This item has the following downloads:
Pinus elliottii Pinus taeda Pinus elliottii
10 number of needles per fascicle (NF) while the opposite was true for base diameter (DB) top diameter (DT), fascicle length (FL), diameter of the fascicle (FD), crown projected area (CPA) and phenological traits: cessation (C), duration (D), day to reach the 50% of the growth (AG50). Average performance (no heterosis) was found for initiation of growth (IN), number of branches (NB), number of nodes (NN), tip moth incidence (TM) sheath length (SL) and specific leaf area (SLA). From path analysis, coefficients of determination of the hypothesized models were 0.664, 0.763 and 0.590 for the pseudo-backcross, loblolly and slash pine, respectively The order of importance of the traits, by relati ve magnitude of effect on total growth was: for the pseudo-backcross CPA, FL, NB, NN, FD, NF and D, while SLA had a negligible effect; for loblolly and slash pine these were CPA, NB, NN, FL, FD and NF, with SLA only relevant for loblolly. The analyses indicated that introgression of loblolly pine alleles into slash pine was effective and novel trait combinations were achieved. The pseudo-backcross had larger variation in earl y height growth than slash pine and was taller than all openpollinated fa milies at the end of the season, and also had lower tip moth incidence than its loblolly pine ancestor. The study indicated that all crown traits considered in the path analysis ha d a moderate effect in total growth with the exception of CPA that showed a consistently large effect on all taxa while SLA, ha d a minimal effect on total growth.
11 CHAPTER 1 INTRODUCTION The importance of the U.S. southern timber production is inarguable; in 1997 it was 58% of the US production and 15.8% of the world timber production (Wear and Greis 2002) with loblolly pine and slash pine as the major contributors in quantity and quality I n 2003 more than 90% of the seedlings planted were of these two species with more than 95% of the seedlings originating from genetically improved material (McKeand et al. 2003). These improved seedlings can achieve 20 to 30% more volume per unit area at rotation age than their unimproved predecessors with an important decrease in rust infection, e.g. 35% when unimproved material will incur 50% infection (White et al., 1993, McKeand et al. 2003, Gezan et al. 2004). Although the use of classic al genetic breeding results in high volume gains, the attention on volume growth has two logical consequences. First, the emphasis results in slow improvement of other traits that are not highly correlated with volume and degradation or aggradation for correlated traits. Second, there is little improvement in traits with low genetic variability in the pure species. However, with the use of hybrids it is possible to combine the most desirable traits from contrasting but related species. The common rule is that hybrids show intermediate characteristics compared to the parents (Wright 1976), but they can perform better or worse than the parental average, a phenomena known as heterosis (Wright 1976; Zobel 1984; White et al. 2007). In the case of slash and loblolly pine, the F1 hybrid has been found to have larger variability than the pure species (Barnes and Mullin 1978) and produced some outstanding individuals (Huber et al. 2000; Nikles 2000). Slash and loblolly pine ha ve unique characteristics that make hybridization interesting S ome of the commercial traits
12 that make this cross attractive are loblolly pine has faster growth than slash pine (Dorman 1976; Xiao et al. 2002) but is more susceptible to tip moth than slash pine (Lopez-Upton et al. 2000). Tip moth can cause decreases in volume of over 20% after 12 years (Stephen et al. 1982; Cade and Hedden 1987). On the other hand slash pine is more susceptible to fusiform rust than loblolly pine (Lopez-Upton et al. 1999). If combining of these traits is possible, then fast growing trees with low rust and tip moth incidence could be obtained. This thesis studies the performance of one pseudo-backcross between slash and loblolly pine. The data for the study comes from a field trial established by the Cooperative Forest Genetic Research Program (CFGRP) under the United States Department of Agriculture (USDA) Conifer Translational Genomic Network (CTGN) grant. CFGRP includes the University of Florida, private industrial and state agencies as cooperators. The breeding program primarily develops genetically-improved slash and loblolly pines for the reforestation of harvested timberlands in the Lower C oas tal Plain of the southern U.S. The field trial was established in December 2007 in Alachua County, Florida (High Springs), and contains three taxa: (1) an original selection open-pollinated loblolly pine family (Lob), (2) an original selection open-pollinated slash pine family (Slash1) and a third cycle open-pollinated slash pine family (Slash3), and (3) a pseudo-backcross (Slash1 x Lob) x Slash3. The F1 hybrid (Slash x Lob) was a selection made in 2002 for growth and rust resistance from a set of eleven hybrids trials that the CFGRP established in 1994. Fourteen repeated height measures during the first growing season allowed estimation of final height, total height growth, daily growth, initiation, cessation
13 and duration of growth. Also, at the end of the first growing season tip moth incidence, basal diameter, top diameter, number of branches, number of nodes, crown projected area, total fascicle length, sheath length, specific leaf area, fascicle diameter and number of needles per fascicle were measured. The objectives of this study were: (1) to evaluate the phenotypic effect of the transfer of loblolly pine alleles into a slash pine background during the first growth season on one North Florida site; (2) to estimate the level of heterosis of the pseudobackcross for each measured trait; (3) to determine if the relationships among traits are maintained in the pseudo-backcross when compared to the parental species; and (4) to evaluate the direct and indirect effect that some traits had on total height growth. In order to meet the first two objectives, a growth curve was fitted by family and later heterosis was investigated for each trait by analysis of variance (Chapter 2). The specific objectives of this chapter were: (1) to compare the differences among family levels for growth, pest incidence, crown architecture and needle traits; and (2) t o determine if the pseudo-backcross showed heterosis for each trait. The third general objective is assessed in Chapte r 3. Here the contribution that crown architectural, needles and phenological traits had on total height growth was studied using path analysis Also, a multivariate analysis was done to investigate the relationships among traits for all taxa. The specific objectives in this chapter were: (1) t o propose a diagram of the relationships between a set of traits and total growth; (2) t o evaluate the necessity of one, three (by taxa) or four (by family) correlation matrices for the path analyses; and, (3) to evaluate the hypothesized path diagram and compare the magnitude of the effects that each trait had on total growth.
14 CHAPTER 2 PHENOTYPIC ANALYSES OF FIRST YEAR TRAITS IN A PSEUDO-BACKCROSS (SLASH X LOBLOLLY) X SLASH AND THE OPEN-POLLINATED FAMILIES OF THE PU RE SPECIES PROGENITORS; GROWTH PHENOLOGY, TIP MOTH INCIDENCE GROWTH AND CROWN ARCHITECTURE Introduction Motivated by the performance of Pinus elliottii var. elliottii (PEE ) x Pinus caribaea var. hondurensis (PCH) in Queensland, Australia, the Cooperative Forest Genetics Re search Program (CFGRP) at the University of Florida established eleven pine hybrid trials with seven taxa including the F1 hybrid between Pinus elliottii var. elliottii by Pinus taeda (PTA) in the Lower Coastal Plain of the southeastern USA in 1994. In Queensland the PEE x PCH hybrid outperforms either parent on lowland sites with poor drainage (Nikles and Robinson 1989). It has comparable wind-firmness, stem form and wood quality to the parental species (Harding and Copley 2000). As a consequence, the hybrid has almost entirely replaced PEE and entirely replaced PCH in southeast Queensland (Nikles 2000) and averages 2.5 times as much volume and with better form than pure PEE in several South African test sites at 13.5 years (Van der Sijde and Roelofsen 1986). Often, one species has many desirable commercial traits but could be deficient in one or two major respects and does not contain enough genetic variability to permit rapid progress by intra-species selection. Hence, contrasting but related species are crossed to capture desirable trait combinations through hybridization (Wright 1976; Fowler 1978; Zobel 1984; Namkoong and Kang 1990). Often these traits pertain to difficult environments, pest resistance, or simply special ty products As an example, hybrids between E. grandis and E. urophylla combine higher resistance to canker
15 (Cryphonectria cubensis ) and higher wood density for clonal propagation (Campinhos et al 1998). In general, a hybrid has phenotyp ic characteristics intermediate to its parents (Wright 1976; Zobel 1984). However, a hybrid can also strongly display a desired characteristic of one parent and not intermediacy, or sometimes the hybrid can be superior to both parents (Zobel 1984). ybrid vigor pply if the hybrids perform better than the average of the parents. Negative hybrid vigor occurs when the hybrid performs worse than the parental average (Wright 1976; White et al. 2007). Nevertheless, tree hybrids have historically had more value as a source of new combinations of genes rather than extra vigor (Zobel 1984) and heterosis in forest tree species hybrids is the exception rather than the rule (Fowler 1978). A cross between a hybrid and either of its parents is called Backcross breeding is a well-known procedure for the introgression of a target trait from a donor line into the genomic background of a recipient line. The objective is to increase the recipient genome content of the progenies by repeated backcrosses to the recipient line (Bouchez et al. 2002) while maintaining desirable trait(s) from the donor line Introgression may be defined as an infiltration of the germplasm of one species into another by repeated backcrosses (Anderson and Hubricht 1938) or as the limited spread of genetic material from one species into another species. Kinghorn (2000) recommends using backcrossing when only two good parental breeds are available and/or when direct heterosis is not important. Yet many times one of the two species being hybridized has commercial interest, and the second has only one or a few traits that are more desirable than those found in the first sp ecies. In tha t
16 case the hybridation followed by backcross to the first species is the logical way to incorporate the traits of interest from sp ecies two into species one Many experiments have used backcrossing to achieve the goal of recovery of the commercial species or the species of interest. The backcross of (shortleaf pine x loblolly pine) x loblolly pine shows that desirable traits can be combined. Most of the offspring were rust resistant (desirable trait from shortleaf pine) and fast growing (desirable trait from loblolly pine) (La Farge and Kraus 1980; Kraus 1986). The interspecific cross of E. grandis and E. globulus and the backcrosses to the parental species were made to combine favorable adaptability characteristics of E. grandis parents with superior wood qualities of E. globulus parents (Myburg et al. 2000) A primary tenant of evolution is that species that share the same geographical region (sympatric) do not easily hybridize (Wright 1976). Some factors that prevent species from crossing are: differences in flowering time, genetic differences in anatomical/morphological flower or pollen, no embryonic development or no chromosomal pairing (Wright 1976; Zobel 1984). Exceptions to this rule are the pine species native to the southeastern US A. Loblolly and slash pine overlap sufficiently in flowering time for natural hybridization to occur (Wright 1976). Loblolly pine occurs throughout the entire south and southeast with the exception of the lower part of Florida. Slash pine grows in the coastal plain from South Carolina to Florida and west to eastern Louisiana. Nevertheless, Barnes and Mullin (1978) reported restrictions to this hybridization (slash pine and loblolly pine) where they found that reciprocal crosses do not yield the same number of seed per cone. The hybrid using slash pine as the maternal parent
17 yi elded an average of 25 viable seed per cone, while using loblolly as the maternal parent yielded less than one seed per cone. Both species, loblolly and slash, ha ve adaptative and morphological attributes which make one more suitable than the other in certain circumstances (Barnes and Mullin 1978). Loblolly pine is the fastest growing species of the southern pine group widely known for high volume production, but generally has less desirable form than certain other species (Dorman 1976). Results from many studies indicated that loblolly pine grows as well as or better than slash pine on most sites except for very poorly drained flatwoods sites on which slash pine outperforms loblolly pine (Borders and Harrison 1989). Loblolly and slash pine have numerous differences in phenology and morphology. Loblolly pine has more branches than slash pine (Xiao et al. 2003) and its branches tend to be long and fairly large in diameter (Dorman 1976) with shorter needles (Richardson 1998; Chmura et al. 2007). What is more, loblolly pine is more susceptible to tip moth ( Rhyacionia spp.) than slash pine (Lopez-Upton et al. 2000). On the other hand, loblolly pine is more resistant to fusiform rust ( Cronartium quercuum ) than slash pine (Lopez-Upton et al. 1999) and also has higher specific leaf area (McGarvey et al. 2004; Chmura et al. 2007) and larger whole tree leaf area (Dallas-Tea and Jokela 1991; Xiao et al. 2003; Martin and Jokela 2004; Emhart et al. 2007) than slash pine. It is well known from the CFGRP work that the mean performance of the slash x loblolly hybrids (F1) at age three is inferior to that of improved slash pine (negative heterosis) for pest resistance and early growth (Gezan et al. 2004). Negative heterosis was also detected for the same hybrid material at year eight for stiffness (Huber et al.
18 2007). However, by year eigh two parental species, and several excellent individuals were found (Huber et al. 2000) Nikles (2000) had similar results where he reported a very heterogeneous slash x loblolly hybrid in Queensland with no heterosis but with some outstanding individuals. Barnes and Mullin (1978) also have reported greater within-family variation in the hybrid between slash and loblolly pine than for the pure species for third-year height. This study examines the transfer of traits from a loblolly pine parent into a slash pine background utilizing an outstanding F1 individual during the first growth season at one North Florida site (High Springs). Growth, pest incidence, crown architecture and needle traits were considered. This evaluation will allow determination of the utility of backcrosses to slash pine in future slash pine breeding. The hypotheses to be tested were that there were no differences among families for all traits evaluated and the BC1 will not show heterosis. Materials and Methods Study Area Characteristics and Description This study was planted on December 18, 2007, in a single block at the High Springs Seed Orchard property of Smurfit-Stone Container Corporation located in Alachua County, Florida (294 W). The average annual precipitation for the area is 1375 mm. The average annual temperature is 20.4 C, with average high and low temperatures in summer of 41.6 and 25.5 C, and with average high and low temperatures in winter of 31.6 and 8 C, respectively (NCDC 1971-2000). The soil is classified as Lake fine sand with a slope from 0 to 5 percent. Slopes are nearly smooth to convex. Typically the surface layer is dark gray fine sand about 18 cm thick. The underlying layer is fine sand to a depth of 208 cm or more. This Lake soil has
19 low available water capacity, rapid permeability, low natural fertility, low organic matter content of the surface layer, low surface runoff, and a water table at a depth of more than 183 cm. Potential productivity of this soil is moderately high for slash, longleaf, or loblolly pines. Seedling mortality is usually moderate because of the droughty conditions of the soil. Weed competition is also moderate (Thomas et al. 1985). Because the land was previously in grass, site preparation included two tillages (mowing plus disking on July 14, 2007, and disking plus leveling on October 31, 2007). The post-plant maintenance included: replacing dead seedlings on February 15, 2008, banded glyphosate 1.1% weed control in May and June, hand-weeding in July (for persistent weeds), and fertilization with NPK (10:10:10) at 270 kg/ha on June 30. PEE (Slash1) PTA (Lob) F1 (SL1) PEE (Slash3) BC1 Figure 2-1. Pedigree of the pseudo-backcross family (BC1). Slash1 and Lob were firstcycle selections and Slash3 was a third cycle. SL1 corresponds to one slash x loblolly F1 hybrid selection. Genetic Material and Experimental Design The genetic material used for the trial was based on the 11 pine-hybrid trials established in 1994 by the CFGRP. The slash pine x loblolly pine (PEE x PTA) F1 hybrid progeny set planted in 1994 was produced by controlled-pollination of first-cycle slash pine mother trees with a loblolly pine pollen mix (Lopez-Upton 1999). Within the PEE x PTA F1 hybrids 30 selections were made for growth and disease resistance at year eight, and needle samples were sent to two genetic marker labs. The results
20 showed that five of the thirty first cycle good quality loblolly male parent. Those genotypes were selected for inclusion in an introgression program (Gezan et al 2005). The F1 hybrid (SL1) is one of these five elite selections where the slash pine female parent (Slash1) was an original slash pine selection. The ps eudobackcross (BC1) of a SL1 x a third-cycle slash parent (Slash3) was performed in 2005 (Figure 21) with seed available in fall 2006. These seeds along with open-pollinated seeds from the slash and loblolly pine ancestors of the cross were grown by Plum Creek Timber Company and the resulting seedlings were used to establish the trial (Huber et al. 2007). Table 2-1. Number of seedlings per family in the High Springs backcross study. Taxon/family Number of seedlings Slash1_OP 282 Slash3_OP 228 Lob_OP 35 2 BC1 738 Total 1600 The trial was planted in a single block (0.89 ha) in a Latinized row-column design with single-tree plots spaced at 1.82 x 3.05 m. The Latinized design was used to improve the allocation of the treatments across the trial (Williams et al. 1999). The trial includes the BC1, one open-pollinated family from loblolly pine (Lob_OP) and two openpollinated families from slash pine (Slash1_OP and Slash3_OP ) (Table 2-1) Traits Evaluated Phenology, growth and pest resistance Evaluation of height (mm) of the backcross and pure species seedlings began o n February 15, 2008, and was follow ed by repeated evaluations until the end of the first growth period, December 30, 2008 (Table 2-2) For each height measurement a
21 graduated pole was used to measure the distance from the ground to the tip of the highest bud. The traits (initiation day, 50% growth day (AG50) and cessation day of height growth) were estimated using linear interpolation to determine when the plants reached 5%, 50% and 95% of their annual growth, respectively (Mirov et al. 1952; Hanover 1963; Jayawickrama et al. 1998; Emhart et al. 2006). Table 2-2. Date and dayof -year for the height measurement of the High Springs backcross study. N Date Day of Year 2008 1 February 15 46 2 Marc h 10 69 3 March 16 75 4 March 24 83 5 March 31 90 6 April 07 97 7 May 08 128 8 June 10 161 9 July 14 195 10 August 13 225 11 September 11 254 12 October 17 290 13 November 14 318 14 December 30 364 *Due to sampling errors the 8 th measurement was not used in analysis; January 1 st is dayof -year 1 The other traits calculated were: duration of the growing season (D, calculated as the difference between the initiation and the cessation day); total growth (TG, difference in height between initiation and cessation date); and average rate of shoot elongation (ARSE cm/day, ratio of TG by D). Evaluation of pest incidence was part of the experimental plan; however, tip moth was the only pest present. Tip moth incidence was recorded for each plant at each of the 14 measurements as a binary variable. Stem and crown architecture Stem and crown architectural traits were recorded at the last measurement. The traits measured were: (1) basal diameter (mm), measured twice at the base of each
22 seedling, one perpendicular to the other utilizing a digital caliper; (2) top diameter (mm), measured just below the bud on the top woody part of the tree, utilizing a digital caliper; (3) total number of nodes on the primary stem; and (4) number of primary branches. Other variables were derived from these measurements: taper, calculated as top diameter divided by base diameter; base diameter divided by total height (B DT H); number of nodes divided by total height (NTH); number of branches divided by total height (BRTH); and number of branches divided by number of nodes (BRN). Crown projected area ( cm 2 ) was estimated utilizing digital images and the threshold technique (King et al. 2008). Digital photographs of each tree were taken with a Nikon D40x camera at a fixed distance and elevation from the tree using a white background with a graduated ruler. The ruler was used as a size reference in image processing. Images were captured from north and south (to avoid shadows) between October and November 2008 in five sessions. Images we re processed with the software ImageJ (Rasband 1997-2005) and included: cropping to the desired area (live canopy of the tree); determining and setting the number of pixels of the known length reference into the software; converting the image to black and white; and calculating the projected area of the crown (John Butnor personal communication). Fully expanded needles were collected from each tree. Needle traits measured were: total fascicle length (mm) measured with a graduated rule to the nearest millimeter, sheath length (mm measured with a digital caliper) and number of needles per fascicle counted in each fascicle sample. For specific leaf area individual needle surface was estimated according to Murty and Dougherty (1997): needle radius was measured with a magnified graduated glass (10x) and length with a digital caliper, then
23 the needles were oven-dried for 48 hours at 65 C and weighed to the nearest 0.0001g (XA -100, Denver Instruments, Denver, CO, USA). Specific leaf area (cm 2 g -1 ) was estimated using the ratio between surface area and dry weight of needles (Gonzalez SLA) that was used for analysis. Growth Curve A growth curve was fit for each tree utilizing the repeated height measures with SAS/PROC NLIN software (SAS Institute Inc. 2002-2003). Six different growth functions with fewer than five parameters were tested. The functions tested were the logistic function, generalized logistic, Chapmanmpertz function (Sit and Poulin-Costello 1994), Richards generalized function (Namkoong and Matzinger 1975), and Lundqvist-Matrn sigmoidal growth function (Danjon 1994). The model that best fit the data was the Gompertz function, obtaining the smallest means square error (MSE) and highest R 2 adj (Sit and Poulin-Costello 1994). This model was (Winsor 1932; Sweda 1984; Zeide 1993; Sit and Poulin-Costello 1994): [2 -1] Where Y corresponded to the height of the tree (mm) at the time x (days). This function has two asymptotes (a lower and upper asymptote at Y= 0 and Y=a ). The parameters a and c control the shape of the curve and the parameter b shifts the curve along the x-axis. The Gompertz function is not symmetric about its point of inflection, having a point of inflection at X=b/c corresponding to the ordinate Y=a/e For the data analyzed, this ordinate was reached at approximately 37% of the final height growth.
24 Analysis of Variance Distributional assumptions of the residuals for hypothesis testing were examined for all traits. Transformations of the response variable were performed where required. Analyses were performed using SAS/PROC MIXED software (SAS Institute Inc. 20022003) except for needle traits that were analyzed with ASReml v.2 software (Gilmour et al. 2006). Models with one to four residual effects (by family) were selected by trait using the BIC criteria (Littell et al. 2002). In addition to the hypothesis of difference among families tested for each trait, a heterosis hypothesis was also tested assuming minor gene effects for all traits except for a major effect for tip moth incidence (Lopez-Upton 1999). The hypotheses were: Minor gene effect H 0 : BC1 =0.5*[( Lob_OP + Slash1_OP )/2] + 0.5*[ Slash3_OP ] [2-2] Major gene effect H 0 : BC1 =0.25*[ Lob_OP ] + 0.75*[( Slash1_OP + Slash3_OP )/2] [2-3] Phenology, growth and pest resistance For tip moth, a compilation of attack at any time during the year was made for analysis as a binary variable, 0 not attacked and 1 attacked. Phenological, growth and pest resistance traits plus the parameter a and c from Gompertz function were analyzed. The statistical linear mixed model was as follows: Y ijk i + r j + c k + e ijk [2.4] Where Y ijk is the value for the respective trait in the ith family (i=1 to 4) in the jth row (j=1 to 40) and kth column (k=1 to 40), is the overall mean, F i corresponds to the fixed family effect, r j is the random row effect ~IID( 0, ), c k is the random effect of the column ~IID (0, ) and e ijk the random residual effect of the ith family in the jth row and kth column ~IID( 0, ) and for some traits ~Diag(0, ) where i is the famil y.
25 Stem and crown architecture Analysis of variance (ANOVA) of base diameter, top diameter, number of nodes, number of branches, taper, base diameter/total height, number of nodes/total height, number of branches/total height and number of branches/number of nodes were performed. A factor for photographic date was included in the model for projected crown area.Using repeatability analysis to estimate the gain in precision from multiple measures (Falconer and Mackay 1996), five fully elongated fascicles were taken from a subsample of 100 trees. The analysis indicated that for most of the needle traits three fascicle measurements were sufficient to reach the desired level of accuracy; however, the number of needles per fascicle required five repetitions (fascicles) per tree. ANOVA of projected specific leaf area (SLA), fascicle length, sheath length, number of needles per fascicle and fascicle diameter, were performed. The linear model 2-4 included the factor of sampling within the tree (fascicle factor). Results Growth Curve Trees in the study displayed a sigmoidal growth curve. Figure 2-2 shows the average shape of the curve for the four families. Three of the six growth functions tested did not converge for the data observed, they were: Chapmantion, Richards generalized function and Lundqvist-Matrn sigmoidal growth function. The generalized logistic failed to converge for 179 trees. The logistic and Gompertz functions converged for all trees in the study with error sum of squares for all trees (SSE) of 61,558 and 54,997 and R 2 adj of 97.3 and 97.3, respectively.
26 Figure 2-2 Average observed height growth curves for the families in the study from 14 repeated measures. Black arrow indicates day 181, day of fertilization. Since both the logistic and Gompertz functions have three parameters, the comparison of goodness of fit can be based on the SSE indicating a better fit for the Gompertz function. From the Gompertz model the predicted average height for the families in the study (Figure 2-3) showed a slight decrease in slope for BC1 late in the growing season while both Slash3_OP and Lob_OP maintained a constant and larger slope than BC1. Phenological, Growth and Pest Resistance All traits were statistical ly significant (p-value<0.05) for differences among families and for the heterosis contrast except for initiation and tip moth (Table A-1). Figure 24 shows the mean levels for the traits evaluated, the dashed line indicates the expected values for BC1. Slash pine families initiated their growth earlier than loblolly pine while BC1 had a value intermediate between its parents (Figure 2-4A). The family patterns were the same as initiation for AG50 except that the BC1 demonstrated heterosis
27 reaching 50% of its growth six days before the parental average expectation (Figure 24B). Figure 2-3 Average predicted height growth curves for the families in the High Springs backcross study. Black arrow indicates day 181, day of fertilization. Cessation and duration of growth had similar mean patterns (Figure 2-4C and 24D). Slash3_OP completed its growth on dayof -year 281 (the longest duration of growth, 223 days). On the other hand BC1 had the shortest duration (212 days) because the family stopped growing earlier than other families (less than the parental average). Total height, total growth and average rate of shoot elongation (ARSE) had similar patterns for the family means in the study (Figure 2-4E, 2-4F and 2-4G). The performance of the two slash families differed considerably; Slash3_OP had the highest rate of growth (0.12 cm day -1 ), total growth, and was the second tallest family at the end of the season while Slash1_OP had the lowest rate of growth (0.09 cm day -1 ), tota l growth, and was the shortest family at the end of the first growing season. Lob_OP had
28 Figure 2-4 Least-square means and error bars for phenological, height growth and tip moth incidence. Slash pine families (Slash1_OP and Slash3_OP), loblolly pine family (Lob_OP) and the pseudo-backcross (BC1) for (A) initiation; (B) AG50, days for reaching 50% growth; (C) cessation; (D) duration; (E) ASRE, average rate of shoot elongation; (F) total growth; (G) total height; and (H) tip moth incidence for 2008 growing season in the High Springs backcross study. The dashed line indicates the average expectation for BC1. If the error bar for BC1 does not touch the dashed line then BC1 showed heterosis. (A) (B) (C) ( D ) ( E ) ( F ) ( G ) ( H )
29 a lower rate of growth than BC1 and Slash3_OP, plus a shorter period of growth and completed the season below Slash3_OP and BC1 for height. Both slash pine entries had low tip moth incidence (10%) where Lob_OP had more than 50% of its individuals affected. The backcross BC1 showed no heterosis at 20% infection when the weighted parental average was 21% (Figure 2-4H). Figure 2-5. Least-square means and error bars for the parameters of Gompertz function. Slash pine families (Slash1_OP and Slash3_OP), loblolly pine family (Lob_OP) and the pseudoHigh Springs backcross study. Dashed line is average expectation for BC1. The asymptote values (parameter a) and the intrinsic rate of growth (parameter c) were statistically significant (p-value<0.05) for differences among families and heterosis (Table A-1). Asymptote mean values had patterns similar to growth. Slash3_OP had the highest asymptote (70.56 cm) while Slash1_OP had the lowest (51.90 cm). BC1 was the second tallest in the test (Figure 2-5A). The intrinsic rate of growth (expressed inversely, lower values indicate higher rate of growth) indicated that Lob_OP and (A) (B)
30 Slash3_OP had the highest rate of growth (0.0077 and 0.0079 respectively). BC1 had a lower rate of growth than Lob_OP and Slash3_OP, but larger than Slash1_OP (Figure 2-5B). The coefficient of variation of Lob_OP and BC1 were lower than the slash families before the 9th measurement; however, after this date the variation more than doubled the original value (Figure 2-6). Slash families had the lowest final CV when compared with BC1 and Lob_OP performing consistently as a species even when they came from different improvement cycles. Figure 2-6. Behavior of the coefficient of variation (CV) for height during the growing season for the four families in the High Springs backcross study. Black arrow indicates day 181, day of fertilization. Stem and Crown Architecture All crown architectural and foliar variables differed among families (p-value<0.05) while the significance of the heterosis hypothesis varied by trait (Figure 2-7A through J, see Table A-2; Table A-3 for details). Slash3_OP had a larger basal diameter than the three other families. BC1 demonstrated negative heterosis for base and top diameter,
31 performing 18 mm and 8 mm less than expected, respectively (Figure 2-7C and 2-7D). The results for top diameter showed the slash pine families had thicker diameters, and Lob_OP had the smallest mean value. Differences in top diameter among families drove the results for taper (Figure 2-7G). Slash pine decreased less in diameter toward the top, as an example Slash1_OP has approximately half of its base diameter at the top. Lob_OP had the largest decrease in diameter from the base to the top. There was no heterosis for taper for BC1. Number of nodes and number of branches had similar patterns for family means (Figure 2-7A, B). Slash3_OP had the highest number of nodes and number of branches, but only slightly higher than Lob_OP family while Slash1_OP had the lowest values. Lob_OP had more branches per node than the other families. Slash1_OP had the lowest number of branches per node (1.9) while BC1 performed as the average of its parents (Figure 2-7E). The ratio of diameter divided by total height indicated that BC1 had the lowest diameter per unit height (negative heterosis) performing outside the parental range and more closely resembling loblolly pine (Figure 2-7F). Number of nodes per unit of height and number of branches per unit of height had similar family mean patterns (Figure 2-7H and 2-7I). Lob_OP had more nodes per unit height and more branches per unit height than other families; BC1 had negative heterosis for both ratios and Slash1_OP has the lowest mean values. Slash3_OP had larger projected crown area (PCA) than the other families in the study (Figure 2-7J), more than 40% greater than Slash1_OP while BC1 had negative heterosis performing 82 cm2 lower than the expectation.
32 Figure 2-7. Least-square means and error bars for crown architecture traits. Slash pine families (Slash1_OP and Slash3_OP), loblolly pine family (Lob_OP) and the pseudo-backcross (BC1) for (A) number of nodes; (B) number of branches; (C) base diameter; (D) top diameter; (E) BRN= number of branches per node; (F) BDTH=base diameter per unit height; (G) taper; (H) NTH= number of nodes per unit height; (I) BRTH=number of branches per unit height; (J) Crown projected area. The dashed line indicates the average expectation for BC1 (see captions in Figure 24) (A) (B) (C) ( D ) ( E ) ( F ) ( G ) ( H ) ( J ) ( I )
33 Fascicle length and diameter had the same mean pattern for the families in the study (Figure 2-8A to E) because they were highly correlated (0.74). Slash families had larger and thicker needles than the loblolly family. There was negative heterosis for both length and diameter for BC1. BC1 did not show heterosis for sheath length performing as its parental average (Figure 2-8D). Lob_OP had slightly more needles per fascicle (3.0) than the slash families, 2.9 and 2.7, (Figure 2-8B). BC1 show ed heterosis for this trait. Slash families had lower specific leaf area (SLA) than the loblolly family and heterosis was not present for SLA (Figure 2-8C). Figure 2-8 Least-square means and error bars for needle traits. Slash pine families (Slash1_OP and Slash3_OP), loblolly pine family (Lob_OP) and the pseudobackcross (BC1) for (A) fascicle length; (B) number of needles per fascicle; (C) SLA, projected specific leaf area; (D) sheath length; (E) fascicle diameter. The dashed line indicates the average expectation for BC1. If the error bar for BC1 does not touch the dashed line then BC1 demonstrated heterosis. (A) (B) (C) ( D ) ( E )
34 Discussion The Gompertz function provided an adequate fit for height growth for slash, loblolly and their backcross with R 2 adj > 0.90 for every tree evaluated. Sweda (1984) reported that Gompertz was the best (over three functions) in a radial stem growth curve for white spruce. Although in our results the family patterns of the asymptote estimation differ ed from the patterns for final height (measured directly). A reasonable explanation for that was the fertilization effect that in creased the slope of the loblolly growth curve for the last part of the growing season. It has been shown that loblolly pine growth is generally more responsive to added fertility than slash pine (Martin and Jokela 2004b; Roth et al. 2007). In our case, fe rtilization increased the loblolly intrinsic rate of growth and affected the asymptote, and the growth curve appear ed to be a segmented function where the day of fertilization mark ed the separation between two potential curves. The date of initiation and cessation determine the duration and those factors plus the daily rate of growth determine the total growth f or the season (Emhart et al. 2006). BC1 started growth earlier than was recorded. This was observed because at test establishment all families were near the same height (not recorded), but BC1 differed at the 1 st measurement (initial height). This meant that the duration of BC1 should be longer than reported. This was also supported by the fact that Slash3_OP and BC1 had the same daily rate of growth and in order to reach similar final height they should have had similar durations for growth. Our phenological results for loblolly pine were in partial accordance with Parisi (2006) where initiation and cessation of growth for seedlings at a Florida site in the second year were the 83 rd and 266 th dayof -year, respectively. Initiation differed by more
35 than twenty days, whereas cessation only differed by eight days. The differences in cessation could have be en caused by site-to -site variation (Parisi 2006) or year-to -year variation (Jayawickrama et al. 1998; Emhart et al. 2006). BC1 demonstrated positive and negative heterosis for several traits and no heterosis for others. While additive gene actions explain parental average traits, heterosis is supposed to be due to non-additive gene action (Shull 1908; Bruce 1910). It had been found to be primarily due to dominant gene action in maize and there is some evidence for epistasis in rice (Franco et al. 2008). Positive heterosis was present for all height growth traits, meaning that BC1 performed better than the parental average while Gezan et al. (2004) found that the F1hybrid had negative or no heterosis for growth traits. Similar results have been seen in other species. With Major et al. (2003) reporting negative heterosis in mature trees for the intercross between red x black spruce F1 hybrid, whereas the backcross to black spruce produced positive heterosis for seedlings and mature trees. The slash pine families were significantly different as was expected. Slash1_OP was an original wild selection and Slash3_OP was a third-improvement cycle selection. The CFGRP analyses indicated that Slash3_OP had approximately twice the breeding value for volume of Slash1_OP. The goal of the backcross program wa s to introduce the fast growing properties of loblolly to slash pine. With a larger base of taxa and families within taxa with the same degree of improvement and several trials, LopezUpton (1999) found that loblolly pine grew faster and was consistently taller than slash pine at three years. In our study the loblolly family was shorter than Slash3_OP but larger than Slash1_OP for height at age one.
36 The influence of loblolly pine alleles was evident in tip moth incidence for BC1 ; however, this was also a successful backcross because the performance of BC1 for tip moth incidence was much closer to slash than loblolly. The confirmation of the successful gene transmission by the recurrent slash parent comes from previous evaluations of the slash x loblolly F1 hybrid at age o ne Lopez-Upton et al. (2000) showed that loblolly pine families and the loblolly x slash F1 hybrid families had higher tip moth incidence than slash pine families. Also, at the third-year evaluation, the F1 families had similar levels of tip moth damage as loblolly pine families. These results were consistent with the hypothesis that loblolly was conferring susceptibility to the hybrid. Even when the slash families came from different improvement cycles, there was no evidence of selection for tip moth resistance. This was consistent with LopezUpton et al. (2000) where there was no statistical difference between improved (1 st cycle) and unimproved slash pine families for tip moth incidence, indicating that selection for growth and rust resistance had not changed the ability of the species to sustain low rates of tip moth incidence. Tip moth can decrease the volume yield in loblolly pine by 28% and 23% after 12 and 20 years, respectively (Stephen et al. 1982; Cade and Hedden 1987).In our study we showed that tip moth resistance could be transmitted from slash pine to the backcross. Considering the importance of this trait and the possible combinability showed in our study the us e of hybrids appears to be a logical strategy to improve tip moth resistance in USA southern pines The BC1 family had more variation in early height growth than the slash families. BC1 was apparently influenced by the loblolly pine parent whose family had the largest
37 CV for height growth while the slash families had consistently low variation. Higher variation could be from higher heritability and then higher genetics gains could be reached in a potential hybrid breeding program. On the other hand this BC1 could be introduced to a slash improvement program as an infusion to increase the variability or provide greater environment al response. The BC1 demonstrated heterosis for the ratio of diameter to height. The BC1 was taller with lower relative diameter than slash and loblolly pine families. This could be a negative issue in an improvement program where the final goal is volume. In pure pine species branch traits are under moderate to strong genetic control (Ehrenberg 1963; Strickland and Goddard 1965 ; Emhart et al. 2007). No apparent heterosis for branch traits indicated that those traits were controlled by additive effects. This could be an advantage in a potential improvement program where, if the observed pattern continued, parents with desirable branch trait levels should result in a desirable hybrid. Slash pine families have longer and thicker needles than the loblolly pine family ; and similar results were reported by Richardson (1998) and Chmura et al. (2007). Chmura et al. (2007) found needle length of 228.3 mm and around 160 mm for slash and loblolly in the second-year evaluation, respectively. These values were larger than for this study for slash (around 200 mm) and loblolly (144 mm). Once more these differences demonstrated the genetic variation within species. More than 50% of the time slash pine had three needles per fascicle contrary to other studies that characterized slash pine as two needles per fascicle with a few threes (Dorman 1976). The higher SLA found in loblolly pine over slash pine families and BC1 was in accord with Chmura et al. (2007), Will et al. (2001) and McGarvey et al. (2004). McGarvey et al
38 (2004) found SLA of 155-187 cm 2 gr -1 and 128-131 cm 2 gr -1 upper and lower crown for loblolly and slash, respectively. SLA did not explain the growth differences between slash and loblolly families as was found by Marron and Ceuleman (2006) for poplar hybrids, while between the slash pine families higher SLA was associated with higher growth. In conclusion, the analyses indicated that introgression of loblolly pine alleles into slash pine was effective and novel trait combinations were achieved. The backcross had larger variation in early growth than slash pine and is taller than both slash families at the end of the season with lower tip moth incidence than the loblolly pine ancestor. Also BC1 was more efficient in terms of height growth given the rate of growth and size at the end of the season with smaller crown area than Slash3_OP. Even when crown area was standardized by size (height and diameter, data not shown), BC1 was still more efficient. Significant differences were found for growth, phenological, crown architectural and needle characteristics among the loblolly, backcross and slash pine families examined for the first growing season at the High Springs site. Positive heterosis was found for all growth characteristics and number of needles per fascicle, while the opposite was true for the phenological characteristics (except for initiation that did not show heterosis), diameter base and top, fascicle length and diameter and crown projected area. Average parental performance for BC1 was found for initiation of growth, branch traits, tip moth incidence, sheath length and specific leaf area. At present the tree improvement programs of the southeastern USA have material with higher levels of improvement for both species These better selections could be
39 used to obtain better hybrids improving the performance of those traits that appeared to be highly influenced by the poor performance of the first-cycle slash ancestor. BC1 could be introduced in the slash improvement program as an infusion of new material to increase the variability and introduction of new alleles of commercial interest. Finally, hybridization and backcrossing have demonstrated high potential as a way to introduce novel traits from one species to the other. In this respect, hybrids can contribute by either increasing growth or maintaining actual growth gains while improving other traits of commercial importance.
40 CHAPTER 3 PATH ANALYSIS OF THE PHENOTYPIC EFFECTS OF FIRST-YEAR CROWN ARCHITECTURAL AND PHENOLOGICAL TRAITS ON TOTAL GROWTH OF THE PSEUDO-BACKCROSS (SLASH X LOBLOLLY) X SLASH AND THE OPENPOLLINATED FAMILIES OF THE PURE SPECIES PROGENITORS Introduction The use of hybrids in breeding has been primarily motivated as an alternative to increase diversity within taxa that have low or almost null genetic variability and as a tool to combine desirable traits. Often in plant science the traits of interest are pest resistance, growth or adaptation to extreme or difficult environments. The common rule is that hybrids show characteristics intermediate to the parents (Wright 1976), but also they can perform better, or worse, than the parental average, a phenomena known as heterosis (Wright 1976; Zobel 1984; White et al. 2007). For example, hybrids between E. grand is and E. urophylla often combine high resistance to canker ( Cryphonectria cubensis ) and high wood density on selected genotypes used for clonal propagation (Campinhos et al. 1998). In southeast Queensland, Australia due to the better performance for volume of the hybrid between Pinus elliottii var. elliottii (PEE) and Pinus caribaea var. hondurensis (PCH) compared to either parent, plantations of PCH have been entirely replaced and PEE is almost entirely replaced by this hybrid (Nikles 2000). In addition to desirable growth, this hybrid has levels of stem form, wood quality and wind-firmness comparable to the parental species (Harding and Copley 2000). The same hybrid has been reported, in a test at 13.5 years, to outperform PEE in South Africa (Van der Sijde and Roelofsen 1986) with 2.5 times the volume and better stem form. The performance of this hybrid in Queensland motivated the Cooperative Forest Genetics Research Program (CFGRP) at the University of Florida to investigate potential pine hybrids for
41 the Lower Coastal Plain of the southeastern USA I n 1994 eleven field trials with seven taxa were established, including the hybrid Pinus elliottii var. elliottii x Pinus taeda (PTA). A backcross is a mating between a hybrid and either of its parents and i s a method used for the introgression of a target trait from a donor line into the genomic background of a recipient line. Bouchez et al. (2002) indicated that the objective of introgression is to increase the recipient genome content of the progenies by repeated backcrosses to the recipient line to capture the desirable trait(s) from the donor line. In pine breeding it has been demonstrated that by using backcrossing desirable traits can be successfully combined. That was the case of (Pinus echinata x Pinu s taeda ) x Pinus taeda where most of the offspring were rust resistant (a desirable trait from P. echinata ) and fast growing ( a desirable trait from P. taeda ) (La Farge and Kraus 1980; Kraus 1986). The use of backcrossing has been recommended when good parental breeds are available (Kinghorn 2000) as is the case of loblolly pine and slash pine in the southeastern US. In this region loblolly pine and slash pine are the most important commercial species (Dorman 1976; Borders and Harrison 1989; McKeand et al. 2003); covering close to 12 .0 and 5.3 million hectares, respectively (Jokela and Long 2000) in naturally regenerated and planted stands. In 2003 more than 95% of the pine plantations contain ed genetically improved material (McKeand et al. 2003). In general, loblolly pine is widely known for high volume production and has less desirable form among the southern pine group (Dorman 1976). Loblolly pine grows as well as or better than slash pine on most sites except for very poorly drained flatwoods sites on which slash pine outperforms loblolly pine (Borders and Harrison 1989). In
42 addition loblolly families tend to be more stable in volume rankings across sites than slash pine families at three years (Lopez-Upton 1999) i.e. showing less genotype-by environment interaction Several morphological characteristics differentiate loblolly from slash pine: loblolly pine branches are larger in length and diameter (Dorman 1976; Xiao et al. 2003) than slash pine. Loblolly needles are shorter (Richardson 1998; Chmura et al. 2007) with higher specific leaf area (McGarvey et al. 2004; Chmura et al. 2007) and greater whole tree leaf area than slash pine (Dallas-Tea and Jokela 1991; Xiao et al. 2003; Martin and Jokela 2004; Emhart et al. 2007). The growth performance of the F1 hybrid field trials at eight years after planting was equal to the average of the two parental species, and several good individuals were found (Huber et al. 2000). Similar results were reported by Nikles (2000) in Queensland where a very heterogeneous slash x loblolly hybrid was detected with no heterosis but with some outstanding individuals. Barnes and Mullin (1978) also reported greater within-family variation in the hybrid than in the pure species (slash and loblolly pine) fo r third-year height. When starting a pine hybrid breeding program, it is of primary importance to understand if the relationships among the traits that could be affecting growth are maintained in the hybrid in comparison to the pure species. This is a simple task when two or even three traits are involved where a simple correlation matrix would give sufficient insight. However, as more variables are considered in the correlation matrix, it is increasing difficult to understand the mutual, positive or negative, associations among traits and also it is more difficult to interpret these relationships (Dewey and Lu 1972). A good alternative is the use of path analysis. This statistical tool was described by Wright
43 (1921) as elation coefficient between variables in a Path analysis has been fully detailed in the literature (Kremer 1985; Bollen 1989; Lynch and Walsh 1998) and has been used as a way to understand the relationship between productivity and its components in several crop plants (Dewey and Lu 1959; Duarte and Adams 1972; Cramer and Wehner 1998; Bidgoli et al. 2006; Babar et al. 2007) and in forestry for this and other purposes (Kremer and Larson 1983; Govindaraju 1984; Lundquist 2000; Sterck et al. 2003; Weisberg 2004; Parisi 2006). However, the usefulness of path analysis is limited to the degree of knowledge that the researcher has of the biological phenomenon underlying the study (Wilkinson et al. 1996). The objective for this study was to determine the relative contribution of individual crown architectural and phenological traits to the first season total growth for a backcross between (slash pine x loblolly pine) x slash pine and its open-pollinated pure species progenitors on one site in North Central Florida. In order to achieve this objective, path analysis was used based on a hypothetical relationship. This represents a system of simultaneous equations where the main advantage is that it presents a graphical representation of the relationships that are assumed to exist among the different traits (Bollen 1989). Path analysis is valuable for comparing the same type of interpretative diagram to different populations through interpretation of the path coefficients (Wright 1960). Materials and Methods Experimental Site and Genetic Material The experiment wa s planted on December 18, 2007 on land managed by SmurfitStone Container Corporation in Alachua County, Florida. The topography is smooth to
44 convex with slopes varying from 0 to 5%. The soil is classified as Lake fine sand which has low water capacity, low natural fertility and rapid permeability. The water table is at a depth of more than 183 cm (Thomas et al. 1985). The climate is humid and subtropical with an average annual temperature of 20.4C and average annual precipitation for the area of 1375 mm (NCDC 1971-2000). Table 3-1. Number of seedlings per family in the High Springs BC1 trial. Family Number of seedlings Slash1_OP 282 Slash3_OP 228 Lob_OP 35 2 BC1 738 Total 1600 Site preparation included two tillages (mowing plus disking on July 14, 2007, and disking plus leveling on October 31, 2007). The post-plant maintenance included: replanting on February 15, 2008, banded glyphosate 1.1% weed control in May and June, hand-weeding in July (for persistent weeds), and fertilization with NPK (10:10:10) at 270 kg/ha on June 30. The experiment was planted in a Latinized row-column design with single-tree plots in a single block (0.89 ha) spaced at 1.82 x 3.05 m. A Latinized design was used to improve the allocation of the treatments across the trial (Williams et al. 1999). PEE (Slash1) PTA (Lob) F1 (SL1) PEE (Slash3) BC1 Figure 3-1. Pedigree of the pseudo-backcross family (BC1). Slash1 and Lob were first cycle selections, while Slash3 was a third cycle selection. SL1 was a slash x loblolly F1 hybrid.
45 The genetic material for this experiment included four families (Table 3-1); (1) an original selection slash pine open-pollinated family (Slash1_OP), (2) a third-cycle of improvement slash pine open-pollinated family (Slash3_OP) (3) an original selection loblolly pine open-pollinated family (Lob_OP) and (4) a pseudo-backcross Slash1 x Lob) x Slash3 (BC1) (Figure 3-1). The F1 hybrid slash x loblolly pine hybrid was tested, along with other taxa, in 11 trials established in 1994 by the CFGRP. At age eight 30 individuals were selected and needle samples were sent to two genetic marker labs. In five of the thirty selections the presence of a good quality loblolly parent was detected and those were chosen to start an introgression program (Gezan et al. 2005). SL1 was an F1 slash x loblolly pine selection that came from elite material and the maternal parent of BC1 (Figure 3-1). Ta ble 3-2. Date and dayof -year for height measurement of the High Springs backcross study. N Date Day of Year 2008 1 February 15 46 2 March 10 69 3 March 16 75 4 March 24 83 5 March 31 90 6 April 07 97 7 May 08 128 8 June 10 161 9 July 14 195 10 August 13 225 11 September 11 254 12 October 17 290 13 November 14 318 14 December 30 364 *Due to measurement errors the 8 th assessment was not used in analysis; January 1 st is dayof -year 1 Traits Evaluated Height (mm) was assessed 14 times during the first growing season (Table 3-2). A graduated pole was utilized to measure the distance from the ground to the tip of the
46 highest bud. Initiation and cessation of height growth were estimated by linear interpolation to determine when the plants reached 5% and 95% of their annual growth, respectively (Mirov et al. 1952; Jayawickrama et al. 1998; Emhart et al. 2006). Total growth (TG) for the period was calculated as the difference in height between initiation (IN) and cessation (CS) dates. Duration of the growing season (DU) was calculated as the difference between the initiation and the cessation day. The following crown architectural and needle traits (from three fully expanded needle fascicles and five fascicles for number of needles per fascicle) collected at the final measurement were recorded for each tree: (1) total number of nodes on the primary stem (NN); (2) number of primary branches (NB); (3) crown projected area (CPA, m 2 ) estimated utilizing digital images and the threshold technique (King et al. 2008) where the digital images of each tree were taken with a Nikon D40x camera and were processed with the software ImageJ (Rasband 1997-2005); (4) fascicle length (FL, mm) measured with a graduated rule to the nearest millimeter; (5) number of needle s per fascicle (NF) counted in each fascicle sample; (6) fascicle diameter (FD, dm) measured with a magnified graduated glass (10x); (7) specific leaf area (cm 2 g -1 ) estimated using the ratio between surface area and dry weight of needles (Gonzalez 2008) wh ere surface area was estimated according to Murty and Dougherty (1997) and needles were oven-dried for 48h at 65 C and weighed to the nearest 0.0001g (XA-100, projected specific leaf area (SLA) for analysis.
47 Statistical Analyses The hypothetical path diagram is illustrated in Figure 3-2. The diagram is divided in two main branches: one based on crown architecture traits and the other on phenological traits. Figure 3-2. Hypothetical path diagram, showing the situation in which NN (number of nodes), NB (number of branches), FL (fascicle length), NF (number of needle per fascicle), FD (fascicle diameter) and SLA (projected specific leaf area) directly impact the CPA (crown projected area) and interact with each other to indirectly impact TG (total growth). Also IN (initiation), CS (cessation) directly impact DU (duration of growth) and interact with each other to indirectly impact TG. Path coefficient (p ij ) is a measure of the direct effect, and correlation coefficients are measures of the pairwise relationships. In this diagram CPA is an endogenous variable whose variance is theorized to be explained by the exogenous variables NN, NB, FL, NF, FD and SLA plus the residual error term E7 that reflects the unexplained variance and measurement error. The variance of the endogenous variable DU is theorized to be explained by the exogenous variables IN and CS plus the residual error term E11. At the same time the endogenous variable TG is theorized to be explained by CPA, DU and the residual term E8, this
48 relationship is indicated by directional arrows where each connection is associated to a p ij (path coefficient) where i indicates the effect and j indicates the cause. The lines with arrows at both ends indicate correlation between both variables r kl where k indicates the first variable and l the second variable. To obtain the coefficients associated with the path diagram, it was necessary to estimate correlations among the different traits. A multivariate analysis of the eleven traits, mentioned above, was performed using the software ASReml v.2 (Gilmour et al. 2006). The statistical linear mixed model was: [2 -1] Where y = i.e. a stacked vector with de column random effects; X and Z are incidence matrices; e is a vector of random errors. The variancestructures: (1) a pooled unstructured error matrix common for all families (55 variance components); (2) an unstructured error matrix for each taxa (165=55x3 variance components); (3) an unstructured error matrix for each family (220=55x4 variance components). LogLikelihood ratio test (Wolfinger 1996) determined if one (pooled), three (by taxa: Slash, Lob and BC1) or four (by family) path analyses were needed. Utilizing the estimated correlation matrices from above, path analyses were performed with SAS/PROC CALIS software (SAS Institute Inc. 2002-2003) and direct, indirect and total effects were calculated according to Dewey and Lu (1959). The direct effect is the influence of the variable on TG unmediated by any other variable of the path model. The indirect effects (IE) are mediated by at least two other variables in the model.
49 Results It was not possible to fit the multivariate analysis with all traits of interest because of strong positive correlation between CS and DU; therefore, CS was dropped from the subsequent analyses. Some details of the fitting and comparison of the multivariate models are presented in Table 3-3 Both three and four error matrices were statistically significant ly different from the model with one error structure. However, the model with four error structures was not significant ly different from the model with three error structures (P-value=0.245). Table 3-3. Likelihood ratio test p-values for tests of the error structure in the multivariate model. model number of p arameters nu mber of traits 2LogL ik Chi Square P value 1 55 10 8530.94 3 165 10 8129.82 9.8E 35 4 220 10 8067.98 0.24512 Therefore, a model with an error matrix for each of the three taxa was used for path analyses (Table B-1). Almost all correlations among variables were significant (pvalue<0.05). No significant correlations were found for: DU-IN, SLA-NB, SLA-DU and NF DU for BC1; DU-NN, DU-NB, CPA-DU, FL-IN, SLA-NB, SLA-DU, FD-IN, FD-SLA, NF -IN and NF-SLA for Lob; DU-NN, CPA-DU, FL-DU, SLA-NN, SLA-NB, SLA-DU, SLACPA, FD-DU, FD-SLA, NF-IN, NF-DU, NF-FL and NF-SLA for the Slash taxa. Path analyses were performed by tax on where the variables included for each taxon varied according to the significance of the relationship between the independent and the dependent traits. For the Lob and Slash taxa the phenological branch of the diagram was not included in the path analysis due to the lack of relationship between DU and TG (p-value=0.44 for Lob and 0.71 for Slash). In the specific case of BC1 taxon
50 the phenological branch was included; however, IN was dropped from the analysis because no significant relationship was present between IN and DU (p-value=0.06). A similar situation occurred for the variable SLA that had no significant relationship with CPA for the Slash taxon (p -value=0.12). The three final path analysis diagrams along with their respective path coefficients and relevant relationships are presented in the Figure 3-3, 3-4 and 3-5 for BC1, Lob and Slash, respectively. The partitioning of the effects is presented in Table B-2 and Figure 3-6, 3-7 and 3-8 for BC1, Lob and Slash, respectively. Figure 3-3. Path diagram path coefficient and relevant correlations for BC1. NN (number of nodes), NB (number of branches), FL (fascicle length), NF (number of needle per fascicle), FD (fascicle diameter), SLA (projected specific leaf area), CPA (crown projected area), TG (total growth) DU (duration of growth), E 7 and E 8 are the residual terms. The sum of the direct and indirect effects is the total effect. Mathematically, the indirect effect will be the sum of all indirect paths. As an example, in the case of the indirect effect of SLA for Lob will be: -0.159 x 0.0342 x 0.8735 + -0.345 x 0.3131 x 0.8735 = -0.099.
51 Figure 3-4. Path diagram, path coefficient and relevant correlations for Lob. NN (number of nodes), NB (number of branches), FL (fascicle length), NF (number of needle per fascicle), FD (fascicle diameter), SLA (projected specific leaf area), CPA (crown projected area) and TG (total growth), E 7 and E 8 are the residual terms. Figure 3-5. Path diagram, path coefficient and relevant correlations for Slash (see captions Figure 3-4 above for abbreviations)
52 Figure 3-6. Direct, indirect and total effect of crown architecture traits and duration of growth on total growth for BC1. Number of nodes (NN) number of branches (NB), fascicle length (FL), projected specific leaf area (SL A) fascicle diameter (FD), number of needles per fascicle (NF), crown projected area (CPA) and duration of the growth (DU). Figure 3-7 Direct, indirect and total effect of crown architecture traits and duration of growth on total growth for Lob. Number of nodes (NN) number of branches (NB), fascicle length (FL), projected specific leaf area (SL A) fascicle diameter (FD), number of needles per fascicle (NF), crown projected area (CPA) and duration of the growth (DU).
53 Figure 3-8 Direct, indirect and total effect of crown architecture traits and duration of growth on total growth for Slash. Number of nodes (NN) number of branches (NB), fascicle length (FL), projected specific leaf area (SL A) fascicle diameter (FD), number of needles per fascicle (NF), crown projected area (CPA) and duration of the growth (DU). Discussion In this study multivariate analysis proved useful in determining the number of correlations matrices needed among different families The inclusion of fixed and random effects in the multivariate model allowed estimates of error and correlation matrices that were properly adjusted by these effects. In this analysis, the original model included eleven traits although high co rrelation between CS and DU did not allow fitting of the complete set of traits. This high correlation has already been reported earlier by Emhart et al. (2006) for different loblolly and slash families and by Parisi (2006) for loblolly pine. The best model (ten traits) had three correlation matrices (one for each taxon). This result suggested that the nature of the intra-plant trait relationships change according to taxa. This could be explained by the different growth strategies that loblolly and slash pine have at the early stages (Colbert et al. 1990) where slash pine
54 partitioned more biomass to stem growth and loblolly pine partitioned more biomass to branch and foliage resulting in larger crown area (Martin and Jokela 2004). For slash and loblolly pine families studied here there was no significant correlation between TG and DU allowing for the removal of the phenological branch of the diagram from the analyses for these taxa. This relationship could depend on the specific family being considered. In an other study, Emhart et al. (2006) found a significant relationship between TG and DU for one slash family and no significant relationship for other slash families. In the case of BC1 this correlation was low, but still significant, while the correlation between DU and IN was not significant. This result was opposite to that reported by Parisi (2006) where DU was significant but weakly correlat ed with IN in loblolly pine while Emhart et al. (2006) found both significant and no n-significant correlations depending on the family considered. Several interesting correlations were observed in Table B-1. A moderate to high correlation (0.20 to 0.74) was found between CPA and needle and crown traits across all taxa with the exception of SLA for slash pine that was not significant. SLA had the lowest correlation with these traits. From the results of path analysis it was possible to rank the traits according to the magnitude of their effect on TG. The path analysis indicated that the important effects in height growth in order of relative importance for BC1 were CPA, FL, NB, NN, FD, NF and DU while SLA had a negligible effect. For loblolly and slash pine these were CPA, NB, NN, FL, FD and NF with SLA only relevant for loblolly. Crown area has been reported to have a consistently positive effect on height for jack pine in other studies based on path analysis (Govindaraju 1984). Although the traits rankings for loblolly and
55 slash pine were similar, the magnitude of the effects of each trait varied considerably (Table B-2). The total effect of CPA on TG was 0.87 and 0.77 for loblolly and slash pine, respectively. These values indicated that crown area is slightly more important for growth for loblolly than slash pine Path coefficients may be interpreted as the change in the dependent variable (in standard deviations) caused by a change in the independent variable (in standard deviations) when all other background variables are held constant (Lynch and Walsh 1998). Path coefficients of the pairs NN-CPA, NB-CPA and FL-CPA were found to be more similar between Slash and BC1 (Figure 3-3, 3-5) than Slash or BC1 with loblolly pine. Such similarity could be explained by the fact that, on average, BC1 shares 75% of the slash genes. In the case of NN-CPA, the path coefficient for loblolly pine was close to zero while that for slash pine and BC1 was on the order of 0.15. The effect NBCPA for loblolly pine was 35% higher than both slash and BC1. This could be because of the importance of branches in loblolly pine that are, in general, larger than slash pine (Xiao et al. 2003; Chmura et al. 2007). In the case of FL-CPA the coefficient was 32% more important for slash and BC1 than for loblolly pine. In the case of slash, this could be due to the different levels that FL has in the two species; in general, slash pine has larger needles than loblolly pine (Chmura et al. 2007) and in particular for the families in this study, the same pattern was observed (Chapter 1 ). In the case of BC1 this could be either for the gene pool that BC1 share with slash pine or because some BC1 trees have extremely short needles with evident problems in growth; those plants appeared to be mutants. The path coefficient of NF-CPA was lower for BC1 and loblolly than slash pine.
56 The path coefficient between two traits divides the correlation coefficient (total effect) into a series of direct and indirect effects, in this case between hypothetical causal traits and total growth. In this context, for BC1 the total effect of NN, FD, NF on TG is explained by the indirect effects in 78, 86 and 75%, respectively (Figure 36) These proportions were higher in loblolly pine where they explained over 90% of the total effect for the same traits (Figure 37) These results indicated that these traits interact with other intermediate traits to finally contribute to the total growth. The opposite is true for NB for loblolly where 78% of the total correlation is due to the direct effect with TG. The coefficients of determination of the full hypothesized models were 0.664, 0.763 and 0.590 for BC1, loblolly and slash pine, respectively. While these values indicated that the path analysis models fit the data reasonably well, we cannot prove causation. These high coefficients of determination were an indication that the biological assumptions considered to construct the path analysis may be valid; however, other models and assumptions may fit the data properly (Bollen 1989), i.e. length of branch could have been included in the analysis to improve the fit of the model. In summary, the use of path analysis allowed differentiation of the influence of different traits on total growth for the three taxa studied. These results indicated that all crown traits considered had a moderate effect on total growth with the exceptions of SLA that had a low effect and CPA that ha d a consistently large effect on all taxa. In the case of phenological traits, duration of growth had a negligible effect on total growth on slash and loblolly, and a small effect on the backcross. Finally, path analysis enhanced
57 the findings of correlation analysis by providing a holistic view of trait associations which would not have been otherwise possible from the use of correlation coefficients only.
58 CHAPTER 4 CONCLUSIO NS Significant differences among the loblolly, slash and the pseudo-backcross families included in the study were detected for growth, phenological, crown architectural and needle characteristics during the first growing season at the High Springs site where the pseudo-backcross and the third-cycle slash pine had better performance for height growth. Positive heterosis was found for all height growth characteristics and number of needles per fascicle while negative heterosis was present for base diameter, t op diameter, fascicle length, fascicle diameter crown projected area and all the phenological characteristics (with the exception of initiation). Average performance (no heterosis) was found for initiation of growth, branch traits, tip moth incidence, sheath length and specific leaf area. Results from multivariate analysis indicated that the nature of the intra-plant relationships changed with the taxa examined. That is, the existing relationship among traits in the pure species was not maintained in the pseudo-backcross; as a consequence, three correlation matrices (by taxa) were used to obtain path analyses. Path analysis allowed differentiating of hypothesized effect for the different traits, as a system, on total height growth for the three taxa studied. These results indicated that all crown traits considered had a moderate effect on total growth with the exception of specific leaf area that had a small effect and crown projected area that ha d a consistently large effect on all taxa The phenological traits had small effects on growth for the pseudo-backcross and negligible effects for slash and loblolly pine. The results of this study indicated that the introgression of loblolly pine alleles into a slash pine background was effective, and novel trait co mbinations were achieved. The
59 pseudo-backcross had larger variation in early height growth than the slash pine families with levels of observed variation closer to the loblolly pine family at the end of the season. The pseudo-backcross was also taller than all open-pollinated families and had lower tip moth incidence than the loblolly pine ancestor. The pseudo-backcross demonstrated efficien cy in terms of growth per unit of crown area when compared with the other families gr owing at similar rates to the thi rd -cycle slash pine (the next tallest ) while allocating less biomass to the crown. All these results indicated the potential of the pseu do -backcross in future breeding: trees with high growth level, less tip moth incidence, and small crown (less aggressive in term of competition with its neighbors). Future studies should explore stability among environments for the backcross The establishment of a series of trials including several backcrosses and its complete set of parents in at least three representative sites of the southern USA is recommended. Also, highly recommended is establishing a test in an area of marginality for the pure species in order to test the hypothes the hybrid should perform better than the pure species i n certain environments in that way identifying the deployment area. At present, tree improvement programs of the southeastern USA have material with high levels of improvement for both species. These could be utilized to obtain better hybrids by improving the performance of those traits that appeared to be highly influenced by the poor performance of the first-cycle slash ancestor observed in this study. The pseudo-backcross of this study could be introduced in the slash improvement program as an infusion of new material to increase the variability and introduction of new alleles of commercial interest. Finally, hybridization and
60 backcrossing have demonstrated high potential as a way to introduce novel traits from one species to the other. In this respect, hybrids can contribute by either increasing gr owth gains or maintaining current growth gains while improving other traits of commercial importance.
61 APPENDIX APPENDIX A SU PPLEMENTAL TABLES
62 Table A-1. P-values for the family differences and heterosis hypotheses, estimates of heterotic effect, least-square-means and two standard errors (in parenthesis) for phenological, growth traits, tip moth incidence, final height (ht_14) and growth curve parameters wth) from Gompertz function for slash pine families (Slash1_OP and Slash3_OP), a loblolly pine family (Lob_OP) and the pseudo-backcross (BC1) for the High Springs backcross study. Initiation (days) Cessation (days) 50% Growth (days) Duration (days) ARSE (cm day 1 ) Total Growth (cm) ht_14 (cm) Tip moth (%) Parameter a Parameter c P value Family <0.0001 <0.0001 <0.0001 <0.0001 <0.0001 <0.0001 <0.0001 <0.0001 <0.0001 <0.0001 hypothesis Heterosis 0.386 <0.0001 0.0051 <0.0001 <0.0001 0.0123 <0. 0001 0.364 0.0029 0.0002 hypothesis Estimated means Lob_OP 59.2 (0.88) 274.0 (3.1) 167.6 (5.1) 214.6 (3.2) 0.11 (0.01) 24.5 (1.3) 53.6 (1.3) 54 (5.8) 67.90 (3.10) 0.008 (0.0007) BC1 57.5 (0.74) 270.7 (2.7) 144.0 (4.7) 212.9 (2.6) 0.12 (0.01) 26.1 (1.1) 59.3 (1.1) 20 (4.8) 68.47 (2.01) 0.010 (0.0007) Slash1_OP 54.6 (0.78) 270.2 (3.5) 123.2 (5.4) 215.4 (3.3) 0.09 (0.01) 18.3 (1.0) 49.3 (1.1) 11 (4.7) 51.90 (1.51) 0.013 (0.0008) Slash3_OP 57.4 (0.83) 281.3 (2.9) 1 56.1 (5.7) 223.9 (2.9) 0.12 (0.01) 27.7 1.2) 58.9 (1.4) 10 (5.1) 70.56 (2.90) 0.008 (0.0006) Expected 57.2 276.7 150.8 219.5 0.11 24.6 55.2 21 65.23 0.009 value BC1 Heterosis 0.36 5.97 6.76 6.51 0.01 1.5 4.13 0.12 3.25 0.0013 gene effect # trees in 1,295 1,295 1,302 1,296 1,299 1,298 1,300 1,505 1,302 1,302 analysis ARSE = average rate of shoot elongation The expected value for the backcross was calculated as follow: Minor gene effect (tip moth) H 0 : BC1 = 0.5*[(Lob_OP + Slash1_OP)/2] + 0.5*[Slash3_OP] Major gene effect H 0 : BC1 = 0.25*[Lob_OP] + 0.75*[(Slash1_OP + Slash3_OP)/2] Heterosis gene effect= Estimated mean BC1 Expected mean BC1
63 Table A-2. P-values for the family differences and heterosis hypotheses, estimate of heterotic effect, least-square-means and two standard errors for crown architectural traits and ratios with final height for slash pine families (Slash1_OP and Slash3_OP), a loblolly pine family (Lob_OP) and the pseudo-backcross (BC1) for the High Springs backcross study. Number nodes Number branches base diameter (mm) top diameter (mm) Diameter/ ht Taper Number nodes/ht Number branches/ht Number branches/ Number nodes Crown projected area (cm 2 ) P value Family <0.0001 < 0.0001 <0.0001 <0.0001 <0.0001 <0.0001 <0.0001 <0.0001 <0.0001 <0.0001 hypothesis Heterosis 0.4605 0.4366 <0.0001 <0.0001 <0.0001 0.8663 0.0175 0.0055 0.2031 0.0001 hypothesis Estimated means Lob_OP 3 .4 (0.15) 7.72 (0.39) 13 .4 ( 0. 47) 3 8 (0.1 4) 2.49 (0.06) 0.30 (0.008) 0.06 (0.0024) 0.14 (0.01) 2.28 (0.08) 638.8 (41.8) BC1 3.33 (0.15) 6.88 (0.36) 13 8 ( 0. 45) 5.1 (0.1 9) 2.34 (0.06) 0.38 (0.010) 0.06 (0.0026) 0.12 (0.01) 2.07 (0.10) 670.0 (41.9) Slash 1_OP 2.31 (0.13) 4.53 (0.35) 13.5 ( 0. 44) 5.9 (0.1 6) 2.74 (0.05) 0.45 (0.008) 0.05 (0.0022) 0.09 (0.01) 1.89 (0.07) 528.7 (38.7) Slash 3_OP 3.7 (0.16) 7.94 (0.45) 17.8 ( 0.5 7) 6.9 ( 0. 22) 3.03 (0.07) 0.39 (0.011) 0.06 (0.0028) 0.13 (0.01) 2.14 (0.08) 921.6 (4 8.4) Expected 3.28 7.03 15 .6 5 .8 2.83 0.38 0.06 0.13 2.11 752.6 value BC1 Heterosis 0.01 0.036 1 8 3 0. 76 0.4875 0.00 0.00025 0.0081 0.05 82.7 gene effect # trees in 1,333 1,331 1331 1,333 1,329 1,330 1332 1,332 1,327 1,315 analysis Taper=top diameter/base diameter, ht=final height (measure 14 th )
64 Table A-3. P-values for the family differences and heterosis hypotheses, estimate of heterotic effect, least-square-means and two standard errors for needles traits and crown projected area for slash pine families (Slash1_OP and Slash3_OP), a loblolly pine family (Lob_OP) and the pseudo-backcross (BC1) for the High Springs backcross study. SLA is specific leaf area. fascicle length (mm) sheath length (mm) needles per fascicle fascicle diameter (mm) SLA (cm*gr 1 ) P value Family hypothesis <.0001 <.0001 <.0001 <.0001 <.0001 Heterosis hypothesis <.0001 0.8757 0.017 <.0001 0.1577 Estimated means Lob_OP 143.8 (4.14) 6.2 (0.20) 3.0 4 (0.02) 1.75 (0.03) 50.8 (1.10) BC1 164.1 (3.71) 7.8 (0.19) 2.9 3 (0.02) 1.94 (0.02) 44.3 (1.00) Slash1_OP 182.8 (4.70) 8.2 (0.22) 2. 66 (0.03) 2.06 (0.03) 39.7 (1.06) Slash3_OP 199.8 (5.05) 8.5 (0.23) 2. 88 (0.03) 2.22 (0.03) 42.4 (1.02) Expected v alue BC1 181.5 7.8 2. 86 2.06 43.8 Heterosis gene effect 17.4 0.013 0.06 0.12 0.46 # trees in analysis 1,293 1,293 1,298 1,299 1,272
65 APPENDIX B CORRELATIONS AND EFFECT OF CROWN AND PHENOLOGICAL TRAITS ON TOTAL GROWTH Table B-1. Phenotypic correlations among crown traits, needle and phenological traits for (a) BC1; (b) Lob; (c) Slash. (a) BC1 NB IN DU CPA FL SLA FD NF NN 0.797 0.397 0.254 0.669 0.417 0.146 0.396 0.206 NB 0.323 0.174 0.670 0.331 0.064ns 0.305 0.183 IN 0.088ns 0.398 0.229 0.142 0.224 0.130 DU 0.293 0.407 0.079ns 0.376 0.092ns CPA 0.693 0.203 0.611 0.291 FL 0.320 0.781 0.228 SLA 0.171 0.188 FD 0.314 (b) Lob NB IN DU CPA FL SLA FD NF NN 0.782 0.200 0.002ns 0.628 0.415 0.15 9 0.292 0.181 NB 0.196 0.071ns 0.735 0.383 0.098ns 0.266 0.232 IN 0.126* 0.196 0.102ns 0.129* 0.087ns 0.107ns DU 0.049ns 0.139* 0.109ns 0.183 0.002ns CPA 0.579 0.195 0.433 0.233 FL 0.345 0.753 0.198 SLA 0.063ns 0.047ns FD 0.266 (c) Slash NB IN DU CPA FL SLA FD NF NN 0.751 0.157 0.066ns 0.546 0.203 0.001ns 0.168 0.206 NB 0.098* 0.137 0.601 0.219 0.068ns 0.129 0.175 IN 0.130 0.264 0.192 0.108* 0.123* 0.031ns DU 0.026ns 0.047ns 0.006ns 0. 012ns 0.018ns CPA 0.516 0.074ns 0.351 0.317 FL 0.195 0.472 0.072ns SLA 0.026ns 0.020ns FD 0.455 *=significant correlation (p<0.05); ns=no significant correlation; all others are highly significant (p<0.01) NN is number of nodes, NB is number of branches, IN is the dayof -year of initiation of growth, DU is the duration of the growth, FL is fascicle length, NF is number of needle per fascicle, FD is fascicle diameter, SLA is projected specific leaf area and CPA is crown projected area.
66 Table B-2. Direct, indirect and total effects of crown architectural traits (NN=number of nodes, NB=number of branches, FL=fascicle length, SLA=projected specific leaf area, FD=fascicle diameter, NF=number of needles per fascicle, CPA=crown projected area) and duration of the growth (DU) on total growth for the first year by taxa (BC1=backcross, Lob= loblolly pine and Slash= slash pine) BC1 Lob Slash Trait Direct Indirect Total Direct Indirect Total Direct Indirect Total NN 0.121 0.431 0.552 0.030 0.519 0.549 0.114 0.314 0.429 NB 0.308 0.245 0.553 0.501 0.139 0.640 0.284 0.185 0.469 FL 0.355 0.216 0.571 0.273 0.233 0.506 0.298 0.115 0.413 SLA 0.006 0.040 0.034 0.020 0.099 0.119 FD 0.068 0.436 0.504 0.023 0.3 53 0.377 0.064 0.263 0.326 NF 0.057 0.182 0.240 0.020 0.182 0.202 0.146 0.102 0.248 CPA 0.825 0.010 0.815 0.874 0 0.874 0.768 0 0.768 DU 0.035 0.242 0.207 .
67 LIST OF REFERENCES Anderson, E. and Hubricht, L. 1938. Hybridization in tradescantia. III. The evidence for introgressive hybridization. Am. J. Bot. 25(6): 396 402. Babar, M., Khan, A.A., Arif, A., Zafar, Y., and Arif, M. 2007. Path analysis of some leaf and panicle traits affecting grain yield in doubled haploid lines of rice (Oryza sativa L.). J. Agric. Res. 45 (4): 245 252. Barnes, R.D., and Mullin, L.J. 1978. Three-year height performance of Pinus elliotii Engelm. var. elliotii x P. taeda L. hybrid families on the three sites in Rhodesia. Silvae Genet. 27 : 217 223. Bidgoli, A.M., Akbari, G.A., Mirhadi, M.J., Zand, E., and Soufizadeh, S. 2006. Path analysis of the relationships between seed yield and some morphological and phenological traits in safflower (Carthamus tictorius L.). Euphytica, 148 (3): 261 268. Bollen, K.A. 1989. Model notation, covariances, and path analysis. In Structural equations with latent variables. Wiley, New York, USA. pp. 10 39. Borders, B.E., and Harrison, W.M. 1989. Comparison of slash pine and loblolly pine performance on cutover site-prepared sites in the coastal plain of Georgia and Florida. South. J. Appl. For. 13 (4): 204 207. Bouchez, A., Hospital, F., Causse, M., Gallais, A., and Charcosset, A. 2002. Markerassisted introgression of favorable alleles at quantitative trait loci between maize elite lines. Genetics, 162 : 1945 1959. Bruce, A.B. 1910. The Mendelian theory of heredity and the augmentation of vigor. Science, 32: 627 628. Cade, S.C., and Hedden, R.L. 1987. Growth impact of pine tip moth on loblolly pine plantations in the Ouachita Mountains of Arkansas. South J. Appl. For. 11 : 128 133. Campinhos, E.N., Peters-Robinson, I., Bertolucci, F.L., and Alfenas, A.C. 1998. Interspecific hybridization and inbreeding effect in seed from a Eucalyptus grandis x E. urophylla clonal orchard in Brazil. Genet. Mo l. Biol. [online]. 21 (3) [accessed 29 November 2007] Available from: http://www.scielo.br/scielo.php?script=sci_arttext&pid=S141547571998000300014&lng=en&nrm=iso Chmura, D.J., Rahman, M.S., and Tjoelker, M.G. 2007. Crown structure and biomass allocation patterns modulate aboveground productivity in young loblolly pine and slash pine. Forest Ecol. Manag. 243 : 219 230.
68 Colbert, S.R., Jokela, E.J., and Neary, D.G. 1990. Effects of annual fertilization and sustained weed control on dry matter partitioning leaf area, and growth efficiency of juvenile loblolly and slash pine. For. Sci. 36 : 995 1014. Cramer, C.S., and Wehner, T.C. 1998. Fruit yield component means and correlations of four slicing cucumber populations improved through six to ten cycles of recurrent selection. J. Amer. Soc. Ho rt Sci. 123 (3): 388 395. Dalla-Tea, F., and Jokela, E.J. 1991. Needlefall, canopy light interception, and productivity of young intensively managed slash and loblolly pine stands. For. Sci. 37: 1298 1313. Danjon, F. 1994. Heritabilities and genetic correlations for estimated growth curve parameters in maritime pine. Theor. Appl. Genet. 89 : 911 921. Dewey, D.R., and Lu, K.H. 1959. A correlation and path-coefficient analysis of components of crested wheatgrass seed production. Agron. J. 51 : 515 518. Dorman, K.W. 1976. The genetics and breeding of southern pines. U.S. Department of Agriculture, Agriculture Handbook 471. Duarte, R.A., and Adams, M.W. 1972. A path coefficient analysis of some yield components interrelations in field beans ( Phaseolus vulgaris L.). Crop Science, 12: 579 582. Ehrenberg, C.E. 1963. Genetic variation in progeny tests of Scot pines (Pinus sylvestris L.). Studia Forestalia Suecica, 10 : 1 35. Emhart, V.I., Martin, T.A., White, T.L., and Huber, D.A. 2006. Genetic variation in basal area increment phenology and its correlation with growth rate in loblolly and slash pine families and clones. Can. J. For. Res. 36 : 961 971. Emhart, V.I., Martin, T.A., White, T.L., and Huber, D.A. 2007. Clonal variation in crown structure, absorbed photosynthetically active radiation and growth of loblolly pine and slash pine. Tree Physiol. 27 (3): 421 430. Falconer, D.S., and Mackay, T. 1996. Introduction to quantitative genetics. Fourth Edition. Longman Inc. New York, USA. Fowler, D.P. 1978. Population improvement and hybridization. Unasylva, 30: 21 26. Franco, G.A.A., Wang, S., Melchinger, A.E., and Zeng, Z. 2008. Quantitative trait loci mapping and the genetic basis of heterosis in maize and rice. Genetics, 180 : 1707 1724. Gezan, S.A., Huber, D.A., Medina, A.M., Parisi, L.M., and Powell, G.L. 2004. Cooperative Forest Genetics Research Program. 46th Progress Report. Univer si ty of Florida, Gainesville, FL.
69 Gezan, S.A., Huber, D.A., Medina, A.M., Parisi, L.M., and Powell, G.L. 2005. Cooperative Forest Genetics Research Program. 47th Progress Report. University of Florida, Gainesville, FL. Gilmour, A.R., Gobel, B.J., Cullis, B.R., and Thompson, R. 2006. ASReml user guide release 2.0. VSN International Ltd. Hemel Hempstead, HP1 1ES, UK. Gonzalez, C. 2008. Interactions between xylem structure and water relations of southern pines. PhD thesis, School of Forest Resources and Conservation, University of Florida, Gainesville, FL. Govindaraju, D.R. 1984. Path analysis of the changes in quantitative characters in jack pine. New Phytol. 97 : 691 696. Hanover, J.W. 1963. Geographic variation in ponderosa pine leader growth. For. Sci. 9 (1): 86 95. Harding, K.J., and Copley, T.R. 2000. Wood property variation in Queensland-grown slash x caribbean pine hybrids. In Proceedings of QFRI/CRC-SPF Symposium: Hybrid Breeding and Genetics of Forest Trees, Noosa, Queensland, Australia, 914 April 2000. Edited by Heidi S. Dungey, Mark J. Dieters, and D. Garth Nikles. Department of Primary Industries, Brisbane, pp. 160 167. Huber, D.A., White, T.L., Powell, G.L., Rockwood, D., and Atwood, R. 2000. Cooperative Forest Genetics Research Program. 42th Annual Report. University of Florida, Gainesville, FL. Huber, D.A., Parisi, L.M., Powell, G.L., and Li, X. 2007. Cooperative Forest Genetics Research Program. 49th Annual Report. University of Florida, Gainesville, FL. Jayawickrama, K.J.S., McKeand, S.E., and Jett, J.B. 1998. Phenological variation in height and diameter growth in provenances and families of loblolly pine. New Forest, 16: 11 25. Jokela, E.J., and A.J. Long. 2000. Using soils to guide fertilizer recommendations for southern pines. Florida Cooperative Extension Service Circular 1230. King, N.T., Seiler, J.R., Fox, T.R., and Johnsen, K.H. 2008. Post-fertilization physiology and growth performance of loblolly pine clones. Tree Physiol. 28 : 703 711. Kinghorn, B.P. 2000. Crossbreeding strategies to maximise economic returns In Proceedings of QFRI/CRC-SPF Symposium: Hybrid Breeding and Genetics of Forest Trees, Noosa, Queensland, Australia, 9-14 April 2000. Edited by Heidi S. Dungey, Mark J. Dieters, and D. Garth Nikles. Department of Primary Industries, Brisbane, pp. 291 302. Kraus, J.F. 1986. Breeding shortleaf x loblolly pine hybrids for the development of fusiform rust-resistant loblolly pine. South J. Appl. For. 10 : 195 196.
70 Kremer, A. 1985. Component analysis of height growth, compensation between components and seasonal stability of shoot elongation in maritime pine (Pinus pinaster Ait.). In Proceedings on managing forest trees as cultivated plants: crop physiology of forest trees, Helsinki, Finland 23-28 July 1984 Edited by Peter M.A. Tigerstedt, Pasi Puttonen, and Veikko Koski.: University of Helsinki, Dept. of Plant Breeding, pp. 303 217. Kremer, A., and Larson, P.R. 1983. Genetic control of height growth components in jack pine seedlings. For. Sci. 29 (3): 451 464. La Farge, T. and Kraus, J. F. 1980. A progeny test of (shortleaf x loblolly) x loblolly hybrids to produce rapid growing hybrids resistant to fusiform rust. Silvae Genet. 29: 197 200. Littell, R.C., Stroup, W.W., and Freund, R.J. 2002. SAS for linear models, 4th ed., SAS Institute, Inc., Cary, NC. Lopez-Upton, J. 1999. Early growth, pest incidence, and cold hardiness of loblolly pine, slash pine, and some slash pine hybrids. PhD thesis, School of Forest Resources and Conservation, University of Florida, Gainesville, FL. Lpez-Upton, J., White, T.L., and Huber, D.A. 1999. Taxon and family differences in survival, cold hardiness, early growth, and rust incidence of loblolly, slash pine and some pine hybrids. Silvae Genet. 48(6): 303 313. Lopez-Upton, J., Blakeslee, G.M., White, T.L., and Huber, D.A. 2000. Effects of the cultural treatments and genetics on tip moth infestation of loblolly pine, slash pine and some slash pine hybrids. For. Gen. 7 (4): 275 286. Lundquist, J.E. 2000. A method of estimating direct and indirect effects of armillaria root disease and other small-scale forest disturbances on canopy gap size. Forest Science, 46(3): 356 362. Lynch, M., and Walsh, B. 1998. Path analysis. In Genetics and analysis of quantitative traits. Sinauer, Sunderland, MA. pp. 823 834. Major, J.E., Mosseler, A., Barsi, D.C., Campbell, M., and Rajora, O.P. 2003. Morphometric, allometric, and developmentally adaptive traits in red spruce and black spruce. II. Seedlings and mature tree assessment of controlled intraand inter-specific hybrids. Can. J. For. Res. 33 : 897 909. Marron, N., and Ceulemans, R. 2006. Genetic variation of leaf traits related to productivity in a populus deltoid es x populus nigra family. Can. J. For. Res. 36 : 390 400. Martin, T.A., and Jokela, E.J. 2004. Developmental patterns and nutrition impact radiation use efficiency components in southern pine stands. Ecol. Appl. 14 : 1839 1854.
71 Martin, T.A., and Jokela, E.J. 2004b. Stand development and production dynamics of loblolly under a range of cultural treatments in north-central Florida USA. Forest. Ecol. Manag. 192 : 39 58. McGarvey, R.C., Martin, T.A., and White, T.L. 2004. Integrating within-crown variation in net photosynthesis in loblolly and slash pine families. Tree Physiol. 24 : 1209 1220. McKeand, S., T. Mullin, T. Bryan and T. White. 2003. Deployment of genetically improved loblolly and slash pines in the south. J. For. 101 (3): 32 37 Mirov, N.T., Duffield, J.W., and Liddicoet, A.R. 1952. Altitudinal races of Pinus ponderosa a 12-year progress report. J. For. 50: 825 831. Murthy, R., and Dougherty, P.M. 1997. Estimating foliage area of loblolly pine shoots. For. Sci. 43 :299 303. Myburg, A.A., Griffin, A.R., Sederoff, R.R. and Whetten, R. 2000. Genetic analysis of interspecific backcross families of a hybrid of eucalyptus grandis and eucalyptus globulus. In Proceedings of QFRI/CRC-SPF Symposium: Hybrid Breeding and Genetics of Forest Trees, Noosa, Queensland, Australia, 9-14 April 2000. Edited by Heidi S. Dungey, Mark J. Dieters, and D. Garth Nikles. Department of Primary Industries, Brisbane, pp. 462 466. Namkoong, G., and Kang, H. C. 1990. Quantitative genetics of forest trees. In: Plant Breeding Reviews, Portland Ore. 9 : 139 188. Namkoong, G., and Matzinger, D.F. 1975. Selection for annual growth curves in Nicotiana-tabacum L. Genetics, 81 : 377 386. NCDC (National Climatic Data Center). U.S. Climate Normals 1971-2000 Products Clim20. Monthly Station Climate Summaries. Climatography of the United States. No.20. [online]. Available from http://www.ncdc.noaa.gov/oa/climate/normals/usnormalsprods.html [accessed 29 November 2007]. Nikles, D.G. 2000. Experience with some Pinus hybrids in Queensland, Australia. In Proceedings of QFRI/CRC-SPF Symposium: Hybrid Breeding and Genetics of Forest Trees, Noosa, Queensland, Australia, 9-14 April 2000. Edited by Heidi S. Dungey, Mark J. Dieters, and D. Garth Nikles. Department of Primary Industries, Brisbane, pp. 27 43. Nikl es D.G. and Robinson, M.J., 1989. The development of Pinus hybrids for operational use in Queensland. In Proceedings of IUFRO Conference: Breeding tropical trees: population structure and genetic improvement strategies in clonal and seedling forestry, Pattaya, Thailand 18 November-3 December 1988 Edited by Gibson, G.L., Griffin, A.R., Matheson, A.C. pp. 272 282.
72 Parisi, L.M. 2006. Shoot elongation patterns and genetic control of second-year height growth in Pinus taeda L. using clonally replicated trials. M.Sc. thesis, School of Forest Resources and Conservation, University of Florida, Gainesville, FL. Rasband, W.S. 1997-2005. ImageJ. U.S. National Institutes of Health, Bethesda, Maryland, USA, [online] Available from http://rsb.info.nih.gov/ij/ [accessed 26 January 2009]. Richardson, D.M. 1998. Ecology and biogeography of Pinus. Cambridge New York, NY, USA : Cambridge University Press. Roth, B.E., Jokela, E.J., Martin, T.A., Huber, D.A., and White, T.L. 2007. Genotype x environment interactions in selected loblolly and slash pine plantations in the southeastern United States. For. Ecol. Manag. 238 : 175 188. SAS Institute Inc. 2002-2003. SAS 9.1 Help and Documentation, Cary, NC: SAS Institute Inc. Shull, G.H. 1908. The composition of a field of maize. Am. Breeders Assoc. Rep. 4 : 296 301. Sit, V., and Poulin-Costello, M. 1994. Catalogue of curves for curve fitting. Biometrics information handbook series. Handbook 4. Ministry of Forests, Victoria, BC, Canada. Stephen, F.M., Wallis, G.W., Colvin, R.J., Young, J.F., and Warren, L.O. 1982. Pine tree growth and yield: Influence of species, plant spacing, vegetation, and pine tip moth control. Ark. Farm. Res. 31: 10. Sterck, F.J., Martinz-Ramos, M., Dyer-Leal, G., Rodrquez-Velazquez, V., and Poorter, L. 2003 The consequences of crown traits for the growth and survival of tree saplings in a Mexican lowland rainforest. Funct. Ecol. 17 : 194 200. Strickland, R., and Goddard, R. 1965. Inheritance of branching and crown characteristics in slash pine. In proceeding 8 th Southern Conference on Forest Tree Improvement, June 16-17. Savannah, Georgia USA. pp. 57 63. Sweda, T. 1984. Theoretical growth equations and their applications in forestry. Nagoya Univ. For. Nagoya, Japan, Bull. Nagoya Univ. For. 7. Thomas, B.P., Cummings, E. and Wittstruck, W.H. 1985. Soil survey of Alachua County, Florida. US Department of Agriculture. SCS. Van Der Sijde, H.A., and Roelofsen, J.W., 1986. The potential of pine hybrids in South Africa. S. Afr. For. J. 136 : 5 14. W ear D.N., and Greis, J.G. 2002. Southern Forest Resource Assessment: Summary of findings. J. Forest. 100 (7): 6 14.
73 Weisberg, P.J. 2004. Importance of non-stand-replacing fire for development of forest structure in the Pacific Northwest, USA. For. Sci. 50(2): 245 258. White, T.L., Hodge, G.R., and Powell, G.L. 1993. An advanced-generation tree improvement plan for slash pine in the southeastern United States. Silvae Genet. 42: 359 371. Wh ite, T.L., Adams, W.T., and Neale, D.B. 2007. Forest genetics. CABI Publishing. Wilkinson, L., Blank, G., and Gruber. C. 1996. Desktop data analysis with SYSTAT. Prentice-Hall, Upper Saddle River, NJ. Will, R.E., Barron, G.A., Burkes, E.C., Shiver, B., and Teskey, R.O. 2001. Relationship between intercepted radiation, net photosynthesis, respiration, and rate of stem volume growth of Pinus taeda and Pinus elliottii stands of different densities. For. Ecol. Manag. 154 : 155 163. Williams, E.R., John, J.A., and Whitaker, D. 1999. Example of block designs for plan t and tree breeding trials. Aust. N. Z. J. Stat. 41 (3): 277 284. Winsor, C.P. 1932. The Gompertz curve as a growth curve. Proc. Natl. Acad. Sci. 18 : 1 8. Wolfinger, R.D. 1996. Heterogeneous variance-covariance structures for repeated measures J Agric. Biol. Environ. Stat. 1 (2): 205 230. Wright, J.W. 1976. Introduction to forest genetics. Department of Forestry Michigan State University. East Lansing, Michigan. Academic Press. (AP) New York. NY. Wr ight, S. 1921. Correlation and causation. J. Agric. Res. 20: 557 585. Wright, S. 1960. Path Coefficients and Path Regression: Alternative or Complementary Concepts? Biometrics, 16 : 189 202. Xiao, Y., Jokela, E., and White, T.L. 2003. Species differences in crown structure and growth performance of juvenile loblolly and slash pine. For. Ecol. Manag. 174 : 295 313. Zeide, B. 1993. Analysis of growth equations. For. Sci. 39 (3): 594 616. Zobel, B., and Talbert, J. 1984. Applied Forest Tree Improvement. John Wiley an d Sons, Inc., New York, N Y.
74 BIOGRAPHICAL SKETCH Patricio R. Munoz was born in Santiago, Chile. He lived is childhood in Pirque, a town located at the foot of the Andes mountains. H e attended high school at the Internado Nacional Barros Arana (INBA), a school located in the center of the capitol. H e then obtain ed the title of Forestry Engineer from the Universidad Catolica de Temuco, located in southern Chile. He worked for Forestal Mininco (CMPC Forestal) for about two years as a data manager in the breeding and tree improvement program before coming to the University of Florida.