NCRPIS Publications and Papers North Central Regional Plant Introduction Station
2-2012 Horticultural Applications of a Newly Revised USDA Plant Hardiness Zone Map Mark P. Widrlechner United States Department of Agriculture, [email protected]
Christopher Daly Oregon State University
Markus Keller Washington State University
Kim Kaplan United States Department of Agriculture
Follow this and additional works at: http://lib.dr.iastate.edu/ncrpis_pubs Part of the Agricultural Science Commons, Agriculture Commons, Agronomy and Crop Sciences Commons, and the Horticulture Commons The ompc lete bibliographic information for this item can be found at http://lib.dr.iastate.edu/ ncrpis_pubs/5. For information on how to cite this item, please visit http://lib.dr.iastate.edu/ howtocite.html.
This Article is brought to you for free and open access by the North Central Regional Plant Introduction Station at Iowa State University Digital Repository. It has been accepted for inclusion in NCRPIS Publications and Papers by an authorized administrator of Iowa State University Digital Repository. For more information, please contact [email protected]. Horticultural Applications of a Newly Revised USDA Plant Hardiness Zone Map
Abstract The ca curate prediction of winter injury caused by low-temperature events is a key component of the effective cultivation of woody and herbaceous perennial plants. A common method employed to visualize geographic patterns in the severity of low-temperature events is to map a climatological variable that closely correlates with plant survival. The .SU . Department of Agriculture Plant Hardiness Zone Map (PHZM) is constructed for that purpose. We present a short history of PHZM development, culminating in the recent production of a new, high-resolution version of the PHZM, and discuss how such maps relate to winterhardiness per se and to other climatic factors that affect hardiness. The new PHZM is based on extreme minimum-temperature data logged annually from 1976 to 2005 at 7983 weather stations in the United States, Puerto Rico, and adjacent regions in Canada and Mexico. The PHZM is accessible via an interactive website, which facilitates a wide range of horticultural applications. For example, we highlight how the PHZM can be used as a tool for site evaluation for vineyards in the Pacific northwestern United States and as a data layer in conjunction with moisture-balance data to predict the survival of Yugoslavian woody plants in South Dakota. In addition, the new map includes a zip code finder, and we describe how it may be used by governmental agencies for risk management and development of recommended plant lists, by horticultural firms to schedule plant shipments, and by other commercial interests that market products seasonally.
Keywords winter injury, zip code, GIS, minimum temperature, Vitis, climate
Disciplines Agricultural Science | Agriculture | Agronomy and Crop Sciences | Horticulture
Comments This article is from HortTechnology 22, no. 1 (February 2012): 6–19.
Rights Works produced by employees of the U.S. Government as part of their official duties are not copyrighted within the U.S. The onc tent of this document is not copyrighted.
This article is available at Iowa State University Digital Repository: http://lib.dr.iastate.edu/ncrpis_pubs/5 when plant tissues optimally achieve a maximal degree of cold acclimation, extreme low-temperature events may still overwhelm adaptive survival mech- anisms. 3) And finally, during the late winter and early spring, plants may deharden when exposed to tempera- tures above freezing, having satisfied physiological rest requirements (Litzow Feature and Pellett, 1980), and can then suf- fer damage from subsequent low- temperature events. Studies that thoroughly document the seasonal progression of acclimation and provide graphical representations of the timing of actual or potential injury Horticultural Applications of a Newly Revised from low-temperature events at all three stages in the annual cycle include those USDA Plant Hardiness Zone Map by McNamara et al. (2002), McNamara and Pellett (1993), Mills et al. (2006), 1,5 2 3 Scheiber et al. (2002), Schrader and Mark P. Widrlechner , Christopher Daly , Markus Keller , Graves (2003), Szalay et al. (2010), and Kim Kaplan4 and Va¨ino¨la¨ et al. (1997). Hardiness-zone basics ADDITIONAL INDEX WORDS. winter injury, zip code, GIS, minimum temperature, Of the three stages when injury Vitis, climate often occurs, the frequency and se- SUMMARY. The accurate prediction of winter injury caused by low-temperature verity of midwinter, low-temperature events is a key component of the effective cultivation of woody and herbaceous events have historically received con- perennial plants. A common method employed to visualize geographic patterns in siderable attention by plant scientists. the severity of low-temperature events is to map a climatological variable that closely Specific weather events causing plant correlates with plant survival. The U.S. Department of Agriculture Plant Hardiness injury on a case-by-case basis may yield Zone Map (PHZM) is constructed for that purpose. We present a short history of insights on factors influencing adap- PHZM development, culminating in the recent production of a new, high- tation [see Bachtell and Green (1985) resolution version of the PHZM, and discuss how such maps relate to winter- for an example from the Chicago hardiness per se and to other climatic factors that affect hardiness. The new PHZM region and Gu et al. (2008) for a de- is based on extreme minimum-temperature data logged annually from 1976 to 2005 at 7983 weather stations in the United States, Puerto Rico, and adjacent tailed discussion of the Spring 2007 regions in Canada and Mexico. The PHZM is accessible via an interactive website, freeze in the eastern and central United which facilitates a wide range of horticultural applications. For example, we States], but such weather events are highlight how the PHZM can be used as a tool for site evaluation for vineyards in not repeatable and do not lend them- the Pacific northwestern United States and as a data layer in conjunction with selves to experimentation and hypoth- moisture-balance data to predict the survival of Yugoslavian woody plants in South esis testing. However, further insights Dakota. In addition, the new map includes a zip code finder, and we describe how it can be gained by shifting emphasis may be used by governmental agencies for risk management and development of away from individual weather events recommended plant lists, by horticultural firms to schedule plant shipments, and by (or even seasons) to longer time frames other commercial interests that market products seasonally. that document multiple, extreme events on a climatic scale. Heinze and Schreiber orticulturists have long recog- plants,’’ so the ability to forecast the (1984) presented a comprehensive re- nized that the accurate pre- risks associated with such factors is ex- view of this topic, detailing the history Hdiction of winter injury is a tremely valuable. and application of long-term climato- key component of the effective culti- Freeze injury to various plant tis- logical data to relate patterns of woody- vation of long-lived woody and herba- sues and organs typically occurs at three plant adaptation to low-temperature ceous perennial plants in many climates. stages in the annual cycle (Larcher, injury. Winter injury can limit long-term plant 2005; Raulston and Tripp, 1994): 1) A BRIEF HISTORY OF HARDINESS survival and vigor and can reduce During the autumn, when plants cease ZONES AND ZONE MAPS. A relatively production of valuable horticultural growth (Kalcsits et al., 2009) and begin simple method used to visualize geo- products, including flowers, foliage, to harden or acclimate to winter con- graphic patterns of the biological se- fruit, and seeds. As noted by Skinner ditions (often signaled by decreasing verity of low-temperature events is to (1962), ‘‘frost dates, length of the grow- photoperiod and temperature), early, map a climatological variable that closely ing season, and minimum winter tem- low-temperature events can exceed a correlates with patterns of plant sur- peratures are among the least readily plant’s (or a specific tissue’s) ability to vival. Rehder (1927) developed the controlled of the major factors gov- withstand the event. 2) During the first such map for the United States, erning the geographic adaptability of lowest temperatures of midwinter, with a mapped zonation system that
6 • February 2012 22(1) related winter minimum temperatures subsequent updates using more recent to represent areas with PH > 4.4 C to the survival of specific woody plants. meteorological data appeared in var- (40 F) that are essentially frost-free. He roughly divided the temperate ious publications in 1951, 1967, and Even in parts of the world that do not portion of the conterminous United 1971 (Wyman and Flint, 1985). normally experience freeze events or States and southern Canada into eight This lack of uniformity in zone a well-defined winter season, plants can zones based on the mean temperature intervals prompted the USDA Agri- experience chilling injury at tempera- of the coldest month, each zone span- cultural Research Service (ARS) to tures below about 10 C (50 F) ning 2.8 C(5 F). develop its own ‘‘Plant Hardiness (Levitt, 1980). Thus, knowledge about Shortly thereafter, Kincer (1928) Zone Map’’ with zones defined uni- geographic patterns of ‘‘warmer’’ low- produced a similar map, but based in- formly by differences in the PH sta- temperature extremes is also horticul- stead on ‘‘the mean annual extreme tistic of 5.6 C (10 F) (USDA, 1960, turally valuable in frost-free regions minimum temperature’’ [the lowest 1965). Discrepancies between the zone and can provide guidance in colder temperaturerecordedin1year—herein designations of Wyman’s and USDA’s regions on how best to manage the referred to as the plant hardiness (PH) maps caused some confusion (Skinner, outdoor cultivation of tender plants statistic], scaled by 5.6 C (10 F) in- 1962), but the USDA’s consistent zone that experience chilling injury. tervals. Although he made this change designations became the standard for Hardiness-zone maps based on without the benefit of extensive data assessing plant hardiness in the United this same general zonation system have demonstrating the superiority of ex- States with the release of a comprehen- been developed for many parts of the treme low-temperature events, recently sive PHZM update in 1990 (Cathey, world, including Australia (Dawson, Quamme et al. (2009) presented a 1990). 1991), China (Widrlechner, 1997a), 90-year dataset from the Okanagan The 1990 version of the PHZM Europe (Heinze and Schreiber, 1984), Valley, BC, Canada, that verified a was based on the PH statistic for the Japan (Hayashi, 1990), southern strong association between such events United States, Canada, and Mexico Africa (Pienaar, 1996), and Ukraine and winter injury for a wide range of (Cathey, 1990; Cathey and Heriteau, (Widrlechner et al., 2001). In various fruit crops. 1990). The map included ten 5.6 C parts of the world, including the United In 1936, a slightly different ap- (10 F) zones; zones 2–10 were sub- States (DeGaetano and Shulman, 1990; proach was taken by Wyman [U.S. divided into 2.8 C(5 F) half zones Sunset, 2010; Vogel et al., 2005), al- Department of Agriculture (USDA), (Table 1). Zone 11 was introduced ternative zonation systems that do 1936], who also produced a revised plant hardiness map for the United States, this time based on the PH Table 1. Comparison of the updated and 1990 Plant Hardiness Zone Map statistic averaged over the years (PHZM) zone temperature ranges. The updated PHZM has full 5.6 C (10 F) 1895–1935. However, one limitation zones, which are defined by the numbers 1–13. These are separated into 2.8 C with the 1936 map was that its zones (5 F) half zones, denoted by an a or b. were not based on consistent tempera- Temperature range ture intervals; some were 2.8 C(5 F), 1990 zones Updated zones C F whereas others were 5.6 C (10 F) or 8.3 C(15 F). This map and 1 1a –51.1 to –48.3 –60 to –55 1 1b –48.3 to –45.6 –55 to –50 2a 2a –45.6 to –42.8 –50 to –45 This journal paper of the Iowa Agriculture and Home Economics Experiment Station, Ames, Iowa, Project 2b 2b –42.8 to –40.0 –45 to –40 No. 1018, was supported by USDA-ARS Research 3a 3a –40.0 to –37.2 –40 to –35 Project 3625-21000-053, Hatch Act and State of Iowa funds, and a Specific Cooperative Agreement 3b 3b –37.2 to –34.4 –35 to –30 between the USDA Agricultural Research Service and 4a 4a –34.4 to –31.7 –30 to –25 Oregon State University. 4b 4b –31.7 to –28.9 –25 to –20 We thank Peter Bretting, Candice Gardner, William 5a 5a –28.9 to –26.1 –20 to –15 Graves, Rita Hummel, S. Elwynn Taylor, and three anonymous reviewers for providing valuable com- 5b 5b –26.1 to –23.3 –15 to –10 ments and suggestions for improving the manuscript; 6a 6a –23.3 to –20.6 –10 to –5 Michael Wisniewski for assistance in supplying useful 6b 6b –20.6 to –17.8 –5 to 0 references; and Mike Halbleib for assistance with graphics. 7a 7a –17.8 to –15.0 0 to 5 Mention of commercial brand names does not con- 7b 7b –15.0 to –12.2 5 to 10 stitute an endorsement of any product by the USDA 8a 8a –12.2 to –9.4 10 to 15 or cooperating agencies. 8b 8b –9.4 to –6.7 15 to 20 1USDA-ARS, North Central Regional Plant Intro- 9a 9a –6.7 to –3.9 20 to 25 duction Station, IA State University, G212 Agron- omy Hall, Ames, IA 50011 (retired) 9b 9b –3.9 to –1.1 25 to 30 10a 10a –1.1 to 1.7 30 to 35 2PRISM Climate Group, 2000 Kelley Engineering Center, Oregon State University, Corvallis, OR 10b 10b 1.7 to 4.4 35 to 40 97331 11 11a 4.4 to 7.2 40 to 45 3Irrigated Agriculture Research and Extension Cen- 11 11b 7.2 to 10.0 45 to 50 ter, Washington State University, 24106 North Bunn — 12a 10.0 to 12.8 50 to 55 Road, Prosser, WA 99350 — 12b 12.8 to 15.6 55 to 60 4USDA-ARS, George Washington Carver Center, 5601 Sunnyside Avenue, Beltsville, MD 20705 — 13a 15.6 to 18.3 60 to 65 5Corresponding author. E-mail: [email protected]. — 13b 18.3 to 21.1 65 to 70
• February 2012 22(1) 7 FEATURE not rely upon the PH statistic have However, many horticulturists as modifying factors to refine the PH been developed, but are generally not apply hardiness zones in a prospective statistic, but the underlying method- widely used in horticultural research, manner, with an expectation that past ology has not been published. Few with the exception of the multivariate climatic records can serve to forecast published studies systematically com- zonation system developed for Canada current and future plant performance, pare the utility of the PH statistic to (Ouellet and Sherk, 1967), which is as related to low-temperature damage. other metrics related to winterhardi- discussed below. Fortunately, for many parts of the ness. In Canada, where plant adapta- Preliminary efforts to update the United States, hardiness-zone bound- tion is typically strongly limited by 1990 PHZM in 2003 (Ellis, 2003) aries have shifted relatively little, as winter severity, Ouellet and Sherk relied on outdated methodologies, evidenced by the 1960 (USDA, 1960), (1967) determined that geographic leading to a draft map that was not the 1990 (Cathey, 1990), and the variation in plant adaptation was best adopted by USDA-ARS. In 2004, recently updated maps, even though explained by a multivariate approach USDA-ARS initiated another effort the zone boundaries are a function of involving a wide range of climatic to update the 1990 map, guided by the extreme weather events in the metrics, with emphasis on the PH modern standards of geostatistical various years used to create the maps. statistic but also including the frost- analysis, accuracy, and resolution. Re- Table 2 illustrates this relative stability free period, summer rainfall, maxi- cently released products of these ef- among hardiness zones, as depicted in mum temperature, snow cover and forts are described in detail below. In the three maps, for the 11 largest Met- wind (Agriculture Canada, 1981). addition, Daly et al. (2012) present ropolitan Statistical Areas in the United Their zonation system has been suc- a thorough discussion of the technical States (U.S. Census Bureau, 2003). cessfully applied in Canadian horti- approaches taken to conduct this But the picture is somewhat dif- cultural research (Richer et al., 2006; work. ferent when differences in the PH Richer-Leclerc et al., 1994, 1996; HOW DO HARDINESS ZONES statistic were evaluated solely for the Rioux et al., 2000, 2004) and formed RELATE TO WINTERHARDINESS (OR two 15-year periods that encompass the basis of an updated map generated LOW-TEMPERATURE SURVIVAL)? The the 30-year interval (1976–2005) used from modern climatic interpolation ability of a plant to survive low tem- to develop the most recent update techniques (McKenney et al., 2001). peratures depends on its genetically (Daly et al., 2012). In that comparison, In the north-central United and environmentally controlled adap- most of the conterminous United States, Widrlechner et al. (1992) com- tations and its acclimation status. The States was one half zone warmer in pared the merits of the PH statistic to utility of plant hardiness zone maps the period 1991–2005 than in 1976– that of January mean temperatures for measuring winter injury is a func- 90. Clearly, the world’s climate is and the proportion of years with tion of the correlation of the PH dynamic (Intergovernmental Panel temperatures below –32 C (–25.6 F) statistic to the frequency and severity on Climate Change, 2007), and zone in multiple-regression analyses designed of low-temperature events that dam- boundaries may change in the future, to explain first-year and overall sur- age aboveground plant parts over ap- based on modeled relationships be- vival of 27 accessions of woody plants propriate time intervals and space. In tween climate change and the occur- collected in Yugoslavia when grown addition, it also is important to rec- rence of extreme low-temperature at 14 sites across the region. They ognize and account for important events (Diffenbaugh et al., 2005). found no significant differences in the modifying factors. These include en- Sabuco (1989) created a modi- strengths of regression models involv- vironmental factors, such as levels of fied plant hardiness zone map called ing these three different measures, moisture, temperature, and light, and Floradapt that incorporated snow suggesting that at least for European proper photoperiod regimens, that cover, cloud cover, and wind speeds plants introduced into their target influence the full expression of the physiological and physical mecha- nisms that promote winter adaptation Table 2. Comparison of plant hardiness zones mapped for the 11 largest and those that hinder it, such as abrupt, Metropolitan Statistical Areas [MSAs (U.S. Census Bureau, 2003)] in the 1960, wide temperature fluctuations and 1990, and recently revised Plant Hardiness Zone Maps. atypical dehardening events (Larcher, MSA 1960 map 1990 map Revised map 2005; Levitt, 1980; Olsen et al., 2004; Raulston and Tripp, 1994). New York–Northern New Jersey–Long 6–7 6b–7a 7a–7b Because of the dynamic nature of Island, NY–NJ–PA climate, the most appropriate use of Los Angeles–Long Beach–Santa Ana, CA 9–10 9b–10b 10a–11a plant hardiness zones is for the retro- Chicago–Naperville–Joliet, IL–IN–WI 5–6 5a–5b 5b–6a active evaluation of plant perfor- Philadelphia–Camden–Wilmington, 7 6b–7a 7a–7b mance within the same time span PA–NJ–DE used to generate the map being ap- Dallas–Fort Worth–Arlington, TX 7–8 7b–8a 8a plied. For example, Widrlechner et al. Miami–Fort Lauderdale–Miami Beach, FL 10 10a–10b 10b (1992) evaluated the performance of Washington–Arlington–Alexandria, 7 6b–7a 7a–7b woody plants from Yugoslavia in the DC–VA–MD north-central United States between Houston–Baytown–Sugar Land, TX 9 8b–9a 9a 1975 and 1989, which roughly cor- Detroit–Warren–Livonia, MI 5–6 5b–6b 5b–6b responds to the period of record for Boston–Cambridge–Quincy, MA–NH 6 6a 6a–6b the 1990 PHZM. Atlanta–Sandy Springs–Marietta, GA 8 7a–7b 7b–8a
8 • February 2012 22(1) region, they were equivalent in value the impetus for the broader, multi- Another important climatic factor, as predictors of winterhardiness. variate climatic zonation system de- moisture balance (typically expressed HOW DO HARDINESS ZONES RE- veloped by Sunset in California, first as a ratio between precipitation and LATE TO GENERAL PLANT ADAPTATION? exclusively for the western United actual or potential evapotranspiration), Clearly, the PH statistic is an impor- States, but expanded in 1997 to the has long been considered a key determi- tant determinant (or a strong corre- eastern United States (Sunset, 1997) nant in vegetation (plant-community) late of determinants) of the long-term and, in 2001, to southwestern Canada, classification (Mather and Yoshioka, survival and adaptation of perennial Alaska, and Hawaii (Brenzel, 2001). 1968; Stephenson, 1990, 1998). It has plants, but other climatic factors can also be very strong determinants (Skinner, 1962; Thompson et al., 2000). Two of the most widely stud- ied of these determinants are warm- season heat and moisture balance. Horticulturists have identified three primary variables related to warm-season heat and plant adapta- tion: night temperature (Deal and Raulston, 1989), annual days above 30 C(86 F) [American Horticultural Society (AHS) heat zones (Cathey, 1997)], and lack of sufficient heat units (Pigott, 1981; Pigott and Huntley, 1981). Notably, there are highly sig- nificant positive correlations between AHS heat zones (Table 3) and plant hardiness zones in the central and southeastern United States (Fig. 1A and B). Strong correlations between these two important climatic deter- minants strengthen the apparent prac- tical value of plant hardiness zones for predicting general plant adapta- tion. However, this relationship is ab- sent in some parts of the western United States, such as central California, where heat zones can vary from 2 [1 to 7 d per year >30 C (86 F)] to 9 [>120 to 150 d per year >30 C (86 F)] within plant hardiness zone 9 [–1.1 to –6.7 C (30 to 20 F)] (Fig. 1C). Attempts to understand the highly variable climate of California on plant adaptation may have been
Table 3. American Horticultural Society Plant Heat Zones (adapted from Cathey, 1997). Avg number of days per Heat zone yr > 86 F (30 C) 1 <1 2 1–7 3 8–14 Fig. 1. Correlations between updated Plant Hardiness Zones (Table 1; x-axis) and 4 15–30 American Horticultural Society (AHS) Heat Zones (Cathey, 1997) (Table 3; y-axis) 5 31–45 for three transects. (A) Two-degree latitudinal intervals along a long. 95 W transect 6 46–60 extending from lat. 30 N to lat. 48 N in the central United States (P value of 7 61–90 regression line < 0.0001); (B) two-degree latitudinal intervals along a long. 82 W 8 91–120 transect extending from lat. 26 30’’N to lat. 36 30’’N in the southeastern United 9 121–150 States (P value of regression line < 0.0003); and (C) fifteen-minute longitudinal 10 151–180 intervals along a lat. 36 45’’N transect extending from long. 120 30’’W to long. 121 45’’W in central California, where patterns of summer heat vary widely because 11 181–210 of maritime influences but winter, low temperatures vary little (P value of regression 12 >210 line < 0.32).
• February 2012 22(1) 9 FEATURE also been used in conjunction with low- The newly revised PHZM of the new PHZM and its development temperature metrics for horticultural A newly updated, interactive are presented by Daly et al. (2012). studies related to plant introduction in PHZM for the United States and Ukraine (Kokhno and Kurdyuk, 1994) Puerto Rico has recently been pro- How the map was created and the north-central United States duced (Fig. 2) and made available to In 2004, the USDA-ARS assem- (Widrlechner, 1994) and to the risk bled a technical review team (TRT), the public via the Internet at http:// that cultivated plants can become nat- including representativesfrom the hor- planthardiness.ars.usda.gov (USDA, uralized (Widrlechner and Iles, 2002). ticulture and nursery industries, public Specific examples where the PH sta- 2012). Methods applied to produce gardens, agro-meteorologists, clima- tistic was used jointly with moisture this high-resolution map, its attributes, tologists, and plant scientists, and they balance (and other environmental var- and key features that enable new appli- developed technical guidelines and iables) include studies to identify cli- cations for horticultural research and suggested ways to present the result- matic analogues as part of the planning management, which were difficult or ing information that maximize its and/or execution of plant explorations impossible to perform with the 1990 value to researchers, the horticultural in China (Widrlechner, 1997b) and PHZM, are described in the following producers, gardeners, and govern- Ukraine (Widrlechner et al., 2001). sections. A more detailed description mental agencies. The TRT followed
Fig. 2. PRISM 1976–2005 Plant Hardiness Zone Maps: (A) conterminous United States; (B) Alaska; (C) Hawaii; and (D) Puerto Rico. Zone shading key (color online) is given at the lower right. See Table 1 for temperature ranges of zones (1 km = 0.6214 mile). (Figure appears in color online and on cover.)
10 • February 2012 22(1) a flexible, multidisciplinary approach, associated data and provided detailed minimum temperature of the coldest incorporating input from horticul- feedback to the PRISM Climate Group. month of the year, derived from the tural industry and professional orga- During the draft map review process, official PRISM climate datasets for nizations, such as the AHS, American the 30-year period also emerged as the USDA (Daly et al., 2008). The Nursery and Landscape Association, the interval that best matched the PHZM was produced at high resolu- and American Public Gardens Associ- expectations, perceptions, and expe- tion (800 m in the conterminous ation, as well as from within the USDA rience of TRT reviewers. United States, 4 km in Alaska, and and the academic community. Temperature data from 7983 400 m in Hawaii and Puerto Rico) The TRT recommended that a stations were obtained for the United and divided into thirteen 5.6 C(10 F) new PHZM incorporates the most States and Puerto Rico, and adjacent full zones and twenty-six 2.8 C(5 F) recent and accurate meteorological regions in Canada and Mexico (to fill half zones (Table 1). Maps of the data, applies the most advanced inter- in gaps in coverage near these inter- standard deviation of the 1976–2005 polation methods, and focuses only national borders). Stations located in PH statistic were created with the on plant cold hardiness, as measured mountainous areas were included in same methods, and incorporated the by the PH statistic. Team members large numbers for the first time. These same stations, as did the PHZM in- recognized that many other climatic included 583 USDA Natural Resources terpolation (Daly et al., 2012). datasets were available, potentially Conservation Service SNOTEL (Snow offering considerable insight into the Telemetry) stations, located at high General features of the geographic patterning of factors re- elevations in the western United updated PHZM lated to plant adaptation. Nonetheless, States. The data were carefully quality- The latitudinal delineation of they decided to retain the existing controlled for outliers and other in- zones in the central part of the con- system because of its widespread adop- consistent values (methods are detailed terminous United States is readily tion, including the availability of esti- in Daly et al., 2008). apparent, ranging from zone 3a in mates of winterhardiness based on the The 1976–2005 PHZM was pro- extreme northern Minnesota to zone PH statistic for thousands of plants duced with the most recent version 10a in extreme southeastern Texas [three online examples include the of parameter-elevation regressions on (Fig. 2). Although the eastern United NGA Plant Finder (National Garden- independent slopes model (PRISM), States has somewhat similar latitudi- ing Association, 2010), Cornell Uni- a well-known climate mapping tech- nal patterns, they are modulated by versity’s Flower Growing Guides nology that has generated official elevational and coastal influences. For (Cornell University, 2006), and the USDA 1961–90 digital climate grids example, zone 6a extends southward UConn Plant Database (Brand, 2001)] and 1971–2000 updates (Daly, 2006; along the Appalachian Mountains and the assortment of compatible Daly et al., 1994, 2002, 2008). PRISM into northern Georgia, but also ex- hardiness-zone maps available interna- develops local regression functions tends along the Atlantic coastline as tionally. In addition, the TRT agreed (one for each grid cell on a map) be- far north as Maine. A combination of to limit the new map’s geographic tween a predictor grid of an explana- latitudinal and maritime influences focus to the United States and Puerto tory variable (most commonly elevation) (particularly noticeable along the At- Rico since Canada had made significant and a climatic variable, for every grid lantic coast, see Fig. 2A) results in progress with its alternative mapping cell. Data from surrounding stations, extremely mild zones that are essen- scheme (McKenney et al., 2001) and, weighted by their geographic similar- tially frost-free in southern Florida at the time, the density of Mexican ity to the grid cell being modeled, pop- (zones 11a and 11b). weather stations with long-term data ulate the regression function. PRISM Zone patterns in the western was unclear. In 2007, on the basis of accounts for the effects of elevation, United States exhibit only an indis- its track record as a leader in climatic terrain-induced air-mass blockage, tinct latitudinal gradient; instead, they map development (e.g., Daly, 2006; coastal proximity, temperature inver- are dominated by relatively mild ma- Daly et al., 1994, 2008), Oregon State sions, and cold-air pooling on extreme rine influences along the West Coast, University’s PRISM Climate Group minimum-temperature patterns (Daly elevational effects in the mountains, was tasked by USDA-ARS with de- et al., 2008). More information about and cold-air pools in many interior veloping the updated PHZM. PRISM can be obtained from PRISM valleys. The Cascade Range in the The period 1976–2005 was cho- Climate Group (2010). Pacific northwestern United States sen as the averaging interval for the Climatologically aided interpola- and the Sierra Nevada in California new PHZM because it represented tion [CAI (Daly, 2006; Willmott and limit the eastward penetration of mild the most recent 30-year period for Robeson, 1995)] was used to inter- Pacific air, creating sharp zonal con- which there were reasonably complete polate the PHZM. Here, instead of trasts along their crests. The Rocky data at the time the project began. A using elevation as the predictor grid Mountains act as a barrier to out- 30-year period was chosen instead of (independent variable in the PRISM breaks of frigid air from northern a shorter period because it is more local regression function), the CAI Canada, resulting in milder zones west stable statistically, samples recent cli- method involves using a previously of the Rockies than to the east. These matological variation more completely, interpolated grid that represents the infrequent arctic outbreaks when they and depicts with higher fidelity the long-term mean of a related climato- do enter the western United States role that past winters have played on logical variable as the predictor. In and Canada often have serious horti- the survival of long-lived plants. In this case, the best predictor for the cultural impacts (see Quamme et al., addition, the TRT conducted exten- CAI of the PH statistic was found to 2009). Topographic effects on the sive reviews of draft maps and be the 1971–2000 mean monthly spatial distribution of the PH statistic
• February 2012 22(1) 11 FEATURE are complex, due to competing effects to its location in the relatively warm hardiness zones colder or warmer than of cooling with elevation and cold-air Caribbean Sea and southern latitude, the mapped zone for a given location, pooling in valleys. The coldest zones has the warmest PH values of all based on standard deviations. Table 4 in the western United States are lo- the regions mapped. An extensive presents data that relate standard de- cated not at the highest elevations, coastal strip of zone 13a delineates viations from Fig. 3 to the frequen- but rather in interior valleys where low-elevation areas moderated by cies of such events, including those persistent cold-air pooling occurs (see nearby ocean temperatures. In the that are one half zone, one full zone, Figs. 4 and 5 in Daly et al., 2008). interior, temperatures cool with ele- and two full zones beyond the map- Not surprisingly, the coldest vation. Interestingly, the coolest zone ped zone. zones in the United States occur in on the island (11b) occurs in an ele- A5 C(9 F) standard deviation Alaska (Fig. 2B). However, zones with vated, interior valley, where cold air translates into about a 13% chance winters as warm as those of southern pools frequently on winter nights, (1 year in 7.6 years) that, in any given Alabama occur along the Gulf of Alaska and in adjacent highlands. year, the PH statistic could be at least coastline and Aleutians. In southern Clearly, plant hardiness condi- two half zones colder than the mean, Alaska, the Chugach Mountains pres- tions vary over time, both within years and another 13% chance that it could ent a formidable barrier between dom- and between them, and the magni- be at least two half zones warmer inant coastal and interior air masses, tude of this temporal variability is not (Table 4). In a few locations in the creating gradients of up to ten half constant across the United States. A intermountain west (Idaho, Montana, zones between the coastal strip and map of the standard deviation of the Oregon, Utah, and Washington), adjacent regions less than 100 km 1976–2005 PH statistic in the conter- standard deviations exceeded 7 C inland. minous United States (Fig. 3) shows (12.6 F). These areas experienced The zones with the warmest PH that variability is greatest [>5 C(9 F)] such large annual fluctuations in the values are found in Hawaii and Puerto in the intermountain region of the PH statistic that winter conditions for Rico (Fig. 2C and D). Again, eleva- western United States and also in the long-lived plants were likely much tion and coastal influence are the southeast midwestern United States harsher than in other, less variable re- dominant factors controlling spatial (Ohio River Valley). Generally, under gions sharing that zone. At such high distribution of zones. Hawaii encom- the assumption that annual extreme standard deviations, one might expect passes a wide range of zones, from 9a minimum-temperature values are nor- a winter that is two hardiness zones on the summits of Mauna Loa and mally distributed over time, it is pos- colder than the mapped zone to occur Mauna Kea to zone 13a in the mildest sible to estimate the frequency of about every 18 years (Table 4). In coastal locations. Puerto Rico, owing extreme events that are one or more contrast, the standard deviation is
Fig. 3. PRISM 1976–2005 map of standard deviation of the plant hardiness statistic for the conterminous United States. Zone shading key is given at the lower right (1 C = 1.8 F, 1 km = 0.6214 mile).
12 • February 2012 22(1) Table 4. Frequency of years with annual extreme minimum temperatures that are colder or warmer than the mapped plant hardiness zone for a given location, based on standard deviations (see Fig. 3) and an assumption of normally distributed temperature values over time. Frequency of years that Frequency of years that Frequency of years that are one half hardiness zone are one hardiness zone are two hardiness zones [2.8 C(5 F)] colder or warmer [5.6 C (10 F)] colder or [11.1 C (20 F)] colder or Standard deviation than the mapped zone warmer than the mapped zone warmer than the mapped zone 1 C (1.8 F) 1 year in 385 years <1 year in 5000 years <1 year in 5000 years 2 C (3.6 F) 1 year in 12.4 years 1 year in 385 years <1 year in 5000 years 3 C (5.4 F) 1 year in 5.7 years 1 year in 32.3 years 1 year in 5000 years 4 C (7.2 F) 1 year in 4.1 years 1 year in 12.4 years 1 year in 385 years 5 C(9 F) 1 year in 3.5 years 1 year in 7.6 years 1 year in 80 years 6 C (10.8 F) 1 year in 3.1 years 1 year in 5.7 years 1 year in 32.3 years 7 C (12.6 F) 1 year in 2.9 years 1 year in 4.7 years 1 year in 18.2 years lowest [£ 2 C (3.6 F)] in the Central As an example, the Walla Walla assessment ability of the updated Valley of California and the desert region in southeastern Washington PHZM is further enhanced when used southwestern United States; there, it and northeastern Oregon has been in conjunction with maps of the 30- is unlikely that minimum-temperature considered a premium wine region year standard deviation in the PH conditions in any given year varied since the 1860s (Irvine and Clore, statistic (Fig. 3; Daly et al., 2012) significantly from the mapped zone. 1997). However, vineyard plantings and of elevation (US Geological Sur- In such areas, more than 300 years there have been limited by compari- vey, 2006). Temperatures < –25 C might transpire between winters that son with other areas in the Pacific (–13 F) kill most European wine are one full hardiness zone colder than northwestern United States, with grape (Vitis vinifera) cultivars even that mapped (Table 4). <5% (about 650 ha) of Washington’s at their maximal level of midwinter wine grape acreage located there cold hardiness (Davenport et al., New features of the PHZM that (Washington Wine Commission, 2008; Mills et al., 2006), which, at enable special applications 2010). A major reason for this re- first glance, would permit successful In the following section, we straint is the historical risk of recur- grape cultivation down to about zone present three new features of the ring cold injury and uncertainty 6a. However, vineyards are typically updated PHZM and describe some regarding potential vineyard sites not viable in regions where such practical applications related primarily that minimize such risk. Nonetheless, killing freezes occur more frequently to horticultural production, market- the 1990 PHZM mapped much of than every 8 to 10 years. The standard ing, and research. the Walla Walla Valley in a warmer deviation map, in conjunction with FINE-RESOLUTION MAPPING AS hardiness zone (7a) than were the Table 4, can help determine the likeli- A TOOL FOR SITE EVALUATION. Cold Columbia and Yakima Valleys (6b) hood of colder than average lowest injury is not a limiting factor in farther to the west (Fig. 5A), which temperatures occurring over longer grape (Vitis spp.) production in most are established grape-growing regions time frames. These considerations California growing regions, but it is and have traditionally experienced less suggest that winter survival is much critical in many other important grape severe winter damage. The updated more likely in zones 7a or warmer production areas in the United States, map corrects that probable misclassi- zones, with cold injury very unlikely such as the Pacific northwestern re- fication (Fig. 5C) and also permits in zones 8a and above. But even within gion, especially east of the Cascade tentative identification of broad mes- zones 7a and 7b, standard deviations Range. Compared with the 1990 oclimates with lower risk than are can vary considerably (e.g., Fig. 5); PHZM, the updated map’s fine reso- typical within the region. The new lower values are associated with lower lution at 800 m for the conterminous map also identifies a region in zone 7b risk (Table 4). Returning to the Walla United States combined with zoom- around Lewiston in western Idaho, Walla example, with the 6 to 7 C ing ability permits a much more ac- northeast of Walla Walla (Fig. 5C), (10.8 to 12.6 F) standard deviation curate assessment of the risk of winter that was absent from the old map that prevails in the valley (zone 7a), injury across existing horticultural pro- (Fig. 5A). Although this area was an- about one in five years can be ex- duction areas (Fig. 4). In addition, this other successful early grape produc- pected to reach minimum tempera- feature can be used as a preliminary tion site (Irvine and Clore, 1997), tures that are characteristic for zone decision-aid tool in site selection, for industry revival following its demise 6a (Table 4), and hence just marginal example, in new orchard and vineyard due to prohibition has only begun dur- to avoid a killing freeze. By overlaying developments. This could be especially ing the last 5 years. According to the the PHZM with the standard devia- useful where industry expansion in- newPHZM,lowwintertemperatures tion and elevation maps, one can iden- volves planting in relatively new areas, here would be expected to be less lim- tify several mesoclimates in the foothills such as is currently occurring in the iting than in many of Washington’s of the Blue Mountains in zone 7a with Pacific northwestern United States, established wine-grape-growing regions. about 1 C (1.8 F) smaller standard where the booming wine industry Most growers expect their newly deviations than those which charac- requires a steady and reliable increase established vineyard to remain pro- terize sites further down in the valley in grape supply. ductive for 30 years or more. The risk- to the west and northwest (Fig. 5E).
• February 2012 22(1) 13 FEATURE
Fig. 4. Comparison of the 1990 Plant Hardiness Zone Map [PHZM (A)] and the updated version (B) for Oregon and Washington. Increased detail in the updated version was achieved primarily by applying interpolation methods that reproduce topographically based climatic patterns. The original color scheme of the 1990 map has been changed to match that of the updated version to allow a more direct visual comparison of the two maps (1 km = 0.6214 mile).
Fig. 5. Comparison of the 1990 Plant Hardiness Zone Map [PHZM (A and B)] the updated version (C and D), and the standard deviation (E and F) for the tri-state corner of southeastern Washington, northeastern Oregon, and western Idaho (left side), and the Lake Chelan region of central Washington (right side). The original color scheme of the 1990 map has been changed to match that of the updated version to allow a more direct visual comparison of the two maps (1 km = 0.6214 mile, 1 C = 1.8 F).
14 • February 2012 22(1) Because this decreases the frequency 1970s, as part of the NC7 Regional Advances in mapping and infor- of zone-6a temperatures to 1 year in 6 Ornamental Plant Trials (Widrlechner, mation technologies during the last to 8 years (Table 4), these sites can be 2004). At the time this study was two decades have facilitated the use of considered less risky for grape published, no high-resolution data- the preceding evaluation data. By us- production. sets for either PH or Im were publicly ing a geographic information system The Lake Chelan region of cen- available, so the equation needed (GIS), we derived a high-resolution tral Washington is another example to be calculated from historical grid of Im (Fig. 6A) from grids of where the 1990 PHZM apparently weather data, which were limited to mean annual precipitation (Daly et al., failed to account for local variation by specific weather stations. With the 2008) and potential evapotranspi- placing the region on the border be- publication of a moisture-balance ration (Thornthwaite, 1948). The tween zones 6a and 5b (Fig. 5B). Al- map for the north-central United PH-statistic grid (Fig. 6B) was then though these zones are too risky for States (Widrlechner, 1999), informa- combined with the Im grid, weighted the long-term survival of European tion from the 1990 PHZM could by the terms of the regression equation, wine grapes, vineyard development then be combined with that from to generate a fine-scale map that pre- has recently been increasing near the moisture-balance map to generate dicts the survival of woody plants from Chelan (Washington Wine Commis- crude predictions of plant survival the former nation of Yugoslavia across sion, 2010). In support of this develop- based on the regression equation. the state of South Dakota (Fig. 6C). ment, the updated PHZM identifies mesoclimates near Lake Chelan rang- ing from zone 6b to 7a with the warmer areas around the town of Chelan (Fig. 5D) and with standard deviations equivalent to or lower than those of the Yakima Valley to the south and considerably lower than those found in the Walla Walla Valley (Fig. 5F). For example, the 4 to 5 C (7.2 to 9 F) standard deviation near Chelan (zone 7a) translates into 1 year in 8 to 12 years likely to experience temper- atures typical of zone 6a (Table 4), which would pose only a minimal risk from killing freezes for most grape cultivars. USE OF THE PH-STATISTIC DATA LAYER IN CONJUNCTION WITH OTHER CLIMATIC DATA FOR PLANT-ADAPTATION ANALYSES. In addition to the new PHZM produced from the high- resolution PH-statistic grid, the grid itself is also a valuable tool for horti- culture. It can be combined with other high-resolution geographic datasets for a broad range of geospatial analyses. For example, Widrlechner et al. (1992) published a regression equation for predicting the survival of Yugoslavian woody plants in the north-central United States in terms of the PH statis- tic and moisture balance, Im. The re- gression equation is SO = 0.0309 PH + 0.0048 Im + 1.2671, where SO is the overall proportion of plants surviving after 10 years, and Im is calculated following Mather and Yoshioka (1968): Im = [(annual mean precipitation/potential evapotranspira- tion) – 1]. This regression equation emerged from analyses of the results of a 10-year, Fig. 6. South Dakota maps of the spatial distribution of (A) moisture index (Im); (B) multisite evaluation of a broad range of updated Plant Hardiness Zone Map (PHZM); and (C) 10-year survival rate (S0)of trees and shrubs introduced from the woody plants introduced from Yugoslavia to the north-central United States, derived former nation of Yugoslavia during the from the regression equation of Widrlechner et al. (1992) (1 km = 0.6214 mile).
• February 2012 22(1) 15 FEATURE
We selected South Dakota as an These are an abbreviated growing connection between plant hardiness example to illustrate such a map for season, with freezes possible in any zones and zip codes. State and local two reasons: first, because there were month, and rapid, strong fluctuations natural resource conservation agencies long-term trial sites that reported data in winter temperatures (National Oce- also consult plant hardiness zones to from four of its neighboring states, but anic and Atmospheric Administration, craft guidelines for recommended none in South Dakota. Thus, this 1980), which, when they occur on trees and landscape plants for specific information would fill a gap, enabling both sides of the freezing point, can jurisdictions. Their recommendations the selection of new trial sites for woody- initiate drastic dehardening and cause can be more quickly and accurately plant evaluation in South Dakota. The winter injury. These fluctuations, pro- developed by entering a location’s zip second reason was because South duced by changes in cold-air drainage code, confirming its zone, and relat- Dakota has an unusual topographic and dynamic chinook winds, are among ing the zones to published informa- and vegetation feature (the Black the most extreme in North America. tion about plant adaptation. Hills) unlike any of the other states Burt (2004) provided details of the 2) Requests from nurseries and in the north-central region. most extreme example, which was doc- other firms that widely distribute Three key aspects of woody- umentedinJan.1943inSpearfishand plants. Many nurseries that fulfill re- plant adaptation can be gleaned from Rapid City. One or more long-term tail and wholesale orders for plants Fig. 6C. First, a general southeast to evaluation sites should be established over large geographic areas, as well as northwest gradient in predicted per- within the Black Hills to incorporate mass retailers with national distribu- centages of plant survival can be ob- this unusual climatic region into the tion, such as Home Depot, have also served, from about 40% survival in the NC7 Regional Ornamental Plant Tri- requested a zip code finder. It allows extreme southeast corner of South als and begin testing hypotheses about them to refine the timing of plant Dakota to less than 10% in the north- the climatic determinants of woody- shipments by location based on delivery west. That gradient conforms well to plant adaptation. zip code, so plants are delivered at an expectations based on data collected THE ZIP CODE FINDER. The optimal time for a nursery’s customers, in surrounding states, where SO values updated PHZM enables users to or for direct sale, in the case of mass varied from 3% in Burleigh County, search for hardiness zones of specific retailers. Mass retailers can also apply ND, to 48% in Carver County, MN, sites by five-digit zip code. This fea- a zip code finder to develop targeted (Widrlechner et al., 1992). Second, the ture has been developed in two forms, lists of plants for sale that are more more local effect of cold-air drainage a freely available single-query tool on likely to survive local winter condi- and pooling, common in certain river the website and a large-scale, batch tions near specific retail outlets. valleys, is readily apparent. In South process tool for commercial appli- 3) Requests from the broader Dakota, cold-air drainage influences cations. Such tools to facilitate zip commercial sector. Beyond the ‘‘green plant survival most notably between code-based website searches were first industry,’’ other firms also market pro- Pierre and Rapid City, along the Bad developed in the 1990s. Buyukkokten ducts based on seasonal temperature and Cheyenne Rivers, where survival et al. (1999) were among the first to considerations. They have expressed was estimated to be <20% (Fig. 6B). describe the concept of geographic interest in zip code-based tools to Third, the Black Hills, a large ‘‘island’’ extraction utilities, including a zip develop more precise criteria for de- of mountainous, forested vegetation code finder, and to propose ways that ciding when to begin marketing or has a substantial influence on woody- these tools could improve data min- otherwise increase activity by location. plant survival. Its native vegetation is ing and increase the overall value of For example, a national termite-control classified by Ku¨chler (1964) as Black websites. firm has requested a way to integrate Hills Pine Forest, which is surrounded The PHZM zip code tools gen- zip codes and plant hardiness zones to by grasslands (Campbell, 1997). SO erate a hardiness-zone value based on predict when termites and other insect survival values > 30% in western South the geographic center for each five- pests may become active. Dakota (Fig. 6C) conform closely to digit zip code region in the United the boundaries of Ku¨chler’s (1964) States. They were developed in re- Concluding remarks Black Hills Pine Forest. Notably, the sponse to the many requests received The development and public model’s highest predicted survival by USDA-ARS for a database that release of the updated PHZM rates (about 60%) also occur in the links zip codes to plant hardiness will advance our understanding of Black Hills, near Spearfish. zones. The bulk of these requests fall landscape-plant adaptation and plant Although extensive woody veg- into three categories: distribution through its presentation etation in the Black Hills helps vali- 1) Requests from governmen- of accurate minimum-temperature date the model’s results, we do have tal agencies. For example, the USDA data at a fine scale via a user-friendly some reservations about the model’s Risk Management Agency (RMA) re- interface readily accessible to researcher accuracy in this particular region. The lates plant hardiness zones to specific and gardener alike. We expect that it model might locally overestimate sur- locations when it sets standards for will be rapidly adopted as a research vival because the Black Hills have two crop insurance for horticultural and dataset in GIS and spatial analyses ‘‘special’’ climatic characteristics det- nursery crops based on the use of crop and will contribute to more effec- rimental to woody-plant adaptation, protection, such as greenhouses or tive planting recommendations, ship- which are unlike the remainder of the rowcovers, and in the development ping management, and horticultural north-central United States, where the of the ‘‘Eligible Plant List and Plant production. model was developed (and unlike much Price Schedules’’ (USDA, 2010). RMA A challenge for developing more of the former nation of Yugoslavia). requested a searchable, automated refined hardiness zonation is how best
16 • February 2012 22(1) to apply GIS tools and sophisticated Cathey, H.M. 1997. Announcing the and T. Meyers. 2008. The 2007 eastern interpolation techniques to the crea- AHS plant heat-zone map. Amer. Gar- US spring freeze: Increased cold damage in tion of zone maps that incorporate dener 76(5):30–37. a warming world? Bioscience 58:253–262. and weight appropriately all the key Cathey, H.M. and J. Heriteau. 1990. Hayashi, Y. 1990. Jumoku ato bukku. climatic factors influencing the adap- Mapping it out. Amer. Nurseryman Abokkusha, Kamakura-shi, Japan. tation of a wide spectrum of perennial 171(5):55–59, 61–63. plants. This is not a trivial challenge Heinze, W. and D. Schreiber. 1984. Eine because these weightings will likely Cornell University. 2006. Flower grow- neue Kartierung der Winterha¨rtezonen ing guides. 22 Apr. 2010.
• February 2012 22(1) 17 FEATURE
McNamara, S., H. Pellett, M. Florkowska, rhododendrons aux diverses conditions CA. 10 Feb. 2010.
18 • February 2012 22(1) Widrlechner, M.P. 1994. Environmental north central United States. Landscape plants in Iowa. J. Environ. Hort. 20: analogs in the search for stress-tolerant Plant News 10(2):10–14. 47–56. landscape plants. J. Arboriculture 20: 114–119. Widrlechner, M.P. 2004. Insights into Widrlechner, M.P., R.E. Schutzki, V.Y. woody plant adaptation and practical ap- Yukhnovsky, and V.V. Sviatetsky. 2001. Widrlechner, M.P. 1997a. Hardiness zones in plications. METRIA Proc. Vol. 13. 9 Aug. Collecting landscape trees and shrubs in China. (Color map - scale ca. 1:16,360,000.) 2010.
• February 2012 22(1) 19