Terry Stanger1, Bill Dvorak2 and Gary Hodge2

Companies and addresses

1Shaw Research Centre, Sappi Forests (Pty) Ltd., P. O. Box 473, Howick, 3290, South Africa
2CAMCORE Cooperative. North Carolina State University, Raleigh, U.S.A.



P. patula, genetic control, basic wood density, densitometry, non-destructive sampling


Pinus patula is a closed-cone pine that has a rather narrow, but long, distribution in Mexico. It ranges from approximately 16oN to 24oN latitude, is generally restricted to humid, sub tropical to near temperate sites, with deep, fertile clay soils, and an estimated mean annual precipitation of 1000 to 2500 mm. Approximately one million hectares of P. patula plantations have been established in the tropics and sub tropics for saw-timber and paper products. The majority of the P. patula forests are located in southern Africa with South Africa having more than 300,000 hectares under operational management. Since 1986 the CAMCORE Cooperative has sampled 25 provenances and 624 mother trees of P. patula in Mexico. The CAMCORE collections represent the most complete coverage of the species natural distribution to date. During December 1990, a series of five trials of open-pollinated P. patula family/provenance seedlots were established adjacent to each other at Maxwell in KwaZulu-Natal, South Africa. These trials are 11 years old and offered a unique opportunity to sample material from the entire geographic range of the species to determine to what extent wood properties are under genetic control at the provenance, family and individual tree level. Nine hundred and seventy-two individual trees were sampled non-destructively by removing 12 mm increment cores for the evaluation of wood anatomical properties using densitometry and image analysis. Preliminary results from wood density assessments indicate that provenance differences are significant and that the trait is under strong genetic control at the family level.


One of the disadvantages of wood in relation to other raw materials used in manufacturing industries is its great variability. However from a tree breeding perspective this provides many opportunities for improvement. In general, genetic studies have shown that most wood physical properties have high heritabilities. This suggests that selection and breeding for these properties should be successful. Zobel and Jett (1) state that wood improvement is most needed for pines that are grown as exotics in the tropics and subtropics. The advantages of exotics are rapid growth and a young harvestable age, but generally they have poor form and a high proportion of juvenile wood. Much of the recent expansion in plantation forestry is occurring in the tropics and subtropics, a number of new species are being tested such as P. greggii, P. maximinoi, and P. tecunumanii. In most cases very little or nothing is known about the wood properties of these new species when they are grown as exotics in plantations.

Pinus patula is a closed-cone pine that has a rather narrow but long distribution in Mexico. It is found from approximately 16oN to 24oN latitude on humid, sub tropical to near temperate sites in areas with deep, fertile clay soils. Mean annual precipitation amounts range from 1000 to 2500 mm (2), with additional moisture being received throughout the year in the form of heavy mists, clouds and fog. Although this pine has been found growing at elevations ranging from 1500 m to 3100 m, most of the principal stands of its central distribution occur from 2100 m to 2800 m altitude. The species can withstand heavy frosts (-10°C) and dry periods of 4 to 5 months but grows much better under warm, humid conditions. Within its native range, it attains heights of 30 to 35 m with diameters of 50 to 90 cm (2).

Approximately one million hectares of P. patula plantations have been established in the tropics and sub tropics for saw-timber and paper products. The majority of the P. patula forests are located in southern Africa with South Africa having more than 300,000 hectares under operational management. Pinus patula was first introduced into South Africa in 1907 (3). It could be said that tree breeding programs really started in South Africa as early as 1953 when a progeny trial comprising 125 open-pollinated families, collected from various plantations in the Mpumalanga province, was established at Border plantation. However it was with the inception of the "D.R. de Wet Forest Research Station" near Sabie, that tree improvement started in earnest. Currently the two major players (Mondi and Sappi) in the Forestry Industry in South Africa have their own P. patula breeding programs but neither routinely include wood properties in their selection criteria.

Genetic improvement of wood quality in young, fast-grown tropical plantations may provide a solution to the high levels of juvenile wood in these plantations. Most tree improvement programs traditionally include adaptability, growth, stem form, disease and pest resistance in their assessments but very few programs routinely include wood properties. Historically very few studies were specifically designed to assess the genetic control of wood properties. However by the mid 1960's it was generally accepted that the genetic control of wood properties was of significant magnitude to be included in breeding programs (1). To be used routinely in a tree breeding program, any pulpwood assessment strategy for wood properties must be convenient, cheap and rapid, and must be able to cope with samples from thousands of trees. Evans et al (4) suggest that for routine assessment of wood properties to be effective, the following criteria must be met:

1) sampling is non-destructive,

2) sample size is small,

3) sampling rate is high,

4) measurement rate is high and repeatable,

5) sample properties reflect those of the whole resource and,

6) resource properties control product properties.

With advances in technology, the first four criteria can be accomplished, and published evidence suggests that the latter two are acceptable (4, 5, 6). It is well recognized that patterns of wood density development vary considerably between trees. However, relating the density measured at limited points in the tree to the average density of the whole tree offers the possibility of rapid assessment of wood density from cores taken at a convenient single sampling point.  This in particular has application in screening selected trees in a tree breeding program where relative values for ranking trees and not absolute estimates of the whole tree density are important. It is almost certain that a tree that exhibits high density wood in the first 4 to 6 years of growth will produce better than average density wood at rotation age (7). Ladrach (5) developed a regression model to predict whole tree specific gravity for P. patula in Colombia from specific gravity determined at breast height. The age of the trees varied from four to 23 years and the R2 for the model based on a sample size of 75 trees was 0.93. Ringo and Klem (8) have also shown that single increment cores extracted at breast height from 25 year old P. patula grown in Tanzania can be used to estimate whole tree basic density. Their results showed that wood density determined using increment cores underestimate whole tree values by about 4%.

In this study, gamma-ray densitometry and image analysis were used to measure basic wood density, tracheid length and cross sectional dimensions on 12 mm increment core samples taken from 11-year-old trees in field trials at Maxwell in KwaZulu-Natal, South Africa. Only the results from the gamma-ray densitometry are reported on in this paper.


Wood density is the term used to express how much wood substance is present in a given volume of wood. It is usually expressed as the ratio of the oven dry weight of wood to its green volume and it is measured in units such as kg/m3 or g/cm3. Wood density is one of the cheapest and easiest wood properties to measure and it can be assessed non-destructively by removing increment cores from a tree. It is a complex characteristic that is interrelated to the proportion of latewood, wall thickness and cell size. Each of these components has an inheritance pattern of its own, but density is easier to measure than the individual components. Studies have shown that the inheritance of wood density is strong in pines (9, 10, 11, 12). Most of the genetic variability is of the additive type, enabling good gains from selection and breeding. Because of its major effect on paper, solid wood products and energy programmes, the genetics of wood density has been studied more than any other wood property (1).  The heritabilities of wood density are generally higher than those for stem form or growth traits, an indication that genetic manipulation of wood density can result in good gains. Published heritabilities for density in pines and hardwoods vary from 0.40 to 0.80 compared to the usual range of 0.15 to 0.25 for many growth traits (1, 13).

One of the biggest sources of variation in wood density in pines is the ratio of earlywood to latewood. The wood characteristics in earlywood and latewood are usually very different.  Earlywood has low density and thin tracheid walls (see figure 1) and it can easily be recognized as the light brown wider rings in the inner part of a growth ring. Latewood has high density with characteristically thick tracheid walls (see figure 1), which make it easy to recognize as the dark brown ring of cells in the outer part of the growth ring. In juvenile wood of pine, the thicker the cell walls the darker the appearance of the latewood band.

Figure 1

Figure 1. Comparison of typical P. patula latewood (on the left) with thick tracheid walls and earlywood (on the right) with thin tracheid walls (x 10 magnification)

The ratio of latewood to earlywood is important because it has a large influence on wood density. The genetic control of the wood density components, which consist of earlywood density, latewood density and latewood percent, are all strong (1). None of these however has so far been found to have a higher heritability than overall density and these components have had limited value in improving selection efficiency for overall density (1).

Altering wood density through genetic manipulation is not simple or easy. Density can be changed by growing trees in different environments, but studies have shown that genotype by environment interaction or rank changes among families, are not significant (11). These reactions (or interactions) frequently occur when exotic species are moved into areas that are different to their indigenous environment. It is not correct to assume that a species, or a provenance, will have the same wood properties in the new environment that it had in its native habitat. It is important that new species and provenances are evaluated as early as possible for wood properties, to determine if their wood is acceptable commercially.

Most wood density studies in pines have shown that wood density has a direct effect on tracheid yield and paper strength, and on strength and utility of solid wood products (1). Van Buijtenen (14) found that increased wood density and latewood percent results in paper with greater tear strength but in decreased tensile and bursting strengths. Similarly, a large proportion of low density juvenile wood results in low tear and high burst and tensile strength. An important aspect of increased pulp yield associated with the improvement in wood density is that pulp output of a mill can be increased without changing wood consumption at the mill. Blair et al. (6) have published predictions of gain in pulp yield and tear strength in young loblolly pine through genetic increases in wood density. They found that yield increases from breeding for high wood density could be in the order of 10 to 15% of the mean pulp yield, and tear strength could be increased by about 10% of the mean. Most publications emphasize breeding for high density wood. However, this is only needed for some products, others are best made from low density wood . It is important to always keep the end product in mind when defining breeding objectives and selection criteria.

Most studies have shown that genetic variation among provenances is negligible, although some studies have shown that there is a slight association between latitude of seed origin and wood density, as well as percent latewood, both of which increased from northern to southern origins (15). Exploitation of provenance variation in wood density could be crucial for the rapid attainment of a target value for the trait (16), and it is important to determine what variation exists at the provenance level when material is grown in an exotic environment. Very little published information is available on the genetic variation in wood density among individual trees and families for P. patula. Of the 58 references provided by Zobel and van Buijtenen (15) on the genetic control of wood density in pines, only one refers to P. patula.


a) Genetic Material
Since 1986 the CAMCORE Cooperative has sampled 25 provenances and 624 mother trees of P. patula in Mexico (17). The CAMCORE collections represent the most complete coverage of the species natural distribution to date. The Cooperative has established 93 provenance/progeny trials in Brazil, Chile, Colombia, Mexico, South Africa and Zimbabwe, as well as a number of ex situ conservation plantings.  In December 1990, a series of five CAMCORE trials of open-pollinated P. patula family/provenance seedlots (see table 1) were established by Sappi adjacent to each other at Maxwell in KwaZulu Natal, South Africa. Maxwell can be described as an ideal P. patula site, at an elevation of 1350 m with a mean annual temperature of 16oC. This trial series was also established in additional locations by other members of the CAMCORE Cooperative in South Africa and overseas.

Table 1. Origin and number of families of the Mexican P. patula provenances represented in the CAMCORE trials at Maxwell

Map Key 1


State or Department

Latitude   (N)

Longitude (W)

Elevation  Range (m)

Rainfall (mm/yr)

No. of Families


Potrero de Monroy


20°  24'

98°  25'

2320 – 2480




Ingenio del Rosario


19°  31'

97°  06'

2770 – 2870






18°  38'

97°  06'

2000 – 2230




Conrado Castillo


23°  56'

99°  28'

1500 – 2060






19°  40'

98°  05'

2750 – 2915




Pinal de Amoles


21°  07'

99°  41'

2380 – 2550






20°  39'

98°  40'

1980 – 2200




Llano de las Carmonas


19°  48'

97°  54'

2530 – 2880




El Manzanal


16°  06'

96°  33'

2350 – 2660




El Tlacuache


16°  44'

97°  09'

2300 – 2620






17°  24'

96°  27'

2600 – 2870




Santa María Papalo


17°  49'

96°  48'

2270 – 2720






17°  10'

96°  21'

2450 – 2770



1 Key to provenance locations in Figure 1.

The standard CAMCORE design, a randomized complete block design with 9 replications and 6-tree row plots established at a 3.0 x 3.0 m espacement, was used for all trials. Four local P. patula seedlots (M2999, M2997, M3404 and 28222) and P. elliottii were included as controls. The P. patula control lots represent material from breeding programs in South Africa at various levels of improvement. The trials were 11 years old and offered an unique opportunity to sample material from the entire P. patula geographic range (see figure 2) grown on one site in South Africa, to determine the extent that wood properties are under genetic control at the provenance, family and individual tree level.

Figure 2

Figure 2. Natural distribution of P. patula identified by CAMCORE in Mexico

b) Sampling Procedure
Twelve of the thirteen provenances planted at Maxwell were selected for the study (see figure 2). The Cuajimoloyas provenance was excluded, because it was represented by too few families. The provenances included in this study span the complete known north to south distribution of P. patula in its natural range. Wood properties often have only a weak or no genetic correlation with growth and stem form traits (1). Many papers covering this topic, and summarized by Zobel and van Buijtenen (15), report little or no relationship between growth rate and wood properties. Although half-sib families and individual trees were selected on the basis of their growth for this study, it is assumed that for purposes of the genetic analysis of the wood properties, that this is a random sample from the population.

The nine best families were selected from each of the twelve provenances using eight-year growth data. The best tree from each row plot in each of the nine replications in the trial, in which the family was being sampled, was chosen for initial, non-destructive sampling. The genetic structure of the material in this study was therefore made up of 12 provenances with nine families per provenance and nine trees per family, or 972 trees, excluding controls. Twenty-six trees from the local South African control lot M2999 were also sampled to provide a baseline representing local landrace P. patula for comparison. M2999 is a 2nd generation bulk collection from a Mondi clonal seed orchard. The total sample size of the study was 998 trees. Where possible, forked and leaning trees were avoided because of the concern that they might have had reaction wood.

Trees were sampled during August and September 2001. Twelve millimeter increment cores were extracted using a Trecor™ HW300 corer. Some authors have reported that neither wood density nor tracheid lengths vary significantly with the compass direction from which samples are collected (18). To maintain a consistent approach in this study, one bark to bark 12 mm increment core was taken in an east-west orientation at 1.2 m above ground from each tree for the determination of wood properties. In some cases where there was no option other than to sample a leaning tree, the sample was taken at 90o to the lean in an attempt to avoid reaction wood. Cores were air dried at the Shaw Research Centre and then transported to the Council for Scientific and Industrial Research (CSIR), Forests and Forest Products laboratories in Durban, South Africa for measurement of wood properties. Each core was separated into two halves, densitometry and tracheid length analysis was carried out on a randomly selected half or radial (pith to bark) core. The second half of the core was retained for the determination of cellulose, lignin and polysaccharide yield using near infrared (NIR) techniques. All analyses were performed on un-extracted cores.

Samples for the densitometer were prepared using a specially developed electric saw with a tungsten carbide tooth sawblade. The 12 mm core was clamped into a moving platform with the tracheid's orientated perpendicularly, as they would be in a standing tree. Each core was machined to produce a 12 mm deep x 2 mm wide section from the centre of the core. The outer portions from the core were retained for tracheid length analysis.

c) Density Measurement
Wood density was determined using 60KeV collimated soft gamma radiation from an Americum241 energy source. Malan and Marais (19) showed that the linear attenuation coefficient of wood correlates well with its gravimetric density. Gamma ray densitometry therefore provides a quick and highly reliable method of assessing wood density. In this study, a fully computer controlled data acquisition system and specimen holder with a stepper motor drive, allowing precise 0.5 mm incremental movement of the sample and a individual scanning field of 0.5 mm, was used. For further details on the equipment readers are referred to the paper by Malan and Marais (19). In this paper, wood density is expressed in g/cm3. Weighted mean density (WMD) was calculated for each core by weighting the density for each scanning field by its area as a proportion of the total core radius.

d) Statistical Analysis
The statistical analysis was carried out using Proc GLM in SAS ver 8.0. The following model was used to calculate provenance means.

yijk = µ + repi + provj + (rep*prov)ij + ,ijk……………………………………….(1)


yijk = weighted mean density of the kth tree in the ith rep and jth provenance

µ  = overall mean

rep = ith rep effect, i = 1,…9

provj = jth provenance effect j = 1,…12

(rep*prov)ij = interaction between the ith rep and jth provenance

,ijkl = random error associated with ith rep, jth provenance and kth tree where ,ijk ~ iid (0,F2)

To calculate family within provenance means and variance components the model was modified to include a family within provenance term.

yijk = µ + repi + provj + (rep*prov)ij + fam(prov)jk ,ijk……………………….(2)

where µ  = overall mean

repi = ith rep, I = 1,…9

provj = jth provenance effect, j = 1,…12

(rep*prov)ij = interaction between the ith rep and jth provenance

fam(prov)jk = kth family within jth provenance effect.

,ijk  = random error associated with ith rep, jth provenance , kth family within jth provenance.

Using the same model variance components were determined using Proc VARCOMP in SAS and the following parameters were calculated:


Results from the analysis have shown that differences between provenances were highly significant (P<0.0001, see appendix 1 for analysis of variance table). Residuals were plotted against fitted values, and showed no detectable trends or patterns. It can therefore be said that the conditions that ,ijk ~ iid (0,F2) have been met for this data and the standard ANOVA assumptions, are valid.

The most northern provenance Conrado Castillo [8], (see figure 1) had the highest area weighted mean density (see figure 2a). Ingenio del Rosario [2] ranked second and three provenances, El Manzanal [4], El Tlacuache [5], and Ixtlán [6] from the southern extreme of P. patula's distribution were ranked third, fourth and fifth respectively. These latter three populations represent P. patula var. longipedunculata not P. patula var. patula. Based on the results from this study it appears that P. patula populations that occur in the Sierra Madre del Sur in southern Mexico have higher density than those that occur in the Sierra Madre Oriental in eastern Mexico. It also suggests that var. longipedunculata has higher wood density than var. patula. Genetic parameters were calculated (see table 2). Provenance variance (P2b) expressed as a proportion of the total phenotypic variance accounted for 4.7% of the total phenotypic variance. These provenance effects are small. Results from the analysis of growth data collected from the same trials, have shown that at eight years of age the local South African seed orchard seedlots and the Potrero de Monroy and Corralitla provenances, were the best for volume production (CAMCORE, unpublished results).  Based on these and earlier assessments, 142 selections were made throughout the trials in South Africa. Forty-five selections were made in families from Potrero de Monroy and thirty-two in families from Corralitla. Selections from these two provenances therefore account for 54% of all selections made in the trial series. Wood density is an important determinant of total fibre yield.  There is cause for some concern that more than fifty percent of the selections based on volume production made for the next generation of breeding come from two provenances with below average values for wood density. These results demonstrate how important are the routine assessments of wood properties in tree breeding programmes. In the absence of this information, a tree breeder could unwittingly be selecting against wood density. This aspect will be investigated in more detail at the family level. Pith to bark profiles were generated for all samples (see figure 3).

Figure 2a

Figure 2a. Provenance area weighted mean density (WMD) at 11 years

Table 2. Variance component and parameter estimates for model 2

Table 2

Figure 3

Figure 3. Typical pith to bark density profiles from three provenances representing extremes of the density range in this study

For example in figure 3, sample 864 has noticeably lower density earlywood bands, all have a density close to 0.350. On the other hand, the earlywood bands in samples 53 and 20 all have densities close to or greater than 0.400. Based on preliminary results, one reason for the provenance variation in wood density may be the proportion of earlywood and its mean density. This is under further investigation.

Differences between families within provenances were also highly significant (P<0.0001, see appendix 1 for analysis of variance table). Family 238 from Conrado Castillo had the highest density (0.488) and family 150 from Ixtlán the lowest density (0.409) (see  table 3). The range in family mean density in this study was 0.079 or 19%. Detailed genetic parameters were calculated (see table 2). The family within provenance individual heritability (h2b) was 0.32. This is considerably higher than typical heritability estimates for growth that are in the range 0.15 to 0.20. A family within provenance individual heritability (h2b) of 0.20 for volume at age eight was reported for this series of South African tests (17). Similarly the family heritability (h2f(b)) was 0.44. These estimates demonstrate that the species exhibits appreciable levels of additive variance for wood density, and potential gains from breeding are likely to be higher if wood density is also included in selection indices. A genetic coefficient of variation was calculated to express the genetic standard deviation as a fraction of the mean. Over cycles of selection, this can be used as measure to determine the magnitude of genetic variation for selection. For this population the GCV was 4.6% and should be used to compare this population with future populations that have been selected for increased wood density.

Table 3. Area weighted mean densities for the top and bottom 10 families

Top 10 Families

Bottom 10 Families















































Trial Mean





This study represents the first detailed evaluation of the wood properties from a comprehensive range-wide collection of P. patula germplasm from its native habitat. The results have shown that differences between provenances are significant. It highlights the fact that in the absence of wood density data selections have been concentrated in two provenances. Both Potrero de Monroy and Corralitla have lower wood density than average. Based on preliminary results, one reason for the provenance variation in wood density may be the proportion of earlywood to latewood and its mean density. This is under further investigation. The individual heritability estimates show that wood density is under moderate to strong genetic control and therefore gains from classical breeding for general combining ability can be realized.

More important in papermaking is the behavior of the fibre in the formation of the sheet. Other aspects of this study intend to examine tracheid properties in detail to determine if they can be improved through breeding and selection and demonstrate to what extent the quantifiable variation impacts pulp and paper properties.


This paper reports on a portion of the research that the senior author is undertaking as part of his Ph.D. studies at North Carolina State University. This research has been funded by Sappi Forests and supported by the CAMCORE Cooperative while the senior author was resident at North Carolina State University. Without the commitment of the Pine Breeding Team at the Shaw Research Centre who planted and maintained the trials, this study would not be possible.


1. Zobel, B.J., and Jett, J.J. 1995. Genetics of Wood Production. Springer-Verlag. pp. 338.

2. Perry, J.P. 1991. The Pines of Mexico and Central America. Timber Press Inc. Portland, Oregon, USA. 231 pp.

3. Poynton, R.J. 1977. Tree planting in southern Africa Vol. 1: The Pines. Department of Forestry. Pretoria. South Africa. 576 p.

4. Evans, R. Downes, G. Menz, D. and Stringer, S. 1995. Rapid measurement of variation in tracheid transverse dimensions in a P. radiata pine tree. Appita 48(2):134-138.

5. Ladrach, W.E. 1984. Wood quality of Pinus patula. Investigacion Forestal, carton de Colombia S.A. Research Report No. 92. pp. 17.

6. Blair, R.L., Zobel, B.J., and Barker, J.A. 1975. Predictions of gain in pulp yield and tear strength in young loblolly pine through genetic increases in wood density. Tappi. Vol. 58, No. 1. pp. 89-91.

7. Plumptre, R.A., 1978. Some wood properties of Pinus Patula (Schiede and Deppe) from Uganda and techniques developed in studying them. C.F.I. Occasional Papers, No. 4. June 1978. Dept. of Forestry. Commonwealth Forestry Institute, University of Oxford.

8. Ringo, W.N., and Klem, G.S. 1980. Basic density and heartwood content in the wood of Pinus patula from Sao Hill. Record No. 14. Division of Forestry. University of Dar es Salaam. Tanzania. pp. 18.

9. Loo, J.A, Tauer, C.G., and van Buijtenen, J.P. 1984. Juvenile-mature relationships and heritability estimates of several traits in loblolly pine (Pinus taeda). Canadian Journal of Forestry Research. Vol. 14. 822-825.

10. Birks, J.S., and Barnes, R.D. 1991. The genetic control of wood density in Pinus patula. ODA Research Scheme R4616. Oxford Forestry Institute. pp. 29.

11. Barnes, R.D., Birks, J.S., Battle, G., and Mullin, L.J. 1994. The genetic control of ring width, wood density and tracheid length in the juvenile core of Pinus patula. South African Journal of Forestry 169. pp. 15-20.

12.  Nyakuengama, J.G., Evans, R., Matheson, C., Spencer, D., and Vinden, P. 1999. Wood quality and quantitative genetics of Pinus radiata D. Don: fiber traits and wood density. Appita 52(5):348-350,357.

13.  Shelbourne, T., Evans, R., Kibblewhite, P. and Low, R. 1997. Inheritance of tracheid transverse dimensions and wood density in radiata pine. Appita 50(1):47-50,67

14.  Van Buijtenen, J.P. 1967. Pulpwood properties and tree breeding: a synthesis. 14th IUFRO Congress Section 41. Munchen, Germany, 243-262.

15.  Zobel, B.J., and van Buijtenen, J.P. 1989. Wood Variation its Causes and Control. Springer-Verlag. pp. 362.

16. Burdon, R.D., Kibblewhite, R.P., and Riddell, M.J. 1999. Wood density and kraft fibre and pulp properties of four Pinus radiata provenances. New Zealand Journal of Forestry Science. 29(2). PP 214-224.

17. Dvorak, W. S., Hodge, G.R., Kietzka, J.E., Malan, F., Osorio, L.F., and Stanger, T.K. 2000. Pinus patula.  In: The Conservation and Testing of Tropical and Subtropical Forest Species by the CAMCORE Cooperative, College of Forest Resources, NCSU. Raleigh, NC. USA.

18.  Ringo, W.N., and Klem, G.S. 1986. Effect of sampling directions on wood density and tracheid length in stems of Pinus patula. Indian Journal of Forestry. Vol. 9 (2). 157-160.

19. Malan, F.S., and Marais, P.G. 1991. Direct gamma ray densitometry of wood. South African Forestry Journal, No 157 ,1-6.

20.  Falconer, D.S. 1981. Introduction to Quantitative Genetics. - 2nd ed. 335 pp.


[Home] [Title] [Author] [Organisation] [Keywords]