American Journal of Botany 100(10): 1923–1935. 2013.
SPECIAL INVITED PAPER
BRANCH ARCHITECTURE IN GINKGO BILOBA: WOOD ANATOMY AND LONG SHOOT–SHORT SHOOT INTERACTIONS1
STEFAN A. LITTLE2,3,6, BROOKE JACOBS2, STEVEN J. MCKECHNIE2, RANESSA L. COOPER4, MICHAEL L. CHRISTIANSON5, AND JUDITH A. JERNSTEDT2 2Department
of Plant Sciences, One Shields Avenue, University of California, Davis, California 95616, USA; 3Centre for Forest Biology, Department of Biology, University of Victoria, Victoria, British Columbia, V8W 3N5, Canada; 4Biology Department, Hillsdale College, 33 E. College St., Hillsdale, Michigan 49242, USA; and 5Department of Plant and Microbial Biology, University of California, Berkeley, California 94704, USA
• Premise: Ginkgo, centrally placed in seed plant phylogeny, is considered important in many phylogenetic and evolutionary studies. Shoot dimorphism of Ginkgo has been long noted, but no work has yet been done to evaluate the relationships between overall branch architecture and wood ring characters, shoot growth, and environmental conditions. • Methods: Branches, sampled from similar canopy heights, were mapped with the age of each long shoot segment determined by counting annual leaf-scar series on its short shoots. Transverse sections were made for each long shoot segment and an adjacent short shoot; wood ring thickness, number of rings, and number of tracheids/ring were determined. Using branch maps, we identified wood rings for each long shoot segment to year and developmental context of each year (distal short shoot growth only vs. at least one distal long shoot). Climate data were also analyzed in conjunction with developmental context. • Key results: Significantly thicker wood rings occur in years with distal long shoot development. The likelihood that a branch produced long shoots in a given year was lower with higher maximum annual temperature. Annual maximum temperature was negatively correlated with ring thickness in microsporangiate trees only. Annual minimum temperatures were correlated differently with ring thickness of megasporangiate and microsporangiate trees, depending on the developmental context. There were no significant effects associated with precipitation. • Conclusions: Overall, developmental context alone predicts wood ring thickness about as well as models that include temperature. This suggests that although climatic factors may be strongly correlated with wood ring data among many gymnosperm taxa, at least for Ginkgo, correlations with climate data are primarily due to changes in proportions of shoot developmental types (LS vs. SS) across branches. Key words: branch architecture; climate; Ginkgo biloba; Ginkgoaceae; mean annual maximum temperature; mean annual minimum temperature; mean annual precipitation; shoot dimorphism; wood development.
Ginkgo biloba L. (Ginkgoaceae) is the single extant species of the order Ginkgoales with a fossil record that extends into the Permian (Seward and Gowan, 1900; Sprecher, 1907; Zhou, 1991, 2009; Zhou and Zheng, 2003; Zhou et al., 2007; Feng et al., 2010). The modern genus is present in the Lower Cretaceous and is often deemed a “living fossil” because it has undergone little apparent morphological change over 100 million years (Royer et al., 2003; Zheng and Zhou, 2004; Feng et al., 2010). As the only ginkgoalean representative of the five main 1 Manuscript received 1 April 2013; revision accepted 7 August 2013. A portion of this work was presented at the Katherine Esau Research Symposium, “Integrating Plant Structure with Function, Development and Evolution,” March 29, 2012, University of California, Davis. We thank Professor Patrick von Aderkas (University of Victoria) for helpful comments on this manuscript, Ellen Roundey for assistance with wood ring observations, Megan Saunders for electron microscopy assistance, and Angie Girdham for plant collection of Ginkgo from Hillsdale, Michigan. We acknowledge support from the Katherine Esau Postdoctoral Fellowship (SAL) and the Grady L. Webster Memorial Research Fund (JAJ). 6 Author for correspondence (e-mail:
[email protected])
doi:10.3732/ajb.1300123
extant seed plant lineages, and because of its central placement in seed plant phylogeny, studies of Ginkgo provide key insights into the development and evolution of seed plants. Our work follows in the extensive tradition of investigations into the distinctive vegetative morphology of Ginkgo (Seward and Gowan, 1900; Sprecher, 1907; Tupper, 1911; Sakisaka, 1928; Gunckel and Wetmore, 1946a, b; Gunckel et al., 1949; Gunckel and Thimann, 1949; Arnott, 1959; Critchfield, 1970; Hoddinott and van Zinderen Barker Jr., 1974; Mundry and Stutzel, 2004; Del Tredici, 1991; Christianson and Jernstedt, 2009; Christianson and Niklas, 2011; Leigh et al., 2011; Niklas and Christianson, 2011; Rudall et al., 2012). Many examinations of the vegetative morphology of Ginkgo have focused on the characteristic shoot dimorphism: long shoots with spaced nodes that extend the canopy, and primarily unbranched short shoots lacking internode elongation and with axillary reproductive structures. Some of the earliest observations of Ginkgo recorded anatomical differences between shoot types (Seward and Gowan, 1900; Sprecher, 1907; Coulter and Chamberlain, 1917; Sakisaka, 1928; Chamberlain, 1935). Short shoots have more primary ground tissue and less dense xylem with more parenchymatous cells. The suite of traits that characterize short shoots,
American Journal of Botany 100(10): 1923–1935, 2013; http://www.amjbot.org/ © 2013 Botanical Society of America
1923
1924
AMERICAN JOURNAL OF BOTANY
[Vol. 100
Fig. 1. Schematic diagram of an ultimate branch segment of Ginkgo biloba. Short shoots are often wider in diameter than the subtending long shoot. The blue boxes represent short shoots, and box lengths represent relative shoot ages. The thick green line represents long shoot segments in a branch. Long shoot segments are typically separated by zones of previous short shoot growth.
including suberized leaf bases, little to no periderm production, wide pith, manoxylic-like xylem (low density of tracheids/abundant parenchyma) and pachycaulous form (unbranched, often tubular to obconic), led several botanists to suggest that the Ginkgo short shoot was anatomically similar to a cycad main axis (Sakisaka, 1928; Chamberlain, 1935; Christianson and Jernstedt, 2009). Later developmental studies compared the shoot apical meristem organization in Ginkgo with that of cycads, finding that both fit Foster’s (1938) zonate model later considered typical of all extant gymnosperms (Gunckel and Wetmore, 1946a; Gunckel and Wetmore, 1946b; Gunckel et al., 1949; Gunckel and Thimann, 1949). These studies concluded that the developmental features examined (i.e., apical organization and origin of primary tissues) were fundamentally similar in long and short shoots. This subsequent downplaying of the cycad-like peculiarity of short shoots has continued generally and may be explained in part by the assignment of Ginkgo as a coniferophyte in the dominant phylogenetic hypotheses prevailing since the mid-1900s (Crane, 1985; Doyle and Donoghue, 1992; Rothwell and Serbet, 1994; Doyle, 2006; Hilton and Bateman, 2006; Rydin and Korall, 2009; Mathews, 2009; Mathews et al., 2010). More recently, there has been renewed interest in characterizing differences between short and long shoots, mostly with regard to leaf traits (Critchfield, 1970; Hoddinott and van Zinderen Barker, 1974; Mundry and Stutzel, 2004; Guo et al., 2005; Christianson and Niklas, 2011; Leigh et al., 2011; Niklas and Christianson, 2011). In particular, specific leaf area, the scaling of specific leaf area, and modeled leaf hydraulics differ significantly between shoot types and between mega- and microsporangiate trees (Christianson and Niklas, 2011; Niklas and Christianson, 2011). Compared with the long history of research on shoots of Ginkgo, investigation of larger scale branch architecture remains relatively uninvestigated (Del Tredici, 1991), and to date, there have been no attempts to integrate morphological, anatomical, and developmental data in the context of branch architecture. The aim of this work is to analyze morphological and anatomical observations of shoot dimorphism and their relationship with branch architecture using a quantitative approach. We are interested in understanding whether the occurrence and proximity of pachycaulous short shoots (Seward and Gowan, 1900; Sakisaka, 1928) is related to wood ring thickness in proximal long shoot wood, and what relationship, if any, climate data has with wood ring thickness in long shoots. Our work addresses the following questions: (1) How does wood ring thickness
differ between years in which branch growth is allocated to canopy expansion (long shoot development) or canopy maintenance (short shoot growth only)? (2) What is the relationship of interannual climatic variation with the production of long and short shoots in branches from micro- and megasporangiate trees? (3) What is the relationship between interannual variation in climate data and patterns of branch wood ring thickness? Our study shows that proximal long shoot wood ring thickness differs between years in which there is distal short shoot growth only and years with at least one distal long shoot. These differences in shoot developmental type and branch architecture influence long shoot wood deposition. Interannual variation in temperature interacts in a complex way with these structural and developmental differences, and mega- vs. microsporangiate trees show differential relationships of wood ring thickness with climate data.
Fig. 2. Two short shoots of Ginkgo biloba showing external features that allow age determination of the axis independent of wood anatomy. White arrows indicate examples of persistent outer bud-scales; blue arrowheads indicate examples of the seam-like zone where ephemeral bracts abscised in spring; plain white lines indicate ovulate stalk scars. The photographs also highlight the persistence of short shoots, indicated by the presence of lichen on the basal regions shown. Eight and nine annual cycles of growth are visible for the shoot on the right and left, respectively.
October 2013]
LITTLE ET AL.—GINKGO BRANCH WOOD
1925
MATERIALS AND METHODS Plant material—Structured sampling for statistical analyses involved the collection of 28 branches in June and July of 2011. Branches were sampled from six trees growing on the campus of the University of California, Davis (UCD): three trees growing near Hunt Hall (two microsporangiate trees and one megasporangiate tree), and three trees adjacent to Haring Hall (one microsporangiate tree and two megasporangiate trees). Each branch was collected with pole-pruners at approximately 3 m above ground, near the periphery of the canopy, and contained at least two long shoot segments per major branch axis. Additional branches were sampled from four trees on the Hillsdale College campus (Hillsdale, MI), although these supplemental branches were not included in statistical models of ring thickness. Instead, they were used to corroborate some of the observations made on samples collected at UCD including instances where the age of axis and wood ring number were not equal. Morphology and anatomy of mapped branches—According to our sampling design, each branch had a minimum of two long shoot segments along the major axis of the branch, although some branches had up to six long shoot segments. The diameters of the long and short shoots were measured using digital calipers for each mapped branch segment (Fig. 1). The age of each long shoot segment was determined by counting the annual cycles of leaf production by observation of the short shoot exterior (Sprecher, 1907; Christianson and Jernstedt, 2009). Three criteria were used to determine the boundaries of annual leaf cycles on a short shoot: (1) the presence of outer bud-scales which do not always abscise completely, (2) the seam-like line where the ephemeral bracts abscise in spring (these tend to abscise with pollen strobili toward the end of April in Davis), and (3) the disruption in parastichies in the phyllotaxy of the suberized leaf scars (Figs. 2, 3). These criteria were used to determine the age of the short shoots occurring along each long shoot segment, thus providing an assessment of the age of each long shoot axis bearing them, independent of the observation of wood rings. Age determinations were verified by counting leaf production cycles of multiple short shoots in each long shoot segment. This mapping of branches and age determination for each long shoot segment is possible because all axils on a long shoot form buds the same year that the long shoot forms. The only exception is the basal most one to three (typically two) axillary positions in the elongate portion of the long shoot, just above the basal pseudo-whorl of preformed leaves (Fig. 3; Critchfield, 1970; Christianson and Jernstedt, 2009). We have observed that these basal most axillary positions will form shoots, but only if there is damage to the long shoot. It is also important to note that the first formed (basalmost) foliar scar series in a short shoot will not be complete because axillary buds formed the same year as long shoot formation are proleptic, bearing only the scales and leaves for the following year’s growth. Anatomical observations were made approximately midway along each long shoot segment and on one short shoot adjacent to each long shoot transverse section (Figs. 3–5). All transverse sections were stained with 0.01% toluidine blue O, or phloroglucinol-HCL (Ruzin, 1999), mounted on a slide with coverslip and water, and observed with an Olympus BH-2 compound microscope (Olympus Optical, Tokyo, Japan) with a 10× objective lens. Images were captured using an Olympus MicroFire digital camera (Olympus America, Melville, New York, USA). Ring thickness and tracheid counts were measured in Adobe Photoshop CS5 (Adobe Systems, San Jose, California, USA). Larger scale images were captured with a Nikon D50 digital SLR camera (Nikon Corp., Tokyo, Japan), and on a dissecting scope also mounted with the MicroFire digital camera. All images were processed in Adobe Photoshop CS5, with linear adjustments applied across the whole image (i.e., sharpening, levels, color balance). Some composite images were assembled using Photoshop autoalign and auto-blend algorithms. Line figures were created directly in Adobe Illustrator CS5, or were exported from JMP (version 10.0.0 [2012], SAS Institute, Cary, North Carolina, USA) and modified in Adobe Illustrator CS5.
Fig. 3. Branch segments of Ginkgo biloba. Left, outer surface of branch segment, showing annual growth increments on short shoots and previous terminal short shoot; the annual increments were used to determine the age of the branch segments; note that the basalmost axil did not grow to form a short shoot. Right, phloriglucinol-HCL-stained median longitudinal cut surface shows wide pith of current and previous short shoots and contiguous piths between short shoots and their subtending long shoots. Scale bar = 5 mm.
1926
AMERICAN JOURNAL OF BOTANY
[Vol. 100
The range of ages for the branch segments observed was 1–39 yr. An age of 1 year indicated that the given long shoot had formed earlier in the spring of the sampling period (June and July of 2011). The 28 branches together represented 169 long shoot segments, with each having one or more rings, representing 884 long shoot wood rings. Of the 169 long shoot segments, 30 were shoots established the year of sampling, which had yet to establish well-developed axillary short shoots. Thus, only 136 short shoots were described. Using the maps of each branch made prior to anatomical observations, we determined the identity (short shoot or long shoot) of the distal stem increment at the time of formation for each long shoot wood ring (designated as developmental context). Each wood ring was categorized based on the presence of at least one long shoot having developed distal to that position in that year (LS) or the presence of only short shoot growth that year (SS). A ring produced the year after distal long shoot development was also classified for developmental context (ALS; after distal long shoot development). In some cases, the number of rings observed did not equal the independently determined age of the axis; these long shoot wood segments with “missing rings” could not be scored for developmental context. The exception to this case was on terminal branch segments; in these cases, the innermost ring produced was scored LS (first year formed as a long shoot), and all subsequent rings were recorded as SS years. Further characterization of branch anatomy with missing rings involved quantification of the radial number of tracheids in xylem. We counted radial tracheid numbers in a subsample of 56 short shoots with a single ring, all long shoots with missing rings (48), and a subsample of 76 long shoot segments with regular, annual rings. The number of tracheids per ring was estimated by counting cells that intercepted a radial transect through a given wood ring. The number of tracheids in a given transect was then divided by the age of the shoot to give a rough estimate of the number of tracheids produced over the shoot life span (tracheids/year). This value (tracheids/year) did not take into account differences in LS, ALS, or SS years. Scanning electron microscopy (SEM)—Several branches collected from trees in Michigan were observed using SEM. Transverse sections of Ginkgo shoots were adhered to aluminum stubs and sputter-coated with (gold (60 s at 30 mA) using a Cressington 108 auto sputter coater (Cressington Scientific Instruments, Watford, UK). The sections were then examined at 5 kV using a JEOL JSM-5510 scanning electron microscope (JEOL USA, Peabody, Massachusetts, USA). Statistical analyses—To characterize the relationship between variation in climate data and long shoot wood ring thickness in G. biloba, we downloaded 50 yr of annual maximum and minimum temperature and precipitation data from the United States Historical Climatological Network (US HCN; website http://www.ncdc.noaa.gov/oa/climate/research/ushcn/). The US HCN data provided mean monthly maximum and minimum temperatures, as well as total monthly precipitation for a weather station in Davis, CA (DAVIS 2 WSW EXP), within approximately 2 km of the trees sampled. Monthly means were averaged to produce mean annual temperature variables; precipitation was summed. For each year in which a ring was measured, we recorded the mean maximum and minimum temperature and total annual precipitation. Variation in mean long shoot wood ring thickness between tree type (microsporangiate vs. megasporangiate trees) and developmental context (LS, ALS, or SS)—The developmental context of individual long shoot wood rings (LS, ALS, or SS) could be definitively assigned for at least one long shoot segment in each of the 26 branches. The mean thickness of the wood rings in each category (LS, ALS, or SS) for each branch was included in this analysis. Terminal long shoot segments with missing rings were included, since the innermost ring represented the year that the axis developed into a long shoot (LS), and the subsequent rings could only represent years of distal short shoot growth (SS). Internal branch segments with missing rings could not be included in the analysis.
Fig. 4. Longitudinal section of a terminal short shoot of Ginkgo biloba, stained with phloriglucinol-HCL. Overall short shoot diameter, as well as pith diameter, increases toward the apex, with a slightly narrower subapical diameter. Leaf traces have not been sealed off by subsequent wood cambium (vascular cambium) development, even at base of the short shoot (oldest part of the axis). Scale bar = 2 mm.
October 2013]
LITTLE ET AL.—GINKGO BRANCH WOOD
1927
Fig. 5. Series of transverse sections of a Ginkgo biloba short shoot adjacent to that shown in Fig 4. Sections stained with phloriglucinol-HCL. Each transverse section is from a relative location (A, near apex of short shoot; F, at base of short shoot). Diameter decreases toward the base of the short shoot, both in overall diameter and pith diameter. Leaf traces (e.g., at arrows) have not been sealed off by subsequent vascular cambium development, even at base of short shoot (F, oldest part of the axis). Scale bar = 2 mm.
A mixed model ANOVA was used to test for differences in mean branch wood ring thickness between megasporangiate and microsporangiate trees and among shoot developmental contexts (proc GLM, SAS 2011). In this model, the mean long shoot wood ring thickness for each developmental context (LS, ALS, SS) per branch was included as the dependent variable (N = 75). The main fixed effects of tree type (mega- or microsporangiate tree), developmental context (LS, ALS, SS), and the interaction between tree type and developmental context were included as fixed effects. The main effect of branch (nested within
tree type) was included as a random effect. When a significant main effect of tree type, developmental context, or an interaction was detected, post hoc Tukey’s tests were used to compare means. Effect of annual variation in climate data and tree type on the likelihood of long shoot production—Two logistic regression models were used to investigate the effect of tree type (megasporangiate tree or microsporangiate tree) and variation in climate data on the likelihood that G. biloba will produce a long
1928
AMERICAN JOURNAL OF BOTANY
[Vol. 100
vs. microsporangiate). In a separate model, we tested for a relationship between proportion of shoot types produced (LS vs. SS) in each year, total annual precipitation, and tree type. In both analyses, years following long shoot production (ALS) were classified as short shoot years (SS). For this analysis, we simplified the classification of developmental context because this test was used to determine whether variation in climate data was a significant determinant of the probability that a given branch would have invested more in canopy maintenance (short shoot growth) or in canopy expansion (production of long shoots).
Fig. 6. Relationship between mean shoot diameter and shoot age for long shoots (blue diamonds and blue regression line) and short shoots (green squares and green regression line) of Ginkgo biloba. Means were calculated for each age class (long shoots N = 29, based on 169 axes; short shoots N = 27, based on 139 axes). Long shoot relationship, R2 = 0.56; short shoot relationship, R2 = 0.52. shoots in branches (proc LOGISTIC, SAS 2011). We tested for a relationship between proportion of shoot types produced (LS vs. SS) in each year, the annual maximum temperature, minimum temperature, and tree type (megasporangiate
Effect of interannual variation in climate data, tree type, and developmental context on ring thickness—It was necessary to assign precise ages to each wood ring observed in the analyses examining the relationship of variation in climate data with ring thickness. Because it was not possible to assign a year for each ring of long shoot branch segments determined to have missing rings, climate data values could not be assigned in these branches. Thus, 18 branches and 565 total wood ring observations were included in these models. Two mixed model ANCOVAs were used to assess the effect of interannual variation in precipitation data (first model) and maximum and minimum temperature data (second model) on branch wood ring thickness (proc GLM, SAS 2011). We included total annual precipitation in a separate model because it was highly correlated with both maximum (r = −0.6655) and minimum temperature (r = 0.6502). We conducted follow up analyses of the relationship between wood ring thickness and precipitation using monthly precipitation records. We chose to include average annual precipitation in the model after less conservative simple linear regression analyses of the relationship between wood ring thickness and average monthly precipitation for each month did not reveal significant relationships. Maximum and minimum temperatures were not highly correlated (r = −0.3030) and were included as covariates in a second ANCOVA model. In both models,
Fig. 7. Transverse sections of Ginkgo biloba long shoots and short shoots. A, B. Phloriglucinol-HCl-stained free-hand sections. C, D. SEM images of transverse cut faces. Each pair represents a set of branch segment observations. (A, C) Long shoots with growth rings. (B, D) Short shoots showing a single ring of xylem. Both pairs are from the same branch segment and are thus the same age; A, B = 8-yr-old axes; C, D = 15-yr-old axes, determined from annual leaf scar series on short shoots on the branch segment. Note: C and D are from a branch segment with nine missing rings in the long shoot wood. Scale bars = 100 µm.
October 2013]
LITTLE ET AL.—GINKGO BRANCH WOOD
Fig. 8. Box plots of radial tracheid density for Ginkgo biloba long shoots and short shoots. The first three box plots represent the number of tracheids along a stem radius, divided by age of the axis. The last box plot (far right) represents the number of radial tracheids in the single ring of short shoot xylem. Short shoots produce more radial tracheids per single ring than annual long shoot wood rings. Long shoots with missing rings have fewer radial tracheids/year than long shoots with annual rings. The number of radial tracheids/year in short shoots is similar to that of long shoots with missing rings. Missing rings are deduced by comparing observed ring number to the age of the branch segment as determined by counting annual leaf scar series of adjacent short shoots (N = 76, annual rings long shoot; 48, long shoot missing rings; 56 short shoots).
1929
Fig. 9. Mean wood ring thickness in branches from megasporangiate (red) and microsporangiate (blue) trees of Ginkgo biloba in years when at least one long shoot was produced distally (LS), the year after a distal long shoot was produced (ALS), and years in which only short shoots were produced distally (SS) on an individual branch. Different letters indicate significant differences in mean wood ring thickness. Least square means and standard errors are shown. (ANOVA, R2 = 0.88, model df = 31, error df = 633, F = 9.99, P < 0.0001).
Description of shoots across mapped branches— Plots of shoot diameter vs. shoot age corroborated preliminary observations that in parts of branches, short shoots may be thicker than their subtending long shoots (Figs. 3–6). Long shoots tend to have narrower diameters than short shoots in their first 1–10 yr. Axis age is a reasonable predictor of shoot diameter in both shoot types (Fig. 6), but short shoots only gradually increase in diameter over their life spans. Branch segments that represented previous short shoot growth (Figs. 1, 13) reflected the pattern expected from short shoots which
have a single xylem ring over their life spans, namely, a single ring for the period of time as a short shoot with additional rings that correspond to the years as a long shoot. The rings in this “previous short shoot” zone of the branch consist of the innermost ring with N rings to the outside, where N = the age of the distal long shoot segment. As an example, consider the branch segment in Fig. 13 with the 15-yr-old previous short shoot. The subtending 19-yr-old branch segment existed for 15 yr, then the terminal short shoot grew out 4 yr ago. This previous short shoot, with 15 leaf scar cycles, will have “1+4” rings—the innermost ring produced during the 15-yr short shoot growth period and four distinct rings from the 4 yr of long shoot growth. We determined the number of radial tracheids in (1) long shoot wood with the full complement of annual rings as deduced from overall branch architecture, (2) from long shoot wood with fewer rings than indicated from counting short shoot annual leaf scar series (missing rings), and (3) in the single ring of short shoots (Figs. 7, 8). The single ring of xylem in short shoots had more tracheids than the number of tracheids/year in both long shoots with annual rings and those with inferred missing rings (Fig. 8). Long shoots with annual rings had more tracheids/year than long shoots with missing rings, whereas the number of tracheids/year in the single xylem ring of short shoots was similar to that of long shoots with missing rings (Fig. 8).
TABLE 1.
TABLE 2.
the thickness of each long shoot ring was included as the dependent variable. Tree type, developmental context (LS, ALS, SS), and their interaction were included as fixed effects. Branch samples were nested within tree type and included as a random effect. To test for variation in the relationship between wood ring thickness and climate, we included all possible two- and three-way interactions between/among tree type, developmental context, and covariates in both models.
RESULTS
Mixed model ANOVA of mean long shoot wood ring thickness of Ginkgo biloba. Tree type (mega- vs microsporangiate tree), developmental context (distal long shoot growth vs. short shoot growth only), and their interaction were analyzed as fixed effects (R2 = 0.88, model df = 31, error df = 633, F = 9.99, P =
