Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
3 Adoption of Precision Agriculture CURRENT STATUS AND LIKELY TRENDS Precision agriculture is still in its infancy, and information about the extent to which precision agriculture technologies are used is fragmentary. This section summarizes what is currently known about use of these technologies and attempts to draw on experience with existing information-intensive agricultural technolo- gies to generalize about likely future adoption patterns. The section begins with a survey of current evidence regarding the adoption of precision agriculture tech- nologies, then reviews the literature on agricultural technology adoption, and fi- nally discusses likely adoption patterns for precision agriculture. Status of Current Adoption Precision agriculture is a suite of technologies rather than a single technol- ogy. Components of that suite of technologies currently in use include GPS re- ceivers; GIS data bases; variable-rate application equipment for seed, fertilizers, and pesticides; grid soil sampling; low-volume irrigation; yield monitors; sensors for soil fertility and weed populations; and remote sensing imagery. Different configurations of components of that suite will be suitable for different opera- tions. It is therefore somewhat misleading to speak generally about the extent of adoption of precision agriculture. By any measure, however, adoption of preci- sion agriculture technologies is extremely limited at present. Variable-rate application equipment is perhaps the most widely used preci- sion agriculture technology. About 1,600 flotation fertilizer-application systems, map-driven variable-rate technology (VRT) systems, and on-the-go sensor trac- tor-based application systems have been sold. Thirteen percent of the respondents 65
66 PRECISION AGRICULTURE IN THE 21ST CENTURY in a Purdue University study of agricultural chemical dealers were applying fer- tilizer by using controller-driven VRT (Akridge and Whipker, 1996). Grid soil sampling also appears to be among the more popular precision agriculture technologies. The same Purdue University study found that 29 per- cent of the agricultural chemical dealers responding to the survey were pulling grid samples for their customers, whereas 15 percent were mapping fields (Akridge and Whipker, 1996). Computers are a central component of information-intensive agriculture. Fewer than 10 percent of the 1.9 million producers and ranchers in the United States own computers. Commercial yield monitors are available for corn, soy- bean, and wheat harvesting. Approximately 2,000 yield monitors (cumulative) were in use in the 1995 growing season. This figure increased to about 9,000 in 1996. Midwest corn and soybean growers are currently the major buyers (A. Meyers, Ag. Leader Technology, personal communication, June 13, 1997). Diffusion of New Technologies There is a large body of economic and sociological literature on technology adoption in general, as well as on information-intensive agricultural technologies such as computers, integrated pest management (IPM), low-volume irrigation systems (drip, center-pivot, and other sprinkler systems), and the California Irri- gation Management Information System (CIMIS), which combines localized weather information with field-level soil moisture monitoring to improve irriga- tion management. The history of the diffusion of these earlier information-inten- sive technologies offers insights into the likely prospects for current precision agriculture technologies. It has long been recognized that new technologies diffuse gradually. Techno- logical diffusion typically follows an S-shaped path over time. In the early years after a new technology is introduced, it is generally used by only a small percent- age of those who could benefit from using it. As time passes, the rate of adoption tends to increase and diffusion becomes more rapid. Finally, after the majority of those who stand to benefit from using the technology have begun using it, the rate of diffusion slows again. It is important to distinguish between two key variables characterizing the diffusion process: the extent (or ceiling) of adoption and the rate of adoption. The adoption ceiling pertains to the long term, when the diffusion process approaches completion. The rate of adoption pertains to the short term, while the diffusion process is in progress. Grilichesâs (1957) work on hybrid corn established that both the adoption rate and ceiling are influenced by economic factors. The adop- tion ceiling is influenced almost entirely by economic factors. In the short term, however, the rate of adoption is influenced by factors such as learning, risk and risk preferences, information, and human capital as well as by profitability con- siderations.
ADOPTION OF PRECISION AGRICULTURE 67 Determinants of Long-Term Adoption Profitability is the principal determinant in the long term of adoption of new technologies in agriculture and elsewhereâtechnologies that are more profitable will be used. However, adoption of any farming technology is unlikely to be universal because agriculture is characterized by a high degree of heterogeneity. Farming conditions vary markedly among regions and crops because of differ- ences in climate, soils, topography, water availability, government programs, and other factors. As a result, the profitability of any given agricultural technology may differ greatly across regions and crops, so that producers in one region find unprofitable what producers in another find extremely profitable. Differences in climate are arguably the major source of heterogeneity in ag- riculture. They lead to differences in crop productivity and thus in the long-term profitability of adopting new agricultural technologies. In his classic study, Griliches (1957) found that ceiling rates of hybrid corn adoption varied across states. Ceiling rates in the Corn Belt approached 100 percent but were much lower in states with lower corn productivity. Differences in climate also result in differences in pest pressure that affect ceiling rates of IPM adoption. For ex- ample, the use of scouting services for cotton crops is more prevalent in the Delta states than in Texas or California (Economic Research Service, 1995b), which are drier and are thus less subject to pest pressure. Differences in land quality (i.e., soils and topography) are a second major reason for heterogeneity in agriculture and for differences in ceiling rates of adop- tion of agricultural technologies. For example, low-volume irrigation technolo- gies (drip and center-pivot) increase the efficiency of water use more on land with sandy soils and greater slopes and thus are more likely to be adopted by producers operating those types of land (Caswell and Zilberman, 1986; Dinar et al., 1992; Lichtenberg, 1989; Negri and Brooks, 1990; Shrestha and Gopalokrishnan, 1993). Adoption of conservation tillage has similarly been more widespread among pro- ducers operating more erodible land (Economic Research Service, 1995a; Ervin and Ervin, 1982; Gould et al., 1989; Lynne et al., 1988). Differences in cost structure are frequently another source of differences in ceiling adoption rates. For example, use of low-volume irrigation technologies is more widespread in areas where water prices are higher, so that savings in water costs are more likely to outweigh initial investment costs for drip systems (Caswell and Zilberman, 1985; Dinar and Yaron, 1990; Dinar et al., 1992; Negri and Brooks, 1990; Shrestha and Gopalokrishnan, 1993). The use of computer services is more widespread among dairy operators than other kinds of producers because computerization reduces the management time and cost involved in herd improvement (Huffman and Mercier, 1991; Putler and Zilberman, 1988). Information-intensive technologies such as low-volume irrigation or chemi- gation frequently increase yield. These yield increases are worth more for higher- value crops, suggesting that ceiling adoption rates of information-intensive tech-
68 PRECISION AGRICULTURE IN THE 21ST CENTURY nologies will be greater for these crops. For example, adoption of drip irrigation systems has been greater for higher-value crops (Caswell and Zilberman, 1985; Dinar et al., 1992), as has the use of CIMIS for irrigation management (Parker and Zilberman, 1996). Adoption of drip irrigation for sugar in Hawaii has been greater on farms where yield differentials between drip and sprinkler systems are higher (Shrestha and Gopalokrishnan, 1993). Adoption of center-pivot irrigation systems in the High Plains has been greater for corn than wheat or sorghum be- cause corn yields are more responsive to irrigation (Lichtenberg, 1989). New agricultural technologies may spur expanded production of specific crops or livestock. Thus, the number of potential adopters of these new technolo- gies may exceed the number of producers currently in the industry. Similarly, the extent of adoption (i.e., as measured in acreage on which the technology is used) may exceed the current size of the industry. New farming methods have fre- quently allowed expansion of cultivation into areas that were previously consid- ered unsuitable. Examples include irrigated corn production on the sandy soils of the High Plains through use of center-pivot irrigation (Lichtenberg, 1989) and expansion of orchards and vineyards onto hilly areas in California through use of drip and sprinkler systems (Caswell and Zilberman, 1986). New agricultural technologies may also turn out to have advantages that were largely unanticipated when they were introduced. As a result, ceiling adoption may differ significantly from initial expectations. For example, sprinkler irriga- tion is used in preference to drip systems for citrus in California because it pro- vides frost protection (Caswell and Zilberman, 1985). Frost protection is simi- larly an important motivation for the use of CIMIS, which was designed mainly to improve irrigation scheduling (Parker and Zilberman, 1996). Determinants of the Speed of Diffusion of New Technologies The speed at which the diffusion of new technologies occurs depends on a variety of factors, including: â¢ how rapidly information about new technologies spreads; â¢ how risky the new technology is perceived to be and how rapidly those perceptions change; â¢ how rapidly producers can adapt to using new technologies, which de- pends in turn on education and other forms of human capital; â¢ how rapidly learning-by-doing and learning-by-using occur; and â¢ how rapidly producers of the technologies improve reliability, cost, and ease of use. (See, for example, David, 1975; Feder, 1980; Feder and OâMara, 1982; Feder et al., 1982; Jensen, 1982; Mansfield, 1963; Stoneman, 1981). Less is known about the performance of newly introduced technologies be- cause producers lack experience using them. As a result, the new technologies
ADOPTION OF PRECISION AGRICULTURE 69 tend to be viewed as less productive, more costly, and riskier than established technologies. Initially, only a few producers will find using them worthwhile. These early adopters tend to have greater human capital or be more accepting of risk. Other producers may believe that the new technologies will not be profitable enough to justify the cost of adopting them (the cost of new equipment, plus the costs of restructuring farm operations and training) or may be too averse to risk to adopt the new technologies even if they appear to be more profitable (until the expected improvement in profit is large enough to outweigh the risk). Risk aver- sion on the part of lenders may also prevent early adoption by limiting producersâ ability to borrow money to invest in new technologies. Aversion to risk may also lead producers to test new technologies partially at first (i.e., trying them out on a field or trying out key components of a system). Larger operators may be more diversified against risk and thus be more willing to test out new technologies partially. Finally, some producers may find the new technologies less profitable than current technologies given the productivity of their existing capital stock (Salter, 1960). As producers gain experience with the use of the new technology, either directly or through improved information, estimates of the return to using them tend to rise whereas perceptions of their risk tend to fall. Perceptions of risk and return for new technologies change over time in response to several factors. The first such factor is the rate at which information about a new technology spreads to potential users. The classic model treats the process of information flow as proceeding largely through word-of-mouth, so that it mimics the spread of an epidemic. The greater the potential profitability of a new technology, the faster information about it is disseminated (Mansfield, 1963). In agriculture, tech- nology-transfer programs within cooperative extension services and marketing by equipment manufacturers, dealers, and consultants all work to spread informa- tion about new technologies. Potential profitability affects the flow of informa- tion and thus the speed of diffusion through its effect on marketing efforts. Tech- nologies that promise greater increases in returns tend to be marketed more aggressively, so that information about them reaches potential users more rap- idly. Once the profitability advantages of a new technology are well established, lenders may also promote adoption of it to their clients. Marketing can also have negative effects on the spread of accurate informa- tion. Overzealous marketing can create inflated expectations about performance. Producers disappointed by the gap between promise and performance in initial trials may come to underestimate the potential profitability of new technologies. For example, the gap between inflated claims about drip irrigation in the 1970s and actual performance of drip systems produced an adverse reaction among pro- ducers that significantly retarded the spread of this technology. The second factor in changing perceptions is the userâs direct experience with the technology. As producers use the new methods they may come to per- ceive them as more profitable and less risky. In some cases increased familiarity
70 PRECISION AGRICULTURE IN THE 21ST CENTURY alone changes perceptions. The spread of computer use in agriculture illustrates this learning-by-using. Producers began with simpler programs (spreadsheets and word processing) and then branched out into more sophisticated management systems after becoming more familiar with computers in general (Putler and Zilberman, 1988). In other cases, user experience with new technology can lead to improvements that actually make the technology more profitable to use, rather than simply changing perceptions about profitability. For example, the perfor- mance of early drip irrigation systems was often poor because drip lines clogged. User experience with clogging problems led manufacturers to redesign the drip systems, which led in turn to accelerated diffusion of these systems (for example, Shrestha and Gopalokrishnan, 1993). Third, the cost of producing and installing equipment for the new technology frequently falls over time because of improvements in technology design and cost reductions resulting from production experience of the manufacturers (i.e., learn- ing-by-doing). Lichtenberg (1989), for example, showed that increases in irri- gated soybean and corn production in the Northern High Plains were due in part to falling costs of center-pivot irrigation systems. Kislev and Shchori-Bachrach (1973) argue that learning-by-doing played a critical role in the spread of winter vegetable production in Israel. Initially, only the most skilled producers were able to produce winter vegetables. Over time, these producers developed standard methods for vegetable cultivation under Israeli conditions, which less-skilled pro- ducers were increasingly able to use. Finally, the profitability of a new technology relative to existing ones may rise over time as the existing capital stock ages and becomes less productive (Salter, 1960). As producers replace their existing equipment, they tend to invest in equipment embodying the new, more productive technologies. Information-intensive technologies may have high fixed costs, either in terms of equipment purchases or in terms of acquiring the skills necessary to use them. The economic theory of investment suggests that adoption of such technologies may be discontinuous over time, because uncertainty about future conditions makes waiting a preferred option (Dixit and Pindyck, 1993). In some cases rapid diffusion has occurred after periods of extreme conditions resulted in substantial alterations in expectations about future prices. Examples include the rapid diffu- sion of drip irrigation in California during the droughts of 1976â1977 and 1988â 1991, when water prices rose sharply and water availability declined sharply; conservation tillage during the energy crisis years of the 1970s, when energy prices rose sharply; and center-pivot irrigation in the High Plains after the grain price spikes of 1973â1974 (Lichtenberg, 1989). It is important to recognize that producers are buying the services of the technology, not the equipment that embodies the technology. Purchasing those services does not necessarily require purchasing the equipment, although equip- ment purchase may be the most economical means for some. There are many other ways of packaging and selling those services, such as equipment rental,
ADOPTION OF PRECISION AGRICULTURE 71 custom operations for hire, or consultant services. For example, in early years, sellers of drip irrigation systems rented equipment to producers rather than re- quiring its purchase. As noted above, scouting services are typically delivered by crop consultants, in the form of pest management recommendations. Laser level- ing, which requires investment of substantial sums in machinery, is typically pro- vided by custom service similar to grain harvesting. Effective and profitable use of information-intensive technologies such as precision agriculture requires high human capital, which thus plays a critical role in the technology diffusion process. The greater the availability of such human capital, the more rapid the spread of technology. Producers with greater human capital are more likely to adopt new technologies, as has been documented for computers and computer software (Huffman and Mercier, 1991; Putler and Zilberman, 1988) and IPM (Fernandez-Cornejo, 1996; Fernandez-Cornejo et al., 1994; Harper et al., 1990; Napit et al., 1988; Wearing, 1988). Human capital can consist of producers themselves if they have the neces- sary education or computer training, as documented in the studies mentioned above. Alternatively, producers may rely on consultants who make recommenda- tions derived from the use of these new technologies. Operations that are suffi- ciently large may hire specialized workers to provide these services. For example, in 1994, 29 percent of cotton producers performed field scouting themselves, 48 percent relied on consultants for scouting services, and 5 percent used employees (Economic Research Service, 1995b). Similarly, diffusion of precision agricul- ture has been more rapid in areas with larger numbers of dealers and consultants to provide advice and service (Wolf and Nowak, 1995). In the short term, lack of access to consultant services may limit the speed of diffusion of new technologies. Over time, however, consultant and training ser- vices should become more available in areas with high demand, as chemical and equipment dealers and independent consultants respond to opportunities to start up or expand business operations. The public sector frequently also plays an im- portant role in training and thus augmenting the supply of human capital in agri- culture. For example, land grant universities have provided the basic training in crop and pest management sciences underlying IPM, and extension agents in many states provide specific pest management training needed to implement IPM (see, for example, Carlson, 1980; Zilberman et al., 1994). The preceding discussion suggests that the diffusion of precision agriculture technologies may proceed unevenly over time. Like computers, precision agri- culture initially is likely to require significant learning, both by users and by equipment suppliers. Time may be needed for these learning effects to improve performance, reduce costs, and increase reliability sufficiently to make this tech- nologies attractive to large numbers of producers. Precision agriculture requires significant supporting infrastructure in the form of skilled labor, software devel- opment, and hardware availabilityâall of which may take time to develop. More- over, some precision agriculture equipment is costly, so that potential usersâ
72 PRECISION AGRICULTURE IN THE 21ST CENTURY whether producers or crop consultantsâmay prefer to wait until uncertainties about costs and reliability have been reduced sufficiently to make investment attractive. Long-Term Potential of Precision Agriculture The long-term potential of precision agriculture depends on its profitability (i.e., the extent to which benefits, such as of increases in yield and savings in input costs, outweigh the cost of purchasing the services). Available evidence indicates that the cost of many precision agriculture services is modest, especially when spread over sufficient acreage, as can be done by a custom applicator (Table 3-1). These costs are comparable to the fees charged for insect and weed scouting on major crops, which are about $3 to $6 per acre. Remote sensing imagery simi- larly has a modest cost when spread over sufficient acreage. Raw satellite imag- ery data can cost as much as $80,000 per season, but data providers can sell processed images for specific fields for as little as $7 to $8 per acre (Lamb, 1996). By way of comparison, fertilizer, lime, and gypsum costs for corn in the United TABLE 3-1 Estimated Costs of Precision Agriculture Services Item Cost per Acre Source Farmer Costa Grid soil sampling $3-7 Lowenberg-DeBoer and Swinton (1997) (plow depth, 3-acre grid) Giacchetti (1996) Grid soil sampling $16-22 Berglund and Freeburg (1995) (4-foot depth, 3-acre grid) Yield monitor $1.45-1.66 Lowenberg-DeBoer and Swinton (1997) GPS receiver $0.75-1.45 Lowenberg-DeBoer and Swinton (1997) Scouting package, weekly $4 Giacchetti (1996) VRT controllers, various applicators $1-5 Lowenberg-DeBoer and Swinton (1997) Variable-rate fertilization application $3-7 Lowenberg-DeBoer and Swinton (1997) (additional cost over uniform) Giacchetti (1996) Dealer Costb DGPS receiver $0.23-0.79 Kohls (1996) Grid soil sample unit $0.62-1.60 Kohls (1996) Yield mapping computer and software $0.33-1.16 Kohls (1996) Liming application unit $1.09 Kohls (1996) VRT fertilization unit $0.22-10.00 Kohls (1996) aAssumes 3-year useful life for equipment, 6 percent interest rate, 3 percent repair cost, and 1,000 acres. bAssumes 3-year useful life for DGPS receiver and yield mapping computer and software, 5-year useful life for liming application unit, 10-year useful life for soil sampling and VRT equipment, 6 percent interest rate, 3 percent repair cost, and 5,000 acres.
ADOPTION OF PRECISION AGRICULTURE 73 States averaged $46 per acre in 1994, custom operations averaged $10 per acre, and total variable cash expenses averaged $147 per acre. Capital replacement on corn averaged $33 per acre (Economic Research Service, 1996). The central characteristic of precision agriculture is the use of detailed infor- mation to reduce the impact of heterogeneity of production conditions on output by allowing producers to calibrate inputs according to conditions at the subfield scale. For example, growers can combine tensiometer information on soil mois- ture at subfield levels with prediction of evapotranspiration derived from weather forecasts hourly or daily, so that producers can vary irrigation water application to match water demand at the subfield level. Similarly, variable-rate applicators combined with fertility mapping allow producers to vary fertilizer application rates in response to natural variations in soil fertility. Calibrating inputs accord- ing to conditions at subfield levels is likely to result in increased yields. For example, diminishing marginal productivity of nutrients suggests that yields ob- tained under variable-rate fertilizer application will generally exceed those ob- tained under uniform application calibrated according to average soil fertility. Within variable-rate application, areas within a field having higher than average fertility should receive lower fertilization than under uniform application, while areas having lower than average fertility will receive higher fertilization than under uniform application. Yields in areas with higher-than-average fertility will be lower under variable-rate application than under uniform application, while yields in areas with lower-than-average fertility will be higher. As long as low- fertility areas account for a sufficiently large share of the field, variable-rate fer- tilizer application will result in an increase in yield for the field as a whole. The preceding discussion suggests that precision agriculture is likely to have a greater profitability advantage than current farming methods in areas where production conditions are more heterogeneous and in areas where input costs are higher, because cost savings from more precise input application are likely to be greater in such cases. Similarly, precision agriculture is likely to have a greater profitability advantage than current farming methods for higher-value crops be- cause yield increases resulting from more precise input application (should they occur) are worth more in such cases. The handful of peer-reviewed, published economic assessments have exam- ined the relative profitability of varying fertilizer application rates on small grains in response to differences in natural soil fertility. These studies thus involved relatively low-value crops and inexpensive inputs, conditions under which preci- sion agriculture technologies should have relatively small profitability advan- tages. Carr et al. (1991) obtained mixed results when comparing profitability of wheat and barley grown in central Montana between fields where nitrogen, phos- phorus, and potassium were applied uniformly and those where application rates of these nutrients were varied to meet recommended application rates for yield goals on different soil types. Returns net of variable costs were significantly higher under variable-rate applications than under uniform-rate applications in
74 PRECISION AGRICULTURE IN THE 21ST CENTURY BOX 3-1 The Paradox of Information Technology and Its Economic Effects Despite the vast sums of investment and the enormous allocation of human talent devoted to the application of information technology in busi- ness, it has been a continual challenge to quantitatively document the economic benefits of information technology. This evaluation problem occurs at the level of the individual firm, the industry, and the economy. Although examining the overall effectiveness of all types of information technology applications is clearly not the purpose of this report, recent research results on this topic can provide lessons to shape our expecta- tions of the likely economic effects of precision agriculture. Innovative applications of information technology tend to attract con- siderable media attention, often based upon anecdotal evaluation. For example, American Airlineâs Sabre reservation system received consid- erable attention not only for gaining advantage for American Airlines but also for altering the way competition occurred in that industry (Buday, 1986). Although instructive and useful, anecdotal and case study evi- dence can tend to be overly optimistic regarding the eventual economic impacts of such innovations. Often, the time and costs associated with the failed projects that were necessary to achieve the successful projects are not accounted for (Leonard-Barton, 1995). Critical examination of reports of information-technology-based success, which do not include careful economic evaluation, is appropriate. Clemonsâs 1986 comments still apply to reports of innovations in information technology in general, as well as for the technologies associated with precision agriculture: âSurely much is media hype or current business fad . . . . There is now a large, and largely anecdotal, literature, most of it referencing similar sto- ries of technologically directed competitive triumphs. How much do we understand? . . . . How many of the stories are true, or accurately re- ported?â (Clemons, 1986). The economic gains from innovative applications of information tech- nologies tend to be spread over several years. This may occur in part because the effects on performance are not immediate. Further, the ef- fects of learning often are critical in determining eventual benefits. Here learning includes becoming proficient in how to operate the technology. More importantly, organizational learning is required to discover how to alter business systems to take full advantage of the innovation. Peffers and Dos Santos (1996), for example, examined the impact of the intro- duction of ATM machines across 2,534 U.S. banks over the period 1974 to 1984. Their findings indicate that the effects on business performance of ATM adoption can be best described by exponential and logistic mod- els, implying that the benefits were small at first but increased rapidly
ADOPTION OF PRECISION AGRICULTURE 75 BOX 3-1 Continued after a few years. Their study stresses the importance of longitudinal analyses, as empirical cross-sectional studies conducted soon after ini- tial implementation would tend to report minimal benefits even if the even- tual benefits were large. Further, the results of this analysis stressed that long run enhancements in performance did accrue to the early adopters of the ATM technology. Reports of information technology innovation tend to stress the poten- tial for innovating firms to gain sustainable competitive advantages in their markets. However, the widespread use of information technology throughout business and society has increased the chance for rapid and successful imitation of innovations based upon information technology. Kettinger et al. (1994) found that five years after information technology innovation, 21 of 30 firms had suffered competitive declines in market share, profit, or both. A recent analysis by Hitt and Brynjolfsson (1996) carefully separates the effects of information technology on productivity, business profitability, and consumer surplus. This study examines 367 large U.S. firms for the period 1987-1991. Their findings indicate that spending on information technology resulted in increased productivity and increased consumer value but unchanged business profitability. These results conform to the notion that adoption of information technology may tend to be more of a strategic necessity than a source of differential sus- tainable advantage (Floyd and Wooldridge, 1990; Kettinger et al., 1994). Although evaluating the effect of different levels of spending for infor- mation technology can be instructive at an aggregate level, doing so pro- vides little guidance to individual managers nor does it explain why firms that adopt similar information technology innovations experience differ- ing levels of success. Exploring the role of information technology in the U.S. retailing sector, Powell and Dent-Micelle (1997) show that adoption of information technology innovations alone is not sufficient to explain differential firm performance. However, these findings demonstrate that some firms have achieved successful performance by leveraging infor- mation technology with intangible, complementary human and business resources. Building upon the resource-based theory of strategy (Barney, 1991; Rumelt, 1987; Teece, 1987), these results identify the interaction between resources (such as a corporate culture, integration of strategy and information technology implementation, and supplier relationships) with information technology innovation. Kelley (1994) documented simi- lar interactions between the specific resources setting of individual manu- facturing plants and the effectiveness of information technology innova- tion in those operations. continued on next page
76 PRECISION AGRICULTURE IN THE 21ST CENTURY BOX 3-1 Continued The preceding studies are from settings and situations that are mark- edly different than those of crop production. However, because aggres- sive adoption of information technology has been under way for some time in those other settings, they provide insights that are potentially rel- evant for the adoption of precision agriculture, as follows: â¢ Initial reports of innovation with information technology are likely to be based upon anecdotal evidence, which may lead to overly opti- mistic expectations and fail to accurately assess economic costs and benefits. â¢ Learning, and the time required to learn, is often essential to achieve total benefits. Therefore, slow rates of adoption should be expected in early years and cross-sectional economic analyses done soon after discovery are likely to understate long-run benefits. â¢ When barriers to imitation are relatively low, information technology based innovations may result in increases in productivity but no change in industry level profitability. In this situation, however, con- sumer well-being increases. â¢ Information technology innovations are adopted within the context of individual firms and effectiveness is significantly affected by the complementary human and business resources of each firm. only one of five field trials; there were no significant differences in returns in the remaining four. In the fifth trial, the increase in returns from variable-rate appli- cation was sufficient to cover likely costs of soil testing (see, for example, Loewenberg-DeBoer and Swinton, 1997). In four out of the five field trials, how- ever, variable-rate application was less profitable once the costs of soil testing were included. Fiez et al. (1994) obtained similar results in a study of wheat production in eastern Washington. Net returns with variable-rate nitrogen application, where the rate was based on a constant nitrogen application rate per bushel of yield goal, were roughly equal to net returns with conventional uniform nitrogen application, where the application rate was determined the same way; inclusion of soil testing costs would make the variable-rate system less profitable. However, when vari- able nitrogen application rates were calibrated according to a non-linear nitrogen- response relationship, the variable-rate system was clearly more profitable, even with soil testing costs taken into account. Wibawa et al. (1993) also found that, because of the cost of soil sampling, uniform nitrogen and phosphorus applica- tion on barley and wheat in eastern North Dakota had lower yields but higher profits than variable-rate applications calibrated to meet standard yield goals.
ADOPTION OF PRECISION AGRICULTURE 77 Recent commercial developments also support the notion that interest in precision agriculture is likely to be greater on higher value crops. For example, sugar beet quality is sensitive to nutrient application and can suffer from both nutrient deficiencies and excesses. American Crystal Sugar Company, the largest sugarbeet producer in the United States, is owned by more than 2,000 North Dakota and Minnesota sugarbeet growers. This cooperative has increased its grid soil sampling in Minnesotaâs Red River Valley from 13,000 acres in 1995 to 130,000 acres (about 35% of the companyâs sugar beet acreage) in 1996 (Lilleboe, 1996). Precision agriculture technologies may have important ancillary uses. For example, the purchase of remote sensing information on crop growth may permit closer marketing ties between producers and grain elevators by improving yield forecasts. As a result, even producers who experience relatively little within-field yield variability may find their share of the increased returns from improved mar- keting sufficient to make the use of such information profitable. Evolution of Precision Agriculture Precision agriculture services may be provided directly (i.e., for custom hire) or may be purchased in the form of hardware and software products embodying those services, or as combination of products and services. In all likelihood, both forms of obtaining precision agriculture services will coexist, with the exact com- bination depending on such factors as the size of the operation, the technical expertise of the operator, and the density of the local market for services. Provision of Precision Agriculture Services New services related to precision agriculture could arise independently but are more likely to evolve from services now offered by crop consultants and input suppliers. Nowlin (1993) estimates that crop consultants provide knowledge- based services on 16 percent of U.S. crop acreage, including 53 percent of cotton and vegetable acreage, 21 percent of corn acreage, and 13 percent of soybean acreage. A study of service offerings by 228 Wisconsin agricultural chemical dealers showed that although precision agriculture services accounted for only a small portion of total gross revenues, the percentage of firms offering these ser- vices was increasing (Wolf and Nowak, 1995). A Purdue University survey of agricultural chemical dealers showed that between 39 and 47 percent of firms offering site-specific services were charging no fee for the service, folding costs into product prices instead. Input suppliers perceive impediments to providing precision agriculture services to include the cost of equipment (61 percent), mismatches with the kind of farming practiced in the area (18 percent), difficulty demonstrating value to producers (11 percent), and the need to train personnel (6 percent) (Akridge and Whipker, 1996).
78 PRECISION AGRICULTURE IN THE 21ST CENTURY Precision agriculture services can be provided to the producer through tradi- tional distribution systems or by consultants. If services are provided directly to the producer, a consultant could design, integrate, and install a precision agricul- ture system (i.e., a combination of a global positioning system [GPS], yield moni- tor, and geographic information system [GIS]) for the operation, much as a com- puter consultant assists a small business with its computing needs. Alternatively, traditional input suppliers could be the primary customer, acting as general con- tractors for specific skills, expertise, and services provided by the technology consultant. If they have access to sufficient financial capital, technology consult- ants could invest in the computers, software, and equipment that suffer from rapid technological obsolescence. If the consultant leases precision technologies to the input supplier, then the producer can have access to the most advanced equip- ment. In return for their investment, the technology consultant can depreciate the capital costs over a larger acreage base. A variation of this scenario could be a technology consultant who either competes with or supplements the services of accountants, marketing consultants, or other financial assistants in automating all of the farmâs electronic accounting and record-keeping functions. Precision agriculture services will likely be provided locally. For example, computer software such as nutrient or pest management recommendation models will need to be adapted to local (even farm-level) conditions (which tend to vary substantially). Firm size will likely be small because the number of producers in a local area would be limited. However, hardware and software support channels for the service providers could be concentrated at the wholesale level. Provision of Precision Agriculture Products An alternative pathway of development of precision agriculture assumes that producers and their crop consultants will buy software and hardware products to implement precision agriculture, thus limiting their purchases of services. That is, the expertise and knowledge needed for precision farming will be embodied in machinery and software to a greater extent than in the services provided by con- sultants. Integrating the various hardware and software components into one sys- tem would provide a seamless flow of data, culminating in either a recommenda- tion or the presentation of alternatives to the producer. Several manufacturers (i.e., John Deere, Case, Rockwell, Ag Chem, and Crop Technology, Inc.) have developed integrated turnkey systems that combine GPS receivers, GIS software, crop yield monitors, and VRT hardware into precision agriculture systems. Object-oriented software modules could facilitate develop- ment of products, including software for mapping and decision support (Environ- mental Systems Research Institute, 1996; Macy and Dondero, 1996). Companies providing precision agriculture products may not need to be lo- cated in most communities. Information technology companies could be concen- trated in technical centers such as Silicon Valley or near universities where tech-
ADOPTION OF PRECISION AGRICULTURE 79 nical expertise is readily available. However, hardware and software support chan- nels would need to be widespread and located at the retail level. Moreover, both software and hardware will likely need to be customized to match local condi- tions, creating a demand for locally based customizing expertise. This expertise could be provided through traditional farm supply sources, such as machinery dealers (especially for proprietary lines). Independent consultants may also prove economically viable in the market for hardware and software customization ser- vices, much as they have in various aspects of computer hardware and software sales, support, and training in other industries. Firm servicing the product market could be large in size and could dominate markets, especially if developed by or acquired by large industry leaders. Combination of Products and Services In reality, precision agriculture is not likely to develop purely as a service or as a product but probably will evolve as a combination of services and products. For example, individual producers may purchase and use some precision agricul- ture hardware and software themselves. Other producers could hire consultants who use the same or similar hardware and software. Still others may purchase the hardware and hire consultants to analyze the data and make recommendations. For example, growers with medium-sized operations and a high level of technical expertise will likely purchase a larger share of precision agriculture services in the form of hardware that they operate themselves. Part-time operators will likely purchase more custom-hired services because of constraints on their time and capital. Very large operators may both purchase equipment and hire full-time employees with the requisite technical expertise. EFFECTS OF WIDESPREAD ADOPTION OF PRECISION AGRICULTURE Three potential effects of precision agriculture have received public atten- tion. First, widespread adoption of precision agriculture may affect employment opportunities in rural communities. Second, precision agriculture may affect the structure of agriculture, particularly the distribution of farm sizes. Third, more efficient and precise use of fertilizers, pesticides, and other purchased inputs may alleviate environmental spillovers from agriculture. Effects on Rural Employment Precision agriculture represents an increase in specialization, that is the divi- sion of labor, involved in agriculture. In adopting precision agriculture techniques, farmers substitute purchased information services (in forms as varied as consult- ant services, specialized equipment, and computer software) for some or all of
80 PRECISION AGRICULTURE IN THE 21ST CENTURY their own personal-observation-based information collection and analysis. Diffu- sion of precision agriculture is thus likely to result in increased employment in agricultural support services, for example, equipment sales, computer software development, customization of equipment and software, and consultant services. The extent to which such increased employment occurs in rural areas depends largely on the combination of direct services and turnkey products through which precision agriculture services are delivered. The greater the extent to which grow- ers purchase precision agriculture services directly, the greater will be the de- mand for locally based skilled technical labor such as crop consultants and com- puter software developers and customizers. Any increases in employment in the production of turnkey precision agriculture products are more likely to occur in areas where equipment manufacturers and software developers are currently lo- cated, which are largely non-rural. Increases in rural employment caused by the spread of precision agriculture are likely to be modest. Farm-related employment is presently quite limited. The farm sector as a whole provides an estimated 1.7 million jobs, or 1.3 percent of total U.S. employment (Edmondson et al., 1996). On-farm labor accounts for at least 800,000 of those jobs (National Agricultural Statistics Service, 1996), so that employment in farm services of all kinds is at most 900,000. Precision agri- culture is unlikely to generate substantial additional labor demand. Some preci- sion agriculture equipment (i.e., variable-rate applicators) replaces other forms of equipment; neither the manufacture nor the sales of such equipment will require expanded employment. Sales of turnkey products will likely not require increases in sales personnel, although more skilled personnel may be needed to service such products. Software development and customization do not generally require extensive increases in employment. Above all, the size of the market for precision agriculture products (both equipment and services) in the United States is limited. In 1994, there were an estimated 1.4 million households containing farm operators or managers in the United States (National Agricultural Statistics Service, 1996) and an estimated 2.1 million farm operations (Economic Research Service, 1996). But only a small percentage of those farm operations generate significant demand for farm equip- ment and services. In 1994, for example, 122,000 farm operations accounted for over 51 percent of total cash expenditures in U.S. agriculture and almost 58 per- cent of net cash income. An additional 224,000 farm operations accounted for over 22 percent of cash expenditures and 26 percent of net cash income. Thus, a total of 346,000 operations accounted for 74 percent of total farm cash expenses and 84 percent of net cash income, indicating a limited customer base for preci- sion agriculture equipment and services (Economic Research Service, 1996). The sales volume of agricultural equipment like that used in precision agriculture is similarly limited. In 1990, less than 25,000 pieces of crop harvesting equipment and about 50,000 power sprayers and dusters were shipped in the United States. In 1991, farmers spent $604 million on planting and fertilizing machinery, $2,158
ADOPTION OF PRECISION AGRICULTURE 81 million on harvesting machinery, $167 million on cultivators and weeders, and $299 million on sprayers and dusters of all kindsâa total of $3.2 billion on the kinds of machinery that precision agriculture is likely to affect (National Agricul- tural Statistics Service, 1996). A market this limited in size is unlikely to support increases in employment that are large on a national scale, even if some rural communities eventually do experience substantial employment growth. As noted above, any gains in employment with precision agriculture will likely require workers with greater skills who would earn higher wages than is typical in rural communities. Effects on the Structure of Farming Some observers have argued that technological change, particularly the dif- fusion of information-intensive technologies such as precision agriculture, may increase concentration and further reduce the number of family farm operations (see, for example, Office of Technology Assessment, 1986). In an industry as competitive as agriculture, farm operations using less-profitable technologies are unlikely to remain in business. Producers using less-productive management approaches and supporting technologies may find it difficult to cover their pro- duction costs. Some producers who have difficulty adapting to new, technologi- cally advanced farming methods may find it more lucrative to sell or rent their land than farm it themselves. A hallmark of precision agriculture is the application of information to im- prove farm decision making. Precision agriculture thus tends to confer a competi- tive advantage on growers more at ease with or more highly skilled in the use of modern information technologies, who are likely to be younger and have more formal education. The spread of precision agriculture may accelerate older, less well-educated farmers into retirement from agriculture. It is important to recog- nize in this connection that exit from farming does not necessarily imply hard- ship, at least for farmers owning land. If precision agriculture is more profitable, its use should lead to increases in the value of land. Those exiting farming by selling or renting their land should thus be able to claim at least a share of the increased profit generated by precision farming methods. It is also possible that the costs and benefits of precision agriculture vary systematically across farm size. There is concern in particular that the net benefits of precision agriculture are greater for larger operations. It is difficult to assess how the net benefits of precision agriculture vary across farm size because preci- sion agriculture is a suite of technologies whose specific components differ across farms. In general, however, there does not appear to be an unambiguous size bias in precision agriculture or similar technologies. Some considerations suggest that smaller operators will benefit less from precision agriculture than will larger operators. Precision agriculture substitutes somewhat for manual information collection and processing by the individual
82 PRECISION AGRICULTURE IN THE 21ST CENTURY producer. Producers whose operations are small enough to allow them to know all parts intimately may benefit very little from automating information collec- tion. Precision agriculture allows producers to reduce heterogeneity in farming operations, which permits improved profitability through reduced input expendi- tures, increased yields, or both. An alternative strategy to reduce heterogeneity, however, is to divide the farm into fields that are relatively homogeneous, permit- ting the same sorts of gains from uniform management. Producers whose opera- tions are of a size that permits such intensity of management are unlikely to gain much from automation of information gathering and processing. Other factors suggest the oppositeâthat smaller operators may gain more from precision agriculture than larger ones. The number of small farm operations is large. In 1992, 35 percent of all farm operators worked off the farm 200 or more days per year, whereas an additional 8 percent worked off the farm 100 to 199 days per year (U.S. Bureau of the Census, 1992). Automation may be more valuable to smaller, part-time producers, who operate under tight labor constraints and may otherwise be unable to increase their monitoring of field conditions. Automation may also narrow differentials in management by allowing smaller operators to access the same sophisticated strategies as larger operators. How- ever, the current competitive advantage of some smaller operatorsâbeing able to focus on individual plants in high-value cropsâmay be lost when the adoption of information-intensive approaches enables larger operators to do the same. There is similarly no clear-cut evidence that precision agriculture exhibits substantial economies of scale, so that larger operators find precision agriculture less costly on average (i.e., per acre) than smaller operators. Equipment such as variable-rate application units, GPS receivers, and computer hardware may de- crease average costs because their capital costs can be spread over larger acreage. The data in Table 3-1 indicate that such economies of scale tend to be limited, however. Moreover, some smaller operators may be able to take advantage of economies of scale by hiring out themselves and their equipment for custom ser- vices. The cost of soil sampling depends on grid size, which in turn depends on the heterogeneity of soils within a field. Larger operators are more likely to have larger, more heterogeneous fields, and may thus find soil sampling more costly. Crop consultants typically charge a flat per-acre fee for their services, in which case the cost of these services does not vary with farm size. Experience with similar information-intensive technologies fails to support the notion of a consistent size bias. As noted, the per-acre cost of scouting (a central component of IPM) is comparable to the per-acre cost of many precision agriculture technologies. The literature shows no consistent relation between farm size and the use of scouting. For example, Napit et al. (1988) examined factors influencing the use of scouting throughout the United States. They found that the value of farm sales was positively correlated with scouting on corn in Indiana, apples in New York, cotton in Texas and Mississippi, and alfalfa in the North- west, but not correlated with scouting on soybean fields in Virginia, peanuts in
ADOPTION OF PRECISION AGRICULTURE 83 Georgia, apples in Massachusetts, or tobacco in North Carolina. Fernandez- Cornejo et al. (1994) found that larger vegetable producers in Florida and Texas were more likely to scout, but that farm size had no significant relationship to scouting on vegetables in Michigan, a finding they attributed to greater familiar- ity with IPM in Michigan. Fernandez-Cornejo et al. (1994) also found that farm size was not correlated with scouting for either insects or diseases for tomatoes in Florida. Additionally, Harper et al. (1990) found that neither the number of rice fields operated nor the acreage per field influenced adoption of sweep nets for monitoring stink bug pests on rice in Texas. Low-volume irrigation involves much higher investment in equipment than most precision agriculture technology. Moreover, low-volume irrigation exhibits some economies of scale (i.e., pump size does not necessarily increase as farm size increases), suggesting that the returns to investment in this technology are likely greater for larger operations than smaller ones. Nevertheless, the literature gives no clear indication that larger farms are more likely to adopt this technol- ogy. Adoption of low-volume irrigation technologies has been more prevalent on larger sugar plantations in Hawaii (Shrestha and Gopalokrishnan, 1993), larger tree crop operations in the southern San Joaquin Valley of California (Green et al., 1996), and larger farms on the west side of the San Joaquin Valley (Dinar et al., 1992). By contrast, neither adoption of sprinkler irrigation in the western United States (Negri and Brooks, 1990) nor adoption of modern irrigation tech- nologies in citrus groves in Israel shows a significant relation to farm size. For both low-volume irrigation technology and scouting, farm size appeared to be positively correlated with adoption in areas which were in the early stages of the technology diffusion process and not correlated in areas in the later stages of the process. For example, low-volume irrigation has been used extensively in Israel for years, whereas interest in low-volume irrigation in California became widespread only during the drought of the late 1980s. Taken together, studies of information-intensive technologies such as low-volume irrigation and scouting show no consistent relationship with farm size over the long term, although they do provide some evidence that larger operators are more likely to be early adopters. Differential access to precision agriculture products and services marketed by consultants is also a controversial issue. Some argue that consultants offer their services preferentially to owners of larger farms. Through its workshops, the committee heard from consultants who indicated that their services were offered on a per-acre basis, suggesting equal access of services to large and small farms. Few data indicate that consultants price their services according to farm size. However, it has been argued that input suppliers in competitive markets may offer their service at no charge or at a discount, so that growers in geographic areas under-served by input suppliers have less incentive to adopt precision agri- culture technologies (Wolf and Nowak, 1995). It is important to put these theoretical concerns in the context of current trends in the structure of farming. The number of farms in the United States has
84 PRECISION AGRICULTURE IN THE 21ST CENTURY declined consistently over time. But the rate of decline has slowed considerably in recent years. Between 1969 and 1982, for example, the number of farms in the U.S. fell from 2.7 million to 2.2 million, an overall 18 percent decline, corre- sponding to an annual average rate of 1.4 percent. Between 1982 and 1994, the number of farms in the U.S. fell to under 2.1 million, a total decline of 8 percent, corresponding to an average annual rate of decline of 0.7 percent. The Office of Technology Assessment (Office of Technology Assessment, 1986) distinguished five size classes of farms. Small farms (annual sales under $20,000) and part-time (annual sales between $20,000 and $100,000) rely prima- rily on non-farm sources of income. Small farms lose money on farming, at least on average. This group includes tax shelters, hobby farms operated as an amenity, and subsistence farms. Part-time farms earn 90 percent or more of their income from off-farm sources. Commercial farms (annual sales over $100,000) are di- vided into three classes: moderate (annual sales between $100,000 and $200,000), large (annual sales between $200,000 and $500,000) and very large (annual sales $500,000 and over). As Table 3-2 indicates, most of the decline between 1982 and 1994 occurred in small and part-time farms, while the number of commercial farms actually increased by 14 percent. Most discussions of farm structure focus on moderate-sized commercial farms. This size category is believed to correspond most closely to the âfamily farmâ that has long been considered the ideal of U.S. agriculture. As Table 3-2 indicates, the number of moderate-size commercial farms increased by 24 per- TABLE 3-2 Trends in Farm Structure, 1982-1994 Number of Farms (Thousand) Percent of Sales Size of Operation 1982 1994 1982 1994 Small 1,355 1,246 5.5 5.9 (Sales Under $20,000) Part-Time 582 472 21.9 17.9 (Sales $20,000-99,999) Moderate 181 224 19.1 23.3 (Sales $100,000-199,999 in 1982 Sales $100,000-249,999 in 1994) Large 94 74 21.0 15.9 (Sales $200,000-499,999 in 1982 Sales $250,000-499,999 in 1994) Very Large 28 48 32.5 37.1 (Sales $500,000 and Over) Total 2,240 2,064 100.0 100.0 SOURCE: Economic Research Service. 1996.
ADOPTION OF PRECISION AGRICULTURE 85 cent between 1982 and 1994. Moderate-size farms increased their share of farm sales as well. Overall, then, the spread of information-intensive agricultural technologies (i.e., computers, drip irrigation, IPM, precision agriculture) over the past 10 to 15 years has coincided with a strengthening of the commercial farm sector, espe- cially its moderate-sized component, at the expense of part-time operations and operations too small to be commercially viable. Processors and Vertical Integration Increased information about market trends may lead to greater vertical coor- dination (i.e., contracting of designer crops and greater vertical integration in livestock industries) and in some cases removal of intermediaries (Beurskens, 1996). Producers may lose independence regarding planting decisions if com- modity buyers (processors) can specify product characteristics (sugar content, size, etc.) and expect their demands to be fulfilled with precision technologies. Some find such increased specialization distasteful, arguing that increased reli- ance on information generated by consultants, input suppliers, and hardware deal- ers necessarily reduces the independence of producers. Environmental Implications Precision agriculture may create a win-win situation by simultaneously im- proving farm profitability and reducing negative environmental effect of agricul- ture. Potential improvements in environmental quality may be a compelling rea- son for society to favor widespread adoption of precision agriculture, but the extent to which producers adopt it depends on economic savings from more effi- cient input use, not on environmental impact. Most of the enthusiasm for precision agriculture off the farm can be traced to the good environmental sense inherent in a concept that matches input applica- tion to plant needs. Precision agriculture could be a more disaggregated version of the kinds of best management practices already recommended at the field scale (Ogg, 1995). Furthermore, precisely matching fertilizer and pesticide inputs to the capabilities and needs of a crop for small areas, and applying them exactly when needed, is a logical way to limit the amounts of these materials added to the environment. Agricultural pollution comes from inputs that do not reach their target (water or nutrients not taken up by plants, herbicides that do not come into contact with weeds, etc.). The inputs that contribute to pollution are thus wasted from a pro- ductivity point of view. Calibrating input usage more precisely should increase the percentage of applied inputs taken up by crops, thereby reducing economic waste and release into the environment. Adjusting fertilizer application rates to match variations in soil fertility at the subfield level should result in less excess
86 PRECISION AGRICULTURE IN THE 21ST CENTURY nutrients and thus less runoff and percolation to groundwater. Applying pesti- cides to only those parts of fields with sufficient pest pressure to warrant treat- ment should result in less excess pesticide in the environment. Calibrating water application rates to match plant uptake rates (i.e., through low-volume irrigation) should reduce excess water applications and thus mitigate drainage problems (Caswell et al., 1990; Dinar et al., 1992; Shah et al., 1995). The extent to which more precise input applications might reduce environ- mental conditions is by no means clear. A number of field-level experimental studies indicate that variable-rate nitrogen fertilizer application can reduce the nitrogen application rate needed to attain given yield levels. But these decreases in application rates have sometimes resulted in little or no discernible decreases in post-harvest soil nitrate concentrations, suggesting few if any, consequent ben- efits in terms of ground or surface water quality improvement (Kitchen et al., 1995; Redulla et al., 1996). Synergy between VRT and biotechnology could also lead to environmental improvements. Biotechnology firms and seed companies are racing to manipulate the genetic makeup of commercial crops to include tolerance to herbicides (Sulecki, 1996). Beginning in the late 1980s, genetically engineered innovations such as imidazolidinone-tolerant corn, sulfonylurea-tolerant soybean, and Buctril- tolerant cotton varieties were developed. The natural insecticidal properties of Bacillus thuringiensis are being incorporated into corn, soybean, and cotton vari- eties. Although these genetically enhanced seeds can confer economic and envi- ronmental benefits when used on a whole field, they are likely to be more effec- tive when applied on a precision basis. For example, if there are yield penalties associated with some of these varieties, they can be variably planted in areas with high weed infestation or areas in which sensors indicate higher organic matter that could be associated with a greater need for pre-emergence herbicide applica- tion. Precision application of seed enhanced by B. thuringiensis could reduce problems with development of resistance that would be accelerated if whole-field application were used. Variable-rate seeding may help match plant populations to soil conditions, and precision seeding with enhanced varieties can be used to apply seed enhanced with B. thuringiensis or herbicide tolerance in potential pest problem areas. Apart from empirical studies, it is possible to envision situations in which precision agriculture both mitigates and exacerbates potential environmental prob- lems associated with crop production. For example, variable-rate fertilization may counteract potential yield differences between steeper shoulder slopes and shal- lower back and foot slopes that were unfertilized or conventionally fertilized (Nolan et al., 1995). Increased soil cover, obtained on steeper slopes through VRT application, could reduce soil erosion, but increased application of nitrogen could increase losses to the environment if yield-limiting factors reduce nitrogen uptake. In another example, areas with droughty soils caused by rapid percolation may have lower levels of soil nitrogen because of greater leaching losses. VRT
ADOPTION OF PRECISION AGRICULTURE 87 nitrogen application could exacerbate leaching if additional nitrogen is applied to counteract losses in these soils, or could mitigate losses if nitrogen were applied in greater synchrony with nitrogen uptake. Uncritical use of precision data indi- cating higher yield potential in certain parts of a field could lead to higher input use, especially if recommendations are formulated without any interest in reduc- ing chemical use. Precision agriculture may make it economically attractive to expand production into marginal lands, which may create new environmental problems. For example, the spread of drip irrigation exacerbated groundwater depletion problems because it facilitated the expansion of tree crop production onto hillsides rather than substituting for less efficient gravity irrigation methods in flat areas. Most studies of the environmental impacts of precision agriculture have con- centrated on field-level effects. But the environmental impacts that matter most are generally those that occur in ambient pollutant concentrations. Unfortunately, there is little empirical evidence currently available that precision agriculture ac- tually reduces delivery of pollutants to surface and groundwater and the atmo- sphere relative to conventional techniques. Moreover, there is good reason to believe that field-level effects do not scale up readily to ambient impacts. Just as with field-level management for potential environmental problems, the effect of reductions in erosion, residual nutrients, and pesticide applications achieved through precision agriculture focused on ambient surface and groundwater qual- ity at the subfield level depends on the position of the field within the watershed and relative to important aquifers. Generally, reductions occurring at great over- land distances from streams and lakes, or away from aquifers contributing to wellheads, will have smaller contributions to environmental improvement than similar reductions occurring closer to these water resources. Spatial patterns of farming activity within the watershed may have more of an impact on environ- mental quality than do improvements in environmental management within farm fields. It is also worth underscoring that environmental improvements by themselves do not generally constitute an incentive for growers to adopt precision agricul- ture. In the long term, potential environmental improvements will constitute an economic incentive for adopting precision agriculture only in areas where pro- ducers bear at least a share of the costs of agricultural pollution. The problem of drainage in the San Joaquin Valley, California, is a case in point. As long as producers were able to dispose of drainage water at low cost into the San Joaquin River and Kesterson Reservoir, they had little incentive to lessen drainage prob- lems by improving irrigation efficiency. After regulatory action resulted in severe restrictions on the use of these low-cost disposal outlets, producers began using low-volume irrigation equipment, leveling fields, shortening irrigation runs, and making other improvements in irrigation efficiency to reduce drainage accumula- tion (Dinar et al., 1992). The use of precision agriculture data for environmental monitoring is also
88 PRECISION AGRICULTURE IN THE 21ST CENTURY potentially a source of contention. The vast amount of site-specific data gener- ated by precision agriculture methods may help allay regulatorsâ concerns about the sources of environmental problems associated with agricultural production in particular areas. However, inappropriate use of such detailed data by regulators can be highly damaging to producersâ interests, and fears of such misuse will undoubtedly make producers reluctant to share such data with government agen- cies. Ways must be found to cooperatively use data generated by precision agri- culture for environmental improvement while safeguarding against abuses or in- appropriate uses. Precision data on nutrient and pesticide soil concentrations and application rates could help producers demonstrate that chemicals were applied in a legally and environmentally sound way, reducing potential liability for environmental problems. Such data could just as easily be used by regulators to prove the oppo- site. Decisions about data ownership and access will determine where the burden of proof lies in use of such data for environmental purposes. CONCLUSION Widespread adoption of precision agriculture may affect rural communities, the structure of agriculture, and environmental quality. Unfortunately, the data currently available are insufficient for judging reliably the effects of precision agriculture on farm profitability, rural employment, or environmental quality. The committee was thus able to arrive only at a few very general conclusions. If precision agriculture becomes widely accepted, employment could increase in rural areas, especially through the provision of services such as crop consulting and software development and customization. Any changes in rural employment are likely to be modest, however. It is clear that training in computer applications will be essential for agronomists. Precision agriculture is one of many factors contributing to change in agri- culture structure. There is concern that small-scale producers will have less ac- cess to consultants providing precision agriculture services. There is no direct evidence indicating differential access. Some precision agriculture technologies exhibit economies of scale, albeit modest ones. Others are characterized by con- stant or decreasing returns in relation to scale. There is thus no unambiguous evidence indicating that precision agriculture favors larger operators. In general, neither economic theory nor experience with earlier information-intensive agri- cultural technologies indicates unambiguously that larger operations will have greater access to or advantage in using technologies such as those characterizing precision agriculture. Moreover, the spread of information-intensive technolo- gies over the past 15 years has coincided with a strengthening of the moderate- size commercial farm sector. The committee found no credible research that contains consistent evidence of environmental benefits from precision agriculture. Current theory suggests that
ADOPTION OF PRECISION AGRICULTURE 89 environmental benefits should be expected in areas where fertilizer inputs are matched to crop needs. It is logical to conclude that precision agriculture should incorporate similar approaches to fertilization. The potential of precision agricul- ture technologies to reduce pesticide applications is still not well known. Some benefits may accrue from localized herbicide treatments. Precision agriculture may also lead to increases in input use, for example, by making profitable the expansion of crop production onto more vulnerable land, or by documenting prior underutilization of fertilizers or pesticides. More research will help to assess the environmental effects of management of small units compared with whole fields. Such research should concentrate on broader-scale effects, however, such as im- pacts at the watershed or ecosystem levels. It is by no means apparent that effects at the field level scale up to broader levels in any readily predictable ways. Finally, experience with previous information-intensive technologies sug- gests that improvements in environmental quality will likely constitute a signifi- cant incentive for farmers to use precision agriculture only if producers bear at least a portion of the costs of agricultural pollution. Precision agriculture is thus unlikely to be a panacea for environmental problems in agriculture in the absence of other regulatory measures.