Modes of climate variability bridge proximate and evolutionary mechanisms of masting

Abstract

There is evidence that variable and synchronous reproduction in seed plants (masting) correlates to modes of climate variability, e.g. El Niño Southern Oscillation and North Atlantic Oscillation. In this perspective, we explore the breadth of knowledge on how climate modes control reproduction in major masting species throughout Earth's biomes. We posit that intrinsic properties of climate modes (periodicity, persistence and trends) drive interannual and decadal variability of plant reproduction, as well as the spatial extent of its synchrony, aligning multiple proximate causes of masting through space and time. Moreover, climate modes force lagged but in-phase ecological processes that interact synergistically with multiple stages of plant reproductive cycles. This sets up adaptive benefits by increasing offspring fitness through either economies of scale or environmental prediction. Community-wide links between climate modes and masting across plant taxa suggest an evolutionary role of climate variability. We argue that climate modes may ‘bridge’ proximate and ultimate causes of masting selecting for variable and synchronous reproduction. The future of such interaction is uncertain: processes that improve reproductive fitness may remain coupled with climate modes even under changing climates, but chances are that abrupt global warming will affect Earth's climate modes so rapidly as to alter ecological and evolutionary links.

This article is part of the theme issue ‘The ecology and evolution of synchronized seed production in plants’.

1. Introduction

In 1997–1998, major plant reproductive events took place in many regions of the world, including tropical forests in South East Asia [1,2] and Central America [3,4], temperate forests of China and Japan [5,6], and boreal forests of northwest and northeast America [7,8]. The spatio-temporal synchrony of seeding in plants displaying variable reproduction (masting) has long interested ecologists [9]. Successful masting depends on the completion of consecutive stages of the reproductive process (i.e. resource uptake, floral induction, pollination and seed maturation), often spanning multiple seasons or years [10–13]. Weather can affect each stage by priming resource uptake, cueing floral bud formation, influencing pollination success or vetoing seed production [11,14]. Consequently, when favourable weather conditions are aligned in time and space over the entire reproductive process, i.e. the Moran effect [15,16], they lead to synchronous seed production in masting plants [13,17–19].

Many environmental patterns that play out over wide geographical areas are controlled by modes of climate variability, such as the El Niño Southern Oscillation (ENSO) [20], the North Atlantic Oscillation (NAO) [21] and other modes that are integral components of Earth's climate system [22–25]. Different climate modes operate at different time scales, affecting weather patterns at frequencies from months to decades [20,21,26]. The spatio-temporal dynamics with which climate modes exert their influence has major implications for the synchronization of ecological processes and ecosystem functioning [22,25,27], including masting. Indeed, there is increasing evidence that masting events are associated with these modes of climate variability across all continents [2,3,28–31] but the existence of a causal link has yet to be ascertained.

In this perspective, we explore the breadth of knowledge on how climate modes control variable and synchronous seed production throughout Earth's biomes. Under the hypothesis that the intrinsic properties of climate modes (periodicity, persistence and trend) drive proximate causes of seed production, we show that they coherently align, over multiple seasons, weather cues conducive to the success or failure of seed production in masting species. Furthermore, we assess the evidence that climate modes influencing the spatio-temporal combinations of proximate mechanisms drive the spatial synchrony/asynchrony of masting in several geographical regions.

Notably, modes of climate variability have tremendous cascading effects on multiple ecosystem processes, such as primary productivity [24,32], natural disturbance regimes [33,34] and animal population dynamics [22], and have the potential to force in-phase ecological processes [27] that interact synergistically with multiple stages of plant reproduction. We explore the hypothesis that climate modes create such synergies via their autocorrelation structure, lagged effects on ecosystem dynamics and density-dependent processes related to the extent of synchrony. From an evolutionary point of view, using climate modes as a pacemaker to time the production of large-seed crops can increase the fitness of an individual plant by promoting the success of its offspring.

Finally, we discuss whether the widespread concordance between seed production and climate modes has emerged by chance or is an evolved strategy, and to what extent it can be disrupted by global climate change.

2. Correlation of seed production with climate modes: a widespread phenomenon

Over the past three decades, studies have associated modes of climate variability and plant reproduction across several continents (figure 1), biomes and species (electronic supplementary material, table S1). Most studies [3,30,40] have analysed the correlation between time series of reproductive effort and large-scale climate indices [22]—often defined as “teleconnection” indices. For example, the NAO is described via the NAO index [21], while ENSO is characterized by the Oceanic Niño Index (ONI) (see electronic supplementary material, glossary for indexes description).

Figure 1.

Regional distribution of selected published studies (extracted from the electronic supplementary material, table S1) reporting correlations between climate indices and masting. Shading indicates regions where climate variability is influenced by the El Niño Southern Oscillation (ENSO; orange) and the North Atlantic Oscillation (NAO; blue).

Seminal research first documented a correlation between ENSO and mass fruiting in Dipterocarp tropical forests of South East Asia [28], where ENSO strongly influence weather patterns (electronic supplementary material, figure S1). ENSO has subsequently been shown to correlate with seed production not only in South East Asia [2,39–44], but also in New Zealand [29], Africa [38], western North America [7,31], South America [3,36,46] and in the Caribbean [4,47].

In the Northern Hemisphere, the NAO is one of the major climate modes concordant with plant reproduction (electronic supplementary material, table S1), particularly in Europe (figure 1) where the NAO index [21] correlates to the timing, variability and synchrony of flowering, pollination and seed production of dominant forest species [13,30,37,48–52].

More generally, several episodes of community-wide mass flowering or fruiting have been tied to the main climate modes influencing a given area [1,3,29,30,35,37,45,53], suggesting that species with disparate life traits may all time their reproduction based on these modes.

The strength and direction of correlations between climate modes and seed production may vary in space and time, following the spatial arrangement of each mode: dipolar or more complex patterns [2,27,31,53,54], depending on the season (electronic supplementary material, figure S1) or on the time window considered [30], and being modulated by local orographical features which can enhance, reduce or revert the large-scale signal. Furthermore, modes interact with each other [33,55–57]: higher frequency climate modes such as ENSO or NAO are affected by lower frequency modes such as the Atlantic multidecadal variability (AMV) [58]. Such complex interactions appear to be key in understanding cascading effects on linked ecological processes [33,59], including masting [7].

3. Climate modes affect the (dis-)alignment of proximate mechanisms for variable reproduction

Masting depends on the completion of multiple reproductive stages (i.e. resource priming, floral induction, flowering, pollination and fruit maturation) that may occur over several seasons or years [13,18,60]. Numerous studies linked reproductive success or failure to weather conditions experienced by plants during single reproductive stages [12,17]. Other studies correlated favourable weather during these stages to modes of climate variability [3,4,13,45,48], suggesting that climate modes are a key regulator of variable reproduction via their direct effect on weather, and particularly on temperature (electronic supplementary material, figure S1).

However, we highlight evidence that climate modes can also synergistically align favourable weather conditions during consecutive reproduction stages (electronic supplementary material, table S1). Climate modes are determined by quasi-oscillatory nonlinear dynamics arising from the interaction of oceanic and atmospheric processes and influence the persistence of circulation patterns over large portions of the globe across multiple seasons [20–22]. For example, ENSO fluctuations between La Niña and El Niño phases display non-stationary periodicity [20]. Climate indices used to represent ENSO, such as ONI (see glossary), display intrinsic autocorrelation (electronic supplementary material figure S2a), positive at lags of six months (i.e. persistence of El Niño/La Niña) and negative at lags of 2 years (i.e. transition from the opposite phase), with a periodicity ranging from 2 to 8 years (electronic supplementary material figure S2b). These fluctuations have the potential for a multi-seasonal alignment of proximate mechanisms, which would occur for example under increased resource acquisition followed by favourable weather during flowering, pollination and fruit ripening.

(a) . Aligned reproductive stages under the El Niño Southern Oscillation

In tropical forests of South East Asia, cool and dry anomalies from December to February are believed to cue mass flowering in Dipterocarp forests: in Malaysia general flowering typically occurs in March and in southwestern Borneo in July [2,19,28,41,45,60]. Our analyses confirm significant negative temperature and precipitation anomalies in the region prior to general flowering events in the last 60 years (electronic supplementary material, figure S3). Ushio et al. [19] and Satake et al. [60] show that general flowering in the region is driven synergistically by low air temperature and drought. Moreover, Chechina & Hamann [2] suggest a cumulative-trigger model to predict flowering in Dipterocarps that incorporates temporal accumulation of resources. Interestingly, these regional masting events occur preferentially during the transition from La Niña (negative ENSO phase) to the onset of El Niño (positive ENSO phase) (electronic supplementary material, figure S4) indicating a potential dynamic role of ENSO in aligning cumulative and synergic cues. A wet period during La Niña stimulates resource priming [45,58] (electronic supplementary material, table S1), while the onset of El Niño reduces cloud cover [58], leading to increased daytime radiation, lower night-time temperatures (electronic supplementary material, figure S5a), and dry conditions (electronic supplementary material, figure S5b) over peninsular Malaysia, northern Sumatra and western Borneo, approximately 12 months before the El Niño peak (usually from December to February). In these Dipterocarp forests, ENSO temporally aligns increased resources and the dry-cool air cue of flowering. A similar synergic alignment of resource availability and flowering cues during the onset of El Niño has been observed in tropical moist forests in Central America [3].

Synergies caused by the transition from La Niña to El Niño also appear relevant for Picea glauca reproduction in western [7] and eastern North America (electronic supplementary material, figure S4). Masting of P. glauca in eastern regions of North America is associated with temperature differences from the two prior summers [61], a cue known as ΔT [17]. Notably, over the last six decades, the summer before the peak of El Niño showed negative anomalies in temperature throughout eastern North America (electronic supplementary material, figure S5g), while the following summer after the El Niño winter peak, positive anomalies occurred across the same region (electronic supplementary material, figure S5h). This mechanism highlights how in eastern North America the transition from La Niña to El Niño aligns the negative summer temperature cue 2 years before masting and the positive temperature cue the summer before masting, resulting in a significant positive ΔT cue (electronic supplementary material, figure S6).

(b) . Aligned reproductive stages under the North Atlantic Oscillation

Another example of multi-seasonal synergies comes from Central Europe, where positive winter NAO and negative summer NAO during the resource priming stage correlate positively with reproduction in several tree species (electronic supplementary material, table S1), including Fagus sylvatica, Picea abies, Quercus robur, Quercus petraea, Betula pendula and Pinus pinea [13,30,37,49,62,63]. In F. sylvatica and P. abies, positive summer NAO during floral bud initiation is associated with masting in the following year [30,37,64]. Positive winter to spring NAO correlates with increased and synchronous pollen influx [48,51] and high seed crops in several species [30,37]. The explanation for these relationships lies in how seasonal NAO phases generate consecutive weather conditions that are optimal for reproduction and thus promote masting (figure 2). In Central Europe, positive NAO in December–February is associated with warm-wet winters (electronic supplementary material, figure S1) and earlier snowmelt [21], which promotes earlier leafing out [48]. Warm-wet winters also increase water availability in spring and the length of the growing season [49] with direct consequences on ecosystem productivity [32]. Positive summer NAO is associated with warm temperature in July–August (electronic supplementary material, figure S1), which induces floral bud differentiation in several species in Central Europe [18,37], while positive spring NAO is associated with warm, dry and windy weather (electronic supplementary material, figure S1) that favours pollination [30,37].

Figure 2.

Large-scale weather patterns associated with negative (left) and positive (right) phases of the NAO during key reproductive stages (resource priming, floral induction, pollination) leading to Fagus sylvatica masting failure (left) or success (right). Positive NAO aligns consecutive mechanisms that promote successful seed production, such as warm-wet winters and extended wet growing seasons favouring resource uptake, hot summers triggering floral bud initiation, dry springs ensuring successful flowering, pollination and fruiting (electronic supplementary material, table S1), leading to large-scale masting in Central Europe in 1995. The opposite occurs in 1966 during prolonged negative NAO phases. The December to March (DJFM) NAO index displayed in the bar plot was calculated using the National Oceanic and Atmospheric Administratoin index. The large-scale beech masting index in Central Europe was derived by Ascoli et al. [30] and normalized (i.e. negative/positive values indicate higher/lower than average seed crops).

Several of these multi-seasonal reproductive cues align synergistically owing to the intrinsic temporal structure of the NAO (electronic supplementary material, figure S7). The NAO index averaged over December to April is positively autocorrelated with a 1-year lag (electronic supplementary material, figure S7a), probably owing to re-emergence of oceanic heat anomalies from one winter to another [65] and to decadal influences by the AMV [56]. Furthermore, the NAO shows a marked persistence from winter to early spring months, as shown by the correlation of the spring with the preceding winter NAO index (electronic supplementary material, figure S7b). In Central Europe periods of positive winter, NAO phases are associated with increased resource uptake in European forests [24,32,66]. Importantly, the lag-1 positive autocorrelation of the winter NAO index (electronic supplementary material, figure S7a) means that a positive winter NAO can persist for consecutive years, thus promoting a prolonged resource gain. Moreover, since positive spring NAO probably follows positive winter NAO (electronic supplementary material, figure S7b), this alignment promotes earlier and synchronized flowering and associated release, dispersal and transport of pollen [48] during a period of increased resources [30].

Another interesting mechanism that links NAO properties to plant reproduction in Central Europe depends on the lagged effect of the winter NAO on summer temperatures. Indeed, summer heat waves in Central-Northern Europe are strong and wide-ranging when positive summer NAO occurs in years of positive winter NAO [67]. For example, this was the case in 1994 before the beech masting in 1995, the largest event in the period 1952–2015 (figure 2). This suggests a potentially higher sensitivity of beech masting to the summer cue during a positive phase of the winter NAO [30]. Together with the higher probability of favourable conditions for pollination in spring, this means that positive trends in winter NAO increase the likelihood of beech masting (figure 2). On the contrary, prolonged negative periods can result in poor resource uptake and vetoes to pollination (wet spring, frost), leading to frequent reproductive failure and longer inter-mast periods, such as those that occurred in Central Europe in the 1960s and 1970s (figure 2).

4. Modes of climate variability affect reproductive synchrony through space and time

The geographical extent of masting is one of its most impressive properties [68]. The spatial coherence of proximate mechanisms of seed production (i.e. the Moran effect) plays a major role in synchronizing plant reproduction over different scales [10,18,54,69]. Synchrony in reproduction occurs when conditions favourable for each reproductive stage, from resource priming to fruit ripening, align not only in time, but also in space [30]. Likewise, large-scale reproductive vetoes affecting single stages may thwart reproduction over extended areas [70].

Climate modes are implicated in synchronizing both resource dynamics [32,59] and weather cues of masting in tropical [2,40], temperate [30,53] and boreal regions [7,8]. For example, in Central Europe, the large-scale synchronization of resource dynamics [32] and weather cues [67] driven by NAO results in spatially synchronized resource priming, bud initiation [30] and timing of pollination at different sites [49]. This led to continent-wide masting events such as those in 1990, 1992 and 1995 [18], which extended over an area of more than 1.3 million km² (figure 2). Conversely, in southeastern Europe, the NAO has a more limited impact on local weather [21], which is more driven by multiple interacting climate modes [25] and a complex geomorphology (e.g. water bodies and orography). Greater local environmental variability differentiates weather patterns through space and time, working against large-scale reproductive synchrony [69]. Similarly, in South East Asia, the complexity created by the Malaysian Peninsula and Bornean and Sumatran coastlines and mountain ridges interacts with the response of atmospheric circulation to ENSO and generates regional weather and masting patterns. In southwestern peninsular Malaysia and western Borneo, flowering and fruiting of Dipterocarp often occurs during the transition from La Niña to El Niño (electronic supplementary material, figure S4). However, flowering starts in March in Peninsular Malaysia and in July in southwestern Borneo, which mirrors the northwest to southeast shift of dry-cool air that precedes the onset of El Niño (electronic supplementary material, figure S5a,c,g,i). Moreover, populations of Dipterocarps on eastern sides of Borneo usually flower and fruit after the onset of El Niño [40,42]. Indeed, the ENSO spatial impact is shaped by the geography of the Indonesian archipelago, which drives opposite responses in sea surface temperatures and cascading effects on weather patterns east and west of Borneo (electronic supplementary material, figure S5a–d). Still, the onset of particularly strong El Niño events, such as those in 1959, 1977 or 1983, coincided with mass flowering throughout most of the region [2,28].

Modes of climate variability often drive dipoles in the spatial synchrony of temperatures and precipitation and hence induce dipolar masting patterns [8,54,71,72]. In Central Europe, the geographical divide exerted by NAO on winter, spring and summer temperatures extends along a ridge from 45° N–3° W to 50° N–26° E, termed the ‘NAO node’ (electronic supplementary material, figure S1). This ridge parallels the boundary between positive and negative synchrony in both weather and large-scale masting of European beech [18,30,69]. Likewise, ENSO drives opposite effects on precipitation over different areas of the western United States, which are reflected by an asynchrony in seed production by Pinus edulis in the southern and northern part of its distribution [31]. The onset of El Niño might explain the peculiar spatial asynchrony that La Montagne et al. [8] observed in P. glauca masting when comparing western and eastern populations of Canada and northern America. Indeed, positive surface temperature anomalies appear in the west the summer before El Niño peaks, and negative in the east (electronic supplementary material, figure S5g), thus cueing masting in the west [7]. However, the dipole switches the summer after El Niño (electronic supplementary material, figure S5h), triggering masting in the east. The resulting west-east shift in the ΔT summer cue (electronic supplementary material, figure S6) is thus key to fully understand why this masting dipole was observed in some periods (i.e. during the onset of El Niño) but not in others.

5. Climate modes drive synergies between masting and reproductive fitness of plants

Climate modes do not only set the timing and extent of masting, but also influence the dynamics of ecosystem processes that affect the success of offspring survival. In the few systems where this hypothesis has been suggested [30,47,73] or tested [3,7], climate modes had concordant effects on proximate mechanisms of seed production and on the environment into which seedlings grow. Such concordance may exist in several geographical areas (figure 3). As discussed, in tropical Dipterocarp forests of peninsular Malaysia, flower initiation is associated with cool-dry air from January to March [2]. A recent study found a correlation between the cool-dry weather cue and wet conditions favourable for seedling emergence a few months later, in October to December [60]. Our analyses confirm such weather oscillation (electronic supplementary material, figure S3). Notably, in Malaysia and western Borneo this oscillation from cool-dry to wet conditions is associated with the transition from La Niña to El-Niño (electronic supplementary material, figure S5), which synergistically aligns the flowering cue and precipitation that favours seed germination, seedling emergence and seedling establishment a few months later (electronic supplementary material, figure S5c,i,q). The transition to El Niño coincides with other advantageous mechanisms (figure 3). Changes in large-scale weather during this transition are tracked by long-distance mobile pollinators, which anticipates general flowering [74]. The drought brought by El Niño can cause the formation of canopy gaps [73] and fire disturbance, which peaks six months before the El Niño in eastern Borneo, and during El Niño in western Borneo [75], prior to seed dispersal. Similarly, in boreal forests of northern America, the onset of El Niño leads to regional drought and heat pulses responsible for both fire disturbance and floral bud initiation in P. glauca (figure 4, left), resulting in masting the ensuing year [7]. This alignment benefits spruce recruitment because seeds dispersed shortly after fire germinate more easily owing to partial litter consumption, and seedlings can establish in canopy openings [76]. Positive ENSO is also responsible for extensive fires in tropical Amazon forests, with negative impacts on seed predators and positive influences on Bellucia sp. fruit production [77].

Figure 3.

Chord diagram showing the potential for climate mode phases (blue sector) to drive synergies between proximate (green sector) and ultimate (orange sector) mechanisms of masting in different geographical regions: Central Europe (left diagram, based on 45 sources) and South East Asia (right diagram, based on 35 sources). The sector size in the outer circle indicates the distribution of synergies, while the flows through the centre of the circle illustrate the relative importance of links between individual agents (as measured by the number of observations reporting on the respective link). Arrows point from the influencing climate mode to the mechanisms being influenced.

Figure 4.

Examples of density-independent (left) and density-dependent (right) mechanisms with adaptive benefits linked to models of climate variability. In northwest America (left), the onset of El Niño time spruce reproduction to anticipate favourable conditions for offspring early-life fitness owing to fire disturbance. In South East Asia (right), the transition from La Niña to El Niño synchronizes community-wide reproduction in Dipterocarp forest setting a ‘regional’ escape from mobile animal predators such as Sus barbatus and Pongo pygmaeus.

Similarly, in Central Europe, the NAO has well-established links with both masting in multiple species (electronic supplementary material, table S1) and natural disturbance regimes that may promote seedling establishment (figure 3). For example, positive winter NAO causes major windstorms in the area [34,78], which supports seedling establishment by creating canopy gaps [79,80]. A large-scale climate cue similar to positive summer NAO is linked to both fire disturbance [81] and beech masting the following year [64], with potential benefits for beech establishment in the post-fire environment [82,83]. Also, a persistent positive NAO from winter to spring curbs populations of seed-eating rodents because of lower snow cover and increased predation [22]. Notably, these seasonal NAO patterns also promote dispersal dynamics. Positive winter NAO correlates with irruptions of key seed dispersers, such as Columba palumbus, Fringilla coelebs, Pica pica, Parus major and Carduelis spinus [84,85]. Similar synergies with seed dispersers were described in North America, where a west-east dipole in temperatures, resembling the pattern occurring in the summers before and after the peak of El Nino (electronic supplementary material, figure S5g,h), modulates both broad-scale masting [8] and the irruption of seed-eating birds anticipating the resource pulse [53], with potential benefits for seed dispersal.

In most other regions where the environment is under the influence of major climate modes, there is evidence that ecosystem processes respond to them, and it will be interesting to see how often that results in an alignment between drivers of seed production and other ecosystem processes improving reproductive fitness. For example, negative phases of ENSO, Indian Ocean Dipole (IOD) and Southern Annular Mode (SAM) concurrently modulate rainfall in central Australia, with direct effects on the productivity [59] and seed output [86–89] of mulga (Acacia sp.) and spinifex (Triodia sp.) vegetation. Increased rainfall triggers multiple cascading processes such as ant population dynamics [90] and wildfires [91]. Consequently, it is not surprising that synchronized flowering, ant abundance and fire in inland Australia have been correlated with La Niña [87,90,91]. In turn, landscape-scale fires limit predators in mulga species [92] and cue seed germination of both Acacia and Triodia [86,89]. The interaction between ENSO-IOD-SAM acts thus as a large-scale mechanism that synchronizes resource priming, flowering, seed dispersal by ants and large-scale fire disturbance with multiple adaptive benefits along with the entire plant reproduction sequence.

6. Relevance of climate modes for the evolution of masting

When considering the role of climate modes in the evolution of masting, we should ascertain whether these modes have existed and maintained their properties (i.e. autocorrelation, periodicity and trends) for timeframes relevant to the evolution of long-lived plants. There is evidence that modes of climate variability such as ENSO have been acting for evolutionary-relevant time periods [93,94]. For example, annular modes of variability (such as the SAM) are observed in extremely idealized climate numerical simulations, pointing to the fact that such modes of variability are an intrinsic feature of the Earth's climate system that have been existing for a long time [95].

We highlight how climate modes drive reproduction and recruitment success not only directly, but also indirectly, by forcing lagged but in-phase ecosystem dynamics conducive to favourable conditions for offspring. By synergistically aligning the proximate mechanisms that cause masting and the processes that improve offspring fitness (figure 3), climate modes make favourable environmental conditions more ‘predictable’ by plants [30], a hypothesis known as ‘environmental prediction’ [79,96]. Environmental prediction based on climate modes is possible since they have predictable dynamics, i.e. inherent autocorrelations and periodicities, and because their cascading effects lag in predictable ways.

The environmental prediction hypothesis has been regarded cautiously. Researchers have favoured adaptive hypotheses related to economies of scale (EOSs) implied by large-seed production events [96]. However, linking reproduction to climate modes may also have implications on EOSs. The two major EOS hypotheses are predator satiation and pollination efficiency. In predator satiation, occasional large-seed crops satiate predators resulting in higher per capita survival of seeds and seedlings [11,97]. In pollination efficiency, large flowering events lower the upper threshold for pollen limitation [96,98,99]. The primary distinction between EOS and environmental prediction is that EOSs are density-dependent, i.e. an individual plant benefits from pulsed flowering only when other individuals are also flowering. By contrast, under environmental prediction, if an individual plant produces seed based on cues that predict a favourable environment for seedlings, this would be enough to increase fitness. Notably, climate modes have the potential to let both emerge.

A major implication of the link between climate modes and reproduction in plants is the spatial extent of synchrony in flowering, pollination and seed production, which is usually larger than a population. EOSs of seed production emerge within a given range of extents in spatial synchrony. In some cases, it is sufficient for trees to synchronize pollination within a stand [100], while avoidance of predation by nomadic vertebrates requires a ‘regional escape’ strategy by ‘community-wide’ masting [1]. In the present study, we found evidence that climate modes synchronize reproduction over different spatial extents, covering the full range of scales needed for known EOSs to emerge. While adaptive benefits at smaller extents (i.e. individual, stand and population) have been extensively studied [97,100], benefits emerging from synchrony at larger scales are understudied and mostly discussed in relation to pollen coupling [98,99]. However, synchrony at scales larger than a population, peculiar of the climate mode-masting relationship, might imply adaptive benefits (figure 5).

Figure 5.

Theoretical scheme of the scale of reproduction synchrony covered by climate modes which is relevant for different density-dependent EOS implying adaptive benefits. EOSs from left to right: seed predation escape (grey), pollination efficiency (green), attraction of seed dispersers (blue), disturbance intersection (red) and gene flow (violet). Escaping predation of low mobile predators (e.g. mice) occurs at the lower spatial scale, the benefit increases sharply and remains constant at increasing spatial scales since the EOS is local. Pollination efficiency benefits from a larger scale, although the cross-fertilization rate owing to the outsource pollen decreases at increasing distance of the pollen origin. Disperser attraction requires a spatial threshold of the reproduction synchrony after which the irruption of long-distance disperser such as birds is triggered. The probability of intersecting a disturbed seedbed increases linearly with the extent of the synchrony but decreases after the regional scale since large-scale disturbances (e.g. megafires) affect negatively post-disturbance seed availability. Gene flow benefits exponentially of reproduction synchrony up to the larger scale when geographically separated population synchronize and exchange genes.

For example, although climate-driven disturbances are predictable in the time domain (i.e. environmental prediction) by species that experience the same environmental cue (e.g. drought), they cannot be predicted in the space domain (i.e. where a disturbed patch will occur). Large-scale and community-wide reproduction increases the chances for a larger number of individuals to release seeds in favourable environments created by disturbances [7,73,79,89]. Regional seed predators may also select for large-scale and community-wide flowering/fruiting synchrony. The spatial scale at which synchrony in seed production effectively reduces losses to predators is influenced by predator's mobility [101,102]. If a predator can move easily between plants, stands or regions, selection will favour synchrony among plants at a scale comparable to the predator's mobility [97]. For long-distance mobile predators, strategies that generate large-scale synchrony in seeding would be favoured [103]. Relevant examples are the interaction between Araucaria araucana and the Austral parakeet, which is highly mobile [104] and effectively satiated by seeding synchrony over distances of 10–100 km [105]. A similar interaction occurs between Dipterocarps and large-bodied, highly mobile generalist predators (Pongo pygmaeus and Sus barbatus) in Borneo [1,106] (figure 4). Over 800 000 pigs were observed migrating out of northeast Kalimantan after the 1983 Dipterocarp mast event and severe El Niño-associated drought [107]. Such ability to move requires seeding synchrony over hundreds of kilometres for predator satiation to produce an effective ‘regional escape’ from predation [1] and is suggestive of similar interactions occurring in ecosystems with similarly mobile predators (e.g. Sus scrofa in Eurasia). Other large-scale examples involve specialist birds like the passenger pigeon, Carolina parakeet, and the Javanese finch [101]. Hence, testing the predator satiation hypothesis at the tree or stand scale may only miss processes occurring at larger scales.

The extent of synchrony needed for pollination efficiency is debated [98,108]. In South East Asia, minor and local flowering events that precede mass flowering in Dipterocarps [2] do not lead to fruit development because of low pollen density [109]. Synchronized flowering of many species in Dipterocarp forests causes an increase in long-distance mobile pollinators through immigration [74], which directly affects pollination efficiency [110].

Synchronous flowering over large distances may also have evolutionary implications by contributing to the flow of adaptive gene pools. Long-distance pollen dispersal contributes to a small percentage of pollination but is highly relevant for gene flow between populations over an evolutionary time scale [111,112]. Genetic similarities among separated populations have been attributed to gene flow due to long-distance pollen dispersal [113], which might occur only during synchronous flowering [114]. Intriguingly, the genetic structure of P. glauca in northern America [115], or of Shorea macrophylla in Borneo [116] resembles the west-east dipole in flowering synchrony shaped by ENSO in both regions. Bogdziewicz et al. [69] suggested that the spatial genetic structure of F. sylvatica in Europe [117] resembles the divide in weather and pollination synchrony that is under the control of the NAO. Interestingly, the same geographical divide in genetic structure is shared by other tree species in Central Europe [118], raising the question of whether the large-scale synchrony in pollination led by NAO plays a role in shaping genetic similarities, in addition to legacies from post-glacial migration dynamics [113,117]. Gene flow is a key process for increasing individual fitness since it allows exchange among diverse, separated or marginal populations, which continue to receive (and spread) genes with adaptive benefits. When masting synchronizes over large scales, the higher chance for long-distance gene flow might reinforce synchrony, since it reduces in-breeding depression and increases local adaptation potential and fitness [119].

7. Climate change, climate modes and masting

Anthropogenic climate change is having a strong impact on local weather in several regions on Earth, and masting patterns appear to be shifting in response to these changes [120]. Climate change can also affect climate modes, although their long-term response remains uncertain. ENSO is expected to remain the dominant mode in a warmer world, but model projections do not agree over a systematic increase in ENSO variability, nor on future changes in ENSO teleconnections [121,122]. However, most models show an increase in the amplitude of ENSO rainfall variability (therefore in the associated extreme events) ascribable to the increase in the mean sea surface temperature and moisture [123]. The SAM is projected to become more positive [124] possibly intensifying the Southern Hemisphere teleconnection with precipitation [33]. The positive phase of winter NAO is expected to occur with higher frequency in a warmer world, following a strengthening and zonalization of the mid-latitude westerly winds [125,126]. On the other hand, predicted future changes of the AMV are largely model dependent [127].

These changes are impacting—and will further impact in the near future—the associated ecological processes and their multiple spatio-temporal synergies [33,128], including masting [30]. In cases where weather influencing proximate mechanisms anticipates weather during germination or seedling development, it is likely that these processes will become decoupled as climates change [129]. Where lags in ecosystem processes result in environmental prediction, it is possible that those processes may remain coupled even as Earth's climate systems change considerably.

8. Concluding remarks

Studies testing for the influence in climate modes over variable and synchronous reproduction in seed plants made use of correlation analyses between climate indices and time series of fruit and seed production. However, in most cases, the causal mechanisms behind those correlations were uncertain. In this perspective, we explain the tight relationships between climate, weather patterns, and the physiology and ecology of plant species, and we demonstrate that climate modes shape the spatio-temporal patterns of reproduction of major masting species in most of Earth's biomes. We highlight how large-scale modes of climate variability, such as ENSO or the NAO, influence reproduction and recruitment both directly, through regional weather patterns that align proximate mechanisms of seed production through time and space, and indirectly, by density-dependent EOSs emerging at increasingly larger scales, and by forcing lagged but in-phase ecosystem dynamics conducive to favourable conditions for offspring. The observation that both ecosystem properties and seed production lag considerably behind climate [22,84] might renew interest in the subject of the evolutionary emergence of masting. We argue that climate modes have the potential to ‘bridge’ proximate and ultimate causes of masting selecting for variable and synchronous reproduction. To better understand the impacts of climate change on plant reproduction, a deeper understanding of changes of climate modes and their relationship with global warming will be critical. The future of such deep interaction is uncertain: processes that improve reproductive fitness may remain coupled even under changing climates, but chances are that abrupt global warming will affect Earth's climate modes so rapidly as to perturb ecological and evolutionary links.

Data accessibility

This article has no additional data.

Authors' contributions

D.A. led the overall study, formulated theses and drafted an initial manuscript. All authors participated in group discussions about research ideas and provided considerable and meaningful contributions to writing the manuscript and editing the figures. P.D. supervised and conducted climate data analysis. All authors gave final approval for publication.

Competing interests

We declare we have no competing interests.

Funding

The work was supported by the Natural Environment Research Council (NERC) grant no. NE/S007857/1.