MSU RETURNING MATERIALS: PIace in book drop to remove this checkout from LIBRARIES 7 1-l-[jlllL your record. FINES wilI be charged if book is ; returned after the date stamped below. RESPONSE OF DRY BEAN ROOTS TO A MANAGEMENT SYSTEM ~DESIGNED T0 ALLEVIATE SOIL RELATED STRESSES. BY Rodney Lynn King A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of MASTER OF SCIENCE Department of Crop and Soil Sciences 1988 ABSTRACT RESPONSE OF DRY BEAN ROOTS TO A MANAGEMENT SYSTEM DESIGNED TO ALLEVIATE SOIL RELATED STRESSES 3:! Rodney Lynn King Field and greenhouse studies were conducted to examine the effects of soil compaction alleviation and other manage- ment factors on dry bean (Phaseolus vulgaris, L.) production. Two crop rotation/tillage systems were studied in combination with irrigation, row spacing, and cultivar variables. Roots were studied by destructive sampling and with minirhizotron observation tubes. A digital image processing system was de— veloped to analyze washed root samples. Soil bulk density was decreased in the plow layer by a deep rooted legume and deep tillage. Soil moisture and aera- tion and pore size distribution were improved in the plow layer by the alternative rotation/tillage management system. Shoot dry weights and root length densities were not affected by treatment combinations. Hinirhizotron root observations were poorly correlated with the destructive sampling results, and it was concluded that the minirhizotron method is not a valid root study tool on fine textured soils with compaction problems. The image analysis system is slow and overestima- ted root length by 23‘ on debris-free samples and dithon sam-' ples with debris. ACKNOWLEDGEMENTS Many people have contributed significantly to this the- sis and I am pleased to acknowledge several of them. I want to express gratitude to my guidance committee. Dr. Alvin J. H. Smucker served as my major professor and his guidance throughout my H. 3. program has been much apprecia- ted. The assistance of Dr. A. Earl Erickson was most help- ful, especially during the weeks of thesis writing and revi- sion. Dr. Ronald L. Perry and Dr. Francis J. Pierce provided valuable assistance in course selection, root analysis tech- niques, and critical review of my thesis. John C. Ferguson provided excellent technical assistance related to the minirhizotron and digital image processing work. The daily exchange of ideas with John was both useful and enjoyable. Fellow graduate students Marie—Claude Fortin and Stephanie Schroeder offered many hours of field and lab- oratory assistance, as well as encouragement and important friendship. I am also indebted to Dr. Amram Eshel and Dr. Juang Vang, both of whom gave freely of their time and exper- tise. Finally, and most importantly, I want to express sincere gratitude to my wife, sandy, who offered constant care, en- couragement, and support throughout this program. 111 TABLE OF CONTENTS LIST OF TABLES LIST OF FIGURES. INTRODUCTION . CHAPTER 1: LITERATURE REVIEW Soils and Crop Production Dry Bean Production on the Fine Textured Soils of the Saginaw Valley of Central Eastern Michigan. Soil Compaction . . . . . . . Bulk Density . Mechanical Impedance and Plant Response. Soil Moisture, Aeration and Plant Response Soil Compaction and Drought Stress Tillage Induced Soil Compaction. Soil Compaction Alleviation . Alleviation by Natural Processes Additional Management Factors Row Spacing. . . . . . . Summary . . . . . . . . . . . References. CHAPTER 2: ALLEVIATION OF SOIL STRESSES ON DRY BEAN ROOT SYSTEMS AS A KEY TO INCREASED DRY BEAN PRODUCTION Introduction. . . . Materials and Methods. iv vii ix 28 32 Results and Discussion Soil Parameters. . . . . . . . . . . . . . . . 40 Soil Moisture. . . . . . . . . . . . . . . . . 50 Plant Parameters . . . . . . . . . . . . . . . 55 Root Responses . . . . . . . . . . . . . . . . 64 Harvest Parameters . . . . . . . . . . . . . . 70 Summary and Conclusions . . . . . . . . . . . . . 83 References. . . . . . . . . . . . . . . . . . . . 87 CHAPTER 3: EVALUATION OF MINIRHIZOTRON OBSERVATION TUBES AS A TOOL FOR ROOT STUDY ON FINE TEXTURED SOILS Introduction. . . . . . . . . . . . . . . . . . . 91 Materials and Methods . . . . . . . . . . . . . . 99 Results and Discussion. . . . . . . . . . . . . . 103 Summary and Conclusions . . . . . . . . . . . . . 116 References. . . . . . . . . . . . . . . . . . . . 118 CHAPTER 4: ROOT LENGTH AND WIDTH DETERMINATION BY DIGITAL IMAGE PROCESSING Introduction. . . . . . . . . . . . . . . . . . . 120 Materials and Methods . . . . . . . . . . 124 Root Extraction and Preparation. . . . . . . . 124 Root Image Recording . . . . . . . . . . . . . 125 Image Analysis . . . . . . . . . . . . . . . . 127 Results and Discussion. . . . . . . . . . . . . . 130 Imaging System . . . . . . . . . . . . . . . . 130 Image Processing . . . . . . . . . . . . . . . 130 Initial Calibration. . . . . . . . . . . . . . 131 Time . . . . . . . . . . . . . . . . . . . . . 139 Tray Size. . . . . . . . . . . 143 Analysis of Debris- Free Roots. . . . . . . . . 144 Analysis of Field Roots. . . . . . . . . . . . 146 Summary and Conclusions . . . . . . . . . . . . . 151 References. . . . . . . . . . . . . . . . . . . . 154 CHAPTER 5: THE EFFECT OF DROUGHT OR STRESS ON GROWTH AND YIELD OF DRY BEANS Introduction. . . . . . . . . . . . . . . . . . . 155 Materials and Methods . . . . . . . . . . . . . . 158 Results and Discussion. . . . . . . . . . . . . . 161 Summary . . . . . . . . . . . . . . . . . . . . . 169 References. . . . . . . . . . . . . . . . . . . . 171 CHAPTER 6: CONCLUSIONS. . . . . . . . . . . . . . . . 173 vi LIST OF TABLES Chapter 2 1. Root length densities of dry beans as affected by rotation/tillage, irrigation, and cultivar. De- structive sampling was in mid-August, which was maximum flowering. . . . . . . . . . Influence of rotation/tillage, irrigation, culti- var, and row spacing on maturity of dry beans as estimated on September 2. Percent maturity was visually estimated as the percent of plants in a plot whose leaves had lost 50% or more of their chlorophyll. . . . . . . . . . . . . . . . . . . . Trend analysis for components of the irrigation- row spacing-rotation/tillage interactions for plant maturity as estimated on Sept. 8. Each line tests significance of row spacing on maturity under given treatments . . . . . . . . . . . . Influence of rotation/tillage, irrigation, culti- var, and row spacing on dry bean yield. Figures are the average of two replications. Influence of rotation/tillage, irrigation, culti- var, and row spacing on dry bean yield components. Data are the average of two replications . . . . . Root efficiency in seed production as influenced by rotation/tillage, irrigation, and cultivar. Root length was determined by destructive sampling and is based on the assumption that the extracted pro- file contained one half of the roots from a single plant. . . . . . . . . . . . . . . . . . . Chapter 4 1. Calibration of digital image processor using trays of strings. Each line represents the same tray of strings analyzed with or without the debris extrac- tion algorithm engaged. Each length was analyzed twice, with the strings mixed and redistributed be- tween analyses . . . . . . . . . . vii 65 71 74 77 81 83 132 2. Width and length calibration data from digital im- age processor. Each row represents one image. wire pieces were 1 cm long except for the 1.6 mm width, which were 2 cm long. . Results of width class calibration. Each row re- presents one image analyzed three times. Figures reported are the average of the three analyses. Each image contained eight 1 cm pieces of wire of the given width. Breaks between rows of data in- dicate break between width classes as determined by digital image processor . Comparison of root length density (RLD, cm cm") determination by the digital image processor (DIP) and the line intersect (LIN) methods. Root length densities reported are calculated from the DIP length determinations. % LIN is the DIP RLD divi- ded by the line intersect RLD. The 18 cases repre- sent the RLDs from 18 cubes of soil, 7.5 cm on a side, from a soil- root profile 7. 5cm thick x 22.5 cm wide x 45 cm deep . . . . . . . . . . . Summary of digital image processing (DIP) root length density (RLD) results compared to line in- tersect results. Figures represent DIP RLDs/line intersect RLDs x 100. Roots are field grown dry beans, ARNST is alfalfa rotation, no secondary til- lage and CONV is conventional management . . . . Chapter 5 l. 2. Dry bean shoot parameters on three dates following stress as affected by flood and drought stress Leaf area and number of leaves immediately post stress and 12 days after stress as affected by flood and drought stress . . . . . . . . . Total root length of dry beans by depth for three dates following flood and drought stress . . . . . Dry bean yield components as affected by flood and drought stress. . . . . . . . . . . . . . viii 136 137 147 149 162 164 165 169 LIST OF FIGURES Chapter 2 1. Soil bulk density of Charity clay at three depths as affected by rotation/tillage. . . . Air-filled porosities (a) and soil moisture reten- tion (b) of Charity clay at different matric suc- tions and at the 0- 7. 5 cm depth as affected by rotation/tillage . . . . Air-filled porosities (a) and soil moisture reten- tion (b) of Charity clay at different matric suc- tions and at the 7. 5- 15 cm depth as affected by rotation/tillage . . . . . Air-filled porosities (a) and soil moisture reten- tion (b) of Charity clay at different matric suc- tions and at the 15-22.5 cm depth as affected by rotation/tillage . . . . . . . . . . . . . . . . . Pore size distribution on Charity clay at the 0- 7. 5 cm depth (a) and 7. 5- 15 cm depth (b) as af- fected by rotation/tillage . . . . . . . . . . . Pore size distribution on Charity clay at the 15-22.5 cm depth as affected by rotation/tillage . Soil moisture (gravimetric) over time in the 0-10 cm (a) and 10-20 cm (b) depths as affected by ro- tation/tillage x irrigation treatment combinations ‘0 Soil moisture as measured by the neutron moisture meter over time at the 30 cm (a) and 45 cm (b) depth as affected by rotation/tillage x irrigation treatment combinations . . . . . . . Rainfall throughout the growing season and soil moisture at the 10-20 cm depth from 40 days after planting until harvest as affected by rotation/ tillage x irrigation treatment combination. As- terisks represent irrigation dates and amounts ix 41 43 44 45 48 49 51 53 54 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. Emergence of dry beans on Charity clay as affected by row spacing at 11 days after planting (a) and 19 days after planting (b) . Emergence of dry beans on Charity clay at 11 days after planting (a) and 19 days after planting (b) as affected by rotation/tillage or cultivar. Dry bean shoot biomass on five dates as affected by row spacing . Above ground biomass of two dry bean cultivars on five dates Above ground biomass of two dry bean cultivars on five dates as affected by rotation/tillage Dry bean shoot biomass on five dates as affected by irrigation. The influence of rotation/tillage on root length densities of dry beans as determined by destruc- tive profile sampling at the time of maximum flowering. . . . . Root length densities of dry beans as determined by destructive sampling at the time of maximum flowering and as affected by cultivar under con- ventional rotation/tillage and irrigated condi- tions. . . . . . . Dry bean root length densities as determined by destructive sampling and as affected by cultivar under an alfalfa rotation/no secondary tillage, nonirrigated management system . . . . . The effects of row spacing, rotation/tillage, and irrigation on dry bean maturity 98 days after planting. Table 3 shows statistical comparison of these treatments. Final population of dry bean plants at three row spacings as influenced by rotation/tillage Dry bean yield on Charity clay as influenced by cultivar (a) and row spacing (b). Yields were averaged across rotation/tillage and irrigation. 56 57 59 60 61 62 67 68 69 72 75 78 Chapter 3 1. Minirhizotron root observations for four rotation/ tillage (ARNST or CONV) and irrigation (IR or NI) treatment combinations in 1985 as affected by date of observation. Data points are the average number of roots observed cm" by ten cm depth increments. Minirhizotron root observations on four dates in 1985 as affected by rotation/tillage (ARNST or CONV) and irrigation (IR or NI) treatment combin- ations. Root observations are reported as the average number of roots observed cm" in ten cm depth increments . . . . . . . . . . . . . . . . . Minirhizotron root observations for four rotation/ tillage (ARNST or CONV) and irrigation (IR or NI) treatment combinations in 1986 as affected by date of observation. Data points are the average number of roots observed cm" by ten cm depth increments. Minirhizotron root observations on four dates in 1986 as affected by rotation/tillage (ARNST or CONV) and irrigation (IR or NI) treatment combin- ations. Root observations are reported as the average number of roots observed cm" in ten cm depth increments . . . . . . . . . . . . . . . Root length density as determined by destructive sampling in mid-August, 1985 and as influenced by rotation/tillage under irrigated conditions. . . . Root length density as determined by destructive sampling in mid-August of 1985. The effects of rotation/tillage under nonirrigated conditions are shown. . . .-. . . . . . . . . . . . . . . . . Root length density as determined by destructive sampling in mid-August, 1986 and as influenced by rotation/tillage under irrigated conditions. . . . Root length density as determined by destructive sampling in mid-August of 1986. The effects of rotation/tillage under nonirrigated conditions are shown. . . . . . . . . . . . . . . . . . . . Minirhizotron observation images representative of those seen under the two crop/soil management sys- tems: a: Conventional management, compacted layer at 20 cm depth; and b: Alfalfa rotation, no sec- dary tillage, compacted layer broken up by alfalfa roots and deep tillage . . . . . . . . . . . . . . 9. xi 104 105 106 107 108 109 110 111 115 Chapter 4 1. Time required for image processing as affected by the use of the debris extraction algorithm and the number of images processed . . . . . . . . . . . Results of tests to determine the fraction of im- ages from a tray of roots which must be processed in order to obtain acceptable estimates of root length. Horizontal lines represent 15% of the ac- tual (DIP) analysis total. Each data point repre- sents the pixel sum (y) of the randomly selected images (x) multiplied by (64/number of images). Comparison of root length determinations by image processing and line intersect methods. Each point represents one tray of roots, or 64 images. Diag- onal line represents 1:1 correlation; top line is linear regression line . . . . . . . . . Comparison of root length density as determined by the digital image processor and the line intersect method for dry bean roots under two crop and soil management systems. Each point is the average of 4 replications . . . . . . . . . . Chapter 5 1. Shoot: root ratios (dry weight. length) of dry beans as affected by flood or drought stress and time following stress . . . . . . . . . xii 141 142 145 150 168 I NTRODUCTI ON Dry beans (Phaseolus vulgaris) are an important crop in the agricultural economy of Michigan. Successful dry bean production depends on the proper management of many interre- lated factors. One of the most significant factors is soil, and proper management of this resource is of primary impor- tance if production is to be optimized. The soil matrix provides a suitable environment for root growth. It must physically support the plant, yet allow for easy root proliferation and penetration. The soil must pro— vide adequate moisture and air for use by the roots and ulti- mately the whole plant system. It must offer adequate nutri- ents for root uptake and plant use. The balance between soil moisture and soil air is crucial, and an excess of moisture can mean an inadequate amount of air for plant needs. Many of the dry beans grown in Michigan are grown on fine textured lake plain soils. These soils are naturally poorly drained and prone to waterlogging, even with artifi- cial drainage. Poor aeration can be an accompanying problem. The soils are subject to compaction by agricultural equipment traffic. Compaction may result in serious mechanical impe- dance to root growth, and may intensify problems of soil wa- ter and air flow and availability to plants. 2 This thesis reports on a study of problems related to soil stresses which adversely affect dry bean production on fine textured soils. The primary objective was to identify soil conditions which would allow the root system to function in a favorable, reduced stress environment, resulting in a healthy whole plant system capable of maximum yield. The study included soil and crop management practices designed to minimize environmental stress. The effects on soil physical properties and on the dry bean plants were mon- itored. The utilization of a deep-rooted legume in the crop rotation was examined for its potential in soil compaction alleviation. Different tillage systems were used to break up existing compaction and to prevent additional compaction. Other management factors included irrigation, row spacing, and cultivar variables. Root studies were conducted to exam- ine the effects of various treatments on the root system and ultimately on whole plant performance. A greenhouse study was carried out to examine the ef— fects of flood and drought stress on dry bean root and shoot performance. Soil compaction can result in effects similar to those of flood or drought since compaction can alter soil water and air flow and availability to plants. The study of root systems is hampered by difficulties in quantifiying root activities. New methods are needed to al- low for faster, more extensive information gathering. Sever- al methods of root study were employed, including two given special attention. The microvideo minirhizotron root 3 observation method has been developed recently and it was used and tested extensively in this study. Digital image processing of extracted, washed root samples is currently being deve-loped and a system was tested on the roots from this study. A chapter on minirhizotron observation tubes on fine textured soils and a chapter on digital image processing for use in root studies are included. Soil management to optimize the rooting zone is impor- tant not only for dry beans on fine textured soils. The exact nature and the effects of soil stresses vary among 3011: but the resulting root stress and decreased production is universal. It is hoped that the results of this study V111 find application to related problems with crop produc- tion on similar or on different soils. CHAPTER 1 LITERATURE REVIEW WIDE. Successful production of any agronomic crop requires careful management of many interrelated factors. Ultimately, it is the combination of management practices which deter- mines if the production enterprise has been successful. The soil resource is a basic element in crop production. Proper management of this resource is essential, both for current production and for future use. Improper management 0f the soil may result in a variety of problems which can have long-term effects. Some of the other factors important in Crop production, e.g. weather, are difficult or impossible to manipulate but this is not the case with soil. The soil tee’Ource can be and is manipulated with relative ease. Soil tillage is practiced in most agriculture production efforts thr oughout the world. The benefits are many and have often been realized. But the potential also exists for treating thQ soil in such a way that the results are counterproduc- t1Ve. There are too many examples worldwide of the misuse of 30113 which have resulted in erosion, compaction, or other negative effects and decreased productivity. 5 Soils are quite diverse; therefore, soil management prac- tices must also be widely varied. Soils differ due to parent material and weathering. Depth, horizonation, texture, structure, natural fertility, water and air relations, micro- bial activity, and organic matter also differ greatly for different soils. The infinite number of combinations of these factors results in soils useful for many different purposes. Hany soils are well suited to one or more particular crops . In other cases crops are grown on soils which may not be especially well suited to the given soil environment but for historical, cultural, economic, or other reasons the 9rowth of a particular crop on a given soil continues. In these cases the management of the soil resource for optimum production and long term conservation becomes even more important . Dry edible beans (Phaseolus vulgaris L.) have been grown commercially in the Saginaw Valley region of Michigan for at 10881: 100 years (Anderson, 1978). The dry bean is a short 8038011 crop which has grown well in the fertile, fine tex- tured lake bed soils. However, in the past two decades dry b9“ Yields in the region have stagnated or even decreased in spite of improved varieties and pest control (Wright, 1978). This trend has been particularly troublesome for producers as 6 the economic pressures on farmers have increased. Plant growth and production is highly responsive to the soil environment. As farmers and researchers have noted the trend toward declining dry bean yields, they have begun to look carefully at management practices, particularly soil management. The problem of soil compaction and related ef- fects has been noted as the most limiting factor to in- creased production (Smucker et al., 1978). Mechanical impe- dance, soil moisture relations, and aeration are interrela- ted factors which are often associated with soil compaction and nay have a significant effect on plant growth and pro- duct 1 on (Hillel, 1982) . W Soil compaction may result from natural or human-induced factors. Among the natural causes are soil formation proces- ses, raindrop impact, wetting and drying'cycles with accom- panying shrinkage, and root growth (Hillel, 1982; Larson and Allmaras, 1971). The major cause of soil compaction is traf- fic and tillage operations (Hillel, 1982; Bowen, 1981). As methods of agricultural production change, soil compaction occurs more frequently and is more intense. A primary factor 15 the increase in the size, weight and power of agricultural tractors and tillage and- harvesting machinery. In addition to the physical compaction imposed on the soil, there are nu- meroug related effects on the soil and plant system. Inclu- ded are changes in the soil moisture relationships, aeration, 7 root restriction, and disease and pest resistance. Fountaine (1959) reported that the productivity of com- pacted soils is altered by increased mechanical impedance, decreased aeration, and altered soil moisture availability. These changes are due to the increase in soil density and accompanying decrease in soil pore space. Most reports show negative effects of increased soil density on productivity acrc>ss a wide range of crops, soil types, and management systzems (Baver, 1944; Douglas and McKyes, 1983; Hakasson, 19855; McKyes et a1., 1979). However, Voorhees (1977) repor- ted .an increase in soybean yields under moderate compaction in em dry year. This likely indicates an improvement in the wateer holding capacity of the soil due to additional densi- ficeation of the soil. It also points out that there can be considerable variation in the effects of compaction and the mechanisms producing those effects. iflfld£_QENELIX Bulk density is the most widely reported measure of soil compaction. Cassel (1982) reported that bulk density is eas- 11Y measured, and is almost always altered by tillage. Cas- sel further noted that bulk density varies temporally and Spatially and sampling must take this into account. In a re- view of plant response to soil compaction, Rosenberg (1964) nOtGdaparabolic yield response to bulk density. He credited vomOcil (1955) with first describing this response. Vomocil stated that below a critical bulk density, yield variation is 8 due to factors other than soil physical conditions. Rosen- . berg concluded that the correlation between yield and bulk density is difficult to quantify due to the numerous factors involved, including soil texture, moisture, aeration, cli- mate, and crop. He further suggested that aeration is often not limiting on compacted soils, especially on coarse and me- dium textured soils.‘ Veihmeyer and Hendrickson (1948) sug- gested that on most soils a bulk density of 1.9 g cm" will prevent root growth of most common plants and that on clay soils the critical maximum is between 1.5 and 1.7 g cm". Zimmerman (1961) found the maximum bulk density for root pen- etrat:ion was about 1.8 and 2.0 g cm“ for a Cherty clay loam and a sandy loam respectively. Mcxyes et al. (1979), in a three year study on soil with a higih clay content, found that the yield of corn silage de- creased significantly as bulk density increased. He also no- ted that, bulk density can be too low for optimum crop yield. Jones (1983) defined critical bulk density as the soil density' at which rooting activity, was at a maximum for a given soil water content. He identified critical bulk densi- ties for several crops (cotton, corn, peas, sudangrass, and sugarcane) on soils with differing textural composition and found a strong correlation between critical bulk density and texture, with soils with a high clay content having lower ‘ Others later disagreed with Rosenberg on this point, as V111 be discussed later. 9 critical densities than those with less clay. He concluded that rooting density above the critical bulk density is af- fected largely by soil strength rather than soil moisture and aeration which figure more prominently in root growth at low- er bulk densities. bEE31ANIEAL_lMEEDAN£E_AND_ELANI_BESBQNSE Roots must penetrate the soil profile to physically sup- port: the plant and take up water and nutrients necessary for plarat growth. If the soil is compacted, the pore size dis- tritaution is altered (Bowen, 1981; Hillel, 1982). The vol- ume decrease due to compaction comes from a compression of the: macropores and rearrangement of soil particles resulting in ean increase in micropores. This leaves fewer macropores thrxaugh which the roots can grow. Wiersum (1957) and Foun- taiiae (1959) noted that roots have to apply pressure to grow . witfllin clods or to deform pores, and that if unable to exert sufficient pressure will be restricted in growth. Gill and Bolt (1956) presented a summary of the classic work by Pfef- fer in 1893 on root growth pressure exerted by plants. Pfef- fer encased a growing root in plaster of Paris and measured the pressure exerted by the root. He found that roots exer- ted Pressure both radially and axially, up to 10 atm. when Pressure was brought to bear on all sides of the root. More Pressure was exerted radially, likely due to much larger sur- face area, and radial pressure was increased when axial 91°Vth was constricted. This work provides a basis for 10 studies on root growth in soil and is important when consi- dering root growth under compacted or constrictive soil con- ditions. By growing roots in a precisely defined granular media, Richards and Greacan (1986) found that fine roots are less subject to mechanical impedance than large roots. They also noted that cylindrical expansion behind the growing root tip is at: least somewhat effective in relieving soil pressure and enabl.ing the root to elongate. Roots have also been found to adapt their own morphology in order to grow in compacted or tortuous soil (Taylor and Arkira, 1981).. This enables the plant root system to overcome a compacted layer or zone in soil. There are limits to roots ' ability to overcome mechanical impedance and there are also» extra costs in terms of growth or photoassimilate parti- tioning. Barley (1961) studied maize roots and found that they continued to grow even when were physically restricted. Growth rate was decreased, however. Increased pressure on the root tip may result in an increase in root diameter due to changes in the internal pressure (Russell, 1977). Russell also suggested that mechanical impedance does not cause roots to become thinner. Rather, mechanical impedance may cause PIOIIEeration of lateral roots which can easily be mistaken for root axes of reduced diameter. The study of the response 05 POOts to mechanical stress is essentially the investiga- tion of the effects of the pressures roots must exert to en- large or create pores. 11 5QLL_MQlEIuBEx_AERAILQN_AND_ELANI_RE£EQNEE Soil compaction can cause or contribute to aeration and soil moisture problems. Warkentin (1971) found that compac- tion alters the water content and movement in soils by modi- fying the pore size distribution with the large macropores reduced first. This leads to reduced water movement, espe— cially on fine textured soils. Reduced water movement can cause waterlogging following precipitation as the macropores no longer exist to provide for good drainage, and waterlog- ging affects plant growth by decreasing or eliminating the necessary soil air. The volume of soil air depends on both total pore space and water filled pore space. Erickson (1982) suggested that 10% air capacity (% of soil volume) is the lower limit for plant growth for most crops, but that no value should be ta- ken as an absolute minimum due to variations in soil condi- tions and plant requirements. In addition to the air filled pore space, it is imperative for good plant growth that oxy— gen diffusion occur rapidly enough to replace that which is taken up by the plant. This includes diffusion through the bulk soil and also through the moisture films surrounding plant roots. The latter is usually the most critical since diffusion through water is approximately 10,000 times slower than through air. Erickson (1982) determined that an oxygen diffusion rate of < 0.2 ug cm" min‘1 is the minimum for oxygen resupply to active roots and > 0.4 ug cm” min-1 is necessary for normal plant growth. The composition of the 12 soil atmosphere can vary markedly from the above ground atmo- ‘sphere. Campbell and Phene (1977) monitored the aeration status of a layered sandy loam in which millet was growing and found that plant growth was unaffected by oxygen concen- trations above 15%, but below 15‘ growth was reduced. In a study by Smucker and Erickson (1977), anaerobic conditions resulted in reduced growth of peas and also contributed to increased exudation of various organic compounds. This in- crease in exudation indicates a loss of photoassimilates and suggests one way in which compaction can result in decreased yield. Letey et al. (1962) found that roots ceased growing if oxygen content was too low for a short time, and periods of low oxygen supply were more detrimental to young plants than to those with larger root systems. Huck (1970) reported that anaerobiosis can result in cessation of root growth and death of some root tips. An anaerobic rhizosphere environment most often results from waterlogging (Cannell and Jackson, 1981). This problem is common on compacted soils since compacted layers may pre- vent adequate drainage. lhile some plants are more tolerant of a waterlogged environment due to root characteristics such as porosity, length, and the ability to form adventitious roots (Cannell and Jackson, 1981), nearly all plants suffer at least some setback when subjected to too much water and the accompanying lack of oxygen. Root growth was generally retarded (Glinski and Stepniewski, 1985) and shoot growth was affected as well. Jackson and Drew (1984) suggested that 13 the primary effect of too much water is asphyxiation of the plant as gaseous diffusion almost ceases. Anaerobic condi- tions predispose the plant root system to attack by pathogens (Stolzy et al., 1965; Miller et al., 1980; Smucker and Erick- son, 1987) and cause the plant to exude compounds which may be toxic to the plant (Schwartz, 1980; Smucker and Erickson, 1987). SQIL_£QMEA£IIQN_AND_DBQHQHT_SIRE&S Cortes and Sinclair (1986) studied the water relations of soybeans grown under drought stress. They found that in soybeans the most important mechanism for sustaining growth under limited moisture supply was the maintenance of root growth into deeper portions of the soil profile to tap exis- ting water supplies. Huck et al. (1986) noted that water stress on soybeans resulted in decreased shoot and seed weight but increased total root length, including an increase in rooting depth. Hoogenboom et al., (1987) and Huck (1986) also reported increased rooting depth during periods of drought stress. These ~studies indicate that a normal re- sponse of a plant to moisture stress is a decrease in shoot: root ratio as the plant partitions more of its photoassimi- lates to the root system in order to sustain growth and ac- cess available water. Shank (1945) found the shoot:root ra- tio of maize changed from 3.4:1 under sufficient (21%) water to 2.5:1 when soil moisture was severely limiting (7.5%). Huck et al., (1986) found that the timing of the moisture l4 stress was also very important. Their work showed the most crucial growth stage in soybeans is the reproduction stage. In the case of crops grown on compacted soils or on soils with compacted layers, rooting depth may be restricted due to mechanical impedance (Miller, 1987; Bennie and Botha, 1986; Raghavan et al., 1977; Bertrand and Kohnke, 1957). As long as there is sufficient water available the plant will not suffer. However, during a short term drought, common on many soils, the shallow root system will not be able to ac- cess the moisture which may be available in the deeper soil horizons. In addition, the common response of roots growing deeper when soil moisture is limiting will be hampered by the continuing mechanical impedance of the soil. IILLAGE:INDUSED.§QIL_§QMEA£ILQN. The effects of increased bulk density on root systems have been discussed previously. Increased bulk density often octurs as a result of tillage operations. Phillips and Kirk- ham (1962) found that a Colo clay soil compacted to various bulk densities by vehicular traffic resulted in reduced stands, maturity, and yield of corn. Raghavan et al. (1979) severely compacted plots on which silage corn was grown by applying 15 passes of tractor traffic. Maximum rooting depth was halved and the depth of dense roots was decreased by one third compared, to the noncompacted plots. In a study on a compacted clay soil, Douglas and Mcxyes (1983) noted a 15 reduction of silage corn yield of up to 40% compared to on noncompacted soils. They suggested that the plants suffered both from mechanical impedance and poor aeration due to inad- equate rainage. Root growth was delayed due to high bulk density levels. The soil moisture content at the time of tillage opera- tions is important in determining the extent of compaction. According to work done by Mckyes (1985), densification can be up to five times as severe on soils tilled at optimum soil moisture when compared to the same soils tilled when quite dry. In laboratory studies Akram and Kemper (1979) found that maximum compaction generally occurred when soils were at or near field capacity for water. Their work was conducted on both fine and coarse textured soils. Soil texture figures prominently in the extent of com- paction on a given soil. McKyes (1985) found an increase of 0.13 g cm" in bulk density of a clay soil resulted in a 50% decrease in yield of maize. A similar increase in bulk den- sity on a coarse textured soil would likely be insignificant. The difference is due to the differences in macropore/micro- pore ratios. 8QIL_£QMEAQIIQN_ALLEIIATIQN. Soil compaction is often caused by tillage operations. Somewhat ironically, tillage may also be used to alleviate compaction. In recent years new tillage practices have be- come popular and have been shown to be of some value in 16 ameliorating compacted soil. Subsoiling may be useful in breaking up tillage-induced hardpan layers in the soil, espe- cially moldboard plow layers.. Other tillage operations may be carried out to alleviate compaction caused by natural pro- cesses or by secondary tillage (Bowen, 1981). Miller (1987), in a study on the effect of subsoiling and irrigation on dry bean production, found that water stress may develop between irrigations if the plants have not developed a deep enough root system. This study was conduc- ted on a sandy loam soil (85% sand, 2% clay, rigid matrix at 30 cm). Subsoiling allowed deeper root growth and resulted in significantly increased yields. However, in a companion study with the same treatments on a loam soil (45% sand, 9% clay, plowpan at 25 cm), Miller found that subsoiling did not affect yields even though rooting depth and foliage density were increased. He attributed this response to the better water holding capacity of the loam soil compared to the sandy loam. Bennie and Botha (1986) studied rooting depth and wa- ter use efficiency for maize and wheat. The work was carried out on irrigated, deep,_ fine sandy soils. Deep ripping and controlled traffic led to an increase in rooting depth, water use efficiency, and yield (30% in maize and 19% in wheat) compared to conventional tillage. ALLE11AIIQN.BX.NAIHRAL.ERQ§E§SES Soil compaction can be alleviated to some extent by natural processes (Bowen, 1981; Larson and Allmaras, 1971). 17 Akram and Kemper (1979) found that wetting and drying cycles improved infiltration rates. The process of freezing and thawing was also found to improve infiltration rates and leave the soil in a generally more friable, less compacted condition. Voorhees (1979, 1983) reported that natural weathering (wetting and drying, freezing and thawing) alle- viated some, but not all, traffic-induced compaction on a Nicollet clay loam. The effect of deep rooted plants or plants with espe- cially hardy root systems has also been studied in relation to compaction alleviation. Radcliffe et al. (1986) examined the effect of a deep—rooted perennial, alfalfa, on subsoil compaction. The combination of alfalfa and the application of gypsum resulted in decreased soil strength as measured by a cone penetrometer. These researchers concluded that the use of a deep rooted legume is effective in loosening com- pacted soil layers as long as the nutritional status of the subsoil is conducive to root growth in that region. In a study on a Charity clay soil Christenson et a1. (1976) found that a crop of alfalfa prior to dry beans resulted in higher dry bean yields than when dry beans were preceded by any other crop. These yield trends are not proof of decreased compaction due to the alfalfa. However, they are strong indicators that this is occurring since the Charity clay is subject to tillage and traffic compaction and the related problems of poor drainage and poor aeration. Chasse et al. (1967) reported that an extensive root 18 system near the soil surface (high organic matter) reduced compaction damage due to machinery loads by about 65% com- pared to bare soil with low organic matter. In a laboratory study comparing three soil textures with varying amounts of organic matter, each at three levels of compaction, Ohu et al. (1985) determined that the presence of organic matter increased the root dry matter and yield of the crop while compaction decreased the same plant parameters. These re- searchers concluded that high levels of organic matter have the potential to improve the productivity of compacted soils. In a contradictory study using a soil bin as well as labora- tory procedures, Gupta et al. (1987) looked at the influ- ence of corn residue on compaction due to wheel traffic and concluded that the presence of this organic matter had little or no effect on soil compaction. AnnIIIONAL_HANA£EMENT_EA§TQRE 891.52A91Nfi, In addition to soil management for the alleviation and prevention of soil compaction, other management factors are important in dry bean production. Row spacing is one such factor. More work has been reported on soybean yield response to row spacing than dry bean yield response to planting pat- terns. The response of both crops will be reviewed. 19 Cooper (1977) documented up to a 25% increase in soy- beans grown in narrow (17 cm) vs. wide (50 or 75 cm) rows, all row spacings having a constant seeding rate. Taylor (1982) found that soybeans grown on a silt loam yielded significantly more in 25 or 50 cm rows than when grown in 75 or 100 cm rows when the moisture supply for the crop was good. However, under conditions of limited water there were no yield differences. In a study designed to determine the reasons soybeans generally yield more in narrow vs. wide rows, Bennie et al. (1982) found that there were few differ- ences in nutrient uptake and accumulation due to row spacing (25 cm and 100 cm row spacings were used). In the same study, Mason et al. (1982) noted a 49% increase in root length density under narrow rows, 52% more roots per unit leaf area, and a differential water uptake rate on the wide rows with more water used from the intrarow space than the interrow space. However, overall water use and plant water potentials were not different under irrigation between row spacings and there were no yield differences due to irriga- tion. These researchers concluded that yield differences between the two row spacings were not due to differential water use which might be expected with large differences in root length densities. Taylor et a1. (1982) participated in the same investigation and reported that radiation interception was greater at the narrow row spacing during most of the growing season and especially at the critical reproductive stages. This is due to increased or more 20 uniform canopy cover. Taylor and associates concluded that this differential radiation interception, rather than differ- ences in plant water status due to rooting differences or differences in nutrient uptake and accumulation, was primari- ly responsible for increased soybean yield under narrow row conditions. Atkins (1961) compared red kidney bean production under different row spacings. He found the 23 and 46 cm rows sig- nificantly outyielded the conventional 92 cm rows but there were no differences between the 23 and 46 cm row spacing yields. Redden et al. (1987) reported up to a 46% increase in dry bean yields when the crop was planted in 18 cm vs. 107 cm rows. In that study the authors also noted a positive yield response to a tripling of plant population (from 112,500 to 337,500 plants ha"). Grain yields closely fol- lowed the number of pods meter", which was positively corre- lated with ground cover between flowering and the middle of pod fill. Thus the increase in yield was attributed primar- ily to canopy cover and radiation interception. EQMMARI. The production of any crop depends on a favorable soil environment. Soil compaction and the related effects of poor water relations, poor aeration, and mechanical impedance are often responsible for weaker and shallower root systems, decreased plant growth, and reduced yields. Dryhbeans do not naturally have a hardy root system which will tolerate 21 poor soil structural conditions, poor aeration, and compacted soil layers. The Charity clay soil in the lake plains in the central eastern part of Michigan's lower peninsula is subject to these detrimental conditions. The production of dry beans on this soil is an acute management challenge. The intent of this study was to utilize a combination of factors to overcome the soil management problems of Phaseolus vulgaris on these soils. The use of a deep-rooted perennial legume, tillage management to alleviate existing compaction and prevent further compaction, timely irrigation, and the use of narrow row spacing were combined in an effort to achieve a maximum yield of dry beans. REFERENCES Akram, M. and w. D. Kemper. 1979. Infiltration of soils as affected by the pressure and water content at the time of compaction. Soil Sci. Soc. Am. J. 43:1080-1086. Anderson, A. L. 1978. The Michigan Dry Edible Bean Industry- History. p. 1-15. In L. S. Robertson and R. D. Frazier (eds.) Dry Bean Production - Principles & Practices. Ext. Bull. E-1251., Mich. Agr. Exp. Sta and Coop. Ext. Service, Lansing, Michigan. Atkins, J. D. 1961. Row spacing influences on yield of snap and dry beans. Farm Research 27:13. Aubertin, G. M. and L. T. Kardos. 1965. Root growth through porous media under controlled conditions. I. Effect of pore size & rigidity. Soil Sci. Soc. Am. Proc. 29:290-293. Barley, K. P. 1962. The effects of mechanical stress on the growth of roots. J. Exp. Bot. 13:95-110. Baver, L. D. and R. B. Farnsworth. 1940. Soil structure effects in the growth of sugar beets. Soil Sci. Soc. Am. Proc. 5:45-48. Bennie, A. T. P., H. K. Mason, and H. M. Taylor. 1982. Responses of soybeans to two row spacings and two soil water levels: III. Concentrations, accumulation, and translocation of 12 elements. Field Crops Res. 5:31-43. Bennie, A. T. P. and F. J. P. Botha. 1986. Effect of deep- tillage and controlled traffic on root growth, water-use efficiency, and yield of irrigated maize and wheat. Soil Tillage Res. 7:85-95. Bertrand, A. R. and H. Kohnke. 1957. Subsoil conditions and their effects on oxygen supply and the growth of corn roots. Soil Sci. Soc. Am. Proc. 21: 135-140. Bowen, H. D. 1981. Alleviating mechanical impedance. p. 21- 58. In G. F. Arkin and H. M. Taylor (eds.) Modifying the root environment to reduce crop stress. ASAE Mono. #4, St. Joseph, Michigan. 22 23 Campbell, R. B. and C. J. Phene. 1977. Tillage, matric potential, oxygen, and millet yield relations in layered soil. Trans ASAE 20:271-275. Cannell, R. Q. and M. B. Jackson. 1981. Alleviating aeration stresses. p. 141-192. In G. F. Arkin and H. M. Taylor (eds.) Modifying the root environment to reduce crop stress. ASAE, St. Joseph, Michigan. Cassel, D. K. 1982. Tillage effects on soil bulk density and mechanical impedance. p. 45-67. In P. W. Unger and D. M. Van Doren, Jr. (eds.) Predicting tillage effects on soil physical properties and processes. Amer. Soc. Agron. Spec. Publ. 44. Chasse, M., G. S. V. Raghavan, and E. McKyes. 1975. Etude sur les problemes de compactage. Quebec Min. Agr. Research Reports: 62 p. Christenson, D. R., C. Bricker, J. Reisen, and R. Voth. 1976. Soil fertility and management for the production of sugar beets, navy beans, and corn. Mich. State Univ. 1976 Res. Report Saginaw Valley Bean - Beet *Res. Farm and Related Bean - Beet Res. Cooper,.R. L. 1977. Response of soybean cultivars to narrow rows and planting rates under weed-free conditions. Agron. J. 69:89-92. Cortes, P. M., and T. R. Sinclair. 1986. Water relations of field-grown soybean under drought. Crop Sci. 26(5):993. Douglas, E. and E. McKyes. 1983. Tillage practices related to limiting plant growth factors and crop yields. Canad. Agr. Eng. 25(1):47-55. . Erickson, A. E. 1982. Tillage effects on soil aeration, p. 91-104. In P. w. Unger and D. M. Van Doren, Jr. (eds.) Predicting tillage effects on soil physical properties and processes. Amer. Soc. Agron. Spec. Publ. 44. Fountaine, E. R. 1959. Physical requirements of plants as criteria for soil structure. Mededelingen Landbouwho- geschool Gent 24:30-35. Gill, W. R. and G. H. Bolt. 1956. Pfeffer's studies of the root growth pressure exerted by plants. Agron J. 47:166- 168. Glinski, J. and W. Stepniewski. 1985. Soil aeration and its role for plants. CRC Press, Boca Raton, Florida. p.149- 152. 24 Gupta, S. C., E. C. Schneider, I. E. Larson, and A. Hadas. 1987. Influence of corn residue on compression and compaction behavior of soils. Soil Sci. Soc. Am. J. 51:207-212. Hakasson, I., L. Henriksson, and L. Gustafsson. 1985. Experiments on reduced compaction of heavy clay soils and sandy soils in Sweden. p. 995-1009. Proceedings of the Intn'l Conf. on Soil Dynamics, Auburn, Alabama. Hillel, D. 1982. Introduction to Soil Physics. Academic Press. Orlando, Florida. 365 p. Hoogenboom, 6., M. G. Huck, and C. M. Peterson. 1987. Root growth rate of soybean as affected by drought stress. Agron. J. 79:607-614. Huck, M. G. 1970. Variation in taproot elongation rate as influenced by composition of the soil air. Agron. J. 62:815-818. C. M. Peterson, 6. Hoogenboom, and C. D. Busch. 1986. Distribution of dry matter between roots and shoots of irrigated and non-irrigated determinate soybeans. Agron. J. 78:807-813. Jackson, H. B. and M. C. Drew. 1984. Effects of flooding on growth and metabolism of herbaceous plants. p.47-128 in T. T. Kozlowski, (ed.) Flooding and plant growth. Academic Press, Inc., Orlando, Florida. Jones, C. A. 1983. Effect of soil texture on critical bulk densities for root growth. Soil Sci. Soc. Am. J. 47:1208- 1211. Larson, l. E. and R. R. Allmaras. 1971. Management factors and natural forces as related to compaction. p. 367-427. In Carleton, et al. (eds.) Compaction of Agricultural Soils. ASAE Mono. ll, St. Joseph, Michigan. Letey, J., L. H. Stolzy, and G. B. Blank. 1962. Effect of duration and timing of low soil oxygen content on shoot and root growth. Agron. J. 54:34-37. Mason, I. K., H. R. Rowse, A. T. P. Bennie, T. C. Kaspar, and H. M. Taylor. 1982. Responses of soybeans to two row spacings and two soil water levels. II.later use, root growth, and plant water status. Field Crops Res. 5:15-29. MCKyes, E., S. Negi, E. Douglas, F. Taylor, and V. Raghavan. 1979. The effect of machinery traffic and tillage operations on the physical properties of a clay and on yield of silage corn. J. Agri. Eng. Res. 24:143-148. 25 . 1985. Soil Cutting and Tillage. Elsevier. Developments in Agr. Eng. 7. Amsterdam. 217 p. Miller, D. E., D. W. Burke, and J. M. Kraft. 1980. Predisposition of bean roots to attack by the pea pathogen, Fusarium solani f. sp. pisi, due to temporary oxygen stress. Phytopathology. 70:1221-1224. . 1987. Effect of subsoiling and irrigation regime on dry bean production in the Pacific Northwest. Soil Sci. Soc. Am. J. 51:784-787. Ohu, J. O., G. S. V. Raghavan, E. McKyes, K. A. Stewart, and M. A. Fanous. 1985. The effects of soil compaction and organic matter on the growth of bush beans. Trans. of the ASAE 28(4):1056-1061. Phillips, R. E. and D. Kirkham. 1962. Soil compaction in the field and corn growth. Agron. J. 54:29-34. Radcliffe, D. E., R. L. Clark, and M. E. Sumner. 1986. Effect of Gypsum and deep-rooting perennials on subsoil mechanical impedance. Soil Sci. Soc. Am. J. 50:1566-1570. Raghavan, G. S. V., E. McKyes, and M. Chasse. 1977. Effects of wheel slip on soil compaction. J. Agri. Eng. Res. 22:79-83. ____, ____, R. Baxter, and G. Gendron. 1979. Traffic - soil - plant (maize) relations. Terramechanics. 16(4):181-189. Redden, R. J., T. Usher, D. Younger, R. Mayer, B. Hall, A. Fernandez, and D. Kirton. 1987. Response of navy beans to row width and plant population density in Queensland. Austral. J. of Exp. Agric. 27(3):455-463. Richards, B. G. and E. L. Greacen. 1986. Mechanical stresses on an expanding cylindrical root analogue-granular media. Austral. J. Soil Res. 24:393-404. Rosenberg, N. J. 1964. Responses of plants to the physical effects of soil compaction. Adv. in Agron. 16:181-196. Russell, R. S. 1977. Plant root systems: Their function and interaction with the soils. McGraw Hill, London. 298 p. Schwartz, H. F. 1980. Miscellaneous problems. In H. F. Schwartz and G. E. Galvez (eds.) Bean production problems. CIAT, Cali, Colombia. Shank, D. B. 1945. Effects of P, N, and soil moisture on top-root ratios of inbred and hybrid maize. J. Agric. Research 70: 365-377. 26 Smucker, A. J. M., D. L. Mokma, and D. E. Linvill. 1978. Environmental requirements and stresses. p 45-61. In L.S. Robertson and R. D. Frazier (eds.) Dry bean production- principles and practices. Ext. Bull. E-1251., Mich. Agr. Exp. Sta and Coop. Ext. Service, Lansing, Michigan. , and A. E. Erickson. 1987. Anaerobic stimulation of root exudates and disease of peas. Plant and Soil 99:423- 433. Stolzy, L. H., J. Letey, L. J. Klotz, and C. K. Labanauskas. 1965. Water and aeration factors in root decay of Citrus sinensis. Phytopathology 55:270-275. Taylor, H. M. and G. F. Arkin. 1981. Root zone modifica- tions: fundamentals and alternatives. p. 3-16. In G. F. Arkin and H. M. Taylor (eds.) Modifying the root environment to reduce crop stress. ASAE Home 84, St. Joseph, Michigan. , I. K. Mason, A. T. P. Bennie, and H. R. Rowse. 1982. Responses of soybeans to two row spacings and two soil water levels. I. An analysis of biomass accumulation, canopy development, solar radiation interception, and components of seed yield. Field Crop Research. 5:1-14. Veihmeyer, F. J. and A. H. Hendrickson. 1948. Soil density and root penetration. Soil Sci. 65:487-493. Vomocil, J. A. 1955. Ph.D. Thesis. Rutgers State Univ., New Jersey, (New Brunswick, N.J.). Voorhees, I. B. 1977. Soil compaction: How it influences moisture, temperature, yield, root growth. Crops and Soils. 29 (6):7-10. . 1979. Soil tilth deterioration under row cropping in the northern corn belt: Influence of tillage and wheel traffic. J. Soil and Fat. Conserv. July/Aug. 1979. p. 184- 186. ____. 1983. Relative effectiveness of tillage and natural forces in alleviating wheel-induced soil compaction. Soil Sci. Soc. Am. J. 47:129-133. larkentin, P. B. 1971. Effects of compaction on content and transmission of water in soils. p. 126-153. In Carleton, et al. (eds.) Compaction of agricultural soils. ASAE, St. Joseph, Michigan. Hiersum, L. K. 1957. The relationship of the size and structural rigidity of pores to their penetration by roots. Plant and Soil 9:75-85. 27 Wright, K. T. 1978. Production trends: World, 0.8., and Michigan. p. 16-30. In L.S. Robertson and R. D. Frazier (eds.) Dry Bean Production - Principles and practices. Ext. Bull. E-1251., Mich. Agr. Exp. Sta and Coop. Ext. Service, Lansing, Michigan. Zimmerman, R. P. and L. T. Kardos. 1961. Effect of bulk density on root growth. Soil Sci. 91:280-288. CHAPTER 2 ALLEVIATION OF SOIL STRESSES ON DRY BEAN ROOT SYSTEMS AS A KEY TO INCREASED DRY BEAN PRODUCTION INTRODUCTION The production of dry beans (Phaseolus vulgaris) in the Saginaw Valley region of Michigan has for many years been an important component of the farm economy. In recent years the average dry bean yield has stagnated or declined (Wright, 1978). This has occurred in spite of the release and wide- spread adoption of cultivars which in many geographic areas have proven to be higher yielding than older varieties. Soil compaction has been suggested as a primary cause of decreased dry bean production (Smucker et al. 1978). Phaseo- lus vulgaris is not a deep rooted plant and does not have a particularly hardy root system. Thus any soil stress is likely to adversely affect root growth and yield. The soils in the Saginaw Valley are particularly prone to compaction related problems. Some of these soils are fine textured and have a high water holding capacity. They are a Charity clay loam (illitic, calcareous, mesic Aeric Hapla- quept) with about 60% clay. The soils are naturally poorly drained but subsurface drainage has been installed on most of the agricultural acreage in the region. 28 29 Soil compaction can result from natural forces such as raindrop impact, wind blown particles, and wetting and dry- ing cycles (Bowen, 1981; Cohron, 1971). However, compaction induced by human activity, primarily mechanized agricultural operations, is the primary cause of serious compaction pro- blems (Bowen, 1981; Cohron, 1971). Specifically, compaction caused by current traffic and tillage practices is known to be detrimental to root growth. Russell (1977) pointed out that increased compaction often results in decreased root proliferation and ultimately decreased crop yield. Mechani- cal impedance can directly affect root growth by restricting root elongation due to high soil strength or a limited number of pores which are of a sufficient size for root penetration (Bowen, 1981; Russell, 1977; Taylor, 1971). Soil water move- ment is decreased due to compaction (Akram and Kemper, 1979), as is air movement through the soil profile (Grable, 1971). A less extensive root system results in less water and nutri- ents available to the plant. Slower water movement through the soil profile results in longer periods of anaerobic rhi- zosphere conditions following a heavy or even a moderate rainfall. It also increases the difficulty in soil manage- ment since there are shorter periods of time when soil mois- ture is at appropriate levels to carry out tillage opera- tions. Oxygen stress, from too much water or too little air, may result in rhizosphere toxicity from ethanol or other organic compound accumulation in the root system and also make the plant more vulnerable to attack by harmful pathogens 30 (Russell, 1977). While soil compaction seems likely to be a major céuse of limited dry bean yields, other factors may also contibute. Row spacing is one such factor. Dry beans have traditionally been grown in 28 inch (71 cm) rows (Erdmann and Adams, 1978). Narrow row spacing has proven beneficial in soybeans (Cooper, 1977; Bennie et al., 1982; Mason et al., 1982; Taylor et al., 1982). In some dry bean studies narrow rows have resulted in increased yields (Atkins, 1961; Redden et a1., 1987) but re- sults were inconclusive in another study (Erdmann and Adams, 1978). Water stress is often a major cause of limited crop pro- duction. The lack of uniform rainfall distribution through- out the growing season may be more acutely noticed on soils which are compacted and do not allow for deep root growth. The compacted soils under the conventional rotation Itillage system in this study would seem to be prime candidates for reduced rooting depth and related water deficit problems. Timely irrigation should alleviate those problems, and the study included irrigation/non-irrigation as a treatment factor to examine that hypothesis. The objectives of this study were: a) to evaluate two rotation/tillage management systems for their effects on soil physical properties, root zone modification, root growth, and whole plant perfor- mance; b) c) d) 31 to evaluate the response of two dry bean cultivars at different row spacings; to study the effect of irrigation on dry bean perfor- mance as related to soil stresses; and to obtain maximum dry bean yield by incorporating vari- ous factors in a management system for soil stress alle- viation and optimization of dry bean genetic potential. MATERIALS AND METHODS We A field study of management practices for dry bean (Pha- seolus vulgaris) production was conducted in 1986 at the Sag- inaw Valley Bean and Beet Research Farm near Saginaw, MI. Each of four tiers of land, 93 by 20 m, was divided into 24 plots. Treatments were rotation/tillage, irrigation, row spacing, and cultivar. The conventional (CONV) rotation/til- lage system included a two year corn-dry beans rotation, moldboard plowing in the fall to a depth of 20 cm, and spring secondary tillage to a depth of 8-10 cm consisting of 3 pas- ses of a Danish S-tyne field cultivator with rolling baskets. The alternative rotation/tillage system was an alfalfa rota- tion, no secondary tillage (ARNST) system. It included a corn-alfalfa-alfalfa-dry beans rotation, late summer subsoil- ing to a depth of 40-45 cm, moldboard plowing in the fall to a depth of 20 cm, and no secondary tillage prior to planting. Subsoiling was carried out both in the same direction as the rows and perpendicular to the rows using a V-ripper with shanks spaced at 75 cm. Two irrigation treatments consisted of no supplemental irrigation and irrigation. Irrigation was applied to half the plots as needed beginning in mid July, or 45 days after 32 33 planting. Row spacings evaluated were 18, 36, and 54 cm. The two cultivars were C-20, a white bean developed in Michi- gan and Black Magic, a black bean. The study was arranged as a split-split plot with the rotation/tillage factors as the whole plot, irrigation ran- domly split on rotation/tillage, and row spacing and culti- vars arranged as randomized complete blocks in the sub-sub plots. The study was replicated four times. Fall tillage was carried out in late August and early September of 1985. In the spring of 1986 the CONV plots were tilled on May 1. The ARNST plots were sprayed with glypho- sate at a rate of 4.8 1 ha"1 on May 27. On June 2 two more passes of tillage were applied to the CONV plots. Planting was carried out on the same date using a grain drill (Inter- national Harvester, Model 5100 Soybean Special, 10 foot width equipped with adjustable press wheels behind the disc open- ers). A single row of spring teeth spaced midway between the seed disc openers was mounted under the grain drill hitch and used on the ARNST plots to level the untilled soil suffi- ciently to create a suitable seed bed. Seeding depth was 3.5-4 cm on the CONV plots and 2.5-5 cm on the ARNST plots, with the greater depth range on the ARNST plots due to the unlevel seed bed. Seeding rate was set to achieve a plant spacing of 7.5 cm (13.3 seeds m") regardless of row spacing. The germination test of the bean seed was low, especially the Black Magic cultivar (73%), so seeding rate was increased to 18 seeds per meter of row. A log chain was dragged behind ' 34 the drill on the ARNST plots to complete covering and improve seed-soil contact. The different row spacings were achieved by plugging the appropriate seed drop holes in the drill. Tires on the tractor were spaced at 2.26 m centers on the .30 m tires, thus providing approximately 2.0'm of soil in each plot which was unaffected by wheel traffic. Wheel traffic was controlled following tillage operations. Fertilizer (22-23-0, with 1% Zn and 4% Mn) was banded at planting at 18 cm spacing regardless of seed row spacing at a rate of 175 kg ha“. Additional fertilizer (same analy- sis) was broadcasted at a rate of 200 kg ha“ on the ARNST plots and 400 kg ha“ on the CONV plots. The different fer- tilizer rates were selected because of the differences in soil fertility due to the previous crop. Disulfoton (Di'syston), a systemic insecticide to con- trol aphids, leafhoppers, mites, Mexican bean beetle, and thrips was banded at planting at the rate of 8 kg ha‘1 in 18 cm bands on all row spacings. Need control was achieved by using a tank mix of chloramben (Amiben) and metolachlor (Du- al) (12 1 ha" and 3.6 1 ha'* respectively in 300 1 ha‘1 wa- ter) sprayed on June 3. Hand weeding was carried out during the growing season as needed. Soil moisture at planting was good below the top 3 or 4 cm of soil. The tillage on the CONV plots caused a loss of existing moisture in the top several cm of soil. Seed place- ment was just into the moist soil. Moisture conditions on the ARNST plots were better than on the CONV but seed 3S placement was less uniform due to the uneven seed bed and im- precise depth control on the drill. Some seeds were placed well into moist soil and others were too shallow to obtain good contact with moist soil. Emergence was uneven and slow due to the uneven seeding pattern. resulting from imprecise seed feeding mechanism on the drill, and the lack of rainfall following planting. It was necessary to spot replant about three weeks after planting. This was accomplished by hand and was carried out from June 26 to July 4. Spot thinning of first planting seedlings was accomplished at the same time as replanting to achieve a final seed spacing of 7.5 cm. Sprinkle irrigation was begun on July 18, 45 days after planting. Irrigation was applied at approximately 2.5 cm per week except when there was sufficient rainfall in a given week. Benomyl (Benlate) was sprayed three times for white mold (Sclerotinia sclerotiorum) control. Applications were at 10% flowering (July 29), at full flowering (August 11), and at late flowering (August 25-primarily second seeding plants). There was little evidence of white mold development in this growing season. Measurements. Soil physical properties in the top 22.5 cm of soil were measured. Undisturbed soil cores 7.5 cm in diameter and 7.5 cm long were obtained on July 17 and 18 using the Uhland dou- ble cylinder hammer method (Blake (1965). ‘Cores were weighed 36 in the field on a portable balance (Ainsworth, Denver, CO, Model SC-2000 Electronic Balance) to determine volumetric wa- ter content. Ten cores from each rotation/tillage treatment and at each of the three depths (0-7.5 cm, 7.5-15 cm, and 15- 22.5 cm) were sampled and taken into the lab for determina- tion of bulk density, soil moisture retention in the 1 to 100 kPa matric suction range, and pore size distribution. Soil cores were saturated by wetting from the bottom for at least 48 hours. Soil water retention in the 1 to 6 kPa range (1, 2, 3, 4, and 6 kPa suction) was determined using a tension table (Leamer and Shaw, 1941). A pressure plate apparatus (Richards, 1965) was~employed to determine soil moisture re- tention at matric suctions of 10, 33.3, and 100 kPa. Cores were oven dried at approximately 104 degrees C for 24 hours and weighed for bulk density determination. Total porosity of the soil was assumed to be equal to the amount of water loss between saturation and oven drying. Air porosity at each matric suction was determined by sub- tracting the measured volumetric water content from the total porosity. Pore sizes were determined on the basis of corre- sponding matric suctions and effective pore size drainage using the capillary rise formula (Vomocil, 1965). Soil moisture was monitored weekly beginning July 14, 42 days after planting, to a depth of 60 cm. Soil cores (2.5 cm diameter) were removed from the 0-10 and 10-20 cm depths and moisture at each depth determined gravimetrically. Two samples for each treatment combination (two rotation/ 37 tillage x two irrigation variables) were collected. Each of the samples contained ten subsamples which were collected from randomly selected sites in two reps of each treatment area. The soil was dried at 105 degrees C for 24 hours and moisture content calculated as the mass of water relative to the mass of dry soil. Volumetric moisture (%) at 30, 45, and 60 cm was determined using a neutron moisture meter (Campbell Pacific Nuclear, Model 50) in aluminum access tubes inserted perpendicular to the soil surface. Emergence counts were taken on June 13 and 21. Bean seedlings emerged were counted in 30.5 m of row from the cen- ter two rows (15.25 m of each row) of each plot. Above ground biomass was measured weekly beginning July 14. Ten successive plants in a row were removed (by cutting at the soil surface) from a center row of each plot, dried 48 hours at 70 degrees C, and weighed. Root growth and distribution was studied by destructive sampling. Destructive root sampling was carried out using the method described by Srivastava et al. (1982). This meth- od involves removal of a soil profile 7.5 cm x 22.5 cm x 45 cm by means of a hammer driven profile sampler mounted on a tractor. Each profile was partitioned into 18 cubes, each of which was 7.5 cm on a side. Profiles were thus divided into a 3 by 6 array of cubes. Profiles were removed at the time of maximum flowering in mid-August. Each profile was taken from the center of the plant to 22.5 cm away from the plant and perpendicular to the row. It was assumed that the 38 profile represented approximately one half of the root system of one plant. The soil cubes were soaked for 8 to 16 hours in a solution of 5% sodium hexametaphosphate to aid in dis- persing the clay and washing out the roots. The soil was washed from the extracted 7.5 cm cubes with a hydropneumatic elutriator (Smucker et al., 1982). Roots were stored in a solution of 20% methyl alcohol until analyzed. Root length was determined by Tennant's line-intersect method (Tennant, 1975), which is a modification of Newman's method (Newman, 1966). A four centimeter square grid was used. By early September pod fill was nearing completion. Ma- turity was estimated by visual observation of the percent of plants in a given plot in which greater than or equal to 50% of the leaves had lost their chlorophyll. Maturity estimates were made on September 2, 92 days after planting, and Septem- ,ber 8, 98 days after planting. On September 9 and 10 nearly 12 inches of rain fell and the plots were inundated for 10 days. Plants were completely submerged and at one point as much as 1.3 m of water stood on the plots. A harvest was attempted on September 25 by pul- ling plants from each plot to obtain yield and yield compo- nent estimates. Most of the plots were still under several centimeters of water. The beans plants in 7.6 m of row were pulled from each plot, bagged, and placed in dryers for sev- eral days prior to mechanical threshing. Two replications were harvested. Following threshing the beans were cleaned using a shaker cleaning mill and a roller mill and then 39 weighed and moisture determined. Yield data was corrected to 16.5% moisture. In addition to the 7.6 m of row, five suc- cessive plants from a randomly selected location along a row unaffected by wheel traffic were pulled and dried for deter- mination of yield components. Again, only two replications were harvested. Final population counts were made following the flood. The two reps which had not been harvested were used for these determinations. The total number of plants in six meters of row was counted. All data was analyzed using analysis of variance. Treatment means were compared using least significant differ- ence (LSD) appropriate for split-split-plot design arranged in randomized complete block design (Little and Hill, 1978). RESULTS AND DISCUSSION We. Soil bulk density was affected by the rotation/tillage factor as shown in Figure 1. Bulk density was measured to a depth of 22.5 cm, or just below the 20 cm plow layer. There were no significant differences at the 0.05 probability level in the bulk density in either the 0-7.5 cm or 7.5-15 cm depths. However, the conventional management system (CONV) did have a slightly higher bulk density at the 7.5-15 cm depth suggesting a moderate compaction effect due to the three passes of secondary tillage. The difference in bulk density at the 15-22.5 cm depth range was significant with the alfalfa rotation, no secondary tillage (ARNST) soils having a lower bulk density than the CONV soils. This dif- ference can be attributed to the plowpan which existed on the CONV soil but which was broken up by the deep tillage and deep rooted alfalfa on the ARNST soil. Douglas and McKyes (1983) found that bulk density of artificially compacted fine textured soil was not decreased in the 15 to 20 cm depth range by chiseling or plowing to a depth of 25 cm but was de- creased by subsoiling to a depth of 45 cm. In another study McKyes et al. (1979) found that chiseling to a depth of 30 cm effectively broke up compacted layers on a fine textured 40 BULK DENSITY (g cm-3) 41 1.45 1.40 1 .35 1 .30 1.25 1.20 1.15 1.10 0-7.5 7.5-15 15-22.5 DEPTH (cm) Figure 1. Soil bulk density of Charity clay at three depths as affected by rotation/tillage. 42 soil. Radcliffe et al. (1986) studied the effect of a deep- rooted legume on soil strength (which is closely correlated .to bulk density for a given soil) and found that soil strength was decreased by the action of deep-rooted plants. In the current study the effects of deep tillage and the pre- vious crop of alfalfa were probably combined which resulted in a decreased bulk density on the ARNST plots. From this study it was not possible to conclude that the decreased bulk density was due to either deep tillage or the alfalfa rota- tion. Soil bulk density is a good indicator of soil compac- tion but is limited in predicting detrimental effects on plants since it reflects only changes in total porosity (Voorhees, 1983). Soil moisture and air relations and changes provide a' better indication of the potential for reduced plant growth due to compacted soil. Soil moisture retention curves and air-filled porosity relationships for the top 22.5 cm of soil are presented in Figures 2, 3, and 4. Values at saturation are presented at 0.1 kPa matric suction (-0.1 kPa matric potential). The CONV soils had a higher volumetric water content than the ARNST soils at all suctions although the difference is not significant at the 95% level at the surface (Figure 2b). The differences are significant at the 7.5-15 and 15-22.5 cm depths (Figures 3b and 4b). The differences in drainage due to rotation/tillage are important since the Charity clay soil is prone to waterlog- ging and related problems of poor aeration and resultant 8 (m3 m-3) Po (m:5 m‘z’) 43 0.30 0.30 ,H 9°NY . . 0.1 "”l ""1 1.0 10.0 MATRIC SUCTION (kPa) Figure 2. Air-filled porosities (a) and soil moisture retention (b) of Charity clay at different matric suctions and at the 0-7.5 cm depth as affected by rotation/tillage. Pa (m3 m-a) 0 (m3 m"3) 44 0.30 : a 0.20- e 0.10: :1: [I I 1.5005 :1: I I H ARNST 0 00: o—o CONV 0.60 ‘ H ARNST l o—o CONV T I U I T t r I I FETUI I U r r I ' ' rl 0.1 1.0 10.0 MATRIC SUCTION (kPa) Figure 3. Air-filled porosities (a) and soil moisture retention (b) of Charity clay at different matric suctions and at the 7.5-15 cm depth as affected by rotation/tillage. ' 300.0 P0 (rrr3 m‘s) 0 (m5 m‘3) 0.30 45 I I I _ 0.204 0.10- 0.00- 1 MUM H ARNST o—o CONV i 0.60 0.50": I III]: 1st5 I I 0.40- “ H ARNST ‘ o—o CONV 0.30 r . . . ”.mr ......, .r..., 0.1 1.0 10.0 100.0 MATRIC sucnon (kPa) Figure 4. Air-filled porosities (a) and soil moisture retention (b) of Charity clay at different matric suctions and at the 15-22.5 cm depth as affected by rotation/tillage. 46 anaerobic conditions for the plant root system. Air filled porosity is shown in Figures 2a, 3a, and 4a. As with moist- ure retention, the differences in the top 7.5 cm are not sig- nificant (Figure 2a). Air filled porosity in the 7.5 to 15 cm depth range is higher by up to 42% on the ARNST soil than the CONV soils, although there is no statistical significance at the 95% level. While the air filled porosities at all moisture levels are lower on the CONV soils, they are always above the 10% (0.10 m’ m”) level which has been noted as a critical level below which root growth may be affected by the lack of oxygen (Vomocil and ‘Flocker, 1961; Grable, 1971; Erickson, 1982). At the 15 to 22.5 cm depth the differences due to rotation/tillage are more pronounced. The air filled porosity in the CONV soils at this depth are at or below 10% until 10 kPa matric suction, indicating the potential for in- jurious oxygen deficiencies due to waterlogging. Following a moderate to heavy rainfall or under irrigated conditions the CONV soils will take more time to reach the same air filled porosity as the ARNST soils, thus subjecting the root system to temporary but potentially serious oxygen deficiency. Erickson (1982) pointed out that aeration on most soils is usually not a problem except under heavy rainfall conditions or irrigation, and that in dry years there may be no benefi- cial effect of a loosened soil due to deep tillage. However, the nature of the Charity clay soil in this study predisposes it to aeration stress due to too much water or too slow drainage. 47 Pore size distribution for the three measured depths is shown in Figures 5 and 6. Pore size distribution, presented here by pore size radii, is essentially responsible for the air filled porosity differences noted above. Larger pores will drain at lower matric suctions than will smaller pores, as explained by the capillary model (Hillel, 1982). Pores with a radius of 2150 um will drain at a matric suction of 1 kPa, (10 cm suction) whereas pores with radii < 4.4 um will retain water at a matric suction of 33 kPa (330 cm). A ma- tric suction of 6 kPa will drain pores with a radius of 25 um. Differences in pore size distribution due to rotation/ tillage in the top 7.5 cm are not significant (Figure 5a). At the second depth (Figure 5b) there is an increase in large pores (radius >150 um) due to the ARNST management system. Although not statistically significant, the trend of de- creased pore size is evident on the CONV soil and is impor- tant when considering the effect of soil compaction. In Figure 5b the increase in pores of radii < 4.4 um is shown. Thus total pore space may or may not be affected by rotation- /tillage and soil compaction factors, but size distribution can be drastically altered with potentially severe consequen- ces for crop growth. Pore size distribution at the 15 to 22.5 cm depth was significantly affected by rotation/tillage (Figure 6). The ARNST system had more than twice as many large pores. This indicates that the compacted layer had been broken up and soil water can drain more freely. On these soils PORE VOLUME (m3 m"3) PORE VOLUME (m3 In“3) 48 0.40 0.32 0.24 0.16 0.08 0.00 0.40 0.32 0.24 0.16 0.08 0.00 >150 150-25 2.5-4.4 <4.4 PORE SIZE DISTRIBUTION (pm, radii) Figure 5. Pore size distribution on Charity clay at the 0-7.5 cm depth (a) and 7.5-15 cm depth (b) as affected by rotation/tillage. PORE VOLUME (m3 m"'3’) 49 0-40 I: CONV 0.32 0.24 0.16 160.05 0.08 I 0.00 >150 150-25 25-4.4 (4.4 PORE SIZE DISTRIBUTION (pm, radii) Figure 6. Pore size distribution on Charity clay at the 15-22.5 cm depth as affected by rotation/tillage. 50 waterlogging following rainfall or irrigation is much less likely than on the CONV soils. Conversely, the CONV soils had significantly more small pores which retain water at higher matric suctions than the ARNST soils but the water in these pores may not be available to plants. The effects of the ARNST management system in breaking up existing compacted layers were demonstrated in the various soil parameters measured. The physical condition of this fine textured soil was notably improved by the combination of a deep rooted legume in the crop rotation and the careful management of tillage. Wm Further evidence of improved soil physical conditions was found by monitoring soil moisture from 40 to 100 days after planting. Soil moisture as measured gravimetrically was almost always lower on the ARNST soil compared to the CONV soil (Figure 7). On these fine textured soils problems most often arise from too much water in the soil profile and accompanying anaerobiosis, as opposed to many less fine tex- tured soils on which water holding capacity is limited. (It is noted that high soil moisture does not necessarily mean high moisture availability to plants because water retained in fine pores may not be readily available since it is tight- ly held and roots may not be able to access it via the small pores. The data presented on pore size distribution suggest that the ARNST soils in this study would likely have better SOIL MOISTURE (Z) SOIL MOISTURE (73) 51 e-m CONVIN J+‘+ CONV IR . m—m ARNST NI m—m ARNSTIR 18_ o—m CONV NI 15 . +-+ CONVIR . q IPIIARNST’“ _ m—m ARNSTIR f F I l7 l l I t I r 40 50 60 70 80 90 DAYS AFTER PLANTING Figure 7. Soil moisture (gravimetric) over time in the 0-10 cm (a) and 10-20 cm (b) depths as affected by rotation/tillage x irrigation treatment combinations. 52 soil moisture availability to plants since more of the pores on this soil are in the range which would foster rapid drain- age of large pores balanced with good water holding capacity. Unger et al. (1981) and Russell (1977) have reviewed this subject in some detail.) Throughout the growing season the moisture content of the ARNST soils was lower than CONV soils both with and without irrigation. In fact, in Figure 7b it can be seen that ARNST irrigated treatment combination had lower soil moisture than the nonirrigated CONV soils seven of nine sampling dates throughout the growing season. Soil moisture at three depths below the plow layer was monitored using a neutron moisture meter probe. Data for the 30 cm (20-30 cm) and 45 cm (35-55 cm) depths are presented in Figure 8 (volumetric moisture content). Soil moisture at the 60 cm (50-70 cm) depth was also measured but is not reported as it was very similar to that at 45 cm. Below the plow lay- er the ARNST—treatment, both irrigated and nonirrigated, con- sistently had lower soil moisture than the CONV Soils. The graphs in Figures 7 and 8 show that fluctuations in soil moisture content due to both irrigation and rainfall are sim- ilar but the ARNST seems prone to wider swings. This is ano- ther indication of the improved drainage due to decreased soil compaction. Figure 9 compares soil moisture fluctuation with rain- fall and irrigation. Irrigation was applied at the rate of approximately 2.5 cm per week. The soil moisture data re- flects the rainfall events more prominently than irrigation. SOIL MOISTURE (x) SOIL MOISTURE (z) 35 53 32- 29-4 m—m ARNST IR m—m ARNST NI +-+ CONV IR o-o CONV NI m- . CONV NI . -I--I- CONV IR . m—m ARNST NI 46 cm " m—m ARNST IR ' I 40 50 r I l r 65 70 so 90 DAYS AFTER PLANTING V Figure 8. Soil moisture (volumetric) as measured by the neutron moisture meter over time at the 30 cm (a) and 45 cm (b) depth as affected by rotation/tillage x irrigation treatment combinations. 54 (z) BaniSIow 'IIOS m— 2. pm VN 5N on ousumwoa Hwom cam common measouw on» smegmaousu Hammcwmm IIIIIIIIUIUIIIIUIII'IIIIIUUTI'UIIIIIIllIIIIIIIIII .mucsoam cam mouse modummwuma economemu mxmfimoum< .moflumcanaoo usefiummuu sawummwuua x mmmafiuu\:0aumuom ha concommm mm umm>mm£ Hausa mewucman momma uses as Scum names 80 omuca Dru um OZ:.Z<.E anti: m>.’o’¢f¢f¢f« 330$. . . Figure 17. Root length densities of dry beans as deter- mined by destructive sampling at the time of maximum flowering and as affected by cultivar under conven- tional rotation/tillage and irrigated conditions. 69 ARNST NI C-20 86 ARNST NI BLMAG 86 Distance Dista from Plant (cm) nce from Plant (cm) Soil Depth (cm) 45.0 Root Length Density (cm cm'al moo man L75 ZJE ’o'o'o'o‘ 33°?“ ' O . . ‘ O O O O f. ..02. :o:o:o:e: ,4. . . ,, g :Ot.’.t e.g...a.a :OIOIOIOD 4.35:... Figure 18. Dry bean root length densities as determined by destructive sampling and as affected by cultivar under an alfalfa rotation/no secondary tillage, nonir- rigated management system. 70 tillage by irrigation treatment combinations. The root data, like the shoot data, revealed few differ- ences due to the rotation/tillage or irrigation treatments. There were significant differences between the two cultivars. W Maturation of dry beans is important in Michigan because harvest is often affected by adverse weather. Cultivars have been developed to take advantage of the full growing season in the state but the beans must mature rapidly and uniformly to offer the best possibility for a timely harvest. Maturity was estimated by visual observation on two dates, Sept. 2 and Sept. 8. Maturity was recorded as the percentage of plants with > 50% of the leaves losing chloro- phyll. Table 2 shows the data as estimated on Sept. 2. The range of maturity percentages was 0 to 60, with an overall mean of 9.1%. The coefficient of variation was 70%. On this date the beans on the CONV plots were more mature than those on the ARNST plots. Nonirrigated plots were significantly more mature than irrigated plots. The C-20 variety was mat- uring more rapidly than the Black Magic cultivar, with signi- ficant differences at some row spacings (generally the more narrow spacings) and depending on the other treatment combi- nations. Maturity on Sept. 8 is presented in Figure 19. There were significant interactions between row spacing and irriga- tion and between row spacing and rotation/tillage. 71 Table 2. Influence of rotation/tillage, irrigation, cultivar, and row spacing on maturity of dry beans as estimated on September 2. Percent maturity was visually estimated as the percent of plants in a plot whose leaves had lost 50% or more of their chlorophyll. , W99. ---- ARNST CONV mm ---- IR NI IR NI 89!. ----------- maturity (%) ----------- Cultixar. 18 12.5 20.0 8.8 27.5 C-20 36 3.8 15.0 7.5 40.0 54 1.3 5.0 5.0 25.0 18 2.5 8.8 o 8.8 BlMag 36 0 6.3 0 6.3 54 0 6.3 0 8.8 LSD.05 - for Rotation/Tillage means at same or different Irrigation, Cultivar, or Row Spacing - 1911‘ for Irrigation means at same Rotation/Tillage, same or different Cultivar or Row Spacing . 811., for Cultivar or Row Spacing means at same Rotation/Tillage and Irrigation - 2‘01, 72 100 90- - -I .. ‘3 8°? : -Ei '70:} d 0 1 (I) 60- - £5 50- - E 40~ .. S 30‘ ' I; T 'I I 20‘ -I—-I- CONV.NI " . o—o CONV.IR . 10" I—I ARNST.NI " ‘ H ARNSTJR 1 0 , , r 0 18 36 54 ROW SPACING Figure 19. The effects of row spacing, rotation/tillage, and irrigation on dry bean maturity 98 days after planting. Table 3 shows statistical comparison of these treatment interactions. 73 No significant differences due to cultivar were detected on September 8. The sums of squares of the interactions were partitioned into single degree of freedom components and sub- jected to F tests to examine more closely the nature of the interaction. The variance components of the interaction are presented in Table 3. The trend analysis in Table 3 shows that the effect of row spacing on maturity was linear for all combinations of rotation/tillage x irrigation, although the differences in maturity due to row spacing were significant only for the ARNST-irrigated and CONV-nonirrigated treat- ments. Figure 19 shows that the effect of row spacing on maturity was opposite for the CONV-nonirrigated plots (in- creasing maturity as row spacing increased) compared to the other treatment combinations (decreasing maturity as row spa- cing increased). The effect of irrigation averaged over the other factors was also significant on Sept. 8, with the irri— gated plants maturing more slowly than the nonirrigated plants. Final population counts were made following the mid-Sep- tember flood. Only two replications were available for plant counts. Based on this determination the only significant differences in final plant population per square meter were between the row spacings as shown in Figure 20. Final popu- lation was 472,000 plants ha"1 on the 18 cm row spacing plots, 276,800 plants ha‘1 on the 36 cm rows, and 177,100 plants ha‘1 at the widest row spacing. There were slightly higher (13%) populations on the ARNST plots at all three row 74 Table 3. Trend analysis for components of the irrigation-row spacing-rotation/tillage interactions for plant maturity as estimated on Sept. 8. Each line tests significance of row spacing on maturity under given treatments. Source of Observed Significant Variation* df SS MS F F(.05) 8 13603 ARNST,NI linear 1 400 .84 quadratic 1 5 -- ARNST,IR linear 1 4096 8.67 * quadratic 1 85 -- CONV,NI linear 1 8100 17.16 * quadratic 1 12 -- CONV,IR linear 1 784 1.66 quadratic 1 Error (c) 60 28320 472 * ARNST = alfalfa rotation, no secondary tillage, CONV conventional rotation/tillage, NI = nonirrigated, IR irrigated. PLANTS (m-Z) 60 75 554 504 45J 40- 351 30~ 254 20- 151 10 (III! CONV m ARNST 1130.05 . III 18 36 *' Row SPACING (cm) Figure 20. Final population of dry bean plants at three row spacings as influenced by rotation/tillage. 76 spacings (Figure 20) but these differences were not signifi- cant On the basis of the two replications. Since this was a management study the treatment effects on yield were a key component. Unfortunately, on September 9 it began to rain and about 30 cm (12 inches) of rain fell within 36 hours. The crop was nearing maturity as indicated by the maturity estimates made on Sept. 8. The plots were flooded for over 10 days with as much as 1.2 meters of water. On Sept. 24 an attempt was made to harvest some of the plots in order to obtain yield estimates. Two replications were harvested by pulling 7.6 m of row from each plot. Yield is reported in Table 4 for all 24 treatment combi- nations. While there are some large differences in yield, significance was difficult to detect since only two replica- tions were harvested. The only significant treatment effects were cultivar and row spacing, as shown in Table 4 and F1- gure 21. The Black Magic cultivar outyielded C-20 by 35%. The row spacing differences were also striking with the 36 cm rows yielding 31% more than the 54 cm rows and the 18 cm rows yielding 72% more than the 54 cm rows. While these differen- ces are large it should be remembered that three times as many seeds were planted on the 18 cm rows vs the 54 cm rows. Dry beans will compensate for some differences in plant popu- lation but this nearly threefold difference is too great for the wider rows to overcome. The mechanism by which yields are increased when plant rows are closer together is thought to be increased radiation interception at the time of 77 Table 4. Influence of rotation/tillage, irrigation, cultivar and row spacing on dry bean yield. Figures are the aver- age of two replications. 895W ---- ARNST CONV Armenian ----IR NI IR NI Cultixar m Spacing ------------ kg/ha ------------ 18 1909 2454 2097 2997 C-20 36 985 1696 1971 2258 54 847 1490 1072 1867 18 2878 3368 2881 3097 BlMag 36 2045 2639 1987 2999 54 1223 2251 1969 1921 LSD.05: -for Rotation/Tillage means at same or different Irrigation, Cultivar, and Row Spacing a -for Irrigation means at same Rotation/Tillage & same or different Cultivar and Row Spacing = 1101 -for Cultivar or Row Spacing means at same Rotation and Irrigation = 185. Significant (p=.95) Main Effects: Treatment—law Cultivar C-20 1803 226 BlMag 2438 Raw Spacing 18 2710 278 (cm) 36 2073 54 1580 YIELD (kg Ila—1) YIELD (kg hd-I) 78 3000 28004 ° 2600- 2400“ 5:5 I 50.05 2200- gig 20004 5;; 1800- ii: 1600- :5; 1400 5:5 . 4 BLACK MAGIC C—ZO CU LTIVAR 3000 . . . 2800- b 2600- 24-00- 1 130.05 2200- 20004 1800- 1600- 1400 . . . 18 36 54 Row SPACING (cm) Figure 21. Dry bean yield on Charity clay in 1986 as influenced by cultivar (a) and row spacing (b). Yields were averaged across rotation/tillage and irrigation. 79 flowering (Taylor, 1982; Bennie et al., 1982). While the differences are not significant at the 95% level, it is interesting to note the direction and magnitude of the differences due to irrigation and rotation/tillage. The irrigated plot average was 1822 kg ha‘”L and the nonirri- gated 2420 kg ha". This 33% difference is significant at the 90% level. The direction of this difference is contrary to that which was expected. There are several possible ext planations. Throughout the growing season there were few differences noted in root or shoot parameters due to irriga- tion. Only late in the season did shoot growth under irriga- tion surpass that of the nonirrigated plants. In a season with near adequate rainfall the effects of irrigation are few and could even be negative. If there was near adequate soil moisture throughout the growing season then irrigation could actually have a detrimental effect if applied to already moist soil. The additional irrigation moiSture could result in short-term anaerobiosis in the rhizosphere. This explana- tion seems unlikely given the timing and limited amount of water applied but no investigations were made to determine rhizosphere anaerobiosis. The soil moisture data (Figures 7, 8, and 9) suggests that any soil anaerobic conditions should have been evident in the CONV plots rather than ARNST plots since soil moisture was higher on irrigated and nonirrigated CONV plots than on irrigated or nonirrigated ARNST plots. Perhaps a more plausible explanation is that the beans were at different stages of maturity at the time of the 80 flood. There was considerable seed damage due to the flood, and beans at different maturity levels may have responded differently to the flooded condition. The nonirrigated plants were more mature than those irrigated. If the more mature seeds withstood a prolonged flood this might be re- flected in a set of results favoring the mature seeds. The average yield on the ARNST plots was 1982 kg ha"1 and on the CONV plots 2260 kg ha". As indicated above this 14% difference is not significant. Again, however, the work- ing hypothesis for this study suggests that yields should be increased when soil physical conditions are improved. That was not the case this year, as soil physical conditions were improved but yields were not. Yield components are presented in Table 5. Pods per plant increased as raw spacing increased with the exception of the ARNST irrigated C-20s which decreased from 18.2 to 13.5 when row spacing increased from 18 to 36 cm. This dif- ference is not significant. There was no clear effect due to irrigation or rotation/tillage, although there was a signifi- cant interaction between these two treatment factors. Pods per plant was higher on the irrigated ARNST plots than the nonirigated ARNST plots but the opposite occurred on the CONV plots. The C-20s had significantly more pods per plant than did the Black Magics. Seeds per plant also increased with increasing row spa- cing. On the CONV soils the irrigated plants had mcre seeds than did the nonirrigated. However, on the ARNST soils no 81 Table 5. Influence of rotation/tillage, irrigation, cultivar, and row spacing on dry bean yield components. Data are the average of two replications. Wises. ----- ARNST CONV mm ----- IR NI IR NI Cultlxar 8.0x Spacing -------- pods/plant -------- 18 18.2 17.2 15.0 14.1 C-20 36 13.5 20.9 23.7 15.0 54 25.0 20.1 26.1 22.2 18 8 3 9.2 9 3 9 2 BlMag 36 13.2 15 8 16 9 15 3 54 18.5 23.1 16 5 21 9 LSD.05: - for Rotation/Tillage means at same or different - Irrigation, Cultivar, and Row Spacing = - for Irrigation means at same or different Cultivar and Row Spacing = 311‘ - for Cultivar or Row Spacing means in the same column = 610* ------- seeds/plant ------- 18 76.7 65.5 60.8 49.7 C-20 36 61.3 80.6 100. 54.5 54 116.4 84.6 106. 87.9 18 47.8 49.5 52. 47.5 BlMag 36 68.9 75.0 93. 66.2 54 106 5 113 5 93. 111.5 LSD.05 for Rotation/Tillage means at same or different Irrigation, Cultivar, and Row Spacing = 164. for Irrigation means at same or and Row Spacing for Cultivar column 21.1.. or Row Spacing means in the same different Cultivar 82 trend was obvious as some row spacing by cultivar combina- tions had more seeds per plant under irrigation while others had more seeds per plant without irrigation. There was a significant difference due to cultivar with C-20 having more seeds per plant than did Black Magic. An attempt was made to determine seed weight but the seeds had been too badly damaged by the flood and useful data could not be obtained. Table 6 presents a comparison of meters of root required to produce a gram of seed under the various treatment combi- nations in the study. According to this data, ARNST system plants required more root length than the CONV plants to pro- duce a gram of seed. The only exception among the treatments examined was the nonirrigated Black Magic combination, in which the ARNST plants produced a gram of seed with 6% less root length. The largest differences were between the irrigated and nonirrigated treatments. The irrigated treatments required 46% more root length than the nonirrigated (averaged across the other treatments) to produce a gram of seed on both rota- tion/tillage treatments. It was expected that irrigated plants would be more efficient in seed production per unit length of root due to less stress on the plant system. How- ever, the root length per gram of seed data is based on seed yield reported in Table 4, which showed that the nonirrigated dry beans yielded more than the irrigated plants. It was suggested that the difference in maturity between the 83 Table 6. Root efficiency in seed production as influenced by rotation/tillage, irrigation, and cultivar. Root length was determined by destructive sampling and is based on the assumption that the extracted profile contained one half of the roots from a single plant. We: --- ARNST CONV Irrigation--IR NI IR NI Caitlin. ---------- meters root/gram seed ---------- C-20 49.77 28.55 39.75 24.76 BlMag 38.07 27.98 35.18 29.88 irrigated and nonirrigated plants at the time of the flood was important in the yield data, and that same factor was im- portant in the root efficiency comparisons shown in Table 6. There was an interaction between the cultivar and irri- gation on root length per unit yield. The C-20 cultivar re- quired about the same or less root length per weight of seed produced on the nonirrigated plots than the Black Magic cul- tivar. However, under irrigated conditions the Black Magic variety appeared to be more efficient in producing seed as evidenced by the lower root length per seed produced than the C-20 cultivar. SUMMARY AND CONCLUSIONS The combination of a deep rooted legume and careful til- lage management was expected to result in an increase in dry bean yield. Soil compaction is a known problem on the fine 84 textured Charity clay soil and root growth may be mechanical- ly impeded or slowed due to a lack of aeration, moisture, or other related problems. The soil physical measurements made confirmed the presence of a compacted layer at about the 20 cm, or plowpan, depth. There were also indications of a les- ser secondary tillage pan between 6 and 9 cm depth. The stu- dy covered only one year, a year that contained some unique problems. In particular the September flood which eliminated the possibility of a harvest made it a difficult year in which to obtain useful data. The variable initial stand of beans and subsequent overplanting resulted in two different sizes of plants throughout the growing season. The differences in the soil physical properties did not result in many significant differences in the plant parame- ters. Shoot biomass was largely unaffected by rotation/til- lage treatment, as was root length density. No differences were detected in bean yield due to rotation/tillage. Soil parameters were measured only to‘a depth of 22.5 cm but root responses were measured to twice that depth. Future studies should include measurement of soil physical proper- ties to a depth of at least 45 cm in order better quantify changes due to deep tillage and the deep rooted legume crop and to interpret root data. Differences due to row spacing were noted for shoot bio- mass as well as yield. This one year of data suggests that decreased row spacing will result in additional yield. Data from previous years at the same location support this finding 85 (Smucker, annual reports from the Saginaw Valley Bean and Beet Farm). In future studies it would be good to decrease intrarow seed spacing on the narrow rows in order to achieve the same population per unit area and thus study response to row spacing without the different populations. The responses of the two cultivars were consistent. The Black Magic cultivar is known to have a more hardy root sy- tem and to be a high yielding variety. In this study the Black Magic variety did have higher root length densities and ultimately a higher yield regardless of the other treatment factors. Since the white navy bean is the bean of preference for most growers in Michigan, further investigation should be conducted to examine the reasons for the superior response of the Black Magic cultivar. Additional studies should be conducted to investigate the effect of row spacing on emergence to verify if the dif- ferences noted in this study were due to planting equipment or to some other factor. The effects of rotation/tillage and its interaction with row spacing and irrigation could also be studied further to answer questions related to physiological maturation. Root studies on different row spacings would be useful to examine the relationship of root growth and distribution and overall plant response to different row spacings. One year of field data is insufficient to draw conclu- sions which can be assumed to be widely applicable. That is true especially of a study in a year of a highly unusual 86 flood as occurred in 1986. This study, with appropriate mod- ifications, could be profitably repeated in order to verify the results and develop recommendations useful to dry bean producers. REFERENCES Akram, M. and W. D. Kemper. 1979. Infiltration of soils as affected by the pressure and water content at the time of compaction. Soil Sci. Soc. Am. J. 43:1080-1086. Atkins, J. D. 1961. Row spacing influences on yield of snap a dry beans. Farm Research 27:13. Bennie, A. T. P., W. K. Mason, and H. M. Taylor. 1982. Responses of soybeans to two row spacings and two soil water levels: III. Concentrations, accumulation, and translocation of 12 elements. Field Crops Res. 5:31-43. Blake, G. R. 1965. Bulk density. In C. A. Black et al. (eds.) .Methods of soil analysis: Physical and mineralogical properties. Part I. Agronomy 9:374-390. Bowen, H. D. 1981. Alleviating mechanical impedance. p. 21-58. In G. F. Arkin and H. M. Taylor (eds. ) Modifying the root environment to reduce crop stress. Am. Soc. Agric. Eng., St. Joseph, Michigan. Cohron, G. J. 1971. Forces causing soil compaction. p. 106- 119. In K. K. Barnes et al. (eds.) Compaction of agricultural soils. Am. Soc. Agric. Eng., St. Joseph, MI. Douglas, E. and E. McKyes. 1983. Tillage practices related to limiting plant growth factors and crop yields. Canad. Agr. Eng. 25(1):47-55. Erickson, A. E. 1982. Tillage effects on soil aeration, p. 91-104. In P. W. Unger and D. M. Van Doren, Jr. (eds.) Predicting tillage effects on soil physical properties and processes. Amer. Soc. Agron. Spec. Publ. 44. Grable, A. R. 1971. Effects of compaction on content and transmission of air in soils. p. 154-164. In K. K. Barnes et al. (eds.) Compaction of agricultural soils. Am. Soc. Agric. Eng., St. Joseph, MI. Hillel, D. 1982. Introduction to 3011 Physics. Academic Press. Orlando, Florida. 365 p. 87 Huck, M. G., C. M. Peterson, G. Hoogenboom, and C. D. Busch. 1986. Distribution of dry matter between roots and shoots of irrigated and non-irrigated determinate soybeans. Agron. J. 78:807-813. Klute, A. 1965. Laboratory measurement of hydraulic conductivity of saturated soil. In C. A. Black et al. (eds.) Methods of soil analysis. Agronomy 9:210-221. Leamer, R. W. and B. Shaw. 1941. A simple apparatus for measuring noncapillary porosity on an extensive scale. J. Am. Soc. Agron. 33:1001-1008 Little, T. and F. J. Hills. 1978. Agricultural experimentation. Wiley and Sons, NY. p. 101-113. Maertens, C. 1987. Ways of using endoscopy to determine growth and quality of root systems. In H. M. Taylor (ed.) Minirhizotron observation tubes: methods and applications for measuring rhizosphere dynamics. Am. Soc. Agron., Madison, '1. Mason, W. K., H. R. Rowse, A. T. P. Bennie, T. C. Kaspar, and H. M. Taylor. 1982.‘ Responses of soybeans to two row spacings and two soil water levels. II.Water use, root growth, and plant water status. Field Crops Res. 5:15-29. McKyes, E., S. Negi, E. Douglas, F. Taylor, and V. Raghavan. ' 1979. The effect of machinery traffic and tillage operations on the physical properties of a clay and on yield of silage corn. J. Agri. Eng. Res. 24:143-148. Newman, E. I. 1966. A method of estimating the total length of root in a sample. J. Applied Ecol. 3:139-145. Phillips, R. E. and D. Kirkham. 1962. Soil compaction in the field and corn growth. Agron. J. 54:29-34. Radcliffe, D. E., R. L. Clark, and M. E. Sumner. 1986. Effect of gypsum and deep-rooting perennials on subsoil mechanical impedance. Soil Sci. Soc. Am. J. 50:1566-1570. Redden, R. J., T. Usher, D. Young, R. Mayer, B. Hall, A. Fernandez, and D. Kirton. 1987. Response of navy beans to row width and plant population in Queensland. Austral. J. of Exp. Agric. 27(3):455 Reicoskey, D. C., H. R. Rowse, W. K. Mason, and H. M. Taylor. 1982. Effect of irrigation and row spacing on soybean water use. Agron. J. 74:958-964. Richards, L. A. 1965. Physical condition of water in soil. In C. A. Black et al. (eds.) Methods of soil analysis. Agronomy 9:128-152. 89 Russell, R. S. 1977. Plant root systems: Their function and interaction with the soils. McGraw Hill, London. 298 p. Smucker, A. J. M., D. L. Mokma, and D. E. Linvill. 1978. Environmental requirements and stresses. p 45-61. In L.S. Robertson and R. D. Frazier (eds.) Dry bean production- principles and practices. Mich. Agric. Exp. Stat. Ext. Bull. E-1251. Lansing, Michigan. S. L. McBurney, and A. K. Srivastava. 1982. Quantitative separation of roots from compacted soil profiles by the hydropneumatic elutriation system. Agron. J. 74:500-503. Srivastava, A. K., A. J. M. Smucker, and S. L. McBurney. 1982. An improved mechanical root sampler for the measurement of compacted soils. Trans. Am. Soc. Agric. Eng. 25:868-871. Taylor, H. M. 1971. Effects of soil strength on seedling emergence, root growth, and crop yield. p. 292-305. In K. K. Barnes et al. (eds.) Compaction of agricultural soils. Am. Soc. Agric. Eng., St. Joseph, MI. , W. K. Mason, A. T. P. Bennie, and H. R. Rowse. 1982. Responses of soybeans to two row spacings and two soil water levels. I. An analysis of biomass accumulation, canopy development, solar radiation interception, and components of seed yield. Field Crop Research. 5:1-14. Tennant, D. 1975. A test of a modified line intersect method of estimating root length. J. Ecol. 63:955-1001. Unger, P. W., H. V. Eck, and J. T. Musick. 1981. Alleviating plant water stress. In G. F. Arkin and H. M. Taylor (eds.) Modifying the root environment to reduce crop stress. Am. Soc. Agric. Eng., St. Joseph, MI. Vomocil, J. A. and W. J. Flocker. 1961. Effect of soil compaction on storage and movement of soil air and water. Trans. Am. Soc. Agric. Eng. 4(2):242-246. . 1965. Porosity, In C. A. Black et al. (eds.) Methods of soil analysis. Agronomy 9:210-221. Voorhesss, W. B. 1983. Relative effectiveness of tillage and natural forces in alleviating wheel-induced soil compaction. Soil Sci. Soc. Am. J. 47:129-133. 90 Vos, J. and J. Groenwold. 1987. The relation between root growth along observation tubes and in bulk soil. In H. M. Taylor (ed.) Minirhizotron observation tubes: methods and applications for measuring rhizosphere dynamics. Am. Soc. Agron., Madison, WI. Wright, K. T. 1978. Production trends: World, U.S., and Michigan. p. 16-30. In L.S. Robertson and R. D. Frazier (eds.) Dry Bean Production - Principles and practices. Mich. Agric. Exp. Stat. Ext. Bull. E-1251. Lansing, Michigan. CHAPTER 3 EVALUATION OF MINIRHIZOTRON OBSERVATION TUBES AS A TOOL FOR ROOT STUDY ON FINE TEXTURED SOILS INTRODUCTION Studies of plant root systems are critical to an in- creased understanding Of the plant system and its interaction with the environment. The current status of rhizosphere knowledge lags behind that of the above ground parts of the plant. Because roots grow below the surface of the soil they cannot be studied directly without destroying them and chan- ging the soil matrix. The conventional method of root stu- dies has been destructive sampling. With this method both soil and roots are extracted and separated and only then can the roots be examined. Destructive root studies are import- ant but have the limitation of not allowing study over time. Thus the effects of a given treatment on the root system may be studied at a given point in time but it is impossible to study effects over time on the same root system. In addi- tion, destructive sampling is both time consuming and costly. Attempts have been made for many years to study root growth in situ. Methods have included glass walled rhi- zotrons of various sizes, glass or clear plastic panels pressed against the sides of trenches dug into the soil, and 91 92 more permanent root laboratories. More recently, glass or plastic tubes inserted into the soil have been used for root observation. These tubes and accompanying equipment have become known as minirhizotrons. According to an historical overview by McMichael and Taylor (1987), G. H. Bates in 1937 was the first to use such a tube. Since 1971 the technique has gained widespread acceptance for root studies. The deve- lopment of improved mirrors, fiber optics, mini or micro video cameras and portable recording equipment has made it possible to more effectively use minirhizotron tubes as in situ root study tools. For a more complete review, see McMichael and Taylor (1987). * Minirhizotron tubes of various sizes are used, ranging from 6 to 150 mm in diameter depending on the boring equip- ment available (McMichael and Taylor, 1987). The most common- ly used size is around 50 mm (Brown and Upchurch, 1987). Most of the tubes used have been round (Upchurch and Ritchie, 1983; Maertens, 1987; Merrill et al., 1987; Levan et al., 1987) but square tubes have also been used (Waddington, 1971). The tubes are installed at various angles, usually from 30 to 45 degrees from vertical (Brown and Upchurch, 1987). Orientation with respect to plants also varies depen- ding on the objectives of the study, personal preference of the researcher, or physical constraints related to installa- tion and root observation. Upchurch and Ritchie (1983) stu- died four different tube orientations including parallel or perpendicular to the row and within the row or between two 93 rows. They found few significant differences in root obser- vations due to tube placement. Many different observation and recording devices have been employed, including a series of mirrors, various scopes with appropriate light source, images transferred to and recorded by 35 mm cameras, and video cameras designed specifically for minirhizotron ob- servation and recording. Brown and Upchurch (1987) have re- viewed this equipment in some detail. Minirhizotrons have been used to study roots of various plants, including trees, grasses, grains, and food crops (Brown and upchurch, 1987). Numerous aspects of rhizosphere dynamics are .being studied. Smucker et al. (1987) suggested there may be as many as 35 categories of information avail- able about roots and the soil system from minirhizotron ob- servations. Data collected includes root color, branching, length, depth, density, diameter, and lateral root spread (McMichael and Taylor, 1987). Minirhizotrons are also used to collect data on soil macro and mesofaunal populations, root turnover rates, nodules on leguminous crops, pathogens, and root diseases (Smucker et al., 1987). Different methods are used for tube installation, inclu- ding hydraulically driven bores (Upchurch and Ritchie, 1983) or hand operated augers (Merrill et al., 1987). Most resear- chers agree that proper installation of the tubes is critical if useful results are to be obtained (Upchurch and Ritchie, 1983; Brown and Upchurch, 1987; Smucker et al., 1987; and Maertens, 1987). Data must be representative of root 94 activity in the surrounding bulk soil. Problems related to installation may affect root growth and thus prejudice data collected. Installation of a tube obviously disturbs the soil and the best that can be hoped for is that disturbance is minimal and will not appreciably affect the growth of roots. If the soil around the tube is compacted, root growth at the tube-soil interface will be restricted and thus not representative of that in the surrounding bulk soil. On the other hand the hole into which the tube is inserted must not be so large as to create unnatural voids at the tube surface as this may favor root proliferation and yield an inaccurate picture of root activity. The ideal installation will result in good contact between the soil and tube without large voids or compacted soil. Factors affecting installation include boring method, cleaning of the hole, and soil conditions at the time of installation. For a review of installation tech- niques see Brown and Upchurch (1987). Minirhizotrons have been used on many different soils. Some problems have been encountered on certain types of soil. Such problems may be related to installation techniques or to the physical properties of the soil. Good results have gen- erally been obtained with sandy soils. However, Vos and Groenwold (1987) encountered a problem on a sandy soil with a compacted layer at the 30 cm depth. In their study the installation of minirhizotron tubes resulted in breakup of the compacted layer, creation of larger vOids at the tube- soil interface than in bulk soil, and root proliferation at '95 the tube which was much greater than in the surrounding bulk soil. Fine textured soils have presented more problems. Maertens (1987) concluded that minirhizotrons may be useless on clay soils if not installed very carefully. This is due to the potential for smearing, soil compaction at the tube- soil interface, or creation of unnatural voids which will favor root proliferation. The capacity of many clay soils to shrink and swell make it difficult to maintain a good tube- soil interface. Maertens suggests that these problems can be overcome by boring a hole larger than necessary, loosening the soil wall with a rotating brush, and introducing a tube with a flexible outer membrane. This outer membrane is then inflated and takes the shape of the soil wall. Merrill et al. (1987) has utilized a similar system to counter the un- natural effects of boring a hole for tube insertion. One of the problems encountered in using minirhizotrons is the poor correlation of data from the top 20-30 cm of the tubes with that of destructive sampling. Upchurch and Rit- chie (1983) found that root length densities from the top 20 cm were severely underestimated by the minirhizotron method compared to the results from destructive sampling. Vos and Groenwold (1987), Beyrouty et al. (1987), and Smucker et al. (1987) noted the same phenomenon. Reasons suggested for this discrepancy include the movement of dry soil away from the tube for an undetermined reason (Upchurch and Ritchie, 1983), the influence of temperature differences at the tube-soil interface (Upchurch and Ritchie, 1983; McMichael and Taylor, 96 1987) and the introduction of light due to soil disturbance during tube installation or improper sealing of the tube to exclude light (Levan et al., 1987). Many researchers have found that the number of replica- tions needed to achieve satisfactory results is approximately twice that of destructive sampling. Upchurch and Ritchie (1983) found very little- correlation between the results of single tube observations and bulk soil root length densities. They suggest that several tubes must be averaged before mean- ingful results can be expected. Vos and Groenwold (1987) in a series of experiments used an average of approximately twice as many minirhizotron tubes as destructive samples in order to obtain reliable results. Data collected has been presented in various forms. Equivalent root length density (Merrill et al., 1987), number of roots observed per unit area (Smucker et al. 1987), root length density (Upchurch and Ritchie, 1983), and root length per area (Vos and Groenwold, 1987) have all been reported. Upchurch (1987) discusses the derivation of root length den- sity from minirhizotron root observations and concludes that this is a valid conversion if certain assumptions and con- straints are understood. Minirhizotron data presents some interesting statistical problems. Upchurch and Ritchie (1983) suggest that since the numbers of root observations per unit area are often quite low and include many zeros the data does not lend itself to classical parametric statistical analysis. They suggest the 97 need for some transformation so that parametric statistical analysis can be used. Alternatively, nonparametric statisti- cal tools could be used. Glenn et al. (1987) found that the variance of treatment means was positively correlated with the mean, a case which violates the assumptions of the anal- ysis of variance. A transformation was used to reduce this correlation in order to subject the data to analysis of var- iance. There is need for additional work in the area of sta- tistical analysis of minirhizotron data. In summary, the minirhizotron is being effectively used for in situ root study. Numerous pieces of information can be gathered with this method. The system may be initially expensive but may prove to be much less costly in the long run than conventional destructive sampling. One major advan- tage is the ability to study root activity over time. How- ever, there are potentially serious problems which may pre- judice results. These include the tube-soil interface ef- fect on root growth, changes in temperature or moisture con- ditions due to the presence of the tube, and a problem of currently undetermined origin with rooting density in the top 30 cm of soil around minirhizotrons. In addition, there are statistical problems associated with handling minirhizotron data which must be addressed. The current study was undertaken as part of a management studylfor dry bean (Phaseolus vulgaris) production. Root re- sponses of two cultivars to different rotation/tillage man- agement systems and to irrigation were studied both by 98 destructive sampling and minirhizotron video recordings. By using the two methods it was possible to consider the rela- tive merits of each. The destructive sampling was considered to be the standard against which the minirhizotron method was compared. The objective of this two year study was to evalu- ate the minirhizotron system for use on fine textured soils and under different soil management practices. MATERIALS AND METHODS A two year study of the minirhizotron microvideo method of root study was conducted as part of a management study for dry bean production. The study was conducted at the Saginaw Valley Bean and Beet Research Farm near Saginaw, Michigan in 1985 and 1986. The soil on the farm is a fine textured Char- ity clay (illitic, calcareous, mesic, Aeric Haplaquept) which is artificially drained. The soil has about 60% clay and is subject to severe compaction by agricultural traffic. Primary and secondary tillage pans are common and can be re- strictive to root growth as well as to air and water move- ment. The management study included two rotation/tillage variables, irrigation vs nonirrigation, and two dry bean cul- tivars. The rotation/tillage variables were: 1) convention- al (CONV), which included a crop rotation of corn-dry beans, fall moldboard plowing, and two to four passes of spring sec- ondary tillage; and 2) alfalfa rotation, no secondary tillage (ARNST) which utilized a four year rotation of corn-alfalfa- alfalfa-dry beans, deep tillage in the fall of the second year of alfalfa followed by moldboard plowing, and no spring secondary tillage. Half of the plots were sprinkle irrigated as needed and the other plots received no irrigation. The two cultivars were C-20, a white navy bean, and Black Magic, 99 100 a black soup bean. Both cultivars were studied in 1985 but in 1986 minirhizotron tubes were installed only on the C-20 plots. Data is reported for the C-20 variety. Row spacing was 50 cm in 1985 and 54 cm in 1986. Minirhizotron tube installation was completed soon after planting and prior to or immediately after crop emergence each year. Clear butyrate plastic tubes 1.83 m long and 51 mm inside diameter with a 3.2 mm wall were installed at a 45 degree angle directly under and parallel to the bean rows. Tube holes were bored using a modified trailer mounted hy- draulic soil sampling probe (Giddings Model GSRP-ST) e- quipped with a cutting bit designed to compact inward rather than outward. After each hole was bored it was cleaned out using a round wire brush. In 1985 the brush was pushed and pulled up and down the hole without rotating it. In 1986 the brush was rotated as it was moved up and down the hole. The change was implemented in 1986 in an effort to decrease the possibility of deep striations or channels resulting from the boring and cleaning operation. Tubes were pushed into the holes by hand or with light tapping. Every attempt was made to insure that the tubes fit snugly. Observations throughout both seasons confirmed that tubes had been inserted without smearing. Tubes were tightly capped at both top and bottom with number 11 rubber stoppers. After insertion the protru- ding portion of tube (approximately 30 cm) was painted with black paint to exclude light and later with white paint to reduce solar heating of the tube. 101 Two tubes were installed in each of four replications for a total of eight minirhizotron tubes for each treatment combination. A microvideo camera (Circon color bore inspection sy- stem, Model MV-9011 Agricultural Camera) was inserted into the tube and video images of the roots recorded using a modi- fied hand held Hitachi monitor-viewfinder, portable video cassette recorder (Panasonic Model NV-8420) and portable com- puter for recording date, depth, and tube identification. Video recording was begun in mid July and was done at appro- ximately one week intervals throughout the summer. Video re- cording was carried out to the depth of the deepest visible root in each minirhizotron. Video taping was done incremen- tally with each frame representing an area 1.2 by 1.8 cm, or 2.16 cm”. The root images were later counted manually using a 13 inch color monitor and recorded as number of root observa- tions per video frame. Destructive root sampling was carried out in mid-August of each year (at or near maximum flowering) using the method described by Srivastava et al. (1982). This method involves removal of a soil profile 7.5 cm x 22.5 cm x 45 cm by means of a hammer driven profile sampler mounted on a tractor. Each profile was partitioned into 18 cubes (3 x 6 array), each of which was 7.5 cm on a side. Profiles were removed at the time of maximum flowering which was in mid-August. Each profile was taken from the center of the plant to 22.5 cm away from the plant and perpendicular to the row. It was 102 assumed that the profile represented approximately one half of the root system of one plant. The soil cubes were soaked for 8 to 16 hours in a solution of 5% sodium hexametaphos- phate to aid in dispersing the clay and washing out the roots. The soil was then washed from the extracted 7.5 cm cubes with a hydropneumatic elutriator (Smucker et al., 1982). Roots were stored in a solution of 20% methyl alcohol until laboratory analysis. Root length was determined by Tennant's line-intersect method (Tennant, 1975), which is a modification of Newman's method (Newman, 1966). A four cen- timeter square grid was used. RESULTS AND DI SCUSSI ON Minirhizotron root Observations for 1985 and 1986 are presented in Figures 1-4. Root observations were compared on the basis of treatment combination (rotation/tillage by irrigation) and these data are presented in Figures 1 and 3. Root growth was also studied over time and the data from four dates is shown in Figures 2 and 4. Root observations were averaged across the eight replications and for every 10 cm of soil depth to smooth the large variability among individual video frame observations. Upchurch and Ritchie (1983) showed that root observations from individual tubes have very little correlation with bulk soil rooting patterns. The destructive sampling data from both 1985 and 1986 is presented in Figures 5-8. Each square in the "rootmaps" re- presents the root length density (RLD) in a cube of soil 7.5 cm on a side. The relative position of the square in the rootmap represents the location of the soil cube in relation to the plant. Few roots were observed in the minirhizotron tubes in the top 20 cm of soil under any Of the treatments. Most roots observed in the tubes were in the 20-60 cm depth range. The rootmaps in Figures 5-8 show that the highest root length density was in the top 7.5 cm of soil, and RLD generally 103 ARNST NI 1985 10" ARNST IR 1985 8001 mm (an-2) m oasmmnous (om-2) _ ‘1. ‘ 6 4 s a b I d OMY11' 6.101111- ::.IIII.Y241 11.101324. ”1.11.st vJIILYze limos ' mAUGB ‘ Figure l. Minirhizotron root observations for four rota- tion/tillage (ARNST or CONV) and irrigation (IR or NI) treatment combinations in 1985 as affected by date of observation. Data points are the average number of roots observed cm"2 by ten cm depth increments. JULY 11. 1985 105 JULY 24. 1985 ROOT oasamnoxs (em-2) ROOT oasenwmous (an-2) IoARNSTl!‘ :tARNSTIU. IICONVIR IICONVII‘ A A JULY 29. 1985 Figure 2. Minirhizotron root observations on four dates in 1985 as affected by rotation/tillage (ARNST or CONV) and irrigation (IR or NI) treatment combina- tions. Root observations are reported as the average number of roots observed cm'2 in ten cm depth increments. 106 ARNST NI 1986 ARNST IR 1986 11007 OBSERVATIONS (em-2) ROOT oassmnous (em-2) 0 1 2 3 4 0 1 z 3 A A m A A A A A A L A A A Figure 3. Minirhizotron root observations for four rota- tion/tillage (ARNST or CONV) and irrigation (IR or NI) treatment combinations in 1986 as affected by date of observation. Data points are the average number of roots observed cm"2 by ten cm depth increments. JULY 31. 1986 10" AUGUST 9. 1986 ROOT mom («114) ROOT oasamnous (an-2) ° - 1 . E . 9 4 o 1 z 3 4 Figure 4. Minirhizotron root observations on four dates in 1986 as affected by rotation/tillage (ARNST or CONV) and irrigation (IR or NI) treatment combina- tions. Root observations are reported as the average number of roots observed cm"2 in ten cm depth increments. Soil Depth (cm) pm LI :3 1'3 II 37.5 108 CONV IR 85 ARNST IR 85 Distance from Plant (cm) a D 4 3:0:0:0:6:0:0:0:0:9‘0 .4. 3.0 0.0.0:... o. . o O . .o.o a: '0’. 9.9.15.9...4000400 04040490004 40...... so oooooeooo ooeoeooo u eooooeooo oooooe {55353535 oooeo 0 404449. e 4. ‘ o ¢%%!#V%%9 ';?§§§§}9§5} . . b. 4 .. .. ., .0 '4’4'0'0’0'0'4'4’0’0‘ ’o'o’o'o’o’o'o’e’o’o‘ 'o'o'o'o’o’o’o'o’ FOOOOOOOOOUO.....O..”..O....O 0....000000......OOOCOOOOO. . ’ ...'....U.O.......4' . o 9' '4‘4 ~’e’o‘e’o°o°e'o’o’o’1 ’o'OIo’o o o o o o I ' O O. .0...’................‘ -:e:e:o:o:o:e:o’o:o:c {o:o:o‘o:o’o:o:o:o:« .’ ’e ’0 ° 3...... .. .‘O4P....... ’66600666564668‘06’.......... > 0...... so 4.. a.» 0.. so so ea on 0.0 0.0 0.9 6.. .4 $’e’o’o’e’o’o’o’o’o‘ ’o'o ’o’e’e'o’o’o’o’ 'o'o’o’o’o’e’e’o’o’e‘ reooeooeoonoooooooeonoooooooo eoooeoeeoeeooooooeoeooo ooeoo r o 9.0 o 9 9.0.9.0; 30.0.0.0...0” o . go o o o e ueeooooooou O. o e o o o o o e e |000090.00090.00090000090000604 ’0.0.0.......0.0...l 30.0.6.0...O...O.. 4 ’O...‘.......O...O.4 Root 0 Length Density (cm cm'3) .80 L75 ass . ‘ 06.9.0.‘ 3.0.0.0... ... 0‘0. . .‘.‘. 4 Figure 5. Root length density as determined by destruc- tive sampling in mid-August, 1985 and as influenced by rotation/tillage under irrigated conditions. Soil Depth (ca) .9 (I o a 109 CONV NI 85 Distance from Plant (cm) 0 0 0 :0. . 9 0.0 0.0.0 0 9‘9‘9’9’0’0’9’9 9 0.0.9. .0.0.0.0 0.0 . 0.0 0 0 9 .0 0 9 0 9 0 0 . . . . . . . .Olo o. xaeexaeasaeazaeaaeexneea .0900 00....909000.0.9.099 9.0 ».9.09.009n.9.9. 999....009..9.. vvvvvvvvv95%%%% >...9..0.9u0....00.9o90009000w .;.;.;.;.;.;.;.;.; O O O O O O O O . 9 0.0.9. . 0 . . 0 9.9.0 ’0’. ..’vv¥vvv0¢v9”’°° ’ ' -.....00.9u...9.9.00 9......0w 5&flflflflfififl’flfifihhhfiflaf‘&&hfi&fl&fifi » , O O O O,“ or. .,. .l. .l.,.l1‘..,.‘.‘. O O O O O C ‘33“?UV5VW5%VVU3WQV5VVWWV QO."O........€'.........‘ . 0.0.999....0.0..0. .... .’.’.‘9‘.’.’.’.°.°9‘ '0 0 . . 0 0 0 9’ ’.‘ 5.9.0’0'0’0'9’0' ‘9' ”’°"”"“°°’°"”'""fififi’°°“ TVVVVVVVVUvvvvvvvvvwvvvvvvvvvv . . 9 9 . . 9 0.0.0.1 -.9 ..0...9..'0.0...a “0.0.0.... ......0 . -...0..00.o 0.......n.0~p.0009 O . -.....0.0.0.0.9...9 Root 0 swuryssss 3.0.9:. .‘.’.’ ’ . OO~ .9. >9. 3.? '0’ 5310:. ' ' ’ O. .‘0 4- Length Density (cm cm‘a) an 75 a. ‘99.9: 0006......9 ARNST NI 85 Distance from Plant (cm) v 9 O O O ‘ 0'.'.'0'.'9 .‘ - 6.9.9.0” {0:03:93 . 9 . "."..090.9.. 0 C O O .O?090.9..4 3533?: Figure 6. Root length density as determined by destruc- tive sampling in mid-August of 1985. The effects of rotation/tillage under nonirrigated conditions are shown. 110 CONV IR 86 ARNST IR 85 Distance Distanc from Plant (cm) e from Plant (cm) 9011 Depth (ca) “.0 Root Length Density (cm cm'a) can 1J5 ass 0 .v.‘ 'Q'Q'O'.‘ ~ 9 0 . 0 ‘ V . . . 59.9.0.4 .’ t ‘ Figure 7. Root length density as determined by destruc- tive sampling in mid-August, 1986 and as influenced by rotation/tillage under irrigated conditions. 111 CONV NI 86 ARNST NI 86 Distance Distance from Plant (cm) from Plant (cm) 5011 Depth (cm) I). 00 Root Length Density (cm cm‘a) mes L75 ass bv.v.v.i ;.;.;.;9; :.:.:.:.9.- “fix”... .- ....i .09. ....‘.".‘.°.'.'. fiDfi6>§§9¥é¢éflk§$fi '....a.d 39.9.9.4 3....9’.‘ '03'03’0. '. Figure 8. Root length density as determined by destruc- tive sampling in mid-August of 1986. The effects of rotation/tillage under nonirrigated conditions are shown. 112 decreased with soil depth. This discrepancy between the root observations in the minirhizotron tubes and rooting patterns in the bulk soil is consistent with reports of other studies (Upchurch and Ritchie, 1983; V05 and Groenwold, 1987; and Le- van et al., 1987). The exact cause or causes of this phenom- enon are not known. Every effort was made to seal the tube from light. It seems unlikely that light within the tubes caused a decrease in root growth as can occur according to Levan et al. (1987). A temperature effect cannot be ruled out but the protruding part of each tube was painted with white paint to minimize excessive heating. Another possibil- ity is that the soil near the surface was disturbed more dur- ing tube installation than the deeper parts of the soil. This may have occurred when the wire brush was moved up and down the hole to clean it prior to tube insertion. The addi- tional soil disturbance may have altered normal rooting pat- terns. Vos and Groenwold (1987) speculated that a disruption to the normal water regime may be a contributing factor to the observed decrease in root growth at the tube surface. The soil disturbance from tube insertion procedures would have altered the normal water regime. The observed effects of rotation/tillage on root growth patterns are quite different between the minirhizotron tubes and the destructive sampling method. As discussed above, that difference is not unexpected in the top 20 cm of the soil profile. However, the minirhizotron tubes showed many more roots cm" in the lower portions of the profile (20-60 113 cm depth) under the CONV management system than under the ARNST system. This difference was evident both years, under both irrigated and nonirrigated conditions (Figures 1 and 3,) and throughout the growing season (Figures 2 and 4). The differences were more pronounced when the crop was not irri- gated in 1985 (compare, for example, Figures la and 1c with lb and 1d), but the opposite was observed in 1986 (Figures 3a and 3c vs 3b and 3d). Despite this observed difference due to irrigation, the differences due to rotation/tillage are striking. The direction of the differences described above was unanticipated. The CONV rotation/tillage treatment had evi- dence of a primary, and possibly a secondary, tillage pan. (See Chapter 2, soil physical measurements.) The ARNST treatment included deep tillage and two years of a deep-root- ed legume prior to the dry bean crop. These plots had lower bulk density and more large pores than the CONV soils, which is an indication that the previously existing tillage pans had been broken up. It was expected that this would result in deeper root growth on the ARNST soils than on the CONV soils. However, the minirhizotron root observations from both years showed much more prolific root growth in the deep- er horizons of the profile on the CONV soils. The destructive sampling data did not show many differences between root length density due .to rotation/til- lage. When analyzed by depth there were almost no differen- ces which were statistically different. 114 The large increase in minirhizotron root observations on the CONV vs the ARNST soils is not substantiated by the de- structive sampling results. This finding raises serious questions about the validity of the minirhizotron method on fine textured soils such as the Charity clay. The proliferation of roots under the CONV system as seen in the minirhizotron tubes must be attributed to an effect of the tubes. Upchurch and Ritchie (1983), McHichael and Taylor (1987), and Vos and Groenwold (1987) have pointed out some of the possible effects of the tube on root growth. The most likely explanation of the effect of the tube on root growth in the current study is that roots on the CONV soils are re- stricted in downward growth by the compacted soil layers ex- ept at the tube. The number of roots which intersect the tube is increased as some of the restricted roots grow later- ally. The roots at the tube-soil interface then proliferate due to less physical restriction since the compacted layers were broken up during tube insertion. There are sufficient voids at the tube surface to allow roots to proliferate. Additionally, there may be an effect due to improved water and aeration conditions along the tube. The individual video frames of the roots support the suggestion that the tube effect is significant in allowing for greater root proliferation. Figure 9 shows minirhizotron observation images representative of each soil management system. The ARNST images show few bundled roots but there are many such images from the CONV tubes. These bundles are 115 Figure 9. Minirhizotron observation images representative of those seen under the two crop/soil management systems: upper: CONV management, compacted layer at 20 cm depth; and lower: ARNST, compacted layer broken up by alfalfa roots and deep tillage. 116 masses of intertwined roots which follow the tube for some length, often 8-10 frames (10—12 cm along the tube) or more. Normal root growth would not include this bundling effect, nor would roots normally grow in a straight path for this distance. The minirhizotron root observations do present a good picture of downward growth over time (Figures 2 and 4). This downward growth was observed on both rotation/tillage treatments. Maertens (1987) states that one of the useful applications of the minirhizotron method is to study rooting depth. The experience of the current study suggests that this may be possible even on fine textured soils. However, the possibility of a tube effect on rooting depth must be examined in order to develop confidence in the minirhizotron method. Since the destructive sampling was done only once, it was not possible to verify the minirhizotron results to insure that there were no tube effects on rooting depth. SUMMARY The minirhizotron root observation system has been found to be useful and accurate in many situations. However, this study showed that the system does not work well on fine textured soils with compacted layers. The data obtained was opposite of that expected and was negatively correlated with that from destructive sampling. The results from the 117 minirhizotron observations showed very few roots in the top 20 cm of soil and a large proliferation of roots under the CONV system vs the ARNST system in the 20-60 cm range. The data suggest that minirhizotrons may be useful for studying rooting depth on fine textured soils. If minirhizotrons are to be useful on this soil, methods will need to be developed to insure that root observations in the tubes are representative of the roots in the bulk soil. This may include changes in installation techniques to obtain better tube-soil contact. Another option might be the use of an inflatable sleeve over the tube which would conform to the shape of the surrounding soil. Such a system, even if it could be perfected for use on fine textured soils, may prove to be too unwieldy for field studies with many tubes. Despite the high cost of destructive sampling, it is still a superior method to the minirhizotron for general root studies on fine textured soils. REFERENCES Beyrouty, C. A. 1987. Characterization of rice roots using a minirhizotron technique. In H. M. Taylor (ed.) Minirhizo- tron observation tubes: methods and applications for measu- ring rhizosphere dynamics. Am. Soc. Agron., Madison, WI. Brown, D. A. and D. R. Upchurch. 1987. Minirhizotrons: A summary of methods and instruments in current use. In H. M. Taylor (ed.) Minirhizotron observation tubes: methods and applications for measuring rhizosphere dynamics. Am. Soc. Agron., Madison, WI. Glenn, D. M. , M. W. Brown, and F. Takeda. 1987. Statistical analysis of root count data from minirhizotrons. In H. M. Taylor (ed.) Minirhizotron observation tubes: methods and applications for measuring rhizosphere dynamics. Am. Soc. Agron., Madison, WI. Levan, M. A., J. W. Ycas, and J. W. Hummel. 1987. Light leak effects on near-surface soybean rooting observed with mini- rhizotrons. In H. M. Taylor (ed.) Minirhizotron observation tubes: methods and applications for measuring rhizosphere dynamics. Am. Soc. Agron., Madison, WI. Maertens, C. 1987. Ways of using endoscopy to determine growth and quality of root systems. In H. M. Taylor (ed.) Minirhizotron observation tubes: methods and applications for measuring rhizosphere dynamics. Am. Soc. Agron., Madison, WI. McMichael, B. L. and H. M. Taylor. 1987. Applications and limitations of rhizotrons and minirhizotrons. In H. M. Taylor (ed.) Minirhizotron observation tubes: methods and applications for measuring rhizosphere dynamics. Am. Soc. Agron., Madison, WI. Merrill, 8. D., E. J. Doering, and G. A. Reichman. 1987. Application of a minirhizotron with flexible, pressurized walls to a study of corn root growth. In H. M. Taylor (ed.) Minirhizotron observation tubes: methods and applications for measuring rhizosphere dynamics. Am. Soc. Agron., Madi- son, WI. Newman, E. I. 1966. A method of estimating the total length of root in a sample. J. Applied Ecol. 3:139-145. 118 119 Smucker, A. J. M., s. L. McBurney, and A. K. Srivastava. 1982. Quantitative separation of roots from compacted soil profiles by the hydropneumatic elutriation system. Agron. J. 74:500-503. , J. C. Ferguson, W. P. DeBruyn, R. L. Belford, and J. T. Ritchie. 1987. Image analysis of video-recorded plant root systems. In H. M. Taylor (ed.) Minirhizotron observa- tion tubes: methods and applications for measuring rhizo- sphere dynamics. Am. Soc. Agron., Madison, WI. Srivastava, A. K., A. J. M. Smucker, and S. L. McBurney. 1982. An improved mechanical root sampler for the measure- ment of compacted soils. Trans. Am. Soc. Agric. Eng. 25:868-871. Tennant, D. 1975. A test of a modified line intersect method of estimating root length. J. Ecol. 63:955-1001. Upchurch, D. R. 1987. Conversion of minirhizotron-root inter- sections to root length density. In H. M. Taylor (ed.) Minirhizotron observation tubes: methods and applications for measuring rhizosphere dynamics. Am. Soc. Agron., Madison, WI. , and J. T. Ritchie. 1983. Root observations using a video recording system in mini-rhizotrons. Agron. J. 75:1009-1015. Vos, J. and J. Groenwold. 1987. The relation between root growth along observation tubes and in bulk soil. In H. M. Taylor (ed.) Minirhizotron observation tubes: methods and applications for measuring rhizosphere dynamics. Am. Soc. Agron., Madison, WI. Waddington, J. 1971. Observation of plant roots in situ. Can. J. Bot. 49:1850-1852. CHAPTER 4 ROOT LENGTH AND WIDTH DETERMINATION BY DIGITAL IMAGE PROCESSING INTRODUCTION Root studies have been hampered by the physical diffi- culties of examining the root system. The processes of root excavation, separation from soil, and analysis of the root system are very laborious. Weaver (1926) and Dittmer (1937) reported detailed studies of root systems. These studies re- quired massive amounts of time and effort. Few researchers since then have been willing to expend a similar amount of energy to study a single root system. The known variability of root systems (Russell, 1977) also frustrates efforts to examine the roots of only a few plants and extrapolate to ge- neral root understandings. Different rhizosphere environ- ments also result in different root responses. Root dry weights have been widely reported due to the relative ease of collecting dry weight data. Dry weights have the limitation of not being well correlated to root activity, especially nutrient and water uptake. Root surface area is a better parameter to measure to gain an understan- ding of total root activity. However, root surface area is almost impossible to measure directly because of the size, 120 121 shape, and number of roots of any single plant. Root length has become the standard parameter for root studies in the past few decades, and has been widely accep— ted in lieu of surface area. Direct measures of root length are almost impossible to obtain. Fortunately, estimation procedures have been developed and shown to be quite accu- rate. Newman (1966) developed a method to estimate root length by counting the number of root intersections with ran- domly placed lines in a tray of well dispersed root sections. Marsh (1971) and Tennant (1975) modified Newman's method. Tennant's modified line intersect method has become the cur- rent standard. In this method, roots are evenly spread in a tray and a square grid is placed under the tray. The number of intersections of roots with the grid lines is then coun- ted. The number of intersections is converted to root length by a simple mathematical formula. While it has allowed for significant advances in knowledge of root function and acti- vity, this method does have several limitations. It is known to overestimate total root length by 5-15‘ depending on the grid size used. The line-intersect method is time consuming and tedious. It is also prone to investigator error due to fatigue as well as differences resulting from different per- sons carrying out the procedure. New methods are needed to improve the accuracy of root length determination, the efficiency with which roots are studied, and to provide investigators with information in addition to root length. Recent developments in computers, 122 specifically image processing, have led to the possibility of greatly enhancing plant root studies through the use of this technology (Smucker et al., 1987). Computer image analysis has several potential benefits. These include: a) decreased labor input required; b) increased accuracy of root data; c) determination of root diameter and surface area; d) determination of branching characteristics such as frequency and angle; e) increased number of samples examined, which would aid in determining statistically significant differ- ences between or among treatments. There are also numerous problems to be addressed in computer image analysis. These include: a) initial cost of image processing equipment; b) preparation and capture of the root image to be analyzed; c) calibration of the system; d) development of appropriate algorithms to gain the desired information from the image. This study summarizes efforts to date in developing a system to analyze roots by using a digital image processor. The system included root sample extraction, washing the roots free of -soil and other material, staining the roots, video taping root images by a computer-driven high resolution video 123 camera, and developing the algorithms to enable the digital image processor to analyze the root video images. The objective of the system development effort was to incorporate current image processing knowledge and technology into a system to more quickly and accurately analyze washed root samples to obtain information on root length. An addi- tional objective was the determination of root width which would allow for calculation of an estimate of total surface area.. Total surface area would be calculated on the assump- tion that roots are round and that root width as determined by the image analysis computer from the two dimensional image is equal to root diameter. A third objective for this system was to obtain information on root branching frequency, bran- ching angle, and other characterization of root branching. MATER I AL 8 AND METHODS Winn Dry beans (Phaseolus vulgaris) were grown in washed sand in the greenhouse or in Charity clay soil in the field. Roots from the greenhouse study were extracted and washed gently in a water bath by hand. Since these plants were grown in sand there was no organic or mineral debris remain— ing after the sand had been washed away. Roots were patted dry and frozen until analyzed. Field roots were extracted at the time of maximum flowering, assumed to represent maximum root growth. The root profile sampler method (Srivastava et al., 1982) was used to extract field roots. Root-soil cubes 7.5 cm on a side were extracted by this method. These cubes were soaked in a 5% solution of sodium hexametaphosphate for 8-16 hours and then washed using a hydropnuematic elutriation chamber (Smucker et al., 1982). The separated roots were stored at 4 degrees C in a 20% methyl alcohol solution until analyzed. These root samples contained varying amounts of organic and mineral debris which was not washed out in the elutriation chamber. Roots were stained with a 5% malachite green solution prior to video taping or hand counting. Root length was determined by Tennant's modified line 124 125 intersect method (Tennant, 1975). A four cm grid was used. This method was considered the standard against which the image processing would be compared. Root_lmase_necordins. Root images were recorded on video tape for future pro- cessing. The process and equipment for this are described below. The roots from one cube of soil (field roots) were con- sidered one sample. Greenhouse roots were divided into sub- samples to achieve the desired amount of roots per tray for counting or video taping. After staining, one sample of roots was placed in a custom made glass tray 43 x 43 cm with 3 cm high glass sides. 750-800 ml of water was added to the tray and the roots were spread evenly throughout the tray. This was accomplished by teasing the roots apart using two small forceps. This procedure is the same for manually counting root—line intersections or video recording the roots for image processing. (For samples containing few roots, a smaller tray, 21.5 x 21.5 cm was used with about 200 ml water. Tray size is discussed more fully in ”RESULTS AND DISCUSSION, Tray Size".) An automated system was developed for video recording the roots. This system consisted of three parts: 1) system control, 2) x y motion (in a horizontal plane), and 3) video acquisition and storage. The system was controlled by an IEM-XT computer with external digital to analog and stepper 126 motor controllers. The x y motion mechanism included two stepper motors that provide power to move a digital video camera (Javelin, MOS) over the root tray. The video acquisi- tion system utilized the video camera and a video recorder (Panasonic, Model AG-6300) to store the images. Images were stored on standard 1/2 inch VHS format video tape. The procedure was as follows. The tray of stained and spread roots were placed on a table backlit with the diffused light of six fluorescent bulbs and positioned to an exact location. The video camera was suspended above the table on an x y scanner. The scan- ner consisted of a frame of two aluminum tables which move in either the x or y direction along four 1/2 inch polished steel shafts fastened to a framework resting on the floor. The backlit table is also attached to this frame. The camera is moved along the ,steel shafts to exact locations by two stepper motors. The camera is moved to one corner of the tray of roots and an image is recorded. The camera is then moved to the adjacent image in the y direction and another image is recorded. The whole tray is ultimately recorded in 64 images (an 8 x 8 grid pattern). Each image represents a 5.4 cm x 5.4 cm square of the tray. The system is fully automated. A computer program writ- ten in Microsoft Basica is utilized to operate the various pieces of equipment needed to produce the video images. A bar code is recorded for each tray of roots prior to any root images. This bar code is entered manually and is used to 127 identify each tray. The bar code is later read by the image processing computer and included with the output for a tray of roots. An audio tone, generated by the computer, is also recorded during each image, including the bar code image. This tone is later used to signal the image processing compu- ter to capture (digitize) the image for processing. Approximately 19 trays of roots, or 1216 images, can be recorded on one two-hour video tape. Six minutes and 4.sec— onds are required to actually video one tray. Total time to prepare (after the roots have been collected and stored) and video one tray is dependent on the time necessary to spread the roots, and was found to vary from 8-25 minutes. Image—Analxsia The video tapes of root images are played back and the images captured in succession and analyzed by a digital image processor (DIP). The hardware used includes the video cas- sette recorder, a time base corrector (Fortel, Model CCDHP), an analog to digital converter (Quasitronix, Model 0-3024), and a digital image processor (Vicom, Model 1800). The Vicom 1800 is a stand alone image processing computer with a Motor- ola 68010 central processing unit. Images are stored in a 512 x 512 pixel array with 16 bits per pixel. The DIP also includes an image digitizer, hard disc storage, computer ter- Iinal, and image display system with monitor. The system is fully automated and is controlled by the DIP. Images from the VHS tapes are first passed through the 128 time base corrector to synchronize the timing with the image digitizer in the DIP. Time base correction is required be- cause video tape is subject to stretching during playback. This stretching can result in inconsistencies in timing be- tween frames, which, if not corrected, can lead to improperly digitized images. The image is then digitized and stored by the digital image processing computer for processing. The software is divided into five sections. These are control, preproCessing and thresholding, skeleton width en- coding, debris extraction, and measurement. The control section of the program controls the video tape, digitizes each image, and deciphers the bar code which is used for identification. Preprocessing involves enhancing the image in order to accentuate the roots and eliminate as much noise as possible. Thresholds are then chosen to iden- tify objects in the image. Thresholding results in a binary image ready for further processing. Binary images are thinned to one pixel in the skeleton width encoding operation. The remaining centerline can be measured for length. The thinning process also allows for width determination by counting the pixels that are thinned away. A debris extraction routine is available for root sam- ples which contain any non-root objects. Most samples will fall into this category unless they have been grown in a med- ium with no organic matter or it is possible to remove any debris. Object shape is used to distinguish between roots 129 and debris. Any object with a 1ength:width ratio less than approximately 3:1 is considered to be debris. Following the above steps, length and width measurements can be made. Length per width class is recorded in pixels. There are five width classes, 1-2, 3-4, 5-6, 7-8, and 9-10 pixels. RESULTS AND DI SCUSS I ON Iminststem The automated video recording system worked well for imaging roots. Approximately 200,000 images have been recor- ded with no significant mechanical or operational problems. Time per tray varied considerably but the video recording sy- stem did not require as much labor as the line intersect counting method. The time required to spread the roots was the same for the two systems. With the automated video sy- stem, one tray could be prepared while another was being re- corded, but with the line intersect method additional time was required to actually count the intersections. It is estimated that the video procedure requires 5-10 minutes per tray less than the line intersect method. Wine The first step in image processing was to test the sys- tem, including the hardware, synchronization, image capture, and analysis algorithm. Approximately 8,750 images of field roots and 5,750 from greenhouse plants (washed sand medium) images were analyzed by the DIP. This represents a total of about 450 trays. All of these trays were also hand counted using the line intersect method. Additionally, several trays 130 131 of different gauge wire and several trays of string were used for initial testing and calibration. W It was necessary to calibrate the DIP in order to con- vert the output in pixels to root length. This was done using string of a known length. Several trays of string were used, each with a different total length of string. Differ- ent thicknesses of string were used and pieces were randomly cut into lengths from 0.5 cm to 3 cm to simulate roots. String thicknesses varied from 0.25 to 1.0 mm and length per tray varied from 2.25 Am to 27 m. This wide range of string length was used to simulate the wide variability in root length among root samples. The string trays were video taped with the same procedure as actual roots and images were pro- cessed by the DIP both with and without the debris extraction algorithm. Each tray of strings was video recorded, mixed and re-spread, and video taped again. In this way the repeatability of the DIP analysis could be tested since there were two trays of exactly the same total string length. Data are presented in Table 1. string length (cm) was per tray. The ”Total Pixels" columns represent the total number of pixels found by the DIP. The values in the "Pixels lcm"columns were calculated by dividing the total pixels by the total string length. These are the figures applicable to actual root image analysis. 9 In Table 1, note that the pixels/cm is a function of 132 Table 1. Calibration of digital image processor using trays of strings. Each line represents the same tray of strings analyzed with or without the debris extraction algorithm engaged. Each length was analyzed twice, with the strings mixed and redistributed between analyses. Without Debris Extraction With Debris Extraction Length Algorithm Algorithm of String Total Pixels Total Pixels (cm) Pixels /cm Pixels /cm 225 26763 118.95 23111 102.72 225 26457 117.59 22957 102.03 450 52666 117.04 45995 102.21 450 51724 114.94 42935 95.41 675 80604 119.41 68281 101.16 675 80051 118.59 68577 101.60 900 88855 98.70 80312 89.24 900 90210 100.20 78840 87.60 1800 174965 97.20 152337 84.63 1800 175263 97.40 151310 84.06 2700 257561 95.40 218738 81.01 2700 256600 95.00 218765 81.02 133 density of the string. As more string was added to the im- ages, the DIP results showed fewer pixels per actual length. This is thought to be due to the effect of overlapping pieces of string. Obviously, as more string is added to the tray, there will be more overlapping. ‘The DIP identifies only one length of string where there may actually be two or more. Thus the pixels/cm figure used to convert the DIP output from root images will vary with density of the sample, or the accuracy will vary with density of the roots. This problem has not been solved to date. Secondly, note that the DIP results are similar for two trays with the same total length of roots. The largest dis- crepancy is in the 450 cm string trays with debris extrac- tion. There are approximately 7% more pixels reported for one tray than the other. Reasons for any discrepancy include differences in overlapping of strings, noise on the video tape, differences in the analog to digital conversion at the time the image is fed into the DIP, and general electronic noise due to current fluctuation. Finally, a comparison of the results between analysis with or without debris extraction shows large differences. The difference is approximately 14‘, and is consistent across the range of string lengths/tray used. Apparently the DIP identifies some "debris” in these trays of string. Since the debris extraction algorithm uses a 1ength:width ratio to dis- tinguish between debris and roots there must be some non- string characters appearing in the image. The most likely 134 cause of the "debris" detected is video noise, which is elec- tronic noise recorded on the video tape or picked up during the analog to digital conversion of the image. Another pos- sible source of the "debris” is surface roughness of the strings which is enhanced by the processing algorithms and appears as small branches. This roughness characteristic may be a problem on roots as well. Some of the root rough- ness may be related to root hairs or branching of very fine roots and thus should be counted as root length. However, the current resolution is not sufficiently high to detect root hairs so it would be more accurate to discard any length due to root surface roughness if that were possible. At present that distinction cannot be made. Additional calibration was carried out using several images with different total lengths and gauges of wire. These analyses were used to test the width determination part of the algorithm. As described previously, the DIP ana- lysis is carried out on a pixel basis and it was necessary to determine the relationship of a pixel width to actual width. The current algorithm was designed to divide the roots into 5 width categories, 1-2, 3-4, 5-6, 7-8, and 9-10 pixels wide. Two sets of tests were conducted. In the first, images used were composed of pieces of wire cut into 1 or 2 cm lengths. Six diameters of wire were used, including 0.1, 0.2, 0.25, 0.6, 1.1, and 1.6 mm. 38 images were recorded using various combinations of wires and various orientations within the images (e.g. x oriented, y oriented, crossed, 135 diagonal, random, systematic, etc.). Data is presented in Table 2. There are several images (rows of data) for which no pixels were found by the DIP. One probable reason is that the wires in these images are too fine to have sufficient contrast in the image for the DIP to find the objects in the image. This could possibly be corrected by setting a differ- ent threshold. However, a lower threshold would allow addi- tional ”noise” to be picked up along with lower contrast ob- jects in the image. The choice of threshold level was made after careful testing. The threshold is reset (automatical- ly) for each image and this may explain why the wires are picked up in some images while in other images the same size and number of wires are not found. This inconsistency raises questions about the reliability of the current system in ac- curately analyzing fine roots. The data in Table 2 show that the separation of wires into different width classes was not conclusive in this test. This is true at all width classes, but especially at the 1-2 and 3-4 pixel width classes. The results of a second series of tests for width class determination are presented in Table 3. Each image contained eight 1 cm pieces of wire. The wires in each image were the same width. Fourteen different widths of wire, ranging from 0.1 to 1.4 mm, were tested. The images were video taped, the wires within each image randomly rearranged, and the image video taped again. This was done three times. The results 136 Table 2. Width and length calibration data from digital im- age processor. Each row represents one image. Wire piec- es were 1 cm long except for the 1.6 mm width, which was 2 cm long. Total Wire Width Classes (pixels) Total Pixels Length Width 1-2 3-4 5-6 7-8 9-10 Length /cm (cm) (mm) ------- length (pixels) -------- (pixels) 1 0.1 0 0 0 0 O 0 0 1 0.1 0 0 0 0 0 0 0 1 0.1 0 0 0 0 0 0 0 3 0.1 160 122 0 0 0 281 93 6 0.1 608 210 0 0 0 818 136 6 0.1 0 0 0 0 0 0 0 8 0.1 523 305 0 0 0 828 104 8 0.1 478 327 0 0 0 805 101 1 0.2 83 26 0 0 0 109 109 1 0.2 0 0 0 0 0 0 0 1 0.2 0 0 0 0 0 0 0 8 0.2 340 453 0 0 0 793 99 1 0.25 58 40 0 0 0 98 98 1 0.25 210 90 0 0 0 300 300 1 0.25 30 84 0 0 0 114 114 8 0.25 223 559 0 0 0 782 98 1 0.6 21 82 18 0 0 121 121 1 0.6 7 97 6 0 0 109 109 1 0.6 0 84 13 0 0 97 97 8 0.6 17 725 51 0 0 792 99 1 1.1 9 9 7 52 37 114 114 1 1.1 17 13 9 69 18 126 126 1 1.1 12 6 5 35 53 111 111 8 1.1 23 48 32 550 135 788 99 2 1.6 50 8 5 7 4 73 37 2 1.6 21 16 5 11 5 57 29 2 1.6 7 12 3 11 5 38 19 8 1.6 25 27 25 32 26 134 17 7 * 215 181 37 41 58 531 76 7 * 102 291 14 80 13 500 71 7 * 79 279 65 75 31 529 76 7 * 106 306 45 84 20 561 80 7 * 104 289 24 9 5 430 61 7 * 133 242 22 12 7 416 59 7 * 134 245 5 6 5 395 56 7 * 119 213 40 12 7 392 56 24 ** 593 1642 91 1 0 2327 97 24 ** 536 1632 176 0 0 2344 98 ,. f 1 cm length of each of the 5 smallest width classes, 2 cm of 1.6 mm width. ** 8 cm length of each 0.2, 0.25, and 0.6 mm width. Table 3. Results of width class calibration. 137 presents one image analyzed three times. are the average of the three analyses. tained eight 1 cm pieces of wire of the given width. Breaks between rows of data indicate break between width classes as determined by digital image processor. Each row re- Figures reported Each image con- Wire Total Pixels Width Width Classes (pixels) Length /cm (mm) 1-2 3-4 5-6 7-8 9-10 (pixels) --------- length (pixels) ----------- 0.10 278.5 528.6 0.0 0.0 0.0 807.1 101 0.20 234.4 571.3 0.0 0.0 0.0 805.7 101 0.25 246.2 582.4 0.3 0.0 0.0 828.9 104 0.35 135.1 667.8 2.2 0.0 0.0 805.1 101 0.45 36.6 749.9 3.5 0.0 0.0 790.0 99 0.50 32.6 780.0 5.0 0.0 0.0 817.5 102 0.60 25.9 746.4 26.6 0.0 0.0 798.8 100 0.75 30.4 82.6 702.3 27.6 0.3 843.2 105 0.85 154.3 59.9 371.3 365.7 0.4 951.5 119 0.90 164.0 77.9 246.8 471.1 0.0 959.8 120 1.05 342.7 74.5 45.3 599.6 107.9 1170.0 146 1.10 197.2 89.9 51.7 466.5 268.2 1073.5 134 1.25 231.9 85.1 50.9 225.3 508.7 1101.9 138 1.40 181.9 100.6 60.2 66.6 54.3 463.7 58 138 presented in Table 3 are the averages of the three images for each width class. The repeatability of the DIP analysis was found to be quite good as the variability among the three analyses was small. The results of this test were more conclusive than the first width class calibration. While the distinctions be- tween the width classes are not precise, the general trends are defined. The division between the two smallest classes, 1-2 and 3-4 pixels, is ambiguous but wires from 0.1 to 0.35 mm are split between the 1-2 and 3-4 pixels width classes, .while wires 0.45 to 0.6 mm appear primarily in the 3-4 pixel width class. The 5-6'pixel wide group includes wires between 0.75 and 0.9 mm. Wires 1.05 and 1.1 mm in diameter are pre- dominantly in the 7-8 mm class. 1.25 mm wires are in the widest class, 9-10 mm, and wires 1.4 mm in diameter were too large to be included in the given width classes. The fact that the distinctions are not precise between or within width classes is due to electronic noise, bounce or shadows which may appear as separate images, or some thresholding differences. The bounce problem is most evident in the wider groups. Note that the total pixels (last col- umn, Table 3) reported for wires in each width class is rela- tively constant until wire width exceeds 0.75 mm diameter. The total pixels then increases appreciably. There is a more pronounced bounce effect with larger wires since the shadows are larger. These bounces are above the threshold and are counted as objects. 139 In summary, current width class distinction is opera- tional in a general way but is not yet sufficiently precise for broad application. The problems appear to be due to im- age recording, transfer, and capture rather than the analysis algorithm. Tables 2 and 3 provide additional data on pixel length calibration. In Table 2 the last column is pixel/cm deter- mined by dividing total pixels by the known length of wire in the image. values range from 33.5 to 300 pixels/cm but most of them fall between 60 and 110. As discussed relative to Table l, pixels/cm is a function of density. In the images with 8 cm total length the pixels/cm is constant at around 100. Images with less than 8 cm are much more varied. The images in Table 3 all contained 8 cm of wire. Again, the last column is pixels per centimeter and is relatively con- stant at around 100 until wire width increases to the point that the bounce problem is more significant. Time. One of the objectives for the development of image pro- cessing for use in root studies is to increase the speed by which root samples can be analyzed. Time for video recording the roots was discussed previously and, while still requiring much time, it is a slight improvement over the conventional line intersect method. DIP processing time, however, adds significantly to the total time necessary for data collec- tion. The current algorithm used to analyze clean root 140 samples (no debris extraction) requires approximately 1.2 minutes per image. Since each sample of roots is spread out in a tray which is recorded in 64 video images, 75 minutes are required to analyze one root sample. The system is auto- mated so that all of the images on a two hour video tape can be analyzed without an operator present. However, the total number of samples analyzed in one 24 hour period is limited by processor speed to a maximum of 18 or 19. The processing of root samples with debris is consider- ably more time consuming. The operation of the debris ex- traction routine approximately triples the time required to process an image which means that about 6 or 7 samples can be processed in one 24 hour time period. Since nearly all field root and many greenhouse root samples contain debris, the ex- traction routine is a critical part of the analysis. A com- parison of processing time required per root sample for dif- ferent root applications is presented in Figure 1. Processing speed must be increased if the system is to be useful in analyzing large numbers of root samples. Op- tions for higher speed include an improved algorithm or a faster computer. Another option would be subsampling, or analyzing only a fraction of the total images. Tests were run to examine the possibility of subsampling and to deter- mine what fraction of the images in a tray would need to be analyzed to achieve acceptable results. Figure 2 summarizes test results. Figure 2 was developed using output from DIP analysis of a tray of roots. Total pixels of length for each TIME/ROOT SAMPLE (hr) 141 5A) 4J5 4J3 215 3J3 2J5 2J1 1.5 1.0 (LS 013 64- :52 0"”) 64 32 DEBRIS EXTRACTION NO oceans EXTRACTION Figure 1. Time required for image processing as affected by the use of the debris extraction algorithm and the number of images processed. PIXELS (thousands) 142 3°04 IAIAI I I I l I I I I I I fi' I I I 1 I I I I I I I I I fI. 4 . 1 I 275- - l . J I) 1 a J 250- - 1 I 2253 1 . 200: J ‘ i 175- . ..r..».r....r...er.. -r.._.r... O 8 16 24’ 32 4O 48 56 64 NUMBER OF IMAGES Figure 2. Results of tests to determine the fraction of images from a tray of roots which must be processed in order to obtain acceptable estimates of root length. Horizontal lines represent :52 of the actual (DIP) analysis total. Each data point represents the pixel sum (y) of the randomly selected images (x) mul- tiplied by (GA/number of images). 143 of 64 images in a tray were compiled. A given numeral of the 64 was randomly selected and multiplied by the fraction 64/1 to yield an estimate of the total pixels in the tray if each image contained the same number of pixels as the selected im- age. This same procedure was carried out for 2, 3, 4, etc., to 64 images. (For example, the sum of pixel lengths of 23 randomly selected images was multiplied by the fraction 64/23.) The results of several runs of this procedure are plotted in Figure 2. The horizontal lines represent plus and minus 5‘ of the total number of pixels as determined by sum- ming the pixels lengths from each 'of the 64 images. This procedure was carried out on several trays of roots and the results were very similar to those presented in Figure 2. These results led to the conclusion that analyzing one half of the images 15‘ a tray and multiplying their sum by two would yield total length estimates which would be acceptable since they are almost always within 5‘ of the total. Reduc- tions below 32 in the number of images analyzed would de- crease the likelihood of obtaining results within the 15\ range selected as acceptable error. Analyzing 32 of the im- ages cuts the time in half which is obviously a major step in increasing processing speed without seriously decreasing accuracy of the results. 1mm Most of the testing and root analysis carried out has been done using the 43.x 43 cm trays for root video taping. 144 However, since some root samples contain very few roots, a smaller tray was developed for use. The smaller tray was used in order to concentrate the roots sufficiently to elimi- nate analysis problems due to limited density. When appro- priate, the small tray also has the advantage of saving time both in video recording (1 min., 47 sec. to video the small tray vs 6 min., 4 sec. for the large tray) and in DIP analy- sis time (16 images in the small tray vs 64 images in the large tray). Comparison of results of analysis using the two tray sizes indicated that the two sizes could be used inter- changeably with only a slight difference in results. The large tray resulted in slightly more pixel length recorded, probably due to the previously described noise problem. We. Roots of dry beans (Phaseolus vulgaris) grown in washed sand in a greenhouse were analyzed by the DIP. The debris extraction routine was not used in processing these images since there was no debris in the root samples. Roots were cut into 1-3 cm pieces and evenly spread and separated prior to counting by the line intersect method or video taping. A comparison of the line intersect and DIP methods is presented in Figure 3. The pixel length conversion factor used was 100 pixels = 1 cm root length (determined from calibration tes- ting as reported in Table 1). The results show that the DIP overestimates root length compared to the line intersect me- thod. The overestimation on the 169 trays of roots compared 145 IMAGE PROCESSING ROOT LENGTH (m) DEBRIS-FREE ROOTS T O I I I I I I iv r I r* r* TV 10 20 30 4o 50 50 LINE INTERSECT ROOT LENGTH (m) Figure 3. Comparison of root length determinations by image processing and line intersect methods. Each point represents one tray of roots, or 64 images. Diagonal line represents 1:1 correlation; top line is regression line. 146 averaged 23%, with a range of -6 to 125%. However, the over- estimation is nearly linear in the range of lengths examined, as shown by the regression line (top line) and the 1:1 ratio line in Figure 3. The overestimation is due to electronic noise, video tape imperfections, bounce, and imperfections in the image transfer from digital to analog and back to digital again. These causes are discussed in other sections of this report. W The DIP was used to analyze a set of field roots and the results compared to those obtained by the conventional line intersect method. Field roots of dry beans (Phaseolus vulgaris) were extracted and processed as described in Chap- ter 2. A summary of results from the analysis is presented in Table 4. In this table root length densities (RLD) are presented for the DIP analysis and compared to the line in- tersect method on a percentage basis. Four treatment combi- nations are presented: ARNST is alfalfa rotation, no secon- dary tillage; CONV is conventional rotation/tillage; IR is irrigated and N1 is nonirrigated. The first column of num- bers, 1-18, identifies the location of the 7.5 cm cube of soil in the extracted soil profile. Cubes 1-3 are from the center of the plant away from the plant and from 0-7.5 cm depth. 4-6 represent the next three cubes, from 7.5-15.0 cm depth, and so on down the profile. Each figure in Table 4 is the average of four 147 Table 4. Comparison of root length density (RLD, cm cm") determination by the digital image processor (DIP) and the line intersect (LIN) methods. Root length densities re- ported are calculated from the DIP length determinations. \ LIN is the DIP RLD divided by the line intersect RLD. The 18 cases represent the RLDs from 18 cubes of soil, 7.5 cm on a side, from a soil-root profile 7.5cm thick x 22.5 cm wide x 45 cm deep. 8011 ARNST ARNST CONV CONV Cube IR A NI 8 IR % NI 8 8 RLD LIN RLD LIN RLD LIN RLD LIN l 3.79 112 , 5.60 145 4. 74 115 4. 89 143 2 4.10 133 4.71 127 4. 47 126 5.68 168 3 4.99 146 3.72 128 L 48 129 5. 73 162 4 6.25 176 3.88 151 6.03 280 3.57 154 5 a 4.11 164 3.55 132 6.58 287 4.17 180 6 5.64 169 4.58 153 6.26 260 4.49 206 7 3.27 145 4.34 162 4. 39 222 3. 41 157 8 2.91 141 3.62 150 3. 72 184 3. 66 173 9 2.47 121 2.73 131 2. 80 150 3. 51 184 10 1.57 104 1.15 76 2.28 136 2.53 126 11 1.54 107 1.81 122 2.79 134 2. 66 138 12 1.43 102 1.25 115 2.27 134 1. 88 137 13 1.08 114 0.83 116 1.93 132 1.67 132 14 0.87 101 ‘ 1.30 129 2.54 169 2.04 123 15 1.53 118 0.89 119 1.58 128 1.35 113 16 0.84 106 1.24 122 1.82 141 1. 77 158 17 1.10 112 1.34 119 1.41 127 1. 92 132 18 1.10 117 0.93 125 1.39 143 1. 61 163 148 replications. Thus each RLD as determined by the DIP in- cludes the results from 256 images (4 replications or trays of roots x 64 images per tray). The results show that the DIP overestimates the root length by a large margin. The range is 76‘ to 280%, with only one point below 100%. Avera- ges for the rotation/tillage treatments at each depth are presented in Table 5, along with the overall averages. The overall average is 144‘ of the line intersect method. Con— version factor for the pixel lengths determined by the DIP was 100 pixels/cm. In addition to the general overestimation problem, there is a problem with the debris extraction routine. The field root samples contain varying amounts of debris, primarily or— ganic debris from previous years' crops. The CONV treatments contain more debris than. the ARNST treatments because corn was grown the previous year on the CONV and alfalfa on the ARNST plots. The cornstalks were plowed down following grain harvest. This is indicated in the DIP RLD data in soil cubes 4-9, which represent the 7.5—22.5 cm depths of soil. It is in this depth range that the majority of the decomposing corn stalks would be expected. The CONV treatments at this depth show the greatest overestimation of RLD by the DIP. These root samples had large amounts of organic debris but did not have- appreciably more roots than the ARNST samples at the same depths (line intersect data). Figure 4 shows the over— estimation of RLD by the DIP for all depths. The points in the higher RLD ranges primarily represent the samples from 149 Table 5. Summary of digital image processing (DIP) root length density (RLD) results compared to line intersect results. Figures represent DIP RLDs/line intersect RLDs x 100. Roots are field grown dry beans, ARNST is alfalfa rotation, no secondary tillage and CONV is conventional management. DEPTH ARNST CONV OVERALL (cm) (ave) (ave) (ave) ------- t of line intersect ------- 0-7.5 131.7 140.7 136 2 7.5-15.0 157 3 227.9 192.6 15.0-22.5 141.7 178.2 160.0 top 22.5 cm 143.6 182.3 162.9 22.5-30.0 104.3 134.2 119.2 30.0-37.5 116.2 132.8 124.5 37.5-45.0 116.8 144.0 130.4 22.5—45.0 cm 112.4 137.0 124.7 Overall Averages 0-45.0 cm 128.0 160.0 143.8 B 'T I I'Trl'ljr' J 4 r."— - ' ‘ I J :, . . §6J '- m'lm d ES- In... .. v d - . q 9 34'- m“ T'.-..l "' m d . f I 53.. 1 .. >’ ‘ .‘Is. . a O. - '. ' ‘ 511 ARNSTMANAGEMENTfi I . O rIrIrr'I‘lrrrr‘ O 1 2 3 4 5 6 7 8 8 . A7-I '? . 5.. 55. v .9... Q I my 5 d <( 3% 0. .‘P 51- CONVMANAGEMENT 0 fifirlrrfrr rl' f ‘31 on o 1 2 :5 4 s 6 LINE—INTERSECT RLD (cm car-3) Figure 4. Comparison of root length density as deter- mined by the digital image processor and the line intersect method for dry bean roots under two crop and soil management systems. Each point is the average of 4 replications. 151 the top 22 cm of soil. Also, those points further from the 1:1 correlation line represent samples with greater amounts of debris. SUMMARY Numerous tests have been run on the root imaging and digital image processing systems. These tests have provided a comprehensive look at the system and indicate the positive aspects as well as various problems to be dealt with prior to using the system to collect useful root data. The imaging system works well. The hardware and software are well integrated and automated. Given the technology used, the system produces high quality root images. The digital image processing hardware is also well inte- grated and functions well. Occasional breakdowns in proces- sing occurred for unknown reasons. Electrical current fluc- tuations are suspected as the primary cause. Many of the problems encountered seem to be related to image resolution. At the current resolution many of the roots are only one or two pixels wide. Much of the "noise" in the images is one, two or three pixels wide and thus is picked up by the DIP and included as roots. A higher resolu- tion would allow for elimination of the video noise by thresholding or other image enhancement techniques. The pro- blem of debris extraction is critical to the success of any 152 root analysis procedure. The current resolution and algo- rithm do not adequately separate the debris from the roots in the images. If resolution were increased, the debris problem could more easily be handled since the distinctions between roots and debris would be magnified. However, this problem will not easily be solved since some of the debris is organic matter which is almost exactly like roots in appearance. Ad- ditional developments in software may enhance the possibili- ties for successfully distinguishing between roots and de- bris. Finally, some samples with excessive amounts of debris may need to be separated into two or more subsamples to de- crease root/debris crossover which increases the difficulty of accurately analyzing the sample. Many of the problems encountered are related to the im- age storage mechanism, video tape. There are several related problems which suggest this technology may be inherently flawed for this application. The electronic and magnetic tape "noise” is a primary source of error. The images must be converted from digital (video camera) to analog (video tape) and back to digital (for processing by the DIP). The conversion process is not precise and introduces potentially significant error. In the long run it will be advantageous to develop the system around a different image storage tech- nology, for example, laser disc. Increased resolution has been suggested as offering sig- nificant improvement in the system. However, increased resoe lution would introduce yet another problem. The time 153 required to video record the images would be increased. Even more importantly, the time required for processing the images with the current digital image processor would be increased from its already prohibitive level. A doubling or quadru- pling of the number of images would require an equal increase in processing time. Image processing has much potential for greatly enhan- cing root studies. However, there are numerous problems to be solved before the system is widely applicable to root stu- dy. The system development and testing reported here are an important step in the direction of root study by digital im- age processing. REFERENCES Dittmer, H. J. 1937. A quantitative study of the roots and root hairs of a winter rye plant (Secale cereals). Am. J. Bot. 24:417-420. Marsh, B. B. 1971. Measurement of length in random arrangements of lines. J. of Appl. Ecol. 8:256-257. Newman, E. I. 1966. A method of estimating the total length of root in a sample. J. Applied Ecol. 3:139-145. Russell, R. S. 1977. Plant root systems:their function and in- teraction with the soils. McGraw Hills, London. p.169-190. Smucker, A. J. M., S. L. McBurney, and A. K. Srivastava. 1982. Quantitative separation of roots from compacted soil profiles by the hydropneumatic elutriation system. Agron. J. 74:500- 503. Smucker, A. J. M., J. C. Ferguson, W. P. DeBruyn, R. K. Bel- ford, and J. T. Ritchie. 1987. Image analysis of video-recor- ded plant root systems. p. 67—80. In H. M. Taylor, Minirhizo- tron observation tubes: Methods and applications for measur- ing rhizosphere dynamics. Am. Soc. Agron. Spec. Publ. No. 50. Madison, Wisconsin. Srivastava, A. K., A. J. M. Smucker, and S. L. McBurney. 1982. An improved mechanical root sampler for the measurement of compacted soils. Trans. Am. Soc. Agric. Eng. 25:868-871. Tennant, D. 1975. A test of a modified line intersect method of estimating root length. J. Ecol. 63:955-1001. Weaver, J. E. 1926. Root development of field crops. McGraw- Hill, New York. 154 CHAPTER 5 EFFECTS OF FLOOD OR DROUGHT STRESS ON DRY BEAN ROOT AND SHOOT GROWTH INTRODUCTION Field grown crops are often subject to stress from too little or too much moisture. The most common cause of stress is insufficient or excessive precipitation. The effects of less than optimum precipitation may be compounded by soil conditions, as will be discussed later. Temporary or perma- nent plant injury from flooding or drought can result in costly damage to the crop. Schwartz (1980) has pointed out that too little soil water can damage plants due to the unavailability of water for plant roots, the accumulation of toxic ions, stomatal closure which results in restricted CO: uptake, and temporary or permanent plant wilt. Laude (1971) noted that stomatal closure results in reduced photosynthetic activity per unit leaf area and an overall decrease in leaf area. A reduction in plant size is typical of plants growing without sufficient moisture. Plants subjected to a mild drought may recover quickly and production may not be greatly affected. Plants have been found to grow more rapidly for a short time following 155 156 rewatering than those which experienced no drought (Laude, 1971). However, plants subjected to a more prolonged drought may suffer irreversible damage. Plants may be damaged to the extent that they are incapable of resuming growth when the drought ends. Flooding can be equally detrimental to plant growth. Schwartz (1980) noted that flooding may leach nutrients es- sential to plant growth, reduce 02 content, induce plant chlorosis, and lead to accumulation of toxic byproducts from anaerobic metabolism. Jackson and Drew (1984) suggest that the primary effect of flooding is asphyxiation of the plant due to decreased gaseous diffusion. Root growth is generally retarded and root survival time varies greatly with species (Glinski and Stepniewski, 1985). Jackson and Drew (1984) also point out that leaf growth is extremely sensitive to flooding and root anoxia. Soil conditions, for example soil compaction, may exa- cerbate the problems related to too much or too little water. A compacted soil or compacted layers in a soil may increase the effects of either problem. A compacted horizon can pre- vent adequate drainage following rainfall or irrigation. The resulting waterlogging will quickly lead to an anaerobic rhi— zosphere environment (Cannell and Jackson, 1981). Anaerobic conditions have been reported as the cause of decreased plant growth rate (Smucker and Erickson, 1987), cessation of root growth (Letey et al., 1962; Glinski and Stepniewski, 1985), and death of some root tips (Huck, 1970). 157 Root systems have been found to grow deeper in a mois- ture shortage situation (Klepper et al., 1973; Cortes and Sinclair, 1986; Huck, 1986; and Hoogenboom et al., 1987). However, downward root growth can be slowed due to compaction related mechanical impedance (Bertrand and Kohnke, 1957; Rag- havan, 1977; and Bennie and Botha, 1986). When this occurs the root system may be prevented from tapping into the avail- able moisture deeper in the soil profile. Shoot:root ratios are usually altered by environmental stress. In general, stress on the root system such as flood- ing or drought leads to a decreased shoot:root ratio as the plant partitions more of its photoassimilate to the root sy- stem (Shank, 1945; Taylor, 1981; Hoogenboom et al., 1987). As this occurs it results in decreased shoot growth which can lead to decreased yield, especially when the stress occurs during a critical growth stage such as reproduction (Huck et al., 1986). The objectives of this study were: a) to examine the effects of drought or flood stress on growth parameters of dry beans (Phaseolus vulgaris); and b) to gain some indica- tion of expected dry bean response to conditions of less than optimum moisture which may occur when the crop is grown on a soil with compaction problems. The study was conducted in a greenhouse in order to control environmental conditions. Both shoot and root responses were monitored. The response and recovery of the plant to the stress was measured at three times following the stress application. MATERIALS AND METHODS W A greenhouse study of the effects of drought and flood on dry bean shoot and root parameters was conducted in the spring of 1986. Dry beans (cultivar C-20) were planted in PVC tubes sealed at the bottom with the exception of a drain hole approximately 1 cm in diameter. The tubes were 70 cm long and 7.6 cm inside diameter. Planting medium was washed silica sand (0.3—0.6 mm). Four seeds were planted in each pot. Soon after emergence each pot was thinned to two uni- form seedlings. Plants were irrigated with half strength Hoaglund's nutrient solution. Drip irrigation was carried out as needed to maintain good moisture throughout the sand profile. Irrigation varied from 4 to 6 times per day for 15 or 30 minutes per irrigation. Differences in irrigation were due to weather conditions and plant size. The first seedlings emerged 4 days after planting and emergence was approximately 90% by the following day. Fluor- escent lighting was used to supplement available sunlight and to extend the photoperiod to 16 hrs day-1. This photoperiod was maintained throughout the study. Stress treatments were applied 42 days after planting. Plants were at maximum flowering at this time. One third of 158 159 the plants received drought stress as follows. A vacuum pump was used to apply suction to the bottom of the tubes. Suc- tion was applied for 3-4 minutes per pot to draw off free water. The drought stress was continued for 7 days with only enough solution added to maintain plant viability. The flood treatment was also applied 42 days after plan- ting. Irrigation solution was pumped into the tubes from the bottom via the drainage port until it overflowed the top. The drainage outlet at the bottom of the tubes was then closed off and flooding maintained for 7 days. The remaining one third of the plants, designated the control, were irriga- ted normally. Ambient weather prior to and during stress treatments was cloudy and cool. Without sunlight the stress effects were less pronounced in a given period of time. The 7 day stress period was chosen in order to subject the plant sy- stems to a severe stress. Following the stress period the drainage ports of the flooded tubes were opened and all plants received normal irrigation. The experimental design was a randomized complete block with four replications. Measurements Whole plants were harvested on three dates. Shoots were cut off at the soil surface. Roots and sand were pushed out of the tubes using air pressure. The sand-root ‘ .._.I .—- "1'“ 160 cores were cut into three sections by depth (0-20 cm, 20-40 cm, and 40—70 cm). The sand was washed away from the roots by gentle motion in a water tub. Roots were patted dry and frozen for future analysis. The first harvest was immediately post-stress or 49 days after planting. Measurements included stem and leaf fresh and dry weights, and leaf area determined by a Licor Leaf Area Meter. Roots were harvested as outlined above. The second harvest’was carried out 12 days after the first or 61 days after planting and measurements were the same. The third harvest was at maturity (90 days after plan- ting, 41 days after termination of stress). Parameters mea- sured included the roots, fresh and dry weights of stems, pods, and beans, and number of beans and pods per plant. Root length was determined using Tennant's line inter- sect method (Tennant, 1975) or by video image analysis as described in Chapter 4. Roots were stained with 5% malachite green prior to counting or video taping. A four cm square grid was used for the line intersect method. RESULTS AND DI SCUSSION Shoot parameters, including both fresh and dry weights, for the three harvest dates are presented in Table 1. The effect of the imposed stress was evident soon after the plants were subjected to either flood or drought. The drought stressed plants showed a greater decrease in growth than did the flooded plants relative to the control plants. Dry weight of the shoots subjected to drought was signifi- cantly lower than the control shoots. The flooded plants had shoot dry weights 15% lower than the control plants but the difference was not significant at the 95% level. Decreased growth was seen in both the stems and leaves. The effects of the environmental stresses were evident throughout the remainder of the study. At 12 days post stress the dry weights of both the flood and drought stressed plant shoots were significantly lower than the control. The shoots subjected to drought weighed 23% less than the flooded plant shoots at this harvest. However, at final harvest the drought stressed plants had surpassed the flooded plants in shoot fresh and dry weights. The recovery of growth of dry beans subjected to drought was greater than the recovery of the flooded beans when measured over a time period of several weeks. At final harvest the shoot dry weights of both 161 162 Ame. .ncv v.v~ o.mm m.m~ o.a m.hm >.ma H.o~ m.vm mo.omq m.oH ~.Hv >.bm n.5h a.va m.~m w.mn o.mm uzmsomn. o.m h.b~ m.hm ~.HHH b.owa o.oq o.Hm «.mm coofim v.am a.as >.vo o.mbH m.pm~ o.mm m.am «.mNH douucoo uuuuuuuuuuuu Am. acmfiom snoum uuunuuuusunu Ana; :2: «.m o.mH H.~ H.b m.o m.~ v.m >.m mo.ama «.5 ¢.m~ o.m m.~H o.mH o.v v.q ~.¢ usmsomo o.m o.o~ o.w m.oH m.v~ H.m «.5 «.ma cooam p.m H.vv ~.oH ¢.>~ ~.om N.» o.w o.m~ .Houucoo nnnnnnnnnnnn Amy unmaom amp nunnnnunuuuu noomw mmoum acmHm no>moa macaw acoam no>moa nmoum usuam mmouum umom name Hv mmouum unom mama NH mmouum unom comma .mnomun uzmsomc new ooon ha copoomwm no nnouan mausoHHOM noumc come» so muoaoamuma moose coon hum .H canoe 163 flooded and drought stressed plants were significantly lower than the dry weight of the control plants. Leaf area is reported in Table 2. The effects of the stress included a statistically significant decrease in total leaf area per plant and area per leaf. The total number of leaves per plant also decreased due to the imposed stress. The plants subjected to drought suffered a greater decrease in leaf number and area than did the flooded plants relative to the control plants. Laude (1971) suggested that drought causes stomatal closure which results in decreased photosyn- thesis and activity per unit leaf area and an overall reduc- tion in leaf area. However, as noted previously in the shoot weight data, the recovery of the drought stressed plants was greater than the flooded plants. Laude (1971) also noted 'that plants subjected to drought may recover quickly and for a short time even grow more rapidly than plants not subjected to insufficient moisture conditions. The drought stressed plants apparently began adding new leaves soon after the end of the stress period as they had 20% fewer leaves than the flooded plants immediately post stress but twelve days later the plants subjected to drought had 18% more leaves than the flooded plants. Area per leaf was less for the drought stressed plants but leaf area per plant was slightly greater compared to the flooded plants, indicating that many of the leaves on the plants subjected to drought were new, smaller leaves. Root length for the three treatments and three harvests 164 Table 2. Leaf area and number of leaves immediately post stress and 12 days after stress as affected by flood and drought stress. Immed. Post Stress 12 Days Post Stress Lf Area I Lf Area Lf Area I Lf Area Plnt" Leaves Lf'1 Plnt"1 Leaves Lf" cma cm’ cma cma Control 3170 80 39.8 4243 116 38.3 Flood 2527 76 33.3 2889 83 34.7 Drought 1852 61 30.2 2995 101 29.9 LSD.05 804 21 3.6 964 35 9.9 (ns) (ns) (ns) are reported in Table 3. Total root length was significantly decreased by the flood and drought stresses relative to the control plants, and the roots of the stressed plants re— mained smaller throughout the study. Rooting depth was also adversely affected by flood and drought. Immediately post stress, the drought stressed plants had the smallest total length of roots. Root distribution from top to bottom of the tube-pots was most even in the control with 68% of the roots in the top 20 cm of sand, 19% in the 20-40 cm depth range, and 13% in the bottom 30 cm of the pots. The flooded plants had a much higher concentration (80%) of roots in the top 20 cm. Only 5% of the total root length on the flooded treatment was in the lower 30 cm of the pot. This was probably due to increased oxygen availability near the surface. Jackson and Drew (1984) pointed out that 165 Table 3. Total root length of dry beans by depth for three dates following flood and drought stress. Immediately Post Stress* Depth (cm) Treatment 0-20 20-40 40-70 Total ------- root length (m) ------ Control 198.1 55.4 36.1 289.6 Flood 172.2 33.3 10.2 215.7 Drought 147.9 37.9 16.1 201.9 LSD.05 67.2 28.0 24.4 12 Days Post Stress** ------ root length (m) ------ Control 286.9 111.3 63.8 462.0 Flood 176.8 80.0 47.3 304.1 Drought 234.7 55.4 33.4 323.5 * One replication only 41 Days Post Stress* ------ root length (m) ------ Control 231.3 67.5 65.0 363.8 Flood 98.5 13.1 5.4 117.0 Drought 127.6 46.6 28.8 203.0 LSD.05 90.8 27.0 32.2 * Average of three replications ** One replication 166 flooding generally retards root growth and that root survival time varies from a few minutes to several days. Since the flood stress in this study was applied over a seven day per- iod, it is probable that some root death occurred. The plants subjected to drought were intermediate, with 73% of their roots in the top 20 cm of the profile, 19% in the 20-40 cm depth range, and 8% in the lower 30 cm of the pots. 12 days after the stress the plants subjected to drought had recovered more than the flooded plants as evidenced by the greater total root length measured at this harvest. This higher rate of recovery by the water deficit plants was also seen in the leaf data, (Table 2) and, by the third harvest, the shoot data (Table l). Apparently the rate of damage or root kill was less on the drought stressed plants than on the flooded beans and this allowed the plant system to recover more rapidly. By the final harvest the plants were mature and total root length had decreased for each treatment from the two earlier harvests. At this harvest the drought stressed plants had greater total root length as well as more even distribution of roots than the flooded plants. 63% of the roots of the plants subjected to drought were in the top 20 cm of soil, 23% in the 20-40 cm range, and 14% of the roots in the lower part of the profile. On the flooded plants the root length distribution by depth from top to bottom was 84, 11, and 5%. The control root systems had penetrated well throughout the profile, with 64, 19, and 18% of their root 167 length respectively from top to bottom. Shoot dry weight:root length ratios are presented in Figure 1. Shoot:root ratios increase with time as expected (Russell, 1977). However, the higher shoot:root ratios ex- pected on the control plants compared to the stressed plants were not evident. Taylor (1981) suggested that a plant will favor the stressed part of the plant, i.e. if the root is stressed it will receive an increased share of photoassimi- late with a resulting decrease in shoot:root ratio. In this study the effects of the stresses were such that root length was severely restricted and the expected decrease in shoot:root ratio was hat evident. Final harvest parameters are presented in Table 4. The effects of the flood and drought stresses are clearly reflec- ted in the yield parameters. The control plants yielded sig- nificantly more beans, both in number of beans as well as fresh and dry weight, than did the stressed plants. The yield response of the two stress treatments were similar. The drought stressed plants outyielded the flooded plants by 16%, (dry weight of beans per plant) although the differences were not significant at the 95% level. The yield advantage of the drought stressed plants over the flooded plants is another indication that the plants subjected to drought re- covered better from the stress than did the flooded plants. SHOOT:ROOT (g m"'1) 168 0.20 E CONTROL 0,13 m FLOOD mm DROUGHT 0.16 0.14 0.12 0.10 0.08 0.06 0.04» 0.02 0.00 4-9 61 90 DAYS AFTER PLANTING Figure l. Shoot:root ratios (dry weightzlength) of dry beans as affected by flood or drought stress and time following stress. 169 Table 4. Dry bean yield components as affected by flood and drought stress. Fresh Wt Dry Wt l I Beans Pods Beans Pods Beans Pods Beans /Pod ------ grams plant“ ------ Control 9.5 31.1 7.5 27.5 34.1 141.0 4.18 Flood 4.1 13.8 3.5 12.3 19.0 84.3 4.45 Drought 8.4 13.9 4.0 14.7 18.9 81.8 4.45 LSD.OS 5.7 10.0 2.9 5.9 11.8 30.8 1.00 (ns) (ns) SUMMARY In summary, the effects of drought and flood were detri- mental to plant growth at the time of the stress. The ef- fects continued on well after the stress had been alleviated. Shoot growth, leaf area, and root growth were all signifi- cantly less for plants subjected to flood or drought stress. The injury to the stressed plants swas too great, and final yield was significantly decreased relative to non-stressed plants. Growth of the drought stressed plants was more negatively affected than the flooded plants at the time of the stress. However, the drought stressed plants responded better following the stress than did the flooded plants as evidenced by shoot, root, and yield parameters over time. The flooded root systems sustained more permanent damage and failed to recover to the same extent as plants with too lit- tle soil moisture. 170 The effects of flood and drought stress are clear. The results provide some indication as to what might be expected under field conditions. If a crop is subjected to stress due to too much or too little moisture, an adverse affect on yield can be expected. It is expected that the effects of such environmental stresses which often occur naturally will be increased when soil problems such as compaction are pres- ent. REFERENCES Bertrand, A. R. and H. Kohnke. 1957. Subsoil conditions and their effects on oxygen supply and the growth of corn roots. Soil Sci. Soc. Am. Proc. 21: 135-140. Bennie, A. T. P. and F. J. P. Botha. 1986. Effect of deep- tillage and controlled traffic on root growth, water-use efficiency, and yield of irrigated maize and wheat. Soil Tillage Res. 7:85-95. Cannell, R. Q. and M. B. Jackson. 1981. Alleviating aeration stresses. p. 141-192. In G. F. Arkin and H. M. Taylor (eds.) Modifying the root environment to reduce crop stress. ASAE, St. Joseph, Michigan. Glinski, J. and W. Stepniewski. 1985. Soil aeration and its role for plants. CRC Press, Boca Raton, Florida. p.149- 152. Hoogenboom, G., M. G. Huck, and C. M. Peterson. 1987. Root growth rate of soybean as affected by drought stress. Agron. J. 79:607-614. Huck, M. G. 1970. Variation in taproot elongation rate as influenced by composition of the soil air. Agron. J. 62:815-818. C. M. Peterson, G. Hoogenboom, and C. D. Busch. 1986. Distribution of dry matter between roots and shoots of irrigated and non- irrigated determinate soybeans. Agron. J. 78:807-813. Jackson, M. B. and M. C. Drew. 1984. Effects of flooding on growth and metabolism of herbaceous plants. p.47-128 in T. T. Kozlowski, (ed.) Flooding and plant growth. Academic Press, Inc., Orlando, Florida. Klepper, B., H. M. Taylor, M. G. Huck, and E. L. Fiscus. 1973. Water relations and growth of cotton in drying soil. Agron. J. 65:307-310. Laude, H. M. 1971. Physiological processes and growth. In K. L. Larson and J. D. Eastin (eds.) Drought injury and resistance in crops. Crop Sci. Soc. Am. Special Publication #2. Crop Sci. Soc. Am., Madison, Wisconsin. 171 172 Letey, J., L. H. Stolzy, and G. B. Blank. 1962. Effect of duration and timing of low soil oxygen content on shoot and root growth. Agron. J. 54:34-37. Raghavan, G. S. V., E. McKyes, and M. Chasse. 1977. Effects of wheel slip on soil compaction. J. Agri. Eng. Res. 22:79- 83. Russell, R. S. 1977. Plant root systems: Their function and interaction with the soils. McGraw Hill, London. Schwartz, H. F. 1980. Miscellaneous problems. In H. F. Schwatz and G. E. Galvez (eds.) Bean production problems. CIAT, Cali, Colombia. Shank, D. B. 1945. Effects of P, N, and soil moisture on top- root ratios of inbred and hybrid maize. J. Agric. Research 70: 365-377. Smucker, A. J. M. and A. E. Erickson. 1987. Anaerobic stimulation of root exudates and disease of peas. Plant and Soil 99:423-433. Taylor, H. M. 1981. Managing root systems to reduce plant water deficits. In R. 3. Russell, K. Igue, and Y. R. Mehta (eds.) The soil/root system in relation to Brazilian agriculture. Proceedings of a symposium on the soil/root system. Published by Fundacao Instituto Agronomico do Parana, Londrina, Parana, Brazil. Tennant, D. 1975. A test of a modified line intersect method of estimating root length. J. Ecol. 63:955-1001. CHAPTER 6 CONCLUSIONS Each of the preceding chapters included the primary conclusions from the study detailed in that chapter. This chapter summarizes those conclusions. Chapter 2: Alleviation of Soil Stresses on Dry Bean Root Systems As a Key to Increased Dry Bean Production A) Soil physical properties of Charity clay were improved by the use of a deep-rooted legume, deep tillage, and no secondary tillage. -Bulk density was significantly less at the 15-22.5 cm depth range on the ARNST soil compared to the CONV soil. -Soil moisture and aeration relations were improved by the ARNST management system. -Pore size distribution was more favorable on the ARNST soils, with more larger pores for improved drainage and fewer smaller pores which would inhibit drainage, air flow, and possibly root penetration. -Hydraulic conductivity was significantly improved on the ARNST soils at the 15-22.5 cm depth. 173 174 B) Plant growth was largely unaffected by the different ro- tation/tillage and irrigation treatments. Differences were detected due to row spacing and cultivar. -Shoot biomass and root length density did not differ significantly due to rotation/tillage and irrigation. -Narrow rows did have higher biomass per unit area.. -Emergence of the Black Magic cultivar was better than that of the C-20 cultivar. C) Yield was not determined because of extensive flooding. Yield estimates showed no significant differences due to rotation/tillage or irrigation. The narrow rows had higher yields than the wider rows. The Black Magic cultivar outyielded the C-20 cultivar. In the one year of this study no advantages to plant growth, root length density, or yield were detected due to the ARNST management system in spite of apparent improvements in soil physical conditions. Chapter 3: Evaluation of Minirhizotron Observation Tubes as a Tool for Root Study on Fine Textured Soils A) The minirhizotron observation tubes did not present an accurate picture of root activity in the top 20-30 cm of soil under either rotation/tillage system in either 1985 175 or 1986. Destructive sampling revealed greater numbers of roots than did the minirhizotron tubes in this soil horizon. B) The minirhizotron method results showed much greater root activity in the 30-60 cm depth range under the CONV soils than on the ARNST soils. This data was contrary to that collected by destructive sampling. The results indicate that the minirhizotron method of root observation is not a useful tool on fine textured soils with compacted zones. Roots apparently proliferate at the tube-soil interface, especially on soils which are compacted. Chapter 4: Root Length and Width Determination by Digital Image Processing A) Washed root samples can be easily video recorded using the system described. B) The digital image processing system was not entirely suc- cessful in determining root length. Major problems were related to debris in the sample and the noise in the sy- stem from the video recording and image processing steps. 176 C) Root width determination is only generally possible with the current system. D) The current system is time consuming. The use of digital image processing for root studies holds much potential. The current system shows that the technology is available to carry out the task. However, much work remains to be accomplished prior to broad scale application to root length and width determination. Chapter 5: The Effects of Flood and Drought on Dry Bean Root and Shoot Growth A) Flood and drought stress are detrimental to plant growth. Dry bean plants were not able to recover completely following subjection to flood or drought. B) Dry beans recovered more quickly following drought than flood. The effects of these two environmental stresses are harmful to plant growth and are long lasting to the extent that yield is significantly impaired.