Understanding mast seeding for conservation and land management

Abstract

Masting, the intermittent and synchronous production of large seed crops, can have profound consequences for plant populations and the food webs that are built on their seeds. For centuries, people have recorded mast crops because of their importance in managing wildlife populations. In the past 30 years, we have begun to recognize the importance of masting in conserving and managing many other aspects of the environment: promoting the regeneration of forests following fire or other disturbance, conserving rare plants, conscientiously developing the use of edible seeds as non-timber forest products, coping with the consequences of extinctions on seed dispersal, reducing the impacts of plant invasions with biological control, suppressing zoonotic diseases and preventing depredation of endemic fauna. We summarize current instances and future possibilities of a broad set of applications of masting. By exploring in detail several case studies, we develop new perspectives on how solutions to pressing conservation and land management problems may benefit by better understanding the dynamics of seed production. A lesson common to these examples is that masting can be used to time management, and often, to do this effectively, we need models that explicitly forecast masting and the dynamics of seed-eating animals into the near-term future.

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

1. Introduction

The production of seed crops is integral to the maintenance of plant communities and to the many organisms and societies that rely, directly or indirectly, on seeds as food. Many plant populations produce seed crops that are variable from year to year and synchronous among individuals, a phenomenon known as masting [1,2]. This ‘boom and bust’ cycle of seed production over time establishes a rhythm that affects the timing of many events in biotic communities. Plant and animal communities are closely tied to one another, with animals playing a role in seed dispersal and predation and plants exerting a significant influence on those animals' populations [3]. Furthermore, seed crops often play an integral role in community structure in natural and human-dominated settings.

Pulsed seed crops (masts) are important to people because of the services they provide. Because of this, people have noted and responded to pulsed seed crops since antiquity [4]. Recently, our understanding of how plants produce mast seed crops has increased dramatically [5,6]. Those advances in understanding the causes and consequences of masting can improve applications in managing ecosystems. For example, in systems where mast seeding has been studied extensively, we can anticipate animal population dynamics and disease rates based on seed crops [7,8]. Simulated scenarios that account for masting can inform management actions and their consequences for species interactions, for example, the reintroduction of birds to islands [9], or the introduction of seed-eating biological control agents [10]. We can develop better expectations of how climate change will impact ecosystems [11,12], and, in some cases, we can predict mast seed crops themselves [13,14].

At a fundamental level, these applications rely on understanding what causes a mast seed crop or seed failure, the spatial or temporal scale at which it occurs, and how such pulsed seed crops affect animals. At present, we understand mast seed crops as a consequence of how a plant responds to weather, how a plant allocates resources over time, and how weather events curtail processes like seed development and pollination [1,5,6,15,16]. The combination of these factors has produced models of masting in some species that explain a high proportion of year-to-year variation in seed production [13,16–18]. In the past three decades, researchers have compiled datasets of seed production that span large geographical areas [1,19–21]. From these datasets, we now understand that mast years can occur synchronously over hundreds or even thousands of kilometres in some species [22–25], while in other species synchrony in seed production is far more localized [26,27], or at some combination of scales [14,28]. These observations further solidify the role of weather in masting, and they point to the scale at which biotic interactions with seeds, such as migrations of birds or irruptions of zoonotic diseases, may occur [29–31]. Finally, it is recognized that populations of seed-eating animals track seed crops over time [32]. We now have a better understanding of when pulsed seed production temporarily increases the local population growth rates of those animals, versus when animals move (sometimes over great distances) to capitalize on those pulsed resources [7]. Often, the effects are not limited to seed-eating animals and have cascading effects throughout food webs [33–35]. Likewise, we have more examples of when variable seed production ‘satiates’ seed predators, resulting in suppressed seed predator abundance over the long term because those predators struggle to find resources in years with poor seed crops [36–38]. We have learned that masting is, at some level, a predictable process, defined by weather and internal dynamics of a plant, and we can use this information to better time management actions.

We highlight advances in the study of masting that provide new knowledge to manage ecosystems. Understanding masting can help solve problems beyond tracking wildlife populations, where masting has been considered important for a long time. We describe how the concept of masting informs management applications to cope with a variety of challenges, from the health of plant populations to the management of seed consumers (table 1). These applications can be divided, imperfectly, into those in which we wish to manage a plant population and those in which the fate of seed consumers and the other organisms (including diseases) with which they interact is of interest (table 1). We provide examples to illustrate how the concepts describing masting have been applied to advance management, and we speculate on how using what we have learned from these examples could improve management outcomes in other systems. We conclude by finding commonalities among the examples that we explore: masting often informs the timing of management actions, and predictive forecasts of masting and subsequent trophic cascades could improve the implementation of a wide range of management actions.

Table 1.

Applications of masting.

application	impacted management activities	example systems
improving forest demographic models	decision-making in forest management, anticipating impacts of global change	North American forests [12,39]
assessing population viability of plants	targeting conservation activities to plant populations and growth stages	perennial sedges [40], limber pine [41]
collecting seed for restoration	timing seed collection trips and restoration projects	island oak [42], New Caledonian forests [43]
timing shelterwood forestry	timing timber harvests during 'establishment cuts'	oak [44], fir forests
biological control using seed predators	forecasting efficacy of biocontrol agents	Russian olive [10]
increasing seed crops to enhance dispersal mutualisms	bolstering bird populations, restoring forests	island scrub-oak–scrub jay [9], pedunculate oak–Eurasian jay [45]
timing of animal relocations relative to mast cycles	reintroducing rare or extirpated seed predators	kākāpō [46]
timing of animal culls relative to mast cycles	culling invasive predators, setting hunting regulations	invasive stoat culls in Nothofagus forests [47]; wild boar harvests [48]
forecasting zoonotic disease and pollen allergies	timing public health campaigns	Lyme disease [29], hantavirus [49,50]
optimizing sustainable seed harvests by people	establishing seed harvest economies, timing harvests	pine nuts [18,51], acorns [52], Araucaria seeds [53]

2. Managing plant populations

Globally, forests and other plant communities are facing unprecedented challenges from fire, pathogens, drought and deforestation [54–56]. The retention of major forests into the future depends on their ability to regenerate, and, for many tree species, masting is central to this process. Future management strategies for forests and other plant populations, especially under climate change, will benefit from considering variable seed production. The interpretation of models that anticipate the likelihood of regeneration in particular forests can be influenced by sporadic seed production, and can, in turn, impact management actions like active seeding, shelterwood harvests and live plantings [57,58]. In a similar way, variable fecundity is a major hurdle to population viability analysis of rare plants [59]. Mast seeding can complicate the planning of even straightforward practices in forest management, like collection of seeds for nurseries or restorations [43,60]. And, in some cases, masting may prove important in the efficacy of invasive plant management via biological control [10]. Here, we give several examples of how masting impacts the management of plant populations, emphasizing projects that have incorporated masting in the past, and possibilities for how our increased understanding of masting could improve projects in the future.

(a) . Application 1: managing forests based on tree demography

Models of forest demography are widely used to anticipate the long-term fate of forests under different management scenarios [61,62], disturbances [63] and climates [64]. These models are used to determine in what situations interventions such as planting, thinning and timber harvest might be useful [65]. The success of early stages of a tree's life cycle often has a large impact on the projected trajectory of a forest, even as seed production and seedling recruitment are among the most difficult processes of forest demography to measure accurately [66] and the least likely to be directly accounted for in demographic models [67]. Over the past two decades, there has been considerable effort to explicitly incorporate processes of seed production and seedling survival into forest demographic models [39].

For tree species that mast, interannual variability of seed production can complicate the interpretation of demographic models in several ways. We use demographic models from North American forests as examples of how masting affects our projections of the fate of forests, with an emphasis on recovery from disturbance and persistence under climate change. First, sporadic seed production adds substantial error to estimates of fecundity or seedling establishment that are based on short-term data [68]. Simply put, any short-term measure of seed production or early seedling abundance may grossly misrepresent the long-term average of seed production in a masting tree. Thus, long-term measures of fecundity and their relationship to weather are important when modelling the demography of masting trees [20].

Second, pulsed seed production may result in more even-aged stands of trees than would occur in a non-masting species. This can make it difficult to distinguish ‘normal’ pulses in age structure from other trends, like long-term declines. For masting species in which seed production and recruitment events only occur at long intervals, years with a coinciding large seed source and a favourable environment for seedling survival would be infrequent, even on the scale of a mature tree's lifespan. For example, pulsed recruitment events, caused by variation in seed production and/or seedling survival, remain a persistent hypothesis to explain forest stand structures, such as in Californian oak woodlands, in which the recruitment of seedlings is rare, but mature trees are common [69].

Third, when seed production is spread unevenly among years, the success of seedlings in a year of pulsed seed production may differ from the success of a seedling in an average year. That is, there may be collinearities over time between the production of seeds and the success of seedlings. Most of the hypothesized fitness advantages of masting anticipate higher seed or seedling survival in years of high seed production. Masting has been shown to satiate predators in some systems, resulting in a lower chance of a seed being eaten by a seed predator during a mast year than in an average or non-mast year [36–38]. Likewise, there is evidence that some plants produce mast crops based on environmental cues that also anticipate a beneficial environment for their seedlings, an idea known as ‘environmental prediction’ [24,70]. For example, white spruce produces mast cone crops along an El Niño weather cycle that also drives the fire conditions in which seedlings prosper [24,71]. Thus, ‘average’ survival rates used in demographic models may not adequately represent the success of seeds or seedlings of masting species.

Finally, in cases where mast seed production is related to weather, it is important to account for this relationship when anticipating the impacts of climate change on forest fecundity and regeneration. For example, only by accounting for the effect of weather on annual reproduction of individual trees could an overall signal of climate change on forest fecundity be detected in North America [12]. In this case, the effect of climate change on forest fecundity in western North American was largely caused by older tree stands in this region, an observation that has far-reaching implications on how stand structure in western North American forests can be managed to cope with climate change [12]. Furthermore, the frequency and intensity of disturbances are predicted to increase with changing climate [56]. Because masting affects both seed dispersal and survival, the combined effects of variable seed production and disturbance can have complicated effects on how forests regenerate [72]. Climate change will also likely affect the frequency and variability of reproduction in masting trees. To date, both theoretical models [13,73–75] and empirical observations [11,76–78] suggest a wide range of possibilities as to how this is happening in different species [79].

(b) . Application 2: assessing population viability of rare plants

Demographic models are also commonly used to predict extinction risk in populations of rare plants. Masting can have a profound effect on our estimates of extinction risk for rare plants, as articulated by [59, p. 75], ‘Episodic seedling recruitment is common for many species. Modelling seedling recruitment rates as constants, as implied by the calculation of finite rates of increase, does not capture this variation.’ For example, temporal variability of flowering of the meadow forb, Pulsatilla patens, substantially decreased its modelled population growth rate [80]. This has been accounted for in various ways. Carlsson & Callaghan [40] modelled the frequency of flowering events in a monocarpic sedge with pulsed sexual reproduction, resulting in a more accurate estimate of population growth for this species. In another example, correct estimates of seedling survival of rare plants depend on whether or not it is a mast year. Because of satiation of pre-dispersal predators, variation in seed production in locally endangered Pinus flexilis in Alberta, Canada resulted in substantially higher success of seeds than in non-mast years [41]. As masting patterns can drive abundance fluctuations that resemble transient dynamics in population and coexistence models, the integration of annual seed crop variation into models may facilitate the identification of critical periods during which protection or restoration efforts should occur.

(c) . Application 3: collecting seed for restoration projects

A lack of local seed sources is one of the largest hurdles to many plant restoration projects [60]. Masting can compound this problem by making seeds virtually absent in certain years. This can cause a problem and result in major costs if planting crews and equipment have already been prepared. A cautionary example comes from the restoration of Quercus tomentella cloud forests on Santa Rosa Island in California's Channel Islands National Park. A large restoration programme was planned and funded for a limited time frame when there was minimal acorn production. The paucity of acorns resulted in a decreased genetic diversity of trees reared for outplanting on the island, though restoration of this forest has continued to improve based on natural regeneration associated with remnant forest patches [42]. Similar challenges resulted from seed collections for restoration of Araucaria nemorosa forests on New Caledonia [43]. Solutions to this type of problem could include planning a longer window for seed collections based on the return interval of large seed crops, or developing forecast models that would allow some anticipation of large seed crops in advance (See box 1: A matter of a piñon).

Box 1. A matter of a piñon: managing for nut production in the southwestern USA.

Piñon pine (Pinus edulis) nuts have been described as the most important food source wherever they are found, in part because the large edible seeds have a caloric content comparable to chocolate [81]. But a changing climate may be altering the availability of piñon pine nuts across the landscape. A mutualistic seed-caching species, the pinyon jay (Gymnorhinus cyanocephalus), is estimated to have declined by 3.6% annually between 1968 and 2015, more than any other broadly distributed land bird in North America [82]. Hotter droughts coupled with bark beetle outbreaks have also driven widespread die-off events of piñon pine across the southwestern USA [83] with limited recruitment decades later [84]. Piñon pine trees in drier climates produce smaller mast crops less frequently [16], and a warming climate is likely driving long-term declines in cone production [85]. Managing piñon pine ecosystems with an eye towards forecasting nut production may not only benefit the conservation of piñon pine dominated ecosystems, but also benefit other management objectives.

The original inhabitants of the southwestern USA consumed piñon pine nuts for millennia and used the tree for fuel, construction, pitch (pine resin or tar) and medicine. Today, woodlands are mostly managed for grazing and wildlife habitat. There are also demands for piñon pine seeds as nursery stock, for research purposes, and for use in cooking and cultural practices, but the lack of predictability of mast years has made it difficult to balance supply, demand and labour across years [86]. Mast years can be predicted by cool, wet climates during the initiation and pollination of seed cones [18], and regional fluctuations in mast crops may be tied to El Nino/Southern Oscillation (ENSO) dynamics [14]. Forecasts of mast years could inform the timing and placement of management actions in piñon pine ecosystems, particularly when managing tree regeneration and wildlife populations, and foster the development of local pine nut economies. Photograph by Andreas Wion.

(d) . Application 4: timing shelterwood forestry

In shelterwood systems, which have replaced clear-cutting approaches in many forests, a series of partial harvest steps is implemented to stimulate forest regeneration without planting seedlings [44,87]. In species that mast, this strategy relies heavily on masting cycles, both to avoid unwanted recruitment and to establish a seedling bank before removing adults. In a preparatory cut, unwanted species that may contribute to the seed or seedling bank are removed. Next, an establishment cut is timed following a sufficient reproductive event, allowing newly established seedlings to grow under reduced competition with remaining adults while avoiding full sun exposure [88]. Finally, the overstorey cut removes most of the remaining canopy to maximize seedling growth. A further refinement in shelterwood systems involves targeting specific tree size categories with the goal of creating uneven-aged forests with canopy heterogeneity. Such management systems are popular for multiple reasons, including low cost for regeneration establishment and the ability to stagger the harvest to adjust to demand.

Planning of shelterwood management strategies would benefit from near-term forecasts of seed production for the targeted species. Particularly in systems that are converted to uneven-aged stands, knowledge about the frequency of suitable reproduction events would allow extended planning of harvest events.

(e) . Application 5: biological control of plants using seed predators

In some cases, the process of masting may limit the success of strategies to control invasive plants. For example, Russian olive (Elaeagnus angustifolia) was introduced from Eurasia into the northern Rocky Mountain region of North America for windbreaks, where it has become an aggressive invader of riparian areas, outcompeting native tree species and changing hydrologic processes [89]. In order to reduce the invasiveness of Russian olive without harming Russian olive stands still used as windbreaks, efforts have been made to develop biological control agents that consume a substantial proportion of its seeds [90]. However, because Russian olive has a crop failure roughly every fifth year [10], predator satiation could reduce the efficacy of these agents by limiting the overall number of seeds they consume. The process of predator satiation is dependent on many aspects of the biology of the masting plant and its seed predator [91], so our ability to anticipate the efficacy of this biological control agent will depend on the geographical scale at which seed crop failures occur as well as the movement, fecundity and life-history of the seed predator. Masting species have been introduced into many regions of the world where they are now considered invasive, e.g. Quercus rubra (Europe) [92], Acacia longifolia (South Africa) [93], Pinus spp. (New Zealand) [94], bamboos (multiple regions) [95], so this may prove to be a common problem.

3. Managing the dynamics of seed predators and the organisms with which they interact

Masting is very important to the population dynamics of many animals and other consumers, which directly or indirectly get their energy from seeds. By understanding mast seeding, we can better harness natural seed dispersal mutualisms, conduct animal reintroductions, manage hunting and animal culls, anticipate zoonotic diseases of humans, and understand our own role as seed harvesters. With increasing understanding of the mechanisms underlying masting, managers can time planned interventions in ways that maximize efficacy (see box 2: kākāpō management). In addition, certain aspects of mast-seeding can be simulated to elicit a response from seed predators, for example, by manipulating available food or other resources.

Box 2. Conservation of the endangered kākāpō in New Zealand (a); mast-seeding gynmoperm species rimu (Dacryidum cupressinum, Podocarpaceae) (b).

Nearly extinct a few decades ago, kākāpō (Strigops habroptilus), unique nocturnal and flightless parrots endemic to New Zealand, illustrate the pivotal role of mast-seeding in the conservation management of vertebrate populations. Once common, the birds were devastated by the introduction of carnivores (stoats, cats) and rodents (black rats) in the late nineteenth century. By the 1950s kākāpō had disappeared from most of the New Zealand mainland, and only 18 surviving males were found in the rugged mountains of South Island [82]. Fortunately, a few females were discovered on an offshore island, allowing managers to commence species recovery efforts on predator-free islands. One of the major challenges of increasing population growth was the irregular breeding effort by the birds, which naturally occurs in 3–4 year cycles, apparently tied to mast-seeding in rimu (Dacryidum cupressinum, Podocarpaceae), a gymnosperm species that produces small, fleshy fruits [96]. Only when approximately 8% of rimu branchlets bore fruit in October did the females initiate nesting, despite males courting in all years. Before onset of the breeding season, the birds feed extensively on unripe fruits, which may trigger both hormonal and cognitive responses, and later, ripe fruits constitute the majority of the food provided to nestlings. Supplemental food provisioning, a strategy commonly implemented to overcome temporal resource limitation, only led to partial success as underweight males and females were now able to breed when rimu crops were sufficient [46]. In an ingenious move by managers, however, all 21 breeding-age females were moved to Whenua Hou/Codfish Island in 2002, when a large rimu mast was developing, and 95% initiated nests, ultimately resulting in 26 fledglings, a 39% increase of the global kākāpō population [46]. Today, masting predictions derived from temperature data (2 years before) and branchlet surveys (1 year) play a crucial role in kākāpō management. Photographs by Andrew Digby, New Zealand Department of Conservation.

(a) . Application 6: increasing seed crops to enhance seed dispersal mutualisms

When foraging on variable resources, many animals show a functional response, such that effort and removal rate scale nonlinearly with available food items [97]. In seed-hoarding animals which keep hoarding even if physiologically satiated, the functional response results in a much larger number of seeds stored than necessary to cover energetic needs [98]. This so-called predator dispersal by scatter-hoarders, animals that distribute food in small caches throughout the landscape, is the main long-term dispersal mechanism for many large-seeded forest trees and can be harnessed to restore tree populations [99].

Historically, German foresters who wished to increase oak stem density would transport baskets of acorns from their source to target areas, thereby effectively simulating a masting event. Eurasian jays (Garrulus glandarius) would, in turn, increase their seed hoarding in the area [100]. When calculating the costs of human labour required to match the seed dispersal ecosystem services by jays when afforesting a Swedish park, Hougner et al. [45] found that each pair of jays replaces as much as US$22 000 of planting effort.

Currently, several field trials are ongoing to test this idea in the context of habitat recovery following disturbances. Preliminary results from one such effort on Santa Cruz Island in California's Channel Islands National Park suggest that oak seedling density in fire scars can be increased by an order of magnitude in the vicinity of platforms on which acorns are provided. The spatial scale of passive restoration on Santa Cruz Island following the removal of key acorn and oak seedling predators was likely driven by the seed-hoarding behaviour of island scrub-jays (Aphelocoma californica) [9,101], and the use of feeding platforms allows managers to target specific areas by simulating local bumper crops. More broadly, by re-establishing seed dispersal interactions at wildland–urban interfaces, for example by encouraging residents to stock bird feeders with native local seeds, passive restoration of frequently disturbed habitat could be conducted on large spatial scales.

(b) . Application 7: timing of animal reintroductions and interventions relative to mast cycles

For species that rely on seeds as a major food resource, the timing of re-introduction efforts in relation to the masting cycle can determine successful outcomes. For example, kākāpō (Strigops habroptilus) relocation is often conducted prior to seed production by rimu (Dacrydium cupressinum), a conifer whose seed development initiates reproductive behaviour in these parrots, in order to avoid moves during the breeding season [46].

Variable population dynamics due to mast crops can complicate monitoring programmes that guide decision-making for rare animals. When monitoring and managing mast-dependent species, it is a challenge to disentangle cyclical responses to mast-seeding from overall trends, such as long-term declines due to other factors. A surprisingly large proportion of tropical bird species (up to 7%, at least 1500 spp.) are considered bamboo specialists. These birds feed on flowers or seeds, as well as the insects that feed on bamboo. Bamboo species undergo a particularly extreme masting-like cycle, in which a mass flowering event is followed by die-offs. Bamboo-dependent birds and mammals therefore experience extensive population fluctuations depending on their dietary specialization: generalists vary the least, insect specialists at intermediate levels, and seed specialists, which rely on the ephemeral resource at the end of the cycle, fluctuate most strongly [102]. Population viability analyses for species in seed-dependent trophic cascades can therefore be improved substantially by including masting [103].

(c) . Application 8: timing hunting limits and animal culls relative to mast cycles

For centuries, mast seed crops have been important to the management of hunted animal populations and to the nutrition of domestic livestock [4,104]. Recent advances in our understanding of masting provide additional tools to manage animal populations.

The management of pigs is a prime example of these applications. In Europe, pigs were raised for millennia in forests, where autumn seed crops provided a significant amount of forage [104,105] This practice, known as ‘pannage’ was so important that forests were often described in terms of the number of pigs that they could support. Owing, in parts, to the unpredictability of mast crops in forests and the increase in grain production, the importance of pannage shifted over time from a primary forage for pigs, to a periodic serendipity for swine production, to a practice largely confined to subsistence agriculture [104,105].

As the practice of pannage in domestic agriculture has waned, masting has remained important in managing wild boar. In Europe, wild boar behaviour, spatial distribution and demography are all affected by masting, and this is used to manage populations. For example, Bieber & Ruf [48] found that population growth rate of wild boar was more dependent on juvenile survival in ‘good’ mast years and more dependent on adult survival during ‘poor’ mast years. This led to the recommendation to use hunting as a population reduction measure for wild boar only in years of poor mast production [48]. Outside of Eurasia, introduced wild boar (actually derived from domestic pigs and not their wild progenitors) cause considerable damage to ecosystems and are actively trapped or hunted to reduce that damage [106]. In the Great Smokey Mountain National Park, USA, wild boar move five times further during winters with low mast crops than in years with high mast crops, likely affecting the detectability of these animals during censuses and control measures [107]. Because detectability is the major impediment to most control measures for invasive wild boar [108], understanding their movement and population dynamics with reference to mast crops may be generally important.

In animal population management, the bottom-up effects of mast-seeding on trophic cascades need to be considered, even when targeting species at higher trophic levels. In New Zealand, introduced stoats (Mustela erminea) are a major threat to native avifauna, which evolved without mammal predation pressure. The New Zealand Department of Conservation uses aerially deployed sodium fluoroacetate to poison rodents, which are then consumed by the stoats [47]. To maximize the impact on the temporal fluctuations in the trophic cascades, theoretical models of masting impacts on population dynamics were developed by Köhnke et al. [47]. The results show that poison deployment in June, when rodent populations peak following seed drop, maximizes the effect on the targeted generalist predator. Because yearly application is not feasible owing to public response and cost, this management strategy benefits greatly from mast predictions, based on a differential temperature cue, for seed production of Nothofagus and other trees [13].

(d) . Application 9: forecasting zoonotic diseases and pollen allergies

Zoonotic diseases, passed between human and other animal populations, are a top concern for human health. Masting has been linked to incidences of human cases of zoonotic diseases because of its impact on populations of seed-eating animals that are the hosts of disease vectors and pathogens (figure 1b and table 2). Information on masting may inform forecasts of those diseases to aid public health.

Figure 1.

Pulsed seed crops affect the timing of events in ecosystems. (a) The abundance of primary consumers that eat seeds may increase following large seed crops. Organisms (predators, parasites, pathogens) that eat those seed consumers may become more abundant following peaks of abundance in their prey. Because of this ‘rhythm’ established by pulsed seed crops, certain management actions may be best timed based on seed crops. As examples, (1) thinning of ‘shelterwood’ forest stands and seed collections may be timed in years following a mast event if a greater number of seedlings are available following a seed pulse, (2) intense management of introduced predators (such as stoats in New Zealand) or pathogens (such as Lyme disease in the eastern USA) may be beneficial during their population peaks, (3) more active management of threatened seed predators during low-points of their populations, and (4) activities (such as prescribed fires) that may increase survivorship of seedlings when administered prior to a mast event. Predictive forecasts of mast events and subsequent trophic cascades may give land managers a head start in any of these activities. Because many of these events represent a chain of interactions (b), predictability decreases the further back in time (and thus further away in the chain of interactions) from the event to be predicted (c). A forecast horizon for these events may be calculated based on how certain a forecast must be in order to improve management actions.

Table 2.

Examples of zoonotic diseases impacted by seed crops.

disease	pathogen (type)	region	linkage(s) to seed crops
Lyme disease	Borrelia burgdorferi and B. mayonii (spirochaete bacterium)	eastern USA, Europe	host populations (rodents) fluctuate with acorn crops [29,33,109]. Abundance of tick predators (bobwhite) fluctuate with pine seed crops [110]
ehrlichiosis	Ehrlichia ssp. (Rickettsiales bacteria)	southeastern USA	abundance of tick predators (bobwhite) fluctuate with pine seed crops [110]
spotted fever	Ricketsia ssp (Rickettsiales bacteria)	southeastern USA	abundance of tick predators (bobwhite) fluctuate with pine seed crops [110]
haemorrhagic fever with renal syndrome/hantavirus pulmonary syndrome	Puumala hantavirus/Sin Nombre hantavirus (virus)	Europe/USA	in Europe, vole populations and human incidence of hantavirus fluctuate with oak and beech mast [8,49,50]
African swine fever	Asfarviridae (DNA virus)	Europe	wild boar fluctuate with beech and acorn mast [48]. Additional feeding to stabilize boar populations is a potential cause of swine fever epidemics [111]
tick-borne encephalitis	tick-borne encephalitis virus (virus)	Europe	seeds increase rodent populations and ticks, which relate to human incidence of tick-borne encephalitis [112]

The two most extensively explored examples involve rodent reservoirs of diseases that fluctuate with acorn and beechnut crops [49,113] (table 2). In northeastern North America, both the incidence of Lyme disease in humans and the number of ticks infected with Lyme disease fluctuate based on acorn crops because rodent populations track acorn crops over time (figure 1b; [29,33,113]). The dynamics of Lyme disease are consistent with a consecutive build-up of rodent populations in the year following a mast event, and then the build-up of tick populations and disease incidence within ticks in the following year, 2 years after a mast crop [113]. Experimental evidence, based on large-scale acorn additions, confirms the causality of acorn production in these dynamics [33]. In Europe, the incidence of hantavirus infection in humans increases following irruptions of vole populations due to beech and oak mast events [8,49,50]. Because oak and beech masts themselves are predicted by weather (summer and autumn temperatures in the preceding two years), these weather events have some predictive power of hantavirus outbreaks in Europe [49,50].

In these well-studied systems, it may be useful to develop near-term disease forecasts based on mast seed production [114,115]. To do so, a predictive modelling framework (anticipating new outbreaks, e.g. [8]) will be needed beyond the retrospective framework (explaining previous outbreaks) that is more commonly in use. A predictive model of seed production–disease dynamics should anticipate disease risk into the future, it should have an explicit geographical and temporal scope (forecast horizons, sensu [116]), and it should accurately convey sources of uncertainty in predictions. The geographical scale of disease dynamics appears to be system-specific. In Lyme disease, human incidence of Lyme disease was poorly predicted at sites more distant from the focal region in New York, where acorns were measured, perhaps owing to regional variation in acorn crops [29]. By contrast, a strong connection of hantavirus to mast crops has been found in The Netherlands [8], Belgium [49,50] and southern Germany [117].

Beyond the two best-studied cases, there are numerous other zoonotic diseases that likely cycle with mast crops (table 2). These include many tick-borne diseases, such as tick-borne encephalitis, spotted fever and ehrlichiosis, as well the emergent disease African swine fever ([110,111,118], table 2). The next problematic zoonosis may well be a disease that is currently poorly described. One of the strongest predictors of animal hosts that are likely to spread zoonotic disease is a high intrinsic population growth rate [119], a mammalian trait also associated with capitalizing on resource pulses like mast crops [7]. As we explore the dynamics of poorly understood zoonoses and experience those that have yet to emerge, it is likely that some portion of those diseases will cycle within animal hosts that feed on mast crops.

(e) . Application 10: seed or fruit harvesting by people

Many seeds and fruits are harvested by people from wild trees and shrubs. These and other non-timber products can be an important human use of forests, which, when harvested judiciously, can promote sustainable ecosystem management [51]. For masting species, year-to-year crop variability complicates the use of wild fruit and nut crops, such as pine nuts, acorns, bilberries and wild durian [120].

In most cases, wild-harvested fruit and nut crops are used locally near where they are harvested, but, in a few cases, large-scale markets exist for these products. Pine nuts (harvested from various Pinus and Araucaria species) and acorns (harvested from Quercus and Notholithocarpus) are examples of this. Global exports of pine nuts have increased dramatically, rising from 6800 metric tonnes in 2008 to 23 600 in 2017 [121]. Even with this rise in exports, the pine nut industry is still considered a ‘supply-limited’ economy [122]. With increasing demand, pine nuts have transitioned from a largely locally used resource, in which individual regions consumed pine nuts from locally available Pinus species, to a global market dominated by supply from East Asia. This global trade has raised alarms about the movement of pests and pathogens as well as poor labelling of products coming from different Pinus species [122]. Consequently, there have been renewed calls to develop local pine nut supplies for markets [51] (see box 1: A matter of a piñon); however, the high year-to-year variability in crop sizes complicates these efforts [51].

Acorns have been consumed throughout history by most cultures that coincided with oak forests [123]. However, current acorn industries are absent from most parts of the world. An exception to this is in Korea, where roughly 15 million kilograms of acorns are consumed annually, chiefly as ‘dotorimuk’ acorn tofu [52]. Roughly 95% of acorns consumed in Korea are imported from neighbouring China, and Korean acorn products are exported globally in modest quantities [52]. Because of demand for alternative flours, small-scale acorn processors have arisen in Europe and North America. Sporadic harvests, in addition to high start-up costs, have been listed as barriers to the development of those regional industries [124]. In each of these cases, better forecasts of mast events, a better understanding of the spatial extent of crop success and failure, and the identification of practices to reduce crop variability might greatly help these industries and result in better conservation of those habitats.

When creating tools to better understand and ultimately exploit mast seed crops for human consumption, it will be important to understand the potential consequences of that exploitation. Mast seed crops have been harvested traditionally by numerous cultures throughout the world. In many cases, traditional methods of harvest and management of mast seed crops are sacred, integral parts of a society [125,126], and new approaches to using those crops would need to respect those traditions. Likewise, a better understanding of seed crops could lead to their over-exploitation. For example, the monkey puzzle tree, Araucaria araucana, is a masting tree species that is endangered throughout its range in southern South America, owing to logging and land conversion to grazing [127]. Traditionally, seeds from this tree were gathered by indigenous people and consumed locally [128]. Recent harvesting practices and commercial markets for these seeds have led to greater exploitation of this resource, despite concern over a lack of regeneration of A. araucana forests [53].

Many commercial fruit and nut crops produce highly variable harvests from year to year, and this variation creates problems for growers and agricultural markets. For example, pistachio harvests exhibit some aspects of alternate bearing (a 2-year cycle of reproduction) and a more complex masting behaviour [129,130] that can be understood using the same processes thought to drive masting in wild trees [130]. Similarly, apple, mango, mandarin, lychee, pecan, and other perennial fruit and nut crops tend to have large crops in alternate years [131]. Much can be learned about masting from these agricultural systems because they lack the environmental and genetic variability that is present in natural forests [129], and because the genetic basis of fruit and seed production is better known in these crops [131]. At the same time, substantial effort has been made to reduce alternate bearing or masting in these crops through breeding [132], chemical sprays that manipulate flowering [133] and pruning techniques [134].

4. Conclusion

The examples highlighted above speak to the diversity of applications that can be improved by understanding masting. In summarizing how an understanding of masting is used, it is important to recognize and embrace this diversity. The tangible benefits of understanding masting might best be achieved through stakeholder-driven research in which experts on masting and users of that information collaboratively design projects to address specific goals rather than relying on generalities learned from other systems [135].

That said, we also find some commonalities in the ways that information on masting is applied for different purposes. Information on masting is often used to time management activities rather than to determine what those activities are. In many cases, the lessons learned from conducting management activities involving masting species is that projects should be prepared for surprises. For example, when conducting restoration activities requiring seed collection of masting species, it may be wise to allow collection to occur over a duration based on frequency of large seed crops [42,43]. And when monitoring the demography of masting species or those animals whose populations fluctuate with mast crops, care should be taken to assess patterns over many years [12,102].

In other cases, we may be able to improve the timing of management activities based on near-term forecasts of mast seed crops (figure 1c). Such forecasts are already available and used in New Zealand to time invasive mammal management and conduct reintroductions of endangered birds [13]. Based on the predictability of mast crops in well-studied systems and the length of time that it takes for seeds to develop in some plants, masting forecasts are a predictive tool that could be used in planning. For example, near-term masting forecasts could help public health agencies plan for zoonotic disease outbreaks [50], help wild seed harvesters plan for boom or bust years, and help restoration programmes achieve reliable seed sources. In another remarkable example, near-term forecasts of masting seem to anticipate the severity of pollen allergy seasons in places where masting trees constitute a major source of allergenic pollen [136,137].

The concept of a forecast horizon [116] is central to these predictions. How far in advance can we anticipate seed crops based on weather conditions or disease outbreaks based on seed crops or other conditions? Part of the answer lies in how predictable a mast seed crop is based on weather events and prior seed crops. The other part lies on how closely the dynamics of other organisms are tied to mast seed crops. For example, because masting is connected to disease incidence in humans, establishing a forecast horizon is akin to asking how far back processes can be in a chain of interactions to still predict disease incidence (figure 1b). The direct effects in the chain of events may vary in strength. For example, an increase in the abundance of an animal host will only lead to an increase in incidence of a disease in some cases but not others [138]. Consequently, the amount of variance in human incidence of disease explained by components of masting–host dynamics appears to be greater when considering parts of that process more closely connected, and closer in time, to human incidence of disease. For instance, ‘now-casts’ of Lyme disease based on the volume of internet searches for relevant key words can be highly predictive of a Lyme disease outbreak [109], whereas more distant processes (like weather and seed production) are still predictive of Lyme disease outbreaks, but to a lesser degree (table 2). To effectively implement ecological forecasts based on masting and to understand their utility, we must always ask what accuracy and time horizon are needed for a forecast to benefit decision-making.

Data accessibility

This article has no additional data.

Authors' contributions

I.S.P. wrote the first draft of the manuscript. All authors contributed expertise, compiled examples of masting applications and edited the manuscript.

Competing interests

We declare we have no competing interests.

Funding

We received no funding for this study.