| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
ANIMAL GENETICS |
* Department of Animal Sciences, Colorado State University, Fort Collins 80523-1171; and
Landcorp Farming Ltd., Rotorua, New Zealand
 |
Abstract |
|---|
 Top Abstract INTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONLITERATURE CITED |
|---|
These EBV were in turn used to calculate annual genetic changes. The average annual genetic changes for weaning weight direct and maternal breeding value were 0.43 ± 0.05 and 0.03 ± 0.22 kg/yr, respectively. Corresponding annual genetic changes for postweaning BW gain, yearling weight, harvest weight, and mature BW were 0.29 ± 0.03, 0.72 ± 0.06, 1.7 ± 0.13, and 0.13 ± 0.09 kg, respectively. The annual change in number of calves weaned per cow lifetime was 0.006 ± 0.001 calves/cow and the change in dressing percentage was estimated to be –0.035 ± 0.003 %/yr. At the end of the program, 3.21 generations of selection had occurred with a mean accumulated selection differential of 3.87 SD. Change in objective traits due to selection was similar to or exceeded change predicted at the onset of the program with the exception of mature BW and dressing percentage. Genetic change in mature BW was not different from zero, whereas the predicted change was 29.3 kg. The overall genetic trend in the breeding objective exceeded that predicted at the onset of the program. Results of this study showed that selection on indexes developed to predict an economically based, multi-trait breeding objective will produce genetic change.
Key Words: beef cattle • breeding objective • genetic change • selection index
|
INTRODUCTION |
|---|
|
TopAbstractINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONLITERATURE CITED |
|---|
). In an effort to overcome this deficiency, economic breeding objectives and associated selection indexes have begun to be delivered to the beef production industry (e.g., Sundstrom and Barwick, 2002; American Angus Association, 2006; American Simmental Association, 2006).
The original concept combining genetic evaluation and economics of production was presented by Hazel and Lush (1942) and Hazel (1943), but until recently was rarely used in the beef industry outside of the research setting. MacNeil (2003) reported on successful genetic improvement using a 2-trait index including yearling weight and birth BW in a research environment. To our knowledge little has been reported on genetic change in an integrated seedstock-commercial beef cattle production system where selection decisions were based on indexes comprising more than 2 traits and where those indexes were designed for a specific economic breeding objective itself including only traits directly influencing profitability. Previous studies reported changes in index and correlated traits (Frahm et al., 1985; Koch et al., 1994), but these indexes were not developed for a specific economic breeding objective. The objective therefore of this study was to determine the long-term genetic change in a commercial beef breeding program resulting from selection for indexes developed for an economic breeding objective and to quantify changes in objective traits given selection pressure on the index.
|
MATERIALS AND METHODS |
|---|
|
TopAbstractINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONLITERATURE CITED |
|---|
Animal Background
Data for this study were drawn from a nucleus Angus seedstock herd owned and managed by Landcorp Farming Ltd., a government-owned but privately managed corporation and the largest farming/ranching enterprise in New Zealand. This open nucleus herd is located on the central North Island of New Zealand at an elevation of 480 to 900 m with 1,000 to 1,800 mm of annual precipitation. The nucleus Angus herd was established in 1970 with cows screened-in from commercial herds owned by the company from throughout the central region of the North Island of New Zealand. Selection of cows to form the nucleus was based on BW of calf weaned with cows weaning the heaviest calves selected. Screening procedures were further described by Dalton and Gibson (1974), but included annual screening of heifers from commercial properties. The primary focus of this nucleus was to produce bulls for distribution and use in the commercial breeding herds of the company in an effort to genetically improve profitability.
Breeding Objective Background
In 1976 the company began selecting animals in this nucleus for an economic breeding objective developed by Morris et al. (C. A. Morris, R. L. Baker, and D. L. Johnson, Ministry of Agriculture and Fisheries, Ruakura Animal Research Station, Hamilton, New Zealand, unpublished data, 1978) and described by Nicoll et al. (1979) and Nicoll and Johnson (1986). Briefly, the breeding objective was defined as
 |
where 0.53, 0.06 = the net income (1976NZ$/kg of carcass) from the harvest of young stock and cull cows, respectively; HW = harvest weight (kg) of surplus progeny at 30 mo of age; DP, DM = dressing percent-age (x 0.01) of harvested progeny and the culled cow, respectively; F = net fertility (calves weaned per cow exposed); and M = BW (kg) of cow at disposal.
The term "4.8 x F" describes the average total number of saleable calves per cow lifetime at the time of implementation. One was subtracted from 4.8 x F to account for the replacement of the cow in the herd. Because F was expressed as a deviation from contemporary group, 4.8 x F represents the number of calves weaned per cow lifetime (NCW) when the cow is culled. Costs of production and income used in the development of this objective were based on data from the Economic Service of the New Zealand Meat and Wool Boards (C. A. Morris, R. L. Baker, and D. L. Johnson, Ministry of Agriculture and Fisheries, Ruakura Animal Research Station, Hamilton, New Zealand, unpublished data, 1978).
The breeding objective did not directly account for costs of production as proposed by Ponzoni and Newman (1989), and subsequently by Newman et al. (1992) for the New Zealand beef industry. Rather than include feed intake in the breeding objective directly, gross income realized through increased carcass weight was adjusted to reflect associated increases in feed intake, and, therefore, feed costs of females and excess harvest progeny. Eleven and 32% of the gross returns resulting from increased harvest weight of surplus progeny and of cull cows, respectively, were allocated to extra feed costs associated with larger carcass weights. When the breeding objective was implemented in 1976, the developers felt that there was insufficient research published on the genetic correlations between feed intake of grazing cattle and the traits likely to be included in the selection index (C. A. Morris, R. L. Baker, and D. L. Johnson, Ministry of Agriculture and Fisheries, Ruakura Animal Research Station, Hamilton, New Zealand, unpublished data, 1978). The resulting breeding objective, as presented above, therefore indirectly accounted for changes in costs of production due to changes in feed consumption and associated costs.
Selection indexes were calculated using traits for which data were normally recorded. Individual performance measures on all traits in the breeding objective were not available, hence the need to use other traits in an index predicting the breeding objective. The index for selection of animals at a year of age included adjusted deviations from contemporary group means for weaning (WW) and yearling weights of the individual (YW), the average lifetime BW of the dam of the calf weaned (WCW), and the fertility of the dam (NCW). These indexes were calculated using the heritabilities and genetic relationships shown in Table 1. Index weighting factors for each trait (Table 2) changed depending on the number of calves weaned by the dam of the calf when the selection index value was calculated for a particular individual. For instance, the index value of a yearling out of a cow that had weaned a total of 4 calves to date would be
|
|
Adjusting weighting factors for number of calves produced by the dam accounts for the increased knowledge of the breeding value of the dam for NCW and WCW due to the repeated observations. This knowledge of the dam therefore changes the weighting factors for all traits in the index. Predicted genetic change in the breeding objective traits per 1 SD of selection (Table 3) was determined at the initiation of the program for differing numbers of calves weaned by the dams. Although dressing percentage of cull cows was included in the objective, no data were collected, and, with a lack of scientific literature on mature cow dressing percentage, it did not receive further consideration in this study.
|
). Animals were selected solely on these values, excepting situations where an animal was identified as structurally or functionally unsound and(or) as likely to be infertile. Mean accumulated selection differential was, therefore, used to quantify actual selection pressure and is discussed further below. Whenever practical, culling of pregnant cows occurred and was based on a cow index that included adjusted contemporaneous deviations of the weaning weight, yearling weight, average lifetime BW of calves weaned, and fertility of the cow, as well as the average lifetime BW of calf weaned and fertility of the dam of the cow. We have chosen not to present the cow weighting factors here, given that over 100 different possible indexes were used in the breeding program (based on up to 10 cow weaning records in combination with up to 10 dam weaning records). Data Description
Genetic improvement in the nucleus and ultimately throughout the company was based solely on selection for this breeding objective from 1976 through 1993. Since 1993 the breeding objective has been modified and did not provide a consistent selection program from which to determine genetic change for the specific objective.
Performance and pedigree information for this study included animals born during the selection period and their parents for a total of 16,189 animals. Data were sifted for useable observations with performance observations coded as unknown if the data were from twins or foster dams, from dams less than 2 yr of age, from unrealistic ages at measurement (e.g., yearling weights recorded before 300 d or after 430 d), and indicated unrealistic weights (e.g., weaning weight less than 80 kg). For animals born between 1976 and 1993 inclusive, data included 11,649 WW and 11,569 YW observations; 9,209 mature cow BW (MW) observations and 3,862 NCW records on 3,964 cows. These totals included 4,401 animals born before 1976 that were either screened-in to form the original nucleus population or were produced in the nucleus herd from 1970 through 1976. Including these foundation animals provided improved pedigree relationships and estimates of levels of production in the nucleus herd before selection on the breeding objective.
Statistical Analysis
Variance components for WW, postweaning BW gain (PWG; calculated as the difference between YW and WW), cow MW, and cow NCW were estimated using ASREML (Gilmour et al., 2002). A 2-trait, multi-component model was used to estimate heritabilities and genetic correlations for WW and PWG as follows:
where y is a vector of observations on PWG (trait 1) and WW (trait 2) with β a vector of fixed effects, µ is a vector of direct (d) and maternal (m) additive genetic effects, c is a vector of permanent environmental effects due to the dam, and e is a vector of random residual effects. The variance structure is
where 
d2 is the direct additive genetic variance, m2 is the maternal additive genetic variance, 
is the permanent environmental variance due to the dam, and e2 is the residual variance. The d1d2 refers to the direct genetic covariance between traits 1 and 2, and e1e2 represents the residual covariance between traits 1 and 2.
A single-trait, repeated-measures model was used to estimate (co)variance components for MW, including a direct additive genetic effect and an uncorrelated permanent environmental effect of the cow. The same model was used to estimate variance components for NCW except that a permanent environmental effect of the cow was not included as there were no repeated measures.
Fixed effects for WW were year of birth and management group as contemporary group, a sex by age of dam effect, and weaning age as a covariate. Fixed effects for PWG were contemporary group as a combination of weaning contemporary group and yearling management group, a sex by age of dam effect, and a covariate for number of days postweaning. Fixed effects for the MW analysis were year of birth, age at measure (year), and weaning contemporary group of the calf. For NCW, fixed effects were year of last measurement and year of birth. Age of dam and age at measurement groups were 2, 3, 4, 5, and 6 yr and older.
The same models used to estimate variance components were also applied to calculate breeding values (EBV) using the previously estimated (co)variance components with exceptions for harvest weight (HW) and dressing percentage (DP). Both of these traits had no available performance observations. To calculate EBV for these traits, they were included in the WW-PWG analysis as traits genetically correlated with both WW and PWG. Phenotypic variance for HW was assumed to be 1,764 kg2 (Koots et al., 1994a) with a heritability of 0.41 (Koots et al., 1994b). The genetic correlation of HW with weaning weight direct (WWD) and PWG was assumed to be 0.39 (Crews et al., 2004) and 0.57 (MacKinnon et al., 1991), respectively. Dressing percentage was assumed to have a phenotypic variance of 3.61 (Koots et al., 1994a) and a heritability of 0.39 (Koots et al., 1994b). Genetic correlations of DP with WWD, PWG, and HW were, respectively, –0.50, 0.16, and –0.23 (Koots et al., 1994a).
Generation intervals for both sires and dams were calculated, as well as proportion of each sex selected to become replacements each year. A generation number was calculated for each calf born in the program using the method of Brinks et al. (1961), where generation coefficient of the calf (GC) is the average generation coefficient of the parents plus 1. Generations of selection are then GC – 1. Foundation animals and animals screened-in (with unknown parentage) were assigned a GC of 0. The mean accumulated selection differential was calculated using the procedure described by Newman et al. (1973) and is equivalent to the cumulative selection differential described by Koch et al. (1994). When calculating the accumulated selection differential, heifers screened-in were assumed to have performance that was average for females born in the nucleus herd in the same year. The differential was standardized using the index variance (Table 3) for a cow having weaned 3 calves. This choice was based on the average female generation interval of 4.4 yr from 1980 through 1993. We did not consider earlier years due to the herd makeup consisting of a large number of older, screened-in females.
The standardized, accumulated selection differential was then regressed on year of birth to determine the predicted genetic improvement per year in the index due to selection applied from 1976 through 1993. Predicted gains for each of the traits in the breeding objective, given 1 SD of selection on the index, are shown in Table 3. These were then compared with actual gains achieved during the selection program, where actual gains were measured using the mean accumulated selection differential and the genetic trends calculated through the BLUP procedures as described below.
Genetic trends for each trait were calculated using the average EBV for each year of birth. A YW EBV was calculated as the sum of the WWD and PWG EBV. The resulting YW EBV was then used for genetic trend calculations. Regression coefficients of yearly, average EBV on year of birth were then calculated using a general linear model (SAS Institute Inc., Cary, NC) that included the fixed effects of period (pre- vs. post-implementation of the breeding objective) and regression of trait EBV on year of birth within period. The subclass regressions were then tested for a significant difference from zero to determine whether long-term genetic changes were evident.
|
RESULTS AND DISCUSSION |
|---|
|
TopAbstractINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONLITERATURE CITED |
|---|
. The estimates of heritability for WWD and weaning weight maternal (WWM) were similar to those reported by Koots et al. (1994b), but the estimate of heritability of PWG in this study was considerably less than reported in that study, although similar to the 0.23 ± 0.13 reported by Bullock et al. (1993).
|
), but was opposite in sign to the average reported by Koots et al. (1994a) of 0.44. The magnitude of this relationship, although within the range of reported values, is one of the stronger negative correlations reported. Koch (1978) discussed this relationship based on results from selection lines of Hereford cattle. He reported WW of calves from a line selected for YW were 4 to 7 kg less than WW of calves from a line selected for WW.
The heritability of MW (0.39 ± 0.03) was less than both the 0.45 ± 0.10 reported by Northcutt and Wilson (1993) and the 0.50 assumed in the calculation of the breeding objective and associated selection indexes (Table 1), but greater than that reported by Morris et al. (1993). The MW permanent environmental effects were estimated at 0.26 ± 0.03% of the total phenotypic variation, which was greater than the 0.00 to 0.12 reported by others (Kaps et al., 1999; Rumph et al., 2002). We suspect that this was due to the variable nature of heifer development on pasture from 1970 to 1993 where from 1 to 2 yr of age, females received no special treatment or supplementation.
The heritability of NCW was greater than expected and greater than many estimates reported in the literature for fertility (e.g., reproductive rate and number of calves in Angus and Hereford cows; Meyer et al., 1990), but was similar to the values reported by Koots et al. (1994b; 0.17), Martinez et al. (2004; 0.17 to 0.21), and MacKinnon et al. (1990; 0.20) in crosses of Brahman, Hereford, and Shorthorn cattle. The NCW reported herein is a lifetime trait and includes variability due to the length of productive lifetime of the cow. As such, heritability of NCW as defined in this paper is expected to show more variability than fertility in one production season as reported by some of the aforementioned studies related to fertility.
Excluding NCW and possibly MW, heritability estimates in Table 4 were similar to those used in the calculation of the indexes for this breeding objective. The "maternal weaning weight" as used in the original selection indexes (i.e., WCW: average adjusted lifetime BW of calf weaned deviated from contemporary group mean) represented total maternal value (0.5 x [BV for WWD] + [BV for WWM]), whereas the WWM estimated in this study refers only to the maternal component of weaning weight.
Average generation intervals for sires and dams and proportion of sires and dams retained as replacements are shown in Table 5. During the course of the selection program, heifers were continually screened-in to the nucleus herd with numbers of these varying depending upon the year (Table 5). For calculations of mean accumulated selection differential, the screened heifers were assumed to have average performance for all heifers born in the same year. At the end of the study, 3.21 generations of selection had occurred (generation number – 1), which was similar to that reported by Frahm et al. (1985) for 2 selection lines where one line was selected for weaning weight and the other for yearling weight. In contrast, Koch et al. (1994) reported almost 5 generations of selection in each of 3 selection lines 20 yr after the selected individuals began entering the breeding herd. However, in this breeding program, heifers were annually screened-in to the nucleus herd and, due to the unknown parentage of these individuals, were necessarily assigned a generation number of 1. Given that many of these heifers were offspring of bulls produced in the nucleus herd and subsequently distributed to the commercial properties, these screened-in heifers were likely of a greater generation number than that assigned. Additionally, essentially twice the number of heifers normally screened-in were introduced from the 1984-born heifers (Table 5) and resulted in a smaller generation number than in previous years.
|
per yr from 1976 to 1993, which is similar to the 0.24 selection differential per yr reported by Koch et al. (1994) for weaning and yearling weight in lines selected for weaning and yearling weight, respectively, but greater than the 0.20 selection differential for index value from the index selection line. In that study the index included yearling weight and muscle score. The regression of generations of selection on year of birth was 0.15 ± 0.009 generations per year in this study.
Genetic trends are presented in Figures 1 and 2 and in Table 6. The lack of significant change in WWM was not surprising given the distinction referred to earlier. Variance in maternal weaning weight, as defined in the original selection index, includes 25% of the additive direct genetic variance, 100% of WWM variance, 100% of permanent environmental (PE) variance, and 50% of the covariance of WWD and WWM. The relatively small increase in WWM should be expected given that WCW is only 1 trait in the 4-trait index, was assumed to be negatively related to WWD, and is only partially composed of WWM. MacNeil (2003) reported positive progress for maternal effects on 200-d BW using an index that included birth and yearling weight (Dickerson et al., 1974), but that study also reported a positive genetic correlation between direct and maternal effects for 200-d BW.
|
|
|
). With 3.87 SD of selection occurring over the course of the breeding program, increase in mature BW would have been expected to be 29.3 kg (Table 7). Although speculative, preadjustment for the increased costs associated with increased feed requirements due to increases in growth rate and mature cow size appears to have applied downward pressure on mature BW and held change in that trait to a minimum. MacNeil (2003) also reported less than expected genetic increases in mature cow size resulting from index selection that was designed to increase growth at 1 yr of age.
|
). In practice, there was a change of 0.006 calves weaned per year for a change of 0.102 over the course of the study—almost double that predicted (Table 7). The difference is likely due to the differences in heritability assumed when the indexes were developed (0.05) compared with that actually estimated from the data (0.20). Increased heritability would result in increased genetic change due to more accurate selection. The indexes used to select the nucleus herd replacements in this study were calculated using least squares adjustments and deviations of adjusted records from contemporary group means, using data from the individual, its dam, and the repeated progeny of the dam. The methodology used in this study to estimate variance components and to calculate breeding values (animal model BLUP) was not feasible due to computing limitations in 1976 and before. The EBV calculated using present-day genetic prediction technologies would be expected to have greater accuracy of prediction due to the use of all available data and full accounting for relationships between animals, and, therefore, would be expected to allow greater genetic progress were selection successful.
Changes in HW and DP were more difficult to quantify as no data were collected on these objective traits and any attempt to quantify actual genetic change is dependent upon the variance components assumed for the calculation of the EBV. Still, having used estimates from a meta-analysis of literature estimates provides some confidence in the predicted change. Based on the indexes used, predicted change for HW and DP was 27.7 kg and –0.08%, respectively. Actual changes, given the assumed variance components and relationships with other traits, were 28.9 kg and –0.60%, respectively. Heritabilities assumed in these analyses (0.41 for HW and 0.39 for DP) were similar to those used to form the original indexes (0.35 and 0.40 for HW and DP, respectively; Table 1). The drastically different change in dressing percentage predicted when the index was developed versus that seen is likely a result of a different genetic relationship between HW and DP when the index was developed (0.04) as opposed to that assumed in the breeding value calculations (–0.23; Koots et al., 1994a).
The nucleus herd was formed in 1970 through the screening of cows having weaned heavy calves in other herds of the company. Selection of nucleus animals based on the breeding objective commenced with animals born in 1976. The rate of genetic change from 1970 to 1975 before the implementation of this breeding objective was not significantly different from zero for any of the traits evaluated in this study (Table 6). This result was not surprising given the time required to formalize a recording and selection program for beef cattle, and to benefit from selections made from 1970 through 1975.
The changes in each of the breeding objective component traits were applied to the breeding objective equation to estimate average change in the aggregate breeding value (H), using the same methodology as applied to the individual traits. Again, there was no significant change in H from 1970 to 1975, but there was a significant improvement in H from 1976 through 1993 (1976NZ$5.30/yr). This assumes that the relationship between DP and the other objective traits is negative as often reported in the literature (e.g., Koots et al., 1994a). However, given that MW remained constant throughout the selection period, it is likely that DP did not change. If this assumption is true, H, net income, increased at an annual rate of 1976NZ$5.72 per cow lifetime.
To put these values on a relatively current (March 2006) US dollar scale, we used the New Zealand consumer price index (CPI; Statistics New Zealand, 2006) for March 1976 (CPI = 174) and for March 2006 (CPI = 1,184) and the ratio of 1,184/174 as a multiplicative conversion factor (M. Frasier, Colorado State University, Fort Collins, CO, personal communication). Using the average March 2006 conversion from New Zealand dollars to US dollars (0.6342), the change in net income per year would be US$22.87/yr change in H, and, if DP was stable, US$24.68/yr. Approximately 6,600 commercial heifers sired by bulls selected for this breeding objective were put into production yearly. The first replacement heifers sired by this program would have entered the commercial herds from 1978 to 1995, resulting in the inclusion of approximately 112,200 heifers into the Landcorp Farming system. Based on the results and breeding objective above, this would have increased the inherited component of profit-earning ability in this enterprise of the company from $2,566,014 (2006 US) to $2,769,096 (2006 US), depending upon which regression coefficient was used.
Discounted gene flow (McClintock and Cunningham, 1974) was not used to recalculate economic values and then to quantify the economic progress from selection for this objective for several reasons. First, the original indexes did not include those adjustments when economic values were originally calculated (C. A. Morris, R. L. Baker, and D. L. Johnson, Ministry of Agriculture and Fisheries, Ruakura Animal Research Station, Hamilton, New Zealand, unpublished data, 1978). Second, subsequent studies have found high correlations between objectives using various methods of calculating economic values (Ponzoni, 1986). Newman et al. (1992) reported small changes in the correlation between index and objective when expense per year or 0, 5, and 10% discounting rates were used. They speculated this was due to the lack of very large negative genetic correlations assumed in that study—a condition that holds true in this study—and that age distributions changed little. From 1980 onwards, there was a relatively constant age distribution of females in this study as well. Finally, most indexes currently used in the United States do not include discounting (M. MacNeil, Ft. Keogh Livestock and Range Research Laboratory, Miles City, MT; personal communication; S. Northcutt, American Angus Association, St. Joseph, MO; personal communication).
To our knowledge this is the first study reporting genetic improvement in a commercial beef cattle breeding program resulting from selection for an economic breeding objective and using indexes that did not contain all traits of economic importance. Other studies have reported progress from selection on indexes (Koch et al., 1994; MacNeil, 2003), but these indexes involved only 2 traits and were designed for specific improvement of the traits in the index. The results reported here focused on indexes designed to change the traits in a distinct economics-based breeding objective where several of the traits in the index did not directly affect profitability (with the exception of number of calves weaned). For instance, the primary source of income for the company was from the sale of harvest-ready surplus progeny and mature cull females, not from the sale of weaned calves.
Given the commercial nature of this breeding program, the study lacks a control selection line. However, the results support the use of multi-trait selection indexes to predict an economic breeding objective in beef cattle genetic improvement programs even when based on historic estimation techniques appropriate at the time of development as in this selection program. Development of breeding objectives that use breeding values estimated via animal model, BLUP techniques should provide even faster rates of genetic change due to the increased accuracy of the breeding value predictions.
Implementation of an economics-based breeding objective in the mid 1970s produced genetic change in most of the traits in the objective. Acknowledging that not all genetic correlations were known at the time of implementation of index selection, and that increases in feed intake and costs were adjusted before calculation of the indexes, the application of index selection methodology was successful in a beef cattle genetic improvement program. This breeding objective was one of the first applied in an integrated seedstock and commercial beef cattle enterprise. At present, the development of breeding objectives and selection indexes is often generalized for national systems, using averages for costs and revenues of production rather than accounting for specific sectors and production systems in the industry. Further research should be conducted to determine if such "generalized" industry-wide breeding objectives would produce similar rates of genetic improvement to breeding objectives tailored to more specific production and marketing sectors.
|
Footnotes |
|---|
2 Corresponding author: Mark.Enns{at}Colostate.edu
Received for publication August 23, 2006. Accepted for publication April 28, 2008.
|
LITERATURE CITED |
|---|
|
TopAbstractINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONLITERATURE CITED |
|---|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
