Leaf Size in Conifers: Global Associations With Climate and Evolutionary History

ABSTRACT

Global analyses of leaf size suggest that large leaves predominate in favourably warm and wet climates, and small leaves occur at climatic extremes. However, these patterns are dominated by data from angiosperms and may obscure drivers of leaf size variation in older, less speciose groups, such as conifers. Here we employ a novel modelling framework, multi‐response phylogenetic mixed models (MRPMM), that identifies trait correlations at both phylogenetic and phylogenetically independent levels to investigate how climate influences leaf size evolution across conifers. We show overall patterns for all conifers and focus on three groups with distinctive leaf architectures: scale‐leaved Cupressaceae, needle‐leaved Pinus and broad‐leaved Podocarpus. We found moderate to strong phylogenetic signal in conifer leaf and climate niche traits. Phylogenetic relationships explained most of the association between leaf size and climate, with temperature revealed as a stronger driver than dry season precipitation, indicating deep‐time co‐evolutionary associations. These patterns reflect trade‐offs associated with leaf hydraulic architecture under contrasting selection pressures. In single‐veined leaves, lateral water transport and thus leaf width is constrained, yet this narrow leaf form can be advantageous for survival under climatic extremes such as drought and freezing. In contrast, anatomical innovations (like accessory transfusion tissue in Podocarpus) allow for broader leaves which are favourable in competitive environments, while also potentially making leaves more vulnerable to climate extremes. Our results support previous evidence for phylogenetic niche conservatism in conifers, where species tend to track their ancestral climatic preferences rather than adapting to new environments. This conservatism, likely controlled by leaf hydraulic architecture, results in strong evolutionary constraints on current bioclimatic distributions and potential responses to changing climates in conifers. This study also highlights the importance of considering phylogenetic impacts on functional trait evolution, especially in evolutionarily conservative groups like conifers.

Global patterns in leaf size are linked to climatic trade‐offs, so we explore these relationships in a major group of plants, the conifers. Using contemporary phylogenetic methods, we show that conifers display a strong pattern of evolutionary conservatism in the association of leaf size with climate, and highlight that these relationships are driven by anatomical idiosyncrasies of different clades. Single veins in the leaves of the major genus Pinus limit leaf width but increase resilience to climatic extremes, and specialised hydraulic tissue in the genus Podocarpus drives an evolutionary link between broad leaves and frost‐free environments. Evolutionary conservatism in leaf size thus underlies conifer distributions and may restrict their adaptation to new and changing climates.

1. Introduction

Leaf size is intrinsically linked to plant function, evolution and adaptation to environments (Koch et al. 2004; Tozer et al. 2015; Wright et al. 2017; Yates et al. 2010). As the primary site for photosynthetic gas exchange and water loss, the size of a leaf is linked to energy trade‐offs for carbon uptake and water use efficiency (Pierce et al. 2022; Wright et al. 2004). Global analyses have shown that large, broad leaves predominate in mesic and warm environments, so called ‘favourable’ abiotic conditions, where competition for light drives leaf size (Li et al. 2020); whereas smaller leaves are more common in dry and/or hot environments, often in poor soils with low nutrients, or in high latitudes and elevations where plants are routinely exposed to frost (Wright et al. 2017).

These global leaf size analyses are dominated by data from angiosperms; a highly speciose group (> 300,000 species; Govaerts et al. 2021) with leaves characterised by high vein density, vessel‐based water transport and diverse hydraulic architecture (Crisp and Cook 2011; Thompson and Ramírez‐Barahona 2023). Here, we aim to investigate the links between climate and leaf size in a smaller but geographically widespread non‐angiosperm group, the conifers (Pinopsida, excluding Gnetophytes), to test whether the leaf size‐climate relationships in this globally significant group differ from those of angiosperms. This is critical not only to gain a more holistic understanding of the factors shaping plant distributions, but also to better understand how the unique evolutionary history and leaf anatomy of non‐angiosperms influence these relationships.

It has been assumed that the relationships between leaf size and climate are similar in conifers and angiosperms, with broad and long leaves in tropical areas, and small narrow leaves in cold and dry climates (Condamine et al. 2020; Wright et al. 2017). However, these relationships have never been verified empirically despite the fundamental differences between conifers and angiosperms. Notably, some conifers possess leaves with extreme tolerances to cold and dry environments (Bigras and Colombo 2001; Brodribb et al. 2005; Chang et al. 2021; McCulloh et al. 2023), feature leaf forms with distinct anatomical constraints on size and shape to those in angiosperms (Brodribb and Hill 1999), and have evolutionary origins millions of years older than angiosperms (Leslie et al. 2018; Miller 1977). Conifers have broad ecological, bioclimatic and distributional ranges (Farjon and Filer 2013; Hill and Brodribb 1999) from warm tropics to cold high latitudes, dry deserts to wet rainforest. With detailed phylogenetic knowledge available due to advancements in molecular phylogenies (Leslie et al. 2012, 2018), the relatively small number of extant species (~600 spp.), and widely accessible trait information (Earle 2024; Eckenwalder 2009; Farjon 2018), conifers present an ideal system for investigating the drivers of functional trait macroevolution across an entire group.

As primary sites of photosynthesis and water loss through transpiration, leaf hydraulic anatomy controls the efficiency of water flow and therefore plays a role in determining leaf size. In angiosperms, this efficiency is enabled by complex, reticulate vein networks dominated by xylem vessels. This system allows leaf size to vary independently of plant water supply (Sack and Scoffoni 2013). Leaf size in angiosperms is also constrained by the balance of energy inputs and outputs, or ‘leaf energy balance’ (Wright et al. 2017), driven by leaf‐to‐air temperature differences, especially as leaves can get very broad in equatorial (warm and or wet) latitudes (Middleby et al. 2025). Conifer leaves generally lack complex venation and xylem vessels, and most species have leaves with a single longitudinal vein containing xylem tracheids, phloem and transfusion tissue (Hu and Yao 1981; Trueba et al. 2022; Zwieniecki et al. 2004). Without specialised lateral water transport tissue, the hydraulic resistance of the mesophyll strongly limits efficient water transport to the leaf margin (Brodribb et al. 2007). As a result, conifer leaves are typically constrained to be narrow, often scale or needle‐like, to ensure adequate irrigation throughout the mesophyll (Brodribb et al. 2010; Han et al. 2019; Mai et al. 2024). However, some conifers have evolved morphological arrangements that partially overcome hydraulic constraints to leaf width, allowing for a variety of leaf forms beyond the typical scale or needle‐like form (Johnson, Brown, et al. 2025). These include parallel veined leaves in Araucariaceae and Nageia (Podocarpaceae) (Hill and Pole 1992), and phylloclades (modified photosynthetic branches) in Phyllocladus (Podocarpaceae) and Sciadopitys verticillata (Sciadopityaceae) (Biffin et al. 2011; Dörken et al. 2021). We also see examples of individual leaves arranged into a single plane (parallel to the shoot axis) functioning as a single photosynthetic unit, much like an angiosperm compound leaf (i.e., Acmopyle, Podocarpaceae; Hill and Brodribb 1999). Perhaps the most unique innovation, enabling a range of leaf widths, is the development of accessory transfusion tissue (ATT); extra‐xylary tissue that extends laterally from the main vein of the leaf, acting as a supplementary water transport system enabling effective irrigation of the entire mesophyll in broader leaves, seen commonly in Podocarpaceae (Brodribb et al. 2010; Sahni and Seward 1997). Nevertheless, these innovations are considerably less sophisticated than reticulate venation in angiosperm leaves, limiting even the broadest conifer leaves to widths substantially narrower than the maximum seen in angiosperms.

Although diverse, these leaf morphologies arose early in the evolutionary history of the conifers (~100 MYA; Biffin et al. 2011), reflecting a broader pattern of niche conservatism and suggesting constraints on climatic adaptation in conifers over the past 300 million years (Brown et al. 2021; Sundaram and Leslie 2021). Conservatism in both leaf size and climate would suggest that leaf architecture and species‐climatic associations may reflect ancient co‐evolutionary divergences. At the same time, more recent climatic radiations associated with considerable variation in leaf size within the large conifer genera, Pinus and Podocarpus (Farjon 2010), indicate the potential for evolutionary lability in leaf size. Therefore, it is important to consider the role of evolutionary processes acting over different scales when investigating these trait‐climate relationships. For example, a strong, temporally unchanging leaf‐climate relationship would also support the use of the rich conifer leaf fossil record for reconstructing paleoclimates (Jordan 2011; Little et al. 2010). An ideal technique for examining the temporal stability of co‐evolutionary relationships between species leaf morphology and climate niche is Multi‐Response Phylogenetic Mixed Models (MRPMM) (Hadfield 2010; Halliwell et al. 2025). This class of models is capable of decomposing trait correlations into phylogenetic and independent components, where phylogenetic correlations reflect evolutionary processes at deep splits in the phylogeny and independent correlations reflect processes acting relatively recently or rapidly within a clade's' evolutionary history (Cornwell and Nakagawa 2017; Revell 2013; Westoby et al. 2023).

This study investigates variation in leaf size across all conifers, then focuses specifically on three phylogenetically and geographically distinct groups, each with a unique leaf architecture: Cupressaceae, Pinus and Podocarpus (Figure 1). Podocarpus species predominantly occur in warm and wet climates of the Southern Hemisphere (Figure 1) and are characterised by single‐veined broad (up to 20 mm wide) leaves with accessory transfusion tissue (ATT). Species within the genus Pinus have single‐veined needle leaves, with one exception (P. krempfii; Brodribb and Feild 2008). Pinus species are almost exclusively found in the Northern Hemisphere (Figure 1), and are known for their extreme frost and drought tolerance (Eckenwalder 2009; Farjon 2018). Given that scale leaves are a common leaf form among conifers, we also examine the broadly distributed scale‐leaved members of Cupressaceae (widespread in both hemispheres; Figure 1). We investigate the evolutionary relationships between leaf size (length and width) and climate across the conifer phylogeny, and assess evidence for deep time associations consistent with niche conservatism using MRPMM. We hypothesise that conifer leaf size is constrained by limitations imposed by the leaf hydraulic architecture, and that this has substantial impacts on species' distributions. Here, we investigate climatic variables that are likely to influence the size of conifer leaves; minimum seasonal temperatures (known to cause freezing‐induced embolism and mesophyll damage; Pittermann and Sperry 2006; Sakai and Larcher 2012), minimum seasonal precipitation (known to cause drought‐induced xylem cavitation; Mackey 1994) and maximum temperature (a predictor of leaf energy balance; Okajima et al. 2012; Wright et al. 2017).

FIGURE 1.

Geographic distribution of the three major conifer groups examined in this study, scale‐leaved Cupressaceae (yellow) and needle‐leaved Pinus (blue) predominating at higher absolute latitudes, and broad‐leaved Podocarpus (red) predominating at lower absolute latitudes. Point records from Larcombe et al. (2018).

2. Materials and Methods

2.1. Data

Species level information about leaf size and climatic distributions was compared using Gaussian generalised linear regressions (log link) and then further analysed by including phylogenetic information with phylogenetic heatmapping and Multi‐Response Phylogenetic Mixed Models (MRPMM). We used point location data for extant conifer species (Larcombe et al. 2018) collated by iteratively searching for additional records of individual species from the literature (Brown et al. 2021). We harmonised the point distribution and grid cell data with Leslie et al.'s (2018) fossil‐ and time‐calibrated phylogeny of the conifers. This phylogeny contains 580 species; however, we excluded 10 species because of a lack of reliable point distribution data. After this data cleaning, 570 species were included in final analyses, representing 70 genera and six taxonomic families.

Leaf minimum and maximum length and width were collected and averaged by extracting species adult leaf characteristics from taxonomic descriptions (Earle 2024; Farjon 2010). To facilitate visualisation of the relationships between leaf form and other traits, species were classified into were into four broad groups of leaf form: scale, bifacial, needle and multi‐veined. Multi‐veined leaves included all leaves with more than one vein, and was broadly defined to include phylloclades. Scale leaves were short (> 5 mm) and/or appressed to the stem. Needle leaves were less than 5 mm wide and lacked accessory transfusion tissue (ATT; defined as lateral water transport tissue that allows leaves to function as broad leaves). Bifacially flattened leaves included single‐veined leaves with maximum width greater than 5 mm, or with ATT. After individual verification, 36 Podocarpaceae species with mean leaf widths less than 5 mm were classified as bifacially flattened due to the presence of ATT. All of these species had bifacial anatomy. Estimates of climatic variables for each point record were derived from WorldclimV2 (Fick and Hijmans 2017), at 30″ (c. 1 km²) resolution. We focused on climatic variables that are known to control species' range limits (Brown et al. 2020) and reflect leaf hydraulic tolerance (Liu et al. 2020): mean precipitation of the driest quarter (PDQ) and mean minimum temperature of the coldest month (MinTCM) (Mackey 1994; Sakai and Larcher 2012). We also included analyses of maximum temperature of the warmest month (MaxT) to test the leaf energy balance theory (Okajima et al. 2012). Climatic correlations are reported in the supporting information (Table S2).

All analyses and data visualisation were performed in R Version 4.3.1 (R Core Team 2023), and figures were produced using phytools 2.0 (Revell 2023), ggplot2 (Wickham 2016) and ggtree (Yu 2020).

2.2. Phylogenetic Comparative Analysis

We used the MCMCglmm R package (Hadfield 2010) to fit multi‐response phylogenetic mixed models (MRPMM). A detailed overview of MRPMM is outlined in Halliwell et al. (2025). We fitted a model with five response variables (leaf length, leaf width, MinTCM, PDQ, MaxT) to the entire conifer phylogeny, and to assess whether the overall patterns occur within lineages independent of the effects of leaf form, we fitted separate models for the three largest groups with consistent and distinctive leaf forms: the genus Podocarpus, characterised by bifacially flattened leaves; Pinus, characterised by true needles; and scale‐leaved Cupressaceae species. We used these analyses to assess differences in phylogenetic signal of leaf and climate traits and leaf climate correlations among and within these major clades. All predictors, climate and leaf size, were log transformed and scaled prior to analysis.

We assessed model convergence by (1) visually inspecting traces and densities of the MCMC posterior estimates; and (2) confirming that potential scale reduction factors ($\widehat{r}$), a convergence diagnostic test that compares within‐ and between‐chain variance, were ≤ 1.01 for all parameter estimates (Vehtari et al. 2021). Estimates for all variance components converged successfully using parameter expanded priors with (variance) V = I _k (an identity matrix of dimension equal to the number of response traits, k), and (degree of belief) ν = k + 1 for random effects (de Villemereuil 2018), 110,000 iterations, a 10,000 burn‐in, and thinning of 100. We used the default independent normal priors with μ = 0 and V = 10¹⁰ for fixed effects. We performed model validation via posterior predictive checks presented in Appendix S1 (Figures S1–S8).

Parameter estimates from models are reported as posterior means with 95% and 50% credible intervals (Table S2), including estimates of phylogenetic signal (Table S1). For significance tests of correlations from MRPMM, we calculated the two‐sided posterior probability (pMCMC; Table S3), which is twice the proportion of the posterior samples below zero (or above zero, whichever is smallest) and closely related to the frequentist p value associated with the two‐sided null hypothesis of zero effect (Shi and Yin 2021). Phylogenetic signal is derived from variance components, which are bounded at 0. This violates the assumptions of pMCMC, therefore a significance test is not possible for phylogenetic signal without providing an arbitrary (non‐biologically informative) threshold for the null expectation (e.g., λ = 0.5).

Large differences in the ages of the four groups studied: (crown ages of 313 MYA for conifers, 145.8 MYA for scale‐leaved Cupressaceae, 92.3 MYA for Pinus and 49.8 MYA for Podocarpus) limits our capacity to directly compare the strength of phylogenetic signal and correlation among these clades. We therefore focus our interpretations on within‐clade patterns rather than directly comparing estimates across clades.

BOX 1. Interpreting phylogenetic signal and phylogenetic versus independent correlations.

With Multi Response Phylogenetic Mixed Models (MRPMM), it is possible to estimate both phylogenetic signal in individual variables, as well as correlations between variables within both phylogenetic and phylogenetically independent components of variance. Phylogenetic signal represents the proportion of the observed variation that can be attributed to the phylogenetic structure assuming a Brownian Motion model of evolution (Felsenstein 1988; Revell et al. 2008). Strong phylogenetic signal indicates phylogenetic trait conservatism (Crisp and Cook 2012; Losos 2008). Conversely, weak phylogenetic signal indicates evolutionary lability.

For two co‐occurring traits, it is possible to have phylogenetic signal in both, one or neither trait. A phylogenetically independent correlation represents the correlation between two traits after controlling for phylogenetic relationships, and can indicate the degree of convergent evolution across the phylogeny—this quantity is similar to a regression coefficient estimated in traditional phylogenetic comparative methodology, that is, phylogenetic generalised least squares (PGLS) (Felsenstein 1985). In contrast, a phylogenetic correlation represents an association between traits deeper in time that manifests in the correlated divergence of trait values among phylogenetic clades. Patterns of phylogenetic and independent correlations have complex but informative interpretations (Figure 2).

FIGURE 2.

Possible patterns of trait correlation and the inferred evolutionary processes that lead to those outcomes. Phylogenetic correlations assume phylogenetic signal in both traits and are typically estimated with greater uncertainty (wider confidence intervals) than independent correlations (Halliwell et al. 2025). While trait correlations may show no phylogenetic signal (i.e., correlated divergences are strictly independent of phylogeny), significant phylogenetic correlations are expected to produce at least some degree of independent correlation, due to phylogenetic uncertainty and the ongoing ecological divergence of species within subclades.

When correlations between traits and climate at the phylogenetic and independent levels are both in the same direction and are of comparable magnitude (Figure 2), evolution is assumed to have operated similarly both within closely related clades and across more distantly related clades in similar climates. This is consistent with processes that have been ongoing throughout the evolution of the group and is considered the most likely evolutionary outcome of functional trait relationships.

Phylogenetically independent correlation (Figure 2) is the classic example of independent trait co‐evolution, with associations emerging repeatedly across distantly or non‐related species. This suggests that the trait is effective in sustaining populations in some contexts (the associated climate) but not others, and that there is an underlying ecological (not merely phylogenetic) mechanism maintaining functional relationships between traits (Westoby et al. 2023).

Phylogenetic correlation (Figure 2) indicates a tendency for traits to change together relatively early in the evolutionary history of the group (i.e., at deep splits) and be conserved in the divergence of clades, consistent with niche conservatism or clade‐level ecological specialisation.

Opposing phylogenetic and independent correlations (Figure 2) suggest omission of phylogenetically structured confounding variables, for example, important compensatory traits that have evolved relatively recently (Anderegg 2023), especially if this is convergent across the phylogeny. Opposing correlations could also indicate that different processes operate over different evolutionary and/or biogeographic scales. While relatively uncommon, this could result from dramatic regime shifts in environment, occurring sometime in recent history (Schneebeli‐Hermann 2020). Scenarios where one correlation is strong and the other negligible can also indicate such changes in regime. However, this pattern can also occur when there is insufficient trait variation, or high uncertainty in trait variation, estimated at one or both levels.

3. Results

We observed two distinctive phylogenetic groupings of conifer leaf size and climate (Figure 3), with Podocarpaceae and Araucariaceae having the widest (bifacially flattened and multi‐veined) leaves. These species tend to be in the wettest and some of the warmest climates inhabited by conifers (Figure 3), whereas Pinaceae, Taxaceae and Sciadopityaceae generally have narrower leaves and tolerate colder winters and drier environments. Cupressaceae are widespread, but generally inhabit climates with colder winters and hotter and drier summers, and have many of the shortest leaves (Figure 3). Pinaceae occur in broadly similar climates to Cupressaceae, although their leaves are longer (needle‐like). The genus Pinus is easily distinguished by exhibiting the longest (but still narrow) leaves present in conifers (Figure 3).

FIGURE 3.

Phylogenetic heatmap of leaf type (outermost bar), size (length; innermost coloured bar, and width; second from inner) and climate (TCM—mean minimum temperature of the coldest month, third bar; PDQ—precipitation of the driest quarter, fourth bar; MaxT—maximum temperature of the warmest month, fifth bar) in conifers. Clades are represented by solid black lines, and families are delineated by the innermost lines, labelled. Clades extracted for modelling (scale‐leaved Cupressaceae, Pinus and Podocarpus) are delineated by the outermost lines, unlabelled. Light blue hues represent lower values; colder temperatures, drier precipitations, shorter and thinner leaves. Dark blue‐purple represent intermediate values. Light red hues are warmer and wetter climates, and longer and broader leaves. S = Sciadopityaceae. Discrete leaf types are represented by: Lightest grey = multi‐veined including phylloclades, medium grey = needle, darkest grey = bifacially flattened, and black = scale. Pinus krempfii is the only bifacially flattened species of Pinus. Continuous values are standardised for visualisation. Non‐standardised heatmaps of Pinus, Podocarpus and scale‐leaf Cupressaceae are presented in Appendix S1.

3.1. Phylogenetic Signal

Phylogenetic signal (λ) was over 50% for almost all trait and climatic predictors (Figure 4), demonstrating a pattern of trait and climate conservatism across the whole conifer phylogeny and within conifer clades. Conifers overall showed a high mean λ across all parameters, displaying very strong conservatism in leaf size (width λ = 93%, length λ = 90%) and climate (λ > 75%). Individual clades showed varying degrees of signal, almost all moderate to high. Perhaps partially a result of being the oldest (crown age of 145.8 MYA) sub‐clade modelled, scale‐leaved Cupressaceae demonstrated strong conservatism in climate; MaxT (λ = 85%), PDQ (λ = 77%) and MinTCM (λ = 69%), and strong signal in width (λ = 80%). Phylogenetic signal in Pinus was strongest in MaxT (λ = 79%), and was similarly strong in PDQ (λ = 73%) and length (λ = 71%). Despite being the youngest clade (crown age of 49.8 MYA), phylogenetic signal in leaf size (width λ = 66%; length λ = 62%) and climate (PDQ λ = 54%; MinTCM λ = 47%; MaxT λ = 53%) was moderately strong in Podocarpus, indicating a pattern of phylogenetic conservatism comparable in scale to much older conifer clades.

FIGURE 4.

Phylogenetic signal (λ) estimated from MRPMM, representing the proportion of variation of each trait attributed to phylogenetic effects. Points represent means, thin lines represent 95% credibility intervals (CI), and thick lines represent 50% CIs of the posterior probability distribution. Conifers overall in black, scale‐leaved Cupressaceae species in yellow, all species in the genera Pinus and Podocarpus in blue and red, respectively (see Table S1 for actual values). See Box 1 for information on interpreting phylogenetic signal among variables.

3.2. Patterns of Leaf Size and Climate

Conifers overall tend to have longer and broader leaves in warmer climates (Figure 5a,c,d), and longer leaves in climates with wetter dry seasons (Figures 5b and 6a), however, these patterns are not consistent within the separate clades we examined. While there is a trend for broader leaves in wetter climates across all conifers (Figure 5e), this relationship is not apparent within any of the major clades examined and was not supported by MRPMM analyses (Figure 6a). In Podocarpus, leaves are longer (r _phy(MaxT) = 0.88, pMCMC = 0) and broader (r _phy(MinTCM) = 0.87, pMCMC = 0.002) in warmer climates (Figures 5 and 6c). Although precipitation does not play a significant role in determining Podocarpus leaf size, they tend to be longer in drier climates (Figures 5b and 6c; r _phy(PDQ) = −0.4, pMCMC = 0.19). In Pinus, leaf width does not vary much due to the needle habit; however, there is a non‐phylogenetic (recently evolved) pattern of broadening in response to colder winters (Figure 6d; r _ind(MinTCM) = −0.14, pMCMC = 0.04). Pinus leaves are longer in warmer (Figures 5a,c and 6d; r _phy(MinTCM) = 0.47; pMCMC = 0.13) and drier climates (Figures 5b and 6d; r _phy(PDQ) = −0.42; pMCMC = 0.9), although this is not significant with MRPMM. There is little relationship between leaf size (length or width) and climate in Cupressaceae (Figures 5 and 6b).

FIGURE 5.

Generalised linear regressions (Ga.ussian, log link) of leaf length and width in all conifers (grey), Pinus (blue), Podocarpus (red), and scale‐leaved Cupressaceae (yellow) against minimum temperature of the coldest month (MinTCM; °C), mean driest quarter precipitation (PDQ; mm) and maximum temperature of the warmest month (MaxT; °C). Lines show fitted values and shaded bands show 95% confidence intervals.

FIGURE 6.

Phylogenetic and residual (independent) correlation coefficients estimated from MRPMM. Points represent the posterior median for each estimate, with 50% (thick bar) and 95% (thin bar) credibility intervals. MaxT is maximum temperature of the warmest month, MinTCM is mean temperature of the coldest month, PDQ is mean precipitation of the driest quarter. Table S2 reports posterior medians and 50% and 95% CI values, and correlations between climatic variables. Table S3 reports pMCMC results. See Box 1 for an explanation of how to interpret phylogenetic and independent correlations.

The strongest relationships between climate and leaf traits were positive phylogenetic correlations between temperature and both leaf length and width in Podocarpus (Figure 6c), indicating that divergences in leaf size and temperature niche are correlated among clades within this genus. Leaf size showed significant phylogenetic correlations with temperature in conifers overall (width r _phy(MinTCM) = 0.36, pMCMC = 0.004; length r _phy(MinTCM) = 0.3, pMCMC = 0.004; Figure 6a), however, these relationships, as well as those within other focal clades, were much weaker than those observed in Podocarpus. Across the board, leaves were broadest in the warmest summer temperatures, indicating leaf energy balance may not be a limiting factor for leaf size.

Correlations between leaf length and PDQ were generally negative (Figures 5 and 6), which is unexpected given the trend of larger leaves in wetter climates among angiosperms. Aside from the independent association of broader leaves with colder climates in Pinus, all independent correlations between climate and leaf size were weak and non‐significant, indicating that correlated divergences between climate niche and leaf size in deeper time are the primary drivers of trait‐climate associations among extant conifer species.

4. Discussion

Our results provide strong evidence that leaf size variation in conifers reflects deep‐time evolutionary associations with climate. Moderate to strong phylogenetic signal in leaf width, leaf length, temperature and precipitation for the conifer clades studied here (Figure 4) suggests that conifers adapt slowly to climate, primarily retaining their ancestral climatic niche rather than displaying rapid phenotypic adaptation to novel climates. The strongest trait‐climate correlations occur between leaf size and temperature within the phylogenetic components of variance (Figure 6), implying that morphological conservatism in leaves and tolerance to extreme temperatures controls conifer species distribution. Maximum summer temperatures do not constrain leaf width, implying that leaf energy balance is not a factor controlling conifer leaf size. Freezing temperatures do, however, constrain leaf size, indicating that the capacity of some conifer groups, particularly Podocarpus, to inhabit cold climates is controlled by how well leaf traits that determine size, likely hydraulic anatomy, tolerate freezing (Figure 6).

Comparing three contrasting conifer groups (needle‐leaved Pinus, broad‐leaved Podocarpus and scale‐leaved Cupressaceae) reveals strong evidence of clade‐level evolutionary associations between leaf size and climate, each different from each other and from all conifers considered together. The presence of these associations within groups with homogenous leaf form is clear evidence against overall leaf size/climate relationships being artefacts of differences in leaf form or among major clades. Associations between climate and leaf size in Cupressaceae were non‐significant, despite particularly strong conservatism in both; indicating scale‐leaves are a successful and generalist leaf form able to tolerate a range of climates across deep time, with little phenotypic adjustment.

4.1. Leaf Architectural Constraints and Freezing

Xylem anatomy directly influences leaf size by determining the distance water can be transported to irrigate photosynthetic tissues (Katifori 2018), and indirectly shapes climatic associations through its control of physiological frost and drought tolerance (Blackman et al. 2010; Brodribb et al. 2007). Angiosperms exhibit extensive diversity in leaf xylem anatomy and consequently leaf size, allowing them to occupy broad climatic niches (Feild et al. 2011; Nicotra et al. 2011). In contrast, conifers are constrained to tracheid‐based xylem (Pittermann et al. 2011), which limits their potential leaf size and likely restricts their bioclimatic distributions.

The contrast between Pinus and Podocarpus illustrates how xylem anatomical differences have led these successful conifer genera to occupy markedly different climates (Jordan and Brodribb 2026). Both groups show phylogenetic correlations between leaf size and temperature, but this relationship manifests only for leaf length in Pinus, while apparent for both length and width in Podocarpus. Accessory transfusion tissue (ATT) in Podocarpus represents an important innovation that overcomes width constraints inherent in single‐veined leaves, facilitating broad leaves that enable efficient light capture particularly in the understory of highly competitive, climatically favourable (warm and wet) niches often dominated by angiosperms (Brodribb et al. 2010). Indeed, many broad leaved Podocarpus species, particularly in tropical and mesic environments, are understorey components (e.g., Podocarpus dispermus). The species that are characteristic of open vegetation tend to have smaller leaves and are associated with cold (e.g., Podocarpus lawrencei, P. nivalis ) or extremely nutrient poor environments (e.g., P. gnidioides or P. rostrata ). It is likely that the amount and length of ATT is somewhat labile, due to the broad range in leaf size across the genus (and the positive though non‐significant independent correlation between width and MinTCM within Podocarpus, Figure 6c), allowing width variation along the temperature gradient. In contrast, the lack of leaf lateral water transport tissue in all but one species of Pinus can explain the limited width variation and the related lack of correlation between width and climate in this genus. The exception proves the rule in this case: in an example of convergent evolution, the only Pinus species to evolve broad leaves and ATT‐analogous tissue, Pinus krempfii, is the only Pinus species successful in warm tropical rainforests (Brodribb and Feild 2008; Willyard et al. 2007). The relative inefficiency of the conifer tracheidal water‐transport system combined with reduced light capture by narrow‐leaves results in a competitive disadvantage against growth rates of broad‐leaved angiosperms in well‐watered forests (Zhang et al. 2020). This underlies the hypothesis that conifers are more common in climatically hostile environments (Bond 1989), though this generalisation notably overlooks broad‐leaved conifers such as Podocarpus.

While accessory transfusion tissue permits Podocarpus (and others like Pinus krempfii) to broaden their leaves and thus occupy low‐light environments (i.e., understoreys of closed forests), Podocarpus are noticeably absent from extremely cold environments. Higher leaf‐vein densities have abundant secondary and tertiary vein‐endings (Gleason et al. 2018), which can lead to ice crystallisation in the mesophyll, irrigated by vein endings (Johnson, Scherer, et al. 2025), during a freezing event. If ATT is functionally similar to the high density reticulate venation found in broad‐leaf angiosperms (Brodribb et al. 2007), perhaps similar consequences for frost tolerance apply. Presence of ATT in conifer leaves likely results in analogous ‘vein endings’ and thus increased sites for freeze nucleation. Pinus needles, which do not contain ATT, minimise frost damage by confining freezing to the lignified endodermis, spatially separating ice crystallisation from the mesophyll (Stegner et al. 2023). Needle thickening in cold temperatures is a more recent evolutionary adaptation in Pinus (indicated by a negative independent correlation with MinTCM; Figure 6d). The most‐cold tolerant Pinus species have developed thicker epidermal cells and cell walls alongside increased resin duct volume insulating the mesophyll from frost damage in their needles (Jankowski et al. 2017). This recent adaptation may be associated with the appearance of extensive and widespread freezing climates in the Northern Hemisphere during the Pleistocene glaciations (Meyers and Hinnov 2010).

Subtleties in leaf anatomy have resulted in contrasting evolutionary histories and temperature associations for the two largest conifer genera, Pinus and Podocarpus (Jordan and Brodribb 2026). Both strategies were successful in maintaining and even dominating their ancestral climates, but dispersal across the equator for either group has seen much less success. The tropics have proven too competitive and light‐limiting for the slender needles of Pinus, whereas the hostility of the arid and freezing continental Northern Hemisphere has proven too challenging for Podocarpus to leave the tropics (Jordan and Brodribb 2026).

4.2. Aridity, Leaf Energy Balance and Leaf Size

In comparison to freezing, the story for precipitation is not as straightforward. Overall, Pinus and Podocarpus leaves are longer in more arid climates (Figure 5b), contrary to patterns shown in angiosperms (Wright et al. 2017). However, neither the corresponding phylogenetic nor independent correlations were significant (Figure 6). Podocarpus is largely excluded from extremely arid climates so it is unlikely that drought is driving this relationship. In Pinus leaves, patterns of conduit tapering with smaller tracheids at the tip than at the base imply that short leaves should be favoured in arid climates (Bicego et al. 2026), contrary to what we saw. Given this context and the non‐significance of the association between long leaves and more arid climates, the unexpected correlations we encountered may be the product of confounding variables correlated with aridity, rather than drought itself.

While drought has been a key evolutionary driver of leaf size in angiosperms (Skelton et al. 2021; Wright et al. 2017), aridity was not a universal process shaping the evolution of conifer leaf size, particularly for lineages that evolved primarily in mesic environments. Conifers currently occupy more environments with freezing as the primary stressor (Leites and Benito Garzón 2023), but only a few groups have managed to tolerate aridity (McCulloh et al. 2023), with some species, like Callitris (a scale‐leaved member of Cupressaceae), doing so very successfully (Larter et al. 2017). Certain leaf anatomical traits that improve tolerance to freezing, such as smaller leaves, lower tracheid density or cavitation‐resistant xylem tracheids, can also be advantageous under drought. This may explain why drought tolerance occurs in some conifers even if it was not the main evolutionary pressure (Camarero et al. 2024). Future studies focusing on leaf anatomy and their physiological tolerance to drought and freezing are needed to clarify these relationships.

Leaf energy balance is another important factor shaping leaf size in angiosperms (Wright et al. 2017), but this does not seem to be the case in conifers. Leaf temperature and associated boundary layer dynamics are dependent on leaf width, a trait that does not vary much in most conifers, particularly in two of our target clades, Cupressaceae and Pinus. Both clades are found in the hottest climates inhabited by conifers (Australia and North America) and have thin or small leaf surface areas that buffer impacts of high temperatures on energy balance (Okajima et al. 2012). Width variation in Podocarpus is temperature related, where leaves are broader in warmer summers and winters, although this indicates that high temperatures are not controlling or limiting leaf size. It is more feasible that leaf size is limited by low temperatures and the freeze–thaw impact on leaf hydraulics. Podocarpus species found in warm climates are often shaded rainforest understorey components where leaf irradiance and VPD are low, and moisture is abundant (Brodribb 2011; Jordan and Brodribb 2026). As a result, constraints to leaf size in conifers are more likely due to the limitations imposed by hydraulic anatomy rather than trade‐offs associated with leaf energy balance.

Conifers that are ancestrally associated with dry or cold climates have thrived by persisting unchanged in environments that exceed most other plants' environmental tolerance, with their small, scale (i.e., Cupressaceae) or needle (i.e., Pinus) leaves well‐suited to tolerate these extreme conditions (Bond 1989; Larter et al. 2017). Species associated with mesic climates (i.e., Podocarpus), while also remaining largely unchanged, have been subject to additional filtering through biotic interactions, chiefly light competition, especially since the rise of angiosperms (Brodribb and Feild 2008; Hill and Brodribb 1999). These pressures suggest that broad‐leaved conifers, unable to survive the extremes of dry or cold climates, have remarkably persisted over 100 million years of angiosperm competition with little hydraulic advantage and virtually no leaf size adaptation (Brodribb et al. 2010; Brown et al. 2021; Condamine et al. 2020).

4.3. Consequences of Morphological and Climatic Conservatism

Our results highlight high phylogenetic signal and strong leaf size‐climate phylogenetic correlations among conifers, supporting the growing body of evidence suggesting that phylogenetic niche conservatism is a key process underpinning global conifer evolution and ecology (e.g., Brown et al. 2021; Crisp and Cook 2012; Losos 2008). Given the advantages of rapid adaptation shown overwhelmingly by the dominance of angiosperms (Berendse and Scheffer 2009), the capacity of conifers to adapt leaf size to exploit new climates is clearly constrained, a pattern that is consistent across their evolutionary history. Aside from early leaf morphological innovations like ATT, conifers are largely unable to modify their leaf size to exploit new environments and thus remain within their paleoclimatic niche space, and retain phenotypes similar to their fossilised relatives. Strong morphological conservatism in leaf hydraulic architecture is thus likely a driver of phylogenetic niche conservatism in conifers.

Conifers in the Southern Hemisphere, in particular, are highly vulnerable to climatic shifts (Offord 2011), likely a consequence of their strong phylogenetic niche conservatism. The most range‐limited conifers today are commonly species from the Podocarpaceae, Araucariaceae and some southern Cupressaceae that had warm and/or wet paleoclimates in their early history and now have become rarer as the continents drifted south (Jordan et al. 2016; Scotese 1998). Similarly, species with cold paleoclimates, such as the Tasmanian highland paleo‐endemics (Fitzgerald and Whinam 2012), are becoming increasingly isolated to high altitude refugia after post‐glacial warming (Gentili et al. 2015; Jordan et al. 2016). The surviving but increasingly rare conifers have found stable, isolated climatic niches (Harrison and Noss 2017), but trait conservatism means these species are at risk of extinction when their ancestral niche disappears, as is occurring worldwide with climate change.

5. Conclusion

Our results reinforce the growing body of evidence that phylogenetic niche conservatism is a key process shaping the global biogeography and evolutionary history of conifers, and that this is likely underpinned by conservatism in leaf morphology and hydraulic anatomy. Phylogenetic patterns emerged between leaf size and climate, most prominently a consistent reduction in size for cold climates. We highlight that the two most diverse conifer genera, Pinus and Podocarpus, demonstrate entirely divergent evolutionary climatic associations, driven fundamentally by the constraints and innovations of the hydraulic architecture within their divergent leaf types.

Despite shifts in global climate, extant conifers have tended to persist within climatic envelopes similar to their ancestral climates throughout much of their evolutionary history (Brown et al. 2021; Cruz‐Nicolás et al. 2024; Wang et al. 2025). This is a clear indication that evolutionary niche conservatism is occurring in the conifers. Phylogenetic correlations indicate leaf size adaptation to climate occurred primarily in the distant past, and more recently species are moving to track these ancestral climates rather than adapting to new ones. One exception is the relatively recent adaptation of increased leaf width in response to freezing temperatures in Pinus, which may have contributed to the success of this genus. Given the strong niche conservatism shown here, many conifers may be disadvantaged in rapidly changing climates, unable to move or adapt adequately to survive the change. This fundamental insight into the evolutionary history of conifers will aid in reconstructing past environments and highlights the importance of climate in shaping past, present and future conifer distributions globally.

Rigorous consideration of phylogenetic effects when modelling leaf‐climate associations is clearly vital to understand the evolution of plant functional traits. This is enabled by advancements in phylogenetic comparative methods, especially the capacity of multivariate models to disentangle trait‐climate relationships that are influenced by shared evolutionary history (phylogenetic signal) and those independent of it (Hadfield 2010; Halliwell et al. 2025). Without this important distinction embedded in these models, the evolutionary associations we have shown to be important for conifers would have been overlooked.

Author Contributions

Katya I. Bandow: conceptualization, investigation, writing – original draft, methodology, validation, visualization, writing – review and editing, formal analysis, data curation. Timothy J. Brodribb: conceptualization, investigation, funding acquisition, writing – original draft, methodology, validation, visualization, writing – review and editing, supervision, resources. Benjamin Halliwell: investigation, methodology, validation, visualization, writing – review and editing, formal analysis. Matilda J. M. Brown: writing – original draft, writing – review and editing, methodology. Kate M. Johnson: writing – original draft, writing – review and editing. Gregory J. Jordan: conceptualization, investigation, funding acquisition, writing – original draft, methodology, validation, visualization, writing – review and editing, formal analysis, project administration, supervision, resources.

Funding

This work was supported by the Australian Research Council Centre of Excellence for Plant Success in Nature and Agriculture.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Figure S1: Model 1 (all conifers) Validation. MCMC posterior predictive density plots for each trait, leaf length, leaf width, minimum temperature of the coldest month (MinTCM), precipitation of the driest quarter (PDQ) and maximum temperature of the warmest month (MaxT). Blue lines represent simulated datasets drawn from the posterior distribution, and the black line shows observed data. Overlaps between observed and simulated densities indicates good model fit.

Figure S2: Model 2 (scale‐leaved Cupressaceae) MCMC posterior density model validation.

Figure S3: Model 3 (Pinus) MCMC posterior density model validation.

Figure S4: Model 4 (Podocarpus) MCMC posterior density model validation.

Figure S5: Model 1 (all conifers) MCMC parameter trace plots model validation. In MCMCglmm, the estimated phylogenetic component for signal and correlation is ‘animal’ and for independent it is ‘units’. Model parameters are: leaf length, leaf width, minimum temperature of the coldest month (MinTCM), precipitation of the driest quarter (PDQ) and maximum temperature of the warmest month (BIO5).

Figure S6: Model 2 (scale‐leaved Cupressaceae) MCMC parameter trace model validation plots. In MCMCglmm, the estimated phylogenetic component for signal and correlation is ‘animal’ and for independent it is ‘units’. Model parameters are: leaf length, leaf width, minimum temperature of the coldest month (MinTCM), precipitation of the driest quarter (PDQ) and maximum temperature of the warmest month (BIO5).

Figure S7: Model 3 (Pinus) MCMC parameter trace plots model validation. In MCMCglmm, the estimated phylogenetic component for signal and correlation is ‘animal’ and for independent it is ‘units’. Model parameters are: leaf length, leaf width, minimum temperature of the coldest month (MinTCM), precipitation of the driest quarter (PDQ) and maximum temperature of the warmest month (BIO5).

Figure S8: Model 4 (Podocarpus) MCMC parameter trace plots model validation. In MCMCglmm, the estimated phylogenetic component for signal and correlation is ‘animal’ and for independent it is ‘units’. Model parameters are: leaf length, leaf width, minimum temperature of the coldest month (MinTCM), precipitation of the driest quarter (PDQ) and maximum temperature of the warmest month (BIO5).

Figure S9: Phylogenetic heatmap of leaf length, leaf width, minimum temperature of the coldest month (MinTCM), precipitation of the driest quarter (PDQ) and maximum temperature of the warmest month (MaxT) for all scale‐leaved Cupressaceae species.

Figure S10: Phylogenetic heatmap of leaf length, leaf width, minimum temperature of the coldest month (MinTCM), precipitation of the driest quarter (PDQ) and maximum temperature of the warmest month (MaxT) for all Pinus species. All species are needle‐leaved.

Figure S11: Phylogenetic heatmap of leaf length, leaf width, minimum temperature of the coldest month (MinTCM), precipitation of the driest quarter (PDQ) and maximum temperature of the warmest month (MaxT) for all Podocarpus species. All species have bifacially flattened leaves.

Table S1: Phylogenetic signal (λ) estimated from MR‐PMM, representing the proportion of variation of each trait attributed to phylogenetic effects. MinTCM is mean temperature of the coldest month, PDQ is mean precipitation of the driest quarter, MaxT is maximum temperature of the warmest month. Cupressaceae models only include scale‐leaved species. Conifer models included leaf form as a fixed effect co‐variate.

Table S2: Phylogenetic and phylogenetically independent correlation coefficients estimated from MRPMM. MinTCM is mean temperature of the coldest month, PDQ is mean precipitation of the driest quarter, MaxT is maximum temperature of the warmest month. Cupressaceae models only include scale‐leaved species. Conifer models included leaf form as a fixed effect co‐variate.

Table S3: Posterior probability (pMCMC) estimated using MCMCglmm. pMCMC values are posterior medians of a two‐sided posterior probability. pMCMC values are similar to the frequentist p value, where significance is determined at < 0.05. Effects are unsupported if they include zero (0), moderately supported if excluded (1) at the 50% credibility interval, and strongly supported if excluded (1) at the 95% credibility interval.

Data Availability Statement

The data and code associated with this study are openly accessible through the Dryad data repository at https://doi.org/10.5061/dryad.3tx95x6vw.