OctOber–December 2006 • VOLUME 60 NUMBER 4 California Agriculture

Lead toxicity: Assessing the risk to U.S. children

University of California | Agriculture and Natural Resources | Research in Agricultural, Natural and Human Resources Editorial

Of Mendel, wikis and open source: New models for knowledge creation

1866, Gregor Mendel pub- collaboration are no longer limited by time or distance, nor lished his paper “Experiments are online reviews, online real-time editing or instant publish- withIN Plant Hybridization” and set a ing. These developments promise to vastly reduce publication foundation for modern genetics and costs. The open-source software development community the laws of inheritance. As with many demonstrates that virtual teams can accomplish highly com- keystone achievements, the importance plex tasks and distribute valuable products. Bob Sams Director, of his paper was not recognized at the These technical and behavioral changes are disruptive. ANR Communication time. Thirty-five years later, long after In fact, Dan Greenstein, university librarian and head of the Services Mendel’s death, rediscovery of his work California Digital Library, has figuratively called them “sub- generated rapid advances in our under- versive.” New forms of popular and academic publishing standing of genetics and marked a starting point for the re- will affect scholarly communications much as Web distribu- search that has improved plant and animal breeding, shown tion has revolutionized the music industry and newspaper us the structure of DNA, and helped us elucidate the risks publishing. New kinds of copyright licenses and new defini- and benefits of biotechnology. tions of intellectual property rights will be required. One sig- However, that is only part of the benefit from Mendel’s nificant effort to define these agreements is the Open Content work and the 140 years of review, replication and analysis Alliance. Web archives and aggregators such as UC’s own that followed. It is not as well known that Mendel’s ex- eScholarship Repository and the national Web project eXten- periments were also studied carefully by Richard Fisher, sion are working to develop open models of content creation, English statistician and evolutionary biologist. Fisher attribution, licensing and ownership. concluded that Mendel’s findings were just too close to For California Agriculture and all ANR publications, these expected results — and while the accuracy of Mendel’s hy- changes are important, difficult and exciting. Communications pothesis is not challenged today, his work is cited as an ex- and information technology professionals see many tantaliz- ample of how smoothing data can lead to confirmation bias. ing, confusing and unknown pathways to the open and broad Fisher’s later career included work at the famed Rothamstead dissemination of peer-reviewed research results. We also see Experimental Station in England and as visiting faculty at a bewildering array of tools to deliver the benefits of new Iowa State University, where he contributed greatly to modern knowledge to society. statistical science. Mendel’s work and Fisher’s analysis are To take full advantage, we must do more than ensure that examples of the tradition of open inquiry and scientific dia- content exists on the Web and that it is appropriately indexed logue that are the reason for publishing this magazine and for and recognized by search engines. We must also engineer reporting research results openly and accessibly. online peer-review processes that are flexible enough to ac- As California Agriculture marks its 60th anniversary commodate the whole range of information and publication (see page 174), the journal is working hard not only to pub- methods, from refereed journals to one-page fact sheets. lish new findings but also to realign production, publication We must explore new forms of community or “salon” and distribution methods with rapid technological advances review that allow editorial or con- in communications. First among these changes is the con- tent changes to Web information tinuing growth of the Internet. The World Wide Web has be- in real time. We must also adapt Links to explore come the dominant information retrieval pathway, especially to changing notions of intellectual eScholarship Repository in science, education and business. Two things have fueled property and copyright. In a letter http://repositories.cdlib.org/ this evolution. Digitized content has grown explosively to the editor (page 173), Lawrence escholarship/ and there is no end in sight. In a recently announced agree- Pitts, chair of the UC Academic ment, UC and Google will undertake to digitize and index Council Special Committee on eXtension http://about.extension.org selected contents of the UC library system, as well as those Scholarly Communication, reminds of other major institutions. Combined with similar initiatives us that all of UC shares an obliga- Gregor Mendel by Yahoo and other search engines, this agreement shows tion to the common good, a value http://en.wikipedia.org/wiki/ Gregor_Mendel that society has implicitly and collectively agreed that this that has always been at the core of Herculean task is both possible and desirable. It also dem- ANR’s mission. At the same time, Open Content Alliance onstrates our confidence that the technology behind today’s we must ensure the quality of our www.opencontentalliance.org powerful search engines is capable of storing, indexing and information, protect its identity and Open-source software retrieving that information. source, and deliver it in an effective www.sourceforge.net Secondly, growing online communities based on “social form to the people of California. networking” permit both direct conversations between indi- We are working to see that California Wikipedia viduals as well as specialized Web publishing through blogs, Agriculture continues its 60-year tra- http://en.wikipedia.org/wiki Web forums, wikis and virtual communities. Cooperation and dition of doing just that.

170 CALIFORNIA AGRICULTURE • VOLUME 60, NUMBER 4 TABLETable of ContentsOF CONTENTS G. DeGrazia/CMSP G. News departments 172 Letters More voices: Making the case for open access

174 Celebrating a work in progress: UC journal delivers research to Golden State and beyond

176 Citrus Research Center – Agricultural Experiment Station COVER: National studies have UC Riverside marks a century of agricultural increasingly linked even low levels of innovation — still thriving in an urban empire lead exposure to IQ defi ciencies and behavioral disorders in young children (page 180). In September, a new study 216 30 years ago in California Agriculture connected attention defi cit disorders to lead, among other factors (see www.ehponline.org).

Research articles 180 is lead toxicity still 200 impact of environmental a risk to U.S. children? factors on fi sh distribution Heneman, Zidenberg-Cherr assessed in rangeland streams Elevated lead levels in paint, dust, Thompson et al. soil, imported pottery and candies, In a Shasta County rangeland water- 185 and ethnic remedies are still of con- shed, fi sh distribution varied across the cern; ongoing surveillance is needed. growing season and was correlated with stream temperature and pool depth. 185 Cost-benefi t analysis conducted for nutrition 207 EU support reductions education in California would benefi t California Block Joy, Pradhan, Goldman tomato growers and processors A conservative analysis using Califor- Rickard, Sumner nia EFNEP data found that, at mini- Lowering trade barriers 50% would raise mum, $1.00 spent on nutrition edu- prices for U.S. processing tomatoes 6%, 200 cation saves $3.67 in medical costs. and increase income for growers and pro- cessors by $53.5 million per year. 192 Radiofrequency power disinfects and disinfests 211 Weather-based yield food, soils and wastewater forecasts developed Lagunas-Solar et al. for 12 California crops A simple telecommunications tech- Lobell, Cahill, Field nology quickly heats various foods Yield predictions for some crops were and materials, killing pathogens and highly accurate and could be made with 207 arthropod pests. longer lead times than current methods.

Correction: In the July-September 2006 issue, the credits for two images on page 116 were incorrect. The California wild grape image was taken by wildscaping.com, and the wild radish was taken by Jeff Abbas. California Agriculture regrets the error.

http://CaliforniaAgriculture.ucop.edu • OctOber–December 2006 171 Letters

WHAt DO YOU tHiNK? More voices: Making the case for open access The editorial staff of California Agriculture welcomes your letters, Agricultural publications online comments and sugges- tions. Please write to us hould peer-reviewed research developed The January-March 2006 California Agriculture edito- at [email protected] or Sat tax-supported institutions such as UC be rial underlined the importance of publishing peer- 1111 Franklin St., freely available to the public? A recent California reviewed agricultural research, the advantages of 6th fl oor, Oakland, CA Agriculture editorial (January-March 2006; 94607. Include your full open-access outlets (including our experiment sta- “California Agriculture delivers access to peer- tion and extension service publications) and the ad- name and address. reviewed research”) explored this issue, noting Letters may be edited vantages of making information available through the impact of soaring journal prices on both li- for space and clarity. braries and individuals. our institutional repositories. Some recent developments: At my own institution we are currently looking at our languishing experiment station series and • In May, U.S. senators introduced legislation discussing what their future should be. As I have requiring agencies with annual research budgets of more than $100 million (e.g., EPA, USDA) to looked at other experiment stations, it seems we are provide public access to research results no more not alone in seeing publication in these once-prolifi c than 6 months after publication. (The National series fall off. Institutes of Health already encourages posting Wouldn’t it be a wonderful contribution to the within 12 months on PubMed.) National Digital Library for Agriculture to have ac- • In July, Rice University announced plans to be- cess to not only the historical record of agricultural come the fi rst all-digital university press; browsers research, but also the current research, readily avail- will be able to read the digital books for free but able to anyone, anytime, anyplace? will pay to download. Constance J. Britton, Librarian • This summer, the journal Nature offered au- Ohio State University, Wooster thors the opportunity to make their submissions available online immediately as preprints and Editor’s note: The National Agricultural Library receive comments from any reader. • In August, UC libraries joined Google Book (NAL) is laying the groundwork to provide comprehen- Search, the massive book scanning and digitiza- sive digital access to agricultural information from part- tion project, marking the largest expansion of nering institutions and universities. that project to date (UC potentially offers the Digitizing research, improving access contents of 100 libraries on 10 campuses). • By September, UC’s eScholarship Repository, The January-March 2006 editorial addressed a key the open-access database of the California Digital issue. This is the fact that researchers and members Library, had posted 3.7 million full-text down- of the public often need access to the results of peer- loads since opening in April 2002. (California reviewed research, but it is no longer easily acces- Agriculture posted close to 1,900 full-text down- sible or affordable. loads between May and September, after initial posting of peer-reviewed articles at the site.) Many agricultural librarians have been con- The letters that follow explore these issues. fronting this for quite a while. Questions come up routinely, such as a graduate student trying to ac- — Editor cess information on Pierce’s disease fi ndings, or a scientist looking for archival data on soil character- istics. At the extension level, someone may want to know how close to plant a fruit tree to a fence built of pressure-treated lumber, for fear of contamina- tion from toxic chromated copper arsenate. At a recent brainstorming session sponsored by USAIN (United States Agricultural Information Network) and NAL, concerns raised included the need for digital archiving and repositories, the unrestricted availability of publications to a wide audience, and an easy-to-use search interface like Amazon or Google. The development of a digital repository by the USDA modeled on PubMed, as well as an expansion of open-access journals like PLoS (Public Like others systemwide, UC irvine students rely Library of Science), are all efforts that need to be on UC library purchases of essential journals and supported and encouraged. online subscriptions. Finding “grey” or “ephemeral literature” has

172 CALIFORNIA AGRICULTURE • VOLUME 60, NUMBER 4 Links to explore always been problematic in many disciplines. Such tems, new measures of impact that include the read- items are not commercially published, widely dis- ership as well as citations of an article, and increased Academic Senate Special tributed or indexed, and not always archived. Many access to original and supplementary data are just a Committee on Scholarly state agricultural documents, especially at the exten- few of the new tools available for ensuring scholarly Communication sion level, fall into this category, and there are now quality and increasing the usefulness of research. www.universityofcalifornia.edu/ senate/committees/scsc/ plans to digitize such materials. Libraries and land- The SCSC developed a series of white papers reports.html grant institutions need to explore further collabora- for consideration by the UC Academic Council tion on publishing and disseminating both extension and Assembly, which endorsed them in May Google Book Search and the more scholarly experiment station research, 2006. Among other things, these papers recom- http://books.google.com given the large amount of material that is no longer mend that: (1) promotion and tenure processes widely distributed outside of the local area. include appropriate alternative forms of dissemi- National Agricultural Library Norma Kobzina, Head, Information Services, nation; (2) researchers manage their copyright to http://naldr.nal.usda.gov Biosciences and Natural Resources Library, UC Berkeley provide greatest access and impact; (3) scholarly societies limit copyright transfer from author to UC faculty and the right to copyright Oxford Journals and society; and (4) journal publishers accept the right NAR Open Access initiative I share the conviction that research and scholarship to fi rst publication and routinely permit posting of www.oxfordjournals.org/our_ should be broadly accessible and free to the public. papers on open-access databases, in some form. journals/nar/announce_ openaccess.html The economic dysfunctions cited in the editorial In May, UC’s Academic Assembly endorsed a (soaring prices of electronic and print journals, due proposal that faculty give UC permission to post PLoS (Public Library of Science) in part to publisher consolidations) were brought their journal articles or conference papers in an and Creative Commons into sharp relief for many UC faculty and admin- open-access repository; fi nal adoption is expected Attribution License istrators during the University’s 2003 negotiations by spring 2007. The SCSC has also proposed www.plos.org/about/ with for-profi t journal publisher Elsevier, and led open-access archiving of stem-cell research results index.html to the establishment of the UC Special Committee funded by California’s Proposition 71. I am pleased on Scholarly Communication (SCSC). But while the that health sciences have encouraged such policies. PubMed (biomedical literature) dysfunctions helped to get our attention, it is the The SCSC does not promote a particular pub- www.ncbi.nlm.nih.gov/Literature opportunities offered by Internet publishing that lishing business model or distribution of respon- have led the faculty at UC and other research insti- SPARC (Scholarly Publishing and sibilities among stakeholders. This is a time for Academic Resources Coalition) tutions to become deeply engaged in these issues investments that encourage experimentation, such www.arl.org/sparc and propose solutions. as UC’s eScholarship Repository, and for faculty to The Internet reduces to near zero the cost of an look outside of their own disciplines to success in additional reader when the content is online instead other domains. of in print form. Internet publication can also be Lawrence Pitts, Professor of Neurosurgery, used to certify, validate, and ensure accuracy and UC San Francisco quality of research results. Online peer-review sys- Chair, Special Committee on Scholarly Communication

California Agriculture (ISSN 0008-0845) is published quarterly and mailed at periodicals California Agriculture postage rates at Oakland, CA, and additional mailing offi ces. Postmaster: Send change of address "Form 3579" to California Agriculture at the above address. Peer-reviewed research and news published by the Division of RAtES: Subscriptions free upon request in U.S.; $24/year outside the U.S. After pub- Agriculture and Natural Resources, University of California lication, the single copy price is $5.00. Orders must include payment by check or inter- national money order in U.S. funds, payable to UC Regents. MasterCard/Visa accepted; VOLUME 60, NUMBER 4 requests require signature and card expiration date. Please include complete address. Executive editor: Janet White For online ordering, go to http://CaliforniaAgriculture.ucop.edu. Managing editor: Janet Byron Art director: Davis Krauter Articles published herein may be reprinted, provided no advertisement for a com- mercial product is implied or imprinted. Please credit California Agriculture, University of California, citing volume and number, or complete date of issue, followed by inclusive Associate Editors page numbers. Indicate ©[[year]] The Regents of the University of California. Photo- graphs may not be reprinted without permission. Animal, Avian, Aquaculture & veterinary Sciences: Edward R. Atwill, UC prohibits discrimination against or harassment of any person on the basis of race, Christopher M. Dewees, Kathryn Radke, Barbara A. Reed; Economics & color, national origin, religion, sex, gender identity, pregnancy (including childbirth Public Policy: James Chalfant, Karen Klonsky, Henry J. Vaux Jr.; Food & and medical conditions related to pregnancy and childbirth), physical or mental dis- Nutrition: Amy Block Joy, Sheri Zidenberg-Cherr; Human & Community ability, medical condition (cancer-related or genetic characteristics), ancestry, marital Development: Ellen Rilla, Alvin Sokolow; Land, Air & Water Sciences: status, age, sexual orientation, citizenship, or status as a covered veteran (special dis- David Goldhamer, Mark E. Grismer, Ken Tate, Bryan Weare; Natural abled veteran, recently separated veteran, Vietnam-era veteran or any other veteran Resources: Adina Merenlender, Kevin O’Hara, Terry Salmon; Pest Manage- who served on active duty during a war or in a campaign or expedition for which a ment: Janet C. Broome, Kent Daane, Deborah A. Golino, Tim Paine; Plant campaign badge has been authorized) in any of its programs or activities. University Sciences: Kent Bradford, Kevin R. Day, Steven A. Fennimore, Carol Lovatt policy is intended to be consistent with the provisions of applicable state and federal laws. Inquiries regarding the University's nondiscrimination policies may be directed to California Agriculture the Affi rmative Action/Staff Personnel Services Director, University of California, Agri- 1111 Franklin St., 6th fl oor, Oakland, CA 94607-5200 culture and Natural Resources, 1111 Franklin St., 6th Floor, Oakland, CA 94607-5200 or call (510) 987-0096. Phone: (510) 987-0044; Fax: (510) 465-2659; [email protected] http://CaliforniaAgriculture.ucop.edu ©2006 The Regents of the University of California

http://CaliforniaAgriculture.ucop.edu • OctOber–December 2006 173 California Agriculture

Celebrating a work in progress: UC journal delivers research to Golden State and beyond 60In the last 6 decades California Agriculture’s format, length and design changed many times. As California grew and diversifi ed, so did the journal. Today, published fi ndings emanate from a broad band of disciplines, covering topics such as demographics, public policy and climate change. However, our goals of publishing rigor- ously reviewed research, and translating that science for a diverse — and interna- tional — audience, are the same. California Agriculture is one of the nation’s oldest, con- tinuously published, land-grant university journals, and our presence extends beyond the printed page to our Web site, California December 1946 Digital Library’s eScholarship Repository and numerous scholarly databases. Below, a brief timeline of events: January 1969: The fi rst editorial appears: “Re- December 1946: Publication of the fi rst issue search is People” by C.F. Kelly, then director, of California Agriculture, a four-page tabloid UC Agricultural Experiment Station. on newsprint. September 1970: The fi rst color cover appears, April 1957 1950s–1960s: In the early years, scientists featuring rusty ribs of head lettuce. Color sec- report a growing diversity of fi ndings, tions become regular in 1978, with the maga- from the effectiveness of DDT, to attendant zine adopting full color on all layouts by 1990. residue problems and increasing insect re- 1978: sistance. Other issues include some of the Peer review begins. At fi rst this consists earliest reports on the rural-urban fringe of a single reviewer for each article. problem (March 1959), integrated control 1970s–1980s: Content evolves in breadth and (September 1960) and gamma ray research depth, embracing topics such as farmwork- (March 1962). ers and the wine industry (March-April 1989). April 1957: The fi rst special issue appears, The journal explores single subjects in greater including 24 reports on the water situation. depth, such as forestry (May 1979), mosquito research (March 1980) and genetic engineering January 1960: Research Briefs fi rst pub- (August 1982). lished; these brief notes about UC research California Agriculture adopts desktop pub- March 1962 run occasionally through December 1978. lishing soon after it becomes available in 1985.

September 1970 May 1979 March 1980 August 1982

174 CALIFORNIA AGRICULTURE • VOLUME 60, NUMBER 4 March-April 1989 November-December 1995 May-June 1998 September-October 2000

1981: The journal switches from monthly to cations, including grandparenting (March- bimonthly and now is longer, running 24 to April 2001), biotechnology and global 32 pages. warming (May-June 2002). In the Year 2000 series, scientists look at scenarios for 1990s: More special issues appear, treating the next 25 years in California, in the areas controversial subjects from diverse view- of demographics, natural resources, food points. Special collections cover pesticides production and food security (September- and food safety (May-June 1994), biodiversity October 2000). (November-December 1995), hunger (Decem- Also in 2000, California Agriculture ber 1995), aquaculture and fi sheries (July- begins posting the entire contents of the August 1997) and urban encroachment on journal on the Web site, including full-text farmland (May-June 1998). PDFs of peer-reviewed research articles. January 1992: Peer review is further strength- March-April 2001 2002: California Agriculture initiates contin- ened with the establishment of the Associate uous page numbering, a standard practice Editor panel. The panel includes subject mat- of peer-reviewed science journals. ter experts from within ANR who oversee the process; to be published, manuscripts must be January-March 2003: Due to the statewide accepted by at least two anonymous review- budget defi cit and UC cutbacks, the journal ers. Increasing requirements for long-term shifts to quarterly issues. Length of maga- data and statistical analysis lead to lengthier zine increases to between 48 and 96 pages articles and higher rejection rates (currently (i.e., Kearney at 40, April-June 2005). 25% or more). August 2005: The fi rst market research begins in the Double-blind procedures form of a survey, to which 66% of readers added to anonymous peer review. The jour- respond. nal begins requesting electronic submission of all new manuscripts. 1994: News sections are established to pro- April 2006: California Agriculture imple- vide a real-world context for research ar- May-June 2002 ticles. Front matter is brief and authored by ments a new electronic peer-review system editorial staff. Shortly after, the fi rst Letters using Berkeley Electronic Press software, column appears. under contract with the California Digital Library (CDL). Prospective authors use the December 1994: The journal begins listing Repository’s secure server to make submis- references for further reading, and in May sions, upload revisions and check on the 1996 extensive use of references and citations status of articles. begins. May 2006: California Agriculture begins January 1996: California Agriculture posts its posting all published, peer-reviewed ar- fi rst Web site, including tables of contents, ticles to the CDL’s eScholarship Repository. news stories and research abstracts. PDFs continue to be posted in full to the California Agriculture Web site. 2000 and beyond: Growing emphasis on special collections with public policy impli- —Editor

April-June 2005

http://CaliforniaAgriculture.ucop.edu • OctOber–December 2006 175 Citrus Research Center David Morgan

AES Centennial

100UC Riverside marks a century of agricultural innovation — still thriving in an urban empire

arly in the 20th century, when the citrus belt facilities, such as a state-of-the-art Insectary and cut a wide swath from Pasadena to Red- Quarantine Facility and two campus-operated fi eld Elands, an outpost experiment station oper- stations totaling nearly 1,000 acres. ated by UC helped to ensure, through science, the “This campus has a legacy of research in agri- prosperity of what was even then a $20 million culture and natural resources that has strengthened industry. one of this state’s most important industries and Today, sprawling suburban developments, helped it remain competitive, whether that meant gleaming midrise offi ce towers and crowded shop- surmounting new insect pests and plant diseases or ping malls have replaced all but a few pockets of developing new products for new markets,” says Southern California’s citrus empire. Yet the citrus Steven R. Angle, dean of the UC Riverside College industry thrives in California, and the experimental of Natural and Agriculture Sciences. outpost known as the Citrus Experiment Station The centennial celebration commenced with a (CES) in Riverside has grown and expanded its mis- barbecue and open house in April, featuring tours sion in support of the state’s diverse, nearly $32 bil- of the on-campus fi eld station and ceremonial lion agricultural industry — still tops in the nation. plantings of the UC Riverside–developed manda- This year marks the centennial of UC rins ‘Gold Nugget’ and ‘Yosemite Gold’. It contin- Riverside’s Citrus Research Center-Agricultural ues this October with groundbreaking of the Experiment Station (CRC-AES). What began $53.9 million Genomics Building on campus and, in 1906 with a staff of two at the base of Mt. next February, a banquet and symposium that takes Rubidoux near downtown Riverside is today a prospective look at the next century of agricul- a multifaceted enterprise involving 140 faculty tural sustainability (see box, page 179). and specialists in the experiment station and “The contributions of UC Riverside are wo- Cooperative Extension, and an annual budget of ven throughout the citrus industry, not only in approximately $22 million. The CRC-AES now California, but nationally and internationally,” tackles a broad array of agricultural, urban and says Ted Batkin, president of the Citrus Research natural resource problems with fundamental and Board in California and chair of the National Citrus applied research in plant biology, pest and disease Research Council. “What is so impressive is the management, and the environment and natural connection between the Agricultural Experiment resources. They are supported with advanced Station and Cooperative Extension specialists. It is

176 CALIFORNIA AGRICULTURE • VOLUME 60, NUMBER 4 ca. 1917

Far left, the parasitic wasp Gonatocerus triguttatus was Herbert John Webber evaluated at the Citrus Experiment Station (CES) for (seated far right), fi rst biological control of the glassy-winged sharpshooter. Left CES director, in a 1916 and above, in 1918, CES moved from Mt. Rubidoux near staff photograph. downtown to the site that would become UC Riverside. seamless in providing a continuum of basic discov- The expanded station moved in 1918 to the east- eries through fi eld application and it is what makes ern edge of the city, on a site that would become the Riverside campus particularly distinctive.” UC Riverside in 1954, occupying a stately Spanish mission–style headquarters — a building now des- A “valentine” for Riverside ignated as a Riverside County historical landmark. The offi cial establishment of the Citrus (Today it houses UC Riverside’s A. Gary Anderson Experiment Station is pegged to the date that Graduate School of Management.) the UC Board of Regents formally approved Groundbreaking research the leases for the Riverside property — Feb. 14, 1907 — though the station had begun operating Experiment station scientists conducted ground- some months prior. Opening of the station fol- breaking research that led to the delivery of new va- lowed a vigorous lobbying effort by leaders of rieties of citrus and other commodities, innovative the Riverside Horticultural Club and Southern pest management techniques utilizing advances in California Fruit Growers Exchange, a mission that biological control and integrated pest management landed Riverside the citrus station and Whittier a (IPM), and new insights on California’s natural pathological laboratory. resources that would inform both environmental When it came time to upgrade and expand the quality and crop production. station, an intense political battle was waged be- Among early accomplishments was the control Alfred M. Boyce, with tween Riverside and the San Fernando Valley, a of the citrophilus mealybug by a natural enemy, one pipe, was CES director then-emerging citrus belt with powerful lobbying of the station’s fi rst ventures in biological control. from 1952 to 1968. interests related to development and construc- In the 1940s, Herbert John Webber and Leon D. tion of the aqueduct from the Owens Valley to Los Batchelor — the station’s fi rst and second direc- Angeles. Riverside viewed itself as the underdog, tor, respectively — published The Citrus Industry, a so there was wild elation in the streets when, on comprehensive two-volume work compiling scien- Dec. 22, 1914, the UC Regents voted to retain the tifi c knowledge about the distribution, botany, ge- station in Riverside. netics and reproduction of citrus. Tristeza disease, In a day now famous in Riverside history, the which threatened to wipe out California’s citrus in- steam whistle at the electric plant blew for 15 dustry starting in the 1930s, was solved after station minutes and Mission Inn owner Frank Miller or- scientists determined that the disease was caused dered the bells at the hotel be rung continuously. by a virus; a new disease-resistant rootstock was Riverside civic leader and citrus pioneer John Henry Reed called it “the most important day in the history of Riverside” in the Riverside Daily Press.

Civic leader and citrus pioneer John Henry Reed was a long-time advocate for CES.

http://CaliforniaAgriculture.ucop.edu • OctOber–December 2006 177 California’s population growth, leading to environmental problems associated with the urban-wildland and urban-agriculture interfaces, has prompted a renewed focus on natural resource protection.

developed by W.P. (Bill) Bitters, who also served as Also in the 1960s, the fi rst International Citrus director of the Citrus Variety Collection for 35 years. Symposium was held at UC Riverside — organized Shortly after its formation, the station expanded by the late Homer D. Chapman — drawing 800 its mission to include work on many other sub- delegates from more than 50 countries and lead- tropical crops. The Riverside station gained a ing to the formation of the International Society of strong reputation in the avocado industry, due in Citriculture (ISC). ISC, still headquartered at UC large part to the efforts of George Zentmyer — Riverside, has convened congresses around the later elected to the prestigious National Academy world ever since, with the next one scheduled for of Sciences — who would be credited with saving 2008 in China. the industry from avocado decline disease with his the second 50 years work on the soilborne fungus Phytophthora cinna- moni and his introduction of resistant rootstock. When it entered its second half-century, the ex- Following World War II, and under the direction periment station carried out a diverse, broad-based of Alfred M. Boyce, the Citrus Experiment Station research program, but citrus remained an important enjoyed its greatest growth. When UC Riverside focus area and scientists retained the strong sense of was designated a general campus in 1960, a College service to the industry upon which the station was of Agriculture was established and new depart- founded. Says plant physiologist Charlie Coggins, ments in agronomy, agricultural engineering and who arrived at the station in 1957, “Good, practi- biostatistics were initiated to augment the station’s cal research was considered to be what we were existing departments. all about, and strong fundamental research was Experiment station scientists investigated new pes- expected as an underpinning to that applied re- ticides and earned a reputation for pioneering studies search.” Coggins’ work on growth regulators begin- on insect resistance to pesticides and the measurement ning in the 1960s dramatically extended California’s s the UC Riverside College of insecticide residues, methods that would become navel orange growing season. of Natural and Agricultural part of the federal evaluation for commercial pesticide Research at the CRC-AES broadened again in the Sciences developed the Core approval. The Citrus Clonal Protection Program was 1970s under W. Mack Dugger, incorporating such instrumentation Facility to founded, today serving as the only approved im- new scientifi c disciplines and techniques as mo- support research in genome biology. portation facility in the nation for citrus. lecular biology, IPM and genetics. Environmental

s protection and the challenges associated with ag- UC Riverside researchers riculture in arid and semiarid regions became new identifi ed a natural enemy to manage the silverleaf focus areas. New crops were developed, including whitefl y. turfgrass varieties with tolerance to soil salinity and air pollution. The campus also became a leader in plant tissue culture. At the three-quarter-century mark, the CRC- AES was led by deans Irwin Sherman, now pro- fessor emeritus of zoology, then Seymour D. Van Gundy, now professor emeritus of nematology. It was an era during which the still-new discipline of molecular biology was maturing, with IPM and bi- ological control remaining robust areas of research. In one particularly dramatic case, the importation

178 CALIFORNIA AGRICULTURE • VOLUME 60, NUMBER 4 Far left, the late George Zentmyer was a world authority Left, Distinguished Professor on Phytophtora cinnamomi, a fungus affl icting avocados Natasha Raikhel, director of the and other tropical plants; inset, avocado thrips. Left, a Center for Plant Cell Biology, and student researcher at the UC Riverside experiment station. students. Above, turfgrass research. Above, an artist’s rendering of the $53.9 million Genomics Building, currently under construction.

and establishment of a tiny stingless wasp brought California’s population growth, leading to the ash whitefl y under control. This bothersome environmental problems associated with the Centennial pest not only caused millions of dollars in damage urban-wildland and urban-agriculture inter- celebration to agriculture, but it also despoiled cars parked faces, has prompted a renewed focus on natural under landscape trees with sticky excrement. It resource protection. UC Riverside’s Center for The UC Riverside was a case that brought Citrus Experiment Station Conservation Biology generates objective scien- Citrus Research Center and Agricultural research to bear on the urban community, bringing tifi c data to inform the development of public Experiment Station widespread media attention to the UC Riverside policy and restoration plans to preserve endan- will celebrate its 100th campus. gered species. The experiment station is also anniversary with a The last quarter-century has also seen the re- expanding its capabilities in environmental and symposium lease of several patented new varieties of citrus, natural resource economics. and banquet on starting with the ‘Oroblanco’ grapefruit in 1981 On the occasion of the centennial, the College of Feb. 14 and 15, 2007. and continuing with the recent release of the Natural and Agricultural Sciences — with guidance For more information, go to: ‘Tango’ mandarin. Likewise, development of the from the Chancellor’s Agricultural Advisory Council http://cnas.ucr.edu/ GeneChip Citrus Genome Array will help scien- and faculty — is undertaking a strategic planning news/special tists quickly examine genetic traits in citrus so process to chart its course for the next one. “We have _events.htm they can develop new varieties and tackle disease served the agricultural industry well for 100 years,” or contact and postharvest problems. Another breeding pro- says Executive Associate Dean Donald Cooksey. Carol Lerner, gram at UC Riverside has yielded cowpea lines “Now we need to determine how best to face the [email protected], (951) 827-5089. that are early-maturing and heat-tolerant, making next century. Are we well-positioned for the future?” them particularly well-suited to the drought con- In the near term, that future is expected to in- ditions of West Africa, helping to reduce hunger clude a continued emphasis on water science with and poverty there. pending formation of a Center for Water Quality Science and Policy to address problems that are Current research initiatives projected to become even more acute as demand by The College of Natural and Agricultural Sciences urban and agricultural consumers increases. Pest has since undertaken a major initiative in genome management, especially as it relates to invasive spe- biology, begun under the leadership of former cies, is expected to remain among the priorities, as dean Michael T. Clegg and continuing today. UC is natural resources, in part through the college’s Riverside’s Institute for Genome Biology and participation in the campuswide Environmental Center for Plant Cell Biology, created to capitalize Research Institute. on the emerging genomics revolution and on exper- There will be no shortage of challenges and op- tise in the experiment station, is applying genomics portunities for California’s agricultural industry. to solve critical health and agriculture-related prob- “As our population increases rapidly, we will lems, such as malaria and abiotic stresses to plants. lose more of our land and water resources for ag- Associated with the institute is the Biotechnology riculture,” says Van Gundy, chair of the centennial Impacts Center, which serves as an “honest broker” celebration planning committee. “We are going to to identify the relevant policy issues related to the have to focus on more intensive agriculture and potential risks of genomic modifi cation, act as a new technologies to feed the world’s growing clearinghouse for credible information and initiate population and to help agriculture in California research that addresses its potential benefi ts and remain economically viable.” consequences. — Kathy Barton

http://CaliforniaAgriculture.ucop.edu • OctOber–December 2006 179 Review Article t Is lead toxicity still a risk to U.S. children?

by Karrie Heneman and Sheri Zidenberg-Cherr

Elevated blood-lead levels put chil- dren at risk for neurobehavioral- cognitive deficits, such as IQ de- ficiency, behavioral disorders and impaired hearing. We examined several factors that contribute to elevated lead levels in U.S. children Prevention Branch California Childhood Lead Poisoning to help define the extent to which lead toxicity from these sources con- tinues to be a problem. The results of our review suggest that elevated levels of lead in paint, dust, soil, imported pottery and ceramic ware, ethnic remedies, and some imported candies continue to be areas of concern, while typical levels in food products appear to be acceptable. It is important to continue monitoring lead levels in children as well as in Chewing and sucking on things is normal in young children, but it can increase the risk of lead poisoning if lead is present. Lead was removed from paint in 1978, but dust from older paint can environmental and food sources. easily be ingested by children via their hands or toys.

he U.S. Environmental Protection points (Canfield et al. 2003), suggesting Other known risk factors for lead Agency (EPA) estimates that 430,000 that any exposure to lead can put a child toxicity among children include: (1) liv- AmericanT children between 1 and at risk. What is most disconcerting is the ing in housing built before 1978, when 5 years old have blood-lead levels at fact that sustained, low-level lead ex- the federal government banned the or above 10 micrograms per deciliter posure can go unchecked because such use of lead-based paint for houses; (2) (µg/dl) (EPA 2004b), costing the United neurological abnormalities may have no recent or ongoing renovation in pre- States an estimated $43.4 billion dollars apparent symptoms (Koplan 2002). 1978 housing; (3) peeling paint in pre- annually (Landrigan et al. 2002). Young children are at the great- 1978 housing; (4) urban residence; (5) Elevated blood-lead levels put children est risk of lead toxicity, as their body low-socioeconomic status; (6) the use at risk for deficits in neurobehavioral- burden (the total amount of lead in the of imported pottery and ceramic ware; cognitive performance that manifest later body at a given time) is influenced by and (7) the use of traditional ethnic in childhood (Needleman et al. 1990). weight. Children between 1 and 4 years medicines (CDC 1991). These deficits include IQ deficiency, be- old face additional risks because of pica In the past, lead in the ambient air havioral disorders and impaired hearing. (ingestion of nonfood substances) and and in food contributed much more In addition, 400,000 more children may hand-to-mouth activity (Mielke and to overall exposure of the general U.S. be yet undiagnosed (GAO 1999). Reagan 1998). These factors are of par- population than it does now. Because Furthermore, these deficits may ticular concern because lead can damage lead from these sources has been drasti- persist even 20 years after exposure a child’s rapidly developing nervous cally reduced, blood-lead levels in U.S. (Bernard and McGeehin 2003), since system, making the threat of permanent children and the general population are lead is sequestered in bone. While the neurological damage greatest during the lower than they were 1 or 2 decades ago body can remove it, some amount re- first 6 years of life (CDC 1991).Research (EPA 2004b). However, despite the de- mains behind and with chronic expo- has shown that the negative effects of creased use of lead in food canning, gas- sure can accumulate. There may be no lead on the cognitive function of children oline and paint, lead will remain in our safe level of lead. For example, IQ has are persistent across cultures, racial and environment for many years. While it is been shown to be adversely affected at ethnic groups, and social and economic critical that recognized sources of lead levels below 10 µg/dl, by as much as 7.4 classes (Nordin et al. 1994). be minimized in a child’s environment,

180 CALIFORNIA AGRICULTURE • VOLUME 60, NUMBER 4 eradicating all sources of excessive lead Although this information is useful, gasoline prior to its ban in the United exposure would be difficult and expen- the prevalence of low-level lead toxicity States in 1996. sive, making continual surveillance for (less than or equal to 10 µg/dl) across In 1999, the EPA estimated that 8% lead toxicity imperative. populations has not been documented. of American children were served by In an effort to detect and treat cases Our research group collected lead toxic- lead-contaminated water systems, pos- of lead toxicity, government agencies ity data in 1996 from children in four sibly contributing up to 20% of a child’s such as the U.S. Centers for Disease California counties, representing both lead exposure. The Safe Drinking Water Control and Prevention (CDC), and low- and high-risk environments (where Act of 1974, created to improve U.S. organizations such as the American high risk was defined as being of low- drinking water by establishing tolerable Academy of Pediatrics, have recom- income and living in houses built before levels of a variety of contaminants (in- mended universal screening for all 1950). We found that approximately cluding chemicals), recommended that children aged 6 months to 6 years 13% of the children had blood-lead lev- no amount of lead in tap water is safe. participating in low-income programs. els greater than 5 µg/dl. This is of con- However, this was solely a recommen- This concept has been met with mixed cern given that other researchers have dation and was not enforced (EPA 2005). responses from the medical community. found detrimental effects on IQ (Canfield Contamination of water by lead Proponents suggest that the plan meets et al. 2003) and reading and math scores generally occurs during the last phases all the cost-benefit criteria for suggested at blood-lead levels from 5 to 10 µg/dl of water delivery, making it difficult to population-based screening and that it (Lanphear et al. 2000). regulate. Upon entering the plumbing would be cost effective if the prevalence We review several factors that con- system of a house or building, the cor- of lead poisoning in the population was tribute to elevated lead levels in chil- rosiveness of water causes lead to leach even as low as 0.1%. Opponents argue dren living in the United States, to help from the pipes (plumbing may have that the prevalence of elevated blood- define the extent to which lead toxicity lead pipes, solder or other materials) lead levels is low, and that there is no from environmental and food sources (EPA 2005). The Lead and Copper Rule, effective therapy for children with low continues to be a problem. It is antici- published by EPA in 1991, aimed to lead levels. Indeed, except in cases of pated that these results will help revise reduce the corrosivity of water through extreme, acute exposure, therapy for children with elevated blood-lead lev- Sustained, low-level lead exposure may result in els consists of education and efforts to neurological abnormalities without any apparent symptoms. remove sources of lead from the child’s environment (DHS 2003a). Current surveillance systems may existing education programs so that the treatment. EPA reviews of water sys- not provide accurate estimates of lead Healthy People 2010 objective (a set of tems in 2004 found that the Lead and toxicity among children. Despite na- national health goals) to eliminate all el- Copper Rule was 96% effective for tional recommendations to screen all evated blood-lead levels in children can systems serving 3,300 people or more children participating in low-income be met (DHHS 2000). (Millett 2005). In response to these find- programs, screening rates vary and Lead in the environment ings, EPA recently issued the Drinking remain low; an estimated 400,000 U.S. Water Lead Reduction Plan to help children with elevated blood-lead levels The U.S. Department of Health and further reduce lead levels (Millett 2005), are undiagnosed (GAO 1999). While Human Services has referred to lead and preliminary monitoring results re- the national childhood blood-lead sur- poisoning as the most common envi- leased by EPA in March 2005 reported veillance program collects state data, ronmental pediatric problem for young that national lead levels in water were reporting is not consistent. In California, children (CDC 1991). Major environ- no longer elevated (Davis 2005). for example, only cases where blood-lead mental sources of lead poisoning in Single paint chips may contain be- levels exceeded 25 µg/dl were reported children are lead-contaminated water, tween 6.45 and 32.25 milligrams lead prior to Jan. 1, 2003. lead-based paint, and soil and dust. per square inch (1 and 5 milligrams per Nationally representative surveys Lead exists in the environment as a square centimeter) (CDC 1991), and and local prevalence data is also col- result of geological activity and human analysis of NHANES III data shows that lected, but these report blood-lead actions, and occurs naturally in soil and children living in housing built prior levels at or above 10 µg/dl. Recent dust (natural lead). Through a global to 1946 are 5.1 times as likely to have estimates from the National Health assessment of atmospheric trace metals, elevated blood-lead levels (Bernard and and Nutrition Examination Survey Nriagu and Pacyna (1988) concluded McGeehin 2003). Estimates from 1998 (NHANES) IV showed that the preva- that lead as an environmental pollutant show that 40% of homes in the United lence of blood-lead levels greater than is essentially caused by human action. States still contain lead-based paint, 16% or equal to 10 µg/dl among American Human activities include the mining of homes have hazardous levels of lead children 1 to 5 years old is 2.2% (Meyer and smelting of ores, the combustion in dust, and 7% of homes have hazard- et al. 2003). Similarily, CDC 2003 sur- of fossil fuels and the dissemination of ous soil lead levels (EPA 2004b). EPA veillance data indicates 2.04% preva- lead through industrial processes; the estimates that between 5.9 million and lence of elevated blood-lead levels in most pervasive is tetra ethyl lead, which 11.7 million children are exposed to lead children under 6 (CDC 2003). was used as an antiknock additive in in dust and soil annually (EPA 2004a).

http://CaliforniaAgriculture.ucop.edu • October–December 2006 181 amounts of lead tetroxide, while greta, a bright yellow or orange powder, also contained high lead levels. Other folk medicines associated with lead poison- ing include ghasard, an Indian remedy; litargirio, a Hispanic folk remedy; and paylooah, a medicine used in the Hmong culture (Sprinkle 1995; CDC 1993). In light of these findings, cultur- ally sensitive warnings about these products are warranted. Photos: California Childhood Lead Poisoning Prevention Branch California Childhood Lead Poisoning Photos: Lead in food Data from 1990 suggests that chil- dren up to 16 years old consume be- tween 3.8 µg and 8.5 µg of lead per day from food, water and beverages, mean- Imported ceramic cookware may contain high levels of lead in the decoration. ing that approximately 16% of a 2-year- Lead leaches into food during cooking and storage, especially with acidic foods. old child’s daily intake of lead may come from food (Abadin and Llados Soil becomes contaminated with lead ing a 4-week period. Consumption of 1999) (table 1). through three methods: the weathering fruit punch from the same container Candy. Recent warnings issued by and chipping of lead-based paint, indus- by others resulted in a 20% increase in the U.S. Food and Drug Administration trial pollution or deposition from the use blood-lead concentrations (Matte et al. (FDA) and state health departments of leaded gasoline. EPA regulations have 1994). Glazes used on traditional pots have raised concern about lead con- greatly reduced the risk of lead exposure imported from both South American tamination in candy imported from from industrial pollution or the use of and Asian countries may contain lead. Mexico (FDA 2004a; DHS 2005). The leaded gasoline, but exposure to lead- Lead can leach into food when these CDC reported that, based on routine based paint continues to be a problem. contaminated pots are used for cooking blood-lead screenings, six California or storage. Cooking with acidic foods children suffered from lead toxicity af- Other sources of lead (such as tomatoes, vinegar, alcohol or ter eating tamarind- and chili-flavored Imported ceramic cookware. soy sauce) in these pots is of special candy imported from Mexico. The FDA Another potential source of lead concern because their low pH enhances then found between 0.3 µg to 0.4 µg of contamination is imported ceramic the leaching process (DHS 2003b). lead per gram of candy. The California cookware. One case study reported Ethnic remedies. The use of folk Department of Health Services inde- that a child who regularly consumed medicine by ethnic groups has also pendently found elevated lead levels fruit punch stored in a lead-glazed urn been associated with lead toxicity in candy as well (> 0.5 µg of lead per had an increase in blood-lead levels (CDC 1993). A recent analysis of 70 ay- gram) (DHS 2005). It is believed that from 22.79 µg/dl to 95.32 µg/dl dur- urvedic products (those based on tra- lead contamination may be introduced ditional medicine in India) from stores from ingredients or occur during candy TABLE 1. FDA analysis of lead concentrations in the Boston area showed that almost processing such as drying, storing and in food products 20% contained lead, ranging from 5 grinding (CDC 2005), or that lead may µg/g to 37,000 µg/g per sample (Saper leach into the candy from tainted wrap- Mean lead Standard Food group concentration deviation et al. 2004). Surma, also know as kohl, pers (Fuortes and Bauer 2000). kajal or al-kahl, is an ethnic remedy Currently, there is no regulatory limit ...... µg/g ...... that is applied to the eyes of children in for lead in food, but the FDA has estab- Dairy 0.002 0.005 some Asian communities. This powder lished a Provisional Daily Total Tolerable Meat, fish, poultry 0.003 0.006 primarily consists of lead sulphide, and Intake of 6 µg of lead per 30-gram food Grain, cereal 0.003 0.006 products use may result in elevated blood-lead serving (0.2 µg/g) (Lynch et al. 2000). Vegetables 0.004 0.006 levels, possibly increasing blood lead The candies described above each Fruit, fruit juices 0.004 0.006 in children by as much as 8.5 µg/dl weighed 30 grams or more per piece, Oils, fats, 0.003 0.007 (Sprinkle 1995). so consumption of a single candy could shortening In Latino communities, the use of deliver lead in an amount greater than Sugar, desserts 0.002 0.007 Beverages 0.007 0.007 azarcon and greta, two herbal rem- the Provisional Daily Total Tolerable Mixed dishes 0.002 0.003 edies used to treat intestinal and stom- Intake (August and Brooks 2004); how- Infant food 0.003 0.005 ach ailments, may result in elevated ever, factors such as nutritional status, blood-lead levels. Samples of azarcon, age and gender influence the amount Source: FDA 2004b. an orange powder, contained large of lead absorbed after consumption.

182 CALIFORNIA AGRICULTURE • VOLUME 60, NUMBER 4 As a result, it is difficult to accurately This, coupled with research suggesting determine the true risk posed by con- that only 5% to 10% of lead contained sumption of these products. in cocoa is bioavailable to the body Chocolate. The issue of chocolate as (Mounicou et al. 2002), suggests that a source of lead exposure among chil- there is limited risk of lead toxicity to dren has also recently surfaced in the regular consumers of chocolate products. media, although warnings issued by Calcium supplements. Historically, state departments of health only pertain calcium supplements have also been to imported candy. Contamination of found to contain lead. Deposits of cal- chocolate by lead is thought to occur be- cium that are mined for supplements cause the majority of beans are grown in may contain lead. Recent lab analysis of locations that still use leaded gasoline. 21 different over-the-counter products Prevention Branch California Childhood Lead Poisoning Photos: Despite high per-capita consumption of completed in 2000 demonstrated that chocolate in the United States, there is a 11 had detectable lead levels, between paucity of data on lead concentrations 1 and 3 µg per dose (Ross et al. 2000). in chocolate products. In California, analyses of 136 products The American Environmental Safety demonstrated that approximately 67% Institute, an environmental advocacy exceeded 1.5 µg of lead per dose, the group based in California, tested a vari- tolerable level set by Proposition 65 ety of chocolate products and reported regulations (Scelfo and Flegal 2000). that they contained between 0.00157 µg (Prop. 65 was enacted to protect and and 0.105 µg lead per gram chocolate. inform Californians about exposure The USDA Total Diet Study (FDA 2004b) from harmful chemicals.) Although found between 0.0 and 0.110 µg lead per intestinal absorption of lead is reduced gram chocolate. Analysis of chocolate in the presence of calcium, it is not samples by our research group yielded completely inhibited. Considering that similar results. Mean lead levels ranged many Americans consume calcium from 0.0010 to 0.0965 µg lead per gram supplements on a daily basis per recom- chocolate (table 2). mendations from their physicians, the The Codex Alimentarius, global potential for lead contamination from Large amounts of lead have been found in food standards developed by the these products should not be ignored. certain ethnic remedies. Top, surma or kohl Food and Agriculture Organization However, consumption of calcium sup- is applied to children’s eyes in some Asian (FAO) and World Health Organization plements effectively reduces the risk of communities. Center and bottom, azarcon and greta are used to treat intestinal and stomach (WHO), limits the lead content of co- osteoporosis, and until further research ailments in some Latino communities. coa powder or beans to 1 µg of lead is completed, it should be maintained per gram product (FAO/WHO 1981). that the benefits of supplement usage Analysis of a variety of chocolate prod- outweigh the risks. products, methods should be developed ucts from various global locations, to reduce their lead content. Ongoing surveillance needed completed by a Swiss research group Due to drastic reductions in lead in 2002, found that the lead contents Ongoing surveillance of lead levels in the ambient air and food, blood- of these items ranged from 0.011 µg to in environmental and food sources is lead levels in children and the general 0.769 µg per gram, below the interna- imperative to ensure that appropriate U.S. population are now lower than tional standards (Mounicou et al. 2003). actions can be taken in a timely fashion. they were 1 or 2 decades ago (EPA Our review suggests that 2004b); however, NHANES and CDC elevated levels of lead oc- data suggests that between 2% and TABLE 2. Lead concentrations in chocolate* cur in paint, dust, soil, im- 3% of American children still exhibit ported pottery and ceramic elevated levels of blood lead (Meyer et No. Mean lead Standard Food samples concentration deviation ware, ethnic remedies, and al. 2003). At the present time, therapy some imported candies. for children with blood-lead levels be- ...... µg/g ...... Dark chocolate 8 0.0965 0.0892 Furthermore, although the tween 10 µg/dL and 20 µg/dL is lim- pieces monitoring of lead levels ited to education and efforts to remove Milk chocolate bar 12 0.0418 0.0705 in candy should continue, sources of lead in the child’s environ- Premium milk 12 0.0949 0.1160 current data suggests that ment, and no action is taken for children chocolate bar Candy-coated 8 0.0010 0.0020 in the United States the risk with blood levels less than 10 µg/dL. chocolate of lead toxicity from these Studies show that the education and

* Food products were collected from Albertson’s, The Nugget and Safeway items is relatively low. If counseling of parents and children is in Davis, CA. Methods available upon request from authors. future analysis shows lead effective in reducing blood-lead levels, contamination of such food in concert with the careful removal of

http://CaliforniaAgriculture.ucop.edu • October–December 2006 183 Fuortes L, Bauer E. 2000. Lead contamination of im- ported candy wrappers. Vet Hum Toxicol 42(1):41–2. [GAO] US General Accounting Office. 1999. Lead Photos: Center for Environmental Health Photos: poisoning: Federal health care programs are not effectively reaching at-risk children. www.gao.gov/archive/1999/ he99018.pdf. In recent years, lead has been detected in children’s jewelry and soft vinyl Kimbrough RD, LeVois M, Webb DR. 1994. Manage- (PVC) lunchboxes. In 2005, the Center for Environmental Health, a nonprofit ment of children with slightly elevated blood lead levels. based in Oakland, initiated lawsuits against retailers and manufacturers of Pediatrics 93(2):188–91. these products, resulting in widespread reformulation to reduce lead Koplan JP. 2002. Managing elevated blood lead levels among children. Centers for Disease Control and Preven- content. In July 2006, the U.S. Food and Drug Administration warned tion, Atlanta, GA. www.cdc.gov/nceh/lead/CaseManage- lunchbox manufacturers to stop marketing vinyl lunchboxes containing any ment/caseManage_main.htm. level of lead that may leach into children’s food. Landrigan PJ, Schechter CB, Lipton JM, et al. 2002. For more information, go to www.cehca.org. Environmental pollutants and disease in American chil-

RECEN T E V EN S dren: Estimates of morbidity, mortality, and costs for lead poisoning, asthma, cancer, and developmental disabilities. Environ Health Perspect 110(7):721–8. hot spots such as peeling paint on win- [CDC] Centers for Disease Control and Prevention. Lanphear BP, Dietrich K, Auinger P, Cox C. 2000. Cog- 1991. Preventing lead poisoning in young children: A state- nitive deficits associated with blood lead concentration dow sills and porch banisters (Rooney ment by the Centers for Disease Control. US Department < 10 microg/dL in US children and adolescents. Public et al. 1994). Most importantly, studies of Health and Human Services, Atlanta, GA. www.cdc.gov/ Health Rep 115(6):521–9. have demonstrated that cognitive func- NCEH/lead/Publications/books/plpyc/contents.htm. Lynch RA, Boatright DT, Moss SK. 2000. Lead- CDC. 1993. Lead poisoning associated with use of contaminated imported tamarind candy and children’s tion improved when moderate blood- traditional ethnic remedies — California, 1991–1992. blood lead levels. Public Health Rep 115(6):537–43. lead levels in children were reduced MMWR 42(27):521–4. Matte TD, Proops D, Palazuelos E, et al. 1994. Acute (Kimbrough et al. 1994). Perhaps the CDC. 2003. Tested and confirmed elevated blood high-dose lead exposure from beverage contaminated by lead levels by state, year, and blood level group for traditional Mexican pottery. Lancet 344(8929):1064–5. parents of children with levels between children < 72 mos. Atlanta, GA. www.cdc.gov/NCEH/ Meyer PA, Pivetz T, Dignam TA, et al. 2003. Surveillance 5 µg/dL and 10 µg/dL should also be lead/surv/database/State_Confirmed_byYear_2004_ for elevated blood lead levels among children — United included in these education programs for%20website.xls. States, 1997­–2001. MMWR Surv Summ 52(10):1–21. CDC. 2005. Lead in candy: Questions and answers. Mielke HW, Reagan PL. 1998. Soil is an important to ensure that they are adequately edu- Atlanta, GA. www.cdc.gov/nceh/lead/faq/candy.htm. pathway of human lead exposure. Environ Health Perspect cated on the risk factors associated with Davis T. 2005. Getting the lead out: The ongoing 106 Suppl 1:217–29. lead exposure. In the future, sustained quest for safe drinking water in the nation’s capital. US Millett J. 2005. EPA to strengthen protection from lead House of Representatives, Committee on Government in drinking water. US Environmental Protection Agency, efforts to reduce the lead exposure of Reform. http://reform.house.gov/GovReform/Hearings/ Washington, DC. Press release.http://yosemite.epa.gov/ American children will be necessary to EventSingle.aspx?EventID=23039. opa/admpress. achieve the Healthy People 2010 goal of [DHHS]. US Department of Health and Human Ser- Mounicou S, Szpunar J, Andrey D, et al. 2002. De- eliminating all elevated blood-lead levels vices. 2000. Healthy People 2010: Understanding and velopment of a sequential enzymolysis approach for the Improving Health. www.healthypeople.gov/Document/ evaluation of the bioaccessibility of Cd and Pb from cocoa. in children. tableofcontents.htm#under. Analyst 127(12):1638­­–41. [DHS] California Department of Health Services. Mounicou S, Szpunar J, Andrey D, et al. 2003. Con- 2003a. Children at Risk. Childhood Lead Poisoning Pre- centrations and bioavailability of cadmium and lead in vention Branch, Sacramento, CA. www.dhs.ca.gov/ cocoa powder and related products. Food Addit Contam childlead/html/clppb.html. 20(4):343­–52. K. Heneman is Postdoctoral Researcher, and DHS. 2003b. Lead in Tableware — Resources. Child- Needleman HL, Schell A, Bellinger D, et al. 1990. S. Zidenberg-Cherr is Cooperative Extension hood Lead Poisoning Prevention Branch, Sacramento, The long-term effects of exposure to low doses of lead Specialist, Department of Nutrition, UC Davis. CA. www.dhs.ca.gov/childlead/tableware/twhome.html. in childhood. An 11-year follow-up report. N Engl J Med DHS. 2005. Lead in Candy — 2005 Laboratory 322(2):83–8. Results. Sacramento, CA. www.dhs.ca.gov/fdb/PDF/ Nordin JD, Rolnick SJ, Griffin, JM. 1994. Prevalence References 05%20candy%20results%20for%20web%20final.pdf. of excess lead absorption and associated risk factors in Abadin H, Llados F. 1999. Toxicological profile for [EPA] US Environmental Protection Agency. 2004a. children enrolled in a midwestern health maintenance or- lead. Agency for Toxic Substances and Disease Registry, America’s children and the environment. Washington, ganization. Pediatrics 93(2):172­–7. Centers for Disease Control and Prevention, Atlanta, GA. DC. www.epa.gov/envirohealth/children/index.htm. Nriagu JO, Pacyna JM. 1988. Quantitative assessment www.atsdr.cdc.gov/toxprofiles/tp13.html. EPA. 2004b. Measure B1: Lead in the blood of of worldwide contamination of air, water and soils by August K, Brooks L. 2004. State Health Department children. Washington, DC. www.epa.gov/economics/ trace metals. Nature 333:134–9. warns consumers not to eat chaca chaca, lead-contami- children/body_burdens/b1.htm. Rooney BL, Hayes EB, Allen BK, Strutt PJ. 1994. De- nated candy from Mexico. California Department of velopment of a screening tool for prediction of children Health Services, Sacramento, CA. www.applications.dhs. EPA. 2005. Lead in drinking water. Washington, DC. www.epa.gov/safewater/lcrmr/fs_consumer.html. at risk for lead exposure in a midwestern clinical setting. ca.gov/pressreleases/store/PressReleases/04-15.html. Pediatrics 93(2):183–7. Bernard SM, McGeehin MA. 2003. Prevalence [FAO/WHO] United Nations’ Food and Agriculture Organization and World Health Organization. 1981. Ross EA, Szabo NJ, Tebbett IR. 2000. Lead content of of blood lead levels > or = 5 micro g/dL among US calcium supplements. JAMA 284(11):1425–9. children 1 to 5 years of age and socioeconomic and de- Codex Alimentarius. Codex Standard for Chocolate 87. mographic factors associated with blood of lead levels www.codexalimentarius.net/web/standard_list.jsp. Saper RB, Kales SN, Paquin J, et al. 2004. Heavy metal 5 to 10 micro g/dL. Third National Health and Nutrition [FDA] US Food and Drug Administration. 2004a. content of ayurvedic herbal medicine products. JAMA Examination Survey, 1988–1994. Pediatrics 112(6 Pt FDA statement on lead contamination in certain candy 292(23):2868–73. 1):1308–13. products imported from Mexico. Washington, DC. www. Scelfo GM, Flegal AR. 2000. Lead in calcium supple- Canfield RL, Henderson Jr CR, Cory-Slechta DA, et fda.gov/bbs/topics/news/2004/new01048.html. ments. Environ Health Perspect 108(4):309–19. al. 2003. Intellectual impairment in children with blood FDA. 2004b. Total diet study statistics on element Sprinkle RV. 1995. Leaded eye cosmetics: A cultural lead concentrations below 10 microg per deciliter. results. Washington, DC. www.cfsan.fda.gov/~acrobat/ cause of elevated lead levels in children. J Fam Pract N Engl J Med 348(16):1517–26. tds1byel.pdf. p 58–68. 40(4):358–62.

184 CALIFORNIA AGRICULTURE • VOLUME 60, NUMBER 4 research Article t Cost-benefit analysis conducted for nutrition education in California

by Amy Block Joy, Vijay Pradhan and George Goldman USDA/Ken Hammond USDA/Ken

Documenting the cost-effectiveness of nutrition education programs is important to justify and determine expenditures and ensure continued funding. A cost-benefit analysis was conducted using the program demo- graphics and food-related dietary behavior of participants enrolled in California’s Expanded Food and Nu- trition Education Program (EFNEP), based on methodology developed by Virginia Cooperative Extension. The initial benefit-cost ratio for nutrition education in California was 14.67 to 1.00. Several sensitivity analyses were done to estimate the effect of changes in key variables. The result- ing benefit-cost ratios ranged from 3.67 to 1.00, to 8.34 to 1.00, meaning ANR Communication Services/Suzanne Paisley that for every $1.00 spent on nutri- tion education in California, between $3.67 and $8.34 is saved in health care costs. These results bolster the argument that nutrition education programs are a good investment and funding them is sound public policy. The impact of nutrition education on health care spending has not been extensively studied; the authors utilize a new cost-benefit analysis methodology, developed by Virginia researchers, to quantify nutrition education impacts in California. Top, a mother receives health counseling at a WIC clinic in Virginia. Above, California EFNEP participants learn healthy cooking choices. uccessful nutrition education pro- grams should be effective in both improvingS dietary health and ensuring Del Tredici et al. 1988). Although the government, public, private and nonprofit a positive economic impact. Evaluations positive improvements in dietary health programs (Sassone and Schaeffer 1978). of the Expanded Food and Nutrition due to nutrition education are well docu- In 1999, Virginia Cooperative Education Program (EFNEP), federally mented, its value in health care savings Extension (VCE) designed, implemented funded nutrition education for low- has not been extensively studied. and published the first cost-benefit income families, have demonstrated Documenting cost-effectiveness is analysis to evaluate the economic value many positive behavior changes over critical if community nutrition programs of EFNEP (Rajgopal et al. 2002; Lambur the last 35 years. These include improve- are to survive under current economic et al. 1999). The Virginia researchers ments in diet (increasing intakes of fruits, conditions. Cost-benefit analysis is one of developed a methodology to measure vegetables and fiber, and decreasing fat the standard ways to document economic the economic impact of nutrition educa- and soda consumption); food shopping benefits. Used primarily by economists tion by using behavior changes among and preparation (saving money, reading for policy- and decision-makers, cost- EFNEP participants. They developed labels, food safety); and nutrition knowl- benefit analysis offers an analytic pro- criteria associated with a number of edge and attitudes (Contento et al. 1995; cedure that is widely applicable to most chronic diseases/conditions and deter-

http://CaliforniaAgriculture.ucop.edu • October–December 2006 185 For every dollar spent on nutrition education, at least Glossary $3.67 is saved on delayed medical treatment costs. Benefits (direct, indirect): Direct benefits are estimated monetary mined the associated medical costs, and the participants who received nutrition amounts for delaying treatment implemented sensitivity analyses on the education lessons (Rajgopal et al. 2002). or preventing a chronic disease; data to adjust for certain assumptions. Both the Virginia and California indirect benefits include increased The VCE calculated a benefit-cost studies used computer software called productivity, longer life span and ratio ranging from 2.16 to 1.00, to 17.04 ERS-3.01 (Evaluation and Reporting reductions in lost workdays. to 1.00, meaning that between $2.16 and System, version 3.01, National EFNEP Cost-benefit analysis: $17.04 was saved in health care costs for Program, USDA; Merrill et al. 1993) to Benefits (direct + indirect) every $1.00 that was spent in Virginia on enter the demographic and dietary be- Costs nutrition education programs for low- havior data of EFNEP participants. The income residents. We calculated similar monetized benefits were direct and indi- Costs (direct, indirect): Actual ex- benefit-cost ratios for California using rect: The direct benefits were the dollars penditures for program imple- program demographics and food-related of postponed health care costs due to mentation. Direct costs include dietary behaviors from participants en- the delay of these chronic diseases and personnel, benefits, travel, sup- rolled in the state’s EFNEP program, uti- conditions, and the indirect benefits plies and equipment. Indirect costs lizing VCE’s methodology. In California, were those that increased productivity include the value of time lost from the EFNEP program is administered due to a healthy lifestyle (e.g., fewer work by participants. through the UC Division of Agriculture sick days) (Rajgopal et al. 2002). Costs (in-kind): Total estimated costs and Natural Resources (ANR). for office space and maintenance EFNEP for low-income families (donated by county budgets) and Virginia methodology The EFNEP program (in both Virginia utilities (not paid for directly). In The VCE studies provided detailed and in California) teaches low-income California these were calculated background on two important factors families how to purchase, safely prepare using estimated proportions of used in their methodology: the relation- and serve a balanced, highly nutri- Virginia’s costs. ship between diet and chronic disease, tious diet. The program emphasizes the Costs ­(tangible, intangible): Tangible and the assigning of monetary values to increased consumption of fruits and costs are easily monetized (direct nutrition outcomes. Cost-benefit analy- vegetables, decreased consumption of and indirect benefits); intangible sis was used to measure economic value fat, and improved skills in food safety, costs are not (e.g., increased self- by comparing “benefits” (monetized in preparation and shopping, with the esteem, improved quality of life). terms of decreased medical treatment long-range goal of improved health and For example, a doctor’s visit is costs) with “costs” (actual nutrition risk reduction for chronic diseases. The considered tangible, while “not education expenses over a set time). The 1990 Dietary Guidelines for Americans feeling well” is intangible. Virginia researchers decided to compute were the basis for instruction during the Equipment costs: The total amount the monetary benefits of EFNEP using time of this study (USDA/DHHS 1995), paid for equipment (e.g., comput- the future savings in health care costs although the recently updated version ers, teaching supplies). accrued by the potential avoidance of is currently used. Participants receive Salaries and benefits: Total funds certain diseases and conditions among intensive nutrition education from spent on all personnel, including program delivery staff, program supervisors, clerical and admin- istrative staff and state office staff for fiscal year 1998-1999 (Oct. 1, 1998, to Sept. 30, 1999). Staff mer- its awarded in July 1999 for work done in 1998 were also included. Supplies and expenses: Total amount paid for state/county programs

to deliver the nutrition education, ANR Communication Services/Suzanne Paisley including the cost of materials, phone, copying, fax, demonstra- tion supplies, teaching materials and utilities. Travel costs: Total travel costs for county-based program staff and state staff (e.g., teaching, training, meetings, conferences). EFNEP provides 4 to 6 weeks of classes in a small-group setting to more than 13,000 low-income California families each year. The program is administered by UC nutrition scientists and educators.

186 CALIFORNIA AGRICULTURE • VOLUME 60, NUMBER 4 trained paraprofessional staff for ap- Hammond USDA/Ken proximately 6 to 8 hours per week over a period of 4 to 6 weeks. Entry and exit dietary-assessment data is collected on every participant enrolled in and graduated from EFNEP. The study in California was done us- ing EFNEP data from 1998, following the methodology developed by the VCE (using their 1996 results). Virginia used health and labor statistics from 1996 in their assessments of monetary values, utilizing national averages. During the time period that we studied (1998), California EFNEP served 13,430 enrolled families in 17 counties. County locations varied from urban (Los Angeles, Orange, San The link between diet and chronic diseases/conditions, such as heart disease, diabetes Francisco) to rural (Tulare, Stanislaus, and obesity, has been established by scientific research. EFNEP uses a food behavior Butte). Enrolled participants were checklist to evaluate changes in nutrition behavior among participants, such as ethnically diverse (62% Hispanic, 9% increases in fruit and vegetable consumption. black, 15% white, 12% Asian and 1% Native American) and primarily female individual was unable to complete These were studied because they have (92%). The median age was 25 to 30 the form). The data was collected and a known dietary association, and the years old, and 10% of the women were entered into the ERS-3.01 computer pro- EFNEP program teaches participants pregnant. Most participants had at gram, and the results were aggregated. how to reduce their risk and delay or least two children (65%), while 35% of The nutrition education lessons were prevent their onset. For California, ad- the families had one child and 8% had taught weekly in a group setting (four ditional nutrition behavior indicators five or more. Approximately 41% were to 10 individuals) over a period of 4 to 6 were included: one question on increas- enrolled in the Special Supplemental weeks, which has been documented as ing fiber consumption; two separate Nutrition Program for Women, Infants, effective in promoting positive dietary questions on increasing fruit and veg- and Children Program (WIC), 34% in changes and meeting program goals etable consumption; three questions on the Food Stamp Program, 19% in Head and objectives (Del Tredici et al. 1988; decreasing fat consumption; and one Start and 35% in the Child Nutrition Joy 2004). Eighty-six percent of the question on decreasing the number of Program; all are federal support pro- families completed the nutrition-educa- visits to fast food restaurants. grams for low-income families. The tion program lessons and the pre- and (2) Estimation of diet-related risk. majority of individuals (98%) met the post-tests, and were continuing in the The Virginia study relied on the fact that federal income level for eligibility (at or program. Of the 14% that did not com- there is a relationship between diet and below 150% poverty level) with 75% of plete the program, the reasons given chronic disease, which it discussed in them at or below the 100% poverty level. included: moved without a contact detail. Likewise, in the literature there is All enrolled families completed address (19%), family concerns (12%), evidence of this relationship, but it is im- two records: the EFNEP family record took a job/returned to school (12%) and precise. In order to justify this assump- for demographic information and the other (47%). tion, the Virginia researchers created a California food behavior checklist for formula to quantify this relationship. For Determining benefits and costs pre- and post-evaluation data. The example, how much of a role does diet California food behavior checklist con- The VCE methodology included a list play in the development of colorectal tained 14 of the same questions used in of program benefits and costs associated cancer? We can estimate that the link is the Virginia study, plus an additional with EFNEP (see glossary), as well as not 100%, nor is it 0%. Other factors are seven questions. Of the 21 questions, the following three major assumptions. involved — genetics, lifestyle, age, eth- seven had been validated in terms of (1) Diet and chronic disease link. The nicity and environmental toxins, as well predicting increased fruit and vegetable link between diet and disease has been as factors that have not yet been identi- consumption (Murphy et al. 2001; established by research. For their analy- fied. The Virginia researchers estimated Blackburn et al. 2006). sis, the Virginia researchers included that 35% of the incidence of colorectal Trained program-delivery staff col- heart disease, stroke, hypertension, cancer is diet-related. For each condition, lected the family record and checklist colorectal cancer, osteoporosis, type 2 this estimation was computed based on (both were self-reported, unless the diabetes, obesity and foodborne illness. published reports from sources cited in

http://CaliforniaAgriculture.ucop.edu • October–December 2006 187 s To establish a cost-benefit relationship, the amount of money saved by delaying a disease or condition for 5 years was estimated. Blood pressure ANR Communication Services monitoring can help to detect heart disease, hypertension and stroke risks.

TABLE 1. Nutritional behavior and scores of EFNEP participants, filtered pre- and post-test for optimal nutritional change for eight chronic diseases/conditions

Optimal Nutritional behavior taught in EFNEP behavior score Healthy People 2000, a publication of approach for the indirect cost-benefits, Colorectal cancer objectives aimed at reducing chronic dis- based on worker productivity (e.g., Increase fruit and vegetable consumption ≥ 4 ease by changing lifestyle factors, includ- health care savings as well as the value Decrease fat consumption ≥ 4 Increase fiber consumption ≥ 4 ing diet (DHHS 1990). of the work lost due to illness). Use food labels to select healthy foods ≥ 4 (3) Behavior changes for 5 years. The tangible benefits of nutrition Satisfy all above criteria simultaneously (150; 7.65%)* Studies have shown that some EFNEP education were the costs saved by

Foodborne illness families who attain optimal nutritional preventing or delaying the onset of a Thaw foods safely 1, 2 behaviors maintain them for 5 years disease or condition. The reason that Store foods safely 1, 2 following the nutrition education the delay is of value is because a given Satisfy all above criteria simultaneously (2,011; 27.95%) (Nierman 1986). Documentation for amount today is worth more in the fu- Heart disease more than 5 years has not been pub- ture. We used 5% for the discount rate, Increase fruit and vegetable consumption ≥ 4 lished. Therefore, the Virginia research- as Virginia did. Virginia developed a Decrease fat consumption ≥ 4 ers did a sensitivity analysis to modify formula for calculating both the direct Increase fiber consumption ≥ 4 this assumption. and indirect benefits, which we utilized Use food labels to select healthy foods ≥ 4 in our study (see box). Satisfy all above criteria simultaneously (150; 7.65%) Monetizing educational benefits Nutritional behaviors as benefits Obesity For the California study, benefits Increase fruit and vegetable consumption ≥ 4 Increase fiber consumption ≥ 4 were constructed and calculated in the Nutrition behaviors taught in Decrease fat consumption ≥ 4 same way as in Virginia. Both states EFNEP (Rajgopal et al. 2002) to pre- Decrease eating in fast food restaurants ≤ 2 use the same nutrition-education de- vent or delay diseases/conditions Use food labels to select healthy foods ≥ 4 livery approaches and teach enrolled were used retrospectively to calculate Satisfy all above criteria simultaneously (122; 6.77%) low-income participants how to choose the benefits for each of the eight con- Osteoporosis and prepare a healthy diet (Lambur et ditions/diseases studied (table 1). Drink milk ≥ 4 al. 1999). Both programs have similar The California results were similar to Improve food selection and preparation = 4 Satisfy all above criteria simultaneously (2,769; 20.12%) behavioral results and associated costs. the Virginia results except for the fol- The VCE monetized direct, tangible lowing: (1) the Virginia researchers Stroke/Hypertension benefits (Lambur et al. 1999) using collected data on infant mortality, but Reduce sodium in diet ≥ 4 costs compiled from Healthy People we did not use this information in our Use food labels to select healthy foods ≥ 4 Increase fruit and vegetable consumption ≥ 4 2000 (USDA/DHHS 1995). National analysis because California EFNEP Increase fiber consumption ≥ 4 figures computed by the Virginia does not collect any medical data on Satisfy all above criteria simultaneously (42; 2.57%) researchers were used for treatment participants; and (2) the California costs. Although California medical food behavior checklist included all Type 2 diabetes Increase fruit and vegetable consumption ≥ 4 costs would likely be much higher than the questions used in Virginia, plus Increase fiber consumption ≥ 4 the national average, we used the same seven additional ones. Decrease fat consumption ≥ 4 kind of cost-benefit data as the VCE in In order for participants to be in- Decrease sugar consumption ≥ 4 Use food labels to select healthy foods ≥ 4 order to provide a more conservative cluded in the California analysis, Satisfy all above criteria simultaneously (87; 5.21%) benefits estimation. Similarly, the in- they had to be practicing “optimal

* No. filtered out of 9,528; % practicing optimal nutritional tangible benefits were calculated using nutritional behaviors,” and stringent behavior based on number of program graduates (not shown). the VCE methods reported in their pro- conditions were applied. The rigor of tocol. The VCE used a lost-workdays the analysis was increased with both

188 CALIFORNIA AGRICULTURE • VOLUME 60, NUMBER 4 TABLE 2. Diet-related diseases/conditions and factors used in cost-benefit analysis of direct, tangible benefits*

Graduates EFNEP Incidence Diet- practicing Graduates to graduates rate in related optimal nutr. accrue benefits Present value Avg. age studied population portion behaviors† (calculated) of behavior Total direct Disease/condition of onset [A] [B] [C] [D] [E] [F] benefit

years no...... % ...... no...... $ ...... Colorectal cancer 36 9,528 15.0 35.0 7.65 38.26683 16,424.75 628,583.11 Foodborne illness 23 9,528 2.8 100.0 27.95 74.566128 18,866.83 1,406,826.40 Heart disease 55 9,528 31.2 26.0 7.65 59.127719 691.76 40,902.19 Hypertension 30 9,528 37.4 45.0 2.57 41.211553 697.87 28,760.31 Obesity 23 9,528 37.0 50.0 6.77 119.33343 11,686.59 1,394,600.80 Osteoporosis 45 9,528 28.0 N/A 20.12 536.7694 65,468.86 35,141,719.00 Stroke 45 9,528 1.7 N/A 2.57 4.1627832 13,143.81 54,714.83 Type 2 diabetes 40 9,528 14.5 45.0 5.21 32.390674 45,898.13 1,486,671.30 Total 40,182,766.94

* Factors as discussed in Rajgopal et al. (2002).* † Calculated based on number of program graduates (not shown). filtering of the Virginia questions as haviors could reasonably be attributed (Virginia used Bureau of Labor Statistics well as the seven additional questions to the nutrition education. Causally data from 1995 for these analyses.) from the California food behavior connected participants who improved In general, the population served checklist. Only EFNEP graduates who in their behaviors (e.g., moved from not by EFNEP is less likely to be employed achieved the greatest benefit (a score of practicing to sometimes practicing) but than the general population represented 4 or more) in all the dietary practices did not achieve the highest score were by this data. One of the sensitivity anal- criteria were considered as practicing screened out of the analysis. yses was done to correct for this gen- optimal nutritional behaviors. All other eralization. The benefits of postponing Cost-benefit calculations graduates were eliminated from the pain and suffering were not included analysis (including those who prac- We used the same values as Virginia in the analysis, though this would be a ticed the optimal behaviors at pre-test). for the cost-benefit analysis formula very large number. Responses were filtered by satisfy- because our data was collected at about To select participants practicing op- ing all the criteria simultaneously. For the same time as theirs, and for the timal nutrition behaviors, a cost-benefit example, the original sample size for purposes of this study represented na- method was used. SPSS (Statistical California was 20,999. After removing tional averages. Package for Social Sciences) was used to participants who did not complete ei- We calculated the benefits as fol- analyze results for each of the diseases/ ther the pre- or post-survey, the sample lows: For direct, tangible benefits, if the conditions studied. All participants who size was 9,528. For example, with heart disease/condition could be prevented met the optimal behavior at the pre-test disease, of the 9,528 who completed the or delayed for 5 years, then the benefit were eliminated from the study to in- pre- and post-tests, only 150 answered was the amount of money saved by the crease the likelihood that the improved all eight questions with a score of 4 or delay in treatment costs for 5 years (us- behavior was a result of the nutrition more. This was 7.65% of the California ing a 5% discount rate) (table 2). For education. population who received EFNEP lessons indirect, tangible benefits, calculations California EFNEP costs, benefits (table 1). The only participants used in were done only for those diseases/condi- the calculations were those whose be- tions where data was available (table 3). The cost of delivering nutrition edu- cation to California participants was derived from the California EFNEP Calculation of direct benefits budget, using all tangible costs. The es- Formula: A x B x C x D = E; E x F = benefits* timated total direct costs for fiscal year 1998-1999 were $2,543,667, and the esti- A = Number of EFNEP graduates studied per year mated indirect costs were $236,883 (cost B = Incidence rate (%) of disease/condition in general definitions and calculations available population (estimation) from authors). C = Incidence rate of disease related to dietary change (estimation) The initial benefit-cost ratio for D = Percentage of graduates practicing optimal nutrition behaviors California was 14.67 to 1.00, meaning (calculated/filtered) that for every $1.00 spent on EFNEP, E = Estimated number of graduates to accrue benefits $14.67 was saved in future medical costs (calculated: A × B × C × D = E) (initial cost calculations available from F = Present value of “benefit” for disease/condition (estimation) author). Even though this result seems high, the researchers used stringent con- * Formulas corrected by author postpublication, 10/6/06. http://CaliforniaAgriculture.ucop.edu • October–December 2006 189 TABLE 3. Diet-related diseases/conditions and factors used in cost-benefit analysis to calculate indirect benefits ditions to apply both the benefit calcula- Avg. delayed Graduates Present Total indirect tions and the cost estimates. In addition, Avg. onset due Avg. to accrue value benefit of our initial result was similar to that cal- Disease/ age of to nutrition annual lost benefits of lost delaying condition* onset† education† workdays†‡ (calculated) earnings† condition culated by the Virginia researchers...... years ...... no...... $ ...... A number of different values for the Foodborne illness 23 65 1.5 74.566 1,600.19 119,326.16 critical variables were used for these Heart disease 55 60 58.0 59.128 693.53 41,008.43 analyses; economists call this sensitiv- Hypertension 30 35 41.0 41.212 4,779.67 196,970.20 Obesity 23 65 1.83 119.333 1,952.23 232,959.60 ity analysis. Therefore, the cost-benefit Stroke 45 50 60.0 4.163 2,084.54 8,671.69 results would change as values for these Type 2 diabetes 40 65 0.6 32.391 228.80 7,410.83 variables change. We used sensitivity Total 606,346.91 analyses to quantify how the cost-benefit * Indirect benefits were not calculated for colorectal cancer and osteoporosis. calculation changed, using VCE’s proto- † Source: Lambur et al. 1999. col. The following four sensiti vity analy- ‡ Rate of $7.60 per hour, discounted 5%. ses were done to be conservative and to prevent overstating results (table 4). Low-income populations. TABLE 4. Sensitivity analyses on reducing the number of participants The practicing optimal behavior to 50% and 25% for direct and indirect benefits Virginia study reported that the inci- dence rates used for diseases/condi- Disease/ Direct value Indirect value tions were for the general population, condition* 100% 50% 25% 100% 50% 25% and there was reason to believe that the ...... $ ...... incidence rate in low-income popula- Colorectal 628,583.11 314,291.55 157,145.77 — — — tions such as EFNEP families would be cancer higher. In Healthy People 2000, “low in- Foodborne 1,406,826.40 703,413.20 351,706.60 119,326.16 59,663.08 29,831.54 come” is noted as a special risk factor for illness chronic diseases. Although the Virginia Heart disease 40,902.19 20,451.10 10,225.55 41,008.43 20,504.22 10,252.11 researchers recalculated their cost-benefit Hypertension 28,760.31 14,380.16 7,190.08 196,970.20 98,485.10 49,242.55 to correct for this possible disparity, we Obesity 1,394,600.80 697,300.40 348,650.20 232,959.60 116,478.30 58,239.15 decided to leave the initial incidence rate Osteoporosis 35,141,719.00 17,570,859.50 8,785,429.70 — — — unadjusted. The incidence rate used in Stroke 54,714.83 27,357.42 13,678.71 8,671.69 4,335.85 2,167.92 our analysis is likely to be conservative Type 2 1,486,671.30 793,335.65 371,667.82 7,410.83 3,705.42 1,852.71 for low-income individuals. diabetes Diet for stroke and osteoporosis. Totals 40,182,766.94 20,091,388.98 10,045,694.49 606,346.91 303,171.97 151,595.98

The Virginia benefit calculations were * Benefit-cost ratios for sensitivity analyses: incidence rate of 50% for osteoporosis/stroke (8.34/1.00); incidence rate based on the assumption that there was of 25% for osteoporosis/stroke (5.17/1.00); long-term benefit for 50% of population (all diseases) (7.33/1.00); a 100% relationship between diet and long-term benefit for 25% of population (all diseases) (3.67/1.00). disease for stroke and osteoporosis. The initial analysis used 100% (e.g., every- one will get this condition at some time) assumption that dietary changes could life-long. The Virginia study did two for the incidence rate, since there was reduce the risk of osteoporosis later in sensitivity analyses to correct for this: no actual data available. However, 100% life. However, there are a number of they assumed that 50% of the popula- is inaccurate for both of these diseases. other risk factors for osteoporosis that tion that showed optimal improvements The first sensitivity analysis done on the are not related to nutrition (e.g., genet- would maintain them over 5 years, and Virginia and California data used a 50% ics, hormones and smoking), and the that 25% of the population would main- incidence rate for both stroke and os- age at which risk prevention is reduced tain them for 5 years. In California, the teoporosis, then an additional incidence may be much younger. The EFNEP same two analyses were done: for 50% rate using 25% as an estimation. For intervention, however, would provide the benefit-cost ratio was 7.33 to 1.00, California, the revised benefit-cost ratio a positive benefit for reducing the risk and for 25% the ratio was 3.67 to 1.00 using 50% was 8.34 to 1.00, and using of osteoporosis. Since the sensitivity (table 4). 25% it was 5.17 to 1.00 (table 4). analysis for diet and osteoporosis (pre- As in all studies using data that is Osteoporosis benefits. The vious paragraph) reduced the incidence self-reported, and due to data collection California and Virginia studies both to 25%, an additional reduction did not procedures in the field, the potential for had the largest tangible benefit in their seem warranted. bias from confounding variables can osteoporosis calculations. Dietary fac- Long-term benefits. The long-term never be completely eliminated. tors play a role in delaying osteoporosis benefits of EFNEP have been studied Good public policy for 5 years. The EFNEP population is for at least 5 years following program mostly female (92%) and younger (25 to participation, but there are no studies Our initial benefit-cost ratio for 30 years old), which would justify the to date showing that the benefits are California was 14.67 to 1.00, and our ad-

190 CALIFORNIA AGRICULTURE • VOLUME 60, NUMBER 4 Based on our findings and those of the tive grant from the UC Division of Agriculture and Virginia researchers, nutrition education Natural Resources. appears to be a good economic invest- ment for the country, especially with an References Blackburn ML, Townsend MS, Kaiser LL, et al. 2006. underserved population likely to have Food behavior checklist effectively evaluates nutrition poor eating habits and higher risk for nu- education. Cal Ag 60:20­­­–4. merous chronic diseases and conditions. Contento I, Balch GI, Bronner YL, et al. 1995. Nutri- We believe this data provides strong evi- tion education for adults. J Nutr Ed 27(6):312­–28. Del Tredici AM, Joy AB, Omelich C, Laughlin SG. dence that nutrition education programs 1988. Evaluation study of the California Expanded Food are cost-effective and that continuing to and Nutrition Education Program: 24-hour food-recall fund them is a sound investment and data. J Amer Dietetic Assoc 88:185–90. [DHHS] US Department of Health and Human Ser- In addition to publishing nutrition guidelines good public policy. vices. 1990. Healthy People 2000: National Health Pro- such as the Food Guide Pyramid (updated in motion and Disease Prevention Objectives. Public Health 2005), USDA spends $255 million annually Service, Washington, DC. Pub #91-50212. on direct nutrition education for low-income Joy AB. 2004. Diet, shopping and food-safety skills consumers. The study provides evidence that A. Block Joy is Specialist and Director of the UC of food stamp clients improve with nutrition education. this spending is cost-effective. Food Stamp Nutrition Education program in Cali- Cal Ag 58:204–8. fornia, Department of Nutrition, UC Davis; and Lambur M, Rajgopal R, Lewis E, et al. 1999. Ap- plying cost-benefit analysis to nutrition education V. Pradhan (deceased) was Senior Statistician programs: Focus on the Virginia Expanded Food and and G. Goldman is Economist, Emeritus, Depart- Nutrition Education programs: Virginia Cooperative Ex- ditional sensitivity analyses provided a ment of Agricultural and Resource Economics, UC tension, Virginia Tech, Blacksberg, VA. range of results from 8.34 to 1.00, to 3.67 Berkeley. The authors thank Ruby Cox, Specialist, Merrill EO, Willis W. 1993. EFNEP Evaluation/Report- ing System User’s Guide, Version 3.01 (ERS-3). ERS Na- to 1.00. The bottom line is that for ev- Virginia Cooperative Extension, who consulted on the methodology of the analysis, and whose tional Committee, Cooperative Research, Education and Extension Service, USDA, Washington, DC. ery dollar spent on nutrition education insight, enthusiastic support and guidance were in California, at least $3.67 is saved on Murphy SP, Kaiser LL, Townsend MS, Allen LH. 2001. invaluable. We also acknowledge Barbara Suther- Evaluation of validity of items for a food behavior check- medical treatment costs. land, EFNEP Director, and the EFNEP Advisors who list. J Amer Dietetic Assoc 101:751–6. We believe that the benefit-cost ratios managed adult EFNEP in their respective counties Nierman LG. 1986. A Longitudinal Study of the Re- presented in our study are low. Several and collected the data used in this study: Mary tention of Foods and Nutrition Knowledge and Practices of Participants from Michigan Expanded Food and Nutri- decisions ensured that our estimates Blackburn (Alameda), Gloria Brown (San Francisco/ San Mateo), Ann Cotter (Orange), Susan Donahue tion Education Program. Doctoral dissertation. Michigan were conservative, including: (1) using State University, East Lansing, MI. (Butte), Nancy Feldman (Stanislaus), Mary Fujii stringent criteria to determine the opti- Rajgopal R, Cox RH, Lambur M, Lewis E. 2002. Cost- (Contra Costa), Chutima Gavanthor (Riverside), benefit analysis indicates the positive economic benefits mal nutrition behaviors associated with Margaret Johns (Kern), Cathi Lamp (Tulare), Con- of the Expanded Food and Nutrition Education Program the delayed onset of several chronic nie Lexion (San Bernardino), Anna Martin (San related to chronic disease prevention. J Nutr Educ Behavior diseases; (2) reducing the incidence rate Joaquin), Yvonne Nicholson (Sacramento), Gwen- 34:26–37. dolyn Stanford (San Diego), Jeanette Sutherlin Sassone PG, Schaeffer WA. 1978. Cost-benefit for osteoporosis and stroke; (3) using Analysis: A Handbook. New York: Acad Pr. 182 p. national averages for treatment costs (Fresno), Barbara Turner (Los Angeles) and Estella West (Santa Clara). In addition, we thank William [USDA/DHHS] US Department of Agriculture and instead of California’s higher costs; (4) US Department of Health and Human Services. 1995. Benford, Chris Hanson, Karrie Heneman and Karl Dietary Guidelines for Americans (4th ed.). Pub no. reducing the population that practiced Schroeder. This study was funded by a competi- IHG232. Washington, DC. the optimal behaviors to 25%, to adjust for the 5-year time frame of treatment delay; and (5) reporting benefit results Tribute: Vijay Pradhan for individuals only when these benefits wo of the authors would like to recognize would be expected to extend to other the outstanding contributions made by family members as well. TVijay Pradhan, who passed away after the Nutrition education is now widely study was completed. He implemented the acknowledged as an important com- statistical analyses for this study, looking at all ponent of the U.S. Department of the possible criteria and meticulously working Agriculture (USDA) budget. The cur- with multiple variables. Pradhan spent 2 years rent EFNEP budget is larger than it was on the design, development and implementation of the analyses for Cali- in 1998, more than $65 million nation- fornia, and he met with the other authors regularly to discuss the results. ally. At the same time, the Food Stamp His dedication to the completion of the study was a reflection of his char- Nutrition Education (FSNE) program acter; he was a devoted, concerned, caring and committed researcher. He also targets food stamp­–eligible clients was a self-motivated professional who took pride in his achievements. with a budget of more than $190 mil- Vijay Pradhan was a man of great integrity and a valuable friend. lion, for a total national USDA nutrition education budget of about $255 million.

http://CaliforniaAgriculture.ucop.edu • October–December 2006 191 Review Article t Radiofrequency power disinfects and disinfests food, soils and wastewater

by Manuel C. Lagunas-Solar, Nolan X. Zeng, Timothy K. Essert, Tin D. Truong and Cecilia Piña U.

Radiofrequency (RF) is an advanced telecommunication technology first invented in the early 1900s, which is in use today for wireless communica- tion worldwide. Because of its ability to penetrate and heat various materi- als, RF has the potential to disinfect and/or disinfest food, agricultural and environmental materials. However, research to validate this approach has been restricted by limited under- standing of how RF photons interact with materials, and by limited access UC Davis researchers have developed a prototype for automatic, computer- to and the high cost of its source controlled batch and conveyorized electronics. Since the early 1990s, we applications of radiofrequency power. The system can disinfect and/or disinfest fresh have conducted research at UC Davis produce, ready-to-eat foods, milk and juices, as well as agricultural products such as rice on continuous RF power applications and wastewater. using nonconventional RF systems and new operational concepts. the adverse environmental impacts of are the technologies most often used Laboratory tests have successfully agricultural practices; and expanding today for disinfection and disinfesta- demonstrated the effectiveness of RF global markets, which impose logisti- tion. New, noninvasive, user-friendly power to disinfect and/or disinfest cal demands on regional, national and and economically viable processing international trade. technologies are needed to meet evolv- fresh produce, rice, soils, agricultural To a large extent, food safety de- ing consumer expectations and trade wastewater, and other foods and pends on the use of adequate disinfec- standards. materials. Likewise, rapid pulses of tion and disinfestation techniques, A research project was established RF are lethal to arthropod pests and while quality is maintained by inte- at UC Davis to study, evaluate and grating multiple handling, packaging, demonstrate new radiofrequency (RF) may provide a nonthermal disinfesta- and storage and distribution practices. processing applications for food, agri- tion process for fresh, temperature- Disinfection is aimed at eliminating cultural and environmental materials. sensitive commodities, as well as a spoilage and pathogenic organisms We review laboratory-scale results for promising alternative to the fumigant to reduce storage losses and prevent several RF processes with potential food-related illnesses. The disinfection commercial applications, and provide methyl bromide. of nonfood agricultural commodities preliminary economic estimates for such as soils, feeds and waste materi- their installation and operation. Our major challenge in producing and als is also needed, either because they experimental methods are described distributing disease- and insect- are recycled or used in food produc- briefly and generally; the scientific Afree foods is the need to maintain their tion, or to protect the environment. and technical details of many of these sensory and nutritional attributes Disinfestation (control of insect pests) results have been or are being re- while minimizing the adverse impacts is needed mostly to comply with ported elsewhere. of treatment. This challenge emerged trade barriers aimed at preventing the How radiofrequency power works because of consumer attitudes and spread of nonnative arthropod pests. market expectations concerning the Pesticides, refrigeration, packaging RF is an advanced telecommunica- safety, quality and condition of foods; and modified atmosphere storage tion technology first invented in the

192 CALIFORNIA AGRICULTURE • VOLUME 60, NUMBER 4 RF thermal power formula

The thermal power induced by RF is given by the following formula: P = 55.61 × 10-14 E2 ν ε” (1) where P is the thermal power gen- erated (W/cm3); E is the electric field strength in (V/cm); ν is the RF frequency in (Hz); and ε” is the dielectric loss factor of the material (intrinsic property). The dielectric loss factor (ε”) largely depends on the Unprocessed (left) and processed (right) apple An Angoumois grain moth larva emerges material’s chemical composition and juice; the photo shows no microbial growth from a rice kernel. In laboratory experiments, on the RF-processed sample, even after both RF processing achieved 100% disinfestation is essentially the ease by which mol- samples were stored for 19 months at room of the moth with no effects on rice quality. ecules can be heated by an RF field. temperature. early 1900s, which is in use today for (those formed by the polarization of viruses and enzymes, as well as arthro- wireless communication worldwide. neutral molecules). These dipole mol- pod pests present in heat-tolerant com- Traditionally, RF energy refers to ecules are forced to reorient within the modities (such as dried fruits, grains, nonionizing electromagnetic radia- changing electronic field, which results nuts, seeds, wastewater and soils). tion with frequencies ranging from in movement or drifting that causes in- Unlike traditional surface heating, approximately 30 megahertz (MHz) ternal friction and creates thermal energy RF penetrates deeply into foods and (wavelength = 11 yards [10 meters]) (heat). The process is known as “RF ther- agricultural materials (see formula 1). to 300 MHz (wavelength = 1.1 yards mal processing” or simply “RF heating.” The surfaces of the treated materials [1 meter]). However, the U.S. Federal At certain frequencies or frequency are slightly colder because of radiation Communications Commission (FCC) bands, some foods and nonfood mate- losses, and the insides are heated homo- allows other frequencies to be utilized rials can be heated preferentially and for industrial, scientific and medical ap- faster, creating rapid thermal effects on plications (e.g., 13.56 and 27.12 MHz) pests but minimal interactions with the Definitions (Kasevich 1998). Our studies focused host material. This is due to the differ- on the use of lower frequencies outside ence in electrical conductivity between Dielectrics: material (isolator) that the FCC domain, ranging from 300 kilo- arthropod pests (high) and the host does not conduct electricity hertz (kHz) (wavelength = 1,094 yards commodity (low). This process is called Dipole: material having two equal [1,000 meters]) to 10 MHz (wavelength “selective or differential RF heating” but opposite charges or magnetic = 36.1 yards [33 meters]). Early tests and could provide an alternative disin- poles demonstrated the potential advantages festation process for thermally sensitive Gram-negative: not forming a color of lower frequencies, in terms of the fruit and vegetable products. In general, precipitate when treated with type and efficiency of RF interactions complex organisms such as arthropod alcohol (Gram’s staining method with different materials. Within the pests are more severely and easily af- for bacteria identification) lower frequency range, very high fected by heat. The higher response of Gram-positive: forming color when (> 80%) overall energy-use efficiencies pests and lesser sensitivity of host com- treated with Gram’s method are achievable with modern design and modities offer a window of opportunity engineering systems. These novel RF sys- for disinfestation with minimal or no tems can be manufactured and operated impact on the commodity (fig. 2). with significant savings, and increased The differential effect is generally less Abbreviations ruggedness and reliability as compared effective with microbial contaminants, cfu: colony-forming units with conventional RF systems. since microbes are significantly smaller Hz: hertz RF power is produced when electric- in mass and are usually well attached kV: kilovolts ity is applied to an RF generator whose to a much larger volume (and mass) kW: kilowatts signal is amplified and delivered to a host, thus being rapidly and effectively kWh: kilowatt hour parallel electrode system (RF cavity), cooled. For disinfection to occur, the MAP: Mycobacterium avium subsp. in which a selected material is placed microbe must reach lethal temperatures, paratuberculosis (fig. 1). Within the RF cavity, an oscillat- which are usually also deleterious to the MHz: megahertz ing electric field is created, and energy host commodity. However, RF heating RF: radiofrequency is transferred to the treated material induces the thermal inactivation of bio- W: watts through electronic-field interactions logical organisms (such as fungi, bacte- with dipole or induced dipole molecules ria, protozoa, parasites and nematodes),

http://CaliforniaAgriculture.ucop.edu • October–December 2006 193 geneously and at controllable rates. In effects can disinfect some biological chemical alternative to pesticides that general, RF heating eliminates surface materials, because it causes cellular merits further investigation. overheating, reducing thermal loads changes including modifications in Fresh fruits and vegetables and allowing a food’s quality and nutri- membrane potential, permeability and tional attributes to be maintained. structural properties (electroporation). Selective RF heating effects were sug- In the laboratory, pulsed RF has gested as a potential insect control for Types of RF treatments induced lethal biological effects on ar- grains several decades ago (Nelson and Capacitive heating. Common thropod pests without thermal effects Charity 1972), but no demonstrations materials such as water (a major on the host (Lagunas-Solar and Essert with fresh produce have been reported. component of most foods), and most 2004). However, for pulsed RF systems We exposed fresh fruits and veg- cellulose-based and plastic materi- to operate reliably with high pulse- etables individually and identically als used in packaging, are generally repetition rates (pulses per second), in a RF research cavity (840 watts [W] inactive or transparent to RF. Other further system design and engineering input power, 6 kilovolts per centimeter materials such as soils, wood, dried is needed. The potential for nonther- [kV/cm], 1 MHz, 4 minutes) and the foods, grains, and nut products absorb mal disinfestation of fresh produce in results were normalized for comparison RF readily and can be heated rapidly. conveyorized operations, with high purposes. Because RF heating effects RF photons easily penetrate these energy-use efficiency, promises a non- are mostly determined by the material’s commodities, allowing them to be pro- cessed in large masses while heating the material thoroughly. This process is known as “capacitive heating.” Conductive heating. Conductive heating is utilized for moist materials such as foods (fresh fruit, vegetables, juices), and agricultural (animal feeds, fishmeal) and environmental (soil, wastewater) materials. The electrical conductivity of these materials is high, promoting the interaction of electri- cal current in the material to gener- ate heat. Conductive heating can be optimized over a low-frequency band (e.g., 150 kHz to 5 MHz), with high energy absorption and energy-use ef- ficiencies (> 90%). When complex mix- Fig. 1. Components of an RF cavity. Permanent and induced dipoles within the cavity are tures of materials having different RF forced to realign to the changing electronic field, creating friction that generates thermal energy properties are processed, each compo- (heat). Molecules that are not permanent dipoles (transparent mass) and cannot be polarized are inactive or transparent within the RF cavity. Insect pests (conductive/absorbent mass) absorb RF, nent can be heated directly at differ- rapidly heating up and dying (disinfestation). ent rates and a threshold temperature reached to assure efficacy (Lagunas- Solar et al. 2003). Ultra-short pulses. RF energy can also be delivered in ultra-short pulses (micro to milliseconds) (µs to ms), creating very high peak (instant) RF power levels. When energy is delivered in ultra-short time pulses, biological targets such as insects and mites are subjected to instantaneous high thermal levels without compromising the host material. Pulsed RF generates mostly nonthermal effects in the host. During the pulse, intense electrical fields induce molecular polarization, which forms Fig. 2. Differential (selective) heating effects for samples individually treated with 840 W at 1 MHz for 4 minutes, with an electric field of 6 kV/cm. Temperatures were measured with dipoles; realignment of dipoles with the nonmetallic (alcohol) thermometers during RF processing; the change in temperature is shown. RF field becomes the major mechanism Similar results were obtained with fresh vegetables (asparagus, carrot, cucumber, eggplant, for energy transfer. These polarization garlic, onion, pepper, potato) (data not shown).

194 CALIFORNIA AGRICULTURE • VOLUME 60, NUMBER 4 TABLE 1. RF thermal disinfection of foods and nonfoods inoculated with human pathogens* chemical composition (see formula, Reduction levels page 193), powdered yeast was used to Sample material temperature E. coli O157:H7 S. thyphimurium M. paratuberculosis represent microbes (e.g., fungi, bacteria, °F (°C) ...... % ...... protozoa) while ants (Pogonomyrmex Ready-to-eat foods/ subdentatus) were chosen as surrogates Japanese bento box for insect and mite pests. We also stud- Cooked rice 201.2 (94) > 99.998 > 99.9993 n/a ied materials commonly used to pack- Salmon 158.0 (70) > 99.999 > 99.9993 n/a Shrimp 179.6 (82) > 99.9993 > 99.9995 n/a age fresh produce, such as foam board, Kamaboko† 161.6 (72) > 99.993 > 99.9993 n/a wood (Douglas fir, pine and redwood) Konnyaku‡ 165.2 (74) > 99.999 > 99.998 n/a and polyethylene. Potato salad 188.6 (87) > 99.999 > 99.998 n/a

Significantly different heating rates Orange juice 129.2 (54) > 99.999 > 99.994 > 9.8 and temperatures were observed be- 145.4 (63) > 99.9996 > 99.9993 > 99.9995 tween the commodities (lowest heating 161.6 (72) > 99.9996 > 99.9995 > 99.9994 rates), packaging materials (medium) Wastewater 113.0 (45) ~ 2 ~ 2 Not detectable and particularly model pest contami- 129.2 (54) ~ 9 ~ 99 ~ 2 nants (highest) (fig. 2). Minor heating 145.4 (63) > 99.9999 > 99.99999 > 99.9999 (~ 54°F [~ 12°C] maximum for cranber- 161.6 (72) > 99.9999 > 99.99999 > 99.9999 ries) was observed in fresh commodi- Fishmeal 140.0 (60) 99.97 99.93 n/a ties with a range of 39°F to 46°F (4°C to 149.0 (65) 99.998 99.992 n/a 8°C) for all others, indicating that a low 158.0 (70) > 99.9995 > 99.99993 n/a thermal load (temperature × time) was 176.0 (80) > 99.9995 > 99.99993 n/a 194.0 (90) > 99.9995 > 99.99993 n/a well tolerated by fresh commodities. 10 Increased temperatures were rapidly * Reduction levels expressed as percentages were obtained from measured log reduction levels. † Kamaboko is a food made from fish paste. lowered by heat losses reaching ambi- ‡ Konnyaku is a gelatinous, noncaloric food derived from a bulbous perennial herb. ent temperature (~ 72°F [~ 22°C]) in a few minutes. Insects reached a level lethal to all arthropod pests (> 194°F to evaluate the efficacy of RF processing 21 days and showed no colony forma- [90°C]). Yeasts heated faster with lower and confirm its economics and logistics. tion. Overall, the RF treatments caused levels of moisture (e.g., dried commod- no sensory changes, and only minor Ready-to-eat foods ities), suggesting that microbial disin- cosmetic damage occurred to fresh let- fection in moist products is unlikely. Processed or partially processed tuce used in some boxes as a decorative All plastic packaging materials heated foods often contain pathogens from nat- item (table 1). at levels similar to the fruits, while ural or humanmade sources (IFT 2002). A panel of Japanese scientists and wood materials heated faster, suggest- For effective disinfection, a threshold visitors (Ishida Ltd. Co., Kyoto, Japan) ing the potential use of RF processing temperature must be reached in all of conducted sensory tests and concluded for disinfestation. the food’s components. With conven- that there were no detectable differences Hot water immersion is currently a tional surface heating, the food surface in marketing quality between the RF- common practice to disinfest fresh com- must be overheated, since heat must be treated and untreated boxes. Clearly, RF modities. Selective RF heating conducted throughout its entire volume. processing through the package is pos- for dry thermal disinfestation appears We tested RF thermal disinfection in sible for many foods, reaching disinfec- possible and may provide an easily two types of freshly prepared Japanese tion levels without the deterioration of adapted alternative for conveyorized bento boxes (n = 30 each). Samples sensory and nutritional properties. operations with single fruits or fruit of each component were inoculated A conveyorized RF process could packages. If developed, it would be with Escherichia coli O157:H7 and/or disinfect ready-to-eat foods quickly, in especially useful for organic products. Salmonella thyphimurium, both important less than 1 minute. With proper design Challenges to developing this process food pathogens. of the RF cavity, multiple packages include measuring the RF properties of Closed plastic packages were treated could be processed continuously. Based fresh produce over a frequency band with the goal of reaching a minimum on operational costs (mostly electric useful to optimize process efficien- temperature of 158°F (70°C) in every power) and without capital cost amorti- cies and minimize uneven heating due component. After processing, the zation, the costs would be an estimated to the often-irregular shapes of fruits treated and control samples were as- $0.0034 per pound ($0.0074 per kg) or and vegetables. Our operational cost sayed using standard microbiology approximately $0.0037 per box. This RF estimates appear to be competitive at procedures at the UC Davis Dairy Food process would be applicable to similar $0.0024 per pound ($0.0054 Safety Laboratory in the School of packaged products for disinfection in per kg) and comparable to other RF pro- Veterinary Medicine. Treated samples a production line, such as grains, dried cesses. Prototyping for small-scale field in standard Petri dishes were also incu- foods, nut products, cereals, animal demonstrations will be needed bated using selective media for about feeds and fishmeal. Furthermore, disin-

http://CaliforniaAgriculture.ucop.edu • October–December 2006 195 fecting or disinfesting food products in the package would help prevent recon- tamination. Fresh fruit juices and liquid milk Fresh, unprocessed fruit juices are potential sources of human pathogens, particularly those containing several different natural, unprocessed (includ- ing organic) ingredients. Contamination can result from field conditions and handling as well as poor hygienic prac- tices during manufacturing. This is why Fig. 3. RF thermal disinfestation effects on naturally infested (Angoumois grain moth) paddy rice. Triplicate samples (155 grams each; 465 grams total) were treated at 95°F to 158°F (35°C to 70°C) juices, and particularly liquid milk, are using 100 W of 385 kHz RF photons, for 5 minutes. Emerging adults were observed for up to 100 heat-pasteurized, but the process can days to include overall emergence (survival) from eggs, larva and pupa stages, and compared to lower nutritional and sensory proper- controls (not treated). Emergence was controlled at 122°F to 158°F (50°C to 70°C). ties, mostly due to surface overheating. (Nonpasteurized fruit juices are avail- ing was reproducible and achieved total Angoumois grain moth (Sitotroga ce- able in some segments of the market.) A destruction of the pathogens, as shown realella), whose larvae and pupae live homogeneous and less energy-intense by the absence of colonies after thermal entirely inside the grain. Chemical fu- thermal process would improve and incubation (~ 7 days, 98.6°F [37°C]). migation with pesticides such as methyl preserve quality. Panels of industry experts determined bromide and phosphine is inefficient Samples (100 milliliters [mL]) of fresh that RF processing did not affect the because the dispensed gases cannot orange juice (Sunkist Growers), apple sensory properties of either smoothies break through naturally occurring air juice and cider (Martinelli, Apple-a-Day), or fruit juices. locks, preventing the fumigant from were innoculated with E. coli O157:H7 A conveyorized RF system for penetrating or diffusing inside the hol- and S. typhimurium at about 106 colony- fresh juices, smoothies and milk could lowed grain. forming-units per milliliter [cfu/mL] provide processing capabilities in the A well-insulated paddy-rice sample and treated with RF (200 W, 13.3 MHz, production line (on-line), and achieve holder (polyethylene and foam board) 1 kV/cm, 1 minute). Samples of liquid disinfection in-the-package in less than was designed and constructed to con- milk inoculated with Mycobacterium 1 minute, while meeting design and tain up to 170 grams of paddy rice and avium subsp. paratuberculosis (MAP), engineering challenges similar to those maintain a homogeneous temperature a bacterium believed to be resistant to of ready-to-eat foods. Operational costs, (± 1.8°F; ± 1°C) during processing. After conventional heat pasteurization, were mostly for electric power, were an esti- treatment, all samples (in triplicate) also RF processed. In addition, nearly mated $0.0031 per pound, or $0.0016 per were transferred to plastic containers a dozen freshly made unpasteurized gallon ($0.0060 per liter). RF systems for with secure lids containing small mesh fruit smoothies were prepared from raw disinfecting fruit juices could be built screens, which maintained moisture and fruits, processed with RF and tested for with a range of processing capacities, oxygen supply and allowed the surviv- differences in sensory properties. allowing its use by small operators or ing insects to grow. Samples were incu- Because of their mixed composition those demanding larger throughput. bated at 82.4°F to 86°F (28°C to 30°C) (including apple, berries, carrots and or- and 35% to 43% relative humidity for up Disinfestation of paddy rice anges), these smoothies and fruit juices to 100 days (approximately three to four heated rapidly and homogeneously in We tested commercial samples of complete insect life cycles). Every 2 to 3 less than 1 minute (table 1). RF process- paddy rice naturally infested with days, the surviving or emerging adult moth populations were determined and TABLE 2. Quality attributes of RF-processed paddy rice* compared to the controls. After each ob- servation, all live adults were removed Test 1 Test 2 Test 3 to allow other life cycles (from eggs, Quality attributes Controls (122°F [50°C]) (140°F [60°C]) (158°F [70°C]) larvae, pupae) to emerge and be as- ...... % ...... sayed. In addition, the quality attributes Moisture 13.5 ± 0.1 13.5 ± 0.1 13.5 ± 0.1 13.5 ± 0.1 of all treated and control samples were Whole kernel 79.3 ± 1.1 81.1 ± 7.9 78.3 ± 0.5 77.9 ± 0.8 analyzed by the California rice industry Total rice 68.1 ± 0.3 68.3 ± 0.1 68.2 ± 0.1 68.0 ± 0.1 Dockage 16.9 ± 4.8 11.7 ± 1.0 12.4 ± 1.6 13.2 ± 1.7 using standard analysis methods. Brown rice 81.1 ± 0.4 81.4 ± 0.2 81.3 ± 0.2 81.3 ± 0.1 Disinfestation levels reached 100% Whiteness 44.2 ± 0.2 44.1 ± 0.2 44.2 ± 0.2 44.3 ± 0.3 control of all Angoumois grain moth * Mean values and standard deviation for triplicate measurements each with 2.2 lbs (1 kg) samples. life cycles (fig. 3), and the milling and quality of the rice were not affected

196 CALIFORNIA AGRICULTURE • VOLUME 60, NUMBER 4 Treating agricultural wastewater We investigated the disinfection of agricultural wastes using wastewater from dairy and animal farms in Tulare County. Wastewater samples (100 mL) were inoculated with S. thyphimurium, E. coli O157:H7 and MAP, then treated with temperatures between 113°F and 194°F (45°C to 90°C). Standard biological procedures (diluting, plating, incubating and counting) with appropriate selective media were used to assay in duplicate all treated and control samples. Measured disinfection effects (percent- age reduction) included corrections for the assay’s detection limit (400 cfu/mL), while the extrapolated values were cal- culated from the initial inoculum levels Researchers are investigating the application of RF power to disinfest and disinfect carpets. Specific frequencies were used to rapidly heat insects; ants since no colony growth was detected were killed in a few seconds. after 21 days of incubation (table 1). In addition, microorganisms present in the (table 2). RF processing appears to be mination of various types of seed, in original wastewater samples — including a promising alternative to fumigation collaboration with private industry gram-positive (data not shown) and other for rice and similar dried commodities (Campbell Soup Research Institute, gram-negative bacteria — were reduced such as seeds, nuts and dried fruits. Davis, Calif.). The preliminary data in- by about 99%. RF processing also effec- Further research to optimize this pro- dicates that overall germination is not tively controlled MAP bacteria, which cess and evaluate sublethal effects on affected, but its rate (vigor) and timing appears to resist conventional heat pas- infesting insects is under way with (synchronization) are greatly improved teurization at similar thermal levels. beetles and moths. at thermal loads (temperature and RF processing could be technically An RF conveyorized system could time) capable of simultaneous disinfec- and economically competitive with operate commercially with large pro- tion and disinfestation. The application chemical (chlorination, ozonation) and cessing capacity, because rice and other of RF to process plant seeds would UV processing to disinfect wastewater grains are handled at rates of many tons decrease the risk of recontamination in in concentrated animal feeding opera- per hour. The logistics of such opera- packaged seed products. Batch or con- tions (Lagunas-Solar et al. 2005). The tions must be further evaluated, but veyorized operations are also feasible. operational costs for wastewater disin- they could offer an effective nonchemi- The operational cost of this treatment fection at a rate of 1 ton per hour would cal alternative to fumigants at opera- is estimated at $2.20 per ton. Because be about $6 per ton. Lacking in chemical tional costs of $2.20 per ton. seeds are not a high-volume commod- residues, treated wastewater could be ity, less-expensive RF systems with rela- recycled as animal feed or for soil fer- Plant seed treatment tively low power capacity can be used. tilization without having detrimental RF processing can disinfect, disin- fest and induce favorable biological activation effects (such as increased TABLE 3. Economic estimates for RF processing at 2,205 pounds/hr (1,000 kg/hr)* rate, vigor, and synchronization of germination) in plant seeds without Economic cost factors Commodity: objectives RF power Cost per pound (kg) affecting their ability to germinate. We demonstrated RF disinfection with vari- kW $ ous seeds including tomato, carrot, pep- Fresh produce: disinfection and disinfestation 48.1 0.0024 (0.0054) Processed foods: disinfection 66.7 0.0034 (0.0074) per, cantaloupe, peas and cauliflower. Fruit juices: disinfection 61.7 0.0031 (0.0069) In contrast to hot-water treatments (~ Liquid milk: disinfection 60.2 0.0030 (0.0067) 131°F [55°C] for 15 minutes), RF pro- Paddy rice: disinfection and disinfestation 22.2 0.0011 (0.0025) cessing allows higher temperatures Plant seeds: disinfection and disinfestation 22.7 0.0011 (0.0025) (activation of germination) because heat distribution is better and Wastewater: disinfection 60.2 0.0030 (0.0067) Fishmeal: disinfection 32.4 0.0016 (0.0036) occurs quickly, in less than 1 minute. We demonstrated and validated * Electricity at $0.10/kWh; overall energy-use efficiency of 81%. the kinetics of activation in the ger-

http://CaliforniaAgriculture.ucop.edu • October–December 2006 197 effects on soil chemistry or transferring thermal energy distribution and volume tional costs than methyl bromide. To im- known plant and human pathogens heating properties, RF allows a lower prove commercial applicability, newly (Lagunas-Solar et al. 2005). thermal load (~ 662°F per minute [350°C designed and constructed pulsed RF per minute]) than conventional [surface] systems are being tested for disinfesta- Fishmeal disinfection heating (~ 2,462°F to 3,272°F per minute tion and soil processing (Lagunas-Solar Fishmeal is used worldwide as a [1,350°C to 1,800°C per minute]). The and Essert 2004). high-protein feed ingredient in aqua- process has been licensed to private in- Nematode control. We demon- culture and animal/poultry production. dustry, which is evaluating its commer- strated nematode (Panagrellus and However, it must be disinfected because cial deployment. Operational costs are Cephalobus sp.) control in agricultural contamination with microbial patho- estimated at about $3.20 per ton (table soils in the laboratory using continu- gens can occur during storage, trans- 3), much lower than chemical alterna- ous and pulsed RF power applied with portation and distribution. The market tives (about $15 to $20 per ton) or con- high energy-use efficiency (> 90%). demands Salmonella-free fishmeal to ventional heating (about $25 per ton). Previously, the lethal temperature for avoid recontamination in the animal- nematodes was experimentally deter- Agricultural soils and pests human food cycle. mined as greater than 131°F (55°C), We investigated RF disinfection at If implemented successfully, RF tech- requiring that soils be heated to in- temperatures ranging from 122°F to niques could provide a rapid, residue- crease their ambient temperature by 194°F (50°C to 90°C) with commercial free and cost-effective alternative to 86°F to 104°F (30°C to 40°C). Control fishmeal samples (Corpesca, Santiago, soil fumigation with methyl bromide, of Fusarium sp. was also demonstrated Chile) inoculated with Salmonella sp. which is being phased out globally due during the same experimentation. and E. coli O157:H7. Between 158°F and to its contribution to ozone depletion. A newly designed, pulsed RF system 194°F (70°C to 90°C), no colonies were Research on soil disinfection with RF was also tested. The system is capable of detected in plated samples after 21 days has focused on containerized and nurs- generating about 200 MW pulses with of observation (table 1), with greater ery soils (Lagunas-Solar et al. 2005b). kHz-level repetition rates using 10 kW than 99.999% reduction in microbial Microwaves, hot water and steam ap- of electric power. Nematode-infested soil pathogens compared to the controls. plications are also being investigated as samples were treated for 5 to 15 minutes Simultaneously, the natural flora pres- alternatives to methyl bromide to disin- using 400 pulses per second. The soil ent was reduced by more than 99.9%, fest and sterilize soils. temperature increased from 73.4°F to adding further quality assurances to the Because soils are good dielectrics and 113°F (23°C to 45°C), indicating that part treated product. RF can easily penetrate them, RF power of the pulsed RF energy was converted By industry standards, the fishmeal offers many advantages; but the use of to heat. After the treatment, nematodes maintained its quality (e.g., protein and conventional, continuous RF power sys- in treated soil samples were extracted lipid contents, moisture) and in vivo tems for soils is currently limited due to into water overnight and then collected digestibility (Lagunas-Solar et al. 2005a) high startup (capitalization), logistical using the Tyler standard screen filter. (fig. 4). In addition, due to improved barriers and somewhat higher opera- Nematode mortality was more than 90% compared with the control. However, 90% control can also be reached with continuous RF power when heating reaches more than 131°F (55°C). Further studies are under way. Other research. Pulsed (nonthermal) RF experiments with adult fruit flies (Drosophila melanogaster) and all life cycles of mites (Amblyseius cucumeris, Tyrophagus putrescientae) in fresh table grapes showed 100% mortality in less than 10 seconds, while inducing negli- gible temperature changes in the grape (< 1.8°F; < 1°C) (unpublished results). This is due to the high electrical con- ductivity of arthropods, which allows enhanced, pulsed, RF-induced electric effects at the expense of thermal effects. Fig. 4. Percentage of commercially packaged fishmeal samples that retained quality attributes Finally, applications of continuous (protein and lipid content, moisture, and protein digestibility) following thermal disinfection at different processing temperatures. Disinfection reached 99.9% at 140°F (60°C) and 100% at 158°F RF power are being developed for the (70°C) and above. Animal feeding studies were conducted at the Norwegian Institute of Fisheries home, particularly to disinfest carpet and Aquaculture Research, Fyllingsdalen, Norway. and garden pests. This application

198 CALIFORNIA AGRICULTURE • VOLUME 60, NUMBER 4 oritize research and development into problems, as defined by industry inter- ests, regulatory demands, and market- ing conditions and opportunities.

Ultra-short pulses of RF power were used to treat adult flies on the surface of a ‘Red Flame’ seedless grape berry. No heating effects were detected in the grape, but microscopic observations indicated acute dryness and charring of the fly’s integument, wing ruptures M.C. Lagunas-Solar is Research Chemist, N.X. Zeng and deformed abdomens, all effects related to rapid heating. is Research Physicist, T.K. Essert is Principal Elec- tronic Technician, T.D. Truong is Research Assis- utilizes specific frequencies or narrow is competitive with conventional heat tant, and C. Piña U. is Staff Research Associate, all at Crocker Nuclear Laboratory, UC Davis. (Truong bands to induce the rapid heating of in- processing (10% to 15% energy-use ef- is also Graduate Student, Department of Chemical sects and mites in RF-transparent (non- ficiencies). No regulatory intervention Engineering and Material Sciences, UC Davis.) The absorbing) materials such as carpets or is expected because RF processes are authors thank the following UC Davis profession- other home materials, thus providing based on either thermal inactivation als for their expert guidance and assistance during rapid, nonchemical pest control. (continuous RF), or a combination of collaborative experimentation in their respective thermal inactivation with electronic- fields: James S. Cullor and Wayne L. Smith, School Economics of RF processing field effects (pulsed RF). While RF of Veterinary Medicine; James D. MacDonald and Linda Bolkan, Department of Plant Pathology; Our preliminary cost estimates for processes would operate outside the al- Edward P. Caswell-Chen and Steven A. Nadler, RF processing included the specific heat lowed FCC frequencies, their operation Department of Nematology; and Zhongli Pan, De- of materials; the thermal load require- is fully shielded to prevent RF emissions partment of Biological and Agricultural Engineer- ments to achieve effects, the cost of to personnel and the environment. ing. This work summarizes research conducted electricity ($0.10/kWh) and process ef- Despite experimental demonstrations under the sponsorship of Hortifrut S.A. (Santiago, ficiencies (table 3). They did not include of its effectiveness and low operational Chile), the California Dairy Research Foundation and the California Rice Research Board. the cost of designing and purchasing costs, there is no evidence that RF is RF equipment for specific applications. currently being used commercially to These estimates are based on the opera- disinfect or disinfest foods or nonfood References tion of UC Davis laboratory prototype products in the United States or glob- [IFT] Institute of Food Technologists. 2002. Emerging Microbiological Food Safety Issues: Impli- systems, which achieve high energy-use ally, nor have any such uses been pub- cations for Control in the 21st Century; Microbial efficiencies (electric-to-RF > 90% and lished in the scientific literature. This is Ecology and Foodborne Disease. http://members. RF-to-thermal > 90%). believed to be due to the lack of under- ift.org/NR/rdonlyres/6ACDE698-6CF1-48EA-9E39- B8915D57B56F/0/ecology.pdf. Although commercial systems cur- standing on the mechanisms of interac- Kasevich RS. 1998. Understand the potential of rently cost from $1,500 to $2,500 per kW, tion between RF photons with foods radiofrequency energy. Chem Eng Progress:78­–81. we were unable to obtain estimates for and nonfood materials, especially in the Lagunas-Solar MC, Essert TK. 2004. Non-Thermal Disinfestation Method with Pulsed Radiofrequency newly designed, lower-frequency RF low-frequency range (< 100 MHz), cou- Power Systems. US Patent Application Serial No. systems. However, we anticipate consid- pled with the challenges associated with 10/900,990. erable cost reductions (10 to 20 times) to designing and operating RF research Lagunas-Solar MC, Cullor JS, Zeng NX, et al. 2005. Disinfection of dairy and animal farm wastewater with about $100 to $200 per kW for manu- prototype systems. Nevertheless, a few radiofrequency power. J Dairy Sci 88(11):412­­0–31. facturing systems using solid-state commercial food (baking and drying) Lagunas-Solar MC, Zeng NX, Essert TK. 2003. electronics and lower single or narrow and nonfood (wood drying and disin- Method for inhibiting pathogenic and spoilage organisms non-FCC-approved frequency bands festation) applications are known, and in products. US Patent #6,638,475. Oct. 28, 2003. Lagunas-Solar MC, Zeng NX, Essert TK, et al. 2005a. (< 10 MHz). These new systems are medical research is being conducted Disinfection of fishmeal with radiofrequency heating for currently being designed and tested by for thermal tumor ablation treatments. improved quality and energy efficiency. J Sci Food Agr our laboratory. However, these applications have lim- 85:2273–80. Lagunas-Solar MC, Zeng NX, Essert TK, et al. 2005b. Pulsed RF systems offer the potential ited impact and have not reached wide Thermal disinfection of soils with radiofrequency power. for even larger energy savings, since acceptance, despite the absence of regu- 2005 Annual Int. Conference on Methyl Bromide Alter- intermittent pulses only demand about latory barriers. natives and Emissions Reductions, p. 38-1 to 38-4. San Diego, CA, Oct 31­–Nov 4, 2005. 10% of energy consumption. If high While RF has tremendous potential, Nelson SO, Charity LF. 1972. Frequency dependence energy efficiency and reduced con- commercial input from the targeted of energy absorption by insects and grains. Trans ASAE sumption are achieved, RF processing industries is needed to focus and pri- 15(6):1099–102.

http://CaliforniaAgriculture.ucop.edu • October–December 2006 199 research Article t Impact of environmental factors on fish distribution assessed in rangeland streams

by Lisa C. Thompson, Larry Forero, Yukako Sado and Kenneth W. Tate

We sampled fish in pools located on tributaries of Cow Creek in the northern Sacramento Valley, and related fish distribution and habitat use to environmental factors across the 2003 agricultural growing season. This rangeland watershed experiences extensive livestock use, and many landowners divert stream water for pasture irrigation. Our goal was to provide landowners and managers with current baseline information about the conditions in Habitat and land-use factors were studied which fish were found. Our results in a northern Sacramento Valley rangeland, provide a basis for the development in order to better understand the life cycle needs of salmon and other fish. Above, Lisa and comparison of irrigation best Thompson, UC Davis assistant specialist in management practices that may Cooperative Extension, holds a water sample collected from a pool in Old Cow Creek. Left, a improve conditions for native fish in temperature logger is inserted into protective PVC case with tethering cable. rangeland streams.

ish population declines and habitat temperature TMDL is being developed. from their nests. Alternatively, exotic alteration across Northern California Sediment and stream temperature are species — such as the bottom-foragers watershedsF have resulted in the listing both important habitat variables, but a common carp, fathead minnows and cat- of numerous species under both the state more holistic approach using a suite of fishes — may invade degraded streams. and federal Endangered Species Acts, interacting habitat and land-use factors There is little specific information on the and instigated the implementation of to- would provide listed fish species with responses of fish in California rangeland tal maximum daily load (TMDL) a greater chance of recovery. However, streams to land-use practices. water-quality strategies for impaired wa- few studies have concurrently examined Thermal refugia ter bodies. These events have highlighted habitat, water quality, fish and agricul- the importance of species recovery and tural practices. This study of Cow Creek The temperature requirements habitat restoration efforts, while increas- in the northern Sacramento Valley is the of salmonids vary with life history ing scrutiny of the potential impacts of first step in such a project for Northern stage and by species (Mihursky and agricultural practices on cold-water fish- California watersheds. Research on how Kennedy 1967). Different species have eries. Agricultural practices of concern in these factors interact will provide vital different optimal temperatures for mi- Northern California watersheds include information for agricultural manage- gration, spawning, egg incubation and irrigation diversions, irrigation return ment, stream habitat restoration and juvenile growth, and also different flows, riparian grazing and roads. fisheries recovery planning. lethal-temperature criteria (Beschta et Restoration and regulatory efforts Past land use can be related to the al. 1987; Thompson and Larsen 2004). often adopt a single-variable approach fish communities currently seen in Since fish are poikilothermic (cold- to species recovery. For example, a streams (Harding at al. 1998). There is a blooded), their metabolic rate and sediment TMDL is being developed tendency in some agricultural streams food needs increase with higher water to aid the recovery of coho salmon for fish species that depend on stream- temperatures. If adequate food is not (Oncorhynchus kisutch) in the Garcia beds for spawning or foraging to be re- available, fish will lose weight and River. To meet the same goal on the placed by species that occupy the water eventually die, even if temperatures south fork of the Trinity River, a stream- column or are able to clean sediments do not reach lethal levels.

200 CALIFORNIA AGRICULTURE • VOLUME 60, NUMBER 4 A more holistic approach using a suite of interacting trout were assumed to be experiencing habitat and land-use factors would provide listed fish the observed maximum temperatures. species with a greater chance of recovery. In some cases the downstream limit of trout distribution in these Idaho streams appeared to be limited by un- Fish in some Northern California bris, overhanging banks or rocks. This suitable habitat (e.g., no water or very streams are known to use thermally may result in increased mortality, due to shallow braided stream channel), but in stratified pools as cold-water refu- low growth rates and increased vulnera- one case flows and channel conditions gia during hot weather (Nielsen at bility to size-selective predation, or due appeared to be suitable, so trout may al. 1994). However, fish may face a to increased predation as fish spend less have been limited by temperatures in tradeoff between temperature and dis- time in refuges in order to feed (Walters excess of 84.2°F. It is not known whether solved oxygen in their use of pools. and Juanes 1993). The distribution of California native fish such as rainbow Stratified pools may have cooler tem- fish relative to thermal refugia, and the trout are able to tolerate similar condi- peratures at the bottom, but also lower timing of this use are undocumented tions in Central Valley streams. Little is dissolved-oxygen concentrations with for most California rangeland streams, known about the specific temperature increased depth (Elliott 2000). In addition which hampers management efforts to tolerances of rainbow trout/steelhead to pools, fish may seek groundwater-fed improve fish habitat. and salmon in the Central Valley. locations in order to experience cooler The upper lethal temperature limit (Rainbow trout and steelhead are the temperatures (Ebersole et al. 2003), but for rainbow trout (O. mykiss) is gener- same species, O. mykiss; however, steel- since groundwater may be lower in ally thought to be about 75.4°F (Beschta head migrate to the ocean, while rain- dissolved oxygen than stream water, et al. 1987). However, Zoellick (1999) bow trout spend their entire life cycle in fish may still experience stress. found that redband rainbow trout freshwater.) Alternatively, levels of dissolved (O. mykiss gairdneri) in southwestern Myrick and Cech (2004) did a com- oxygen in groundwater may be higher Idaho occupied stream reaches with prehensive review of the published than those of stream water depending maximum daily water temperatures of information available for the effects of on how much dissolved oxygen was in 80.1°F to 84.2°F, and tolerated tempera- temperature on juvenile Central Valley the groundwater when it first entered tures above 78.8°F for up to 4.4 hours. Chinook salmon (O. tshawytcha) and the ground, since dissolved oxygen Flows decreased to 0 cubic feet per steelhead. They looked at survival of tends to change little once water goes second (cfs) in some streams, and some eggs, alevin and juveniles, as well as underground. In their search for ap- of the streams subsided underground. juvenile growth rates, and reported that propriate temperature and dissolved- Despite the low or nonexistent flows, the number of published studies was oxygen levels, fish may locate farther Zoellick observed that pools at the “surprisingly low.” They reported that from optimal feeding sites, or away lower ends of the occupied reaches did survival of eggs declined above 54°F from refuges such as large woody de- not show temperature stratification, so for Chinook salmon, but they found no published, peer-reviewed studies on the Glossary effects of temperature on Central Valley steelhead eggs and alevins. The thermal Alevin: A fish that has just hatched from the egg. In the case of salmon and tolerance of juvenile Chinook to chronic trout, the alevin lives in the gravel at the bottom of the stream until it ab- elevated temperatures is believed to sorbs its yolk sac and is ready to emerge and begin foraging. be approximately 75°F, but no studies Assemblage: A group of species commonly found together in the same habi- of this sort have been conducted on tat area, such as a stream, lake or estuary. Species within an assemblage Central Valley steelhead. may share some habitat preferences, such as water temperature or velocity. Furthermore, no published data Julian day: The day of the year, from 1 to 365 (366 in a leap year). was available regarding the thermal Rainbow trout/steelhead: Fish of the species Oncorhynchus mykiss. Rainbow tolerance of Central Valley Chinook to trout live their entire life cycle in freshwater. Steelhead are the anadro- acute temperature elevation, which is mous form; they migrate to the ocean to feed and grow, then return to dependent on the temperature to which freshwater to spawn. a fish is first acclimatized, and such Refuge: Location (e.g., under a log, rock or undercut bank) where fish can peer-reviewed data was also unavail- escape from the risk of mortality due to predators such as birds, otters or able for Central Valley steelhead. The larger fish. growth rate of juvenile Chinook salmon Refugia: Isolated location in which fish are able to survive during times when in the studies reviewed was optimal most of the surrounding habitat has become unsuitable (e.g., due to high from 63°F to 68°F, with slower growth water temperatures). outside of this range. Myrick and Cech Thermal stratification: A gradation of temperature, warm to cool, from top to (2005) studied the growth rates of juve- bottom in a pool or lake, that occurs during summer months. Increases in nile steelhead obtained from Nimbus flow or wind may mix the water to an even temperature throughout, dis- State Fish Hatchery at 52°F, 59°F and rupting the stratification. 66°F. They found that the steelhead growth rates were highest at 66°F, but

http://CaliforniaAgriculture.ucop.edu • October–December 2006 201 Fig. 1. Sampling sites on Old Cow Creek (OC-1, OC-2), South Cow Creek (SC-1, SC-2, SC-3) Top, cows graze near Cow Creek; above, trout and the mainstem of Cow Creek (CC-1). in a pool in Old Cow Creek in midsummer. noted that the study did not test higher of native populations, particularly in holders can methodically assess streams temperatures. environments that have been modified of concern, identify priority stream to be warmer, slower flowing and more reaches, examine current management Stream biodiversity and exotics nutrient-rich than normal (Moyle 2002). practices potentially affecting priority Historically, native fish in Central The modification of stream water flow reaches, and develop first approxima- Valley streams were distributed in a series and habitat may affect the interaction tions of BMPs to reduce impacts. of assemblages according to elevation between native and nonnative fish spe- We utilized Cow Creek in Shasta (Moyle et al. 1998). Cold-water fish such cies. Moyle et al. (2003) assessed fish County as an example of methodologies as rainbow trout, riffle sculpin (Cottus assemblages in the Cosumnes River to develop region-specific information. gulosus) and speckled dace (Rhinichthys os- and found that most nonnative spe- We attempted to determine how differ- culus) were found at mountain elevations cies were in low-elevation, agricultural ent fish species were distributed along (above approximately 1,800 feet) (Moyle sections of the river, whereas native the length of the creeks from higher to 2002), and down into the upper foothills rainbow trout were mainly at higher lower elevation; whether preferred na- (approximately 1,200 feet). Fish such as elevation, higher gradient segments of tive species such as rainbow trout use California roach (Lavinia symmetricus) the river. Apparently, nonnative species pools; whether fish distribution and the and hardhead minnow (Mylopharadon were intolerant of cooler waters with use of pools change across the summer; conocephalus) lived at foothill elevations faster flows. Moyle et al. (2003) sug- whether pools stratify in terms of tem- (approximately 300 to 1,800 feet). Fish gested that the management of water perature and/or dissolved-oxygen con- such as Sacramento suckers (Catostomus to provide increased flows and cooler centrations; and how fish distribution occidentalis) and Sacramento pikemin- water would benefit native species. relates to environmental factors such as now (Ptychocheilus grandis) appeared in temperature, flow, elevation and pool Developing stream-specific data the foothills, and were also present on the characteristics. valley floor (approximately 0 to 300 feet). The development and implementa- Location and data collection. We Species tolerant of warmer water, such as tion of successful best management chose six pools in the Cow Creek drain- Sacramento perch (Archoplites interruptus) practices (BMPs) to alleviate rangeland age, at a range of elevations representa- and splittail (Pogonichthys macrolepidotus) and irrigated pasture impacts on native tive of rangeland habitat in this part were present on the valley floor. fish and their habitat must be based of the Sacramento Valley (fig. 1). We The introduction of nonnative fish upon a clear understanding of: (1) cur- sampled temperatures at each pool con- species is a widespread phenom- rent habitat and fish assemblage condi- tinuously from May through November, enon across the United States (Rahel tions; (2) how environmental factors by placing two temperature loggers 2000), including California streams. such as temperature determine fish dis- (Onset Optic Stowaway) near the bot- Introduced fish species in California tribution and habitat use; and (3) how tom of each pool, and one temperature tend to prey upon or compete with na- management is modifying environmen- logger in the air (e.g., attached to a tree tive species, frequently to the detriment tal factors. With this information, stake- near the pool). We measured maximum

202 CALIFORNIA AGRICULTURE • VOLUME 60, NUMBER 4 depths to estimate the cover fish would several months, there is a potential for tive fish such as salmonids (Bjornn and find in a given pool, and length and codependence between observations Reiser 1991); but this was expected, width to allow estimates of pool area made at each site at different times. The since we took our measurements at and fish density per pool. (Structures codependence introduced by repeated midday when aquatic plants produce such as overhanging banks, boulders, measurements of a single site through oxygen. It is probable that dissolved- logs and overhanging branches may time can be efficiently addressed using a oxygen levels declined overnight, and also provide cover, but the measure- linear mixed-effects analysis. were lowest just before dawn. An ad- ment of these stream features was out- equate supply of dissolved oxygen is Pool characteristics side the scope of this study.) important to fish, especially salmonids We defined the upstream bound- The Cow Creek pools we studied such as salmon and trout, during all ary of a pool as the point at which ranged in elevation from 400 to stages of life (Bjornn and Reiser 1991; water cascaded in from a higher point, 1,718 feet (table 1), and in surface area Brown 1985). The swimming ability or where ripples ceased and the water from 165 to 854 square yards. Maximum of adult salmonids has been shown to became flat. The downstream end of the pool depth ranged from 3 to 7 feet, decline at dissolved-oxygen levels less pool was defined as the point at which and usually occurred at the upstream than 6.5 to 7.0 ppm. For spawning fish, riffles started again. We sampled water end of the pool. Flows varied between dissolved-oxygen levels should reach flow, temperature and fish, monthly from pools and across the season, with lower at least 80% saturation with tempo- June to October. Flow was measured at elevation pools tending to have higher rary levels no lower than 5.0 ppm. The the downstream end of the pool, using flows. None of the pools were stratified growth rate of juvenile salmonids de- a handheld flowmeter (Global Water during our visits. Temperatures at the clines when dissolved-oxygen levels are FP101). During each monthly visit top and bottom of given pools showed below 5.0 ppm, and death occurs at less we used a handheld temperature and differences of 0.10°F or less. Dissolved- than 1.0 to 2.0 ppm of dissolved oxygen. dissolved-oxygen probe (YSI 550A) to oxygen concentrations were also similar Fish distribution in Cow Creek measure these variables at the surface, at the top and bottom of each pool, giv- middepth and bottom of the pools. We ing further evidence that the pools were We observed fish in all pools from did snorkel counts to estimate fish spe- well mixed. June to October 2003 (fig. 2a). The na- cies present, density and age class. One Dissolved-oxygen levels observed tive fish species observed between snorkeler moved slowly through the across the season ranged from 7.52 to the months of June and October were: pool, first working upstream along one 10.79 parts per million (ppm), well California roach, Chinook salmon, bank, then floating down the middle, within the preferred ranged for sensi- rainbow trout/steelhead, Sacramento then upstream along the other bank and again down the middle. Fish observed during the second pass down the mid- dle were assumed to be repeat counts, so they were only counted if a fish of a given species and size had not been seen in the pool up to that point. Data analysis. We analyzed our data graphically and with regression- based linear mixed-effects analysis (S-Plus version 6.1 software; Pinheiro and Bates 2000), in order to relate fish distribution and habitat use to environ- mental factors. Seven individual linear mixed-effects analyses were conducted to determine if there were statistically significant (P < 0.05) differences for the water temperature, pool depth and area, flow, and elevation of pools occupied by trout compared to pools not occupied by trout, between June and October. Linear mixed-effects analysis has dis- tinct advantages for repeated-measures field studies such as this one, and has been used successfully to analyze other California stream data sets (Tate, Lile et al. 2005). Because we repeatedly Fig. 2. Total number of (A) fish and (B) rainbow trout observed at six pools sampled each site over the course of in tributaries of Cow Creek, 2003.

http://CaliforniaAgriculture.ucop.edu • October–December 2006 203 TABLE 1. Characteristics of study pools in Cow Creek watershed pikeminnow, Sacramento sucker and Maximum Flow Temp. speckled dace. Nonnative species ob- Pool Elevation Area depth* min.-max. stratification† trout refugia‡ served were: green sunfish (Lepomis ft yd2 ft cfs °F cyanellus), largemouth bass (Micropterus OC-1 1,331 339 4.0 4 –155 -0.10 Yes salmoides) and smallmouth bass OC-2 568 854 7.0 21 –170 -0.01 No (Micropterus dolomieu). Chinook salmon SC-1 1,718 165 4.5 29 –141 -0.03 Yes SC-2 1,053 227 4.5 11 –57 -0.08 Yes were observed only in the fall, during SC-3 614 685 3.0 32 –136 -0.10 No their upstream spawning migration, CC-1 402 303 6.0 36 – 356 -0.03 No when most irrigation diversions had * Maximum depth observed at lowest stream-flow conditions during study period. † Temperature at bottom of pool minus temperature at top (ºF). ceased. Since rainbow trout/steelhead ‡ At least one trout was observed in the pool during midsummer observation. are of great interest to fishery managers, and potentially of concern to landowners who use stream water for irrigation, we (P = 0.058) for the period from June Pools with trout in midsummer were focused on rainbow trout/steelhead, and to October. Young-of-the-year trout cooler than pools without trout (fig. 3a, grouped the rest of the species together. emerge from the gravel in early spring, table 2). For example, the period maxi- The declines in fish numbers be- and thereafter total trout densities may mum temperature for August in pools tween June and July reflect the mortality be expected to decline due to mortality. without trout in midsummer was over of large numbers of young-of-the-year The increase in trout density in higher 90°F, while in pools with trout it was fish, likely due to predation by birds elevation pools may indicate that trout well below 80°F. This corresponds with or aquatic predators (fig. 2a). Rainbow are moving upstream toward cooler the temperature tolerance value of 75°F trout/steelhead were observed in at waters, or out of riffles and into pools for juvenile Chinook salmon. The pe- least one pool in each of the sampling in search of depth (cover) or food. riod average temperature in pools with visits, including the hot summer Trout were observed in midsummer trout in midsummer was about 65°F months. Trout numbers increased in (sample visits June 30 to Sept.2) in three in August, which corresponds to the higher elevation pools as the season pools (table 1). We considered these optimal growth temperature ranges progressed (fig. 2b). Trout density was pools to be summer temperature refu- observed for juvenile Chinook salmon positively correlated with Julian day gia, although the pools did not stratify. and steelhead. Stream temperature was correlated with elevation (fig. 3b). Higher elevation pools were cooler and contained trout in midsummer. Pools that contained trout in midsummer contained only na- tive fish species. Other pools contained a higher proportion of native species in spring and fall, when water tempera- tures were cooler. Flows declined across the sum- mer months, then began to increase in October (fig. 4). Flows were higher in pools with no midsummer trout, because these pools were farther downstream in the system where more tributaries, sur- face runoff and groundwater flow had joined the stream. This suggests that flow volume alone is not a good predictor of the presence of trout, and that other factors such as water temperature and dissolved oxygen should also be consid- ered. In general, it is important to deter- mine whether increased flows in lower pools reflects the return of potentially warm tailwater from irrigated pastures, Fig. 3. Stream water temperature across (A) agricultural growing season and (B) elevation in a determination critical to BMP develop- pools in tributaries of Cow Creek, 2003. Values are grouped according to whether or not pools ment and implementation. were occupied by rainbow trout in midsummer. Error bars are one standard deviation. Each point represents the average of five monthly samples (June to October) for one pool. Max. temp. is the For the sites on South Cow Creek highest temperature observed since the previous fish observation; average daily max. temp. is and Cow Creek, water flow did not the average of the highest temperature observed each day since the previous fish observation; increase steadily with decreasing eleva- average temp. is the overall average temperature observed since the previous fish sampling. tion (fig. 5). Throughout the season,

204 CALIFORNIA AGRICULTURE • VOLUME 60, NUMBER 4 Fig. 4. Flow (cubic feet per second) across the agricultural Fig. 5. Flow (cubic feet per second) and elevation on South Cow Creek and growing season in tributaries of Cow Creek, 2003. Values are Cow Creek (SC-1, SC-2, SC-3, CC-1), June 3, Aug. 4 and Sept. 29, 2003. Flows grouped according to whether or not pools were occupied by at site SC-2 were consistently lower than flows at site SC-1, likely due to the rainbow trout in midsummer. diversion of water to the South Cow Creek Powerhouse upstream of site SC-2. flows at site SC-2 were lower than at or flow (table 2). The model term “trout cies such as rainbow trout/steelhead site SC-1, which was farther upstream. present” in table 2 can be directly inter- and salmon in spring and fall, but not This observation is due to the diversion preted as the average difference of pools in midsummer when pool temperatures of water from the mainstem of South occupied by trout compared to unoccu- increase. In contrast, higher elevation Cow Creek between sites SC-1 and SC-2, pied pools. pools may provide cold-water fish for use at the Pacific Gas and Electric For example, the maximum daily habitat throughout the year. It is likely Company’s (PG&E) South Cow Creek temperature of pools occupied by trout that rainbow trout/steelhead popula- Powerhouse. The water is returned to was 12.86°F cooler than pools not oc- tions are limited by the stream area that South Cow Creek between sites SC-2 cupied by trout. Conversely, pools oc- is habitable in midsummer, and that and SC-3 via a PG&E canal, restoring cupied by trout were 0.52 feet deeper on lengthening the part of the stream that much of the creek’s volume. average compared to unoccupied pools. is cold enough year-round to support Linear mixed-effects statistical analy- Two relatively deep pools, OC-2 and cold-water species will allow popula- sis was done on the full data set (i.e., CC-1, did not have trout in midsummer, tions to increase. six pools, with five monthly visits per likely because temperatures were too Since species distribution is elastic pool between June and October). The high (table 1). over the season, the species observed resulting seven models indicated that in a given pool depends on the time Fish and farm implications the presence of trout was negatively chosen for sampling. This phenomenon correlated with temperature and posi- Our study looked at fish distribution may assist in the design of habitat moni- tively correlated with pool depth, but in the Cow Creek watershed across the toring and fishery restoration projects not correlated with pool area, elevation spring-summer-fall season, roughly cor- where funds are limited. Midsummer related with the is likely to be the most efficient time of

TABLE 2. Results of linear mixed-effects analysis agricultural grow- year to assess the abundance, size and to predict pool characteristics by the presence of trout ing season and age composition of resident rainbow the use of stream trout populations and other cold-water Pool characteristics Model term Coefficient (S.E.)* P value water for irriga- native species, since cold-water fish are Max. temp. (oF) Intercept 82.13 (2.07) < 0.001 tion. The results confined to cooler (generally higher el- Trout present -12.86 (2.92) 0.012 suggest that the evation) areas. Avg. daily max. temp. (oF) Intercept 77.60 (1.43) < 0.001 Trout present -13.10 (2.02) 0.003 distribution of fish Irrigation diversions may affect pool Avg. temp. (oF) Intercept 72.13 (1.59) < 0.001 species in pools in characteristics such as flow, maximum Trout present -11.46 (2.24) 0.007 rangeland streams depth and temperature. If irrigation Pool depth (ft) Intercept -0.72 (0.10) < 0.001 Trout present 0.52 (0.14) 0.020 varies across the practices tend to cause stream water Pool area (yd2) Intercept 273.56 (70.12) < 0.001 growing season, temperatures to increase, it may be pos- Trout present -122.08 (99.16) 0.286 and is correlated sible to adjust agricultural water-use Elevation (ft) Intercept -300.56 (72.30) < 0.001 Trout present 133.68 (102.25) 0.261 with factors such practices so that tailwater is no warmer Flow (cfs) Intercept 84.00 (19.46) 0.002 as stream tem- than the stream water to which it is Trout present -31.93 (22.98) 0.237 perature and pool returned. It would then be possible to * Significant coefficients for “trout present” model term are directly interpreted as depth. Lower el- extend the length of the stream that is the difference in pool characteristics for pools where trout are present compared to pools without trout (e.g., pools with trout averaged 0.52 feet deeper than pools evation pools may cool enough to be habitable for trout without trout). provide habitat for throughout the summer, and increase cold-water spe- the minimum habitat area available.

http://CaliforniaAgriculture.ucop.edu • October–December 2006 205 Interactions. University of Washington, Institute of Forest Resources. Contribution No. 57. p 191­­–32. Bjornn TC, Reiser DW. 1991. Habitat requirements of salmonids in streams. Am Fish Soc Spec Pub 19:83­–138. Brown GW. 1985. Forestry and Water Quality. Col- lege of Forestry, Oregon State Univ Publ, Corvallis, OR. 142 p. Ebersole JL, Liss WJ, Frissell CA. 2003. Thermal het- erogeneity, stream channel morphology, and salmonid abundance in northeastern Oregon streams. Can J Fish Aquat Sci 60:1266­–80. Elliott JM. 2000. Pools as refugia for brown trout during two summer droughts: Trout responses to ther- mal and oxygen stress. J Fish Biol 56:938–48. Harding JS, Benfield EF, Bolstad PV, et al. 1998. Stream biodiversity: The ghost of land use past. Proc Natl Acad Sci USA 95:14843–7. Mihursky JA, Kennedy VS. 1967. Water temperature criteria to protect aquatic life. Am Fish Soc Spec Pub 4:20­­–32. To spawn and thrive, fish have specific water temperature and flow requirements. Best Moyle PB. 2002. Inland Fishes of California. Berkeley, management practices for irrigation can minimize adverse effects on stream habitat. Author CA: UC Pr. 502 p; fig. 5. Thompson snorkels in South Cow Creek to observe and count fish. Moyle PB, Crain PK, Whitener K, Mount JF. 2003. Alien fishes in natural streams: Fish distribution, assem- blage structure, and conservation in the Cosumnes River, California, U.S.A. Env Biol Fish 68:143–62. Preferably these management practices of the current study — and future stud- Moyle PB, Marchetti MP, Baldrige J, Taylor TL. 1998. would be implemented without exces- ies to assess relationships between BMPs, Fish health and diversity: Justifying flows for a California stream. Fisheries 23(7):6–15; fig. 2. sive costs to irrigators. Such practices stream temperature and flow — will Myrick CA, Cech JJ Jr. 2004. Temperature effects on would have the added benefit of reduc- allow more informed choices about juvenile anadromous salmonids in California’s central ing regulator concerns about the im- which BMPs to implement and their valley; What don’t we know? Rev Fish Biol Fisheries 14:113–23. pacts of rangeland agricultural practices prioritization. This should result in Myrick CA, Cech JJ Jr. 2005. Effects of temperature on streams and native fish. fewer adverse effects on stream habi- on the growth, food consumption, and thermal toler- This study provides a foundation tat, and the avoidance of unwarranted ance of age-0 Nimbus-strain steelhead. N Am J Aquacult 67:324­–30. for understanding the current condi- impacts on agricultural stakeholders in Nielsen JL, Lisle TE, Ozaki V. 1994. Thermally strati- tions and relationships among stream the watershed. fied pools and their use by steelhead in northern Califor- habitat, fish distribution and habitat nia streams. Trans Am Fish Soc 123:613–26. use, as they correspond to the period of Pinheiro JC, Bates DM. 2000. Mixed Effects Models in S and S-PLUS. New York: Springer-Verlag. 528 p. agricultural irrigation-water extraction L.C. Thompson is Assistant Specialist, Department Rahel FJ. 2000. Homogenization of fish faunas and return to Cow Creek. Fish habitat of Wildlife, Fish, and Conservation Biology, UC across the United States. Science 288:854­–6. use and distribution is not static and is Davis; L. Forero is County Director and Farm Advi- Tate KW, Lancaster DL, Morrison, JA, et al. 2005. driven by factors (temperature and pool sor, UC Cooperative Extension, Shasta County; and Monitoring helps reduce water-quality impacts in flood- irrigated pastures. Cal Ag 59(3):168–75. depth) that could logically be affected by Y. Sado is Postgraduate Researcher and K.W. Tate is Specialist, Department of Plant Sciences, UC Da- Tate KW, Lile DF, Lancaster DL, et al. 2005. Statistical stream-based irrigation diversion and analysis of monitoring data aids in prediction of stream vis. The authors gratefully acknowledge the land- return irrigation practices (Tate, Lile et temperature. Cal Ag 59(3):161–7. owner cooperators in the Cow Creek watershed Thompson LC, Larsen R. 2004. Fish Habitat in al. 2005; Tate, Lancaster et al. 2005). The who gave access to their properties for this study; Freshwater Streams. Farm Water Quality Planning next focus of research on the potential Mike Berry, California Department of Fish and Series Reference Sheet 10.3. UC DANR Pub 8112. 12 linkage between irrigation management Game, for his advice on field protocols; and Peter p; table 1. http://anrcatalog.ucdavis.edu/merchant. ihtml?pid=5632&step=4. and habitat factors should be strategic B. Moyle and anonymous reviewers for reviewing the manuscript. Walters CJ, Juanes F. 1993. Recruitment limitation as stream temperature and flow monitoring a consequence of natural selection for use of restricted of lower elevation pools. Tate, Lancaster feeding habitats and predation risk taking by juvenile References fishes. Can J Fish Aquat Sci 50:2058–70. et al. (2005) provide an example of such a Beschta RL, Bilby RE, Brown GW, et al. 1987. Stream Zoellick BW. 1999. Stream temperatures and the monitoring strategy in similar streams in temperature and aquatic habitat. In: Salo EO, Cundy TW elevational distribution of redband trout in southwestern northeastern Modoc County. The results (eds.). Streamside Management: Forestry and Fishery Idaho. Great Basin Naturalist 59:136–43.

206 CALIFORNIA AGRICULTURE • VOLUME 60, NUMBER 4 research Article t EU support reductions would benefit California tomato growers and processors

by Bradley J. Rickard and Daniel A. Sumner

Many countries apply import barri- Morning Star Company ers for processing tomatoes, but the European Union is the main producer that uses export and production subsidies. We modeled and measured the potential impacts on global mar- kets and the California industry that would result from reductions in trade barriers (such as import tariffs) and subsidies for the European Union’s processing tomato industry. A multi- equation simulation model showed that reducing trade barriers in Europe California growers produced $670 million worth of processing tomatoes in 2004, making it the state’s 11th most valuable crop. More than 10% of the California crop is processed into products and elsewhere (including the United for export, such as tomato paste and sauce. States) by 50% would raise the mar- ket price for California tomatoes by About $250 million of processed to- The Doha Round of trade ne- about 6%, improve net returns to Cali- mato products were exported in 2004, gotiations under the World Trade fornia processing tomato producers by accounting for approximately 12% of Organization (WTO) contained a num- ber of proposals to reduce agricultural $34 million per year, and improve net the crop; the industry ranked eighth among California agricultural com- subsidies, lower import barriers and returns to California tomato proces- modities in value of exports (Bervejillo eliminate export subsidies on a global sors by $19 million per year. We also and Sumner 2005). basis. A relevant option based on the found that a 50% reduction in EU The United States and European discussions in those negotiations in- Union each supply approximately cludes the following: eliminating export domestic support would improve the one-third of the world’s processing subsidies and a 50% cut in both tariffs net returns of California producers tomatoes (fig. 1). There is little or no and domestic support for agricultural and processors by about $8.5 million direct subsidy for processing tomatoes commodities (WTO 2004). Our analysis per year. Based on these results, we in the United States; however, process- shows what the California processing ing tomato production is directly sub- tomato industry can expect if these ne- believe that negotiating reductions sidized in the European Union with gotiations are successful in reducing EU in subsidies, and especially in global payments to growers. The EU subsidy subsidies and protection. trade barriers, would make economic regime for processing tomatoes is part Support for EU industry sense for the California processing of their overall system of subsidy, which also applies to other fruit and The EU policy for processing toma- tomato industry. vegetable industries. toes includes domestic support in the We investigated the consequences form of subsidies, import tariffs and an alifornia produces 95% of the of EU processing tomato subsidies export subsidy (European Commission processing tomatoes grown in and global trade barriers for tomato 2005). The subsidies are payments tied theC United States, and the processing producers and processors, especially to the production of processing toma- tomato industry is an important com- in California. Through a simulation toes. Import tariffs are a tax applied to ponent of California agriculture. Its model, we show quantitatively how processed tomato products entering the total revenue was $670 million in 2004, the removal of EU production subsi- European Union, and export subsidies ranking processing tomatoes 11th dies would reduce EU production and are paid to EU processors for selected among all agricultural commodities exports, and raise prices in the global tomato products that are exported. produced in California (USDA 2005). market. We also show that trade barriers From 1978 to 2000, EU domestic sup- Processed tomato products are also a (such as import tariffs) have even larger port included a complex array of direct major California export commodity. effects than subsidies. transfers to processors (processor aid),

http://CaliforniaAgriculture.ucop.edu • October–December 2006 207 Model parameterization The proportional changes in prices and quantities are functions of vari- ous elasticity and share parameters. (Elasticities are used to represent the ra- tio between proportional change in one variable and proportional change in an- other.) The price elasticities of demand for the five processed tomato products were calculated from an overall price Fig. 1. Average processing tomato production globally, 1999 to 2003. elasticity of demand for all processed Source: Tomato News 2004. tomato products, an elasticity of sub- stitution between processed products and consumption shares (Armington minimum prices for growers, and a of 12.5% to processed tomato products, 1969). The overall elasticity was set to quota that set a limit on the quantity and the average (nonweighted) tariff in -0.5, and was based on estimates from eligible for the processor aid and mini- other tomato-importing regions is ap- George and King (1971) and Huang mum price. In 2000, processor aid was proximately 20%. (1985). The elasticity of product sub- approximately $165 (€163) per ton of to- stitution was set to 5.0, reflecting the The simulation model mato paste; the minimum price for pro- fact that tomato products are relatively cessing tomatoes was $81 (€80) per ton; We used an economic simulation close substitutes for each other. and the minimum price was applied model to assess the effects of lower EU The consumption shares were based to approximately 7.7 million tons of domestic support and reduced border on per-capita-consumption rates for processing tomatoes grown in Europe. measures on the global processing to- processed tomato products (USDA (In January 2000, $1 was equivalent to mato industry. There are three regions 2005). The price elasticity of supply for €0.99; in January 2006, $1 was equiva- in our model: the European Union, processing tomatoes was assumed to lent to €0.82.) the United States and the rest of the be relatively inelastic and was set to This subsidy program was changed world. The model accommodates five 0.5, based on an estimate from Chern beginning with the 2001 crop, and processed products that are less-than- and Just (1978). The price elasticity of the changes have further stimulated perfect substitutes for each other in the manufacturing input supply was EU production (Rickard 2003). Since consumption: (1) European Union­– set to 1.0. The cost share for processing February 2001, EU growers of process- produced canned tomato products, tomatoes was set to 45% for tomato ing tomatoes have received a per-ton (2) canned tomato products from other paste and 20% for canned tomato subsidy from the European Union, as sources, (3) European Union–produced products (based on estimates from a long as total EU production does not paste, (4) U.S. paste and (5) paste pro- survey of industry experts). We al- exceed the threshold limit of 9.1 mil- duced in other countries. In each of the lowed for some substitution between lion tons (8.25 million metric tons). In three regions, two inputs (raw tomatoes processing tomatoes and the manufac- 2005, the subsidy rate was approxi- and other inputs) are used in the pro- turing input, and this parameter was mately $39 per ton and therefore, ap- duction of these five processed tomato set to 0.1. (A value of 0 would indicate proximately 43% of per-unit revenue. products. There is trade among the re- no input substitution; for perfect sub- Since 2001, the EU tariff has been set gions in processed tomato products but stitutes, the elasticity would be infi- at 14.4% for processed tomato products not in raw tomatoes. nite.) The simulation model included such as tomato paste and tomato sauce. The simulation model is used to per- raw tomatoes and processed tomato The tariff rate has been reduced by one- form experiments in policy reform; that products, and we used a conversion fifth since 1995, in accordance with the is, to examine the effects that alterna- rate of 6.1 tons of tomatoes per ton of Uruguay Round GATT deal, which is tive policy scenarios would have on the tomato paste and 1.2 tons of tomatoes administered by the WTO. This tariff is processing tomato industry. We focus per ton of canned tomato products. refunded when the imported product on reductions in domestic subsidies, ex- The effects of alternative values of is used in, or offset by, exports of pro- port subsidies and tariffs. The European several of these parameters, especially cessed tomato products. The European Union has the world’s only significant supply and demand elasticities, were Union also allows reduced or zero tar- program of domestic subsidies for pro- also examined to test sensitivity in iffs for imports from selected develop- cessing tomatoes, but many countries, Rickard (2003). Our results are robust ing countries. Export subsidies apply including the United States, have im- to changes in key parameters across a to certain canned tomato products, port tariffs. It is implausible that import plausible range, and the main results which make up a relatively small share tariffs in the European Union would fall for the most likely parameter values of total processed tomato production in unilaterally; therefore, we considered are reported here. the European Union. For reference, the multilateral reductions in import tariffs United States applies an import tariff across all tomato-producing regions,

208 CALIFORNIA AGRICULTURE • VOLUME 60, NUMBER 4 TABLE 1. Simulated effects of policy changes on prices and quantities TABLE 2. Simulated effects of policy changes in the U.S. processing tomato industry* 50% reduction in: 50% reduction in: EU domestic import tariffs + Variable import tariffs support EU domestic support EU domestic import tariffs + Benefit or cost to: import tariffs support EU domestic support ...... % change ...... EU tomatoes change in U.S. $ millions Grower price 1.4 –9.3 –7.9 U.S. tomato 34.6 5.5 40.3 Processor price 1.4 12.2 13.6 producers† Quantity 0.7 –4.6 –3.9 U.S. tomato 18.9 3.0 22.0 EU processor inputs processors Price 0.8 –3.1 –2.3 U.S. govt. budget –2.4 –1.6 –3.2 Quantity 0.8 –3.1 –2.3 (tariff revenue) U.S. tomatoes U.S. consumers of pro- –19.8 –6.2 –24.6 Price 6.2 1.0 7.2 cessed tomato products Quantity 3.1 0.5 3.6 Total U.S. economy 31.3 0.6 34.5 U.S. processor inputs Price 3.4 0.5 3.9 * Effects for the European Union and the rest of the world are available from the authors. Quantity 3.4 0.5 3.9 † California growers would earn more than 90% of the benefits. The gain of $34.6 million represents 6.4% of total producer revenue and a significantly larger percentage of net revenue. combined with reductions in EU export export more to third markets, which be- lion per year for U.S. growers and subsidies and EU domestic support. gin with higher tariff rates (table 1). The $3.0 million per year for U.S. processors. The simulation model used a set of price and quantity of processor-supplied The reduction in tariffs on a global ba- equations to describe the supply and inputs would also rise. sis would have a significantly larger ef- demand conditions for the processing Domestic support. If EU production fect on tomato producers and processors tomato sector. Equilibrium adjustments subsidies were cut by 50%, we found that in the United States. The benefit to U.S. can be simulated by specifying changes per-unit grower returns in the European producers of tomatoes would be approx- in the policy parameters, such as changes Union would fall, and market prices imately $34 million per year, with about to EU domestic support or tariff rates. paid by EU processors for tomatoes $32 million per year of that for growers The model is used to simulate propor- would rise by 12.2% (table 1). The result in California. The increase in benefits to tional changes in prices and quantities is a decline in the quantity of tomatoes U.S. processors would be approximately (and ultimately benefits or costs to pro- used and a decline in processor-supplied $19 million per year, with almost all of ducers and processors) for selected input and output markets in the processing Negotiating reductions in subsidies, and especially in tomato industry (see box, page 208). global trade barriers, makes economic sense for the Effects of policy reform California processing tomato industry. The simulation results describe how changes in EU export subsidies, global inputs as well. Cutting EU domestic that amount again benefiting California tariffs and EU domestic support would support by this magnitude would have processors. Benefits to tomato producers affect prices, quantities and net benefit positive effects in the United States. and processors would total $53.5 million measures, such as revenues and govern- Import tariffs and domestic support. per year. Part of this net revenue increase ment expenditures. The focus is on the If the European Union cut production would come from U.S. markets and part effects in the European Union and, espe- subsidies and all countries cut tariffs, from additional export revenue. The cost cially, in California. there would be an increase in the price to U.S. consumers from higher prices for Export subsidies. The complete paid for tomatoes and a decrease in U.S. tomato products would be about elimination of export subsidies would the price received by growers in the $20 million. The United States would lower EU export tonnage by only European Union (table 1). Reducing also lose $2.4 million in tariff revenue, so 0.6%. The small impact of export sub- global tariffs and EU subsidies together the net gain for the United States would sidy reform is mostly attributed to the would raise prices and quantities in the be about $31 million per year. fact that the export subsidy rate is low United States. Implications for the U.S. industry and applies to only a small portion of Benefits to U.S. producers and total EU production. Because the im- processors. The changes in prices and Farmers and processors in the pact of the EU export subsidy on the quantities (table 1) were then used to United States would benefit more from California industry is so small, the rest calculate changes in net producer rev- reductions in import tariffs than from of our analysis focused on the effects enues, net government expenditures reductions in EU domestic support, of reductions in import tariffs and EU or tariff revenues, and consumer ben- even though that would also mean re- domestic support. efits from tomato consumption in the ductions in the U.S. tariff. However, re- Tariffs. We found that a 50% cut in tar- United States (table 2). The annual net ductions in import tariffs would place iffs would increase the price and quan- benefits to producers and processors pressure on the EU domestic support tity produced in the European Union and from cutting the domestic subsidy in regime. Reducing import tariffs would the United States because both would the European Union would be $5.5 mil- increase the production of tomatoes

http://CaliforniaAgriculture.ucop.edu • October–December 2006 209 Morning Star Company

If the European Union reduced its trade barriers for processing tomatoes by 50%, California producers and processors stand to gain an estimated $53.5 million annually.* in the European Union, and thereby inputs to the tradable product. Third, Agricultural Issues Center, and Frank H. Buck, Jr., Chair Professor, Department of Agricultural and increase the taxpayer cost of the EU the EU domestic support program Resource Economics, UC Davis. domestic support regime. This would drives a wedge between the price that place additional pressure on EU bud- growers receive and the price that pro- References gets that could lead to reductions in cessors pay for tomatoes. Reducing EU Armington PS. 1969. A Theory of Demand for subsidies as a response. domestic support would reduce that Products Distinguished by Place of Production. IMF Staff Papers 16 (March):159–76. International Monetary Fund, Producers and processors in the wedge, and the burden of any reduction Washington, DC. United States would gain about $8.5 mil- would be shared between the grower Bervejillo JE, Sumner DA. 2005. California’s Inter- lion annually from a reduction in the EU and processor. national Agricultural Exports in 2004. AIC Brief 30. UC subsidies, and the gain from tariff elimi- Agricultural Issues Center, Davis, CA. Effects of trade negotiations Chern WS, Just RE. 1978. Econometric Analysis of nation would be even larger. This result Supply Response and Demand for Processing Tomatoes of the simulations may seem surprising, Trade negotiations have the potential in California. Giannini Foundation Monograph 37. UC because the initial tariff is only 14.4%, to reduce trade barriers and farm sub- Berkeley, Berkeley, CA. European Commission. 2005. Agricultural Markets: while the subsidy is 43%. The relative sidies on a global basis. The California Fruits and Vegetables. http:/europa.eu.int/comm/ magnitude of the two impacts is driven processing tomato industry has long agriculture/markets/fruitveg/index_en.htm (accessed by three factors. been concerned with subsidies and Aug. 15, 2006). Brussels, Belgium. George PS, King GA. 1971. Consumer Demand for First, the supply response of process- import barriers in the European Union. Food Commodities in the United States with Projections ing tomatoes in the European Union to Our research shows that this interest is for 1980. Giannini Foundation Monograph 26. UC Da- lower per-acre returns (including policy well placed, and although the effects vis, Davis, CA. Huang KS. 1985. U.S. Demand for Food: A Com- benefits) is relatively inelastic in the of domestic subsidies are significant, plete System of Price and Income Effects. USDA Eco- intermediate time-frame because — for to increase net returns the negotiations nomic Research Service, Washington, DC. Tech Bull this analysis, and in the context of trade should emphasize trade barriers more 1714. Rickard BJ. 2003. Domestic support and border mea- negotiations — we envision reductions than domestic subsidies in Europe. We sures for vertically linked and differentiated goods: An in support for processing tomatoes as also show that the California processing examination of EU policy in the processing tomato in- part of a larger, multicommodity pack- tomato industry would receive consid- dustry. Ph.D. dissertation. UC Davis, Dept. of Agricultural and Resource Economics. age. Second, EU domestic support ap- erable benefits from global tariff reduc- Tomato News. 2004. Worldwide production of plies to the farm-produced product, tions, even though that would mean tomatoes for production: 1990 to 2003. www.toma- and import tariffs apply to processed giving up some of its own protection tonews.com/processing.php. products. The farm product represents from imports. [USDA] US Department of Agriculture. 2005. Na- tional Agricultural Statistics Services. Agri Stat. www. only 45% of the cost of tomato paste usda.gov/nass/pubs/agstats.htm (accessed Aug. 15, and 20% of the cost of canned tomato 2006). products, and barriers that apply at the B.J. Rickard is Assistant Professor, Agribusiness De- [WTO] World Trade Organization. 2004. Decision adopted by the General Council on 1 August 2004. WT/ border have bigger effects on trade than partment, California Polytechnic State University, L/579, Geneva: WTO. www.wto.org/english/tratop_e/ do subsidies for raw materials that are San Luis Obispo; and D.A. Sumner is Director, UC dda_e/draft_text_gc_dg_31july04_e.htm#par1b. * Photo at left added postpublication, 10/6/06. 210 CALIFORNIA AGRICULTURE • VOLUME 60, NUMBER 4 Research Article t Weather-based yield forecasts developed for 12 California crops

by David B. Lobell, Kimberly Nicholas Cahill, and Christopher B. Field

Crop-yield forecasts provide useful information to growers, marketers, government agencies and other us- ers. Yields for several crops in Cali- fornia are currently forecast based on field surveys and farmer interviews, although official forecasts do not exist for many crops. Because broad- scale crop yields depend largely on the weather, measurements from existing meteorological stations have the potential to provide reliable, timely and cost-effective predictions. We developed weather-based models of statewide yields for 12 major Cali- fornia crops and tested their accuracy using cross-validation from 1980 to 2003. Many of the weather-based forecasts were highly accurate, as Weather-based yield predictions were developed for 12 major California crops, based judged by the percentage of yield on more than 20 years of daily weather records and actual yield data. The highest correlation between weather and yield was seen in almonds. variation explained by the forecast, the number of yields with correctly predicted direction of yield change, arrange greater storage capacity, and used It is common knowledge that one a 2005 forecast of reduced almond pro- of the main factors causing yields to or the number of yields with cor- duction to allocate limited quotas among change from year to year is climate vari- rectly predicted extreme yields. The preferred customers (D. Flohr, CASS, ability — no two growing seasons expe- most successfully modeled crop was personal communication). Similarly, in- rience exactly the same weather. Indeed, almonds, with 81% of yield variance dustries involved in handling and trading grower expectations of crop yields commodities often use information on fu- are likely to be based at least partially captured by the forecast. Predictions ture harvests to make logistical decisions on subjective weather observations for most crops relied on weather (Hammer et al. 2001). and perceived relationships between measurements well before harvest Each year, the California Agricultural weather and yields. To our knowledge, time, in many cases allowing longer Statistics Service (CASS) estimates the size objective, quantitative weather mea- of the coming harvest for various major surements are not currently used in ex- lead times than existing procedures. California crops, including almonds, isting yield-forecast procedures. Such an grapes, olives, oranges and walnuts approach would be attractive because orecasts of crop yields can provide (NASS 2005a, 2005b). These estimates are yields could potentially be forecast at important information about com- categorized as either subjective or objec- lower cost, with greater accuracy and modityF markets and are frequently used tive. The former are based on phone inter- longer lead times. by growers, industry and government to views with hundreds of farmers to assess Building forecast models make decisions (Vogel and Bange 1999). their opinions of crop development, and For instance, growers may use forecasts the latter are based on field samples taken To test the ability of weather mea- to plan their harvest, storage and distribu- from hundreds of fields. Forecasts are surements to forecast crop yields prior tion strategies: California growers used generally made public 1 to 3 months be- to harvest, we studied the statistical the 2004 forecast of a large rice harvest to fore the end of harvest (NASS 2005a). relationships between historical weather

http://CaliforniaAgriculture.ucop.edu • October–December 2006 211 Yields could potentially be forecast at lower cost, TABLE 1. Economic value and national importance of production for crops studied and with greater accuracy and longer lead times.

Crop 2003 value* U.S. production TABLE 2. Months and weather variables* used for yield forecasts $ millions % Grapes, wine 1,828 96 Year prior to harvest Year of harvest Lettuce 1,634 88 Crop† Aug Sept Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sept Almonds 1,506 99 Strawberries 973 83 Grapes, wine ppt tmn ppt Grapes, table 953 91 Lettuce tmx tmx tmx Hay 950 12 Almonds ppt tmn Oranges 949 22 Strawberries all Cotton 774 10 Grapes, table ppt ppt tmn tmn Tomatoes, 571 95 Hay ppt ppt processing Oranges tmn ppt Walnuts 434 99 Cotton tmx tmn Avocados 402 95 Tomatoes, processing tmx tmx tmn Pistachios 173 99 Walnuts tmx ppt Avocados tmx ppt tmn * Values are taken from CASS (2004a), and are based on free-on-board (FOB) prices that include value-added * tmn = average minimum temperature; tmx = average maximum temperature; ppt = total rainfall; all = all three variables. items such as packing and inspections. † No weather variables are shown for pistachios, which were modeled using only previous years’ yields. and crop-yield records. We selected 12 crops (wine grapes, lettuce, almonds, strawberries, table grapes, hay, oranges, cotton, processing tomatoes, walnuts, avocados and pistachios) that are among the most valuable in California (table 1) (CASS 2004a), and obtained state yield data for 1980 to 2003 from California county agricultural commis- sioners (CASS 2004b). Several crops have exhibited significant positive yield trends since 1980 due to management and cultivar changes, so we removed a linear trend from each crop to produce a time series of yield anomalies, or depar- tures from expected yields. A positive anomaly indicates yields higher than expected based on time trends, and a negative anomaly indicates yields lower than expected. Daily weather records for the same period were obtained for 382 stations throughout California from the California Climate Change Center at the Scripps Institution of Oceanography (M. Tyree, staff scientist, personal communication). The average daily minimum and maxi- mum temperature and precipitation for each month in each county were then computed, resulting in a monthly time series of three variables for 24 years. For each crop, a statewide monthly time se- ries for each meteorological variable was calculated by weighting each county by the relative area of the crop in that county in 2003 (Lobell et al. 2006). The weather and yield data were Fig. 1. Observed and forecast yields from 1980 to 2003. Forecasts were derived for each then combined in linear regression year using a model fit to data in all other years. Coefficient of determination 2(R ) between models to test how well yield anoma- observed and forecast yields is shown for each crop.

212 CALIFORNIA AGRICULTURE • VOLUME 60, NUMBER 4 TABLE 3. Summary of forecast accuracy and timing for crops evaluated

Fraction yrs. Last month Months between R2 using only with forecast in used in Month of USDA Peak harvest forecast and Crop R2cv* RMS† (%) previous yields‡ correct direction forecast forecast§ period harvest end

Grapes, wine 0.59 6.4 n/a 0.61 June July–August August– 4 October Lettuce 0.44 4.0 n/a 0.61 April Continuous — Almonds 0.81 7.8 0.17 0.73 February May (subjec- August– 8 tive); June October (objective) Strawberries 0.49 4.6 n/a 0.48 Previous April Continuous — November Grapes, table 0.62 6.7 n/a 0.61 July July–August July– 4 September Hay 0.44 3.9 0.01 0.55 June August March– 5 November Oranges 0.69 8.8 0.22 0.68 May Navel: November– 6 September; May; May–Oct¶ Valencia: March Cotton 0.56 6.3 n/a 0.54 June June–August October– 6 December Tomatoes, 0.49 3.1 n/a 0.67 September May and June– 2 processing September November Walnuts 0.43 7.3 0.06 0.57 February September September– 9 November Avocados 0.57 16.7 n/a 0.70 May Continuous — Pistachios 0.35 27.5 0.42 0.70 n/a August September– — November

* R2cv: Cross-validated R2, the proportion of yield variance explained by the weather predictor variables. † RMS: Root mean squared difference between forecast and observed yield, expressed as a percentage of average yield for 2000 to 2003. ‡ Only crops that exhibited alternate bearing were modeled with previous years’ yields. § Available in California Crop Production Reports (www.nass.usda.gov/ca). ¶ The first period refers to navel orange harvest and the second to Valencia oranges.

lies could be predicted before harvest squared variables and previous years’ Forecast accuracy based on monthly weather measure- yields — ranged from four to eight. ments. Between two and four weather (The model equations are omitted for The results of the cross-validation variables were selected for each crop, brevity, but can be obtained from the analysis suggest that yields of some based on a combination of objective authors.) crops can be forecast with fairly high (good model accuracy) and subjective An important step in building sta- accuracy based on objective weather (physiologically reasonable) criteria tistical models is to independently test measurements (fig. 1; table 3). For (table 2). Because temperature and model predictions, because tests using many crops, the model forecasts cap- precipitation can have a nonlinear ef- the same data that was used to calibrate tured close to or more than 50% of the fect on yields, with yields maximized the model tend to be overly optimistic variability in yield anomalies, mean- at intermediate values, we included (Hastie et al. 2001). The straightforward ing that the selected weather variables the squared values of the weather approach of reserving part of the data explained over half of the variations variables in the regression model during model calibration, however, is observed in crop yields. Interestingly, along with the variables themselves. problematic when the quantity of data is the models did fairly well at forecast- For crops such as pistachios that limited. An alternative approach, which ing extremely low yields, such as al- are known to exhibit alternate bear- we employed here, is “leave-one-out” monds in 1995, oranges in 1991, and ing, with years of high reproductive cross-validation. In this approach, a sin- processing tomatoes and cotton in 1998 growth (high yields) alternating with gle year is left out of the calibration step (fig. 1). Almonds were particularly well years of high vegetative growth (low and subsequently compared to model modeled, with over 80% of variance yields), yield anomalies from previ- predictions in that year. This comparison captured by the model. ous years were also included in the is done for each year, in this case result- For a few crops, some of the power model. The total number of predictors ing in 24 comparisons between model of the models came from knowing the — including the weather variables, predictions and observations. previous year’s yield (table 3). For

http://CaliforniaAgriculture.ucop.edu • October–December 2006 213 Jack Kelly Clark/UC Davis Kelly Jack instance, including weather informa- wrong direction — cor- tion did not improve the pistachio responded to years with model, where the biological pattern moderate yields, so the of alternate bearing seemed to domi- forecast was in fact not nate effects on yield more than any far from the observed weather signal. For all other crops, yield. None of the crops however, most or all of the predictive exhibited any years with power came from weather variables. a forecast error of three As an alternative measure of fore- classes. cast skill, we considered the fraction To test the significance of years in which the model correctly of these class accura- predicted the direction of yield ano- cies, we performed 1,000 moly (table 3). That is, we examined simulations using two the frequency with which the model 24-year, random-noise correctly predicted whether the yield variables with a normal would be above or below the trend. distribution. The average For a random forecast, this statistic percentage of years with has a distribution whose mean is 0.5 an error of zero, one, and whose 90th percentile is 0.625 for two or three classes was a 24-year record (15 out of 24 years). 28%, 45%, 22% and 5%, Six of 12 crops had a forecast with respectively. Only 10% of skill greater than a random forecast the simulations had more using this criterion and significance than 40% of years (10 out level. Three others (wine grapes, table of 24) classified correctly grapes and lettuce) fell slightly below by chance, while all crops this value. except strawberries, Another criterion is the ability of pistachios and walnuts It is well known that climate is an important factor influencing crop yields from year to year. Weather-based yield forecasts can forecasts to correctly predict unusually met this criterion. This be developed at lower cost than field surveys, and with longer high or low yields, which is of particu- indicates that the forecast lead times. Above, a weather monitoring station. lar interest to many forecast users. For accuracies for most of the each year, both the forecast and the crops were statistically actual yield were classified into one of significant by this measure. our models offer significant timing four classes: below one standard devia- advantages over existing forecasts for Importance of timing tion (SD) from zero, between minus almonds (3 to 4 months earlier than one SD and zero, between zero and Forecast timing can be as impor- current forecasts), hay (2 months ear- one SD, and above one SD from zero. tant as accuracy. A “forecast” made lier), strawberries (5 months earlier) The first and fourth of these classes after harvest, for example, would not and walnuts (7 months earlier). represent unusually low or high yields, be very valuable. Most of our models Potential improvements respectively, and the middle two rep- are capable of providing forecasts at resent more moderately negative or least several months before the end of The current analysis was limited to positive years. We then computed the harvest, giving growers and others an only a dozen of the many crops grown number of years when the forecast cor- opportunity to use the information to in California and considered only state- rectly predicted the yield class, was off make decisions (table 3). For instance, wide yields. We chose to aggregate by one class (in either direction), two our models for almonds and walnuts several crops over different subcrop classes or three. relied mainly on winter weather, groupings, such as by combining va- Most crops did not exhibit any while harvest does not begin until rieties of hay and lumping navel and years when the forecast was off by late summer. Valencia oranges together. In addition, more than one class. There were We compared the times that our we used only monthly averages of three some exceptions; for example, let- modeled yield predictions could be meteorological variables (number of tuce yields in 1981 were forecast to made available to growers with the frost days per month was also consid- be slightly negative but were actu- times that currently available USDA ered, but did not substantially improve ally very high (above one SD), and forecasts are released (table 3). The any of the models). the reverse was true for hay in 1995. two approaches were similar for wine These decisions reflect an explicit Overall, the forecasts were usually no grapes, table grapes and cotton, and desire to test forecasts of state yields for more than one class off. Most of the existing forecasts were available major crops using commonly reported cases discussed above — where the 4 months earlier for processing to- climatic data. However, data for many forecast predicted an anomaly in the matoes than our models. However, additional crops is currently available at

214 CALIFORNIA AGRICULTURE • VOLUME 60, NUMBER 4 both state and county levels, as are ad- before the forecasts from USDA became (EPA) Science to Achieve Results (STAR) Gradu- ditional weather measurements at time available in May and June. ate Fellowship to D.B. Lobell; a David and Lucile Packard Foundation Stanford Graduate Fellowship scales from hourly to monthly. Open Although field-based surveys are to K.N. Cahill; and the Office of Science, Biological questions are how well other crops can likely to be more accurate than weather- and Environmental Research Program (BER), U.S. be modeled and whether different scales based forecasts, it is important to Department of Energy (DOE), through the West- of analysis and meteorological indices consider the tradeoff between forecast ern Regional Center of the National Institute for would substantially improve forecast accuracy, cost and timing. The low cost Global Environmental Change (NIGEC). This work accuracies. Additional information such and long lead times that are possible was performed under the auspices of the DOE as remote sensing data might also aid with weather-based models would National Nuclear Security Administration, by the UC Lawrence Livermore National Laboratory, under predictions. likely provide a useful complement to contract No. W-7405-Eng-48. It is also possible that different model existing approaches for crops that are formulations could improve results. For currently surveyed. For crops that are References example, in certain situations, process- not currently forecast by USDA, such Almond Board of California. 2004. Almond Alma- based models that rely on a mechanistic as avocados, these models present an nac. Modesto, CA. www.almondboard.com. understanding of crop growth and yield opportunity to develop forecasts with [CASS] California Agricultural Statistics Service. 2004a. California Agricultural Commissioners’ Data. may outperform statistical models such minimal cost by using existing weather www.nass.usda.gov/ca/bul/agcom/indexcac.htm. as the ones developed here, which are measurements. CASS. 2004b. California Agricultural Statistics 2003: derived from observed relationships Overview. USDA-NASS, Sacramento, CA. without explaining the mechanisms Hammer GL, Hansen JW, Phillips JG, et al. 2001. Ad- vances in application of climate prediction in agriculture. causing the relationships. Alternative Agric System 70:515­­–53. D.B. Lobell is Lawrence Research Fellow, Energy statistical approaches to the multiple and Environment Directorate, Lawrence Livermore Hastie T, Tibshirani R, Friedman J. 2001. The Ele- linear regression that we used may also ments of Statistical Learning: Data Mining, Inference, National Laboratory; K.N. Cahill is Ph.D. Candi- and Prediction. New York: Springer. 552 p. improve accuracies. (For example, we date, Department of Global Ecology, Carnegie Lobell DB, Cahill KN, Field C. 2006. Historical effects tested the use of regression trees, which Institution, Stanford, Calif., and Interdisciplinary of temperature and precipitation on California crop did not perform as well.) Whether these Graduate Program in Environment and Resources, yields. Climatic Change. In press. more sophisticated approaches offer Stanford University; and C.B. Field is Director, De- [NASS] National Agricultural Statistics Service. 2005a. California Objective Measurement Reports, worthwhile improvements can be tested partment of Global Ecology, Carnegie Institution, Stanford. The authors are grateful to Mary Tyree 1997­–2005. www.nass.usda.gov/ca/rpts/om/indexom. htm. only on a case-by-case basis, using ac- for providing the climate data, Doug Flohr for tual observations and well-defined cri- NASS. 2005b. California Subjective Measurement information on current forecast applications, and Reports, 1997–2005. www.nass.usda.gov/ca/oth/ teria for an ideal forecast. anonymous reviewers for comments that improved indexfor.htm. the manuscript. Funding for this work came in Vogel F, Bange G. 1999. Understanding crop statis- Weather promising for forecasts part from an Environmental Protection Agency tics. www.usda.gov/nass/nassinfo/pub1554.htm. The models developed in this study are promising for forecasting statewide crop yields based on weather measure- ments. Because the significance levels for the models depend on specified COMINGUP performance criteria, the eventual value of such forecasts will depend on the Blue oak regeneration and survival acceptable types and magnitude of er- Blue oak trees are a valuable eco- rors for particular applications. The nomic and aesthetic resource in Cali- potential to forecast yields also depends fornia oak woodlands, which provide on crop type. In general, almonds had some of the richest wildlife habitat in significantly greater forecast accuracies the state. However, their current re- than the other crops that we considered. generation rates may not be adequate Ralph L. Phillips Ralph L. Because almonds are California’s most to replace mortality, due to a variety valuable export crop and account for of factors. In some regions, the regen- over 80% of global almond production eration of blue oak is limited by the (Almond Board of California 2004), such ability of seedlings to survive long forecasts could be of great relevance to enough to become larger saplings. Also: almond trade and management deci- Two related, long-term studies by UC sions. For example, an almond grower researchers examine the growth of Coyote lure operative device revisited could have used data on January rain- blue oak seedlings and the effect of ▼ fall and February nighttime tempera- exclosures — which protect seedlings Size trends in fresh-market peaches tures to correctly predict the low yield from livestock and wild-animal graz- ▼ in early March 1995 and adjust cultural ing — on their survival. Evaluating quality control in the packingshed or marketing practices accordingly, well

http://CaliforniaAgriculture.ucop.edu • October–December 2006 215 that environmentalism and consumerism are here to stay, and Editor’s note: In honor of our 60th anniversary, California welcome the interest of a wide variety of previously uninterested Agriculture has been publishing excerpts from past issues. individuals, organizations, and agencies. Thirty years ago, California Agriculture was a mostly black- and-white, monthly, 16- to 24-page magazine, with a smat- Hard Tomatoes and Hardy Myths (March 1976) tering of color photos and inks on the cover and inside. A A myth, according to one of Webster’s de nitions, is “an ill- timeline celebrating the journal‘s 60 years is on page 174. founded belief held uncritically especially by an interested group.”  ere is a growing body of mythology concerning our food supply which seems to  t Mr. Webster’s de nition. Its general theme is The early 1970s were a time of increased oversight and scru- that those responsible for our food supply, including the agricul- tiny for agriculture. The U.S. Environmental Protection Agency tural scientists and land-grant universities, have devised a system was formed in 1970 with broad powers to regulate land, air and to drench soil and crop plants and embalm meat animals with water; Jim Hightower published a widely read report on land- harmful chemicals later reinforced by a deadly array of additives grant colleges, “Hard Tomatoes, Hard Times”; and Ralph Nader’s and preservatives. All of this provides us with food that is tasteless, consumer movement was increasingly infl uential. Editorials in expensive, poisonous and nutritionally de cient, and best exempli- California Agriculture by J.B. Kendrick, UC vice president for agri-  ed by the tomato which has been designed with the performance cultural sciences, refl ected these new realities. For headlines from characteristics and taste of a tennis ball. 1976, go to http://California Agriculture.ucop.edu. Our Client “ e Consumer” (June 1976) Accentuating the Positive (February 1976) Our diet and our lifestyle have been transformed because we O en it has been said that agriculture has a poor image, that it have sought new ways to make nature work for us. Scienti c, agri- talks to itself instead of delivering its message to the urban major- cultural and food research has freed us from problems of survival ity, and that it is the only major U.S. industry without an e ective and food preparation and done wonders for the American menu. public relations program directed to the ultimate consumer. We have What has been accomplished has been one of the most success- a story to tell — a factual, positive story of an industry that is indis- ful, but little recognized consumer movements the world has ever pensable to the welfare of every citizen. I submit that now is the time known. It must be admitted that consumer bene ts have evolved to get o the defensive and accentuate the positive! largely as the by-product of the free enterprise system and the Because it is large and complex, agriculture is like the elephant necessity for agriculture to remain economically viable in the face being examined by blind men.  e public’s concept of it may depend of rising costs. But nevertheless, the  nal participant in the food on the part of it with which they have had contact. We must accept complex system — the consumer — is the one who bene ts most.

[email protected] Phone: (510) 987-0044 Fax: (510) 465-2659

visit California Agriculture on the internet: http://CaliforniaAgriculture.ucop.edu216 CALIFORNIA AGRICULTURE • VOLUME 60, NUMBER 4