A kiwifruit in transverse section (Figure 2. 1 7) can be separated into several components tissues, each of which is made up of cell s of distinct and different sizes. Hopping ( ] 976b) described kiwifruit as being made up of three distinct parts; core, inner pericarp and outer pericarp. The central core is made up of a relatively homogeneous population of large parenchyma type cells and is conveniently separated from the inner pericarp by clusters of vascular bundles which are associated with each locule (ventral carpellary vascular bundles (Schmid, 1 978) see chapter three, plate 3 .4). Inner pericarp tissue which is associated with the seeds and locules can be conveniently separated from outer pericarp by a second ring of vascular bundle clusters (dorsal carpeUary vascul ar bundles (Schmid, 1 978), see chapter three, plate 3.2). Inner pericarp cells are distinctly elongated in the radial plane particularly within the locule walls (see chapter three, plate 3 .3 ) . Outer peri carp cell s are composed of a heterogeneous population of two distinct cell types (see chapter three, plate 3 . 1 ). Large ovoid cells are interspersed among small starch containing cell s (Patterson et al ., 1 993) .
In longitudinal section (Figure 2. 1 8), kiwifruit can be separated into the core, inner pericarp and outer pericarp tissue as for transverse sections. However at the proximal and distal end of the fruit, distinction between the tissues becomes less clear, as the tissues lose their differentiation. For longitudinal sections used in this study, the sections were taken from the same point as transverse sections, i.e. at the point of minimum equatorial diameter, approximately half way up the fruit.
2.7.1 Fresh sections
Fresh sections were used in some experimental work due to the speed, ease and low cost of preparation. Fresh sections had the disadvantages of lower quality than wax imbedded sections, restriction to fresh material and inability to be stored for inspection at later dates. Transverse thin sections (approximately 20j.lm thick) of each fruit at the point of
2-42 2: General Materials and Methods. Cell measurements maximum circumference was made using a Cambridge rocking microtome with a CO2 freezing attachment. Freshly cut sections were transferred from the microtome blade to a water bath, floated onto a microscope slide and covered with a glass cover strip. Sections were viewed immediately following preparation.
IP
OP
Core
Figure 2. 1 7 Diagrammatic representation of a transverse slice of a kiwifruit. The box demonstrates the position of a transverse section and the arrows show the transect l ine of cell counts made across the minimum equatorial diameter of the core, inner pericarp (IP) and outer peri carp (OP).
2.7.2 Permanent sections using wax imbedding technique
Permanent wax-imbedded sections were used for most experimental work, as these were able to be prepared and viewed at leisure and stored for re-inspection. Tissue slices for transverse or longitudinal sectioning were taken at the point of maximum diameter and were fixed in ethanol (60%)/formaldehyde/acetic acid 20: 1 : 1 v:v:v. (Hopping, 1 976b) prior to embedding. Tissue was wax imbedded in a Shanton Citadel 2000 programmable automatic tissue processor. Imbedding took place over a 1 6 hour period and involved a
2: General Materials and Methods. Cell measurements 2-43 progressive ethanol series up to 1 00% ethanol, two xylene baths and three molten paraffin wax baths. Air was removed from embedded slices by placing them in a vacuum oven for 5 hours at 60°C . Thin sections (7j.lm) of cold wax imbedded tissue were made with a Heitz ] 5 1 2 rocking microtome. Sections were floated onto slides in a water bath and fixed to slides at 60°C overnight. Sections were de-waxed in xylene, rehydrated through an ethanol-water series, stained with Gills haemotoxylin and then rinsed in ' Scotts' tap-water. Finally sections were dehydrated through an alcohol series, cleared in xylene and permanently mounted in DPX.
2.7.3 Histology
Sections were viewed under an Olympus compound microscope fitted with an eye-piece micrometer.
In transverse sections, the number of cells in the core, inner pericarp and outer pericarp were counted along a straight line transecting the outer and inner pericarp on one side of the fruit and the entire diameter of the core (Figure 2. 1 7). The mean cell diameter in each tissue was obtained by dividing the width of each tissue by the total number of cells counted across it. Core cell numbers presented were halved to represent a core radius. In order to determine the possible influences of the two types of cell present in outer pericarp tissue, the number and mean width of outer pericarp 'large' and 'small' cells were counted across the radial line. Estimation of large cell number and width was achieved by counting the total number and the width of each large cell along the radial transect. By subtracting these two measures from total cell number and width, mean number and width of small cells could also be estimated. Large and small cells were not determined in fresh sections.
In longitudinal sections, cell lengths of cells were determined from four positions within the fruit at the point of minimum equatorial diameter (Figure 2. 1 8).
2-44 2: General Materials and Methods. Cell measurements
1 . The total length was measured (longitudinally) of 1 00 cells in a straight line at the mid-point of the core.
2. The total length was measured (longitudinally) of 50 cells parallel to the epidermis at the mid-point of the outer pericarp. Cells were counted as large or small cells. Counts were made at two randomly selected points in each fruit.
3. Average longitudinal cell length was estimated at a point 1 0 cells inside the epidermis. Extended counts were not found to be possible, so cell lengths were instead estimated by counting the number of cells in ] mm parallel to the epidermis at four randomly selected points.
4. Average longitudinal cell length was estimated just inside the outer edge of the inner pericarp as in 3 . • • • • " • • " " • • • • " ,
i
• • • •\
I • • 'I II •I
•I
•I,
... ... ... .. I I • • • • I I I • • • •i
\,
• • • • • • • • \ • • • • �.,. •OP
.-.�.,Core··· ....
I
Figure 2. ] 8 Diagrammatic representation of a longitudinal slice of a kiwifruit. The boxI
demonstrates the position of a longitudinal section and the arrows show the direction andI
location of longitudinal cell counts made in the core, inner pericarp (rp) and outerI
2 : General Materials and Methods. Statistical A nalysis
2.8 Data Handling and Statistical Analysis
2.8.1 Data handling
2-45
Where possible, data was collected directly onto computer e.g. using a portable laptop computer with input directly from a balance or digital callipers. This reduced the occurrence of transcription errors for such data. Datasets were checked for recording and transcription errors using three separate methods:
1 . Each variable collected in a dataset was formatted to a fixed number of decimal places. Datasets were sorted by treatments and a visual scan was made of data in columns to identify any possible unusual data-points, which were manually checked for transcription errors from data-sheets.
2. Datasets were sorted individually in ascending sequence by each variable collected and by meaningful ratios such as percentage dry weight. Observations at the lowest or highest end were manually checked. Impossible or unlikely data such as kiwifruit weighing 600g were deleted from datasets. Scatterplots were often made of meaningful pairs of variables e.g. seed number and fruit weight and outliers were able to be identified and checked.
3 . Inspection of residuals plots was made for each data set and outlying points were identified and checked. Identification of outliers was easily made using the interactive SAS Insight module.
2.8.2 Experimental design and analysis of variance
Most experimental designs utilised were randomised complete block designs. Individual kiwifruit vines or rows of vines were used as blocks, to remove random variation due to the position or prior history of the vines. In most cases, a lateral shoot was the minimum experimental unit to which treatments were applied. As there were normally two or more fruit present on a lateral shoot, the individual variables collected for each fruit on a shoot were always averaged because of the non-random selection of such fruit which could
2-46 2: General Materials and Methods. Statistical Analysis
otherwise lead to an incorrect estimation of the random variation for analysis of variance (Mead et aI., 1 993).
Analysis of variance was undertaken using the general linear models procedure (proc GLM) in the SAS system for statistical analysis available as SAS for windows release 6. 1 0. Where data required transformation in order to conform to assumptions underlying ANOYA (Cochran, 1 947), the method of Box and Cox ( 1 964) summarised by Fernandez ( 1 992) was used to estimate an appropriate transformation. For presentation, transformed means were back transformed to the original scale.
2.8.3 Multiple comparison testing
The ANOY A test statistic F was used to test model hypothesis that all treatment means were the same. If this hypothesis was rejected due to a significantly high value of F, an alternative hypothesis was accepted that at least one of the treatment population means differed from the rest. In order to compare treatment means, three different types of comparison were used.
1 . Contrasts (SAS proc GLM , contrast statement) were used for pre-planned comparisons of individual means and to provide estimates of the probability that a pair of means are different for specific comparisons of interest. In addition to individual comparisons, this procedure also allowed comparison of groups of treatments with other groups, for example control treatment versus all other treatments. Estimates of the value of a combined treatment mean could be obtained using the estimate statement in proc GLM. A further use of contrasts was made in testing for polynomial trends between treatment levels with an orthogonal structure.
2. Fishers protected least significant difference (LSD) procedure was used for
multiple comparisons between treatments from balanced datasets. The LSD procedure controls the comparison wise type I error (falsely accepting the hypothesis that a pair of means are not different), but tends to inflate the overall or experimentwise error rate (falsely rejecting the hypothesis that a pair of means are not different) as the number of
Chapter 2: General Materials and Methods. Statistical Analysis 2-47 comparisons increases (Ott, 1 993). However if Fishers LSD is applied only after the F test for treatment differences has been shown to be significant, it becomes a protected LSD and the experimentwise error rate is weakly controlled at a level approximately equal to the a-level of the F-test (Ott, 1 993). Fishers protected LSD has been observed to have the highest 'correct decision rate' in simulation studies made with ten commonly used multiple comparison procedures (Carmer and Swanson, 1 973).
3. Where missing observations occurred, least-squares means otherwise described as population marginal means (PMM) (Searle et aI., 1 980) were used, as these provide a better estimate of the true population mean in the presence of missing values. Arithmetic means of data separated into cells (blocks, or treatments) with missing values give mis leading results, as means are weighted towards cells with fewer missing data. PMM estimates an average of the population cell means for all cells using parameters of a linear model (Searle et al., 1 980). The GLM procedure in SAS computes PMM using the lsmeans statement and can also provide a significance test that individual PMM's are different using a modified t-distribution.
2.8.4 Comparison of correlation coefficients for fitted lines
A comparison of correlation coefficients can be made by transforming the correlation coefficient (r) into a transformed value (zr) (Edwards, 1 976):
1
Zr = [Loge(1 + r) - log e( 1 -r)l
2 .
where r = correl ation coefficient, n = sample size
The distribution of Zr is approximately normal and has a standard error (jzr which is related simply to the sample size. To test the significance of the difference between two values of r, a Z test statistic of the difference between the two Zr values can be obtained: (j (zr l
Zrl - Zr2
Z = ---
2-48 2: General Materials and Methods.
2.9 References
Badenoch-Jones. J . , Parker, C.W. and Letham, D.S. ( 1 987). Use of isopentenyladenosine and dihydrozeatin riboside antibodies for the quantification of cytokinins. Journal of Plant Growth Regulation, 6, 1 59- 1 82.
B ieleski, R.L. ( 1 964). The problem of halting enzyme action when extracting plant tissues. Analytical Biochemistry, 9, 43 1 -442.
Box, G.E.P. and Cox, D.R. ( 1 964). An analysis of transformations. Journal (){ the Royal Statistical Society Series B, 26, 2 1 1 -243 .
C armer, S.G. and Swanson, M .R. ( 1 973). A n evaluation of ten pairwise comparison procedures by Monte Carlo methods. journal of the American Statistical Association, 68, 66-74.
Chan, D.W. ( 1 987). General principles of immunoasay. in: C han, D.W. and Perlstein, M .T. (editors) Immunoassay: A practical guide. Academic press, San Deigo, USA. pp. 1 -23.
Chard, T. ( 1 990). Radioimmunoassay and related techniques. Elsevior science publishers B V . , Amsterdam, The Netherlands.
Cochran, W.G. ( 1 947). Some consequences when the assumptions for the analysis of variance are not satisfied. Biometrics, 3, 22-38.
Dotti, C. and Castagnetti, C. ( 1 978). Non-specific count subtraction in radioimmunoassay (a criticism). Ricera in Clinica e in Laboriato, 8, 30 1 -3 1 1 .
Edwards, A.E. ( 1 976). An introduciton to linear regression and correlation. W.H. Freeman and Co., S an Fransisco.
Erlanger, B .F. and Beiser, S.M. ( 1 964). Antibodies specific for ribonucleosides and ribonucleotides and their reaction with DNA. Proceedings of the National A cademy of Sciences of the United States (�f America, 52, 68-74.
Ernst, D. ( 1 986). Radioimmunoassay and gas chromatography! mass spectrometry for cytokinin determination. in: Linskens, H.F. and Jackson, J.F. (editors) Immunology in plant sciences. Springer-Verlag, Berlin. pp. 1 8-49.
Feldkamp, C.S. and Smith, S.W. ( 1 987). Practical guide to immunoassay method evaluation. in : Chan , D . W . and Perlstein, M.T. (editors) Immunoassay: a practical guide. Academic press., San Diego, USA. pp. 49-95.
Fernandez, G.c. ( 1 992). Residual analysis and data transformations: Important tools In statistical analysis. Hortscience, 27, 297-300.
Galfre, G. and Butcher, G.W. ( 1 986). Making antibodies. in: Wang, T.L. (Editor) Immunology in plant science. Cambridge University Press., Cambridge, England. pp. 1 -25.
Green, A.E., McAneney, K J . and Astil l , M .S. ( 1 990). An instrument for measuring kiwifruit size. New Zealand journal of Crop and Horticultural Science, 18, 1 1 5 - 1 20 .
Hopping, M.E. ( 1 976a). Effect o f exogenous auxins, gibberellins and cytokinins o n fruit development in chinese gooseberry (Actinidia chinensis Planch.). New Zealand journal (){ Botany, 14, 69-75. Hopping, M.E. ( 1 976b). Structure and development of fruit and seeds in chinese gooseberry (Actinidia
2: General Materials and Methods. References 2-49 Jones, H.G. ( 1 987). Correction for non-specific interference in competitive immunoassays. Physiologia
Plantarum, 70, 1 46- 1 54.
Lai, R. ( 1 987) Leaf - fruit relationships in kiwifruit (Actinidia deliciosa (Achev) (F.Liang et AR. Ferguson) Phd Thesis. Massey University, Palmerston North, New Zealand.
Lai, R., Woolley, DJ. and Lawes, G.S. ( 1 990). The effect of inter-fruit competition, type of fruiting lateral and time of anthesis on the fruit growth of kiwifruit (Actinidia deliciosa). Journal of Horticultural Science, 65, 87-96.
Lewis, D.H. ( 1 994) Cytokinins and fruit growth III kiwifruit. PhD Thesis. University of Otago., Dunedin, New Zealand.
Lewis, D.H., Burge, G.K., Schmierer, D .M . and Jameson, PE ( 1 996). Cytokinins and fruit development in the kiwifruit (Actinidia deliciosa) . 1. Changes during fruit development. Physiologia Plantarum, 98, 1 79- 1 86.
MacDonald, E.M.S., Akiyoshi , D.E. and Morris, R.O. ( 1 98 1 ). Combined HPLC - radioimmunoassay for cytokinins. Journal oj Chromatography, 214, 1 0 1 - 1 09 .
MacDonald, E.M.S. and Morris, R . O . ( 1 985). Isolation o f cytokinins b y immunoaffinity chromatography and analysis by high-performance l iquid chromatography and radioimmunoassay. Methods in Enzymology, 1 10, 347-358.
Mead, R., Curnow, R.N. and Hasted, A.M. ( 1 993). Statistical methods in agriculture and expelimental biology. Chapman and Hall, London.
Mertens, R., Deus-Neumann, B. and Weil er, E.W. ( 1 983). Monoclonal antibodies for the detection and quantification of the endogenous plant growth regulator, abscisic acid. FEBS Letters. , 1 60 , 269- 272.
Morris, R.O. ( 1 986). Analysis of eytokinins by immunological methods. in: Yopp, J .H., Aung, L.H. and Steffens, G.L. (editors) Bioassays and other special techniques for plant hannones and plant g rowth regulators. Plant growth regulator society of America, pp. 1 23-207.
Neuman , D . S . and Smit, B .A . ( 1 990) . Interference from xylem sap in an enzyme-linked immunosorbent assay for zeatin riboside . Physiologia Plantarum, 78, 548-553.
Norman, S .M . , Pol ing, S.M. and Maier, V.P. ( 1 986). An indirect enzyme-linked immunosorbent assay for (+ )-abscisic acid in plant tissue and Cereospora rosicola cultures. Abstract no, 72. Symposium on Immunological Techniques, A CS national meeting Anahein, CA.
Ott, R.L. ( 1 993). An introduction to statistical methods and data analysis. Duxbury Press, B elmont, Calif.
Patterson, K.J., Mason, K.A. and Gould, K.S. ( 1 993). Effects of CPPU (N-(2-chloro-4-pyridyl)-N' phenylurea) on fruit growth, maturity and storage quality of kiwifruit. New Zealand Journal ()f Crop and Horticultural Science, 2 1 , 253-26 1 .
Pengelly, W.L. ( 1 986). Validation of immunoassays. in: Bopp, M. (editor) Plant growth substances 1 985. Springer-Verlag, Berlin Heidelberg. pp. 3 5-43.
Perlstein, M .T. ( 1 987). Immunoassays: Qual ity control and troubleshooting. in: Chan , D.W. and Perlstein, M .T. (editors) Immunoassay: A practical guide. Academic press, San Deigo, USA. pp.
1 49- 1 63.
Scatchard, G. ( 1 949). The attraction of proteins for small molecules and ions. Annals ()f the New York academy of science, 5 1 , 660-672 .
2-50 Chapter 2: General Materials and Methods. References Schmid, R. ( 1 978). Reproductive anatomy of Actin idia chinensis (Actinidiacease). Botanische
jahrbucher fur systematik, pjlanzengeschichte und pjlanzengeo, 100, 1 49- 1 95 .
Searle, S.R., Speed, F.M. and Millikan, G .A. ( 1 980). Population marginal means in the linear model : A n alternative to least squares means. The American Statistician, 34, 2 1 6-22 1 .
Snelgar, W.P., Martin, P.1. and Manson, P.J. ( 1 992). Influence of shelterbelts on pollination of kiwifruit. Acta Horticulturae, 297, 263-268.
Technicon ( 1 973). Industrial method no 98/70w. Technicon, Tarrytown, New York.
Twine, J.R. and Williams, C.H. ( 1 97 1 ). The determination of phosphorus in kjeldahl digests of plant material by automatic analysis. Commercial soil science and plant analysis, 2, 485-489.
Walker-Simmons, M. ( 1 987). ABA levels and sensitivity in developing wheat embryos of sprouting resistant and susceptible cultivars. Plant Physiology, 84, 6 1 -66.
Weiler, E.W. ( 1 980). Radioimmunoassay for trans-Zeatin and related cytokinins. Planta, 1 49, 1 55- 1 62. Weiler, E.W. ( 1 980). Radioimmunoassays for the differential and direct analysis of free and conjugated
abscisic acid in plant extracts. Planta, 148, 262-272.
Weiler, E.W. ( 1 986). Plant hormone immunoassays based on monoclonal and polyc\onal antibodies. in : Linskens, H.F. and Jackson, J.F. (editors) Immunology in plant sciences. Springer-Verlag, berlin Heidelberg. pp. 1 - 1 7 .
Yalow, R.S. and Berson, S .A. ( 1 960). Immunoassay of endogenous plasma insulin i n man. Journal of clinical investigation, 39, 1 1 57- 1 1 75.
3 : Lateral Shoot of Kiwifruit. Introduction 3 - 1