127x Filetype PDF File size 3.21 MB Source: givnishlab.botany.wisc.edu
vol. 195, no. 6 the american naturalist june 2020 Historical Perspective The Adaptive Geometry of Trees Revisited Thomas J. Givnish* Department of Botany, University of Wisconsin–Madison, Madison, Wisconsin 53706 Submitted September 10, 2019; Accepted December 10, 2019; Electronically published April 7, 2020 abstract:TheAdaptiveGeometryofTreeshadanimportantcon- ences in canopy geometry maximized carbon acquisition ceptual influence on plant ecology and helped inspire many new undersunnyversusshadyconditions,thatasaresultmul- approaches to understanding succession, plant adaptation, and tilayers wouldhaveagrowthadvantageearlyinsuccession plant competition. Its central model provided an elegant potential andmonolayersanadvantagelater,andthatthedensityof explanation for how optimal canopy form should shift with ecolog- shadecastbyforestswouldperforceincreasethroughtime ical conditions, change those conditions through time, and thus after canopy closure. The Adaptive Geometry of Trees was helpdrivesuccessionandbeaconsequenceofit.Yetoncloseexam- aseminalandhighlycreativecontributiontoplantecology, ination, this deeply inspirational model does not lead to the predic- explaining how optimal plant form should shift with eco- tionsforwhichitiswidelyknown.HereIshowthattheHornmodel logical conditions, change those conditions through time, actually favors monolayer canopies over multilayers under all light conditions if relative growth rate (growth per unit investment) is andthusbothhelpdrivesuccession and be a consequence maximized. Horn’s conclusion that multilayers would be favored of it. over monolayers in brighter sites is an artifact. I propose that self- Horn’sslenderbookhadanoutsizedinfluenceonthink- shadingmultilayersmightgainanadvantageinbrightlylitsitesbyre- ingaboutoptimalitytheory,plantcompetition,andsucces- ducingwaterloss,reducingthecostsofbranchconstructionandmain- sion, rackingup1,224citations(GoogleScholar,April2019), tenance, reducing photoinhibition, increasing light capture in sidelit includingmanybyinfluentialpublications(e.g.,Grime1979; microsites,andincreasingwaterandnutrientsupplies(orleaflongev- Givnish 1982, 1988; Tilman 1987, 1994; Canham et al. 1990, ity) when combined with one or more of the previous potential ad- vantages.IconcludewithabriefdiscussionconnectingHorn’smodel 1994; Pacala et al. 1996; Weiher et al. 1998; Westoby et al. tootherconceptualframeworksinplantecologyandoutliningpossi- 2002 [all cited 1275 times each]). For those of us reading ble future extensions. it at the time, The Adaptive Geometry of Trees was highly Keywords: adaptation, monolayer, multilayer, growth maximiza- stimulatingbecauseitshowedhowsimpleprinciplesmight tion, optimality theory. lead to quantitative predictions of how competitively opti- malplantformshouldvarywithenvironmentalconditions, providingpotentialexplanationsforspeciesdistributionsin The greatest homage that can be paid to an empir- timeandspace,trait-environmentcorrelations,andtempo- ical theoryistheconstructivecriticismthatmakesit ral and spatial patterns in plant community composition obsolete at an early age. (Horn 1971, The Adaptive andstructure—themesthatall of the authors just cited ex- Geometry of Trees) plored in depth. Yet Horn’s model is, in terms of its original formula- Nearly half a century ago, Horn (1971) used two simple tion, flawed, and it does not yield the predictions for principles—the nonlinear response of photosynthesis to whichitiswidelyrecognized.Thisfacthasescapedallno- photon flux and the filtering of sunlight within tree can- tice, and it removes the only explanation we had for the opies—to explain why early-successional trees in temper- early dominance of multilayers and the later dominance atedeciduousforestsoftenscatteredtheirleavesinmultiple of monolayers in temperate forest succession. Here I layers (e.g., birch, aspen) while late-successional species of- briefly lay out the problem and outline a number of other ten packed their leaves in a single, densely packed layer factors that may instead drive the multilayer-monolayer (e.g., beech, hemlock; fig. 1). He argued that these differ- shift. Horn’s (1971) model assumes horizontal leaves, a sta- * Email: givnish@wisc.edu. tionary sun directly overhead, no wind or clouds, a non- ORCIDs: Givnish, https://orcid.org/0000-0003-3166-4566. reflective forest floor, and a closely packed forest canopy Am.Nat.2020.Vol.195,pp.935–947.q2020byTheUniversityofChicago. that eliminates sidelighting. There are two central assump- 0003-0147/2020/19506-59478$15.00. All rights reserved. tions. The first is that net photosynthesis P shows a DOI: 10.1086/708498 Michaelis-Mentenresponsetoincidentphotonfluxdensity 936 The American Naturalist Multilayer 4 Betula Quercus 2 0 -3)20 Carpinus Tilia 2 m 15 Monolayer 10 Leaf area density (m5 0 0 0.5 1.0 0 0.5 1.0 Relative height in canopy Figure 1: Multilayer and monolayer phenotypes. Left to right: Drawings of typical differences in branching pattern, with several vertically overlapping branches in multilayers and a shell of branches in monolayers; multilayered canopy of early-successional, sun-adapted quaking aspen (Populus tremuloides) from Colorado versus monolayered canopy of late-successional, shade-adapted witch hazel (Hamamelis virginiana), an understory tree in a Pennsylvania forest; orthotropic shoot (with erect axis and leaves scattered in loose spirals) of quaking aspen (orthotropy is characteristic of many multilayered trees and is adapted for energy capture and canopy growth in sunny environments [Givnish 1995]) versus plagiotropic shoot (with horizontal axis and leaves packed tightly in two horizontal ranks) in American beech (Fagus grandifolia; plagiotropy is characteristic of many monolayers and is adapted for energy capture and canopy growth in shady environments [Givnish 1995]); and plots of leaf area density as a function of relative height in the canopies of four tree species in temperate German forests (Hagemeier and Leuschner 2019). Shade-tolerant, late-successional Carpinus betulus and Tilia cordata strongly concentrate their foliage in a single layer, while shade-intolerant, early-successional Betula pendula and Quercus petraea scatter their foliage more evenly across several layers. Note, however, that all species shown hold their leaves in multiple layers and that the total leaf area index (m2 leaves m22 ground area occupied) is slightly lower in the multilayers, contrary to the Horn model. The photograph of quaking aspen was taken by Brady Smith, USDAForestService, Coconino National Forest, and the photograph of witch hazel was taken by Nicholas A. Tonelli; both images are avail- able for reproduction via Wikimedia Commons (CC BY-2.0). The photographs of shoots of quaking aspen and American beech were taken by the author. 22 21 integral is taken from the top of the canopy to the desired I (PFD; mmol photosynthetically active radiation m s ) with half-saturation at I p k, dark respiration R, and as- depth.Allparametersusedinthisarticlearelistedintable1. ymptotic approach to Pmax 2 R: Giventheseconstraints,Horn(1971)askswhichoftwo Typesetter error trees occupying a given ground area A—amonolayer, see final page PmaxI with leaves packed in a single shell, or a multilayer, with P p : ð1Þ for correct form (I 1k)2R leaves scattered over several layers—will have the highest used in total carbon gain G p Ð AP(h)F(h)dh, integrated from The second is that the PFD penetrating the canopy to a calculations the top to the bottom of the canopy. The best monolayer given depth obeys Beer’s law: ð under these conditions involves complete coverage of a I pI0exp 2 F(h)dh , ð2Þ single layer and has a total return of Typesetter error see final page where I0 is the PFD at the top of the canopy, F(h) is the PmaxI0 fraction of ground covered by leaves at height h, and the GpA (I 1k)2R : ð3Þ for correct form 0 used in calculations Adaptive Geometry of Trees Revisited 937 Table 1: Parameters used in this article Parameter Description P Net photosynthetic rate per unit leaf area (mmol CO m22 s21) 2 P Maximum gross photosynthetic rate per unit leaf area (mmol CO m22 s21) max 2 k Photon flux density that results in half-saturation of net photosynthesis in the Michaelis-Menten model (mmol photons m22 s21) R Respiration rate per unit leaf area (mmol CO m22 s21) 2 C Instantaneous light compensation point (mmol photons m22 s21) 22 21 I Photon flux density in the photosynthetically active spectrum (mmol photons m s ) I Photon flux density at the top of a plant’scanopy(mmol CO m22 s21) 0 2 P(h) Net photosynthetic rate as a function of leaf height (mmol CO m22 s21) 2 P(I) Net photosynthetic rate as a function of photon flux density (mmol CO m22 s21) 2 F(h) Fraction of ground area occupied by horizontal leaves at height h (unitless) A Ground area occupied by a plant canopy (m2) G Total instantaneous carbon gain for a plant canopy (mmol CO s21) 2 2 22 LAI Leaf area index (m leaf area m ground area) L Total leaf area (and mass) of a canopy (m2 leaf area or g leaf mass) a Cost of leaf construction (g CO m22 leaf) 2 Q Net photosynthetic return per unit investment per unit time (s21) LMA Leaf mass per unit area (g leaf tissue m22 leaf tissue) SLA Specificleafareap 1/LMA (m2 leaf tissue g21 leaf tissue) B Annual cost of branch construction for a monolayered canopy B0 Annual cost of branch construction for a multilayered canopy L Branch length p canopy radius (m) b Allometric exponent relating branch construction cost to branch length m Number of leaf layers in a multilayered tree canopy AccordingtoHorn’smodel,theoptimalmultilayerun- Canopy geometries that return more energy under a der the same conditions will add leaves to the bottom of given set of conditions are assumed to yield a competitive the canopy until the return P goes to zero at the instanta- advantage under those conditions. According to Horn neousleafcompensationpointC p Rk=(Pmax 2R),result- (1971), monolayers have an advantage in low light be- ing in a total return of cause they exhibit no self-shading, which would decrease I 1k I or negate photosynthesis under shady conditions (eq. [1]); GpAPln 0 2Rln 0 : ð4Þ multilayers have an advantage under brighter conditions max C1k C because they can maintain several layers of leaves Ð When G per unit ground area is plotted against incident ( F(h)dh 1 1) at full or at least nonnegative rates of net PFD for a monolayer and multilayer for plants with the photosynthesis. Horn (1971) uses this model to predict same photosynthetic parameters, the curves cross, with that multilayers will dominate early succession, soon after the optimal monolayer having an energetic advantage at a disturbance removes the canopy and creates sunny con- low I and the optimal multilayer having an advantage at ditions,butthatasthoseplantsgrowandshadetheground 0 high I (fig. 2). Furthermore, the optimal multilayer will they favor saplings below them with fewer and fewer 0 hold a greater leaf area index (LAI; ratio of leaf area to layers, more and more densely packed, so that canopy ge- ground area occupied) the greater the amount of light at ometrytendsincreasingly toward monolayers and under- the top of the canopy. Given Beer’slaw,I0 exp(2LAI) p story shade increases through succession. C, where C is the instantaneous leaf compensation point The fundamental but previously unrecognized prob- (see above). Consequently, according to Horn’s model, lem with this model—in both its mathematical and verbal 2 22 form—isthatit involves a comparison between big plants the optimal LAI (leaf area per unit ground area, m m [unitless]) for an individual tree crown would be (multilayers) with lots of leaf tissue and small plants (monolayers) with much less. If we instead ask whether LAI p2ln C plnI 2lnC, ð5Þ a monolayer or a multilayer will yield a greater photosyn- I 0 0 thetic return for a given total investment in leaf mass (as- implying that optimal LAI should increase with the loga- sumedproportionaltoleafarea),thenwehavetodivideG rithm of light availability. inequations(3)and(4)bytotalleafmass.If,forsimplicity, 938 The American Naturalist a 2 wechooseourunitssothatA p 1andleafmassperunit ) Low k, low R multilayer area p 1, then an optimal monolayer will have a total -1 s leaf area (and mass) L p 1, while the optimal multilayer 2 will have total leaf area (and mass) defined by C p 1.5 I exp(2L) (eq. [1] and Horn’s optimality criterion of 0 P p0atthe bottom of a multilayer). Dividing G by L to get the photosynthetic return on a given investment (g 1 21 21 CO g leafday )—akeydeterminantofwhole-plantrel- monolayer 2 21 21 ative growth rate (g g plant day ; e.g., see Kruger and Volin 2006)—we invariably find that monolayers outper- 0.5 formmultilayersatalllightlevels(fig.3).Thereisnocross- Carbon uptake (µmol CO over; monolayers always win. Cutting through all the equations, it is easy to see why. 0 If we assumeaMichaelis-Mentenphotosyntheticresponse tolight(eq.[1])andself-shadingwithinamultilayer’scan- b opy,eachunitarea(ormass)ofleafcandonobetterthana 2 Low k, high R leaf at the top of a monolayer’s canopy and will often do -1) substantially worse. Consequently, monolayers always win, s2 often by a proportionally very large amount (fig. 3). Gen- 1.5 erally, the bigger k is and the greater R is relative to P , max the bigger the advantage of monolayers when the costs of leaf construction are ignored; large values of k and R in- 1 crease the negative effect of self-shading on the return on investment of the lower leaves. However, when leaf con- struction costs are included, monolayers are favored over 0.5 multilayersregardlessoftherelativemagnitudeofkandR. 22 If a is the cost of leaf construction (g CO m leaf), then 2 Carbon uptake (µmol CO G/L is proportional to G/(aL), the ratio of return rate to initial cost (which we used above as a growth metric), 0 and is linearly related to net energetic return per invest- c mentperunittime:Q p (G2aL=T)=aLpG=aL21=T, 2 High k, high R where T is leaf longevity and aL/T is the cost of construct- -1) ing a unit area of leaf, amortized over its lifetime but not s2 taking into account opportunity costs (see Givnish 1984; 1.5 Givnish et al. 2004). This approach to assessing optimal canopy geometry, based on optimizing the returns on a given energetic in- 1 vestment,issimilartotheoneIusedtoreanalyzetheclas- sic dataofBjörkmanetal.(1972)onphotosyntheticadap- tation of individual leaves to high, intermediate, and low PFDs (fig. 4; Givnish 1988). Björkman et al. (1972) used 0.5 measurements of photosynthesis per unit leaf area as a Carbon uptake (µmol CO function of PFD—P(I)—for leaves of Atriplex triangu- laris grownat I p 920,290,and92mmolm22s21toshow 0 that those response curves crossed and to make the para- 0 400 800 1200 1600 2000 digmatic argument that, as a result, the differences in ac- -2 -1 climation shown were adaptive, with plants grown at low PFD (µmol m s ) light having an energetic advantage at low light levels, those grown at intermediate light having an advantage Figure 2: Sample calculations of photosynthetic returns per unit area for optimal monolayers (red) and multilayers (blue) as a func- tionofphotonfluxdensity(PFD),basedontheHorn(1971)model. curves, with monolayers having an advantage at low PFD and All assumethatPmax p 1;forA,k p 200,R p 0:1;forB,k p 200, multilayers having an advantage at high PFD, with the crossover Rp0:25;andforC,kp600,Rp0:25.Notethecrossoverinthe point increasing with k and R.
no reviews yet
Please Login to review.