock water and to keep the water table elevated before the winter months [26].

This region has greater hydrologic connectivity than the surrounding regions. Much of the land in the irrigation district adjacent to the New Fork River overlies a shallow, unconfined alluvial aquifer system comprising gravel and sand layers [27]. The subsurface water across the district is hydrologically connected within the unconfined alluvial aquifer system. Diverted irrigation water returns to the stream system through overland, subsurface, and channel flow, called “return flows”. On average, approximately 70% of water used for flood irrigation returns to the New Fork River. Of this quantity, 90% returns during the spring and summer of the year in which it was diverted; the remainder returns during lower flow winter months [26]. These late-season return flows help maintain elevated base flow water levels longer into the season and provide downstream users with water that would not be available during the winter months otherwise.

2.2. Analytical Framework

Private landowners whose lands are not protected under conservation easement have the right to alter their land management and related water use. As flood-irrigated agriculture accounts for most water diverted in the Upper Green River Basin [13], such private decisions regarding land and water use have impacts that extend beyond the ranch boundaries to the local ecosystem and downstream users.

The following framework is used to assess how different land uses available to private landowners in the area may affect the disposition of water, to the benefit or disadvantage of other water users and the downstream system. The study framework (Figure 2) is relevant for mountain valleys like the New Fork area, in which return flows are substantial and consequently create significant hydrologic interconnectivity with downstream users and the environment.

The hydrologic response (the timing and quantity of return flows) to the potential alternatives (the adoption of efficient irrigation technology, the conversion of hay land to pasture, and residential development) is quantified (Figure 2). Hay yield estimates, market price estimates, and production cost estimates are used to quantify the agricultural production value under each scenario. Sublette County is a popular destination for anglers; the presence of water late in the season improves fishing conditions. The brown trout is used as a proxy for recreational fishing because estimates from existing research can be used to predict the population response of brown trout to reductions in late-season stream flows. Angler benefits associated with the presence of more fish in the stream are then estimated. This permits a comparison of the water’s value to producers and anglers under current use to the water’s value in each of the examined alternatives.

3. Methods

3.1. Potential Alternatives

Private landowners may undertake any number of different activities on their land that may affect their economic returns. Likely alternative activities are described below, along with reasonable assumptions regarding how these land uses would affect water use patterns and flows. Methods for quantifying the hydrologic, ecological, and economic impacts associated with these potential alternative land uses are also outlined.

3.1.1. Flood Irrigation (Current Practice)

Water in the NFID is currently diverted from the New Fork River and applied to hay fields using flood irrigation. Over 80 km of ditches deliver irrigation water to approximately 5900 ha for native grass hay production [26]. Typically, the efficiency of flood irrigation ranges between 40 and 60% [28]. On the NFID, efficiency is lower; an average of 70% of the water applied returns to the stream through return flow [26]. The consumptive irrigation requirement for meadow mountain hay in the region is estimated to be 0.43 hectare-meters per hectare (ha-m/ha) per season [29]. Assuming an irrigation efficiency of 50%, 0.86 ha-m/ha must be applied for flood irrigation.

A partial enterprise budget for native (timothy) grass hay is developed to determine agricultural net projected returns from hay production under current flood irrigation practices [30]. The budget is adopted from eastern Oregon [31] and updated based on interviews with NFID producers to reflect local typical operation and production costs. Local producers primarily use their hay to feed livestock over the winter. Livestock production is used in the calculation of water’s value in hay production because hay could be purchased from local and regional markets. (Omitting livestock value in the computation of hay irrigation benefits is standard practice [32]. In parallel fashion, existing enterprise budgets of livestock operations in Wyoming do not include hay operations but rather use the opportunity cost or market value of hay as an input to livestock operations based on historical hay market prices [33,34]. This practice ensures that profit estimated for each enterprise of an agricultural operation recognizes the true opportunity cost of inputs [35]).

3.1.2. Center-Pivot Irrigation

Producers elsewhere in the Upper Green River Basin have transitioned from flood to center-pivot irrigation to improve irrigation efficiency and/or reduce salinity [13,36]. When the purpose of the transition is to reduce salinity, the Natural Resource Conservation Service (NRCS) will often subsidize 50% of installation costs. Less water is generally applied under center pivot than under flood irrigation, and significantly more of the water applied is consumptively used by the crop.

Center-pivot irrigation is a more efficient method of irrigation that provides greater yields due to a more even distribution of water across a field (center-pivot efficiency ranges from 60 to 80% [28]). Higher yields increase revenues, but installation and operation costs often make this system uneconomic for native hay production. The economic viability of center pivot is evaluated with and without a 50% subsidy from NRCS. Native grass hay yields might increase from 2.2 to between 3.3 and 4.4 metric tons per hectare (mt/ha) under center-pivot irrigation [37]. A range of yield estimates is included in the analysis because it is not known how native grass hay yields will respond under center-pivot irrigation in the NFID, though a doubling of yields is likely unrealistic given absent significant change in climate conditions (specifically increased growing season length).

Soil, slope, and stream characteristics limit the potential for center-pivot adoption in the NFID. The district is classified as having shallow soils that may be damaged by center-pivot systems’ weight and movement. The New Fork River is also somewhat braided; the reach of the river is made wider than it would otherwise be by the existence of smaller channels, which restrict the quantity of land available for center-pivot systems. These characteristics likely add risks and costs to individual producers that are difficult to capture in budget analyses. Only 50% of adoption of center-pivot irrigation is modeled in recognition of these factors.

3.1.3. Residential Development

Like many counties in Wyoming and other states in the Rocky Mountain West, Sublette County is at risk of ranchland conversion to residential development. Since 1998, 1130 ha has been subdivided and platted for rural residential development within the NFID boundary. Sublette County’s population nearly doubled between 2000 and 2010 due to a recent energy boom in oil/gas development, though declining energy extraction activity has more recently slowed this trend. The county’s breathtaking scenery and proximity to Yellowstone and Grand Teton National Parks will likely continue to drive rural residential development in the long term [14]. Subdivisions also place disproportionate pressure on water resources because people in the region prefer to build homes near water [38].

Half of irrigated land in NFID is assumed to be converted from hay fields to residential development in this scenario. Subdivision may result in small parcels of between 1 and 2 ha or in somewhat larger parcels of 8 to 16 ha. Residents of some subdivisions may continue irrigating hay fields, either to raise a few horses or to sell hay to nearby ranches. In these instances, return flow and stream flow patterns are virtually unaffected by the land-use change. Conversely, other subdivisions abandon water rights, and water is no longer applied in that area. Based on the pattern of subdivision development in Sublette County to date, it is assumed that 50% of subdivisions will maintain their water rights and irrigate while the remaining subdivisions will effectively fallow their land. For this scenario, one quarter of all irrigated land in NFID is therefore assumed to be no longer irrigated. All residential development requires water for household use, but this water comes from deep groundwater wells not connected with water in the New Fork River.

3.1.4. Pasture

If producers were to fallow their hay fields, the alternative agricultural land use would be non-irrigated pasture, used to graze livestock during the spring, summer, and fall months. Producers would need to purchase hay to supplement winter feed supplies or convert their ranches to stocker operations (yearlings sold in the fall) rather than cow-calf and cow-calf/yearling operations typical to the region, which provide feed (hay in winter months and grazing pasture forage in the remaining months) to livestock all year long [33,39]. Thus, for this scenario, 50% of irrigated land is assumed to be converted to pasture.

3.2. Hydrology

The Wyoming State Engineer’s Office maintained a stream gauge below the New Fork dam nearly continuously between 1992 and 2009 (except 1995 and 2003). NFID is the first point of diversion below the dam; this gauge provides the best available estimate of flows entering NFID. Streamflow levels recorded at gauges at the top of the reach are assumed to be maintained throughout NFID (in essence, this is an assumption that the streamflow levels recorded below the New Fork dam represent a natural system without diversions). Zero transmission losses from the channel of the New Fork River into the shallow aquifer are also assumed.

Diversions and return flows are not continuously monitored anywhere in the New Fork watershed. However, intensive gauge activity over a four-year period [26] can be used to estimate diversions and return flows under flood irrigation. Average monthly diversion and return flow percentages estimated in [26] (

Table 1. Impact of irrigation method on quantity and timing of flows.

	Flood Irrigation		% ∆ Flows from Flood-to-Pivot Conversion
	% Flows Diverted	% Diversions Returned to the River
June	85	10	67
July	71	39	0
August	40	12	−20
September	62	8	−19

Notes: Flood irrigation parameters are based on stream gauge data (1984–1987) from [26]. Center pivot parameters are from a water mass balance study [36] using stream gauge data (1954–2000).

) are applied to historical flows as measured at the gauge above NFID.

The conventional wisdom for high-elevation mountain valleys is that conversion from flood irrigation to center pivot may decrease diversions, thereby increasing streamflow early in the season, increasing consumptive use, and reducing return flow later in the season. Ref. [36] uses a water mass balance approach to quantify changes in flows as a result of 75% conversion from flood to sprinkler in nearby Star Valley, located approximately 65 km to the west of NFID (Table 1, final column). The same approach is applied here.

In the residential development and pasture scenarios, water that was previously diverted for irrigation remains in the stream. This reduction in diversions will likely result in some shift in the hydrograph to earlier, higher peak flows and lower late-season flows, resulting in return flows equal to 75% and 50% of the return flows associated with full flood irrigation, respectively. (Development would likely increase runoff from roads and other paved surfaces and affect riparian ecosystems other than simply through stream flow patterns. Variation in alluvial aquifer response throughout the NFID is also important, as development in certain areas may have more impact than in other areas. Estimating these potential impacts is outside the scope of this analysis). NFID is assumed to act as a unified system, with every landowner making similar land-use choices. Landowners within the irrigation district have historically tended to make similar land-use choices, and incentives to land-use patterns would tend to affect all landowners equally. The main exception is heterogeneity in land quality, which could present some landowners with more enticing options than others for real estate development. This is discussed further below.

3.3. Ecological Production Function

The impact of altered hydrology on an ecosystem is difficult to measure due to the vast number of species and ecosystem services dependent on riverine and riparian habitat, and the dynamic relationships that exist between species and ecosystem function. To overcome these difficulties, indicator species are often designated as proxies to represent the condition of the whole ecosystem [40]. Brown trout (salmo trutta) is used as an indicator species to measure the potential impacts of hydrologic change in the NFID on overall aquatic habitat.

The New Fork currently supports a productive and healthy brown trout fishery. Nearly 30 of the 110 km of the New Fork is classified by the Wyoming Game and Fish Department (WGFD) as a blue-ribbon trout fishery [38]. High base flow levels are important for brown trout in the New Fork watershed. Instream flows between 2.7 and 3.7 cubic meters per second (m³/s) are needed in segments of the New Fork to support healthy fisheries. Summer months between July and September require increased flows of 3.7 m³/s. The flood irrigation practices typical in the New Fork in 1989 (and also today) meet the instream flow needs during summer months 96% of the time [41].

Brown trout have specific habitat criteria related to hydrologic conditions, which, if altered, affect species’ population. Trout populations are estimated as a function of important habitat attributes in Wyoming streams in a log-linear regression model following [42] (Equation (1)). Total brown trout presence is $\widehat{Y},$ measured in kg/hectare contributing land. The parameter X₁ is late summer stream flow, calculated by [42] as critical period flow (CPF) divided by season average daily flow (ADF). The original variable is categorical, but here a stepwise function is generated with inflection points at the midpoints of each category to represent finer ecological response. The parameter X₂ is annual stream flow variation (see

Table 2. Habitat Quality Index parameters.

Parameter	Late Summer Stream Flow (CPF as % of ADF)	Annual Stream Flow Variation
Symbol	X1	X2
0 (worst)	Inadequate to support trout (CPF < 10 ADF)	Intermittent stream
1	Very limited; potential for trout support is sporadic (CPF 10–15% ADF)	Extreme fluctuation, but seldom dry; base flow very limited
2	Limited; CPF may severely limit trout stock every few years (CPF 16–25% ADF)	Moderate fluctuation, but never dry; base flow occupies up to two-thirds of channel
3	Moderate; CPF may occasionally limit trout numbers (CPF 25–55% ADF)	Small fluctuation; base flow stable, occupies most of channel
4 (best)	Completely adequate; CPF very seldom limiting to trout (CPF > 55% ADF)	Little to no fluctuation

Note: Two key parameters for determining habitat quality from hydrological conditions [42] are X₁, late summer stream flow, which is critical period flow (CPF) as a percentage of seasonal average daily flow (ADF); and X₂, annual stream flow variation determined by visual inspection.

). The parameter J in Equation (1) represents all other attributes included in the specification in [42]: food index, shelter index, nitrate, eroding stream banks, substrate, maximum summer stream temperature, cover, water velocity, and stream width. These other attributes are assumed to remain constant across the land-use change scenarios (this marginal benefit of water flow change (with other variables held constant) is likely a lower-bound conservative estimate of trout response).

$${log}_{10}\left(\widehat{Y}+1\right)=\left[0.807{log}_{10}\left(\widehat{X_1}+1\right)+0.877{log}_{10}\left(\widehat{X_2}+1\right)+J\right]\ast 1.12$$

(1)

3.4. Economic Benefits Transfer

The brown trout fisheries in the New Fork River support tourism by attracting anglers to the region. The same river that supports diversions for native grass hay production also supports a blue-ribbon trout stream. The private actions of landowners affect the quality of publicly accessible recreational and environmental water-based activities, including angling, through the relationship between return flows and late-season stream flows. Yet landowners do not directly account for these benefits when they make their own private resource-allocation decisions. A full comparison of the direct benefits of alternative land-use scenarios should consider how land uses change not only the measurable agricultural benefits but also nonmarket benefits of angling to residents and tourists alike. Angling studies do not exist for this region. Recommendations from [43] on how best to impute angling benefits from studies performed elsewhere are applied. (Angling value is the present study’s focus as this is most likely the primary economic benefit. Stream flow changes may also provide benefits for other wildlife. As such, these estimates of angling benefits may be a lower bound of the total environmental benefits provided by different return flow patterns).

Benefits transfer is the application of valuation studies to similar goods and services in a new location [32,42]. Benefits transfer is often used to determine how policy changes may affect the overall value of a good or service when the resources necessary to conduct a new valuation study are not available. Results from multiple studies are often considered to provide more robust value estimates [32]. The benefits transfer method reduces the time and expense associated with conducting valuation studies.

Table 3. Relevant valuation studies.

Study	Location	Value (per Day)
[22] Dalton, Bastian, Jacobs (1998)	Wyoming	USD 194
[22] Dalton, Bastian, Jacobs (1998)	Wyoming	USD 253
[23] Loomis, J. (2006)	Southwestern Wyoming	USD 117
[21] Harpman, Sparling, Waddle, (1993)	Taylor River, Colorado	USD 42
[21] Harpman, Sparling, Waddle, (1993)	Taylor River, Colorado	USD 62
Trimmed mean estimate from studies		USD 124

presents point estimates from other studies in the region. An insufficient number of relevant studies exists for deriving a statistically valid benefit function. Further, no single estimate from other study sites matches the characteristics of the Upper Green River Basin well. Thus, a measure of central tendency of the existing relevant angler benefits is presented. This approach is consistent with the recommendations and examples provided in [43] (pp. 237–240).

Results from non-market valuation studies of recreational fishing in the western USA (primarily studies from the Colorado River Basin and states adjacent to Wyoming) are used to estimate the value of recreational fishing in the Upper Green River Basin. Trout stream fishing values in these studies are variously measured in per-day, per-trip, and per-water volume units [21,22,23]. These units are converted to activity-days and adjusted to 2020 price levels (Table 3). High and low estimates are removed to calculate a trimmed mean that is equal to USD 124 per angler-day, following [43].

Economists often use a combination of stated preference with trip choice data to understand how changes in a policy or recreational site may change total benefits (see, for example, [23,44,45]). Ref. [23] provides an estimate of behavior related to change in angling days given a change in the fishery. That application closely relates to the present study in terms of recreational activity and fishery change scenarios. That study’s approach is thus used to estimate angling benefits associated with changes in brown trout presence.

4. Results

4.1. Hydrology

The first row of

Table 4. Daily flow characteristics and change in trout presence by scenario.

	Daily Flow Characteristics					Trout Presence
Scenario	Early Season Average (m3/s)	Late-Season Average (m3/s)	Full Season Average (m3/s)	Late Season to Full Season Ratio	Within-Season Variability	HQI Parameters		Brown Trout Presence (kg/ha)
						X1	X2
Gauge Above NFID	6.6	0.6	4.0	15%	0.91	-	-	-
Flood Irrigation	1.7	1.2	1.5	82%	0.52	4.12	3	45.0
Center Pivot	2.1	1.0	1.6	60%	0.53	3.82	3	42.8
Residential Development	2.9	1.1	2.1	50%	0.58	3.26	2	29.8
Pasture	4.1	0.9	2.7	33%	0.72	2.63	1	18.0

Note: Early-, late-, and full-season averages are average daily flows for 1 June–31 July, 1 August–15 September, and 1 June–15 September, respectively. Within-season variability is the coefficient of variation (standard deviation divided by mean) of daily flows over the full season. X₁ and X₂ indicate parameter values of each scenario calculated using the Habitat Quality Index model [42], summarized in Table 2. Final column indicates brown trout presence (kg/ha) in response to hydrological conditions by scenario.

indicates the average daily flows for the full season (1 June through 15 September), early season (June–July), and late season (August–15 September) at the gauge. As is typical for the region, flows are substantially higher at the start of the season due to spring snowmelt than they are later in the season, as evidenced by the late-season to full-season daily flows ratio of 15% and high within-season variability (0.91).

Flood irrigation substantially alters flows in the New Fork watershed (row 2, Table 4). For example, early in the season, 74% of flows are diverted for irrigation as producers flood their fields to saturate the soil, leaving just 1.7 m³/s in the stream. Late-season flows are increased from 0.6 to 1.2 m³/s as water diverted early in the season returns to the streambed through the alluvial soils. This late-season streamflow augmentation by return flows increases the ratio of late-season to full-season flows and reduces within-season variability.

Changes in flows associated with the conversion from flood irrigation to center-pivot irrigation are applied to historical flows measured at the gauge above NFID. For example, early season flows increase by 23% due to the conversion from flooding to sprinklers. Percentages from Table 1 are applied to the 50% of land assumed to be converted to sprinklers in this scenario. Full-season average daily flows are not significantly altered relative to the flood irrigation scenario, but early season flows are higher and late-season flows are lower (row 3 Table 4). Within-season variability is also somewhat higher than the flood irrigation scenario.

Three-quarters of the amount of water diverted under flood irrigation is assumed to be diverted in the residential development scenario (row 4 Table 2). The residual water remains in the stream, increasing the level of early season streamflow (2.9 m³/s) and decreasing the amount of late-season streamflow (1.1 m³/s) by foregoing return flows resulting from flood irrigation.

Half of NFID-irrigated land is assumed to be converted to pasture in the final scenario (final row, Table 4). The reductions in diversions and subsequent return flows lead to relatively high early season flows (4.1 m³/s) and low late-season flows (0.9 m³/s). This scenario most closely mimics the flow patterns that would have prevailed pre-development: low late-season to full-season ratio and high within-season variability. Increasingly higher levels of pasture conversion would approach the pre-development flow patterns reflected in the first row of Table 4.

4.2. Changes in Habitat and Recreational Angler Value

Most important for the brown trout habitat are the late-season to full-season streamflow ratio (a larger percentage is better for the species) and within-season streamflow variability (less variability is preferred). Flood irrigation late-season streamflow is high due to return flows returning to the system late in the season. Streamflow fluctuations exist but are relatively small. These conditions yield the HQI parameter values and predicted brown trout presence shown in the final three columns of

Table 5. Change in trout value as a result of habitat change.

Scenario	Trout Presence		Fishing Days		CS	CS/ha
	(kg/ha)	% ∆	Number	% ∆	(USD 1000 s)	(USD)
Flood Irrigation	45	-	2359	0.00	USD 293	USD 50
Center Pivot	43	−4%	2292	−3%	USD 284	USD 48
Residential Development	30	−33%	1856	−21%	USD 230	USD 39
Pasture	18	−60%	1453	−38%	USD 180	USD 31

Notes: Trout presence (and changes from the base flood irrigation scenario) is calculated using the brown trout Habitat Quality Index [42]. Fishing days (and changes from the base flood irrigation scenario) are calculated using impacts on angler-day traffic from [23]. Total CS is total consumer surplus associated with all angler activity on the NFID.

(this specification predicts well for the New Fork; the measured presence was also 45 kg/ha. This specification also predicts well for other streams in Wyoming: R² = 0.985 for the 20 streams included in the regression and R² = 0.901 for an additional 16 streams not included in the regression).

The parameter and predicted brown trout presence values for the three alternative scenarios are also shown in the final three columns of Table 4. The conversion from flooding to sprinklers increases base flows slightly but not substantially, and late-season flows are only slightly lower (reflecting reduced return flows returned to the system). Although the late-season to full-season ratio is lower and within-season variability is higher than in the flood irrigation scenario, neither change, based on average conditions, changes the HQI significantly. This may reflect the coarseness of the HQI rather than a lack of impact on the brown trout habitat. The residential development alternative has higher base flows than the flood irrigation scenario, reflecting lower diversion rates. However, it does not score as well as flooding or sprinklers in the HQI due to its substantially lower late-season to full-season flow ratio and higher within-season variability. Base flows in the pasture scenario are even higher than in the residential development scenario because no diversions occur on any of the converted lands (unlike residential development, where half of the converted land continues to be flood-irrigated).

Table 5 indicates the angling benefits associated with changes in brown trout presence. Table 5, columns 1 and 2, contain brown trout presence estimates from Table 4 and the associated percentage change from the baseline flood irrigation scenario. Absent information on how popular or productive the New Fork is relative to other streams and rivers in Wyoming, angler-day traffic is imputed to the 110 km of the New Fork under the baseline flood irrigation scenario on a proportional basis: 2359 angler days per year on the New Fork. (The number of angler-days per year in Wyoming is 2.696 million, 45% of which are on rivers and streams (rather than lakes and reservoirs [46].) The total number of Wyoming perennial stream kilometers is approximately 58,000 [47]). Ref. [23] estimates from angler survey data that a 100% improvement in the trout catch rate results in a 64.5% increase in angler use and corresponding economic impact in stretches of the Snake River located in Southwestern Wyoming and southeastern Idaho (approximately 80 km northwest of the New Fork) (the Loomis study accounts for an increase in angler-days by existing anglers but not the possibility of new anglers). The change in catch rate is assumed to correspond proportionally to the change in brown trout presence (although the catch rate may vary based on the skill of anglers). This relationship is applied to the brown trout’s presence to determine the reduction in the number of fishing days associated with each scenario (Table 5, columns 3 and 4).

The final four columns of Table 5 provide estimates of the consumer surplus for the entire New Fork and on a per-hectare basis for each scenario (assuming that benefits are spread proportionately across the New Fork). Consumer surplus is estimated to be USD 293,000 in the flood irrigation scenario (USD 50/ha). Consumer surplus under the pivot scenario is diminished from the flood irrigation scenario only slightly (USD 48/ha); changes in consumer surplus under the residential and pasture scenarios are more significant (USD 39/ha and $31/ha, respectively).

4.3. Economic Outcomes

Agricultural producers consider net returns when they make land-use decisions. Returns from flood-irrigated native grass hay [48,49] less cash costs (primarily ditch maintenance and harvest costs, insurance, fees, and taxes) and non-cash costs (primarily machinery and equipment depreciation) are USD 153/ha (the solid gray portion of bar 1, Figure 3). This is the private benefit of flood irrigation to NFID producers in the longer run. (Producers may tend to focus on returns over variable costs in the short term, but if their land use changes occur over multiple consecutive years, they will shift their focus to thinking about the returns over total costs presented in Figure 3.)

Center pivot irrigation costs are adapted from an Oregon enterprise budget [50]. Energy costs are higher under center-pivot irrigation than flood irrigation, though labor requirements are lower. Even with a subsidy, agricultural net returns over total costs are only USD 17/ha for the 3.3 mt/ha yield scenario (solid gray portion of bar 2, Figure 3). If producers were able to achieve a 4.4 mt/ha yield, returns would increase to USD 183/ha (solid and striped gray portions of bar 2, Figure 3), which is financially preferable to flood irrigation. NFID producers currently do not use center pivot irrigation due to limitations in soil, slope, and stream characteristics. If changes in climatic, technological innovation for irrigation, and/or economic conditions improved the probability of higher yields and reduced risks with implementing center-pivot irrigation, this could change in the future.

Absent publicly available information on real estate transactions in the area, the value of residential development is assumed to be equal to the going rate for renting irrigated cropland in the region, USD 203/ha [51] (solid gray portion of bar 3 Figure 3), though prices will vary based on land proximity to natural amenities [52]. The per-hectare value of land sold for the purpose of residential development could be significantly higher or lower than this estimate. Regardless, residential development will continue to occur on the New Fork and throughout the Upper Green River Basin, replacing agricultural land over time.

The current value of pasture land is assumed to be equal to the going rate for renting pastureland in the region, USD 12/ha [51] (the solid gray portion of bar 4, Figure 3). Wyoming and other states in the Colorado River Basin are currently considering a voluntary and compensated water conservation program (the pilot is called the System Conservation Pilot Program [SCPP]) as a strategy for addressing water availability concerns caused by steadily increasing demand and reduced supply projections [15,16]. In 2018, SCPP compensation for estimated reductions in consumptive water use averaged USD 1216/ha-m (USD 526/ha). No NFID producers have applied to participate in the SCPP, demonstrating a revealed preference for flood irrigation even at extremely high prices, likely due to the livestock operation risks associated with reduced hay production. Depending on future climatic and economic conditions, net returns from water conservation program participation may approach levels sufficient to entice NFID producers to participate. Potential net returns for water conservation of USD 250/ha are included in Figure 3 (striped gray portion of bar 4) in recognition of this potential additional revenue stream.

5. Angler Impacts

Given current climatic, hydrologic, and economic conditions (solid gray bar portions Figure 3), producers interested in remaining in agriculture should generally prefer flood-irrigated agricultural production, though cost-shared center pivot, residential development, and pasture may be preferred where feasible or if producers already plan to leave agriculture for other reasons. Flood-irrigated agricultural production is also the preferred land use from the perspective of anglers, based on the brown trout habitat quality information currently available. Trout presence in the flood irrigation scenario is estimated to be 45 kg/ha (black line Figure 3) for an estimated economic recreational angler value of USD 50/ha (dotted bar portions in Figure 3). Any decrease in flood irrigation in the NFID will decrease the value to anglers, given the interconnectivity between irrigation and species habitat.

In the future, there may be increases in returns to center-pivot irrigation and/or pasture (striped bar portions, Figure 3) that increase private incentives to transition land use away from flood irrigation. The total economic value of flood irrigation and residential development would remain at USD 153/ha and USD 203/ha, respectively, while residential development and pasture would increase to USD 183/ha and USD 262/ha, respectively.

Significant angler benefits exist in the return flows generated by flood irrigation. The addition of recreational angler value to net returns from private land use alters the value gaps between the alternatives, but only marginally. The total economic value of flood irrigation becomes USD 203/ha, that of a center pivot is USD 231/ha, that of residential development is USD 242/ha, and that of pastures is USD 293/ha. The addition of ecosystem service value does not change the overall ranking of the alternatives compared to the long-term private net returns (including potential future improvements for center-pivot irrigation and pasture). The economic value of other fisheries or recreational activities would need to be incorporated before the ranked order of alternatives would vary from the perspective of private net returns.

If policymakers sought to slow the transition of land from flood-irrigated agriculture to other land uses to protect brown trout habitats and other ecosystem services, they could provide a subsidy or credit equal to brown trout angler benefits for keeping land flood-irrigated. They could also facilitate landowner efforts to capture some of the surpluses through fishing leases or per-rod fees on private land. However, internalizing angler benefits in this fashion is unlikely to be sufficient to keep land flood-irrigated in the presence of significant pressure for residential development or pasture. This is true even if the largest per-hectare recreational angler value from Table 5 was used to estimate angler benefits. (At $253/angler-day, per-hectare angler benefits for the flood, pivot, development, and pasture scenarios would increase to USD 101, USD 98, USD 79, and USD 62, respectively.)

6. Conclusions

This study provides a framework for quantifying how land-use changes and associated return flow patterns affect the economic value of water across uses in a hydrologically connected, shallow alluvial aquifer system. The analysis was conducted for an area in Wyoming, USA, facing considerable pressures and potential policy changes that may alter water use patterns significantly in the future. The inclusion of recreational angler values with agricultural values alters the magnitude of returns, though it does not alter the overall ranking of alternatives. These results highlight the potential heterogeneity of conclusions to be drawn regarding water use efficiency, depending on the degree of hydrologic connectivity and the economic value of water in different uses; in some instances, flood irrigation practices, though inefficient, may yield a higher total economic value than other irrigation technologies or water uses.

This study also highlights gaps in existing data availability and modeling needs. First, the framework is applied to a single irrigation district for which reliable return flow data are available. The expansion of the study to a larger portion of the high-mountain valleys that constitute the headwaters of the Colorado River Basin would be of use to water managers and policymakers. Improved hydrologic data (including continual datasets for stream flows) would be needed to expand the geographical scope of the study.

Second, this study focused on the impact of changes in hydrologic conditions on brown trout anglers because a science-based habitat quality metric exists for brown trout. In reality, private land and water use decisions likely affect a broader range of recreational users than simply brown trout anglers. Future studies could be expanded to include other aquatic and avian species (through bird-watching and artificial wetland provision for migratory species) along with the agricultural returns and brown trout angler benefits from different land uses included here. Such future studies could also be broadened to recognize the possibility that some species and ecosystem services may benefit from reduced irrigation diversions.

Finally, future studies of alternative land-use impact on return flow patterns and recreational angler value in the region could also consider indirect community impacts on the local regional economy in the form of jobs and income. The present study demonstrates that changes in hydrologic function, land-use patterns, and return-flow patterns affect the number of angler days in the region. It follows that tourism revenues from fishing—and other recreational activities such as hunting, wildlife-watching, and birding if they were also to be included—would also be affected.

This study nonetheless provides a framework for incorporating social tradeoffs associated with flood irrigation into water resource decision making and for internalizing externalities on a landscape where the implications of private decision making for social benefits can be significant. Improved data and models and the incorporation of the full scope of economic losses and benefits associated with water management changes would enhance our understanding of how agricultural and non-agricultural water uses interact. This is especially important in a semi-arid state like Wyoming where water is scarce and the pressure to allocate scarce resources effectively is increasing.