We obtained daily maximum (T_max) and minimum (T_min) temperatures from the high-resolution (4 km, 1979–2022) GridMET Climatology Lab dataset (Abatzoglou 2013).

Our study assesses the spatial climatology over the current climate normal period (1991–2020) and trends over the past four decades (1979–2022) across the contiguous U.S. We use U.S National Climate Assessment regions shown in figure 1 for discussion of observed trends (USGCRP 2018). We also examine the time series of the climate metrics across the leading apple-producing counties: Yakima (Washington), Wayne (New York), and Kent (Michigan) (figure 1). Yakima county, Washington has a semi-arid climate with cold and wet winters and hot and dry summers while Kent and Wayne County have humid continental climates characterized by warm and humid summers. These counties account for 53,703 acres (13.1%), 27,761 acres (6.8%), and 14,153 acres (3.4%) of the total national acreage of 411, 262 (USDA NASS CoA 2022).

We examined the following six climate metrics that capture climate conditions over the apple phenological cycle and are relevant for assessing risks to growth, quality, and harvest (figure 2):

• (a)  
  Chill portions (CPs): For assessing the transition between endo- and eco-dormancy, certain number of hours of chill accumulation between specific cold temperatures are vital (Bowling et al 2020). Insufficient chilling can lead to non-uniform bud break, increased bud abscission, and reduced flower and fruit quality (Atkinson et al 2013). Optimal temperatures that are beneficial for chilling are between −2 °C and 13 °C, with the most optimal temperatures occurring between 6 °C and 8 °C (Erez et al 1990). CPs are calculated over 1 September–31 March using the dynamic model (Erez et al 1990, Fishman et al 1987a, 1987b), which accounts for the chill 'effectiveness' of different temperatures. We analyze the CPs accumulated as of 31 March that represents the chilling accumulation from 1 September to 31 March. The chillR package (Luedeling et al 2023) was used to calculate CPs. CP requirements can vary between 35 and 79 CPs for different apple varieties (Darbyshire et al 2016, Díez-Palet et al 2019, Parkes et al 2019, Noorazar et al 2022).
• (b)  
  Cold degree days: The accumulation of cold conditions over a given period is useful for evaluating the threat of damage from extended periods of freezing temperatures (Rochette et al 2004). We define cold degree days as the integration of daily average temperatures below 0 °C (Bhatnagar et al 2018) in degree C on all days (d) between 1 November–31 March, as follows:

$$\begin{align}{\text{Cold}}\,{\text{degree}}\,{\text{days}} &= \sum\limits_d \left( {0^\circ C - \frac{{{T_{{\text{max}}}} + {T_{{\text{min}}}}}}{2}} \right);\nonumber\\ &{\text{if}}\frac{{{T_{{\text{max}}}} + {T_{{\text{min}}}}}}{2} < 0^\circ C \ldots .\end{align} \tag{ 1 }$$

• (c)  
  Frost risk: Changes in the last day of frost for the season could affect the overlay between bud break timing and frost exposure and lead to loss of yields due to frost damage (Pfleiderer et al 2019, Park et al 2021). We examine changes in frost risk based on the last day of frost, defined as the last day with T_min ⩽ 0 °C between 1 January and 31 July.
• (d)  
  Growing Degree Days (GDD): Accumulation of degree days between growth-conducive temperatures is critical for flower and fruit development and growth. GDD is a measure of this accumulation between a baseline temperature and upper temperature threshold outside of which growth is minimal (Zhou and Wang 2018). We calculate GDD in degree C over January–April as relevant for bud break & flowering and January–September for the overall growth of apples (Küçükyumuk and Erdal 2012). We use 6 °C for the baseline and 28 °C for the upper threshold following the guidelines provided by the Cornell Institute for Climate Smart Solutions (2023):

$$\begin{align}{\text{Growing Degree Days}} &= \left\{ \begin{array}{*{20}{c}} {\sum\limits\frac{{{T_{{\text{max}}}} + {T_{{\text{min}}}}}}{2} - 6^\circ C);if{ }\,{T_{{\text{min}}}} > 6^\circ C{ } \, and{ } \, {T_{{\text{max}}}} \lt 28^\circ C{ }} \\[3pt] {\sum\limits\frac{{28C + {T_{{\text{min}}}}}}{2} - 6^\circ C);if \,{ }{T_{{\text{min}}}} > 6^\circ C{ } \, and{ } \, {T_{{\text{max}}}} \ge 28^\circ C} \\[3pt] {\sum\limits\frac{{{T_{{\text{max}}}} + 6^\circ C}}{2} - 6^\circ C);if \,{ }{T_{{\text{min}}}} \le 6^\circ C{ } \, and{ } \, {T_{{\text{max}}}} \lt 28^\circ C} \end{array} \ldots \ldots \right.\end{align} \tag{ 2 }$$

• (e)  
  Extreme heat days: Apples can suffer damage from sunburn when exposed to air temperatures exceeding 34 °C during the warm growing season (Darbyshire et al 2015). Fruit surface temperatures are greater than air temperatures (Willsea et al 2023) and cause sunburn browning damage when the fruit's surface temperature hits 42 °C–46 °C and necrosis when surface temperatures exceed 52 °C (Kalcsits et al 2017). These thresholds are variety specific and may only damage a percentage of the crop (Racsko and Schrader 2012). Given the availability of long-term air temperature data, we examine sunburn risk using daily maximum air temperatures during June–August exceeding 34 °C, during which fruit surface temperatures can likely exceed 42 °C.
• (f)  
  Warm nights: Optimal color development, crucial for a marketable apple, occurs when anthocyanin synthesis begins on the peel and apples start developing red and pink hues (Arakawa et al 1999, Willsea et al 2023). Harvested apples have been known to show greater red coloration with exposure to cooler nighttime temperatures during the late growing season and harvest (Fang et al 2019). While the temperature thresholds are still an area of active research, we use 15 °C based on personal communications from tree fruit experts and previous studies (Wang et al 2000) and define warm nights as days with daily minimum temperatures between August–September exceeding 15 °C.

Zoom In Zoom Out Reset image size

We employ simple linear regression to quantify long-term (1979–2022) trends and p-values to determine statistical significance of trends from the historical GridMET dataset. We considered trends as significant if p-values < 0.05 (α = 0.05). This is done using the 'linregress' function from the SciPy library in Python (Virtanen et al 2020). All p-values are found in supplementary figure S1.

To evaluate the overall impact of changes in multiple metrics, we define a potential climate damage index (PCDI) that synthesizes the six key climate metrics outlined in section 2.2. PCDI is the number of climate metrics that show trends in a direction that has potential to adversely impact the yield, quality aspects such as sweetness, color, cosmetic damage, storability or other characteristics that affect marketability. This index highlights regions where apple production may face increased risks due to climatic trends. Here, we assume that positive trends in cold degree days, GDD during bud-break and bloom, GDD for overall growth, extreme heat days, and warm nights, and negative trends in CPs have the potential for negative impacts. An additional assumption is that local production is adapted to local climatology. We note the caveat here that the impact of these climate trends on apple production characteristics can vary across different regions of the U.S. For instance, increased GDD might be beneficial in colder regions while detrimental in relatively warmer southern regions.

Note, we did consider using a weighted index of these metrics. However, these metrics have relevance for different aspects of apple growth and quality and there is no research to justify weights to assign to different metrics.

Figure 3 shows the spatial patterns of the climatology and decadal trends in two metrics—cold degree days that capture the risk of lethal conditions and CPs that capture chilling accumulation that is beneficial for apples. Parts of the Northern Great Plains, Midwest, and Northeast typically experienced an average of 900–1500 cold degree-days per season (figure 3(A)). Several southern states experience fewer cold degree-days (∼0–150). Cold degree-days decreased across most of the northern U.S., with the most pronounced declines of up to 70-degree-days/decade across the Rocky Mountains located in the Northern Great Plains and parts of the Southwest (figure 3(i)).

Zoom In Zoom Out Reset image size

Different varieties of apples grown across the U.S. require different CPs for endo-dormancy. Average CPs on 31 March representing chilling accumulation between 1 September–31 March, were the highest for coastal areas of the Northwest (∼140 CP; figure 3(B)). Much of the Northwest, Southwest, Northern Great Plains, and Northeast have average CP of ∼84–112 while CP is lower in the Southern Plains, Southeast, and southern Southwest regions (figure 3(B)). Over the past four decades, CPs increased across the Northwest and Northern Great Plains, exceeding 5 CP/decade in some areas (figure 3(ii)). CPs also increased in the upper Midwest and Northeast by ∼1–2 CP/decade), while the southern states of the U.S. experienced little to no change of between 0–2 CP/decade (figure 3(ii)). While there was increased seasonal chill accumulation from November through March in the northern latitudes, the increase was driven primarily by the late winter season changes (February and March; figure S3).

Next, we evaluated the climatology and decadal trends in GDD for two periods—bud break and flowering (January–April) and overall growth (January–September)—that capture the accumulation of heat between certain temperatures (figure 4). GDD leading to apple bud break and flowering varied from ∼900–1500 for the Southern Plains, Southeast, and Southwest (figure 4(A)) to 0–300 degree-days in the upper Midwest, Northwest, and Northeast. Similarly, GDD for overall growth spanning January-September varied from 3600–6000 degrees-days over parts of the southern and midwestern U.S. to ∼600–2400 degree-days across the Northwest and ∼1200–2400 degree-days in the upper Midwest and Northeastern U.S. (figure 4(B)). Across several southern U.S. states, GDD over the bud break and flowering period increased by >25 degree-days/decade and experienced moderate declines of ∼0–10 degree-days/decade in the upper Midwest and Northwest (figure 4(i)). In contrast, GDD for overall growth increased across most of the US with the largest changes exceeding 70 degree-days/decade across all regions except the Northern Great Plains and parts of the Midwest (figure 4(ii)). Trends in January–September GDD are more moderate or negligible in parts of the Northern Great Plains (figure 4(ii)).

Zoom In Zoom Out Reset image size

High daytime temperatures in the summer can pose sunburn risk for apples during fruit development and warm nighttime temperatures in fall can affect apple color development (figure 2). Figure 5 shows the climatology and decadal trends in two climate metrics that capture these conditions- extreme heat days representing daily maximum temperatures >34 °C during June–August and warm nights representing daily minimum temperatures >15 °C in August and September. The number of extreme heat days are typically highest in parts of the Southwest and Southern Plains that experience an average of 54–90 days per season. Although in the northern latitudes, key apple growing regions like eastern Washington (Northwest) experienced 9–27 extreme heat days (figure 5(A)) per season that pose sunburn risk to apples. The Northwest and Southwest U.S. along with the Southern Plains experienced increases in extreme heat days (1–5.5 days/decade; figure 5(i)). Conversely, parts of the Midwest, Northern Great Plains, Northeast, and Southeast experienced decreases in extreme heat days ranging from 0.5–2.5 days/decade (figure 5(i)).

Zoom In Zoom Out Reset image size

Despite relatively low likelihood of summer daytime heat, the likelihood of warm nights during the early fruit maturity and coloration period (August–September; figure 2) was substantially higher across the Southeast region of the U.S. (figure 5(B)). Much of the Southern Plains, Southwest, and Southeast experienced on average 48–60 warm nights. The number of warm nights is considerably lower (<24) across much of the Northwest, upper Midwest, Northern Great Plains, and Northeast. Nighttime temperatures are increasing across most of the U.S with the largest changes across the Southwest and Northeast regions, which have the potential to adversely impact apple sweetness and color development (figure 5(ii)).

Changes in these six-climate metrics collectively illustrate the potentially compounding negative effects of climate change on the yields and quality of apple production across the U.S. The spatial variations in these metrics across the U.S. are broadly driven by the spatial patterns of trends in maximum and minimum temperatures (figures S3 and S4). For instance, the greater increases in CPs across the northern half of the U.S. (figure 1) are consistent with the higher rate of warming in the higher latitudes, particularly in January. The more moderate changes in cold degree days, CPs, and bud-break and flowering GDD across the Great Plains and parts of the northern U.S. are likely driven by the regional cooling trend in February (figures S3 and S4). Further, the relatively more widespread increases in January–September GDD and summer heat extremes across the western U.S. are associated with the relatively higher rate of warming across the region compared to the eastern U.S. in several months of the year. These regional differences are particularly amplified in the summer (figures S3 and S4).

Currently, most major and minor apple growing counties are located in the Northwest, Midwest, and Northeast and some minor apple growing counties are in the Southwest and parts of the Southeast (figure 1). States in the Northwest that contribute the highest to national apple production have experienced significant and large changes in several metrics including decreases in cold degree days and increases in CPs, GDD for overall growth, extreme heat days, and warm nights (figures 3–5(i–ii)). The Midwest and Northeast have also experienced similar but more moderate changes in chilling accumulation metrics, increases of similar magnitude in GDD for overall growth, and more widespread increases in GDD for bud-break and flowering and warm nights. These trends have the potential to reduce apple yields by disturbing the phenological cycle, accelerating growth into periods that would result in higher exposure to harmful temperatures, and to reduce apple quality via sunburn and lack of color development.

To assess changing risks in the highest apple-producing counties, we quantify the characteristics (supplementary table 1) and 43-year trends (figure 6) in the six-climate metrics across the leading apple-producing counties—Yakima County, WA (Black), Kent County, MI (Blue), Wayne County, NY (figure 1). Among these three counties, Kent County has the highest interannual variability (standard deviation) in most metrics. All three counties have experienced changes in climate conditions during multiple phenological stages of apples. However, trends across these counties vary substantially due to their geographical locations, which influence their climatology, exposure to climate trends, andconsequently, their vulnerability to changing climate conditions. Yakima County, the highest producer of apples in the U.S., experienced the most significant (at the 5% level) and largest changes amongst these three counties in five of the six metrics.

Zoom In Zoom Out Reset image size

Cold degree days significantly decreased by 2.5-degree days/year or 107.5-degree days from 1979 to 2022 (figure 6(A)) and last day of frost significantly advanced by ∼28 days since 1979 (0.65 days/year; figure 6(B)) for Yakima County. Kent and Wayne counties also had fewer cold degree days, but those changes are not significant despite the magnitude of the trend in Kent County being larger than in Yakima, due to high interannual variability. Despite these trends towards fewer cold degree days, 2013–2014 and 2014–15 recorded the highest cold degree days in these counties. These were associated with an amplification of the North American Winter Temperature Dipole (Wolter et al 2015, Singh 2016). These counties also did not experience significant changes in the last day of frost. It is noteworthy that Wayne County experienced little interannual variation for the last spring frost day in Wayne County, so relatively small deviations had notable effects. For instance, the late frost in April 1981 substantially impacted apples and other tree fruits with fruit growers in the county losing an estimated $28 million due to frost damage (figure 6(B), green line) (Faber and Times 1981).

Early season GDD leading into bud break and flowering showed no significant change amongst all the counties (figure 6(C)). However, all three counties experienced significantly higher GDD over January-September, with Yakima (∼6 degree-days/year), Kent (∼4 degree-days/year), and Wayne (∼2.4 degree-days/year) over 1979–2022 (figure 6(D)). In addition to experiencing the largest increase in January-September GDD, Yakima was the only county that experienced significant increases in the number of extreme heat days, averaging about ∼0.15 days/year or 6.45 days between 1979–2022. Notably, Yakima County experienced 20 days with temperatures >34 °C in 2021, the highest on record, followed closely by 16 days in 2022. While Kent and Wayne County typically experienced fewer extreme heat days with no significant increases during the study period, Kent County had two notable summers ‒ 1988 and 2012 ‒ with 21 and 29 extreme heat days respectively, which were substantially higher than the average (figure 6(E)) (Holmstrom and Ellefson 1990, Whetstone 2014, Lupher et al n.d.). All three counties have experienced significant increases in fall warm nights exceeding 0.10 days per year or ∼4.3 days over our entire analysis (figure 6(F)). Yakima also had the third highest number of warm nights in 2021, which also contributed to the impacts on WA apple producers in 2021.

The maps in figure 7 illustrate the spatial distribution of the PCDI across the U.S. These maps highlight regions subject to multiple climate trends that are potentially harmful to healthy apple growth, quality, and production. The left panel shows that the Northwest, particularly Yakima County—the largest apple-producing county in the U.S.—exhibits up to five detrimental trends, while other areas, such as the Northern Great Plains, show two to three detrimental trends. In contrast, the right panel, which applies a stricter >10% change per decade threshold, reveals that only the Northwest, Southwest, and Northeast exhibit more than two large detrimental trends. This suggests that while multiple regions are experiencing adverse climate impacts, the severity and extent of these trends are more concentrated when using the stricter threshold. These areas with large potentially detrimental trends across multiple metrics overlap with the top apple production regions of the US (figure 1).

Zoom In Zoom Out Reset image size