| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
* CoRFiLaC, Regione Siciliana, 97100 Ragusa, Italy
Department of Animal Science, Cornell University, Ithaca, 14853
Instituto de Ciências Biomédicas Abel Salazar and Centro de Investigação em Biodiversidade e Recursos Genéticos, Universidade do Porto, 4485-661 Vairão, Portugal
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSACKNOWLEDGEMENTSREFERENCES |
|---|
Key Words: genotype x environment interaction • heterogeneous genetic variance • milk yield • environmental definitions
Abbreviation key: GxE = genotype by environment interaction, HYSD = herd-year standard deviations, ME = Mature equivalent, WSCS = weighted somatic cell score
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSACKNOWLEDGEMENTSREFERENCES |
|---|
Licitra and collaborators (1998) found substantial peakless lactations in herds in Mediterranean southeastern Sicily (Italy) lacking normal detectable ascent to peak milk yield followed by a rapid (convex) decline. These abnormal outcomes signified substantial sacrifices in milk production with restricted within-herd variation. Potential causes of opportunity loss in milk were inadequate diet from low availability and poor quality of forage, insufficient reserves of body tissues to support early lactation, and inadequate health management. Extending the results from other studies to this Sicilian case, where genetic parameters of dairy production have not yet been estimated, the unequal input use (environmental opportunity) associated with abnormal lactation would be expected to underwrite significant GxE in dairy performance.
Therefore, the objectives of this study were to evaluate potential GxE in yields of milk, fat and protein, and in somatic cell scores of Friesian and Brown Swiss cows by comparing genetic parameters and expected correlated responses in contrasting Sicilian herd environments alternatively defined. Farmers need to know the opportunity losses in daughter response due to management inputs, which are known to influence selection decisions (Holmann et al., 1990; Blake, 1992).
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSACKNOWLEDGEMENTSREFERENCES |
|---|
Test day milk yields and somatic cell counts of the first three lactations were provided by CoRFiLaC, Sicily’s main dairy research center, and the Associazione Provinciale Allevatori of Ragusa, Sicily. At least four test day records were required per lactation. Observations for DIM less than 5 or greater than 310 and intervals less than 15 d or greater than 75 d between observations were not considered. Weighted SCS (WSCS) per lactation was created from the monthly somatic cell scores for each cow with at least one SCC. SCS, which is homoscedastic among samples, was calculated for each SCC observation as: (logeSCC/loge2)-(loge12.5/loge2)+12. Because SCC is inversely related to daily milk yield (Jones et al., 1984), the WSCS was calculated using test day milk (mi) associated with SCS as a weighting factor:
 |
Data on management practices were from interviews with a special questionnaire (Raffrenato, 2002) of 168 Friesian and 71 Brown Swiss herd owners conducted in 2000 through CoRFiLaC. The survey included 82% of the Friesian and 86% of the Brown Swiss herds enrolled in the local dairy recording program (Associazione Provinciale Allevatori), the majority of the populations of herds for these breeds in Sicily’s primary milk shed.
Definitions of Contrasting Herd Environments
Within herd-year standard deviation for lactation milk yield.
The phenotypic HYSD for 305-day ME milk yield was used to discriminate herds like in other studies (Boldman and Freeman, 1990; Dong and Mao, 1990; Stanton et al., 1991a; 1991b; Cienfuegos-Rivas et al., 1999; Costa et al., 2000). Low opportunity environments had HYSD <1330 kg for Friesian herds and HYSD <950 kg for Brown Swiss herds. High opportunity environments were HYSD >1370 kg for Friesian herds and HYSD >990 kg for Brown Swiss herds. Accordingly, 132 Friesian herds and 37 Brown Swiss herds were allocated to the low HYSD class, and 76 Friesian and 30 Brown Swiss herds were allocated to the high class. There were 315 (of 825) Friesian sires and 63 (of 220) Brown Swiss sires with recorded daughters in both HYSD environments.
Detectable peak daily milk yield.
Data files of test day milk records were obtained for cows in two parity groups: cows in first lactation and those in second and third lactations. The days in milk observations for these records were divided into 31 classes for 18 age-at-calving groupings. The final data sets consisted of 195,705 records for 228 Friesian herds and 39,916 records for 78 Brown Swiss herds. Solutions were obtained with the following mathematical model for each DIM class for each parity group in each herd:
 |
where yijklmn is the nth test day observation at herd-year-season i, age at calving class j, herd k, parity l and DIM class m; HYSi is the level i for the herd-year-season fixed effect; Agej(Hk) is the level j of the fixed effect for age at calving nested within each herd; DIMm(HkPl) is the mth level of the fixed effect for DIM nested within herd and parity; and eijklmn is the random vector of residuals assumed to be normally distributed. Relationships among animals were ignored.
The DIM solutions for each herd and parity class were fit to an incomplete gamma function (y = atbe-ct; Wood, 1967). Only cows with at least five test day records per lactation were analyzed. Herds permitting estimation for only one parity class were ignored to restrict classification errors. Detectable peak lactation was determined by testing the significance and sign of the b parameter using a significance threshold of 
= 0.05.
High opportunity Friesian herds were those with detectable peak lactation (i.e., b parameter value exceeded zero; P < 0.05) in primiparous and multiparous cows. Low opportunity Friesian herds were those with abnormal lactation in either parity group (i.e., b parameter value less than or not significantly different from zero). Low opportunity Brown Swiss herds were those with abnormal lactation in both parity groups, otherwise herds were assigned to the high environment. This response behavior would be expected to differentiate productive opportunity by nutritional and health inputs and orchestrated physiological, or homeorhetic, drivers of lactation. As a result there were 91 Friesian herds and 38 Brown Swiss herds in the low opportunity environment, and 136 Friesian and 38 Brown Swiss herds in the high environment. Similar proportions of herds were sought in the contrasted environments to assure sire representation and genetic ties between them.
Management practices that enhance milk production.
Interviews of owners indicated there were no important changes in herd management across the time period covered by the lactation performance data. Therefore, survey responses to 17 questions about management inputs (Table 1
) were used to form environmental clusters with contrasting input use frequencies. Input categories included nutrition management, milking practices, health practices, and animal handling. Each practice was assigned an asymmetric binary value (0 = no, 1 = yes) whether it would be expected to enhance milking performance. The management environmental distance of Jaccard (Kuo, 1997) between two farms, x and y, was calculated as
|
 |
where n = total number of questions considered, 
x,y = 1 if xi 
yi, x,y = 0 otherwise and vx,y = 1 if xi = yi, vx,y = 0 otherwise. By this method a specified practice is weighted more than its absence. The resulting matrices of management distances between farms were used to form two groups of herds in each breed using the flexible beta clustering method (Lance and Williams, 1967), where the value of beta governs the distance between points merged into a cluster. A beta value of -0.25 was used as recommended for conditions of variable response and widely spaced clusters (Milligan, 1989). In our case this hierarchical agglomeration method signified similar causality in each environment, or predisposing dietary adequacy, udder health and other conditions affecting yield response. This criterion resulted in 95 Friesian and 27 Brown Swiss herds in the low environment, and 73 Friesian and 44 Brown Swiss herds in the high environment.
Some herds were classified as providing the same general opportunity with our definitions. For example, there were 68 Friesian herds and 22 Brown Swiss herds common to the low HYSD and abnormal lactation environments; and 37 Friesian and 16 Brown Swiss herds with abnormal lactation and infrequent use of yield-enhancing practices. This set of criteria identified the same 25 (37) Friesian herds and 8 (6) Brown Swiss herds common to the low (high) environments.
Within-Environment Analysis
The components of variance and covariance and genetic parameters for first-lactation yields of milk, fat and protein, and weighted somatic cell score in each breed were estimated using a multiple-trait linear mixed sire model with unequal design matrices and a missing observations structure. All traits were analyzed for the average herd (all data) and low and high opportunity environments.
The model in matrix notation was:
 |
where Y = vector of records of traits within an environment; h = fixed herd-year-season effects; X = incidence matrix relating herd-year-season to records; ZQ = incidence matrix relating group of sires to daughters’ records; g = fixed genetic group effects; Z = incidence matrix relating sires to daughters’ records; s = vector of random sire effects; and e = vector of random errors. Assumptions were:
 |
where 
= refers to the Kronecker product; Go = the genetic variance-covariance matrix; Ro = the environmental variance-covariance matrix; V(e) = the residual variance.
Between-Environment Analysis
Components of covariance and genetic correlations between environments were estimated for each trait in each breed using a two-trait linear mixed sire model with unequal design matrices. This model was the same as for the within-environment analysis with identical assumptions. The size of A for the respective breeds was the same for every analysis regardless of environmental definition.
Correlated Response and Genetic Trend
Coefficients of correlated response in low opportunity environments from sire selection in the high opportunity environment were estimated by genetic regression
 |
is the estimate of sire covariance for pairs of traits in the contrasting environments and 
is the estimate of sire variance in the high opportunity herds. Genetic trends in contrasting environments were estimated for each trait and breed by regression of the weighted (by number of daughters per sire) average sire predicted transmitting ability (PTA) on year of birth of their progeny. These PTA were obtained from the between-environment analysis by adding sire solutions from the mixed model equation to respective group solutions in the same system of equations. Average PTA was expressed as a deviation from the mean PTA of cows born in 1991.
|
RESULTS AND DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSACKNOWLEDGEMENTSREFERENCES |
|---|
contains descriptive statistics for average herds (all data) and HYSD environments for Friesian and Brown Swiss cows. The HYSD definition of contrasting environments excluded 20 Friesian and 11 Brown Swiss herds to restrict overlap between them. One Friesian herd and two Brown Swiss herds were excluded for missing information to determine peak lactation. Sixty Friesian and seven Brown Swiss herds could not be clustered by management practices because they were no longer operating at the time of the survey.
|
for Friesian cows and in Table 4 for Brown Swiss cows. Sire and residual variances were smaller in low than in high opportunity environments of both breeds. This compression of variance was like that observed in diverse situations in North America and Latin America (Castillo et al., 2000; Costa et al., 2000; Cienfuegos-Rivas et al., 1999; Dong and Mao, 1990; Meinert et al., 1988; Stanton et al., 1991b). The resulting heritabilities ranged from 0.21 to 0.28 in the two breeds, which, like the associated correlations (Raffrenato, 2002), were similar to other published estimates (Castillo et al., 2000; Costa et al. 2000; DeJager et al., 1987; Maijala et al., 1974; Manfredi et al., 1984; Roman and Wilcox, 2000; Seykora and McDaniel, 1983; Schutz et al., 1990; Welper et al., 1992). Unfortunately, components of covariance from a national study of Italian Friesian (Burnside et al., 1992) were not published, which precluded comparison with our results.
|
|
Weighted somatic cell score.
Sire and residual covariances for weighted somatic cell score (Table 3
) were similar between high and low Friesian environments, except for the possible smaller genetic variation with frequent abnormal lactation. Heritabilities in contrasting HYSD environments were 0.10 in both breeds (Raffrenato, 2002), which agreed with findings by Castillo-Juarez and co-workers (2000). Genetic correlations between score and yield traits were slightly favorable in high HYSD Friesian herds (range: -0.16 to -0.04), but they tended to be antagonistic in the low HYSD environments in both breeds (range: 0.02 to 0.42). Therefore, besides yield performance, genetic expression of weighted somatic cell score may also be affected in low opportunity environments.
Between-Environment Analysis
Multivariate analyses revealed important differences in genetic components of variance in differentiated environments. Therefore, estimates from bivariate analyses helped to further evaluate the expected impact of GxE on opportunity loss, or sacrifice, in genetic gain in the most restrictive environments. The resulting components of covariance for all definitions of low and high opportunity Friesian environments with heritabilities and genetic correlations for HYSD environments are in Table 5, and in Raffrenato (2002) for Brown Swiss herds.
|
Sire and residual variance components for milk yield from low HYSD Friesian herds were about half as large as corresponding estimates from the high HYSD environment (Table 5). The genetic correlation (0.63) between these environments differed from unity (P < 0.0001) and was similar to the correlation in first-lactation milk yield between herds in the US and Mexico (Cienfuegos-Rivas et al., 1999), which suggests major re-ranking of sire breeding values. Components of variance for yields of fat and protein showed the same pattern. Genetic correlations between environments were 0.66 for fat and 0.48 for protein, both less than unity (P < 0.0001), also signaling substantial re-ranking in breeding values. Like the multivariate analysis sire and residual variances and the heritability of weighted somatic cell score (0.10) were similar in the contrasted environments.
Incidence of abnormal lactation.
A similar pattern of response in yield traits was found, where genetic variation was less in herds with abnormal (peakless) lactation either in primiparous or pluriparous cows. Genetic variation was also compressed in low opportunity Brown Swiss herds (Raffrenato, 2002). These diminished responses may have been caused by too few nutritional and health inputs and disrupted homeorhetic pathways (physiological orchestration) in the early postpartum period. Genetic correlations between environments were large and positive (range: 0.95 to 0.99), indicating no important effect on the ranking of sire values. Sire variances for weighted somatic cell score were similar with no important change in breeding values indicated by the genetic correlations.
Differential management practices.
As for other definitions of contrasting environments, the sire component of variance for milk yield was also smaller in Friesian herds where nutritional and udder health inputs and practices were used relatively infrequently. This finding is consistent with the result from multivariate analysis for this definition (Table 3). Less genetic variation also occurred in protein yield, but genetic variances were similar for yield of fat. Sire variances for yield traits of Brown Swiss cows also were least in the low input environment (Raffrenato, 2002).
Importantly, the sire variance for weighted somatic cell score in the low opportunity environment (both breeds) was less than in herds relying on more management (e.g., udder health). Consequently, the genetic correlation between environments differed from unity (rg = 0.83, P < 0.05), indicating re-ranking in breeding values. Thus, differentiation of production environments based on our survey of management inputs and environmental clustering was effective in identifying heterogeneous genetic expression in these herds.
Overall, discriminating herd environments by any of these definitions, including the endogenous HYSD variable, revealed a consistent pattern of unequal genetic expression, or GxE, in yield traits. Furthermore, evidence was also found for GxE in weighted somatic cell score in both breeds. Collectively, this information indicates that the GxE occurring in about one-half of the dairy herds in Sicily’s principal dairy region is mainly from compression in genetic expression fostered by input constraints.
Correlated Response and Genetic Trend
Compression in sire components of variance translates into diminished genetic gain from selection in low opportunity environments compared to herds with fewer limitations. The expected correlated responses in these environments for Friesian and Brown Swiss are in Table 6. Estimates generally indicate 20% to 60% less genetic gain in milking performance from indirect selection based on information from more privileged environments.
|
The expected correlated response in somatic cell score for both breeds was similarly depressed about 30% in herds with relatively little milk yield-enhancing management. This outcome may be due to less use of recommended udder health and milking practices. Coefficients for the other environmental definitions exceeded 0.80, except in low HYSD Brown Swiss herds.
Less sire variation in low opportunity environments reduced the expected selection response in both breeds, regardless of the definition. This result is consistent with findings from around the world (Boldman and Freeman, 1990; Carvalheira et al., 1998; Castillo-Juarez et al., 2000; Cienfuegos-Rivas et al., 1999; Costa et al., 2000; Dong and Mao, 1990; De Veer and Van Vleck, 1987; Meinert et al., 1988; Short et al., 1990; Stanton et al., 1991b) that unequal genetic progress and, thus, unequal net economic returns from genetic decisions are functions of identifiable input constraints. Net economic returns are depressed and optimal decisions are altered by diminished daughter milk response and predisposing environmental limitations (Blake, 1992; Holmann et al., 1990). Greater and more rapid genetic improvement in productivity and larger net economic payoffs would be expected by ameliorating the restrictions that foster GxE, including the ones in this study to identify low opportunity herds.
Figure 1 shows genetic trends for milk yield for Friesian (panels a and b) and Brown Swiss (panels c and d) in the contrasting environments defined by incidence of abnormal lactation and the use frequency of management inputs to increase milk yield. Genetic trends in herds with frequent abnormal lactation were inferior to those in environments with normal lactation, undoubtedly due to less expression of genetic variation in these herds. Annual trends with low inputs were mostly smaller than in less constrained environments.
|
Positive genetic trends in alternative environments in both breeds indicated farmers’ attention to sire selection. The differences between environments in annual genetic change for yield traits were consistently positive across criteria, which demonstrate the greater returns from selection that accrue in the more favorable environments. Annual genetic gain in weighted somatic cell score was positive but close to zero in every environmental definition.
|
CONCLUSIONS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSACKNOWLEDGEMENTSREFERENCES |
|---|
These definitions revealed consistent unequal responses in contrasting high and low opportunity environments. Sire and residual variances for all traits were typically smaller in the low opportunity environments for within- and between-environment analyses. Heritability estimates for yield traits and for somatic cell score with these definitions were similar to those from previous reports. In general, genetic correlations were less favorable in the low opportunity environments. Antagonistic genetic correlations between yield traits and somatic cell score may have been attenuated by greater inputs in the high opportunity herds.
Except for the HYSD definition, genetic correlations between contrasted environments were near unity for yields of milk, fat and protein, indicating similar genetic controls. However, as in other studies relying on endogenous definition, compressed variance in breeding values (i.e., scaling effect) rather than change in rank was the key evidence for GxE. The expected correlated responses in yield traits were about 40% less in low opportunity herds when selection was based on information from more privileged environments. Furthermore, small correlated responses in somatic cell score in both breeds were detected in environments with infrequent use of preferred management, including udder health and milking procedure. This is additional evidence for environmental restriction of genetic expression (variation), which signifies less gain from selection regardless of specific selection goals.
Ideally, it would be best to cluster herds using exogenous information, as in this study. Nonetheless, findings clearly showed that the consequences are not severe when using HYSD to differentiate environments. Contrasting environments defined by the incidence of peakless lactation has potential advantages because nutritional and health inputs and homeorhetic mechanisms control or predispose this behavior. An appealing alternative, because it would be easy to implement, would be to depict herd responses as a continuous environmental gradient (reaction norm) of normal and peakless lactation curves using b parameter solutions (or their lower confidence limits) from the incomplete gamma function.
|
ACKNOWLEDGEMENTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSACKNOWLEDGEMENTSREFERENCES |
|---|
Corresponding author:
E. Raffrenato; e-mail:
er53{at}cornell.edu.
Received for publication October 16, 2002. Accepted for publication February 18, 2003.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSACKNOWLEDGEMENTSREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  H. Hammami, B. Rekik, C. Bastin, H. Soyeurt, J. Bormann, J. Stoll, and N. Gengler Environmental sensitivity for milk yield in Luxembourg and Tunisian Holsteins by herd management level J Dairy Sci, September 1, 2009; 92(9): 4604 - 4612. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
H. Hammami, B. Rekik, H. Soyeurt, C. Bastin, J. Stoll, and N. Gengler Genotype x Environment Interaction for Milk Yield in Holsteins Using Luxembourg and Tunisian Populations J Dairy Sci, September 1, 2008; 91(9): 3661 - 3671. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. Vasconcelos, F. Santos, A. Bagnato, and J. Carvalheira Effects of Clustering Herds with Small-Sized Contemporary Groups in Dairy Cattle Genetic Evaluations J Dairy Sci, January 1, 2008; 91(1): 377 - 384. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. M. Shariati, G. Su, P. Madsen, and D. Sorensen Analysis of Milk Production Traits in Early Lactation Using a Reaction Norm Model with Unknown Covariates J Dairy Sci, December 1, 2007; 90(12): 5759 - 5766. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
R. Ruiz-Sanchez, R. W. Blake, H. M. A. Castro-Gamez, F. Sanchez, H. H. Montaldo, and H. Castillo-Juarez Short Communication: Changes in the Association Between Milk Yield and Age at First Calving in Holstein Cows with Herd Environment Level for Milk Yield J Dairy Sci, October 1, 2007; 90(10): 4830 - 4834. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. P. L. Calus, L. L. G. Janss, and R. F. Veerkamp Genotype by Environment Interaction for Somatic Cell Score Across Bulk Milk Somatic Cell Count and Days in Milk J Dairy Sci, December 1, 2006; 89(12): 4846 - 4857. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
W. J. Nauta, R. F. Veerkamp, E. W. Brascamp, and H. Bovenhuis Genotype by environment interaction for milk production traits between organic and conventional dairy cattle production in The Netherlands. J Dairy Sci, July 1, 2006; 89(7): 2729 - 2737. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
E. G. Cienfuegos-Rivas, R. W. Blake, P. A. Oltenacu, and H. Castillo-Juarez Fertility responses of Mexican Holstein cows to US sire selection. J Dairy Sci, July 1, 2006; 89(7): 2755 - 2760. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. J. Windig, M. P. L. Calus, B. Beerda, and R. F. Veerkamp Genetic Correlations Between Milk Production and Health and Fertility Depending on Herd Environment J Dairy Sci, May 1, 2006; 89(5): 1765 - 1775. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S. Konig, G. Dietl, I. Raeder, and H. H. Swalve Genetic Relationships for Dairy Performance Between Large-Scale and Small-Scale Farm Conditions J Dairy Sci, November 1, 2005; 88(11): 4087 - 4096. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
N. Gengler, G. R. Wiggans, and A. Gillon Adjustment for Heterogeneous Covariance due to Herd Milk Yield by Transformation of Test-Day Random Regressions J Dairy Sci, August 1, 2005; 88(8): 2981 - 2990. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. P. L. Calus, J. J. Windig, and R. F. Veerkamp Associations Among Descriptors of Herd Management and Phenotypic and Genetic Levels of Health and Fertility J Dairy Sci, June 1, 2005; 88(6): 2178 - 2189. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. J. Windig, M. P. L. Calus, and R. F. Veerkamp Influence of Herd Environment on Health and Fertility and Their Relationship with Milk Production J Dairy Sci, January 1, 2005; 88(1): 335 - 347. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
N. R. Zwald, K. A. Weigel, and T. J. Lawlor Jr. Genetic Parameters for Conformation Traits in Herds that Differ in Mean Final Score and Completeness of Pedigree and Performance Data J Dairy Sci, January 1, 2005; 88(1): 386 - 391. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. Vasconcelos, A. Martins, M. F. Petim-Batista, J. Colaco, R. W. Blake, and J. Carvalheira Prediction of Daily and Lactation Yields of Milk, Fat, and Protein Using an Autoregressive Repeatability Test Day Model J Dairy Sci, August 1, 2004; 87(8): 2591 - 2598. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
