Potential Effects of Drought on Carrying Capacity for Wintering Waterfowl in the Central Valley of California

The Central Valley of California supports one of the largest concentrations of wintering waterfowl in the world despite loss of > 90% of its historic wetlands (Heitmeyer et al. 1989; Fleskes 2012). For instance, about 60% of the Pacific Flyway's waterfowl winter in the Central Valley, with a third or more of North America's northern pintail (Anas acuta) and almost all the continental population of tule white-fronted geese (Anser albifrons elgasi) and Aleutian cackling geese (Branta canadensis leucopareia) wintering in the region (Gilmer et al. 1982). The 6–7 million waterfowl that winter annually in the Central Valley rely upon a mix of wetland and agricultural food resources to meet their energetic needs. These foods are primarily obtained from managed seasonal wetlands as well as rice, corn, and other agricultural fields. Rice fields that are flooded after harvest (i.e., winter flooded) to decompose rice straw provide many of the same habitat values for waterfowl as the wetlands that they replaced (Elphick and Oring 1998; Eadie et al. 2008). Management of most Central Valley wetland and agricultural habitats for waterfowl require surface water delivered by the federally operated Central Valley Project and the State Water Project. Water supplies are limited and fully allocated for urban and agricultural use, management of wetlands, and maintainenance of in-stream flows to address water quality in the Sacramento–San Joaquin River delta and support various fisheries. Water supply shortages have the potential to negatively affect waterfowl habitats, reduce waterfowl food resources, and lower the carrying capacity of the Central Valley for wintering waterfowl, making it more difficult to meet continental waterfowl population goals established by the North American Waterfowl Management Plan (NAWMP).

In the summer of 2014, faced with a third consecutive year of drought in an 8-y period having seven below-average to critically dry water years (California Department of Water Resources 2015), state, federal, and private wetland managers in California met to discuss options for minimizing impacts of impending water supply reductions on wintering waterfowl. Reductions of surface water supplies from the Central Valley Project and State Water Project in 2014 were projected to affect both wetland and agricultural habitats important to wintering waterfowl in the Central Valley by reducing the area of 1) managed seasonal wetlands flooded from early fall through early spring, 2) managed seasonal wetlands that receive summer irrigation treatments to increase production of waterfowl foods, 3) planted rice and corn fields, and 4) winter-flooded rice fields. Actual impacts would depend upon final water supply allocations and when those water supplies were applied to which habitats. A desired outcome from the meeting was to tailor use of water supplies targeted for waterfowl habitat to best provide habitat in locations and during times needed to meet the bioenergetic needs of wintering waterfowl. A consensus among managers at the meeting was the need to evaluate a range of scenarios that varied in timing, type, and area of habitats to determine which, if any, met the bioenergetic needs of the Central Valley wintering waterfowl population.

Our objective was to evaluate the potential effects of the California drought on waterfowl carrying capacity in the Central Valley by modeling a range of plausible drought scenarios. These scenarios were informed by our understanding of likely water restrictions in 2014, as well as interviews with public and private wetland managers and water contractors that supply water for wetland and agricultural habitats important to waterfowl. Our intent was to help waterfowl managers understand how water shortages in the Central Valley may affect waterfowl carrying capacity during the current drought and into the future.

The Central Valley of California was regionally comprised of the Sacramento Valley in the north, the San Joaquin Valley in the south, and Suisun Marsh and Sacramento–San Joaquin River delta (delta) east of San Francisco Bay. Agricultural and urban development reduced the estimated 1.6–2 million ha of original wetlands in the Central Valley by > 90% by the early 1900s (USFWS 1978; Gilmer et al. 1982; Fleskes 2012), with the magnitude of loss and types and amounts of remaining waterfowl habitat differing among regions. Wetland loss was extensive throughout the Central Valley; however, many wetlands in the Sacramento Valley were converted into rice fields and in the delta to corn or other grain fields that retained high value to wintering waterfowl, especially when flooded after harvest.

Most Central Valley wetlands were managed so they were dry during summer, except for periodic irrigations to promote seed production, and flooded during fall and winter. Most managed flooding of wetlands occurred from late August through late November, varying regionally and annually depending upon the opening date of duck-hunting season. Winter flooding of rice fields varied annually depending upon timing of harvest, surface water availability, and start and end dates of duck season. Before 1989, winter flooding of rice fields in the Sacramento Valley was almost exclusively for the purpose of duck hunting. After 1991, increasing restrictions on rice straw burning required by the 1991 California Assembly Bill 1378 (California State Assembly Bill 1378; Hill et al. 1999) resulted in a growing number of fields flooded after harvest for straw decomposition (Fleskes et al. 2005a). Managed postharvest flooding of cropland in the delta was mostly winter flooding of harvested corn and small amounts of rice and other crops for waterfowl hunting. Some postharvest flooding of (mostly nonrice) croplands also occurred in the San Joaquin Valley, although this area has declined since the 1970s (Fleskes et al. 2013).

We used the bioenergetics model TRUEMET (Central Valley Joint Venture 2006; hereafter CVJV 2006) to evaluate the impact of California's drought on food supplies for wintering waterfowl in the Central Valley of California under a range of habitat and waterfowl population scenarios. We compared waterfowl food supplies and waterfowl food demand for 10 scenarios: 1) recent nondrought conditions and two projections of waterfowl abundance and 2) eight drought scenarios where foraging habitats varied by amount, temporal availability, and food biomass provided, with an assumed abundance of wintering waterfowl. We present duck results for all scenarios, with goose results only for selected scenarios. Our emphasis on ducks when evaluating the effects of the California drought on waterfowl aligns with the CVJV's attempt to ensure adequate food resources for dabbling duck species such as northern pintails, which are well below established population objective levels, and less emphasis on geese, which are well above population objective levels (U.S. Fish and Wildlife Service 2014).

The CVJV was established in 1989 under the auspices of the NAWMP. (Joint Ventures are self-directed partnerships of public and private entities that have accepted responsibility for implementing the NAWMP within their geographic area of responsibility, including the development of conservation implementation plans for waterfowl.) The CVJV completed its most recent conservation plan in 2006, which established waterfowl population abundance objectives for the Central Valley, as well as landscape conditions needed to support these populations. The biological assumptions and data used in establishing population and habitat objectives for the Central Valley are fully described in the CVJV's plan (CVJV 2006); we believe the CVJV's planning efforts for waterfowl provide the best context for evaluating the potential effects of drought on waterfowl carrying capacity.

The CVJV implementation plan previously described the amount of waterfowl habitat in the Central Valley available during nondrought years and evaluated the capacity of these habitats to meet the energetic needs of waterfowl populations in the Central Valley from mid-August through late March (CVJV 2006). To evaluate waterfowl carrying capacity on the basis of food resources, the CVJV used the bioenergetics model TRUEMET (CVJV 2006). Conceptually, TRUEMET was built upon the framework of a daily ration model (Goss-Custard et al. 2003) that assumes that birds are ideal free foragers and do not incur costs associated with traveling between habitat patches. Most estimates of waterfowl energetic carrying capacity have relied on a daily ration model in which the total food energy required by a defined population of birds is first estimated. Total food energy available to this population is then estimated from foraging habitats included in the model (Pearse and Stafford 2014; Williams et al. 2014). Energetic carrying capacity is often expressed as a maximum value within a daily ration model, regardless of when food resources become available (Reinecke and Loesch 1996). This expression is appropriate where maximum food energy occurs at the beginning of the time period being modeled; however, waterfowl often experience temporal variability in food supply availability (Williams et al. 2014). For example, food resources in managed wetlands may become available over an extended period, either because of the time needed to fully flood these habitats or differences in flooding schedules among wetlands. Daily ration models that do not incorporate temporal variation in food availability may fail to represent the dynamic relationship between population energy demand and energy supply through time. Such models implicitly assume that if a food resource is available at any point during the planning period it is functionally available during the entire planning period, which reduces the ability to detect food deficits that may occur over shorter time frames within the planning period.

In contrast to some daily ration models, TRUEMET allows the user to define when foraging habitats become available within the time period being modeled. As a result, the relationship between population energy demand and energy supply can be examined for any point in time for multiple foraging guilds, and exploitive competition for food resources among foraging guilds can be accounted for (e.g., the effects of goose consumption on dabbling duck food resources is accounted for in all period-specific estimates of dabbling duck energy supply). There are eight explicit inputs required for each TRUEMET model run: 1) number of days or time periods being modeled within the overall season of interest, 2) population objectives or estimates for each waterfowl foraging guild within each time period, 3) daily energy expenditure of a single bird in each foraging guild within each time period, 4) habitat types used by each waterfowl foraging guild to satisfy daily energy requirements, 5) area and availability of habitat types during each time period, 6) biomass of food in each habitat type at the start of the overall season of interest, 7) nutritional quality (i.e., true metabolizable energy content), and 8) decomposition rate of each food type (Table 1).

Within TRUEMET, the total energy demand (TED; in kcal) of a foraging guild in a time period is calculated as:

where TED_jk = total energy demand of foraging guild j in time period k, POP_jk = population size of foraging guild j in time period k, D_k = number of days in time period k, and DEE_jk = daily energy expenditure (kcal) of an average bird in foraging guild j in time period k. The total energy supply (TES; in kcal) available to a foraging guild in a time period is calculated as:

where TES_jk = total energy supply available to foraging guild j in time period k, and NEFH_ijk = net energy available in foraging habitat i to foraging guild j at the beginning of time period k. This equation assumes that foraging guild j has been given access to foraging habitat i within the model.

The TRUEMET model requires the user to identify the maximum area of foraging habitat (FH_i) possible within the time frame being modeled. This habitat is placed in a “reservoir” where it can be made available incrementally over time by the user, including releasing all of it in a single time period. For example, managed wetlands can be released from the reservoir at a rate that reflects their flooding schedule. Conversely, foraging habitats can be retrieved by the model and placed back in the reservoir where they are no longer available to the birds (e.g., where managed wetlands are drained before the final time period). The rate at which a foraging habitat is released from the reservoir or retrieved is dependent on user inputs that define the availability of this foraging habitat over the time (i.e., the user builds “availability curves” within the model). Thus, NEFH_ijk is a function of the cumulative sum of food energy released from the reservoir before and including time period k, the cumulative sum of waterfowl food consumption and food decomposition that occur in time periods before k, and the cumulative energy of foraging habitat i returned to the reservoir in time periods before k (e.g., due to drying conditions). The model calculates NEFH_ijk as follows:

where EFH_ijk = the energy of foraging habitat i released from the reservoir at the beginning of time period k to which foraging guild j has access, CFH_ik = total consumption of food energy in foraging habitat i during time period k, DFH_ik = decomposition of food energy in foraging habitat i during period k, and R_ik = energy of foraging habitat i returned to the reservoir at the end of time period k (e.g., due to drying conditions). The model calculates EFH_ijk as follows:

where FBFH_i = the food biomass per unit area of FH_i that resides in the reservoir (i.e., starting condition), MEFH_i = the true metabolizable energy (e.g., kcal/g) of foods provided by FH_i, and HFH_ijk = area of FH_i released from the reservoir at the beginning of period k to which guild j has access. TRUEMET calculates CFH_ik as follows:

where CFH_ik = consumption of food energy in foraging habitat i in period k by all guilds having access to habitat i. Finally, TRUEMET calculates DFH_ik as follows:

where DFH_ik = decomposition of food energy (kcal) in foraging habitat i in period k, TEFH_ik is the total energy of foraging habitat i that exists outside the reservoir in period k, and DRFH_ik is the decomposition rate applied to the food in foraging habitat i in period k expressed as a fraction.

The equation for CFH_ik illustrates an important assumption of the model. For each time period, birds in a foraging guild are assumed to consume a food in proportion to its availability, where availability is defined in energetic terms. For example, assume that birds in a duck guild are given access to managed wetlands and that this foraging habitat provides 40% of all the food energy available to ducks in time period k. Within time period k, ducks would meet 40% of their food energy needs from managed wetlands (if TED_jk > TES_jk, then the food resources provided by managed wetlands would be completely exhausted within time period k, though this foraging habitat could provide food energy in future time periods if additional managed wetlands were made available in these future periods).

The assumption that foods are consumed in proportion to their contribution to total food energy may be violated in some model scenarios. Birds may show some selection in the foods they eat, and thus deplete some foods at a faster or slower rate than what would be predicted by relative energy abundance alone. Most applications of the model are more concerned with the total energy available to a guild in each time period, as opposed to accurately predicting how quickly a given foraging habitat is depleted. The biological assumption is that birds will switch to less favored foods as more desired foods are depleted. However, our ability to accurately model food energy for each foraging guild using TRUEMET is strongly dependent on our understanding and assumptions about how foraging guilds overlap in their use of habitats, and the exploitive competition for food resources that results from this overlap. Thus, careful consideration must be given about the habitats that are assumed to be used by each foraging guild.

Migrating and wintering waterfowl were present in the Central Valley from mid-August through the end of March. The CVJV modeled waterfowl population energy demand and food energy supply at 15-d intervals between August 15 and March 28. We used the same 15-d time intervals here to be consistent with CVJV planning when modeling various drought scenarios.

The CVJV recognized three waterfowl foraging guilds; 1) ducks, 2) dark geese, and 3) white geese. Approximately 92% of all ducks were dabbling ducks, whereas the remainder were diving ducks. Diving ducks were pooled with dabbling ducks in the TRUEMET model to account for their potential competition for food resources with dabbling ducks, especially wetland plant seeds in managed seasonal wetlands. Semipermanent and permanent wetlands that are more closely associated with diving ducks make up only 10 to 15% of all wetland habitat in the Central Valley (CVJV 2006). Dark geese included white-fronted geese (Anser albifrons), western Canada geese (Branta canadensis moffitti), and Aleutian cackling geese. White geese included lesser snow geese (Chen caerulescens) and Ross's geese (C. rossii). Tundra swans (Cygnus columbianus) were also included in the white goose foraging guild (CVJV 2006). However, recent work on the diets of dark geese in the Central Valley suggests that dark and white geese should not be separated into different groups on the basis of food consumption (Skalos 2012). As a result, we combined all dark and white goose species, as well as tundra swans, into a single foraging guild.

Duck population objectives for each 15-d interval represent the number of birds that are expected to winter in the Central Valley when continental breeding duck populations are at NAWMP goals. These continental breeding population objective goals were translated into mid-winter (early January) population objectives for Joint Ventures that support migrating and wintering waterfowl, including the CVJV (Reinecke and Loesch 1996; Koneff 2003). The CVJV mid-winter objective was combined with information on duck migration chronology in the Central Valley (Fleskes et al. 2005b) to establish duck population objectives at 15-d intervals between August 15 and March 28 as described in the CVJV implementation plan and elsewhere (CVJV 2006; Petrie et al. 2011); we adopted those 15-d population objectives here (Table 2). We chose duck population objectives, not population estimates, to evaluate the effects of drought for two reasons. First, comprehensive surveys needed to estimate duck populations over this time period are not available for the Central Valley (i.e., monthly or biweekly surveys are not currently conducted). Second, the CVJV has focused on creating the landscape conditions necessary to support duck populations in years when these populations reach NAWMP goals (CVJV 2006). This “setting the table” approach to planning is common among Joint Ventures, and it provides a framework for evaluating the effects of the drought in the larger context of an international waterfowl plan.

Many North American goose populations have exceeded their population objectives (U.S. Fish and Wildlife Service 2014), and Joint Ventures have been advised to use recent goose counts when developing implementation plans (Koneff 2003). The CVJV used direct counts of geese to estimate the number of geese and swans in the Central Valley from late August through late March (Fleskes et al. 2005b; CVJV 2006). However, the number of geese wintering in the Central Valley has increased over the past decade, and some species, such as white-fronted geese, are now migrating earlier into the Central Valley (Skalos 2012). As a result, we updated goose population objectives for each 15-d interval using recent counts of geese (Olson 2015) and information on goose migration chronology (Fleskes et al. 2005b; Pacific Flyway Council 2006; Petrie et al. 2011; USFWS unpublished data; Table 2).

Most Joint Ventures estimate the DEE of ducks, geese, and swans by multiplying the resting metabolic rate (RMR) of an “average” bird by a factor of three to account for the energy costs of free living (Williams et al. 2014). We used the following equation from Miller and Eadie (2006) to calculate the RMR for geese and swans, and multiplied RMR by a factor of three to estimate the daily energy intake of an average bird within the goose/swan foraging guild:

Body mass estimates for geese and swans were obtained from Bellrose (1980); adult mass was used to provide conservative estimates of DEE as age ratios for goose populations in the Pacific Flyway are usually skewed in favor of adults (Bellrose 1980). Because the relative abundance of species included in the goose foraging guild varied by time period (Table 2), we calculated a weighted body mass for all time periods on the basis of the relative abundance of each species in each time period. Finally, we converted kJ to kcal by dividing the latter by 4.18 (Table S1).

The CVJV did not use an estimate of RMR to estimate DEE for ducks. Instead, they relied on the period-specific estimates of DEE for pintails by Miller and Newton (1999) between August and March that were derived from pintail body mass and carcass composition in the Central Valley, and we adopted those same estimates of DEE here. Weighted body mass for ducks in the Central Valley is 0.86 kg (CVJV 2006), which is similar to pintails (0.94 kg) that make up 46% of the duck objective included in the TRUEMET model (CVJV 2006; Table 3; Table S1).

The CVJV assumes that ducks in the Central Valley rely on three major foraging habitats, including managed seasonal wetlands, harvested rice fields that are winter flooded, and harvested grain corn fields that are flooded and unflooded (CVJV 2006). However, some species of ducks that winter in the Central Valley may only meet a portion of their food energy needs from these foraging habitat types. Population objectives for these species were modified for use in the TRUEMET model because failing to do so may overestimate the impact these species have on food resources that were included in the model (CVJV 2006; Table 3). Finally, geese were assumed to forage in harvested rice fields and harvested grain corn fields regardless if they are flooded, and believed to use wetlands mostly for roosting purposes. We summarized data on the area of each of these foraging habitat types under nondrought conditions, and estimated the area of foraging habitat that would be available to waterfowl under several drought scenarios (Table 4).

To estimate how drought might affect the amount of managed seasonal wetlands and winter-flooded rice available to waterfowl during fall–spring 2014–2015, we considered surface water delivery restrictions described in the Central Valley Project and State Water ProjectDrought Operations Plan and Operational Forecast (2014), consulted water contractors that traditionally supply water for wetlands and winter flooding of rice, and interviewed public and private wetland managers about their understanding of likely water availability in 2014–2015. Wetland managers and water contractors were interviewed throughout the Sacramento and San Joaquin valleys, the Suisun Marsh, and the Sacramento–San Joaquin River delta, and asked to estimate: the area of managed wetlands that would be flooded within their geographic area, what percentage of these wetlands would be summer irrigated to promote seed production, and whether these habitats would be flooded over the same duration as in nondrought years. These same individuals were also asked to estimate the amount of rice that would be winter flooded in 2014–2015. As a result of these investigations, we assumed that 25% of all managed seasonal wetlands would be dry during 2014–2015 because of insufficient surface water supplies (Table 4). However, our interviews with wetland managers indicated that traditional flooding schedules would be maintained for most managed wetlands that received water in 2014. As a result, we adopted the CVJV's flooding curve for wetlands when modeling drought scenarios (Table S2).

An average of 225,647 ha of rice was harvested in the Sacramento Valley between 2011 and 2013 (U.S. Department of Agricultural National Agricultural Statistics Service 2015). Approximately 63% of harvested rice fields are winter flooded in nondrought years, or an estimated 142,158 ha (Point Blue Conservation Science, unpublished data). Of the approximately 83,500 ha of harvested rice fields that are not winter flooded, the CVJV assumes that 25% of these hectares are deep plowed and provide no waterfowl food resources (CVJV 2006). This leaves nearly 63,000 ha of harvested rice fields that are not winter flooded or deep plowed after harvest in a nondrought year (Table 4).

Harvested rice in 2014 declined to an estimated 171,728 ha because of the drought (U.S. Department of Agricultural National Agricultural Statistics Service 2015); we estimated that winter-flooded rice fields would also decline to between 20,235 and 30,351 ha in 2014–2015 on the basis of the professional judgment of water contractors. If the CVJV's current assumption about deep plowing was maintained, we forecasted that between 37,873 and 35,344 ha would be deep plowed in 2014 depending on the level of winter flooding. A lack of surface water supplies for winter flooding could encourage some rice producers to consider alternative methods for decomposing straw, including more deep plowing of fields, so we included model scenarios where deep plowing rates exceeded the CVJV's current assumption.

The CVJV assumes that winter flooding of rice begins in early October, with most fields flooded by mid- to late December, and we maintained that assumption here. “Maintenance” water is usually applied to winter-flooded rice throughout fall and winter to maintain favorable water depths. In fall 2013, some rice fields received no maintenance water after October because of restricted surface-water supplies, and these fields were dry by late November or early December. To account for this possibility, we included some model scenarios that applied this abbreviated flooding schedule to 20,235 ha of harvested rice (Table S2). These 20,235 ha were in addition to the 20,235 to 30,351 ha for which maintenance water was assumed available.

Most of the harvested grain corn used by waterfowl in the Central Valley is grown in drainage basins north of the San Joaquin Valley. Between 2011 and 2013, an average of 50,479 ha of grain corn was grown in these drainage basins (U.S. Department of Agricultural National Agricultural Statistics Service 2015). The CVJV assumes that 50% of these hectares provide no waterfowl food resources because of postharvest practices (e.g., plowing, deep disking; CVJV 2006). Thus, on average, an estimated 25,240 ha of harvested grain corn provided food resources to waterfowl between 2011 and 2013. Harvested grain corn declined to 28,450 ha in 2014 because of the drought (U.S. Department of Agricultural National Agricultural Statistics Service 2015); we maintained the CVJV's assumption that half of these 28,450 ha provide no waterfowl food for all drought scenarios (Table 4).

We used waterfowl food biomass estimates for managed seasonal wetlands and harvested grain corn fields presented in the CVJV Plan (2006), but updated those estimates for rice habitats on the basis of more recent information (Fleskes 2012). We also slightly adjusted food biomass estimates for managed seasonal wetlands after reviewing the study on which these estimates were based (Naylor 2002). Approximately 56% of the managed seasonal wetlands that were sampled by Naylor (2002) were summer irrigated, and moist-soil seed biomass in these irrigated wetlands was nearly 60% greater than in nonirrigated wetlands. On the basis of results of wetland manager interviews, we assumed that only 10% of managed seasonal wetlands would be summer irrigated in 2014, and adjusted our estimate of average moist-soil seed biomass in managed seasonal wetlands to account for this decline in summer irrigation when modeling drought scenarios (Table 5). Waterfowl do not consume all the foods available in wetlands because foraging efficiency declines with decreasing food biomass (Reinecke and Loesch 1996). As a result we adopted the CVJV's “foraging threshold” of 37 kg/ha, which represented the amount of food remaining in managed seasonal wetlands in the Central Valley at the end of March (Naylor 2002). The CVJV also applied this foraging threshold to agricultural habitats, as we did here, as values specific to these habitats were lacking. The nutritional quality, or true metabolizable energy, of waterfowl foods was also taken from the CVJV Plan (Table 5). We also used estimated decomposition rates for seeds in managed wetlands and rice and corn fields from the CVJV Plan, which are based on earlier work by Nelms and Twendt (1996) and Naylor (2002).

Although seed production from moist-soil plants accounts for most of the food energy available to ducks in managed seasonal wetlands, invertebrates can make up 25% of the diet from January through March (Euliss and Harris 1987). To recognize the potential importance of invertebrates during late winter, the CVJV assumes that managed seasonal wetlands provide 31 kg/ha beginning January 1 (CVJV 2006). This estimate is based on late winter estimates of invertebrate biomass for seasonal wetlands in the Mississippi Alluvial Valley (Manley et al. 2004; Table 5). Although winter-flooded rice undoubtedly provides some invertebrate resources as well, the CVJV chose not to include these foods in the model. Most managed wetlands in the Central Valley remain flooded through March. In contrast, many winter-flooded rice fields are quickly dewatered in late January after close of the hunting season (Table S2), which leads to considerable uncertainty about the invertebrate food resources provided by these habitats.

We modeled 10 scenarios for ducks in the Central Valley (Table 6). Scenario 1 (i.e., Current Hab–Goal Pop) reflected our current understanding of habitat conditions in a nondrought year and assumed that ducks using the Central Valley are at NAWMP goals (Table 2). Scenario 2 (Current Hab–80% Pop) also assumed Central Valley habitat conditions during a nondrought year but reduced the overall population objective for ducks in the Central Valley by 20% because pintails were below their NAWMP goal in 2014. Pintails make up 46% of the CVJV duck population assumed in the TRUEMET model (Table 3), and the species was at 58% of its NAWMP goal in 2014 (U.S. Fish and Wildlife Service 2014). Although this pintail deficit may have argued for an even greater reduction in the duck objective (i.e., greater than 20%), some species like green-winged teal (Anas crecca) were actually above NAWMP goals in 2014 and we believed the 20% reduction was reasonable. Scenarios that assumed duck populations were at 80% of NAWMP goals may have better represented the 2014/2015 drought, whereas scenarios assuming NAWMP goal duck populations may be most useful for CVJV planning. Our updated estimates of the number of geese and swans wintering in the Central Valley were included in all TRUEMET simulations (Table 2), which did not differ by model scenario.

We considered the possible effects of the California drought on duck habitat conditions in eight additional scenarios. Scenario 3 (20K–Goal Pop) assumed that 20,235 ha of all harvested rice fields will be winter flooded, and maintained current assumptions for deep plowing of rice (i.e., 25% of all harvested rice fields that are not winter flooded are deep plowed and provide no food resources). This scenario also assumed that ducks were at NAWMP goals. Scenario 4 (20K–80% Pop) was identical to scenario 3, except that ducks were at 80% of NAWMP goals. Scenario 5 (30K–Goal Pop) assumed that 30,353 ha of all harvested rice fields were winter flooded and also maintained assumptions for deep plowing of rice. Scenario 6 (30K–80% Pop) was identical to scenario 5, except that ducks were at 80% of NAWMP goals. Scenario 7 (20K–2× Plowed–80% Pop) assumed that 20,235 ha of all harvested rice fields were winter flooded and the rate of deep plowing doubled to 50%. Ducks were at 80% of NAWMP goals for this and all remaining scenarios. Scenario 8 (30K–2× Plowed–80% Pop) assumed that 30,353 ha of all harvested rice fields were winter flooded, and the level of deep plowing increased to 50%. Scenario 9 (20K+20K–80% Pop) assumed that 20,235 ha of all harvested rice fields were winter flooded; however, an additional 20,235 ha of harvested rice fields were assumed flooded but would receive no maintenance water. As a result, these winter-flooded rice fields are assumed to be dry by early December and no longer available to foraging dabbling ducks. The current assumptions for deep plowing of rice were maintained in Scenario 9. Scenario 10 (30K+20K–80% Pop) assumed that 30,353 ha of all harvested rice fields were winter flooded and an additional 20,235 ha of harvested rice fields were also flooded but received no maintenance water. The current assumptions for deep plowing of rice were also maintained in Scenario 10.

To depict results of these scenarios, we plotted the relationship between TED and TES for each foraging guild for all 15-d intervals between August 15 and March 28. For ducks we also estimated the food deficit or food surplus associated with each scenario expressed as duck energy days (DED), where 1 DED represents the food energy equivalent of the DEE of a duck (Williams et al. 2014). Food deficits were calculated by first identifying the 15-d time periods where TED was less than TES, and the kilocalorie deficits associated with each of these time periods was summed. This kilocalorie total was subsequently divided by the DEE of a single bird in the duck guild, also expressed in kilocalories. This DEE was an average of DEE values for ducks in time periods where TED was less than TES (Table S1). Food energy surpluses were estimated by dividing the amount of unconsumed food energy remaining at the end of the final time period (kcal) by the DEE of a single duck in that period.

In general, duck food supplies were exhausted by late winter and goose food resources were exhausted by late winter or early spring in all drought scenarios that we evaluated. Results from scenario 1 (Current Hab–Goal Pop) indicated that duck food supplies in the Central Valley under nondrought conditions were adequate in all time periods except late March, even when duck populations were at NAWMP goals (Figure 1). Food supplies for geese in scenario 1 were adequate in all time periods, though differences between food energy demand and food energy supply were greatly reduced by spring (Figure 2). When duck numbers were reduced to 80% of NAWMP goals under nondrought conditions (scenario 2), food supplies for this foraging guild were adequate for all time periods (Figure 3).

Results for scenario 3 (20K–Goal Pop), the first drought scenario that we examined, indicated that duck food supplies would be exhausted by mid-January (Figure 1), whereas goose food supplies would be exhausted by mid-February (Figure 2). Duck food supplies were forecasted to be completely depleted by late January under these drought conditions even if assumed duck numbers were reduced to 80% of NAWMP goals (scenario 4, 20K–80% Pop, Figure 3). Increasing the amount of winter-flooded rice by 10,000 ha did little to increase dabbling duck food energy supplies, regardless if we assumed that duck populations were at NAWMP goals (scenario 5, 30K–Goal Pop, Figure 1) or 80% of NAWMP goals (scenario 6, 30K–80% Pop, Figure 3); in both scenarios, food supplies were exhausted well before spring. Increases in the amount of deep-plowed rice in scenario 7 (20K–2× Plowed–80% Pop) appeared to have little effect on duck food supplies relative to other drought scenarios (Figure 3), though goose food supplies were further reduced (Figure 2). Simultaneous increases in both the amount of winter-flooded rice and deep plowing (scenario 8; 30K–2×X Plowed-80% Pop) again had little effect on duck food supplies compared with other drought scenarios (Figure 3). Adding 20,235 ha of winter-flooded rice that received no maintenance water did little to increase late winter–early spring food resources for ducks in either scenario 9 (20K+20K–80% Pop; Figure 3) or scenario 10 (30K+20K–80% Pop; Figure 3).

Food surpluses for ducks only occurred in scenario 2 (80% NAWMP goals–nondrought conditions), and were equivalent to 47.8 million DED (Table 7). For all other scenarios, food energy deficits ranged from 24.4 million to 307.6 million DED. As expected, the greatest food deficits occurred in drought scenarios where duck populations were at NAWMP goals (Table 7).

Our results indicate a substantial impact of the California drought on carrying capacity for wintering waterfowl in the Central Valley. In general, duck food supplies were exhausted by late winter, and goose food resources were exhausted by late winter or early spring in all drought scenarios. In nondrought years, food resources in the Central Valley are projected to be at surplus for current populations of waterfowl and adequate for most of winter (i.e, except in late March) even if populations reach NAWMP goals. These results are consistent with other Central Valley studies that report increased body mass (Thomas 2009; Fleskes et al. 2016) and winter survival (Fleskes et al. 2007) of several dabbling duck species since the late 1980s, which were attributed to gains in managed wetlands and winter-flooded rice. In contrast, under all drought scenarios we evaluated, food resources fall short by mid- to late winter at meeting dabbling duck population needs even when duck numbers were only at 80% of NAWMP goal. Our drought results were strongly related to the large projected declines of winter-flooded rice that were common among all drought scenarios. Winter-flooded rice currently provides about 45% of all the food energy available to dabbling ducks in the Central Valley in nondrought years (Petrie et al. 2014).

Although wetland water supplies are likely to be restored in nondrought years, the effects of the drought on winter-flooded rice could be long lasting. Rice straw that remains after harvest is high in silicate and other components that make it difficult to decompose, but eliminating this straw before the next growing season is important for seedling establishment and avoiding disease (Bird et al. 2000). Before the drought, winter flooding of harvested rice fields provided an economic means for eliminating straw and enhancing the foraging quality of these fields, especially for dabbling ducks (Eadie et al. 2008). However, the long-term feasibility of winter flooding is likely dependent on reliable and affordable water supplies. Rice producers that do not winter flood typically incorporate rice straw into the soil by plowing or disking (i.e., dry incorporation; Miller et al. 2010). The degree to which dry incorporation reduces the abundance of waste rice is likely dependent on the field implements used and the number of field passes made. However, its reasonable to assume that winter flooding, not dry incorporation, optimizes foraging conditions for ducks. The extended California drought has greatly restricted availability of surface-water supplies for winter flooding, and long-term uncertainty around these water supplies could result in many rice growers permanently adopting dry incorporation to decompose straw.

Even with large declines in winter-flooded rice, dabbling duck food surpluses from October through early December persisted across all drought scenarios, though these surpluses were greatly reduced compared with nondrought years. It appears that effects of drought in fall and early winter are partially buffered by traditional management practices that essentially “front-load” wetland food supplies in the Central Valley by flooding most of these habitats before opening of the hunting season in late October. Although an estimated 25% of all managed wetlands were scheduled to receive no water in 2014, most that did receive water were expected to be flooded during the traditional time frame on the basis of our interviews of wetland managers. Although food production in some of these managed wetlands was reduced because they were not irrigated in summer, ≥ 60,000 ha of wetlands were expected to be inundated in late summer and early fall when duck numbers were still low compared with December and January (Table 2).

Although geese exhausted their food supplies in all drought scenarios, we caution interpretation of these results. Geese in the Central Valley rely heavily on green forage in February and March (Skalos 2012). However, green forage was not included in TRUEMET because we lacked information on the area, biomass, and nutritional quality of this food source. As a result, the food deficit we predicted for geese under drought conditions may be overcome by geese switching to green forage in late winter and early spring, as has been observed elsewhere (Alisauskas et al. 1988).

Our decision to include geese in this study was largely driven by the need to account for the effects of goose consumption on dabbling duck food resources, especially rice. Rice is the most important food in the diet of white-fronted geese in the Central Valley from October through January (Skalos 2012), and probably white geese as well. Regardless of drought scenario, dabbling duck food shortages occurred before the time geese begin consuming green forage. Thus, our decision not to include green forage in the model probably had little effect on our attempt to model the competitive effects of geese on duck food sources. Finally, the number of geese wintering in the Central Valley has almost doubled during the past 15 y, due mostly to large increases in greater white-fronted and white geese (Olson 2015). Further increases in the number of geese wintering in the Central Valley could worsen the impact of drought on dabbling ducks as geese consume a growing proportion of duck food supplies.

Although this study may help to understand the effects of drought on waterfowl carrying capacity, it also suggests that we begin to evaluate long-standing wetland practices that may not be suited to any future plagued by chronic water shortages. Current flooding schedules for managed seasonal wetlands in the Central Valley contribute to a temporal “mismatch” between duck population energy demand and energy supply that produces large food surpluses well into November. This fall surplus persisted even for drought scenarios where duck populations were at NAWMP goals. Wetland flooding schedules that are better timed with waterfowl migration in the Central Valley may help alleviate mid- to late-winter food shortages for ducks that were common to all drought scenarios, as well as reducing the amount of water needed to manage seasonal wetlands from late August through late March. However, additional research is needed to better understand the benefits and challenges that delayed flooding would pose. This includes investigating the effects of delayed flooding on other wetland-dependent species and waterfowl hunting opportunities, as well as developing a better understanding of how delayed flooding may be constrained by California's existing water delivery system.

Recent concerns over surface water supplies for waterfowl are not limited to the Central Valley. Water conflicts have emerged in many areas that are continentally important to waterfowl including the Klamath Basin of Oregon and California, Great Salt Lake, Utah, and Texas Gulf Coast (Engilis and Reid 1996; Downard 2010; Nielson-Gammon 2011; Fleskes 2012). Securing adequate water supplies for waterfowl and other wetland-dependent birds is likely among the greatest challenges facing resource managers, especially in the increasingly arid west.

Our results indicate that drought-related shortfalls in duck food supplies in the Central Valley will most likely occur during mid- to late winter, and that much of this shortfall is related to large declines in winter-flooded rice. In contrast, ducks continue to have access to a surplus of food supplies in early and late fall even under drought conditions. Managers could re-examine traditional flooding practices that now provide most duck foods by early fall, well before most birds have migrated into the Central Valley. By identifying alternate flooding schedules, managers may help alleviate mid- to late-winter food shortages that result from drought.

Please note: The Journal of Fish and Wildlife Management is not responsible for the content or functionality of any supplemental material. Queries should be directed to the corresponding author for the article.

Table S1. Microsoft Excel file containing data on the daily enegy expenditure of ducks and geese used in bioenergetics model TRUEMET to evaluate the impact of California's continuing drought on wintering waterfowl food resources in the Central Valley under a range of landscape and waterfowl population scenarios. Data shown are daily expenditure of dabbling ducks in kcal/d (column B), and daily energy expenditure of geese in kcal/d (column C) by time period (column A).

Found at DOI: 10.3996/082015-JFWM-082.s1; (11 KB XLSX).

Table S2. Microsoft Excel file containing data on habitat availability used in bioenergetics model TRUEMET to evaluate the impact of California's continuing drought on wintering waterfowl food resources in the Central Valley under a range of landscape and waterfowl population scenarios. Data shown are the percentage of managed seasonal wetlands assumed to flooded (column B), the percentage of rice area assumed to be harvested (column C), the availability of winter-flooded rice as a percentage of maximum flooded hectares where maintenance water is applied (column D), and the availability of winter-flooded rice as a percentage of maximum flooded hectares (column E) where maintenance water is not applied by time period (column A). Note: all other data used in the scenario modeling are presented in Tables 2, 4, 5, and 6.

Found at DOI: 10.3996/082015-JFWM-082.s2; (12 KB XLSX).

Reference S1. Central Valley Joint Venture. 2006. Central Valley Joint Venture Implementation Plan Conserving bird habitat. Sacramento, California: U. S. Fish and Wildlife Service.

Found at DOI: 10.3996/082015-JFWM-082.s3; (16440 KB PDF).

Reference S2. Fleskes JP, Yee JL, Casazza ML, Miller MR, Takekawa JY, Orthmeyer DL. 2005b. Waterfowl distribution movements and habitat use relative to recent changes in the Central Valley of California. Final report U.S. Geological Survey Western Ecological Research Center, Dixon Field Station, Dixon, California.

Found at DOI: 10.3996/082015-JFWM-082.s4; (10129 KB PDF).

Reference S3. Koneff M. 2003. Derivation of regional waterfowl planning objectives from NAWMP Continental Population Objectives. U.S. Fish and Wildlife Service Division of Migratory Bird Management. Laurel, Maryland.

Found at DOI: 10.3996/082015-JFWM-082.s5; (88 KB PDF).

Reference S4. Olson SM. 2015 Pacific Flyway Data Book. U.S. Fish and Wildlife Service, Vancouver, Washington.

Found at DOI: 10.3996/082015-JFWM-082.s6; (3711 KB PDF).

Reference S5. Pacific Flyway Council. 2006. Pacific Flyway management plan for the Aleutian goose. Aleutian Goose Subcommittee, Pacific Flyway Study Committee (c/o U.S. Fish and Wildlife Service), Portland, Oregon.

Found at DOI: 10.3996/082015-JFWM-082.s7; (981 KB PDF).

Reference S6. Petrie M, Brasher MG, Soulliere GJ, Tirpak JM, Pool DB, Reker RR. 2011. Guidelines for establishing Joint Venture waterfowl population abundance objectives. North American Waterfowl Management Plan Science Support Team Technical Report, 3–10.

Found at DOI: 10.3996/082015-JFWM-082.s8; (262 KB PDF).

Reference S7. Petrie M, Brasher M, James D. 2014. Estimating the biological and economic contributions that rice habitats make in support of North American Waterfowl Populations. Stuttgart, Arkansas: The Rice Foundation.

Found at DOI: 10.3996/082015-JFWM-082.s9; (3639 KB PDF).

Reference S8. U.S. Fish and Wildlife Service. 2014. Waterfowl population status, 2014. Washington, D.C.: U.S Department of the Interior.

Found at DOI: 10.3996/082015-JFWM-082.s10; (3769 KB PDF).

We dedicate this paper to our friend, colleague, and coauthor Mike Wolder, who passed away unexpectedly in December 2015. Mike's practical knowledge of the issues facing waterfowl and wetland managers in the Central Valley was fundamentally important in the development of this study. Mike's passion for and dedication to the natural resources of the Pacific Flyway were extraordinary, and he is sorely missed by those who were fortunate to know or work with him. We also thank the numerous public and private wetland managers, water district managers, Resource Conservation District personnel, nongovernmental organizations, and local landowners throughout the Central Valley who provided information that helped in the development of these scenarios. Julie Yee, Michael Brasher, and John Coluccy provided helpful comments on earlier drafts of the manuscript, whereas Elliott Matchett and Samantha Yeo prepared the figures. We also thank D. Haukos and two anonymous reviewers for significantly improving the manuscript. Funding for this analysis and summary was provided by the U.S. Geological Survey.

Any use of trade, product, website, or firm names is for descriptive purposes only and does not imply endorsement by the U.S. Government.