The study revealed large differences among species in their extent of spring phenological avoidance, phenological stage during avoidance, and total annual irradiance and its seasonal distribution. Furthermore, as predicted, species that exhibited greater spring avoidance exhibited more rapid expansion of leaves coupled with rapid development of physiological characteristics that would enhance C gain. Autumn avoidance, in contrast, was minimal and would bring little or no C gain. Senescence was well advanced prior to much canopy opening, and leaves did not maintain or enhance their physiological capacity after canopy leaf drop began.
The wide range of phenological avoidance among the study species is representative of juveniles of canopy tree species in the same forest, where spring avoidance by fully expanded leaves ranges from a few days to 3 weeks (Augspurger & Bartlett 2003). Autumn avoidance is uncommon and minimal, as found in this study. Frost damage to young leaves indicates that the species are currently near the limit of spring phenological avoidance in this locality. A similar extent of shade avoidance occurs in understorey woody individuals of other temperate deciduous forests (dePamphilis & Neufeld 1989; Gill et al. 1998).
The mechanism underlying phenological avoidance of Aesculus is associated with earlier accumulation of thermal degree hours in the understorey than in the canopy (Augspurger 2005). In that study, Aesculus seedlings were grown either at ground level under the shelter of the forest canopy, or exposed at near canopy height. The latter, because of greater radiation cooling at night, accumulated degree hours more slowly and showed delayed leaf development. The mechanism for avoidance in the other species may be similar to Aesculus, given that bud break in many temperate woody species has been linked to winter chilling followed by accumulation of thermal degree days (Hunter & Lechowicz 1992; Heide 1993; Chuine & Cour 1999).
Both spring and autumn phenology were initiated earlier in understorey than in edge individuals, especially in Aesculus. Temperature accumulation in spring appears to be faster for understorey than for edge individuals. Both juvenile groups, however, predated canopy conspecifics (data not shown). Thus their phenology somewhat reflects ontogeny, but the environment created by canopy trees further modifies it.
The study species differ as to whether they experience phenological avoidance throughout their lifetime. Shrub/treelet species are permanently in the understorey, and presumably display avoidance throughout their lifetime. The phenology of Acer and Aesculus, in contrast, shifts through time to become more similar to conspecific canopy trees (Augspurger & Bartlett 2003). Their extent of phenological avoidance diminishes as they age.
light gains via phenological avoidance
A relatively small increase in spring avoidance brings a disproportionate gain in light. Estimated total irradiance of Aesculus with the greatest spring avoidance is five times greater than in Asimina with the least spring avoidance. Nevertheless, irradiance of understorey individuals of Aesculus and Acer is still two orders of magnitude lower than edge and canopy individuals.
Light gains via avoidance are much greater in spring than in autumn. As in other temperate deciduous forests (Hutchison & Matt 1977; Gill et al. 1998), the period of closed canopy fell mostly after the summer solstice and peak radiation. The greater light received by understorey individuals in spring than in autumn strongly reflects this asymmetry. It was surprising that peak irradiance in the open was 19 days after the summer solstice. This delay may be due to greater cloud cover in spring.
Light gains in spring are not uniform across the forest due to differential timing of leaf development among species of canopy trees (Shogo & Akira 2002). This study summarized light and canopy closure at the forest level, but an individual understorey plant will experience higher or lower irradiance during spring depending on the specific species of canopy tree above it. The distribution of understorey plants with respect to species of canopy trees is unknown.
Our estimates of total light interception did not take into account transient sunflecks that increase light availability and, potentially, C gain under the closed canopy in summer (Naumburg & Ellsworth 2002). Canham et al. (1994) report that they contribute <10% of the summer radiation beneath late successional species, such as those that predominate at our study site. The estimates of sunflecks were based on hemispheric photographs of the closed canopy, and thus their radiation values did not incorporate light gain during spring or autumn avoidance.
All species except Asimina, the species with minimal avoidance, showed a physiological capacity to capitalize on the high light during spring, but not autumn, avoidance. Photosynthetic capacity, Nmass and Narea were highest in young, expanding leaves but low in old leaves. Actual photosynthesis, as well as potential, would be higher in spring because soil moisture is more favourable, temperature stress and photorespiration are less, and stomatal conductance is higher (data not shown). In general, older leaves have lower photosynthetic capacity than young leaves (Gill et al. 1998; Bond 2000), including in A. saccharum (Jurik 1986).
Some forest herbs that spend part of their growing season in canopy shade show photosynthetic acclimation (Rothstein & Zak 2001). The study's woody species, however, did not show evidence of acclimation. Acer saccharum has a limited ability to acclimate to increased irradiance (Ellsworth & Reich 1992).
In contrast to the other traits, total chlorophyll did not peak until the middle of leaf life span in all species, as has been shown in shade leaves of A. saccharum (Koniger et al. 2000). We do not know the reason for the slow accumulation of chlorophyll. The peak occurred at different dates for different species, indicating that it is not related to an increase in leaf area index. The leaf area index peaks at the same time as full expansion of canopy leaves (C.A., unpublished data), while chlorophyll accumulation continues well past that time.
The seasonal pattern of SLA can be used to infer what contributed to the seasonal decline in leaf N. The rapid decline in young leaves was due, at least in part, to a dilution effect from leaf expansion. The continuing, but slow, decline in N after SLA stabilized indicates that N was being translocated from the leaf or leached throughout the mature phase of the leaf, well before the period of leaf senescence. The N did not supply a new cohort of leaves, as all species produced only one leaf cohort. At leaf drop, N content was <1·5% in all species.
The rate of physiological development was similar in understorey and edge individuals of Acer and Aesculus, with the understorey individuals simply initiating development earlier. Once begun, understorey individuals did not compress leaf development in spring or delay it in autumn, relative to edge individuals. As with phenology, the rate of physiological development appears to be under ontogenetic control, while its initiation is somewhat modified by the environment.
The study predicts that spring phenological avoidance coupled with high physiological capacity will increase C gain and growth. Carbon gain by all the forest's understorey individuals via phenological avoidance should be incorporated into any model considering overall and seasonal C gain of a forest ecosystem (Wilson, Baldocchi & Hanson 2001). Indeed, deciduous juveniles in deciduous forests gain a substantial amount of annual C (Gill et al. 1998) and growth (Jones & Sharitz 1989; Seiwa 1998) in periods without canopy closure, but more in spring than in autumn (Harrington, Brown & Reich 1989). In contrast, evergreen understorey plants in deciduous forests can only adjust their physiology to changing light availability. For example, Juniperus virginiana has its highest photosynthetic rate in spring, but a high autumnal rate as well (Lassoie et al. 1983).
The study has implications for categorizing a species’ shade tolerance. Aesculus glabra has been classified as shade-tolerant (Wenger 1984). Our study demonstrated, however, that it receives 98% of its light prior to 100% canopy closure; it has young, full-sized leaves with high photosynthetic capacity then, and thus must rely heavily on spring avoidance for C gain during its juvenile stage. Direct evidence for its shade-intolerance has been demonstrated by artificially shading understorey individuals during their normal period of spring phenological avoidance. In that case, we found accelerated leaf senescence, shortened leaf life span, lower growth and higher mortality after only 1 year of treatment (C.A., unpublished data).
The five coexisting species differed greatly in estimated total irradiance, maximal photosynthetic capacity and, presumably, overall C gain. Other things being equal, these differences may result in variation in competitive ability and would be reflected in their relative growth and survival, thus affecting their relative abundance. Indeed, Aesculus may assume its prominence in the understorey of this forest because of its great phenological avoidance. Aesculus should be less dependent on gaps for recruitment and growth. This temporal partitioning of light may be one mechanism promoting coexistence of these species.
While it is clear that phenological avoidance does not result in total annual light interception or C gain even remotely equivalent to that received by unshaded individuals, its importance for understorey individuals is equally clear. Enhanced C gain via phenological avoidance appears to be the norm for many herbs, saplings of canopy trees, and understorey treelets and shrubs. Phenological avoidance of the low light environment of the closed canopy is thus a common mechanism to enhance C gain of understorey plants in temperate deciduous forests.