Leaf out times of temperate woody plants are related to phylogeny, deciduousness, growth habit and wood anatomy

Authors


Summary

  • Leaf out phenology affects a wide variety of ecosystem processes and ecological interactions and will take on added significance as leaf out times increasingly shift in response to warming temperatures associated with climate change. There is, however, relatively little information available on the factors affecting species differences in leaf out phenology.
  • An international team of researchers from eight Northern Hemisphere temperate botanical gardens recorded leaf out dates of c. 1600 woody species in 2011 and 2012.
  • Leaf out dates in woody species differed by as much as 3 months at a single site and exhibited strong phylogenetic and anatomical relationships. On average, angiosperms leafed out earlier than gymnosperms, deciduous species earlier than evergreen species, shrubs earlier than trees, diffuse and semi-ring porous species earlier than ring porous species, and species with smaller diameter xylem vessels earlier than species with larger diameter vessels. The order of species leaf out was generally consistent between years and among sites.
  • As species distribution and abundance shift due to climate change, interspecific differences in leaf out phenology may affect ecosystem processes such as carbon, water, and nutrient cycling. Our open access leaf out data provide a critical framework for monitoring and modelling such changes going forward.

Introduction

In many temperate and boreal regions of the world, the timing of spring leaf out is advancing due to the warming effects of climate change (Menzel & Fabian, 1999; Chmielewski & Rötzer, 2001; Menzel et al., 2001, 2006; Richardson et al., 2006; Wesolowski & Rowinski, 2006; Delbart et al., 2008; Carroll et al., 2009; Morin et al., 2009; Chen & Xu, 2012; Cong et al., 2012). Modelling of future leaf out phenology suggests that this trend will continue for most species, except when limited by insufficient winter chilling (Morin et al., 2009; Bennie et al., 2010; Cook et al., 2012). Earlier leaf out will have major consequences for ecosystem processes, including carbon sequestration, hydrology, biomass accumulation and timber production (Lopez et al., 2008; Vitasse et al., 2009; Fridley, 2012). Changes in leaf out phenology could also influence ecological interactions, for example, spring ephemeral and bark epiphyte life cycles that synchronise with high spring light levels in the forest understory, outbreaks of insect pests that feed on leaves and numerous trophic interactions with birds, mammals and fungi (Polgar & Primack, 2011). Despite the growing attention to phenology and its importance in climate change biology, ecosystems and physiological ecology, interspecific variation in leaf out phenology has been relatively overlooked.

Only a few phenology surveys have compared differences in the leaf out times of different species and those that have, have been restricted to between 7 and 73 tree and shrub species (Lechowicz, 1984; Sun et al., 2006; Davi et al., 2011; Liu et al., 2011; Fridley, 2012). Despite the paucity of data, studies have revealed considerable interspecific variation in leaf out times and a consistent order in the timing of leaf out (Lechowicz, 1984; Kramer, 1995; Wesolowski & Rowinski, 2006; Bennie et al., 2010). In addition, Lechowicz (1984) showed that woody species with smaller xylem vessel diameter and diffuse porous wood anatomy tended to leaf out earlier than species with larger vessel elements and ring porous wood anatomy. The difference is, perhaps, because small-vessel species have better tolerance to freezing. More recently, two additional studies in China observed that small-leaved woody species leafed out earlier and had slower leaf expansion rates than larger-leaved species (Sun et al., 2006; Liu et al., 2011). The authors hypothesized that these traits were adaptations to minimise leaf herbivory (Sun et al., 2006; Liu et al., 2011). In a survey of seven tree species, Davi et al. (2011) found that deciduous species leafed out before evergreen species. Understory species and saplings were also observed to leaf out earlier than canopy tree species, presumably to better utilise available light before canopy closure (Seiwa, 1999; Augspurger & Bartlett, 2003; Sun et al., 2006; Lopez et al., 2008; Richardson & O'Keefe, 2009; Liu et al., 2011; Fridley, 2012; Rollinson & Kaye, 2012).

In addition to ecological factors, it has been hypothesized that differences in leaf out times might be related to the evolutionary history of species (Lechowicz, 1984). However, leaf out phenology has not been examined explicitly in a broad phylogenetic context (Lechowicz, 1984; Gazal et al., 2008; Davi et al., 2011; Fridley, 2012; Davies et al., 2013) despite its relevance to large-scale phenological analyses (Willis et al., 2008, 2010; Davis et al., 2010; Wolkovich et al., 2013). While previous studies have demonstrated that species vary in their leaf out phenology, research to date lacks a phylogenetic context that includes a large and diverse sample of plant species from across multiple geographical locations.

The objective of our study is to assess factors of interspecific variation in leaf out phenology by examining a broad spectrum of temperate woody plants in the Northern Hemisphere. To accomplish our objective, we address the following questions: How much interspecific variation exists in the leaf out dates of temperate woody plant species? How does leaf out phenology relate to deciduousness, growth habit, wood anatomy and phylogenetic relatedness? And finally, how does leaf out phenology vary across geographic locations? Our leaf out study utilises the broadest taxon sampling compiled to date and incorporates a rigorous phylogenetic framework to clarify leaf out patterns in northern hemisphere woody plants.

Materials and Methods

We recorded weekly leaf out dates of woody species in the spring at eight northern hemisphere temperate botanical gardens and arboreta in 2012 (Table 1). We monitored 1597 tree, shrub and vine species, representing 88 angiosperm families and seven gymnosperm families at all sites combined. Not every species was monitored at every site but on average each species was monitored at at least two sites (Table 1). In the prior year, 2011, 616 species were monitored at the Arnold Arboretum and 60 species of Quercus L., Juglans L. and Carya Nutt. at Morton Arboretum.

Table 1. Mean leaf out day of year, standard deviation (SD) range of leaf out dates in days, number of species, and numbers of trees (t), shrubs (s) and vines (v) for 2012
 Latitude, longitudeAltitude (m)Mean annual temp. (°C)Mean March 2012 temp. (°C)Mean day of year leaf out dateSD leaf out date (d)Leaf out date range (d)Number of species, t, s, v
  1. Leaf out dates monitored at Arnold Arboretum, Boston, MA, USA; Beijing Botanical Garden, Beijing, China (Beijing BG); Botanic Garden and Botanical Museum Berlin-Dahlem, Berlin, Germany (Berlin BG); Garden in the Woods, Framingham, MA, USA; Morton Arboretum, Lisle, IL, USA; Munich Botanical Garden, Munich, Germany (Munich BG); Ottawa Arboretum, Ottawa, Canada and US National Arboretum, Washington, DC and Beltsville, MD, USA (US Nat. Arboretum), with their respective latitude, longitude, altitude, mean annual temperature and mean March temperature over 1961–1990 for Beijing and North America and over 1971–2000 for Munich and Berlin.

US Nat. Arboretum39°54′N, 76°58′W19–5013.77.68410.746187, 112, 74, 1
Beijing BG39°59′N, 116°12′E7411.95.21047.655162, 97, 57, 8
Morton Arboretum41°49′N, 88°03′W223103.38912.268416, 258, 146, 12
Arnold Arboretum42°18′N, 71°07′W2210.73.410115.6881112, 548, 514, 50
Garden in the Woods42°20′N, 71°25′W5710.73.411114.577162, 50, 107, 5
Ottawa Arboretum45°23′N, 75°42′W805.8−2.712216.559182, 149, 31, 2
Munich BG48°10′N, 11°30′E5239.24.510015.571444, 203, 230, 11
Berlin BG52°27′N, 13°17′E579.24.610117.7105772, 385, 360, 27

We defined the leaf out date for deciduous species as the date on which young leaves were unfolding or expanding with their final shape at least partially visible on at least three branches of an individual plant. This definition is comparable to the definitions used by the USA National Phenology Network (USA-NPN National Coordinating Office, 2012) and the International Phenological Gardens of Europe (IPGE, 2013). Our definition of leaf out date was similar for evergreen species. We defined the leaf out date for species of Pinaceae as the date on which the needles began emerging from their bundles and the needle tips were separated from each other. New scale leaves on certain Cupressaceae species were distinguished by their distinctive lighter colour. We excluded some species of Thuja L., Juniperus L. and Chamaecyparis Spach. because there was no clear difference in colour or shape between new and old leaves.

We scored all species for the following traits: angiosperm vs gymnosperm, deciduous vs evergreen (deciduousness) and growth habit (tree, shrub or vine). Species that grow as trees or shrubs were classified by their dominant growth form. We additionally scored wood anatomy of 1076 angiosperm species as ring porous, semi-ring porous or diffuse porous, and 661 angiosperm species with vessel diameter classes of ≤ 100 μm or > 100 μm based on data from FFPRI (1989), Schweingruber (1990), Schweingruber & Landolt (2010) and InsideWood (2012). Wood anatomy and vessel diameter data were not available for all species in the study.

For the statistical analysis not accounting for phylogeny, we used standard least squares modelling to determine how much variation in leaf out dates was explained by sites, angiosperm vs gymnosperm, deciduousness, growth habit, wood anatomy and vessel diameter.

In order to facilitate comparisons among sites in 2012 we calculated an adjusted leaf out date for each species so that all sites had the same mean leaf out date. This helped to deal with the fact that large numbers of species were not shared between sites and that different sites had different climates and consequent mean leaf out dates. To accomplish this, we first calculated a site adjustment factor for each site, which was the difference between the mean leaf out date of all species at the site in 2012 and the mean leaf out date of all species at all sites in 2012. Then we calculated the adjusted leaf out date for each species by adding the site adjustment factor to the leaf out dates of each species at a site and then averaged the leaf out date + site adjustment across the sites for each species. This method is similar to Primack et al. (2004) and Panchen et al. (2012). We used JMP10 (SAS Institute, Cary, NC, USA) for all statistical analyses.

In order to account for the potential effect of the shared evolutionary history of species, we performed the same set of statistical analyses correcting for phylogenetic relationships (Davis et al., 2010). We ran all phylogenetic analyses using two different phylogenies to address limitations of taxon sampling and phylogenetic resolution (see Wolkovich et al., 2013 for a similar approach). The first phylogeny (hereafter referred to as the PHYLOMATIC tree) was a composite tree based on the literature and had high taxon sampling, but low phylogenetic resolution. The second phylogeny (hereafter referred to as the PHLAWD phylogeny) was based on sequence data and had low taxon sampling, but greater phylogenetic resolution.

The PHYLOMATIC tree was assembled with all 1597 species using the program Phylomatic (Webb & Donoghue, 2005). Relationships in the PHYLOMATIC tree were based on the APG III (2009) phylogeny and were further refined based on the Angiosperm Phylogeny Website (Stevens, 2012). Branch lengths of the PHYLOMATIC tree were adjusted to reflect divergence time estimates based on the fossil record (Bell et al., 2010; Smith et al., 2010) using the function ‘bladj’ implemented in Phylocom 4.1 (Webb et al., 2008). The PHLAWD phylogeny was constructed with maximum likelihood methods using the program PHLAWD (Smith et al., 2009). DNA sequence data was collected from GenBank for 451 species (see Methods in Wolkovich et al. (2013) for further details on markers and tree assembly).

We tested for phylogenetic signal in the adjusted leaf out dates using Pagel's λ (Pagel, 1999) and Blomberg's K (Blomberg et al., 2003) using the ‘phylosig’ function in the package ‘phytools’ v0.2-1 in R v2.15.1 (R Foundation, Vienna, Austria) (Revell, 2012). To test which factors best explain leaf out dates while controlling for phylogeny, we used the phylogenetic generalized least squares function ‘pgls’ in the R package ‘caper’ v0.5 (Orme et al., 2012). The adjusted leaf out date was treated as the dependent variable, while angiosperm vs gymnosperm, deciduousness (deciduous or evergreen) and growth habit (tree, shrub or vine) were treated as the predicted variables.

We used the Welch difference of means test with α = 0.05 to determine if there were significant differences in mean adjusted leaf out dates and site mean leaf out dates for each of the factors of angiosperm vs gymnosperm, deciduousness, growth habit, wood anatomy types and vessel diameter categories. The Welch test does not assume equal variance between samples so this test was chosen because our sample sizes are unequal.

As a first step to gaining a broad overview of whether closely related species have similar leaf out phenology, we calculated the average leaf out date for each genus with three or more species and for each family with five or more species. Although taxonomic ranks are arbitrary clade identifiers, we chose to present average leaf out dates for clades that correspond to the rank of genus and family because of their prominent recognisability. Deciduous species and evergreen species for each genus or family were averaged separately. We calculated these mean genus and family leaf out dates for the entire set of adjusted leaf out dates and separately for the 2012 leaf out dates at the Arnold Arboretum and at the Berlin Botanic Garden. As a second step we used the phylogenies to assess whether closely related species have similar leaf out phenology. Here we determined the clades that had either significantly earlier or later average leaf out dates than expected relative to a random sample of species. We used the ‘aotf’ function in program Phylocom v4.2 (Webb et al., 2008) and 1000 randomized runs to calculate the mean adjusted leaf out date of each node under the ancestral averaging algorithm (Node.mn) and identified all nodes with significantly early average adjusted leaf out dates (Nmn.RankLow < 25, Nmn.RankHigh > 975) and significantly late average adjusted leaf out dates (Nmn.RankLow < 975, Nmn.RankHigh > 25).

We ran linear regression models to compare the order of leaf out at the Arnold Arboretum with each of the other seven sites. The Arnold Arboretum was used as the standard because this site had the most species monitored. We also ran linear regression models to compare the order of average leaf out for genera and families at the Arnold Arboretum and Berlin Botanic Garden, the two locations with the greatest number of species and genera in common. Additionally, for the species monitored in both 2011 and 2012 at the Arnold and Morton Arboreta, we ran linear regression models to determine the extent to which species order of leaf out was consistent across these 2 yr.

Results

Leaf out dates at each of the eight sites were distributed over periods of 6–15 wk (Table 1). Leaf out took place over an average of 71 d across the sites monitored in 2012, ranging from 46 d at the US National Arboretum to 105 d at Berlin Botanic Garden (Table 1). Leaf out took place over c. 3 months at the Arnold Arboretum (88 d) and the Berlin Botanic Garden (105 d), the two sites with the largest number of monitored species.

Much of the variation among species was explained by site, angiosperm vs gymnosperm, deciduousness, growth habit, wood anatomy and vessel diameter in the standard least squares model (R2 = 0.44, < 0.0001, = 1731). All terms in the model were significant (< 0.001) except angiosperm vs gymnosperm. The models including site and one other factor still explained some of the leaf out date (angiosperm vs gymnosperm: R2 = 0.29, < 0.0001, = 3437; deciduousness: R2 = 0.31, < 0.0001, = 3437; growth habit: R2 = 0.31, < 0.0001, = 3437; wood anatomy: R2 = 0.33, < 0.0001, = 2497 and vessel diameter: R2 = 0.35, < 0.0001, = 1747).

There was significant phylogenetic signal in the adjusted leaf out dates for both the PHYLOMATIC (λ = 0.69, < 0.0001, Blomberg's K = 0.61, = 0.011) and PHLAWD trees (λ = 0.89, < 0.0001, Blomberg's K = 0.04, = 0.001). Evidence of phylogenetic signal in leaf out date indicates the importance of accounting for phylogenetic relatedness in our regression analysis. When controlling for phylogeny, the adjusted leaf out date had a significant relationship with deciduousness and growth habits for both the PHYLOMATIC and PHLAWD trees (Table 2).

Table 2. Results from phylogenetic generalized least squares (PGLS) examining the effects of deciduousness (deciduous vs evergreen), growth habit (shrub vs tree or vine) and angiosperm vs gymnosperm on the adjusted leaf out dates for the low resolution PHYLOMATIC tree, a composite phylogenetic tree of all woody species in our study (= 1597) and the high resolution PHLAWD phylogeny, a subset of the woody species included in the study (= 451)
Termdfβ t P
  1. Parameter estimate (β) is equivalent to the days of difference in leaf out between the pairs.

PHYLOMATIC tree1592   
Deciduous vs evergreen 9.719.90< 0.0001
Habit (shrub vs tree) 6.098.00< 0.0001
Habit (shrub vs vine) 3.632.560.011
Angiosperm vs gymnosperm −1.02−0.250.806
PHLAWD tree446   
Deciduous vs evergreen 9.714.51< 0.0001
Habit (shrub vs tree) 6.594.39< 0.0001
Habit (shrub vs vine) 0.9870.330.740
Angiosperm vs gymnosperm 5.460.290.769

Patterns in species level variation of leaf out times

The mean adjusted leaf out date was significantly earlier for: (1) angiosperms than for gymnosperms (Fig. 1, Table 3); (2) deciduous species than for evergreen species (Fig. 2, Table 3); (3) shrubs than for trees or vines (Fig. 3, Table 3); (4) semi-ring porous species than for both diffuse porous and ring porous species (Fig. 4, Table 3); and (5) species with vessel diameters ≤ 100 μm than for species with vessel diameters > 100 μm (Fig. 5, Table 3). These differences remained significant at all sites with the following exceptions: the mean leaf out date for shrubs was significantly earlier than for vines at the Arnold Arboretum and the Berlin Botanic Garden (= 1.96, < 0.001) and there was no significant difference in mean leaf out date in the wood anatomy or vessel diameter categories at the Garden in the Woods (Supporting Information Table S1).

Table 3. Welch difference of means test for the adjusted mean leaf out date comparing angiosperm vs gymnosperm, deciduousness, growth habit, wood anatomy and vessel diameter
ComparisonDays Dif. t N 1 N 2 P
  1. Days Dif., the number of days of difference in mean leaf out dates; N1, the number of species that leafed out significantly earlier in the pairwise comparison (the trait listed first in the comparison column); N2, the number of species that leafed out significantly later (the trait listed second in the comparison column).

Angiosperm vs gymnosperm1912.891477120< 0.001
Deciduous vs evergreen1715.351368229< 0.001
Shrub vs tree growth habit1014.32792736< 0.001
Shrub vs vine growth habit73.60792690.001
Diffuse vs ring porous wood anatomy99.90651224< 0.001
Semi-ring vs ring porous wood anatomy129.75201224< 0.001
Vessel diameter ≤ 100 μm vs > 100 μm1111.34482179< 0.001
Deciduous angiosperm vs deciduous gymnosperm51.371356120.279
Evergreen angiosperm vs evergreen gymnosperm105.37121108< 0.001
Deciduous angiosperm vs evergreen angiosperm157.2393277< 0.001
Deciduous gymosperm vs evergreen gymnosperm275.18994< 0.001
Figure 1.

The adjusted leaf out dates of angiosperms (a) vs gymnosperms (g) for 1597 species in 2012 across the eight temperate botanical garden and arboretum sites (Table 1), showing that angiosperms leaf out significantly earlier than gymnosperms (= 1.96, α = 0.05, = 1477 angiosperms, 120 gymnosperms). The day of year is the number of days from 1 January 2012. The box plot shows the adjusted leaf out date range, quartiles and a mean adjusted leaf out day of year of 99 for angiosperms and 118 for gymnosperms, the box width indicates the relative sample sizes.

Figure 2.

The adjusted leaf out dates of deciduous species (d) vs evergreen species (e) for 1597 species in 2012 across the eight temperate botanical garden and arboretum sites (Table 1), showing that deciduous species leaf out significantly earlier than evergreen species (= 1.96, α = 0.05, = 1368 deciduous species, 229 evergreen species). The day of year is the number of days from 1 January 2012. The box plot shows the leaf out date range, quartiles and a mean adjusted leaf out day of year of 98 for deciduous species and 115 for evergreen species, the box width indicates the relative sample sizes.

Figure 3.

The adjusted leaf out dates related to growth habit type (s, shrub; t, tree; v, vine) for 1597 species in 2012 across the eight temperate botanical garden and arboretum sites (Table 1), showing that shrubs leaf out significantly earlier than trees and vines (= 1.96, α = 0.05, P < 0.0001, N = 736 trees, 792 shrubs and 69 vines). The day of year is the number of days from 1 January 2012. The box plot shows the leaf out date range, quartiles and a mean adjusted leaf out day of year for each growth habit type (95 for shrubs, 105 for trees and 102 for vines), the box width indicates the relative sample sizes.

Figure 4.

The adjusted leaf out dates related to wood anatomy type (d, diffuse porous; r, ring porous; sr, semi-ring porous) for 1076 species in 2012 across the eight temperate botanical garden and arboretum sites (Table 1), showing that diffuse porous species leaf out significantly earlier than ring porous species and that semi-ring porous species leaf out significantly earlier than both diffuse and ring porous species (= 1.96, α = 0.05, < 0.0001, < 0.0001 and 0.0014, respectively, = 651 diffuse porous, 224 ring porous and 201 semi-ring porous). The day of year is the number of days from 1 January 2012. The box plot shows the leaf out date range, quartiles and a mean adjusted leaf out day of year for each wood anatomy type (96 for semi-ring porous, 99 for diffuse porous and for 108 ring porous), the box width indicates the relative sample sizes.

Figure 5.

The adjusted leaf out dates related to vessel diameter (x ≤ 100 μm and y > 100 μm) for 661 species in 2012 across the eight temperate botanical garden and arboretum sites (Table 1), showing that species with vessel diameters ≤ 100 μm leaf out significantly earlier than species with vessel diameters > 100 μm (= 1.96, α = 0.05, = 482 diameter ≤ 100 μm, 179 diameter > 100 μm). The day of year is the number of days from 1 January 2012. The box plot shows the leaf out date range, quartiles and a mean adjusted leaf out day of year of 98 for species with vessel diameter ≤ 100 μm and 109 for species with vessel diameter > 100 μm, the box width indicates the relative sample sizes.

Furthermore, the mean adjusted leaf out date for evergreen angiosperms was significantly earlier than for evergreen gymnosperms. However, there was no significant difference in mean adjusted leaf out dates between deciduous angiosperms and deciduous gymnosperms (Table 3). When angiosperm and gymnosperm species were considered separately, the mean adjusted leaf out date for deciduous angiosperms was significantly earlier than for evergreen angiosperms and likewise for deciduous gymnosperms and evergreen gymnosperms (Table 3).

In order to illustrate the patterns described above, at the Arnold Arboretum, where the largest number of species was monitored, the earliest species to leaf out on 10 March 2012 were angiosperm deciduous shrubs including Chaenomeles japonica (Thunb.) Lindl., Daphne mezereum L., Ligustrum compactum (Wall. ex G.Don) Hook.f. & Thomson ex Decne., Lonicera tatarica L., Ribes alpinum L. and Sorbaria sorbifolia (L.) A. Braun. The last species to leaf out in the spring of 2012 at the Arnold Arboretum were the gymnosperm evergreen trees Pinus aristata Engelm., P. contorta Douglas ex Loudon and P. pumila (Pall.) Regel and the evergreen angiosperm trees Ilex maximowicziana Loes. and Magnolia grandiflora L. on 30 May and the angiosperm evergreen shrub Rhododendron makinoi Tagg. on 6 June. Similarly, at Berlin Botanic Garden, where the second largest number of species was monitored, the angiosperm deciduous shrubs Ribes alpinum L., Rubus idaeus L., Sambucus nigra L., Sorbaria kirilowii (Regel) Maxim. and S. sorbifolia were the first to leaf out on 7 March 2012, and the last to leaf out were the gymnosperm evergreen trees Picea torano (Siebold ex K.Koch) Koehne, Pinus aristata Engelm., P. armandii Franch. and P. bungeana Zucc. ex Endl. on 6, 13 and 20 June, respectively, and the angiosperm evergreen shrub Rhododendron fauriei Franch. on 16 May. Most of the earliest leafing out angiosperm species have diffuse porous stems with small vessel diameters, and most of the latest leafing out angiosperm species have ring porous stems with large vessel diameters. A list of the species that leafed out first and last at each site and in the adjusted leaf out dataset is given in Table S2 and in the online dataset.

Differences among genera, families and clades

Some clades in the PHYLOMATIC and PHLAWD trees exhibited significantly conserved early or late average adjusted leaf out dates relative to a random sample of species from across the phylogenies (Tables 4, 5, Figs 6, S1, S2). For example, some of the major early leaf out clades include Rosaceae, Dipsacales (including Viburnum and Lonicera) and Ribes that contain predominantly deciduous species. The major late leaf out clades include gymnosperms (including Pinus and Picea) with the exception of Larix and Ginkgo, Ericaceae (including Rhododendron), Fagales (including Fagus, Quercus, Carya and Juglans) and Magnolia. There was agreement in the early and late leaf out clades between the PHYLOMATIC and PHLAWD trees (Tables 4, 5).

Table 4. Major clades that have either significantly earlier (day of year < 94) or later (day of year > 107) average adjusted leaf out dates than would be expected relative to a random sample of species from across the PHLAWD phylogeny
CladeMain genera in cladeNtaxaNode.mnNode.SD
  1. Ntaxa, the number of taxa in the clade; Node.mn, the average adjusted leaf out date of the clade calculated using Phylocom's ‘aotf’ function using the ancestral averaging algorithm (Webb et al., 2008); Node.SD, the standard deviation of subtending nodes' average adjusted leaf out dates.

Paeonia  2750.24
Ribes  3750.92
Larix  2818.51
Syringa + Ligustrum 6831.25
RosaceaeSpiraea, Cotoneaster, Malus67906.60
Malpighiales + CelastraceaeSalix, Hypericum17901.94
Dipsacales + Ilex + AraliaceaeViburnum, Lonicera, Hedera, Eleutherococcus39948.33
Fabales + FagalesGleditsia, Fagus, Quercus731091.93
Ericaceae + Cyrillaceae + ClethraceaeRhododendron, Cyrilla, Clethra271106.06
Magnolia  611418.28
Gymnosperms (excluding Ginkgo)Tsuga, Cedrus, Abies, Picea, Pinus341134.75
Fraxinus + Forestiera + Chionanthus 141142.01
Catalpa  51170.57
VitaceaeVitis, Ampelopsis31181.27
Table 5. Major clades that have either significantly earlier (day of year < 94) or later (day of year > 112) average adjusted leaf out dates than would be expected relative to a random sample of species from across the PHYLOMATIC tree
CladeMain genera in cladeNtaxaNode.mnNod.SD
  1. Ntaxa, the number of taxa in the clade; Node.mn, the average adjusted leaf out date of the clade calculated using Phylocom's ‘aotf’ function using the ancestral averaging algorithm (Webb et al., 2008); Node.SD, the standard deviation of subtending nodes' average adjusted leaf out dates.

Ribes  16746.81
DipsacalesViburnum, Sambucus, Lonicera, Diervilla, Abelia1107813.63
Ligustrum  118513.04
Deutzia  15868.52
RosaceaePrunus, Spiraea, Malus, Cotoneaster371877.72
Philadelphus  20905.00
Syringa  119010.00
Salix   359511.43
Magnolia  1710912.19
Quercus + Ilex 611116.24
Rhododendron  4311216.55
Rutaceae + Meliaceae + SimaroubaceaePhellodendron, Zanthoxylum, Picrasma, Ptelea, Melia, Toona191132.87
Gymnosperms (ex. Ginkgo + Ephedra)Cedrus, Abies, Tsuga, Picea, Pinus, Taxus, Juniperus1171162.89
Erica  41185.88
Moraceae + Boehmeria bilobaMorus, Broussonetia91188.08
GentianalesGardenia, Cephalanthus, Periploca61180.42
PaliureaeHovenia, Paliurus, Ziziphus31218.31
Figure 6.

Adjusted leaf out dates (day of year) for the PHLAWD phylogeny: dotted line, indicating the mean adjusted leaf out date; black line, the day of year above which species leaf out significantly later than would be expected relative to a random sample of species from across the phylogeny; grey line, the day of year below which species leaf out significantly earlier than would be expected relative to a random sample of species from across the phylogeny; circle, a deciduous species; triangle, an evergreen species. The PHLAWD phylogeny highlights major clades with significantly early average adjusted leaf out dates (in grey), significantly late average adjusted leaf out dates (in black), with numbered circles representing clades with predominantly deciduous species, and numbered triangles representing clades with predominantly evergreen species. The average adjusted leaf out date at each node was calculated using Phylocom's ‘aotf’ function and the ancestral averaging algorithm (Webb et al., 2008).

In the nonphylogenetic comparison of families and genera, the mean leaf out date for the majority of deciduous genera was earlier than for evergreen genera (Tables S3–S5). Across the Arnold Arboretum and Berlin Botanic Garden and the adjusted leaf out datasets, the deciduous shrub genera Chaenomeles Lindl., Ligustrum L., Lonicera L., Niellia T.T. Yu & T.C. Ku, Ribes L. and Sambucus L. consistently had mean leaf out dates in the first or second week of the spring flush and the deciduous tree genera Catalpa Scop. and Morus L. were consistently among the last deciduous genera to leaf out (Tables S3–S5). Among evergreens, species of Cotoneaster Medik. consistently had the earliest mean leaf out date across the three datasets and Pinus L. the latest mean leaf out date (Tables S3–S5). For genera in which there were species of both evergreen and deciduous species, such as Berberis L., Cotoneaster Medik., Ilex L. Rhododendron L., Lonicera and Viburnum L., the deciduous species had an earlier mean leaf out date than the evergreen species. In the genera Magnolia L. and Prunus L., in which most species are deciduous, the last species to leaf out were the evergreen M. grandiflora L. and Prunus laurocerasus L.

Across the Arnold Arboretum, Berlin Botanic Garden and the adjusted leaf out datasets, the mean leaf out date for families (deciduous members only) Grossulariaceae, Caprifoliaceae, Berberidaceae and Pinaceae were consistently the earliest (Tables S6–S8). Species of Bignoniaceae, Moraceae and Vitaceae consistently had the last mean leaf out dates of deciduous species. Among evergreen species, the mean leaf out date for the family Rosaceae was notably early (due largely to many Cotoneaster species). Pinaceae, Taxaceae, Ericaceae families (mainly Rhododendron and Erica species) and Aquifoliaceae (mainly Ilex species) had late mean leaf out dates. For families with both deciduous and evergreen species such as Aquifoliaceae, Berberidaceae, Ericaceae and Pinaceae, the mean leaf out date for the deciduous species was earlier than the mean leaf out date for the evergreen species in the same family.

Differences among sites and between years

There was a strong relationship between the order of leaf out of species at the Arnold Arboretum in 2012 and each of the other seven sites in 2012 indicating that species leaf out in essentially the same order at all study sites (R2 = 0.32–0.72, < 0.0001, = 100–549, depending on site) (Table 6). There were particularly high R2 values (0.63, 0.72 and 0.70, respectively) for the relationship of leaf out dates at the Arnold Arboretum in 2012 with the leaf out dates at the Berlin Botanic Garden (Fig. 7), Garden in the Woods and the Ottawa Arboretum (Table 6), despite the mean leaf out dates at the Garden in the Woods and Ottawa Arboretum being c. 1.5 and 3 wk later, respectively, than at the Arnold Arboretum (Table 1). Within this general pattern, there were also certain species that leafed out in a somewhat different sequence at separate locations. There was a strong relationship between the order of genus mean leaf out dates at Arnold Arboretum and Berlin Botanic Garden in 2012 (R2 = 0.68, = 69, < 0.0001) and between the order of family mean leaf out date at these two sites in 2012 (R2 = 0.83, = 33, < 0.0001); that is, genera and families leafed out in the same order at the two locations, even though the species represented in each genus and family differed.

Table 6. Linear regression (R2, slope, P) of order of leaf out of species at Arnold Arboretum in 2012 with Arnold Arboretum in 2011 and with the seven other sites
Site R 2 Slope P N MeanSDMinMaxAA12 meanAA12 SDAA12 minAA12 max
  1. AA2011, Arnold Arboretum in 2011; USNA, US National Arboretum; Munich, Munich Botanical Garden; Berlin, Botanic Garden and Botanical Museum Berlin-Dahlem; Beijing, Beijing Botanical Garden; Morton, Morton Arboretum; GitW, Garden in the Woods; Ottawa, Ottawa Arboretum. N, Mean, SD, Min and Max are, respectively, the number of species, mean leaf out date, earliest leaf out date and last leaf out date at each site for species common to that site and Arnold Arboretum and AA12 Mean, AA12 SD, AA12 Min, AA12 Max are, respectively, the number of species, mean leaf out date, earliest leaf out date and last leaf out date at Arnold Arboretum in 2012 for species common to Arnold Arboretum and each site.

AA20110.690.66< 0.000154926-Apr12.5226-Mar4-Jun6-Apr15.709-Mar29-May
USNA0.390.50< 0.000111020-Mar9.845-Mar15-Apr5-Apr12.2912-Mar2-May
Munich0.520.81< 0.00013049-Apr15.956-Mar9-May4-Apr14.159-Mar14-May
Berlin0.630.94< 0.00015439-Apr16.666-Mar19-Jun6-Apr14.319-Mar29-May
Beijing0.320.34< 0.000110012-Apr8.1316-Mar10-May8-Apr13.509-Mar14-May
Morton0.380.48< 0.000136627-Mar12.2514-Mar21-May5-Apr15.829-Mar29-May
GitW0.720.80< 0.000110819-Apr14.4814-Mar30-May14-Apr15.299-Mar18-May
Ottawa0.700.88< 0.000114230-Apr16.1325-Mar23-May11-Apr15.799-Mar29-May
Figure 7.

Order of leaf out of species at Arnold Arboretum and Berlin Botanic Garden in 2012, indicating that species leaf out in roughly the same order (R2 = 0.63, < 0.0001, = 549). Many of the individual points are difficult to see due to their overlapping values.

The order of leaf out dates of the species at the Arnold Arboretum in 2012 and 2011 were highly correlated (R2 = 0.69, < 0.0001, = 551) (Table 6). The high R2 value indicates that the order of leaf out of species remains very much the same in consecutive years, despite the Arnold Arboretum mean leaf out date being significantly earlier, by 20 d, due to a substantially warmer spring in 2012 than 2011 (= 1.96, = 616). The order of leaf out dates of Quercus, Juglans and Carya species at the Morton Arboretum in 2012 and 2011 were also correlated (R2 = 0.36, < 0.0001, = 60).

We have made the dataset available as supplemental data for other researchers to use in other research projects (Tables S9–S11). We request that scientists using this data acknowledge its source and inform us of its use.

Discussion

Overall, leaf out dates were spread over a long period during spring, with specific timing related to multiple aspects of a species' biology. Interspecific variation in leaf out dates was associated with ecological and anatomical factors, even when corrected for phylogenetic nonindependence. First, angiosperms, on average, leafed out earlier than gymnosperms. This difference in mean leaf out date held true when evergreen species were analysed separately but not when deciduous species were analysed separately, possibly because of the small sample size of deciduous gymnosperms surveyed (nine species) and because the evergreen and deciduous gymnosperm samples were dominated by groups of closely related plants in single clades. Of the species monitored in this study, half of the deciduous gymnosperms were in one genus (Larix) and 2/3 of the evergreen gymnosperms were in Pinaceae, so it is possible that the results reflect, at least in part, characteristics of these clades rather than gymnosperms as a whole.

A second factor related to leaf out phenology is deciduousness. Across all comparisons, deciduous species leafed out on average before evergreen species. This was true when analysing angiosperms, gymnosperms, trees and shrubs separately. In general, for a particular group, deciduous species leafed out 1–2 wk earlier than evergreen species. A likely explanation is that leaves from past years on evergreen species can photosynthesise during the period in the spring before the risk of frost has passed while deciduous species must produce new leaves as soon as possible to replenish nutrient supplies (Davi et al., 2011; Michelot et al., 2012). Also, the leaves of evergreen species are energetically more expensive to produce and are therefore more carefully protected from potential frost damage (Villar & Merino, 2001). For deciduous species, the first process at the end of dormancy is recovery of photosynthetic capacity by producing new leaves. Evergreen leaves change physiologically in autumn to protect the leaf photosystem from winter damage and then change back in the spring (Öquist & Huner, 2003). Evergreen plants repair their existing leaves before beginning new leaf production, and this may explain the tendency of evergreen species to leaf out later than deciduous species (Lundmark et al., 1988).

The third key factor is that mean leaf out dates differed significantly for different growth habits, with shrubs leafing out before trees, as was described for a much smaller sample size of species in China by Sun et al. (2006) and Liu et al. (2011). In our study, shrubs leafed out earlier on average than trees in angiosperms and gymnosperms and in deciduous and evergreen species. Early leaf out may be particularly advantageous for shrubs in allowing them to become photosynthetically active before the canopy trees produce leaves and reduce the supply of direct sunlight (Seiwa, 1999; Augspurger & Bartlett, 2003; Lopez et al., 2008; Richardson & O'Keefe, 2009; Rollinson & Kaye, 2012). Shrubs may also be better able to tolerate the loss of their first crop of leaves to frost damage because replacing damaged leaves may be less expensive for shrubs than for trees due in part to shrubs having smaller leaf size than trees (Sun et al., 2006; Liu et al., 2011). Differences in early spring microclimates at different heights may expose trees and shrubs to different selective pressures (Geiger et al., 2009; Vitasse, 2013). This will require further research to determine. Vines tended to leaf out later than shrubs and earlier than trees, though the pattern is not as clear, possibly due to both the small number of vine species (69 species) surveyed and to vines' heterogeneous growth forms and ecology.

Another important characteristic affecting leaf out times is wood anatomy, also noted in a previous study (Lechowicz, 1984). Species with diffuse porous stems and semi-porous stems leafed out 1–2 wk earlier on average than species with ring porous wood anatomy. Species with smaller vessel diameters leafed out earlier on average than species with larger vessel diameters. In our study, species with the narrower diameter vessels associated with diffuse and semi-ring porous anatomy are presumably less likely to be damaged by embolisms caused by freeze–thaw events than are the large diameter vessels of ring porous species (Essiamah & Eschrich, 1986; Michelot et al., 2012) and hence can leaf out earlier in the spring. Diffuse and semi-ring porous species are also more likely to retain functioning undamaged vessels from the previous season that can facilitate early spring growth (Essiamah & Eschrich, 1986; Suzuki et al., 1996; Michelot et al., 2012). Ring porous species may need to initiate new wood growth in the spring before bud break can occur (Aloni & Peterson, 1997). There are likely to be other traits beyond wood anatomy that could be important drivers of leaf out times and warrant further research, including leaf area, leaf mass, leaf thickness and twig thickness (Sun et al., 2006; Liu et al., 2011).

The depth of phylogenetic signal in leaf out phenology varied widely across the phylogeny

In several instances, clades with significantly earlier or later leaf out times corresponded to recognized taxonomic ranks. For example, Rosaceae species, on average, leafed out significantly earlier in the year, while the gymnosperms (excluding Ginkgo, Ephedra and Larix) leafed out significantly later. However, several clades do not conform to traditional taxonomic ranks, but represent a range of evolutionary depths. An example would be the Fagales + Fabales clade, which leafed out significantly later relative to the rest of species in the study. The presence of clades that do not have formally designated taxonomic ranks emphasizes the importance of taking a phylogenetic perspective, as opposed to a traditional taxonomic perspective. The phylogenetic signal in leaf out date may be due in part to its correlation with other phylogenetically conserved traits. For instance, of the major conserved clades, those that leafed out significantly early contain predominantly deciduous species while the late leaf out clades can be either predominantly deciduous or evergreen.

We found that the order of leaf out times is consistent at different locations and shows little inter-annual variation. This consistent ordering of species leaf out has been found in other studies of woody plants (Lechowicz, 1984; Wesolowski & Rowinski, 2006). Some of the leaf out differences of individual species between different locations could be due to different microsites or to geographic variation in the plant material source. Determining growing degree days for each species at each site would be another approach to confirm that leaf out order is consistent across years and sites. However, both approaches give similar results and each has advantages and disadvantages (Polgar & Primack 2011; Archetti et al., 2013; Laube et al., 2014). We did not distinguish between native and non-native species in our analysis. Non-native species growing in botanical gardens outside of their native range, in conditions of novel temperature, photoperiodic and chilling cues may introduce additional variance that needs to be considered (Alberto et al., 2013; Polgar et al., 2014; Zohner & Renner, 2014). Native species at each garden may be represented by more individuals and therefore greater genetic variation than non-native species, again potentially influencing observations. We are now collecting more years of observations at these sites, data that will clarify the extent to which warmer temperatures associated with climate change will impact different species. In the current study, we used weekly observations, which was practical given the large number of species under observation. The accuracy of leaf out timing could be improved with more frequent visits, but weekly observation should be adequate (Miller-Rushing et al., 2008).

Our results have a variety of implications for forest and ecosystem ecology. There is currently great interest in determining how leaf out and flowering phenology will be affected by climate change. Many studies have demonstrated that warming spring temperatures have already resulted in earlier leaf out phenology and may cause even earlier leaf out in coming decades (Menzel & Fabian, 1999; Menzel, 2000; Chmielewski & Rötzer, 2001; Menzel et al., 2001, 2006; Delbart et al., 2008; Carroll et al., 2009; Jeong et al., 2011; Cong et al., 2012; Fridley, 2012). A warming climate will affect species composition of forests as well, with complicated effects on ecosystem processes. Climate change projections predict that by the end of this century, the characteristic maple–birch–beech forests of the northeastern United States will be replaced by the oak–hickory forests more typical of the southeastern United States and pine forests in the southeastern United States displaced by oak–pine forests (Karl et al., 2009). If species composition is changing with climate change, predicting leaf out times in the future for a particular geographical area becomes more complicated. In New England, a warming climate resulting in early leaf out birch and maple species being replaced by later leaf out oak and hickory species will make it difficult to predict community-wide phenological changes and ecological interactions. Climate change models are also predicting that the warming climate will delay leaf out times in certain species if they are unable to meet their winter chilling requirements (Morin et al., 2009; Bennie et al., 2010; Cook et al., 2012; Rollinson & Kaye, 2012). Understanding such species differences in winter chilling requirements for leaf out times is a clear priority for forest ecology research linked to climate change (Laube et al., 2014; Polgar et al., 2014). Climate change models sometimes assume all woody plants leaf out at the same time while our results show that woody plants behave differently depending on phylogenetic affinities, growth habit, deciduousness and wood anatomy. While warming conditions will tend to cause some trees to leaf out earlier, the changing composition of the forest and different responses in interspecific leaf out times will also have implications on ecosystem processes and need to be factored into climate change model predictions.

Acknowledgements

The authors wish to thank Ulrike Lohmann (Botanic Garden and Botanical Museum Berlin-Dahlem), Eva Schmidbauer (Botanical Garden Munich), and Kevin Conrad and Robert Webster (US National Arboretum) for assistance with monitoring leaf out times; Danielle Fraser and Thomas Hossie (Carleton University) for advice on the phylogenetic analysis; and Richard Webster (Carleton University) for running the CPU intensive phylogenetic analysis on his computer. Funding in China was provided by the Natural Science Foundation of China (Y32H3A1001) and in Munich by the KLIMAGRAD project of the Bavarian Ministry for the Environment. Comments on the manuscript were provided by Abe Miller-Rushing. We thank the botanical gardens for permission to carry out this work.

Ancillary