Physiography and Vegetation

The physiographic regions of Arizona are illustrated in figure 2. Most of northern Arizona lies within the Colorado Plateau, characterized by high elevation and mostly flat-lying or slightly dipping layered sedimentary rocks. In contrast, southern and far western Arizona fall within the generally lower elevation Basin and Range, characterized by a heavily faulted, folded, and eroded landscape. Between these two physiographic provinces is a topographically rugged area known as the Transition Zone (fig. 2; Spencer and Reynolds, 1989). Also referred to as the Central Arizona Highlands, the Transition Zone makes up about 15 percent of Arizona’s total land area but contributes about 50 percent of the streamflow originating within the State (Montgomery and Harshbarger, 1989). The Fort Apache Reservation straddles the boundary between the Transition Zone and the Colorado Plateau Physiographic Province (fig. 2). The western part of the reservation, within the Transition Zone, is characterized by heavily eroded, sloping terrain with deep canyons and stream valleys. The eastern part of the reservation, within the Colorado Plateau, is also rugged—despite the Colorado Plateau’s typical flat-lying sedimentary rocks—because this area is overlain by the eroded remnants of a volcanic field.

A large escarpment called the Mogollon Rim and the Apache-Sitgreaves National Forests form much of the northern boundary of the Fort Apache Reservation (fig. 1). Where the Mogollon Rim is present, it forms the boundary between the Colorado Plateau Province and the Transition Zone. The Fort Apache Reservation borders the Apache-Sitgreaves National Forests to the northeast and east within part of the White Mountains (fig. 1). The White Mountains are the erosional remnants of a volcanic field that was actively erupting lava, volcanic cinders, and volcanic ash between 9 and 2 million years ago (Bezy and Trevena, 2003). The largest feature in the volcanic field is Mount Baldy, the erosional remnant of a large stratovolcano (Bezy and Trevena, 2003). The summit of Mount Baldy (Baldy Peak) at 11,422 feet (ft; USGS, 2025), is the highest point in the White Mountains and on the Fort Apache Reservation (fig. 1). The Salt and Black Rivers form the southern boundary of the Fort Apache Reservation and separate it from the adjoining San Carlos Reservation (fig. 1). The Tonto National Forest forms the western boundary of the reservation (fig. 1).

The topography of the Fort Apache Reservation is diverse, featuring elevations that range from 11,422 ft at the summit of Mount Baldy (Baldy Peak) near the reservation’s eastern boundary, down to about 2,600 ft along the Salt River where it leaves the reservation to the west (USGS, 2025; fig. 1). The headwaters of several substantial streams are entirely or partially within the Fort Apache Reservation. These streams include the Black, White, and Salt Rivers, and Carrizo, Cibecue, and Canyon Creeks (fig. 1). Numerous smaller tributaries to these streams are also present on the Fort Apache Reservation (fig. 1).

Moore (1968) identified four physiographic subprovinces within the Fort Apache Reservation (fig. 3). These four subprovinces are characterized below based on the descriptions from Moore (1968) and by using the Landscape Fire and Resource Management Planning Tools (LANDFIRE) to identify vegetation types for the subprovinces (LANDFIRE, 2024).

Moore (1968) called the deep canyons of the Salt River, Canyon Creek, and lower parts of their tributary canyons, at the west and southwest edges of the reservation, the Canyon Creek–Salt River Subprovince (fig. 3). Within the canyons, the main vegetation types are cacti and scrub—narrow bands of riparian areas parallel streams where water is sufficient. The main vegetation types in the uplands above the canyons are woodlands containing pinyon, juniper, and oak trees (LANDFIRE, 2024).

East of the Canyon Creek–Salt River Subprovince, Moore (1968) identified an area extending east to Highway 73 and the White River, which he called the Carrizo Slope Subprovince (fig. 3). He described this area as a south-southwest-facing slope that has been incised into a badlands topography. The dominant vegetation types in this area are woodlands containing pinyon, juniper, ponderosa, and oak trees. Grasslands are also present to a lesser extent, and riparian areas parallel streams where water is sufficient (LANDFIRE, 2024).

Moore (1968) referred to the area containing large volumes of volcanic rock in the form of lava flows and pyroclastic material that were extruded and piled up to form mountains east of Highway 73 and bounded by a line from Fort Apache to Poker Mountain, as the White Mountain Subprovince. The dominant vegetation types in this subprovince are subalpine spruce-fir forest and woodland, upper montane conifer-oak forest and woodland, Ponderosa pine woodland, aspen forest and woodland, and subalpine-montane mesic meadow (LANDFIRE, 2024).

The fourth subprovince identified by Moore (1968) is a roughly triangular-shaped area south of the White River called the Bonito Prairie Subprovince. The Black River forms its southern boundary. The Bonito Prairie Subprovince is a gently rolling plain floored by basalt flows (fig. 3). The dominant vegetation types in this area are pinyon-juniper woodland, semidesert grassland, juniper savanna, and warm desert lower montane riparian woodland (LANDFIRE, 2024).

Geology

This diversity of the landscape of the Fort Apache Reservation is due to the complex geologic history of the region. Although some rocks on the reservation date back to the Proterozoic, more recent geologic events probably have a greater impact on the present landscape. Two relatively recent geologic events that helped shape the current landscape of the Fort Apache Reservation are the nearby crustal extension that began in the Miocene, which formed the Basin and Range Physiographic Province south of the reservation, and the prolific volcanism in the region that started during the Miocene and ended in the Pleistocene (Bezy and Trevena, 2003; Merrill and Péwé, 1977).

The oldest rock formations that are present at or near the surface on the reservation are Proterozoic in age (Richard and others, 2000). The rocks, which have been heavily metamorphosed, faulted, and folded, are present in the Canyon Creek–Salt River Canyon area of the reservation. Moore (1968) postulated that even though these rocks are exposed only in this area of the reservation, similar rocks are presumed to underlie the entire reservation. Proterozoic rock types present include intrusive and extrusive igneous rocks, sedimentary rocks, and metamorphic rocks.

The surficial geology of the Carrizo Slope area of the reservation consists primarily of dissected sedimentary rocks ranging in age from Paleozoic to Cenozoic (Richard and others, 2000). The slope forms a connection between the higher Colorado Plateau Province to the north and the Salt River at the southern boundary of the Fort Apache Reservation. The slope likely began to form during the Miocene in response to faulting and eroding associated with the Basin and Range Province’s crustal extension. The top of the slope is marked by the Mogollon Rim, which is the southern edge of the Colorado Plateau Physiographic Province.

For this report, the physiographic subprovince called the “White Mountain Area” in Moore (1968) is used to delineate the boundaries of the White Mountains. However, the boundaries of the White Mountains tend to be variable depending on the source referenced. The geologic report describing the White Mountains by Merrill and Péwé (1977) limits their extent to the area covered by Mount Baldy. Likewise, the U.S. Board on Geographic Names (2023) limits its extent to the Mount Baldy area, as does the map of physiographic areas in Arizona by Trapp and Reynolds (1995). However, the extent of the White Mountains on some U.S. Forest Service maps has a considerably larger area, including Mount Baldy and large adjacent areas.

Mount Baldy is a large stratovolcano that erupted predominantly between 9 and 2 million years ago, although earlier volcanic activity in the area dates back to the Eocene and Oligocene (Merrill and Péwé, 1977). Despite being heavily eroded, Mount Baldy is still the highest feature in the region; Baldy Peak reaches an elevation of 11,422 ft. Several other features near Baldy Peak also have elevations above 11,000 ft, including Mount Ord, Mount Warren, Mount Thomas, and Paradise Butte (fig. 1). Although these features have their own names, they are still part of a single stratovolcano which some geologists have unofficially called the “Mount Baldy Volcano” (Merrill and Péwé, 1977; Bezy and Trevena, 20034). Eruptions of this volcano were often explosive, blasting large amounts of volcanic ash into the atmosphere and periodically leading to ash-rich mud flows (lahars) that carried loose material from the slopes to the base of the volcano. No glaciers are in the White Mountains today, but glaciers were present on Mount Baldy during Pleistocene ice ages (Merrill and Péwé, 1977).

The Bonito Prairie Subprovince is relatively flat, topographically, compared to the rest of the reservation. The flat terrain is mainly formed by lava flows that have experienced less erosion than other reservation areas. Based on adjacent mapping by Wrucke and others (2004), the rock units at or near the surface in this area appear to be primarily Miocene lava flows. Moore (1968) reported that the volcanic rocks in this area overlie Paleozoic and Mesozoic sedimentary rocks.

Climate

The Fort Apache Reservation spans several climate zones because of its wide range of surface elevations. This report uses the Köppen climate classification system (Kottek and others, 2006) to describe the climate variability within the reservation. Based on empirical relationships between climate and vegetation, the Köppen system provides a way to describe climatic conditions defined by multiple variables and seasonality (Chen and Chen, 2013).

The Köppen climate classifications within the reservation boundary are illustrated in figure 4 using a 2.5-mile (4 kilometer) grid. These classifications are based on data computed by Godfrey (1999) for the conterminous United States, using gridded estimates of average precipitation, temperature, and elevation from the Parameter-elevation Regressions on Independent Slopes Model (PRISM; PRISM Climate Group, 1999). The averages for precipitation and temperature were estimated for the 30 years from 1961 to 1990. Five climate classifications—midlatitude steppe (BSk), interior Mediterranean (Csa), coastal Mediterranean (Csb), humid continental (Dsb), and continental subarctic (Dsc)—are shown in figure 4 and described in more detail below. Their distribution across the Fort Apache Reservation roughly parallels the elevation contours of the landscape.

A small part of the southwestern corner of the Fort Apache Reservation with the lowest surface elevation is classified as midlatitude steppe (BSk), which represents a cold semiarid climate (fig. 4). The use of “cold” here means the annual average temperature is below 64 degrees Fahrenheit (°F). Semiarid implies the zone receives less precipitation than the potential for evapotranspiration but more precipitation than a desert climate (Critchfield, 1983).

The remaining part of the reservation’s southwestern area is classified as interior Mediterranean (Csa), which represents a hot-summer, Mediterranean-like climate (fig. 4). These zones typically experience hot, dry summers, in which the warmest month averages temperatures above 72 °F. Winters are milder and wetter; the wettest winter month receives about three times more precipitation than the driest summer month (Critchfield, 1983).

To the north and east of the Csa zone is an area that roughly forms a band classified as Csb for coastal Mediterranean (fig. 4). This Csb classification represents a warm-summer Mediterranean-like climate similar to the Csa classification except that monthly average summer temperatures remain below 72 °F (Critchfield, 1983). The Csb zone has higher elevations than the Csa zone along the slope below the Mogollon Rim and on the west side of the White Mountains (fig. 4), so lower temperatures are expected. The Csb classification also implies that at least three times as much precipitation falls in the wettest month of winter as in the driest month of summer (Critchfield, 1983).

Paralleling the Csb zone (fig. 4) to the north and east is the humid continental zone (Dsb). The Dsb classification represents a climate with dry cool summers, in which the average temperature of the warmest month is greater than 50 °F. In comparison, the average temperature of the coldest month is less than 32 °F. As with the Csa and Csb classifications, at least three times as much precipitation falls in the wettest month of winter as in the driest month of summer (Critchfield, 1983). Winters are severe with snowstorms, strong winds, and bitter cold. The Dsb zone is higher in elevation on the slope below the Mogollon Rim and on the west side of the White Mountains than the Csb zone (fig. 4).

The continental subarctic zone (Dsc) is the highest in elevation of all the Köppen climate regions on the Fort Apache Reservation, covering the highest parts of Mount Baldy in the northeast part of the reservation (fig. 4). A subarctic climate implies long, cold winters and warm to cool summers. The Dsc classification also implies that at least three times as much precipitation falls in the wettest month of winter as in the driest month of summer, and winters are severe with snowstorms, strong winds, and bitter cold (Critchfield, 1983).

Estimated temperature data available from the PRISM Climate Group (2023) indicated that the 30-year (1991–2020) mean annual temperatures on the Fort Apache Reservation varied, in degrees Fahrenheit, from the high 30s to mid-60s, depending on elevation (fig. 5A). The 30-year mean annual precipitation on the Fort Apache Reservation, according to the PRISM Climate Group (2023), varied from around 15 to 45 inches (in.), again largely depending on elevation (fig. 5B). Mean annual temperature and precipitation for the Fort Apache Reservation were estimated using spatial regression methods that incorporated data from traditional weather stations and high-altitude meteorological sites (Daly and others, 1994).

Hydrology

As described in the “Physiography and Vegetation” and “Climate” sections, the topography of the reservation has considerable relief and is dominated by dendritic (branching) patterns of stream erosion (fig. 1). Much of that erosion is powered by the high elevations of the White Mountains and Mogollon Rim. About 8,800 ft of relief separates the highest reaches of the White Mountains from the point where the Salt River leaves the reservation (fig. 1). The straight-line distance between these features is only about 70 miles (mi), resulting in steep stream gradients. The elevation change between the point where the Salt River exits the reservation and the crest of the Mogollon Rim exceeds 4,000 ft across much of the reservation. Streams originating at higher elevations have dissected the terrain as they flow toward lower elevations.

The major streams on the reservation are the Salt River and its large tributaries, the White and Black Rivers. The Black River and its continuation as the Salt River form the boundary between the Fort Apache and San Carlos Reservations (fig. 1). The North Fork White River currently supplies municipal water to communities around Whiteriver through a centralized water system that also uses groundwater. A new rural water system is planned that will receive water from an 8,600 acre-feet (acre-ft) reservoir to be constructed on the North Fork White River about 10 mi north of Whiteriver (Bureau of Reclamation, 2022). The reservation is also crossed by many smaller tributary streams, including Big Bonito, Canyon, Carrizo, Cibecue, Diamond, and Tonto Creeks (fig. 1).

The clear, cool headwaters of the White and Black Rivers on the reservation provide critical habitat for Oncorhynchus apache (Miller, 1972; Apache trout), a species sacred to the White Mountain Apache Tribe of the Fort Apache Reservation, Arizona, and is the State fish of Arizona (U.S. Department of the Interior, 2024). The Apache trout is one of only two trout species native to the State (the other is Oncorhynchus gilae [Miller, 1950; Gila trout]). Originally listed as endangered under the Federal Endangered Species Preservation Act of 1966 (Public Law 89–669), the Apache trout later became federally protected with the passage of the Endangered Species Act in 1973 (16 U.S.C. 1531–1544). In 1975, the Apache trout was downlisted to threatened, and in 2024, delisted from the Federal list of endangered and threatened species—becoming the first American sportfish to achieve delisting (U.S. Fish and Wildlife Service, 2024a; 2024b). This remarkable recovery is due in large part to the management and conservation efforts of the White Mountain Apache Tribe of the Fort Apache Reservation, Arizona (U.S. Fish and Wildlife Service, 2024a; 2024b). Cool mountain streams are rare in Arizona, making the cold-water habitats on the Fort Apache Reservation critical for the continued long-term survival of the species in its native range.

The Williams Creek Unit of the Alchesay-Williams Creek National Fish Hatchery Complex at the head of Williams Creek on the Fort Apache Reservation (fig. 1) is the only facility in the Nation that has a captive breeding population of Apache trout. Other fish species produced at the Williams Creek Unit include brook, brown, and rainbow trout (U.S. Fish and Wildlife Service, 2024c). The hatchery is fed by large springs issuing from basalt flows that at times have a combined flow of about 1,800 gallons per minute. In recent years, flow from the springs has been lower than in the past. In 2022, flow rates reached their lowest levels since recordkeeping began in the 1960s. Additionally, in recent years, the Williams Creek Unit has found it challenging to meet the requirements of its National Pollutant Discharge Elimination System permit for the concentration of total phosphorus (Bradly Clarkson, U.S. Fish and Wildlife Service, oral commun., 2023).

The Alchesay National Fish Hatchery, about 7 mi north of Whiteriver, Arizona, along the North Fork White River (fig. 1), is the other unit of the Alchesay-Williams Creek National Fish Hatchery Complex. Water for this hatchery comes from the North Fork White River. Large springs issuing from Paleozoic sedimentary rocks on the east side of the river, less than a mile upstream from the hatchery, contribute to the streamflow. The Alchesay facility serves as a “grow-out rearing unit” for brown and rainbow trout fingerlings that are transferred from the Williams Creek facility. Fish from the Alchesay National Fish Hatchery are used to stock Tribal reservoirs, lakes, and rivers annually across the southwest.

Many natural springs emerge on the Fort Apache Reservation, in addition to the large springs previously mentioned in association with the Alchesay-Williams Creek National Fish Hatchery Complex. The density of springs is especially high on and below the Mogollon Rim and in the White Mountains area (fig. 3). These springs provide critical habitat for vegetation and wildlife.

The Fort Apache Reservation has numerous lakes, providing habitat for aquatic and other species as well as recreation for Tribal members and guests. Some of the larger lakes are Hawley Lake, Sunrise Lake, Reservation Lake, Pacheta Lake, and Christmas Tree Lake (fig. 1). Fishing permits and camping and lodging fees are important sources of revenue for the White Mountain Apache Tribe of the Fort Apache Reservation, Arizona (Javis Davis and Emery Hoffman, White Mountain Apache Tribe of the Fort Apache Reservation, Arizona, Water Resources Program, oral commun., 2023).

Groundwater is not widely used across the Fort Apache Reservation, even though geologic maps (Wilson and others, 1959; Finnell, 1966a, b; McKay, 1972; Skotnicki, 2002) indicate that Paleozoic formations, which are typically good aquifers where present in Arizona, are found under much of the reservation. Additionally, Merrill and Péwé (1977) suggested the pre-Oligocene stratigraphy of the White Mountains likely contains Paleozoic and Mesozoic formations in at least some areas under the younger volcanic deposits.

The largest use of groundwater on the reservation is for the centralized water system that supplies water to Whiteriver and surrounding communities (Javis Davis, White Mountain Apache Tribe of the Fort Apache Reservation, Arizona, Water Resources Program, oral commun., 2024). Water for this system comes from surface water from the North Fork White River and by groundwater from a wellfield on Miner Flat approximately 11 mi north of Whiteriver. The wells in the Miner Flat wellfield draw water from the regional Coconino aquifer. Although the Coconino aquifer is a large regional aquifer, Kaczmarek (2002) reported drawdown and depletion in the Miner Flat area was caused by over pumping required to meet the water demands of the local population. Although wellfield production rates have continued to decline, groundwater from Miner Flat continues to be an important source of water for the Tribe’s centralized water system and is planned to be part of the new rural water system currently under development (Bureau of Reclamation, 2022).

Additional groundwater uses include supplying water for the communities of Carrizo and Cibecue through wells installed in stream alluvium and partially in rocks of the lower part of the Supai Group (Kaczmarek, 2015). The Sunrise Park Resort in the White Mountains uses groundwater from wells completed in volcanic rock. Several wells equipped with solar pumps are used for stock watering; however, many of these wells are now dry or go dry for at least part of the year (Javis Davis and Emery Hoffman, White Mountain Apache Tribe of the Fort Apache Reservation, Arizona, Water Resources Program, oral commun., 2024).

Periods Used for Trend Analyses

Multiple time periods were used for trend analyses. In most cases, a period of record data analysis was done for each dataset using all available data beginning with the first complete year for which data are available (assuming no subsequent gaps in the data) and ending in 2023. If there were gaps in the data, the analyses started with the first complete year of data after the final year containing a gap. For datasets going back to at least 1980, the data from 1980 to 2023 were chosen for additional trend analyses.

The main reason for selecting this period is that the temperature dataset used in the study has two distinct trends for mean annual air temperature; a breakpoint between the two trends happened in the late 1970s. The timing of this breakpoint coincides with a known shift in the rate of change of air temperatures for the continental United States (Meehl and others, 2016; Vose and others, 2017; U.S. Environmental Protection Agency, 2025). Streamflow trends showed inflection points in the late 1980s when data were analyzed for the entire period of record. For these reasons, the trend analyses of a variety of climate metrics from 1980 to 2023 were used to test whether the rate of change on the Fort Apache Reservation was also stronger than that for the preceding decades. Trends identified from 1980 to 2023 may be part of natural climatic oscillations and not part of a long-term change in climate. However, because the purpose of this study is to specifically look at the change in climate during the lifetime of the current population living on the Fort Apache Reservation, the 1980 to 2023 period is important. If datasets extended back to the 1930s or earlier, the period from the beginning of their relevant records to 1979 was analyzed for trends. Again, trends seen during these periods may be part of natural climatic oscillations and not part of a long-term change in climate, but nonetheless affected the current population, either directly or because of ecological responses.

Another complication with the time periods used in the trend analyses is that three different types of years (calendar year, water year, and climate year) were used depending on the type of data analyzed. Calendar year is self-explanatory. A water year is the 12-month period from October 1 through September 30 of the following calendar year. The water year is designated by the calendar year in which it ends; for example, water year 1981 was the period from October 1, 1980, through September 30, 1981. A climate year is the 12-month period from April 1 through March 31 of the following calendar year. The climate year is designated by the calendar year in which it ends; for example, climate year 1981 was the period from April 1, 1980, through March 31, 1981.

Near-Surface Air Temperature Record

Near-surface air temperature data analyzed for this study came from the USHCN monthly data version 2.5 dataset (Menne and others, 2009; Menne and Williams, 2012; Mason, 202637). Only data from station Whiteriver 1 SW, Arizona (NCEI USHCN station USH00029271) were used for trend analysis (table 1) because this is the only USHCN station on the Fort Apache Reservation (fig. 1) and because long-term air temperature records can be subject to several forms of systematic bias (Menne and Williams, 2009; Menne and others, 2009, 2010). The raw values for this dataset came from the NOAA COOP station with the same name (Whiteriver 1 SW). The NCEI USHCN Whiteriver 1 SW station (USH00029271) is a virtual version of the NOAA Whiteriver 1 SW COOP station that uses the COOP station’s location and available raw data to construct a bias-corrected dataset and to provide estimated data for periods when raw data are missing. The NOAA Whiteriver 1 SW COOP station was located in the community of Whiteriver in the east–central part of the Fort Apache Reservation at an elevation of approximately 5,120 ft (fig. 1) and had temperature data from 1900 to 2008 (Western Regional Climate Center, 2023a). The station was discontinued in 2008. Although actual data collection at Whiteriver 1 SW occurred from 1900 to 2008, the USHCN version 2.5 dataset contains temperature data from 1893 to the present (2023). Data outside the 1900 to 2008 period were estimated, as described below. Estimated data from 2009 to 2023 were used in this study, but estimated data from 1893 to 1900 were excluded because of some missing monthly data. Data available from the USHCN version 2.5 dataset include monthly maximum, minimum, and average temperature, and monthly total precipitation.

When deemed necessary, the NCEI adjusted USHCN version 2.5 data from the raw value to account for things like changes in the timing or location of observations; location, replacement, or recalibration of instruments; and alterations to the land use or land cover surrounding a measurement site (Menne and Williams, 2009). According to the NCEI (Menne and Williams, 2012), the version 2.5 data are either (1) raw (nonadjusted, though flagged for possible data-quality issues), (2) adjusted because of time of observation bias, or (3) put through the pairwise homogenization algorithm.

Menne and others (2009) reported that in many cases, the timing of daily observations has changed through time. Prior to 1940, observers recorded temperature data near sunset in accordance with U.S. Weather Bureau instructions, resulting in a slight positive (warm) bias during the first half of the century. Since then, a switch to morning observation times has become more common to satisfy operational hydrological requirements and has resulted in a broad-scale reduction in mean temperatures (Menne and others, 2009). Adjustments for this time of observation bias have been applied to USHCN version 2.5 data when detected.

A pairwise homogenization algorithm is used to make these adjustments to the USHCN version 2.5 data. The rationale and methodology used to apply adjustments to station data are described in Menne and Williams (2009). In short, an automated homogenization algorithm using a pairwise comparison between the target station dataset and a reference series is used to apply corrections to the target dataset. The reference series is commonly constructed by averaging values from other measurement stations near the target station that have fully adjusted temperature values (Menne and Williams, 2009).

Estimates for missing monthly temperature values were added to the USHCN version 2.5 data by NCEI when data were missing from the raw dataset. The estimates are generated using an optimal interpolation technique, which makes use of the fully adjusted temperature values at neighboring measurement stations (Menne and Williams, 2009). Data from USHCN version 2.5 are being adjusted as new data are collected. This creates a time series going back in time that reflects the current status of a station’s configuration (for example, current instrumentation, observation time, and site characteristics), but without the localized inhomogeneities detected by the pairwise homogenization algorithm. This can cause the data to change slightly depending on when it is obtained, but ideally, the overall climate trend of the adjusted record should be reasonably stable through time. The USHCN version 2.5 data used in this study was downloaded on August 21, 2024 (Mason, 2026).

The adjusted monthly temperature record from Whiteriver 1 SW was used to test for trends in the average monthly mean, maximum, and minimum, near-surface air temperature at the station using a Mann-Kendall test. The Theil-Sen slope estimate was used to compute the average change in temperature between the record’s beginning and ending periods. In addition, the average monthly mean, maximum, and minimum near-surface air temperatures for each year were used to analyze trends in the average annual mean, maximum, and minimum near-surface air temperatures.

Another possible source of temperature data for this study was NOAA’s Monthly U.S. Climate Gridded Dataset (NClimGrid; Vose and others, 2014). NClimGrid contains gridded data for the continental United States containing precipitation and maximum, minimum, and average temperature records. The data are based on stations in the Global Historical Climatology Network daily (GHCNd). The Whiteriver 1 SW station used in this study is part of the GHCNd. The Whiteriver 1 SW data were used exclusively instead of NClimGrid data because the comparisons of interest were changes through time, rather than spatial comparisons. Also, no other stations in the area offer a similar length of source records within the GHCNd. Because of the lack of other stations, useful or statistically valid differentiation of temporal trends across gridded locations within the scale of the study area is not feasible.

Precipitation Records

Nine sets of precipitation records were analyzed for this study. These records came from three NRCS SNOTEL stations, one NRCS snow course, one NCEI USHCN station, and one NOAA COOP network station (table 1).

Streamflow Records

Daily mean streamflow values from six USGS streamgages that capture flow originating from the Fort Apache Reservation were used in this study. In addition, the PHDI also was used to assess drought through time.

Snow-Derived Streamflow Statistics

Statistics on the annual spring streamflow center of mass were computed for USGS streamgage East Fork White River Near Fort Apache, Arizona (station 09492400; hereafter, East Fork White River Near Fort Apache). This streamgage was chosen to examine the magnitude and timing of streamflow associated with the annual spring snowmelt period because it lies within the White Mountain Physiographic Subprovince identified by Moore (1968). Snowmelt peak statistics were not computed for data from the other five streamgages because although they all receive some flow from snowmelt, they receive enough rain runoff during the winter and spring months to confound the timing and magnitude of annual snowmelt runoff.

The total volume of streamflow during the months of March through June for each year and the day of the year when the center of mass of that volume occurred were analyzed. The March through June timeframe used to capture the spring snowmelt runoff was chosen by visually inspecting annual streamflow hydrographs from this streamgage and assessing which months typically had snowmelt runoff. The analysis included a Mann-Kendall test and Theil-Sen slope estimate applied to the data using the same methods described in the “Precipitation Records” section above.

Using the streamflow record from this streamgage to look for trends in snowmelt runoff is confounded by the fact that flow at this streamgage is periodically influenced by rain on snowpack events that can occur in the winter and spring in this region. In some years, the highest daily mean streamflow recorded during the winter to spring months can be rain on snowpack events. An example is the January 8, 1993, peak that occurred likely as a result of rain on the snowpack. House and Hirschboeck (1997) provide a description of the rainstorms of January and February 1993 in Arizona and their interaction with snowpack. A hydrograph of daily mean streamflow at USGS streamgage East Fork of the White River Near Fort Apache, from December 1, 1992, through June 30, 1993, is shown in figure 6. The steepness and magnitudes of the rising and falling limbs of the January and February peaks (fig. 6) are a good indicator that these peaks are affected by rain and not just snowmelt. The peaks in April and May are more typical of peaks generated by the more gradual process of melting snow due to solar radiation and above-freezing air temperatures. This becomes even more obvious when you look at streamflow collected from the same time span but collected at 15-minute intervals (fig. 6B). Fifteen-minute data are only available from this streamgage starting in 1987; therefore, daily mean values are used in the analysis because they extend back to August 1957. Not only does the steepness of the rain-runoff-generated peaks become more pronounced, but the daily diurnal peaks, caused by the changes in solar radiation and air temperature between day and night, also become visible in the snowmelt peaks from mid-March through May. Another thing illustrated in figure 6B is the relatively small area under the hydrograph from rain or rain on snowpack peaks compared to the overall area under the hydrograph from mid-March through early June. Because the area under the hydrograph curve is directly proportional to the volume of streamflow, the volume of streamflow generated by snowmelt is considerably larger than the volume generated by the rain-associated peaks. The use of the total streamflow volume for the snowmelt months and that volume’s center of mass as indicators of change through time lessens the effect that rain on snowpack peaks has on the analysis of change. Although the January and February peaks of 1993 are not a factor in the analyses because the period used was March through June of each water year, a handful of possible rain on snowpack peaks that occurred through the period of record during the months included in the analyses were identified.

The total streamflow volume was computed by converting daily mean streamflow in cubic feet per second into daily acre-feet and then summing all the daily acre-feet values for March through June of each water year. The center of mass for this volume was computed using the following equation:

Day of year for streamflow center of mass

$$=\frac{{\displaystyle \sum _{n=61}^{182}(dv\times n)}}{{\displaystyle \sum _{n=61}^{182}(dv)}}$$

,

(1)

where

dv

is the daily streamflow volume, in acre-feet, and

n

is the numeric day of the year.

A Mann-Kendall test and Theil-Sen slope estimate were applied to the data using the same methods described in the “Precipitation Records” section.

Annual Near-Surface Air Temperature

The annual mean, maximum, and minimum near-surface air temperatures from 1901 to 2023 from the USHCN version 2.5 dataset for Whiteriver 1 SW (Menne and Williams, 2012; Mason, 2026)37) were analyzed for statistical trends using the Mann-Kendall test and Theil-Sen slope estimate. The annual mean, maximum, and minimum near-surface air temperatures were calculated by summing the average monthly values from the USHCN version 2.5 dataset and dividing by 12. The monthly averages from the USHCN version 2.5 dataset themselves were calculated from daily averages. For all three temperature parameters, the annual data are not distributed evenly when plotted (fig. 7), and an apparent inflection in the plotted data points occurs around 1980 (fig. 7A, D, G). For this reason, in addition to examining data from 1901 to 2023, additional analyses were done for data from 1901 to 1979 (fig. 7B, E, H) and 1980 to 2023 (fig. 7C, F, I).

The annual mean near-surface air temperature from 1901 to 2023, using adjusted data from the U.S. Historical Climatology Network version 2.5 dataset for Whiteriver 1 SW, is presented in Figure 7A. The Theil-Sen slope estimate line is shown along with the p-value from a Mann-Kendall test. The slope of the Theil-Sen line is positive, and the p-value from the Mann-Kendall test indicates this trend is significant (p<0.05). The slope of the line indicates that mean annual temperatures have increased by 1.35 °F during this period (fig. 7A; table 2). Average mean annual air temperature from 1901 to 1979 decreased by 1.57 °F (fig. 7B; table 2), whereas average mean annual air temperature from 1980 to 2023 increased by 2.48 °F (fig. 7C; table 2).

The average annual maximum near-surface air temperature from 1901 to 2023 is presented in Figure 7D. The Theil-Sen slope estimate line is shown along with the p-value from a Mann-Kendall test. The slope of the Theil-Sen line is positive, and the p-value from the Mann-Kendall test indicates this trend is significant (p<0.05). The slope of the line indicates that average annual maximum near-surface air temperatures have increased by 2.87 °F during this period (fig. 7D; table 3). Average annual maximum near-surface air temperatures from 1901 to 1979 are listed in table 3 and are plotted along with a fitted Theil-Sen slope line in fig. 7E. The line has a slightly negative slope, but the p-value from the Mann-Kendall test indicates this trend is not significant (p>0.05). Average annual maximum near-surface air temperatures from 1980 to 2023 increased significantly (p<0.05) and by 3.48 °F during this time period (fig. 7F; table 3).

The average annual minimum near-surface air temperature from 1901 to 2023 is presented in Figure 7G. The Theil-Sen slope estimate line is shown along with the p-value from a Mann-Kendall test. The slope of the Theil-Sen line is negative, but the p-value from the Mann-Kendall test indicates the trend is not significant (p>0.05; table 4). Average annual minimum near-surface air temperature from 1901 to 1979 decreased by 3.34 °F (fig. 7H; table 4), whereas the average annual minimum near-surface air temperature from 1980 to 2023 increased by 1.45 °F (fig. 7I; table 4).

Monthly Near-Surface Air Temperatures

The average monthly near-surface air temperatures from 1980 to 2023, using adjusted data from the USHCN version 2.5 dataset for Whiteriver 1 SW, are presented in Figure 8A. This period was chosen for analysis based on the observed inflection point detected in the annual air temperature data described in the “Annual Near-Surface Air Temperature” section. The Theil-Sen slope estimate line is shown, along with the p-value from a Mann-Kendall test. Twelve plots, one for each month of the year, are shown in figure 8A. The Theil-Sen slope lines are positive for all of the months, and significant trends (p-values<0.05) were determined for the months of March and June–November (fig. 8A). Among the months with significant trends, June had the largest monthly average temperature increase (3.91 ˚F) from 1980 to 2023 (table 5).

Similar plots for average monthly maximum near-surface air temperatures are shown in figure 8B. The Theil-Sen line slopes are positive for all months, and significant trends (p-values<0.05) were determined for half of the months (March, April, June, August, September, and November). March had the largest average monthly maximum temperature increase (5.39 °F) from 1980 to 2023, as predicted by the Theil-Sen lines for months with significant trends (p<0.05; table 6).

Plots with Theil-Sen lines for the change in average monthly minimum near-surface air temperatures from 1980 to 2023 are shown in Figure 8C. All months have Theil-Sen lines with positive slopes except for February and December, which have Theil-Sen lines with negative slopes. However, only the months of June through September had significant trends (p<0.05; table 7). All four of these months had increases in average monthly minimum near-surface temperatures of at least 3 °F, and June had the largest increase (3.25° F; table 7).

Annual Precipitation

The average total annual precipitation from 1901 to 2023, using adjusted data from Whiteriver 1 SW, is presented in Figure 9A. The Theil-Sen slope estimate is negative, and the Mann-Kendall test indicated the trend is significant (p<0.05). Using the Theil-Sen slope estimate, the average total annual precipitation decreased from 19.10 in. in 1901 to 15.30 in. in 2023, a 19.9 percent decrease (fig. 9A; table 8). As with the annual near-surface air temperature record, the annual precipitation record from Whiteriver 1 SW was broken into two periods (1901–79 and 1980–2023) to further analyze for trends. The Theil-Sen slope estimate was positive from 1901 to 1979, but the Mann-Kendall test indicated there was no significant trend (p=0.454; fig. 9B; table 8). The data from 1980 to 2023 had a negative Theil-Sen slope estimate, and the Mann-Kendall test indicated the trend was significant (p<0.001; table 8). Using the Theil-Sen slope estimate, average total annual precipitation decreased from 20.7 in. in 1980 to 10.98 in. in 2023, a 47.0 percent decrease (fig. 9C; table 8).

Total annual precipitation data from Baldy 310, Maverick Fork 617, and Wildcat 866 were found to have significant negative trends (p<0.05; fig. 9D, E, F; table 9). Baldy 310 and Maverick Fork 617 had precipitation data available from 1981 to 2023, whereas Wildcat 866 had data available from 1985 to 2023. The decrease in average total annual precipitation calculated using the Theil-Sen slope estimates ranged from 5.70 in. at Baldy 310 to 10.07 in. at Wildcat 866 (table 9). These decreases were 18.7 and 34.7 percent, respectively, for these two stations.

The total annual precipitation from 1934 to 2023, as recorded at McNary 2N, is displayed in Figure 9G. The Theil-Sen slope estimate is negative, but the trend was not significant (p>0.05; fig. 9G; table 10). As with the total annual precipitation record from Whiteriver 1 SW, the record from McNary 2N was broken into two periods (1934–79 and 1980–2023) to further analyze for trends. The Theil-Sen slope estimate was positive from 1934 to 1979, but the trend was not significant because the p-value for the Mann-Kendall test was 0.079, which is slightly greater than the p<0.05 threshold (fig. 9H; table 10). The period from 1980 to 2023 had a negative Theil-Sen slope estimate, and the p-value from the Mann-Kendall test (p=0.010) indicated the trend was significant (p<0.05; fig. 9I; table 10). Using the Theil-Sen slope estimate, average total annual precipitation decreased from 29.11 in. in 1980 to 20.53 in. in 2023, a 29.5 percent decrease (fig. 9I; table 10).

Monthly Precipitation

Monthly precipitation data from 1980 to 2023 for Whiteriver 1 SW (NCEI, 2024) were analyzed. This period was chosen for analysis because of the apparent inflection point in the annual total precipitation data around 1980, as described in the “Annual Precipitation” section. March, April, and August were the only months with significant trends (p<0.05) in average monthly total precipitation (fig. 10A; table 11). All three trends were negative, but August had the largest average decrease (1.86 in.) from 1980 to 2023 (table 11).

Baldy 310 and Maverick Fork 617 each had data available from 1981 to 2023, whereas Wildcat 866 had data from 1985 to 2023 (NRCS, 2024a, b, c5253). For Baldy 310, the only month with a significant trend (p<0.05) for average monthly total precipitation was April (fig. 10B; table 11). The trend was negative and indicated an average decrease in April precipitation of 0.88 in. (table 11). Monthly precipitation recorded at Maverick Fork 617 had significant trends (p<0.05) for the months of March and April (fig. 10C; table 11). The trends were negative for both months and indicated average decreases in monthly precipitation of 1.61 and 1.02 in., respectively. Monthly precipitation recorded at Wildcat 866 only had a significant trend (p<0.05) for April (fig. 10D; table 11). The trend was negative and indicated an average decrease in April precipitation of 1.01 in. (table 11).

McNary 2N (Western Regional Climate Center, 2023b) had precipitation data available from 1934 to 2023. However, only data from 1980 to 2023 were chosen for analysis because of an apparent inflection point in the annual total precipitation data occurring around 1980, as described in the “Annual Precipitation” section. The only months with significant trends (p<0.05) for this period were March, April, and November (fig. 10E; table 11). All three trends were negative, but March had the largest decrease in average monthly total precipitation (2.08 in.) from 1980 to 2023.

Snowfall and Snowpack

Snowfall and snowpack records were analyzed using a Mann-Kendall test and the Theil-Sen slope estimate to detect significant positive or negative trends at Baldy 310 and Maverick Fork 617, Fort Apache 09R05 (NRCS, 2024a, b, c5253), and McNary 2N (Western Regional Climate Center, 2023b).

Annual maximum SWE values and the day of year when the annual maximum SWE occurred at Baldy 310 and Maverick Fork 617 were analyzed for trends. In addition, an analysis was performed on the day of year when SWE at both stations went to zero.

The analyses of peak SWE from Baldy 310 and Maverick Fork 617 resulted in Theil-Sen slope estimates with negative trends, but only the trend from Baldy 310 was significant (p<0.05; fig. 11A, B). The trend from Baldy 310 indicated a decrease in the average peak SWE of 4.5 in., or a 45.0 percent decrease from 1981 to 2023 (table 12). The p-value from Maverick Fork 617’s trend analysis (p=0.055) was close to the p<0.05 threshold (table 12). The decrease in annual peak SWE calculated from Maverick Fork 617 was 4.9 in. (table 12), or a 45.0 percent decrease from 1981 to 2023.

The trend analyses performed on the day of year when peak SWE was reached at Baldy 310 and Maverick Fork 617 also produced negative Theil-Sen slope estimates for both stations, but only the trend for Baldy 310 was significant (p<0.05; fig. 12A, B; table 12). The trend from Baldy 310 indicated the average date of peak SWE from 1981 to 2023 was around March 18 early in the record, but by 2023, the average date of peak SWE was around February 13. This shift to an earlier date for peak SWE is about 33 days for this period (table 12).

Trend analyses of the day of year when SWE reached zero at Baldy 310 and Maverick Fork 617 produced negative Theil-Sen slope estimates, and both trends were significant (p<0.05; fig. 13A, B). At Baldy 310, the trend indicates that the average date the SWE reached zero early in the record was around April 25, but by 2023, this date had shifted to around March 25. This shift to an earlier date for zero SWE is about 31 days for this period (table 12). Similarly, at Maverick Fork 617, the average date when the SWE went to zero early in the record was around April 28, shifting to around April 2 by 2023, 26 days earlier table 12).

Trend analyses of early April SWE from Fort Apache 09R05 resulted in negative Theil-Sen slope estimates for the two periods tested (1951–2023 and 1981–2023), but the trends were not significant (p>0.05; fig. 14; table 13). A trend analysis was also done on the total annual snowfall recorded at McNary 2N, indicating a significant (p<0.05) decrease in annual snowfall between 1935 and 2023 (fig. 15). Average annual snowfall during this time period decreased from 106.3 in. to 61.2 in., representing a 42.4 percent decline (table 14).

East Fork White River Near Fort Apache Streamgage

The East Fork White River Near Fort Apache streamgage has been in continuous operation since August 1957. The streamflow measured at this streamgage should largely represent natural conditions with minimal human-made interference. The streamgage is about 15 mi southeast of Baldy Peak (fig. 1) at an elevation of 6,050 ft in the White Mountain physiographic subprovince identified by Moore (1968). The drainage area upstream from the streamgage is 38.8 square miles (mi²; U.S. Geological Survey, 2024). The drainage area upstream from the streamgage is primarily undeveloped forested land (LANDFIRE, 2024; fig. 3). No known regulation or diversions of the East Fork of the White River occur upstream from the streamgage.

Statistical analyses were done using daily mean streamflow data from 1958 to 2023. Plots of the 1-, 7-, and 30-day minimum streamflow, the 1-, 7-, and 30-day maximum streamflow, the annual median and annual mean streamflow, along with the p-value and a LOWESS-based smoothed trend line (using a 30-year smooth window) are presented in figure 16A. No trends were significant (all p-values>0.05), but the p-value for the annual 30-day maximum streamflow trend (p=0.050) was equal to the threshold of p<0.05 (table 15).

For all annual streamflow plots analyzed, an inflection point appears in the smoothed trend line in the late 1980s (fig. 16). Prior to the late 1980s, the trend lines are mostly flat; however, after this period, the lines have a clear negative slope. For this reason, additional trend analyses were done for the record from 1980 to 2023. For this period, all trends were negative and significant (p<0.05) except for the 1-day maximum streamflow; although the 1-day maximum streamflow did have a p-value near the p<0.05 threshold (p=0.072; fig. 16B; table 15). The annual mean streamflow in cubic feet per second can be converted into annual runoff in acre-feet. Using the equation for the LOWESS smooth line with a 20-year smooth window, the annual runoff decreased from about 33,590 acre-ft in 1980 to about 16,360 acre-ft in 2023—a reduction of 17,230 acre-ft, or 51.3 percent (table 16).

Snowmelt Peaks

Because the headwaters of the East Fork White River originate in the White Mountains (fig. 1), streamflow data from East Fork White River Near Fort Apache were analyzed to identify potential trends related to mountain snowmelt runoff. The center of mass for the total streamflow from March through June of each calendar year was used to represent the annual snowmelt peak. The “Data Sources and Methods of Analysis” section provides a detailed description of how the center of mass was calculated. In addition, the total springtime flow volume was analyzed for possible trends.

Using data from 1958 to 2023, the Mann-Kendall test applied to the day of year when the spring snowmelt peak occurred at this streamgage identified a significant (p<0.05) negative Theil-Sen slope estimate (fig. 17A; table 17). The Theil-Sen line indicates that the average date of the spring snowmelt peak was around May 4 early in the record, but by 2023, it had shifted to around April 22 (fig. 17A). This represents a shift of 12 days to an earlier date for the spring snowmelt peak (table 17).

A similar analysis was done to look for a trend in the day of year for the annual spring snowmelt peak from 1980 to 2023 because other significant trends were detected in streamflow data from this period. Again, the Mann-Kendall test produced a negative Theil-Sen slope estimate that was significant (p<0.05; table 17). The Theil-Sen line indicates the average date of the spring snowmelt peak was around May 2 early in the record, but by 2023 the average date of the peak has been occurring around April 20 (fig. 17B). This is a shift to an earlier date for spring snowmelt peak of 12 days (table 17).

The total annual springtime (March–June) streamflow volume from 1958 to 2023 was tested for a trend using the Mann-Kendall test. Although this test produced a negative Theil-Sen slope estimate, the trend was not significant (p>0.05; fig. 17C; table 17).

An analysis of data from 1980 to 2023 revealed a significant negative trend (p<0.05) in the total annual springtime (March–June) volume of streamflow (fig. 17D; table 17). Despite the variability in the data for this period, the trend line indicates that the average volume of springtime streamflow early in the period was about 17,900 acre-ft, but by 2023, the average volume of springtime streamflow had decreased to about 6,700 acre-ft (table 17). This is a 62.6 percent decrease in the average volume of springtime streamflow. During this period, the smallest volume of recorded springtime streamflow was about 1,700 acre-ft in 2002, whereas the largest volume of recorded springtime streamflow was about 38,200 acre-ft in 1993 (fig.17D).

Table 17.    

Change in the timing and magnitude of the spring snowmelt peak, using Mann-Kendall test and Theil-Sen slope estimate, for U.S. Geological Survey streamgage East Fork White River Near Fort Apache, Arizona (station 09492400), 1958–2023 and 1980–2023.

[All data are from the U.S. Geological Survey National Water Information System database (U.S. Geological Survey, 2024). p-value, probability value; acre-ft, acre feet; beg., beginning]

Table 17.    Change in the timing and magnitude of the spring snowmelt peak, using Mann-Kendall test and Theil-Sen slope estimate, for U.S. Geological Survey streamgage East Fork White River Near Fort Apache, Arizona (station 09492400), 1958–2023 and 1980–2023.

Period of record used	p-value for the center of mass streamflow between March 1 and June 30 each year*	Average date for spring snowmelt peak		Change in date of annual snowmelt peak (days)†	p-value of total volume of streamflow between March 1 and June 30 each year*	Average total volume of spring streamflow		Change in total volume of spring streamflow (acre-ft)
		Beg.	End			Beg. (acre-ft)	End (acre-ft)
1958–2023	<0.001	May 4	April 22	−12	0.141	16,700	11,100	5,600
1980–2023	<0.001	May 2	April 20	−12	0.015	17,900	6,700	11,200†

^*

p-value from Mann-Kendall trend analysis

^†

Significant trend at p<0.05.

Palmer Hydrological Drought Index

The PHDI computed for Arizona Climate Division 4 (East Central) by the National Centers for Environmental Information (NCEI, 2024) was compared to streamflow data from East Fork White River Near Fort Apache. The PHDI is a water-balance model that analyzes precipitation, temperature, and surface and groundwater supplies to detect long-term drought. The PHDI for Arizona Climate Division 4 (East Central) and the annual mean streamflow measured at East Fork White River Near Fort Apache from 1958 to 2023 are plotted in fig. 18. Both parts of the figure contain a LOESS smooth line that was generated using the ggplot2 R package (Wickham, 2016; Wickham and others, 2024), in contrast to the other streamflow figures described in this section that contain a modified LOWESS smooth line that was generated using the EGRET R package (Hirsch and De Cicco, 2015; Hirsch and others, 2023). The LOESS smooth line for the Arizona Climate Division 4 (East Central) PHDI data (fig. 18A) is shaped similarly to the LOESS smooth line for the streamflow data (fig. 18B). Both curves show a positive slope from the beginning of the period in 1958 to sometime in the early 1980s, followed by a negative slope that eventually starts to gradually increase again around 2010 and continues to increase through the end of the period in 2023. One notable difference between the two smooth lines is that the line for PHDI ends considerably lower in 2023 than the level at which it started in 1958 (fig. 18A). In contrast, the line for annual mean streamflow begins and ends at roughly the same level (fig. 18B). Nonetheless, the similarities in the two smooth lines likely indicates the streamflow recorded at East Fork White River Near Fort Apache was, at least in part, responding to climate trends that were occurring in Arizona Climate Division 4 (East Central).

White River Near Fort Apache, Arizona, Streamgage

The White River Near Fort Apache, Arizona, streamgage has been in continuous operation since October 1957. Several small, man-made recreational lakes receive streamflow upstream from the streamgage, and some water is diverted upstream from the streamgage for irrigation (U.S. Geological Survey, 2024). In addition, surface water from the North Fork of the White River is used as a municipal water supply for communities around Whiteriver, Arizona. The design capacity for the intake used in this system is about 4 million gallons per day, or 6 cubic feet per second (ft³/s; Bureau of Reclamation, 2022). Because of these anthropogenic water uses, streamflow measured at this streamgage does not represent natural conditions. However, the record from this streamgage is still informative for assessing whether the perceived decrease in water availability on the Fort Apache Reservation could be quantified using existing hydrologic data. The streamgage is about 4.5 mi upstream from the White River’s confluence with the Black River and about 1 mi west of Fort Apache (fig. 1) at an elevation of 4,368 ft in the Carrizo Slope Physiographic Subprovince identified by Moore (1968). The drainage area upstream from the streamgage is 632 mi² (U.S. Geological Survey, 2024). Most of this area is covered with ponderosa pine and pinyon-juniper woodlands; however, there are areas with other vegetative cover and areas of urban development as well (LANDFIRE, 2024; fig. 3).

Statistical analyses were done using daily mean streamflow data from 1958 to 2023. Plots of the 1-, 7-, and 30-day minimum streamflow, the 1-, 7-, and 30-day maximum streamflow, and the annual median and annual mean streamflow, along with the p-value and a LOWESS-based smoothed trend line (using a 30-year smooth window) are presented in figure 19A. Only the 30-day maximum streamflow showed a significant trend (p<0.05). However, the 30-day minimum, 7-day maximum, annual median, and annual mean streamflows had p-values near the p<0.05 threshold (table 15).

For all annual streamflow plots analyzed, an inflection point in the smoothed trend line occurs in the late 1980s. Prior to the late 1980s, trend lines are mostly flat or slightly positive, but after this period, they show clear negative slopes. For this reason, additional trend analyses were done for the data from 1980 to 2023. For this period, all trends were negative and significant at the p<0.05 level except for the 1-day maximum streamflow (fig. 19B; table 15). Using the equation for the LOWESS smooth line with a 20-year smooth window, the annual runoff decreased from about 200,760 acre-ft in 1980 to about 69,280 acre-ft in 2023—a reduction of 131,480 acre-ft, or 65.5 percent (table 16).

Black River Near Fort Apache, Arizona, Streamgage

The Black River Near Fort Apache, Arizona, streamgage (USGS station 09490500; fig. 1) has been in continuous operation since October 1957. There is negligible storage in several small recreational lakes upstream from the streamgage and one transbasin diversion (Willow Creek diversion) that removes water from the river for industrial and municipal use at the Black River pump station, about 0.4 mi downstream from the Black River’s confluence with Freezeout Creek (fig. 1). Flow in the diversion was measured from 1945 to 2002 by the Black River streamgage. During this time period, maximum daily mean diversion flows ranged from 0 to 28 ft³/s (U.S. Geological Survey, 2024). Because of this diversion, streamflow measured at the Black River streamgage cannot be considered to represent the natural condition. However, the record from this streamgage is still informative for assessing whether the perceived decrease in water availability on the Fort Apache Reservation could be quantified using existing hydrologic data.

The streamgage is about 5 mi upstream from the Black River’s confluence with the White River and about 14 mi west of Fort Apache (fig. 1) at an approximate elevation of 4,345 ft in the Carrizo Slope Physiographic Subprovince identified by Moore (1968). The drainage area of the stream upstream from the streamgage is 1,232 mi² (U.S. Geological Survey, 2024). Much of the drainage area upstream from the streamgage is covered with ponderosa pine and pinyon-juniper woodlands, and some grasslands; however, smaller areas have other vegetative cover (LANDFIRE, 2024).

Statistical analyses were done using daily mean streamflow data from 1958 to 2023. Plots of the 1-, 7-, and 30-day minimum streamflow, the 1-, 7-, and 30-day maximum streamflow, and the annual median and annual mean streamflow, along with the p-value and a LOWESS-based smoothed trend line (using a 30-year smooth window), are presented in figure 20A. No trends were significant at the p<0.05 level (table 15).

For all annual streamflow plots analyzed, an inflection point in the smoothed trend line occurs in the late 1980s. Prior to the late 1980s, the trend line is either mostly flat or has a positive slope, but after this period, the line has a clear negative slope. For this reason, additional trend analyses were done from 1980 to 2023. For this period, all slope estimates were negative, and all trends were significant (p<0.05) except for the 1-, 7-, and 30-day maximum streamflow and the annual mean streamflow. However, the annual mean streamflow had a p-value near the p<0.05 threshold (p=0.070; fig. 20B; table 15). Using the equation for the LOWESS smooth line with a 20-year smooth window, the annual runoff decreased from about 351,850 acre-ft in 1980 to about 154,930 acre-ft in 2023—a reduction of 196,920 acre-ft, or 56.0 percent (table 16).

Cibecue Creek near Chrysotile, Arizona, Streamgage

The Cibecue Creek near Chrysotile streamgage (fig. 1) has been in continuous operation since May 1959. Small diversions are reported upstream from the streamgage near Cibecue, Arizona (diversions not shown in fig. 1). Streamflow measured at this streamgage might not be considered the natural condition because of these small diversions. However, the record from this streamgage is still informative for assessing whether the perceived decrease in water availability on the Fort Apache Reservation could be quantified using existing hydrologic data. The streamgage is about half a mile upstream from Cibecue Creek’s confluence with the Salt River and about 7 mi north of the former mining town of Chrysotile, Arizona (fig. 1) at an elevation of 3,185 ft in the Canyon Creek–Salt River Canyon Physiographic Subprovince identified by Moore (1968). The drainage area upstream from the streamgage is 295 mi² (U.S. Geological Survey, 2024). Upstream from the streamgage, the drainage area consists of woodlands containing pinyon pine, juniper, and oak trees (LANDFIRE, 2024).

Statistical analyses were done using daily mean streamflow data from 1961 to 2023. Plots of the 1-, 7-, and 30-day minimum streamflow, the 1-, 7-, and 30-day maximum streamflow, and the annual median and annual mean streamflow, along with the p-value and a LOWESS-based smoothed trend line (using a 30-year smooth window), are presented in Figure 21A. Only the 7- and 30-day maximum streamflow and annual mean streamflow had a significant trend (p<0.05; table 15).

For most annual streamflow plots analyzed, an inflection point in the smoothed trend line occurs around 1990. Prior to 1990, most trend lines are mostly flat or have a positive slope, but after this period, the line has a clear negative slope. For this reason, additional trend analyses were done for data from 1980 to 2023. For this period, all trends were negative and significant at the p<0.05 level except for the 1-, 7-, and 30-day maximum streamflow, although the 30-day maximum streamflow had a p-value (p=0.082) slightly above the p<0.05 threshold (fig. 21B; table 15). Using the equation for the LOWESS smooth line with a 20-year smooth window, the average annual runoff decreased from about 38,590 acre-ft in 1980 to about 14,480 acre-ft in 2023—a reduction of 24,110 acre-ft, or 62.5 percent (table 16).

Carrizo Creek near Show Low, Arizona, Streamgage

The Carrizo Creek Near Show Low, Arizona (USGS station 09496500; hereafter, Carrizo Creek streamgage) streamgage has been in continuous operation since April 1977. The streamgage is near the intersection of U.S. Highway 60 and State Highway 73 (fig. 1) at an elevation of 4,750 ft in the Carrizo Slope Physiographic Subprovince identified by Moore (1968). The drainage area upstream from the streamgage is 439 mi² (U.S. Geological Survey, 2024) and primarily consists of woodlands containing juniper, pinyon pine, ponderosa pine, and oak trees (LANDFIRE, 2024). A transbasin diversion from Show Low Creek affected the streamflow record from this streamgage prior to 2006. Additionally, diversions for irrigation of areas less than 300 acres are reported upstream from the streamgage (U.S. Geological Survey, 2024). Because of these diversions, streamflow measured at the Carrizo Creek streamgage cannot be considered to represent the natural condition. However, the record from this streamgage is still informative for assessing whether the perceived decrease in water availability on the Fort Apache Reservation could be quantified using existing hydrologic data.

Statistical analyses were done using daily mean streamflow data from 1978 to 2023. Plots of the 1-, 7-, and 30-day minimum streamflow, the 1-, 7-, and 30-day maximum streamflow, and the annual median and annual mean streamflow, along with the p-value and a LOWESS-based smoothed trend line (using a 20-year smooth window), are presented in figure 22. The trends were all negative and significant (p<0.05) apart from the 1-day maximum streamflow (p=0.214; table 15). For consistency, the equation for the LOWESS smooth line was used to calculate how much the annual runoff has changed from 1980 to 2023 (in contrast to using the full record from 1978 to 2023). The average annual runoff decreased from about 35,140 acre-ft in 1980 to about 6,380 acre-ft in 2023—a reduction of 28,760 acre-ft, or 81.8 percent (table 16).

Salt River Near Chrysotile, Arizona, Streamgage

The Salt River Near Chrysotile streamgage has been in continuous operation since September 1924. The streamgage is 5.7 mi northeast of Chrysotile and upstream from the U.S. Highway 60 bridge (fig. 1) at an elevation of 3,355 ft in the Canyon Creek–Salt River Canyon Physiographic Subprovince identified by Moore (1968). The drainage area upstream from the streamgage is 2,849 mi² (U.S. Geological Survey, 2024). The drainage area upstream from the streamgage is primarily forest and grassland (LANDFIRE, 2024).

The Salt River Near Chrysotile record has been affected by minor irrigation diversions from tributary streams, a transbasin inflow from Show Low Lake, and a transbasin outflow from the Black River to Willow Creek in the Gila River Basin. For these reasons, streamflow measured at Salt River Near Chrysotile cannot be considered to represent the natural condition. However, the record from this streamgage is still informative for assessing whether the perceived decrease in water availability on the Fort Apache Reservation could be quantified using existing hydrologic data.

Statistical analyses were done using daily mean streamflow data from 1925 to 2023. Plots of the 1-, 7-, and 30-day minimum streamflow, the 1-, 7-, and 30-day maximum streamflow, and the annual median and annual mean streamflow, along with the p-value and a LOWESS-based smoothed trend line (using a 30-year smooth window), are presented in Figure 23A. In all cases, the overall slopes of the smoothed trend lines were negative, and the 7- and 30-day minimum streamflow trends were significant (p<0.05). In addition, the p-value for the annual 1-day minimum streamflow (p=0.074) was close to the p<0.05 threshold (table 15).

Two obvious inflection points are in the smooth lines of all the annual streamflow plots analyzed from 1925 to 2023. The first inflection point, which occurs around 1960, is discussed in later report sections. The second inflection point occurs in the mid- to late-1980s, the same period observed in most of the other streamflow records described in this section. Between about 1960 and the 1980s, the trend lines either have positive slopes or are mostly flat, but after this period, the lines have clear negative slopes. For this reason, additional trend analyses were done for the period from 1980 to 2023. For this period, all trends were negative and were significant (p<0.05), except for the 1-, 7-, and 30-day maximum streamflow; although, the 30-day maximum streamflow had a p-value (p=0.061) near the p<0.05 threshold (fig. 23B; table 15). Using the equation for the LOWESS smooth line with a 20-year smooth window, the average annual runoff decreased from about 658,810 acre-ft in 1980 to about 273,660 acre-ft in 2023—a reduction of 385,150 acre-ft, or 58.5 percent (table 16).

Palmer Hydrological Drought Index

The PHDI computed for Arizona Climate Division 4 (East Central) by the National Centers for Environmental Information (2024) was compared to streamflow data from Salt River Near Chrysotile. Although the streamflow record from Salt River Near Chrysotile is affected by irrigation diversions and transbasin inflows and outflows, a comparison of the record with the PHDI is nonetheless informative. Plots and LOESS smooth lines of the PHDI for Arizona Climate Division 4 (East Central) and annual mean streamflow at Salt River Near Chrysotile from 1925 to 2023 are shown in Figure 24. The smooth lines in both plots have similar shapes (fig. 24). Both smooth lines indicate negative trends from the beginning of the period in 1925 to sometime around 1960, followed by a positive trend until the mid-1980s, and then another negative trend until the end of the period, in 2023. This comparison indicates that the streamflow at Salt River Near Chrysotile was likely responding, at least in part, to regional climate trends across Arizona Climate Division 4 (East Central).

Climate and Seasonal Patterns

Examples of perceived changes in climate and seasonal patterns recorded by Gauer (2019) from interviews of White Mountain Apache Tribe of the Fort Apache Reservation, Arizona, community members are shown in table 18. Warmer temperatures with less predictable weather and changes in seasonal patterns are a common theme among the community members interviewed. The statistical analyses of near-surface air temperature recorded at Whiteriver 1 SW described in the “Trend Analyses Results” section support these observations. These results show that this station's overall average annual air temperature increased by 1.35 °F from 1901 to 2023 (fig. 7; table 2). However, a more striking change was observed when that period was broken into two periods, from 1901 to 1979 and from 1980 to 2023. The average annual temperature decreased by 1.57 °F from 1901 to 1979, then increased by 2.48 °F from 1980 to 2023 (fig. 7; table 2), indicating that recent warming has been greater than earlier in the record. Another notable observation is that the average monthly mean and average monthly maximum temperatures from March for this station increased significantly during the period from 1980 to 2023. The average monthly mean temperature for March increased by 2.91 °F, whereas the average monthly maximum temperature for March increased by 5.39 °F (fig. 8A, B; tables 5, 6).

Warming temperatures alone can affect the character of winter precipitation. Knowles and others (2006) evaluated seasonal (November–March) snowfall liquid water equivalent from NOAA COOP stations in the 11 westernmost states from 1949 to 2004 and compared it to the seasonal total precipitation. The largest reductions in snowfall liquid water equivalent they documented appeared to be shifts from snowfall to rainfall, driven by warming at temperate, low to moderate elevations. This pattern is apparent in observed changes in snowfall amounts and the timing of snowmelt-derived runoff on the Fort Apache Reservation described in the following four sections.

Aridification

Examples of perceived aridification and its effects reported by Gauer (2019) are listed in table 18. Nine community members reported lower snowfall as an example of aridification. Specific examples described in interviews include a shorter ski season in recent years at the Sunrise Park Resort with an earlier end to the snowpack on Mount Baldy. Other examples included perceived decreases in snowfall or persistence of snow in community areas like McNary. Seven community members reported drier weather and five reported lower rainfall. One member reported that soaking summer rains are decreasing in frequency. Five community members reported lower river water levels. In addition, the Director of the White Mountain Apache Tribe of the Fort Apache Reservation, Arizona’s Water Resources Program reported that changes in the flow regimes of streams and increasing water temperatures are causing increased algal growth, which has led to challenges operating part of the Tribe’s drinking water system that relies on flows from the North Fork of the White River (Cheryl Pailzote, White Mountain Apache Tribe of the Fort Apache Reservation, Arizona, Water Resources Program, oral commun., 2023).

Increased Risks

The increased risk most often reported to Gauer (2019) was more frequent wildfires (table 18). This observation is supported by several other recent studies that show the annual area burned by wildfire in the western United States has increased since the mid-1980s (Dennison and others, 2014; Abatzoglou and Williams, 2016; Westerling, 2016; Holden and others, 2018). A couple of examples of recent findings are Mueller and others (2020) and Parks and Abatzoglou (2020). Mueller and others (2020) looked for climate relationships with increasing wildfire in Arizona and New Mexico from 1984 to 2015. They found that the total burned area increased at a rate of 9.5 percent annually, or about 11,000 acres per year, whereas the rate of increase for areas burned at high severity was 17.3 percent, or about 2,000 acres annually (Mueller and others, 2020). Similarly, Parks and Abatzoglou (2020) found that areas burned at high severity had an eightfold increase in western United States forests from 1985 to 2017.

A common theme in many studies examining the increase in wildland fires in the western United States is the rise in vapor pressure deficit (VPD) of the atmosphere, partly driven by anthropogenic temperature increases. Like relative humidity (RH), VPD measures the atmosphere’s ability to hold water vapor. Seager and others (2015) explain that VPD is a more useful absolute measure of the atmospheric moisture state, unlike RH, which is a ratio of two known quantities expressed as a percentage. RH is the ratio, expressed in percent, of the amount of atmospheric moisture present relative to the amount that would be present if the air were saturated. Because RH depends on temperature, it is not a consistent metric to describe the potential loss of water by plants to air and by direct evaporation of other terrestrial water reservoirs. For example, an RH of 20 percent at an air temperature of 60 °F provides much less evaporation potential than an RH of 20 percent at an air temperature of 100 °F.

Vapor pressure deficit is the calculated difference between the saturating pressure (water vapor at equilibrium with condensation) of water in air and the actual water vapor pressure of the air, measured in units of pressure. Although VPD is affected by air temperature, the calculation of VPD is done in such a way that the calculated VPD has the same evaporation potential regardless of air temperature. For example, if the calculated VPD is 0.7 kilopascal (kPa; a value suitable for healthy plant growth), the evaporation potential is the same whether the air temperature is 60 °F or 100 °F. This makes using VPD to assess the atmospheric moisture state more straightforward.

Plants use evapotranspiration to move water and nutrients from the roots through the shoots, so some degree of evaporation potential is desirable. Additionally, excessively wet conditions can lead to the growth of plant pathogens. In general, a VPD between 0.7 to 1.5 kPa is considered good for most plant growth, whereas a VPD greater than 1.5 kPa is considered dry air that can lead to rapid water loss from plants, leaving them more susceptible to wildfires. Seager and others (2015) reported that VPD was extremely high in advance of the June 2002 Rodeo–Chediski fire that burned 468,600 acres in east–central Arizona, including part of the Fort Apache Reservation. Ffolliott and others (2010, p. 27) reported that the fire “damaged or destroyed ecosystem resources, disrupted hydrologic functioning, and altered loadings of flammable fuels in much of the ponderosa pine (Pinus ponderosa) forests exposed to the burn.”

Traditional Resources

Tribal community members reported a decrease in the availability of traditional plants and medicines and a decline in the elk and deer populations (table 18; Gauer, 2019). No readily citable dataset records changes in traditional resources on the Fort Apache Reservation, but the observed climate and seasonal changes, aridification, and wildland fires previously discussed have the potential to affect the availability of traditional resources.