1. Introduction
As global temperatures have risen in recent years, many regions of the contiguous United States have been observed to experience earlier onset of spring relative to twentieth-century averages (Monahan et al. 2016; Lipton et al. 2018; Ault et al. 2015; Allstadt et al. 2015), defined by a variety of parameters including first bloom (flower emergence) and budburst (the emergence of new leaves) (Allstadt et al. 2015). With the onset of an earlier springtime comes an increased risk of “false springs” or the chance of a return to the much colder (and historically typical) early-spring temperatures after a substantial warm-up (Ault et al. 2013; Allstadt et al. 2015; Chamberlain et al. 2019). This type of variability in late-winter and springtime temperatures has been illustrated to be particularly harmful both economically and ecologically in temperate climates since plants become far more vulnerable to below-freezing temperatures after active growth begins than during winter dormancy. Once growth of sensitive leaf and flower tissue has begun, late-spring frosts (LSFs) can have significant negative impacts on plant growth and reproduction, with damages ranging from reductions in crop yields (Faust and Herbold 2018; Papagiannaki et al. 2014; Snyder and de Melo-Abreu 2005) to impacts upon biodiversity such as the alteration of plant/animal ranges or the creation of ecological mismatches for integrating species, such as plants and their pollinators (Knudson 2012; Ault et al. 2013; Lipton et al. 2018; Chamberlain et al. 2019; Saino et al. 2011; Schweiger et al. 2008).
In the Northeast, the impacts of a warmer climate upon changing seasons have been highlighted in major research studies such as the Fourth National Climate Assessment, where the effects of milder winters and earlier springs upon rural ecosystems, environments, and economies were highlighted in the first key message of the Northeast regional section of the report (Dupigny-Giroux et al. 2018). Of nine different U.S. regions, the Northeast has been identified as having the greatest shift in earlier budburst (1.6 days decade−1) and the third greatest shift in first bloom (1.2 days decade−1) from 1955 to 2013 (Ault et al. 2015). As divisions between spring and winter become more blurred, and warmer temperatures emerge earlier in the year, vulnerabilities of ecosystems ranging from forests to salt marshes are expected to be amplified (Nosakhare et al. 2012; Swanston et al. 2018; Dupigny-Giroux et al. 2018).
Within the Northeast, New Jersey is one of the most densely populated and geographically diverse states, with mountainous terrain in the north of the state, the coastal plain in the southern region of the state (where much agriculture is located), and a long coastline (State of New Jersey: Office of Emergency Management 2019; U.S. Census Bureau 2021). New Jersey’s Scientific Report on Climate Change recently noted that the state is warming at a rate faster than both the Northeast and global averages, with an annual temperature increase of 1.94°C (3.5°F) since 1895. The rate of warming in New Jersey has also increased in the past 50 years, and the 10 warmest years on record have all occurred since 1990 (New Jersey Department of Environmental Protection 2020). Furthermore, the report noted that winter temperatures in the state have seen the greatest increase over time, with average winter temperatures increasing by 2.83°C (5.1°F) in the north, 2.78°C (5.0°F) along the coast, and 2.56°C (4.6°F) in the south since 1895 (New Jersey Department of Environmental Protection 2020). While such exceptional warming in New Jersey may be partially driven by urbanization (New Jersey Department of Environmental Protection 2020), it is also possible that the state’s coastal location plays a role, as sea surface temperatures have also risen exceptionally fast over the northwest Atlantic shelf in recent decades (Karmalkar and Horton 2021). Given the observed increases in average winter temperatures (New Jersey Department of Environmental Protection 2020), and the diverse range of ecosystems and agricultural industries in New Jersey that may impacted by changing transitions from the winter to spring season (New Jersey Department of Environmental Protection 2020; Dupigny-Giroux et al. 2018), it is important to more explicitly understand how springtime temperature variability in the state has already changed. In this work, a study focusing on eight sites from different regions of the state aims to understand how late-winter and springtime temperature variability has evolved in New Jersey—“the Garden State”—from 1950 to 2019. Results show critical spatial and temporal differences in the number of instances of extreme temperature variability as well as the magnitude of temperature variations in New Jersey that suggest a need for both climate adaptation and mitigation efforts moving forward.
2. Methods
a. Data sources and site selection
Data for this project was obtained from National Oceanic and Atmospheric Administration’s (NOAA’s) National Centers for Environmental Information (NCEI) through their Climate Data Online archive (https://www.ncei.noaa.gov/cdo-web/). The data are open access and freely available. Eight sites that include locations at relatively higher elevations in northern New Jersey (Sussex, Belvidere, and Flemington), locations within the coastal plain in central and southern New Jersey (Hightstown and Millville), and locations along New Jersey’s expansive coastline [Cape May, Atlantic City Marina (hereinafter Atlantic City), and Newark] were chosen for analysis (Fig. 1). As a caveat, note that stations for locations in northern New Jersey (Sussex, Belvidere, and Flemington) may be in valleys but are still in distinctly different terrain and located at relatively higher elevations than other sites used in this study. For all locations, daily maximum and minimum temperature records from 1950 to 2019 are assessed, and all temperature records including the months of February–May are at least 94% complete over this time period.
Site map showing the state of New Jersey and the eight locations used in the study. Latitude and longitude values are shown on the map borders. Elevation color-bar units are meters.
Citation: Journal of Applied Meteorology and Climatology 63, 2; 10.1175/JAMC-D-23-0152.1
As with any observational or instrumentation study, analyses presented here require careful control for factors that may have influenced data quality, including station relocations, changes in instrumentation, or changes to the time of day that measurements are collected. To help alleviate many of these concerns, we use only data points that have already passed the rigorous quality control regulations in place for the Climate Data Online archive. As documented by (Menne et al. 2012), quality assurance protocols for the Global Historical Climate Network daily datasets used here include tests for “large, systematic jumps in the annual mean of a record (such as might be caused by a shift in reporting units)” as well as a check for U.S. temperature datasets that include information about the time of observation that tests “for inconsistencies between the reported observation time and the reported temperatures.” Across all datasets used in this analysis, 0.74% of spring data points had to be removed from our analyses due to quality control checks, typically because the data points were flagged as internally inconsistent by NCEI.
Station location changes may produce more subtle variations in temperature that could be missed by the quality control protocols in place for the Climate Data Online archive but that would nonetheless have important impacts on our analyses. We therefore calculate the mean and 95th percentile confidence bounds of minimum temperature, maximum temperature, and daily temperature range on either side of major changes (usually the installation of a new station or a station relocation) at each site used in this study (Table S1 in the online supplemental material). In most instances, neither the means of daily minimum/maximum temperatures, nor the mean of the daily temperature range are statistically different [95% confidence interval (CI)] before and after changes at the station, providing confidence that such station changes do not significantly affect key temperature variables. Even in instances where the means of temperature variables do differ for a 95% CI, most are only marginally statistically different, and are not statistically different for a higher (99% CI) threshold.
Temperature variables that do exhibit statistically different means on either side of a station change even for a 99% CI level are 1) Belvidere minimum temperature and daily range on either side of the station relocation that occurred in 1959; 2) Sussex daily range on either side of the station relocation that occurred in 1960 (a variation that is also associated with a statistically different mean minimum temperature at the 95% CI); 3) Sussex minimum temperature on either side of the station relocation that occurred in 1995; and 4) Sussex minimum temperature and daily temperature range on either side of the station relocation that occurred in 2008. Given these substantial differences, further investigation into the records at these sites is discussed below.
First, an analysis of the mean daily minimum springtime temperatures at both Belvidere from 1954 to 1963 (a time period centered around the station relocation that occurred in January 1959) and Sussex from 1955 to 1965 (where the central year of 1960 is omitted, due to a station relocation March of that year) reveals that both of these northern New Jersey locations experienced unusually cool but regionally consistent minimum temperatures in 1963. It appears that these unusually frigid springtime lows are the driving factor behind statistically lower minimum temperatures that occur for the 5 years after station relocations at Belvidere (1959) and Sussex (1960) relative to the 5 years prior to both station relocations. The increase in daily minimum temperature range that occurs at each of the sites after the station relocation is also consistent with the statistically lower daily minimum temperatures in the later time period at each site that occur without any similar shift in the daily maximum temperatures in either location.
Second, mean daily minimum springtime temperatures at Sussex from 1990 to 1994 are statistically greater than mean daily temperatures at Sussex from 1996 to 1998, which represents the time period after the station was relocated in 1995, but prior to an additional relocation that occurred in June 1998. This statistical difference appears to be driven substantially by unusually warm springtime low temperatures that occurred in 1991 and 1994, prior to the station relocation. While the mean daily low temperatures are warmer in these years than in adjacent years at Sussex, they are regionally consistent with other nearby values. For example, the 1991 mean daily low temperature at Sussex is 34.9°F, which aligns with a similarly warm mean daily low temperature for Belvidere during spring of that year (36.2°F). These relatively warm mean daily low temperatures that occurred during springtime at Sussex prior to the 1995 station move are the likely driver behind statistical differences that are noted before and after the station relocation, rather than the station relocation itself.
Last, mean daily minimum springtime temperatures at Sussex from 2003 to 2012 generally exhibit an upward trend, with significantly warmer means for the 5 years after the station move in October 2007 than for the 5 years prior to the move. However, these warmer temperatures from 2008 to 2012 are once again consistent with other stations in the region, as similarly warm mean daily minimum temperatures also occurred at Belvidere during this time. Mean daily low temperatures during springtime months at Belvidere are in fact within 0.5°F of their Sussex counterparts for 2008, 2010, 2011, and 2012, suggesting that the warmer temperatures during these years are regionally consistent, and not likely to be driven by the station relocation at Sussex. Statistically different temperature ranges at Sussex over these time periods are consistent with the simultaneous occurrence of statistically different minimum temperatures but statistically similar maximum temperatures before and after the station relocation.
b. Statistical analyses
To assess changes in temperature variability during late winter and spring, the full temperature datasets from NCEI are parsed to isolate daily maximum and minimum temperatures between 1 February and 31 May for the years 1950–2019, creating the datasets of late-winter and springtime temperatures. Several different metrics are used to assess temperature variability during these months.
Extreme temperature variation events during these months are defined as instances when the maximum temperature (daily high temperature recorded at each site) rises to or above 15.5°C (60°F), followed by the outdoor minimum temperature (daily low temperature recorded at each site) falling back below 0°C (32°F; freezing). The threshold of 15.5°C is biologically meaningful as this is approximately the temperature where significantly higher photosynthetic activity and growth has been observed in some agricultural plants (Guedira and Paulsen 2002; Hendrickson et al. 2004). Here, the term “extreme” is used in a statistical sense, to describe a relatively rare or tail-area event. We set no artificial bounds on the amount of time between when the temperature first rises above 15.5°C and then falls below 0°C; but record any instance in which this happens during the spring months of a single calendar year as an extreme variation. For example, if on 16 February, the temperature rose above 15.5°C at a given site, and then fell below 0°C again on 2 March at the same site, that would be marked as a single temperature variation. If the temperature at the same site again rose above 15.5°C on 4 March and fell again below 0°C on 7 March, that would be marked as a second extreme temperature variation for that year. This process is followed throughout February–May at all sites in all years of the study. To assess how the frequency of temperature variations change over time at the sites in the study, the number of extreme temperature variations are tracked and tallied across decades, and R2 and P values are calculated to assess the drivers and significance of these shifts [section 3a(1)].
Other analyses assess how the temporal distribution of late-winter and springtime extreme temperature variations changes over time. In this approach, the number of extreme temperature variations in each month (February, March, April, and May) are analyzed at each site on a decadal basis to determine what percentage of total springtime temperature variations occur at each site in each month by decade [section 3a(2)]. For example, if a total of 100 extreme temperature variations occurred at a particular site during the 1950s, and 8 of them occurred in February, 70 occurred in March, and 22 occurred in April, this would mean that in the 1950s at that site, 8%, 70%, and 22% of extreme temperature variations occurred in February, March, and April, respectively, with 0% occurring in May. This analysis is performed for each site during each decade in the study, allowing us to understand how the percentage of total extreme temperature variations occurring in each month (February, March, April, or May) has changed over time at each site.
Using these percentage values, the average percentage of extreme temperature variations in each month during each decade of the study is then compared across specific groupings of sites in the earliest decade in the study (1950) and the latest decade in the study (2010s). Credible intervals to assess significance of monthly mean temperature variations are found by bootstrapping the original datasets. Bootstrapping requires resampling the datasets n times (where n is the number of bootstrap samples) with replacement to produce new samples that are equal in length to the number of samples contained in the original dataset (Efron 1979; Efron and Tibshirani 1993). An illustrative example of the bootstrapping process is provided in Fig. 2. For extreme temperature variation results, 10 000 bootstrap samples were generated to determine the 90% credible interval of monthly mean temperature variations during different time periods.
Schematic to illustrate an example of the bootstrapping process. In this simple example, 12 bootstraps of the original dataset are generated. The means of each bootstrap sample are then calculated and sorted to construct a 66.7% CI for the mean of the original dataset. Note that this is only an illustrative example and that a similar approach could be used to calculate a CI for any statistic of the original dataset (median, mode, etc.).
Citation: Journal of Applied Meteorology and Climatology 63, 2; 10.1175/JAMC-D-23-0152.1
Finally, changes in average variability of day-to-day maximum and minimum temperatures are also analyzed (section 3b). For this assessment, average daily maximum (and minimum) temperatures are calculated for each day from 1 February until 31 May across each decade in the study. For example, at Belvidere, we find the mean high temperature that occurred on 1 February across years 1950–59, the mean high temperature that occurred on 2 February across 1950–59, etc., for all days until 31 May 1950–59. We do the same for all other decades to find the decadal average of the daily high temperature for each day in each of our springtime months, for all decades from the 1950s to the 2010s. The same is repeated for daily minimum temperatures, and this process is carried out at all sites. The daily difference in those average maximum and minimum temperatures is then computed. For example, at Belvidere, the average maximum temperature on 1 February from 1950 to 1959 was 2.83°C (37.1°F), and the average temperature on 2 February from 1950 to 1959 was 4.1°C (39.4°F). This means that the first daily difference in springtime average maximum temperatures at Belvidere from 1950–59 is 1.27°C (2.3°F). This process is carried out for all day-to-day differences for both maximum and minimum decadally averaged daily temperatures at each site. Distributions of day-to-day variations in decadally averaged daily maximum and minimum temperatures for the first two decades (1950–69), and last two decades (2000–19) are then compared at each site. Quantile–quantile (Q–Q) plots are used to compare distributions. When points in the Q–Q plots deviate from a one-to-one line, it indicates that the two distributions being compared are significantly different from one another.
3. Results
a. Instances of extreme temperature variations
1) Changes in total instances of extreme temperature variations
From 1950 to 2019, five of eight sites included in the study exhibit an increase in the number of times that the temperature rises to or above 15.5°C (60°F) before once again falling below 0°C (32°F; hereinafter “extreme temperature variations”) during the months of February–May (Fig. 3). Four of the five sites at which increases in the number of extreme temperature variations are observed (Millville, Hightstown, Atlantic City, and Cape May) are in the central or southern part of the state, located either in the coastal plain, or along the coastline. The fifth site at which increases in the number of extreme temperature variations are observed (Newark) is farther north than the other sites but is still located in a relatively coastal location.
Bar plots showing the total instances of extreme temperature variability per decade for each of the eight sites in the study: (a) Sussex, (b), Belvidere, (c) Newark, (d) Flemington, (e) Hightstown, (f) Millville, (g) Atlantic City, and (h) Cape May. The colors of the bars show the total number of temperature variations each decade that occur in February (purple), March (red), April (tan), and May (brown). Each bar represents one decade from 1950–59 (1950s) through 2010–19 (2010s).
Citation: Journal of Applied Meteorology and Climatology 63, 2; 10.1175/JAMC-D-23-0152.1
Of these five locations, the two most southern sites (Millville and Cape May) both exhibit statistically significant increases in the number of extreme temperature variations across time (P < 0.05; Table 1). R2 values at these two sites suggest that 74%–88% of the variability in the total number of such temperature swings each decade is explained by time—that is, the decade in which variations occur (Table 1, Fig. 3).
Site-by-site statistics of extreme temperature variation trends from 1950 to 2019.
Three additional locations (Hightstown, Newark, and Atlantic City) also exhibit an upward trend over time in the number of extreme temperature variations (Fig. 3). At these sites, all P values fall between 0.05 and 0.1 (Table 1), with the exception of Newark, which falls just over 0.1 (0.1028). In these three locations, R2 values indicate that up to ∼50% of variability in the total number of extreme temperature variations is explained by the decade in which the temperature variations occur (Table 1, Fig. 3).
2) Long-term changes in monthly extreme springtime temperature variations
Across all eight sites included in the study, there are changes in both the temporal distribution of late-winter and springtime extreme temperature variations and the magnitude of late-winter temperature variations from 1950 to 2019.
At all sites, there is an increase over time in the percentage of extreme temperature variations that occur during February. During the 1950s, the average percentage of extreme temperature variations during February across sites is 9.2%. By the 2010s, the average percentage of extreme temperature variations per decade that occur in February increases to 23.7%–an increase that is significant for a 90% CI (Fig. 4a). Similar statistically significant increases in the percentage of total extreme temperature variations during the month of February are noted for both the subsets of coastal sites (Newark, Atlantic City, and Cape May) and inland sites (Sussex, Flemington, Belvidere, Hightstown, and Millville). At coastal sites, the percentage of February temperature variations increases from an average of 11.4% decade−1 in the 1950s to 32.4% decade−1 in the 2010s (Fig. 4a). At inland sites, the percentage of February temperature variations increases from an average of 7.8% decade−1 in the 1950s to 18.5% decade−1 in the 2010s (Fig. 4a).
The left-hand side of each panel shows line plots that depict the percentage of extreme temperature variations during (a) February, (b) March, (c) April, and (d) May for Sussex (light blue), Belvidere (gray), Flemington (pink), Newark (purple), Hightstown (dark blue), Millville (dark orange), Atlantic City (light orange), and Cape May (green). The right-hand side of each panel shows the average percentage of extreme temperature variations during the 1950s (white dot) and 2010s (black dot) for all sites, coastal sites, and inland sites. Error bars with average points represent the 90% credible interval.
Citation: Journal of Applied Meteorology and Climatology 63, 2; 10.1175/JAMC-D-23-0152.1
During the month of March, most sites show a decrease in the percentage of extreme temperature variations per decade; however, the only subset of sites for which this decrease is statistically significant (90% CI) is the set of coastal sites. At coastal sites, the average percentage of extreme temperature variations during March decreases from 73.3% decade−1 during the 1950s to 55.9% decade−1 during the 2010s (Fig. 4b).
There are no statistically significant shifts in the percentage of extreme temperature variations per decade that occur in either April or May. Across all sites, as well as the coastal and inland subsets of sites, there is minimal change in the average percentage of extreme temperature variations per decade that occur in April (Fig. 4c). During May, there are no extreme temperature variations that occur during any decade at any of the coastal sites—a result that is not terribly surprising given that many of these sites typically experience their final freeze much earlier in the season (Office of the New Jersey State Climatologist 2023). At inland sites, there is a decrease in the percentage of May temperature variations from 7.9% during the 1950s to 3.0% during the 2010s; however, this decrease is not statistically significant at the 90% CI (Fig. 4c).
It is also useful to consider the magnitude of extreme temperature variations during each month. Though most months show little change in the magnitude of temperature variations over time, the upper end of the magnitudes of extreme temperature variations during the month of February increases over time (Fig. S1 in the online supplemental material). During February, the 75th percentile of the magnitude of extreme temperature variations increases from 21.9°C (39.5°F) during the 1950s to 23.9°C (43°F) during the 2010s. The total number of extreme temperature variations with a magnitude of 22.2°C (40°F) or greater during February increases from seven (less than 1 yr−1 across sites) during the 1950s to 33 (more than 3 yr−1 across sites) during the 2010s (Fig. S1).
b. Patterns in day-to-day temperature variability
In addition to changing numbers and magnitudes of extreme temperature variations from the 1950s to the 2010s, there is also an evolution in day-to-day differences in daily maximum and minimum temperatures during the months of February–May at many sites considered for the study. At 50% of sites (Sussex, Belvidere, Atlantic City, and Cape May), the magnitude of difference in day-to-day temperature for both maximum and minimum observed temperatures increases from the first 20 years of our study (1950–69) until the final 20 years of the study (2000–19; Figs. 5a,b,g,h). Furthermore, at both Flemington and Hightstown, although there is little change in the variability of day-to-day minimum temperatures over time, there are significant increases in variation of day-to-day maximum temperatures from 1950 to 1969 to 2000–19 (Figs. 5d,e). At Newark, while there is minimum change in day-to-day maximum temperature variability, there is a substantial increase in day-to-day minimum temperature variability by 2000–19 relative to 1950–69 (Fig. 5c). These variations in day-to-day temperatures across sites are most notable in the upper tails of the day-to-day temperature differences, suggesting that at most sites, the largest day-to-day differences in either minimum temperature, maximum temperature, or both have increased since the mid-twentieth century. Note that here, analyses focus on a comparison between only the earliest two decades of data and the most recent two decades of data, to ensure that we assess differences between two periods that are likely to be substantially different from one another in terms of anthropogenic climatological forcing. It is expected that results including a comparison of the intervening decades would produce distributions with magnitudes lying between the two distributions compared in Fig. 5 but that nonetheless align with the overall trend found when comparing the earliest and latest decades.
Q–Q plots comparing the distribution of average day-to-day temperature difference from February to May during 1950–69 (x axis) and during 2000–19 (y axis) for (a) Sussex, (b), Belvidere, (c) Newark, (d) Flemington, (e) Hightstown, (f) Millville, (g) Atlantic City, and (h) Cape May. Orange points show differences in daily maximum temperatures, and purple points show differences in daily minimum temperatures. The dashed gray line is the one-to-one line.
Citation: Journal of Applied Meteorology and Climatology 63, 2; 10.1175/JAMC-D-23-0152.1
4. Discussion and conclusions
Agriculture is strongly influenced by weather and climate, and the consequences of “early springs” coupled with LSF events can be economically devastating. Across North America and Europe, LSFs cause more crop losses than any other climate-related phenomena (Körner et al. 2016; Lamichhane 2021; Snyder and de Melo-Abreu 2005), and the impact of these events is exacerbated by early springs that cause plants to prematurely produce sensitive growing leaf and flower tissues. In Europe, a single LSF across the continent in spring 2017 resulted in economic losses of EUR 3.3 billion (Faust and Herbold 2018).
Although warmer temperatures in winter that can lead to the potential for early springs caused by anthropogenic climate change have been well documented (Dupigny-Giroux et al. 2018; Gulev et al. 2021; Lipton et al. 2018), our work adds to the previous literature by revealing a number of specific patterns in New Jersey late-winter and springtime temperature variability that require both mitigation and adaptation efforts to deal with the impact of changing seasonal temperatures.
An assessment of extreme temperature variations during February–May from 1950 to 2019 reveals that more than 50% of sites included in the study (Millville, Hightstown, Newark, Atlantic City, and Cape May) exhibit increases over time in the number of instances when springtime temperatures climb to or above 15.5°C (60°F) and then fall back to 0°C (32°F) or lower (Fig. 3). All five of these sites are located either along the coast or within the coastal plain and have historically experienced relatively stable temperature transitions during the winter–spring season (typically well under 60 extreme temperature variations per decade from the 1950s–1980s; Fig. 1). Given the coastal location of many of these sites, it is plausible that warming sea surface temperatures have played some role in the increased variability of springtime land temperatures (Karmalkar and Horton 2021). Northern and more inland sites (Sussex, Belvidere, and Flemington) that have historically had larger numbers of extreme temperature variations (often 60–120 days decade−1) have a less clear trend (Figs. 1 and 3)—a result that does not change even if the upper threshold for these more northern/inland sites is lowered to 12.8°C (55°F; Fig. S2 in the online supplemental material).
Furthermore, although increases in extreme temperature variations are mainly present at southern and coastal locations, nearly all sites exhibit more variation in day-to-day late-winter and springtime temperatures over time, leading to temperatures that are generally less consistent by 2000–19 in comparison with 1950–69 (Fig. 5). All sites also show a temporal shift in the timing of extreme temperature variations, with an increase in the percentage of extreme temperature variations that occur in February. While early springs are more likely to induce budburst and early blooming in March than in February (due in part to colder nights and lower solar radiation in February than March), this shift nonetheless highlights an important trend, especially since it is plausible that with further warming of the planet, an increase in February extremes could become more likely to induce early budburst or blooming, given that nights are warming more quickly than days (Seneviratne et al. 2021).
There are several factors that may contribute to the changes in late-winter and springtime temperature variability identified here. While warming sea surface temperatures may play a role in the increased variability for many coastal locations (Karmalkar and Horton 2021), it is also possible that changing precipitation patterns in the region impact seasonal temperature variability. For instance, recent studies have suggested potential increases in extreme springtime rainfall in the Northeast (Whitehead et al. 2023). Cooler, overcast conditions that coincide with these extreme events in a warming world could conceivably contribute to increased temperature variations, while also compounding the challenges facing the agricultural industry due to potentially flooded fields that could delay planting (Wolfe et al. 2018). As another example, growing urbanization in many New Jersey locations, and especially in Newark, could contribute to evolving temperature variation, given that past work has shown that urbanization plays a role in temperature variability, particularly on a day-to-day basis (Tong et al. 2022; Tam et al. 2015). Additionally, it is worth noting that elevation varies substantially across sites, with several of the most northern sites (Essex, Belvidere, and Flemington) located at substantially higher altitudes that most other locations in the study (Fig. 1), which could impact temperature variability results at these locations (Ohmura 2012). Additional research that can fully explore all the driving factors of increased late-winter and early-springtime temperature variability would be highly beneficial to further understand how these changes may continue to evolve in a warming climate.
An increase in winter–spring temperature variability at sites located along New Jersey’s coast and in the coastal plain could have important implications for crops typically grown in these areas as well as fragile ecosystems that rely upon more consistent temperature patterns as spring arrives. For instance, New Jersey, also known as the Garden State, is well known for its orchards and fruit crops; in 2018 the state was ranked third in the nation for production of popular fruits like peaches and cranberries, and sixth in the nation for production of blueberries (New Jersey Department of Agriculture 2019). The same study found that in 2018, peaches and blueberries alone accounted for more than 13 000 acres in New Jersey and had a production value of over $100 million (New Jersey Department of Agriculture 2019). The vast majority of orchards and berry farms are located in the southern portion of New Jersey, in counties that lie either along the coast or in the coastal plain (New Jersey Department of Agriculture 2019). Given that these types of flowering fruits can be highly susceptible to damage during spring frost events, there may be a need to adapt farming practices for such fruit (Reig et al. 2013; Lin and Pliszka 2008) in the face of increasing variability of late-winter and springtime temperatures in the southern portion of the state (Figs. 3 and 4, Table 1). For instance, it may be advantageous to plant varieties of fruit that either bloom later (since data suggests that fewer extreme temperature variations could happen in May, but more are occurring in February over time; Fig. 4) or are shown to be more tolerant of extreme temperature variations (Reig et al. 2013; Lin and Pliszka 2008).
Climate change impacts on temperature variability are furthermore altering the phenology (timing of development and life stage) for many plants and animals, frequently creating ecological mismatches between species and their ecosystems (Cohen et al. 2018; Thackeray et al. 2016; Parmesan and Yohe 2003; Walther et al. 2002). For example, the timing of budburst and leaf development in temperate forests has advanced due to higher early-spring temperatures, which in turn alters the emergence of plant-feeding insects; some insect predators have been able to adjust their phenology to track this food availability, but others have not (Usui et al. 2017). Similarly, due to late-winter/early-spring temperature increases, some annual bird migrations have been shifting earlier in recent decades, as many bird species adapt to these temperature changes (Horton et al. 2020; Usui et al. 2017; Zaifman et al. 2017); however, not all species are able to adjust the onset of their migration, especially long-distance migrants (Usui et al. 2017). Those species that are not leaving wintering grounds earlier are experiencing a mismatch between their arrival to breeding grounds and the peak availability of their insect food (Zaifman et al. 2017), which could have significant negative consequences for the migrants, including increased mortality or decreased fecundity. Furthermore, advanced plant phenology followed by LSFs can kill leaf tissues, as well as the insects that feed upon them (Lombardero et al. 2021; Marquis et al. 2019), resulting in lower food availability for birds during a critical time in their life cycle.
Mutualistic organisms may also be sensitive to the effects of spring temperature shifts and increased variability. Even if species phenologies are all advancing, they may not all be doing so at the same rate, creating the potential for mismatches between species that were tightly linked prior to recent climate changes. For instance, early springs are causing early flower development, but that does not always result in a perfectly matched corresponding shift in pollinator phenology. In fact, early-spring advancement in phenology can be significantly more rapid in some species of butterfly pollinators than their plants (Parmesan 2007), and this decoupling may lead to reduced food availability for pollinators, as well as reduced seed set for their plant mutualists. As another more comprehensive example, a recent modeling study based on 1420 pollinators and 429 plants indicated that climate change has caused phenological shifts and interaction mismatches between flowers and their pollinators (Burkle et al. 2013). Although some other research suggests that in many cases, plants and their pollinators may remain largely in sync phenologically (Bartomeus et al. 2011; Hegland et al. 2009), it is important to note that even if mismatches occur for relatively few species, there could be outsize impacts on vulnerable ecosystems (Brosi et al. 2017; Brosi and Briggs 2013).
Last, early springs coupled with LSF events can negatively impact wild nut-producing trees in New Jersey—such as hickories, walnuts, beeches, oaks, and others—via damage to flowers (Augspurger 2009; Cecich and Sullivan 1999). Flower losses lead to a reduction in nut production, and this results in reduced food availability for numerous species of wildlife (e.g., mice, squirrels, deer) that depend on these resources to survive the winter. It is logical that these effects could have cascading impacts on predators of the aforementioned species.
While the focus of this work is on New Jersey and the evidence available from observational data that suggests potential challenges to ecosystems and agriculture due to greater variability of springtime temperatures, the research also paves the way for broader analyses. For example, additional studies that take a more regional approach (e.g., the Northeast, or the Eastern Seaboard) or focus on longer time periods (e.g., the entire twentieth century) would provide a larger region and more extensive time period over which trends may be observed, with the potential to reveal more complex relationships in temperature variations at different locations. Furthermore, additional studies may benefit from an analysis focused on site-specific drivers of increased springtime temperature variability. While this study presents clear evidence of changing springtime temperatures that have occurred as humans have warmed the planet (IPCC 2021; New Jersey Department of Environmental Protection 2020) that will have an impact on agriculture and ecosystems in New Jersey, additional studies that consider the influence of land use or urbanization on the magnitude of change in springtime temperature variability may provide more direction in how best to mitigate the temperature changes that are occurring in specific locations.
Like many areas in the Northeast, observations indicate that locations around New Jersey have experienced an increase in variability of late-winter and springtime temperatures from the mid-twentieth century to the present. Such variations will be critical to address in this state, which is well known for both its agriculture and unique wildlife and ecosystems. Given that anthropogenic climate change has already led to an observable difference in temperature variation, and that additional warming is expected in coming centuries (Gulev et al. 2021), this work highlights the vital need for both adaptation strategies to ensure sustainable agriculture and ecosystems throughout the New Jersey, as well as mitigation measures to limit the impacts of potential future climate change upon seasonal temperature patterns in the state.
Acknowledgments.
Study conception and design, data collection and analysis, and figure development were performed by author Garner. Both Garner and author Duran contributed to interpretation of results and writing/editing of all drafts of this paper.
Data availability statement.
Datasets used to conduct this study were obtained NOAA’s NCEI through their Climate Data Online archive (https://www.ncei.noaa.gov/cdo-web/). The data are open access and freely available.
REFERENCES
Allstadt, A. J., S. J. Vavrus, P. J. Heglund, A. M. Pidgeon, W. E. Thogmartin, and V. C. Radeloff, 2015: Spring plant phenology and false springs in the conterminous US during the 21st century. Environ. Res. Lett., 10, 104008, https://doi.org/10.1088/1748-9326/10/10/104008.
Augspurger, C. K., 2009: Spring 2007 warmth and frost: Phenology, damage and refoliation in a temperate deciduous forest. Funct. Ecol., 23, 1031–1039, https://doi.org/10.1111/j.1365-2435.2009.01587.x.
Ault, T. R., G. M. Henebry, K. M. De Beurs, M. D. Schwartz, J. L. Betancourt, and D. Moore, 2013: The false spring of 2012, earliest in North American record. Eos, Trans. Amer. Geophys. Union, 94, 181–182, https://doi.org/10.1002/2013EO200001.
Ault, T. R., M. D. Schwartz, R. Zurita-Milla, J. F. Weltzin, and J. L. Betancourt, 2015: Trends and natural variability of spring onset in the coterminous United States as evaluated by a new gridded dataset of spring indices. J. Climate, 28, 8363–8378, https://doi.org/10.1175/JCLI-D-14-00736.1.
Bartomeus, I., J. S. Ascher, D. Wagner, B. N. Danforth, S. Colla, S. Kornbluth, and R. Winfree, 2011: Climate-associated phenological advances in bee pollinators and bee pollinated plants. Proc. Natl. Acad. Sci. USA, 108, 20 645–20 649, https://doi.org/10.1073/pnas.1115559108.
Brosi, B. J., and H. M. Briggs, 2013: Single pollinator species losses reduce floral fidelity and plant reproductive function. Proc. Natl. Acad. Sci. USA, 110, 13 044–13 048, https://doi.org/10.1073/pnas.1307438110.
Brosi, B. J., K. Niezgoda, and H. M. Briggs, 2017: Experimental species removals impact the architecture of pollination networks. Biol. Lett., 13, 20170243, https://doi.org/10.1098/rsbl.2017.0243.
Burkle, L. A., J. C. Marlin, and T. M. Knight, 2013: Plant-pollinator interactions over 120 years: Loss of species, co-occurrence, and function. Science, 339, 1611–1615, https://doi.org/10.1126/science.1232728.
Cecich, R. A., and N. H. Sullivan, 1999: Influence of weather at time of pollination on acorn production of Quercus alba and Quercus velutina. Can. J. For. Res., 29, 1817–1823, https://doi.org/10.1139/x99-165.
Chamberlain, C. J., B. I. Cook, I. García de Cortázar-Atauri, and E. M. Wolkovich, 2019: Rethinking false spring risk. Global Change Biol., 25, 2209–2220, https://doi.org/10.1111/gcb.14642.
Cohen, J. M., M. J. Lajeunesse, and J. R. Rohr, 2018: A global synthesis of animal phenological responses to climate change. Nat. Climate Change, 8, 224–228, https://doi.org/10.1038/s41558-018-0067-3.
Dupigny-Giroux, L.-A., and Coauthors, 2018: Northeast. Impacts, Risks, and Adaptation in the United States: The Fourth National Climate Assessment, Vol. II, U.S. Global Change Research Program, 669–742, https://doi.org/10.7930/NCA4.2018.CH18.
Efron, B., 1979: Bootstrap methods: Another look at the jackknife. Ann. Stat., 7, 1–26.
Efron, B., and R. J. Tibshirani, 1993: An Introduction to the Bootstrap. Chapman and Hall, 456 pp.
Faust, E., and J. Herbold, 2018: Spring frost losses and climate change—Not a contradiction in terms. Munich RE, accessed 16 December 2022, https://www.munichre.com/en/insights/natural-disaster-and-climate-change/spring-frost-losses-climate-change-2018.item-63b78dc6572d54d4c380b34189b6c273.html.
Guedira, M., and G. M. Paulsen, 2002: Accumulation of starch in wheat grain under different shoot/root temperatures during maturation. Funct. Plant Biol., 29, 495–503, https://doi.org/10.1071/PP01006.
Gulev, S. K., and Coauthors, 2021: Changing state of the climate system. Climate Change 2021: The Physical Science Basis, V. Masson-Delmotte et al., Eds., Cambridge University Press, 287–422, https://www.ipcc.ch/report/ar6/wg1/downloads/report/IPCC_AR6_WGI_Chapter02.pdf.
Hegland, S. J., A. Nielsen, A. Lázaro, A. L. Bjerknes, and Ø. Totland, 2009: How does climate warming affect plant–pollinator interactions? Ecol. Lett., 12, 184–195, https://doi.org/10.1111/J.1461-0248.2008.01269.X.
Hendrickson, L., M. C. Ball, J. T. Wood, W. S. Chow, and R. T. Furbank, 2004: Low temperature effects on photosynthesis and growth of grapevine. Plant Cell Environ., 27, 795–809, https://doi.org/10.1111/j.1365-3040.2004.01184.x.
Horton, K. G., and Coauthors, 2020: Phenology of nocturnal avian migration has shifted at the continental scale. Nat. Climate Change, 10, 63–68, https://doi.org/10.1038/s41558-019-0648-9.
IPCC, 2021: Climate Change 2021: The Physical Science Basis. V. Masson-Delmotte et al., Eds. Cambridge University Press, 2391 pp., https://www.ipcc.ch/report/ar6/wg1/downloads/report/IPCC_AR6_WGI_FullReport.pdf.
Karmalkar, A. V., and R. M. Horton, 2021: Drivers of exceptional coastal warming in the northeastern United States. Nat. Climate Change, 11, 854–860, https://doi.org/10.1038/s41558-021-01159-7.
Knudson, W. A., 2012: The economic impact of this spring’s weather on the fruit and vegetable sectors. Strategic Marketing Institute Working Paper 01-52012, 4 pp., https://legislature.mi.gov/documents/2011-2012/CommitteeDocuments/House/Agriculture/Testimony/Committee1-5-30-2012.pdf.
Körner, C., D. Basler, G. Hoch, C. Kollas, A. Lenz, C. F. Randin, Y. Vitasse, and N. E. Zimmermann, 2016: Where, why and how? Explaining the low-temperature range limits of temperate tree species. J. Ecol., 104, 1076–1088, https://doi.org/10.1111/1365-2745.12574.
Lamichhane, J. R., 2021: Rising risks of late-spring frosts in a changing climate. Nat. Climate Change, 11, 554–555, https://doi.org/10.1038/s41558-021-01090-x.
Lin, W., and K. Pliszka, 2008: Blueberry flower spring frost injury in central Poland. Small Fruits Rev., 1, 43–49, https://doi.org/10.1300/J301v01n04_05.
Lipton, D., and Coauthors, 2018: Ecosystems, ecosystem services, and biodiversity. Impacts, Risks, and Adaptation in the United States: The Fourth National Climate Assessment, Vol. II, U.S. Global Change Research Program, 268–321, https://doi.org/10.7930/NCA4.2018.CH7.
Lombardero, M. J., F. Castedo-Dorado, and M. P. Ayres, 2021: Extreme climatic events affect populations of Asian chestnut gall wasps, Dryocosmus kuriphilus, but do not stop the spread. Agric. For. Entomol., 23, 473–488, https://doi.org/10.1111/afe.12448.
Marquis, R. J., J. T. Lill, R. E. Forkner, J. Le Corff, J. M. Landosky, and J. B. Whitfield, 2019: Declines and resilience of communities of leaf chewing insects on Missouri oaks following spring frost and summer drought. Front. Ecol. Evol., 7, 396, https://doi.org/10.3389/fevo.2019.00396.
Menne, M. J., I. Durre, R. S. Vose, B. E. Gleason, and T. G. Houston, 2012: An overview of the Global Historical Climatology Network-Daily database. J. Atmos. Oceanic Technol., 29, 897–910, https://doi.org/10.1175/JTECH-D-11-00103.1.
Monahan, W. B., A. Rosemartin, K. L. Gerst, N. A. Fisichelli, T. Ault, M. D. Schwartz, J. E. Gross, and J. F. Weltzin, 2016: Climate change is advancing spring onset across the U.S. National Park System. Ecosphere, 7, e01465, https://doi.org/10.1002/ecs2.1465.
New Jersey Department of Agriculture, 2019: 2019 annual report and agricultural statistics. 45 pp., https://www.nass.usda.gov/Statistics_by_State/New_Jersey/Publications/Annual_Statistical_Bulletin/2019/2019%20NJ%20Annual%20Report.pdf.
New Jersey Department of Environmental Protection, 2020: 2020 New Jersey scientific report on climate change. 184 pp., https://dep.nj.gov/wp-content/uploads/climatechange/nj-scientific-report-2020.pdf.
Nosakhare, O. K., I. T. Aighewi, A. Y. Chi, A. B. Ishaque, and G. Mbamalu, 2012: Land use–land cover changes in the lower eastern shore watersheds and coastal bays of Maryland: 1986–2006. J. Coastal Res., 28, 54–62, https://doi.org/10.2112/JCOASTRES-D-09-00074.1.
Office of the New Jersey State Climatologist, 2023: Historical monthly station data. Rutgers University, accessed 28 November 2023, http://climate.rutgers.edu/stateclim_v1/monthlydata/index.php.
Ohmura, A., 2012: Enhanced temperature variability in high-altitude climate change. Theor. Appl. Climatol., 110, 499–508, https://doi.org/10.1007/s00704-012-0687-x.
Papagiannaki, K., K. Lagouvardos, V. Kotroni, and G. Papagiannakis, 2014: Agricultural losses related to frost events: Use of the 850 hPa level temperature as an explanatory variable of the damage cost. Nat. Hazards Earth Syst. Sci, 14, 2375–2386, https://doi.org/10.5194/nhess-14-2375-2014.
Parmesan, C., 2007: Influences of species, latitudes and methodologies on estimates of phenological response to global warming. Global Change Biol., 13, 1860–1872, https://doi.org/10.1111/j.1365-2486.2007.01404.x.
Parmesan, C., and G. Yohe, 2003: A globally coherent fingerprint of climate change impacts across natural systems. Nature, 421, 37–42, https://doi.org/10.1038/nature01286.
Reig, G., I. Iglesias, C. Miranda, F. Gatius, and S. Alegre, 2013: How does simulated frost treatment affect peach [Prunus persica (L.)] flowers of different cultivars from worldwide breeding programmes? Sci. Hortic., 160, 70–77, https://doi.org/10.1016/j.scienta.2013.05.021.
Saino, N., and Coauthors, 2011: Climate warming, ecological mismatch at arrival and population decline in migratory birds. Proc. Roy. Soc., 278B, 835–842, https://doi.org/10.1098/rspb.2010.1778.
Schweiger, O., J. Settele, O. Kudrna, S. Klotz, and I. Kühn, 2008: Climate change can cause spatial mismatch of trophically interacting species. Ecology, 89, 3472–3479, https://doi.org/10.1890/07-1748.1.
Seneviratne, S. I., and Coauthors, 2021: Weather and climate extreme events in a changing climate. Climate Change 2021: The Physical Science Basis, V. Masson-Delmotte et al., Eds., Cambridge University Press, 1513–1766, https://www.ipcc.ch/report/ar6/wg1/downloads/report/IPCC_AR6_WGI_Chapter11.pdf.
Snyder, R. L., and J. de Melo-Abreu, 2005: Frost Protection: Fundamentals, Practice and Economics. Food and Agriculture Organization of the United Nations, 223 pp.
State of New Jersey: Office of Emergency Management, 2019: 2019 New Jersey State Hazard Mitigation Plan. https://www.nj.gov/njoem/mitigation/2019-mitigation-plan.shtml.
Swanston, C., and Coauthors, 2018: Vulnerability of forests of the Midwest and northeast United States to climate change. Climatic Change, 146, 103–116, https://doi.org/10.1007/s10584-017-2065-2.
Tam, B. Y., W. A. Gough, and T. Mohsin, 2015: The impact of urbanization and the urban heat island effect on day to day temperature variation. Urban Climate, 12, 1–10, https://doi.org/10.1016/j.uclim.2014.12.004.
Thackeray, S. J., and Coauthors, 2016: Phenological sensitivity to climate across taxa and trophic levels. Nature, 535, 241–245, https://doi.org/10.1038/nature18608.
Tong, X., P. Wang, S. Wu, and M. Luo, 2022: Urbanization effects on high-frequency temperature variability over South China. Urban Climate, 42, 101092, https://doi.org/10.1016/j.uclim.2022.101092.
U.S. Census Bureau, 2021: Northeast region: 2020 census. Accessed 9 August 2022, https://www.census.gov/library/stories/state-by-state/northeast-region.html.
Usui, T., S. H. M. Butchart, and A. B. Phillimore, 2017: Temporal shifts and temperature sensitivity of avian spring migratory phenology: A phylogenetic meta-analysis. J. Anim. Ecol., 86, 250–261, https://doi.org/10.1111/1365-2656.12612.
Walther, G. R., and Coauthors, 2002: Ecological responses to recent climate change. Nature, 416, 389–395, https://doi.org/10.1038/416389a.
Whitehead, J. C., and Coauthors, 2023: Northeast. Fifth National Climate Assessment, A. R. Crimmins et al., Eds., U.S. Global Change Research Program, https://doi.org/10.7930/NCA5.2023.CH21.
Wolfe, D. W., and Coauthors, 2018: Unique challenges and opportunities for northeastern US crop production in a changing climate. Climatic Change, 146, 231–245, https://doi.org/10.1007/s10584-017-2109-7.
Zaifman, J., D. Shan, A. Ay, and A. G. Jimenez, 2017: Shifts in bird migration timing in North American long-distance and short-distance migrants are associated with climate change. Int. J. Zool., 2017, 6025646, https://doi.org/10.1155/2017/6025646.
