The first stage in any regenerative process is flowering, fruiting and seed production. Although seed limitation has been reported as a main bottleneck in natural regeneration of forest species ( Muller-Landau et al ., 2002 ), there still exists a wide gap on the knowledge on specific topics such as total amount of seeds annually produced, climate factors affecting fruit and seed production, initial and final ages for bearing fruits and masting habit.

P. nigra is assumed to be a masting species, with good seed crops occurring every 6 years ( Alejano et al ., 2008 ), the minimum seed bearing age being between 15 to 40 years. Existing observations of P. nigra reproductive ecology may support both the pollination coupling and the predator satiation hypotheses for masting ( Lucas-Borja et al ., 2011 ). The wind pollination hypothesis states that wind-pollinated plants obtain reproductive benefits by synchronizing large flowering efforts, because it increases the probability of pollination ( Smith et al ., 1990 ). It has been observed that P. nigra produces higher percentages of empty seeds (unpollinated) in low flowering years ( Tíscar, 2007 ). Similarly, the predator satiation hypothesis states that large seed crops are likely to satiate seed predators, which thus would destroy a lower percentage of crop ( Kelly, 1994 ). Moreover, seed production is significantly lower at the altitudinal limit of the species distribution than in the most favourable habitat ( Lucas-Borja et al ., 2012 ), confirming that pines under stressed conditions tend to produce solely male cones or even stop reproducing ( Shmida et al ., 2000 ). Seed production in mast years can be up to 10-fold greater than that of a non-mast year, with average values over 70 seeds/m2 on mast years and favourable locations, and less than 10 seeds/m2 on non-masting years and/or altitudinal limits ( Lucas-Borja et al ., 2011 , 2012 ). Larger values of seed production were identified in high dense stands. Nevertheless, these values are notably smaller than those obtained in other non-limiting areas such as Cazorla (average values of seed dispersal over 600 seeds/m ² ; Tiscar, 2007). Thus, seed availability is one of the main factors limiting the natural regeneration.

P. sylvestris in the Central Range of Spain starts producing a high amount of cone and seeds at 25-30 years old (in isolated trees) or 40 years old (when growing in canopy) holding this capacity up to very high ages ( Ruiz de la Torre & Ceballos, 1979 ; Rojo & Montero, 1996 ). According to the classical trade-off theory by Smith & Fretwell (1974) a small-seeded species like P. sylvestris will produce a sufficient amount of seeds every year. Confirming this, during a four-year study (2005 to 2009) annual seed production was high enough to discard either source or dispersal seed limitations ( Calama et al ., 2015a ). Annual seed density mean values ranged from 150-300 seeds/m ² ( Calama et al ., 2015b ), except in the lower lands ( Fig. 2 ) where the species occurs in mixed stands with Quercus pyrenaica Willd. (average production 36 seeds/m ² ). A certain masting is to be found in the species, with about two good crops every seven years ( Montero et al ., 2008a ), although null crops are absent ( Hilli et al ., 2008 ). In the aforementioned four-year study, 16-fold average differences between years were detected, mainly related with the precipitation occurring in the spring two years before cone maturation ( Calama et al ., 2015b ). It is worthwhile to mention that the observed values for P. sylvestris in Central Range are remarkable if compared with that of the relictic forests of the species in Andalusia (average annual production 2-20 seeds/m ² ; Castro et al ., 1999 ). Seeds from trees in the timberline stand showed a different interannual variation to those at mid-altitudes (1400 – 1600 m), where seed production seems to be maximized. Despite these spatiotemporal differences, natural regeneration for P. sylvestris seems not to be limited by the production of seeds, except on the lower altitudinal limit and/or in years with very low seed production ( Calama et al ., 2015a ).

Cone and seed production has been largely studied for P. pinea , given the edible character of the nut. In this respect, there is a major feature that makes P. pinea different from all other Iberian pines, as cone development occurs over a 3-year period, in contrast to the 2 years of the rest of pines ( Calama et al ., 2012 ). Additionally, P. pinea exhibits a large size seed (over 700 mg ± 200 mg, with 20 mm in length), which prevents wind dispersal from occurring. Cone and seed production is delayed up to an age of 15-20 years, although it can be anticipated in grafted plantations ( Mutke et al ., 2012 ). On the opposite, trees hold their productive capacity up to 160-180 years.

Given its large size, and according to the trade-off theory, we should expect smaller seed crops than in other species. This was confirmed by the average seed production of 9.4 seeds/m ² observed over a 5-year experiment on seed dispersal in P. pinea ( Calama et al ., 2015a ). However, large differences in cone and seed production are detected within the study area. Lower crops are associated with sandy soils with low water holding capacity ( Calama et al ., 2012 ). Low density stands, where trees can expand their crowns freely, are more favourable for cone and seed production than overdense stands ( Calama et al ., 2011 ).

Additionally, the species shows a noticeable, climate-mediated masting habit, which can result in interannual 20-fold magnitude differences in cone production at regional level, largely synchronized within the study area ( Calama et al ., 2012 ). The rate of trees not bearing cones at all in very bad years can reach values close to 80% ( Calama et al ., 2016 ). The occurrence of favourable climatic conditions during key phenological stages as bud formation, bud differentiation, conelet survival through first summer and winter and cone enlargement determine mast events ( Mutke et al ., 2005 ; Calama et al ., 2011 ). In addition, masting in P. pinea is secondarily ruled by a three-year delayed resource depletion effect, resulting in lower than expected cone production three years after an exceptional bumper crop ( Calama et al ., 2016 ).

Previous findings indicate that the lack of seed availability could be considered a main limitation for P. pinea natural regeneration in unfavourable crop years, especially in overdense stands within low producing sites. The results of the above mentioned seed dispersal study came to corroborate this, as regeneration limitation was related to seed shortage in three out of the five studied years (2005, 2008 and 2009) ( Calama et al ., 2015a ).

Contrarily, P. pinaster seed availability has never been considered a limiting factor within the Northern Plateau, due to abundant and frequent cone production and the silviculture approach implemented that always left a sufficient number of trees in the stands during the regeneration phase ( Gordo et al ., 2012a ). While a serotinous species, this trait in P. pinaster is not so marked in the Northern Plateau as in other regions ( Tapias et al ., 2001 ). However, recent studies have given exact figures by counting cones on standing trees and collecting seed from traps ( Ruano et al ., 2015 ), resulting on average values around 7-24 seeds/m ² , though showing large spatial and temporal availability (3.6 fold magnitude between years). Juez et al . (2014) found that just a small fraction of mother trees are responsible for a high proportion of seed production. Ruano et al . (2015) identified a climatic control over masting habit in the species, with the precipitation during the period of secondary growth of the cones positively affecting cone production. Thus, water stress may limit seed production in extreme dry years. Additionally, they identified that following very intensive regeneration fellings, seed availability could be a limiting factor.

P. halepensis presents a combination of several remarkable reproductive attributes related with cone and seed production that helps explaining the great colonization potential of this species and its high resilience to intense fire events ( de las Heras et al ., 2012 ). P. halepensis starts as female the reproductive stage, prior to the appearance of male structures in secondary branches and it shows an extremely short juvenile period, with an early onset of cone and seed production from 3-5 years of age ( Tapias et al ., 2001 ; Espelta et al ., 2008 ). Annual production of cones is often massive and quite regular across years ( Verkaik & Espelta, 2006 ) and part of the crop of mature cones is retained in the canopy for several years as serotinous (closed) cones (45–80% according to Thanos & Daskalakou, 2000 ). According to these authors, considering both serotinous and non-serotinous cones, canopy seed bank can reach values between 115-790 seed/m ² , with 10-20% ( del Río et al ., 2008 ) that can be released during summer without exposure to fire.

In fact, serotiny has probably been one of the most analysed reproductive attributes of P. halepensis . This pine is a weak or partial serotinous species as rupture of serotinous cones is certainly induced by fire, but it may also occur spontaneously, especially when canopies are suddenly exposed to high radiation levels ( Verkaik & Espelta, 2006 ) or after drought episodes ( Espelta et al ., 2011 ). Serotiny degree of P. halepensis decreases with ageing and tree size, a pattern suggested to be caused by the costs of maintenance for the plant of these long-lasting structures competing with younger cones for resources ( e.g . water, in Espelta et al ., 2011 ). In Spain, a large variability in the degree of serotiny has been observed among populations across geographical gradients revealing that serotiny may increase with stand density and from northern to southern populations ( Moya et al ., 2007 , 2008 ; de las Heras et al ., 2012 ). Recently, studies conducted in common garden experiments have demonstrated, after controlling for the effects of tree size, that serotiny has a genetic basis and it is lower in pines from harsh (dry and cold) sites compared with mild sites ( Martín-Sanz et al ., 2016 ), while is more abundant on post-fire stands than on stands established in fire-free conditions ( Ne’eman et al ., 2004 ).

In spite of the well-known importance of the occurrence of fire, the abundant production of cones in P. halepensis ensures a massive availability of seeds even in the absence of this disturbance ( del Río et al ., 2008 ). Therefore seed availability cannot be considered a limiting factor for the species, except in very dry conditions, or when the lapse between two fire events is shorter than the expected time for younger plants to reach sexual maturity ( del Río et al ., 2008 ; Espelta et al ., 2008 ).

Seed dispersal is a key process in natural regeneration, determining the arrival of seed to favourable sites for establishment and the spatial location of the seedlings. Wind-dispersed seeds are expected to arrive to every location within the stand, even travelling over long distances. Contrarily, heavy seeds tend to be clustered beneath the crown of mother trees, resulting in a clumper distribution with gaps in those locations where seeds cannot arrive.

P. nigra , P. sylvestris , P. pinaster and P. halepensis are species mainly dispersed by wind, a fact that facilitates its wide distribution, even in low dense stands. P. nigra disperses seeds from January through May, with a maximum in March, and seed rain tends to be uniformly distributed throughout soil, especially on dense stands (Tiscar, 2007). Literature in P. sylvestris shows that fifty percent of the seeds remain under the crown, while other 40% of the seeds are found at distances as long as 2 to 4 times the tree height ( Burschel & Huss, 1997 ). This means that between 30 and 75% of the seeds are found not more than 18 m from the tree ( Montero et al ., 2008a ). In our specific study area only 20% of the seeds were released under the crown, while in a distance of three crown radii the number of dispersed seeds is similar to that beneath the crown ( Calama et al ., 2015b ), resulting in a close to uniform distribution of the seed rain on the soil surface. For P. pinaster median dispersal distances in the territory varied between 14.1 to 24.5 m, with maximum distances over 54 m ( Juez et al ., 2014 ). Dispersal events are concentrated in only one period, from June to August, with maximum dispersal events associated with violent summer storms. Finally, in P. halepensis spatiotemporal variation of seed rain reveals dispersal curves with ca. 95% seeds dropping less than 20 m away from the mother tree, maximum seed dispersal up to distances of 100-120 m, and occasional long distance dispersal events spreading 0.2 % of the seeds up to distances over 1 km ( Nathan & Ne’eman, 2000 ). Based on the information above, no limitations associated with seed dispersal are expected in these species.

Concerning seed dispersal, P. pinea is mainly referred to as a gravity-dispersed species, due to the morphology of the large and wingless seeds produced. Manso et al . (2012) found a strongly aggregate spatial primary dispersal pattern in a P. pinea stand in the Northern of Spain, only 1% of the seeds being expected to drop beyond 2 crown radii. Similar results are due to Masetti & Mencussini (1991) in their study in Toscana (Italy). This aggregated pattern of seed dispersal will result in a clustered distribution of seedlings and samplings under the influence area of the crown of mature trees, conforming overdense small patches that will require further liberation to enhance development ( Barbeito et al ., 2008 ). From a temporal perspective, seed release is climate-controlled, at least in central Spain locations: cones break open when mean monthly temperature reaches a thermal threshold (usually in May), whereas the subsequent release rate is positively related to precipitation, taking place from summer until early fall ( Manso et al ., 2012 ), showing a similar temporal pattern than its conspecific P. pinaster .

Accordingly, low-distance dispersal by gravity prevent seed arrival into large gaps (radius > 10 m), thus avoiding natural regeneration in these areas ( Fig. 3 ). Additionally, though dispersal of P. pinea seeds has been often assumed to be secondarily animal-mediated, a belief probably linked to the highly nutritional content of the nuts, this fact has never been clearly confirmed. In fact, Manso et al . (2014b) did not find any obvious trace of secondary dispersal by rodents within the regeneration blocks. The authors detected a minority seed removing by corvids, though. Although the fate of these seeds could not be tracked, this can be behind the occurrence of isolated seedlings growing in the closer vicinity of managed forests, far beyond the dispersal distance expected from primary dispersal models.

Pre-dispersal predation by birds ( e.g . cornbills), jays, mammals ( e.g . squirrels), or insect pests attacking cones and seeds in the early stages of maturation can be responsible for severe seed losses. However, post-dispersal seed predation has been reported as a main limiting factor in many pine species, as the seeds reaching the soil can be consumed in the ground in the different seasons of the dissemination period. A large variety of potential consumers (from insects to mammals and birds) have been identified, thus resulting in spatio-temporal heterogeneity in predation due to factors as abundance of seeds or abundance of predators. Seed predators can severely reduce the total amount of seeds available to create a seed bank, but also change the spatial distribution of seeds after the initial seed rain ( Castro et al ., 1999 ).

For P. nigra in Cuenca mountains three main group of predators were observed ( Lucas-Borja et al ., 2010 ): ants, birds ( Fringilla coelebs L., Parus caeruleus L., Parus major L.) and rodents ( Apodemus sylvaticus L., Mus musculus L.). Predators present different seed removal rates depending on the season of the year, with special incidence on winter and lower rates on spring, coinciding with the uprising of temperatures and subsequent lesser activity of rodents ( Fig. 4 ). Seed predation was related with seed rain patterns ( Lucas-Borja et al ., 2016 ) being lower in the high seed rain year (11.2±3.7% of removed seeds) than in the low seed rain years (84.3±6.6% of removed seeds), a finding in agreement with the predation satiation hypothesis ( Janzen, 1976 ). For the species, a good year of recruitment has to be preceded by a good seed production year, in which case post-dispersal seed predation is not as important as in low seed fall years. In high seed production years, birds were the most important predators and rodents were the less important predator group in both typical and relict forest areas for this species. Due to the high seed removal percentage by all predator groups, no conclusion can be obtained in low seed production years ( Lucas-Borja et al ., 2010 ).

Though no specific studies on seed predation on P. sylvestris in Central Spain have been carried out, this topic has been widely studied in other locations of the species. For instance, the great spotted woodpeckers Dendrocopos major L. is one of the main predators of seeds in closed cones of P. sylvestris in central and eastern Europe ( Myczko & Benkman, 2011 ), while the crossbill ( Loxia curvirostra L.) feeds on ripening seeds in southern Spain ( Castro et al ., 1999 ). Predation pressure shows an interannual variability that may be related to temporal variation in cone production ( Matias et al ., 2009 ). Although pre- and post-dispersal predation can reduce the seed pool of P. sylvestris to almost 50% ( Worthy et al ., 2006 ) or even over 90% in relict P. sylvestris forests in southern Spain ( Castro et al ., 1999 ), it is unlikely to be a serious problem for the regeneration of the species in Sierra de Guadarrama, except in the lower limit, where seed production is much more scarce.

In the case of P. pinea , it is important to consider the effect of both pre- and post-dispersal predation. Pre-dispersal seed predation has only been described as a serious threat for seed production very recently. Classical native pests, as Pissodes validirostris Gyll. or Dioryctria mendacella Staud. have been reported to produce damage to a limited extent, damaging about 20% of cones ( Calama et al ., 2017 ). Moreover, in recent years, a North-American invasive seed-feeding bug, Leptoglossus occidentalis Heid., has been postulated as the main cause for the dry cone syndrome, resulting in recent decay in cone production as well as the rate of filled seeds ( Mutke et al ., 2015 ). In this sense, the rate of damaged seeds in the Northern Plateau has increased up to 50% in the recent years, resulting in both economic and ecological losses. Additionally, it is worthwhile to notice that this bug species is a conifer-specific pest, thus the potential damage over other Iberian pines should not be neglected and the monitoring of its activity and its effect on seed availability it is highly recommended.

The size and nutritional value of P. pinea seeds may not go unnoticed by post-dispersal seed predators. Manso et al . (2014b) observed that the wood mouse ( A. sylvaticus ) was the main predator of seeds of P. pinea in the Spanish Northern Plateau, being responsible for more than 80% of seed removal. Secondary agents for predation were identified as azure-winged mapie (Cyanopica cyanus Pall.) and common raven ( Corvus corax L.). A. sylvaticus acts exploiting almost all available seeds during drought-free periods, being able to consume almost 100% of available seeds during winter. The favourable summer period for seed survival is a consequence of both the decreasing rodent populations due to the effect of drought and the higher seed availability in summer, when dispersal takes place. This fact suggests the existence of a dual climate prey-predator control, driven by summer conditions and the climate-mediated masting habit of the species. From a spatial perspective, predation is slightly higher in the close vicinity of trees and potential rodent shelters ( Manso et al ., 2014b ).

A similar process has been identified in the conspecific P. pinaster in the Northern Plateau. Ruano et al . (2014) showed that predation can reduce seed density up to values below 1 seed/m ² , even in locations where seed rain reached values over 40-50 seeds/m ² . Although no specific experiment for detecting potential predators’ species was carried out, the authors indicate that together with A. sylvaticus , C. ciyanus and C. corax , ants should be considered a potential predator for the seed of P. pinaster . The authors found a positive correlation among density of seeds escaping from predation at the end of the year and total seed rain, thus higher predation rates occur when the amount of available seed is lower. Additionally, climate variability influences predator populations, showing larger rates in autumn-winter and almost no predation in summer. As for P. pinea , predation is favoured in the vicinity of stumps and in denser areas (potential sheltered locations), while the presence of grasses and needles on the floor hide seeds and therefore prevents predation ( Ruano et al ., 2014 ).

In P. halepensis , notwithstanding an initial abundance of seeds, their availability may be reduced by intense seed predation, especially when no other seed sources are available ( e.g . after a fire event), when seed losses can reach up to 90%. Rodents have been observed to be the main predators of P. halepensis seeds, while predation by ants or birds is considerably lower (Broncano et al ., 2008). Concerning the spatiotemporal patterns of seed predation, Broncano et al . (2008) showed that predation by rodents is very high in all situations and seasons.

Seed germination, seedling emergence and early survival success will define the spatial pattern of seedling location, and further spatial pattern of adult plants, redefining the spatial pattern associated with seed shadow. In this sense, different requirements can be identified for each process, thus, conditions and traits beneficial for one developmental process could be disadvantageous for another.

Stand density, soil preparation, dispersal season and site conditions had a significant influence on P. nigra seed germination and initial seedling survival, though showing conflicting situations for both processes ( Lucas-Borja et al ., 2011 ). The effect of the overstory can be summarized as positive for seed germination, with rates reaching values up to 40-60% in denser stands and below 30% in open stands ( Lucas-Borja et al ., 2011 ). This effect is more evident on drier and warmer years, where germination and seedling emergence is also favoured by shrub protection, factor with no effect on cooler and wetter years. Germination is also favoured in mid-altitude locations, while largely restricted in the upper limit area. Site preparation had a significant influence on seed germination, which was favoured by scalping, but no effect on seedling survival at long term ( Lucas-Borja et al ., 2011 ).

Higher light availability (low basal area) promotes seedling survival, thus a seed germination-seedling survival conflict can be observed. This conflict is more patent on wetter years, since the probability of summer seedling survival in drier and warmer years is very low, irrespective of crown coverture ( Lucas-Borja et al ., 2011 ). All the results point out that, for P. nigra , annual climate is the major factor controlling seed germination and seedling survival, with a secondary control due to light availability.

Light is a determinant factor for the germination of P. sylvestris in Central Spain. The optimal radiation for germination suggested for P. sylvestris is 35% of full sunlight, thus germination is favoured under the mid to high shadow conditions that are found when the group shelterwood system is applied and with selective cuttings, while is minimized under the large gaps created after a clearcutting ( Calama et al ., 2015b ). Apart from the direct limiting effect of radiation on seed germination, high rates of light are often associated with the presence of a dense herbal layer which prevents seed germination and seedling emergence. Under moderate light levels the presence of this herb layer had a weak effect on seedling emergence, suggesting that seedlings and herbs can occupy the same sites ( Pardos et al ., 2007 ). On the contrary, an increase in the light levels led to a decrease in regeneration performance because of an increased competition for resources ( Módry et al ., 2004 ). Although P. sylvestris can germinate in many different types of seedbed ( Cañellas et al ., 2004 ), the litterfall accumulated increases the depth of organic matter and thus improves microsite conditions for germination ( Pardos et al ., 2008 ), except if layer thickness reach values over 10 cm. This result contrasts with previous findings for the species in Central Europe ( Hille & den Ouden, 2004 ), or even in Northern Spain, where soil scarification is recommended, and reflects the particular ecological conditions for the species in the mountains of Southern Europe.

The results from a 3-year (2009 to 2011) study in Valsaín forest also suggested a climate control that limits germination at least in the timberline ( Calama et al ., 2015b ), partly related to the occurrence of spring frost events and low water availability during summer ( Barbeito et al ., 2009 ). In any case, under the current climate conditions, germination rates in P. sylvestris are high, with a great variability between years and sites. For instance, germination rates recorded in the studied area were over 70%, the larger rates obtained during spring and autumn at 1600 m. On the contrary, lower rates of germination were observed at timberline areas ( Calama et al ., 2015b ).

Despite high rates of seed germination and seedling emergence, the regeneration occurs as a temporary pulse of seedlings, due to seedling mortality that could be close to 100% after a hot and drought summer ( Calama et al ., 2015a ). Water availability during the summer seems to be the key factor driving P. sylvestris regeneration at Sierra de Guadarrama, similarly to what has been recorded at its southernmost distribution limit ( Matias & Jump, 2014 ). With respect to light requirements, although in the Eurosiberian forests P. sylvestris behaves as a light demanding species, results from different studies in central Spain, suggest the preference of the species to moderate light conditions, at least during the early stages of growth ( Pardos et al ., 2008 ; Calama et al ., 2015a ). Seedling survival is also threatened after extemporal cold springs and in highly shaded environments ( Calama et al ., 2015b ). Mortality in P. sylvestris steadily increases during the first two years after emergence, but a notable stabilization in seedling survival is found after three years ( Pardos et al ., 2007 ).

Seed germination in P. pinea is governed by thermal variables in combination with different degrees of soil and air humidity. In the field, Manso et al . (2013a , b) were able to empirically link seed germination of P. pinea seeds to climatic and stand variables in the Northern Plateau of Spain. Optimal conditions occur over few weeks either in autumn (linked with warm conditions) just after seed dispersal, or in the following spring (linked with humid conditions). Large interannual variability in germination rates, conditioned to the occurrence of this optimal conditions, is observed. Over a 4 years study, germination rates varied between 9% and 90%, with two years showing autumn germination events, one year with spring germination events, and one year with almost no germination at all ( Manso et al ., 2013a ). Moreover, extremely poor overstory cover leads to low probabilities of germination. Implications for natural regeneration in P. pinea are relevant as the process becomes seriously limited in the absence of this climatic optimal conditions. Occurrence of cold fall season prevents germination, thus the remaining seeds are highly vulnerable to the action of predators during winter, the season where rodents are more active. Additionally, if a cold winter is followed by a dry spring, germination does not occur ( e.g . 2008-2009 campaign, germination rate < 9%).

Studies carried out in the Northern Plateau show that seedling mortality of P. pinea occurs as a result of a combination of low water potential and negative assimilation rates ( Calama et al ., 2015c ). These circumstances usually take place during the summer season, with older seedlings being able to better thrive due to carbon reserves stored in previous favourable seasons. This picture matches up with the empirical evidence found in the same area, where the vast majority of emerged seedlings perished in the first two years after emergence, except if optimal summer conditions occurs. Interestingly, winter frost does not significantly affect seedling survival ( Pardos et al ., 2014 ). From a spatial perspective, optimisation of the carbon balance and survival is attained when seedlings are placed in mid-shaded locations, which are those locations more favourable for germination. This is in full accordance with the findings of Awada et al . (2003) , which reveal that P. pinea seedlings can actually tolerate some degree of shading in the initial stages of development. The same observations show that individuals emerging in autumn and closer to adult trees are more likely to survive than seedlings from spring cohorts and isolate individuals ( Manso et al ., 2014a ).

Seed germination for P. pinaster in the Northern Plateau seems not to be a limiting factor, with this provenance showing larger germination rates in response to drought than Atlantic provenances ( Núñez et al ., 2013 ), and positive response to thermal shocks ( Herrero et al ., 2004 ). Concerning germination rates on field conditions, Ruano et al . (2009) observed, over an 18 months experiment, a significant effect of remaining basal area on germination rates. Higher germination rates (around 60%) were found at remaining basal area of 6.6 m ² /ha (25% of reduction) on harvest operations. Surprisingly, values of germination rate about 40% were obtained both in non-harvested as well as on clearcut areas. Germination occurs in either fall or spring season, seasonality that could explain the observed little effect of summer rainfall on germination rate.

Seedling establishment and survival processes in P. pinaster are closely related to seed germination, thus more favorable sites for germination are those more favorable for survival. However, initial seedling survival is largely threatened by water stress during the summer ( Fig. 5 ), as well as interactions with other factors. In particular, mid-shaded conditions, the presence of shrubs acting as nurse plants ( Rodríguez-García et al ., 2011 ) and needle soil cover ( del Peso et al ., 2012 ) facilitate P. pinaster germination, emergence and initial survival.

Unless severe predation or topographic constraints limit seed availability, germination of P. halepensis seeds and early survival is often successful and occurs at high rates (80%) under contrasting environmental conditions: i.e . from closed (10% full sunlight) to open (80% full sunlight) canopy cover. In this sense, while shading favours earlier germination ( Broncano et al ., 1998 ), previous studies in Southern France have reported complete lack of regeneration in very dense stands ( Prévosto et al ., 2012 ). Germination is favoured in exposed mineral seedbeds while thick layers of organic matter may limit this process ( Broncano et al ., 2008 ). As a light-demanding and drought-tolerant species, once established, seedlings are able to tolerate high levels of radiation exposure ( Fig. 6 ) and moderate water stress ( Eugenio et al ., 2006 ). Indeed, these are environmental conditions particularly occurring after intense stand-replacing wildfires and therefore, in this situation, regeneration is often successful ( de las Heras et al ., 2012 ), with densities of young seedlings varying from 0.1 up to 10 seedlings/m ² ( Rodrigo et al ., 2004 ). The fact that P. halepensis shows similar environmental conditions for both germination and initial seedling establishment, as well as a wide plastic response, confirm the lack of a seed-seedling conflict in this species ( Broncano et al ., 1998 ). Thus, the lack of natural recruitment in some cases could be associated with the absence of effective dispersal. Interestingly, in contrast with the myriad of studies in Spain that have analysed the regeneration of P. halepensis forests after stand-replacing wildfires and that claimed for the high resilience of this species to these disturbance, the regeneration in the absence of fire has been seldom explored ( Beltrán et al ., 2011 ), and this is a main constrain to predict the dynamics of undisturbed forests (but see Zavala et al ., 2000 , 2011 ) and their response to global challenges.

Findings from the previous section allow us to identify the main bottlenecks for natural regeneration in Iberian pines ( Fig. 7 ), and give insight into the potential threats for the future.

Seed production can be a key limiting factor in non-mast years for masting species such as P. nigra and P. pinea , especially in the ecological limits of their distribution. For the rest of species, even though some interannual variability in cone and seed production is observed, seed availability seems sufficient except under very limiting environments ( e.g . at lower altitudinal limits for P. sylvestris , or in very dry years). Little is known about the process of non-fire-associated ripening of serotine and non-serotine cones in P. halepensis , though the total amount of seeds produced and dispersed even in non-fire events seems sufficient. Given the climate control over masting on pine species, limitation due to seed availability could be a limiting factor for regeneration of pines under future climate scenarios.

Concerning seed dispersal, the only species severely affected by ineffective dispersal seems to be P. pinea , which is the unique species showing a main gravity dispersal process. Low current densities of mature P. pinea stands-associated with a common silvicultural practice applied in the territory during the 20th century to promote cone production-resulted in large gaps that cannot be occupied by the seeds. The postulated animal mediated seed dispersal in the species seems not to be sufficient to fill these gaps, thus resulting in a patched and clumped distribution of seeds under the crowns of adult trees.

Seed predation is a key limiting factor for the main part of the species, especially on low seed years. In this sense, masting can be considered a regeneration strategy to deal with this, by satiating predator populations on mast years and starving them on non-mast years, according to the postulated predator-satiation theory ( Salisbury, 1942 ; Janzen, 1976 ). At within-year temporal levels, we detect a delay among the season of seed fall and the maximum activity of predators (mainly rodents), which would allow seeds to escape predation if favourable conditions for germination are attained. On the contrary, complete losses of seed crop can be found on unfavourable years for many of the studied species.

Seed germination and seedling emergence is mediated by both environmental control and soil conditions where seed arrives. Seed germination is mainly regulated by temperature, water availability and light. Freezing temperatures can limit autumn germination in P. pinea , as well as germination of P. sylvestris and P. nigra on their altitudinal limits. Dry years can result in lack of spring germination for all the species. With respect to light, germination is generally favoured under mid-shaded environments, except in P. halepensis , whose germinative ability remains high under a wide range of cover conditions. Finally, for the small seeded species, as P. nigra , P. sylvestris and P. halepensis , the presence of a dense herbal layer limiting the access of seeds to mineral soil can prevent seed germination and seedling emergence. In this sense, light exerts a multiple control, since in very open stands the presence of herbs and competitors is enhanced, while soil water content is minimized. On the opposite, in very dense stands the presence of a very thick litter layer together with low light availability could be detrimental to emergence and survival ( Sayer, 2006 ).

Shortage in water during the first summer seasons together with excessive light irradiance seems to be the key factors impairing seedling survival in all the studied pine species, except, once more, in P. halepensis , which shows large rates of survival even at high exposures. The rest of species require initial shading for seedling survival, and initial survival is more likely in the close vicinity of parent trees, or even under the presence of shrubs acting as nurse plants. Nevertheless, as seedlings grow, different degrees of tolerance are identified among species. In this sense, and according to their low shade tolerance, P. halepensis seedlings require earlier liberation ( Scarascia-Mugnozza & Schirone, 1984 ), followed by P. pinea and P. pinaster ( Ruano et al ., 2009 ; Manso et al ., 2014a ). On the opposite side, P. sylvestris and P. nigra are the species whose seedlings can survive longer periods under the shade of parent trees ( Calama et al ., 2015b ; Lucas-Borja et al ., 2016 ).

The main bottlenecks identified for all the species in the studied areas are climate-mediated. This means that it is very likely to have unfavorable conditions for one or several of these processes in a given year, thus favorable conditions for the regeneration of the species occur within a lapse of years. Under changing climate scenarios we should expect, in general, more unfavorable conditions for seed production, seed germination and seedling initial survival. On the other hand, processes as autumn-winter germination can be favored by increasing temperatures, which could also affect the population dynamics of predators as rodents. In any case, natural regeneration of the species in the studied areas would surely continue to exhibit an intermittent temporal pattern, which should be aggravated under drier scenarios.

Based on the previous findings, new proposal for managing forests in order to attain the natural regeneration have been proposed for the species in the studied areas.

In the case of P. nigra it has been proposed (Lucas–Borja et al ., 2016) an extension of both the rotation and the regeneration periods up to 150 and 30 years, respectively, in order to ensure the occurrence of favorable conditions permitting successful natural regeneration. In addition, less intense and more gradual and continuous regeneration fellings should be applied. These fellings should be programmed following a mast year, and applied in combination with soil treatments ( i.e . scalping) oriented to prevent soil compaction and to favour soil humus. Shrubs cover might have an important role regarding natural regeneration. In drier years, shrub cover should be promoted since seedling emergence has been enhanced under moderate light (about 25 m ² /ha) and shrubs protection. However, in wetter years, forest managers should promote P. nigra regeneration by clearing shrubs after seedling emergence to increase light availability and to avoid shrub competition. Overall, the natural regeneration of P. nigra has to go through conflicting situations between conditions suitable for seedling emergence (medium basal area interval, shrub cover) and conditions suitable for seedling survival (outside shrub cover without basal area influence), these effects being modulated by the climate of a given year, seed predation and site quality. These issues make many of these recommendations difficult to be implemented in the field.

Similar, but even more flexible schedules are proposed for P. sylvestris in the Central Range ( Cabrera & Donés, 2010 ). Uniform shelterwood system is being now replaced by a more gradual group regeneration system, where the shelterwood fellings are gradually applied on small patches and trees are harvested through several fellings. In this way, natural regeneration establishes progressively under the older trees canopy during a 40 years regeneration period and a 140 years rotation length, with some seed trees still remaining after the last cutting. For example, in Valsaín forest, the aim is to gradually substitute the old stand by a thicket stand with 250 trees per ha. This initial density will decrease by natural mortality and harvesting operations. Under these proposed practices, P. sylvestris stands would naturally turn to semiregular or multi-aged structures, making the stand more vulnerable to fire. It is important to note the adaption of the silviculture to the peculiarities of the different stands, with important restrictions to the harvesting operations in those stands that hold protected species as the European black vulture ( Aegypius monachus L.).

The main innovative proposal in managing natural regeneration in P. pinea ( Gordo et al ., 2012b ) is that in order to ensure seed arrival into gaps, thinning schedules should target to densities about 125-150 stems/ha at the beginning of regeneration fellings. This means that less intensive and more gradual thinnings should be applied during the whole cycle. Early thinning operations are also recommended in order to favor initial lateral expansion of the crowns, facilitating seed arrival into the gaps. A second concern is related to the need of initial shading conditions to favor seed germination and seedling survival, especially under more severe climate scenarios. Intense uniform shelterwood system should be replaced by more gradual fellings. An initial single preparatory-seeding felling, reducing stand density up to 80-100 stems/ha, is recommended. With this cutting we aim to remove dominated, co-dominant and dominant trees in a bad state, with inadequate stem form, malformed crown and low cone production. The remaining trees will act as seed trees, also providing initial shelter for the recently emerged seedlings. Simultaneously, it may be necessary to eliminate the understory vegetation and nonviable advanced regeneration. Once a sufficient number of viable seedlings had been already established beneath the crown cover of the seed trees (after 5-10 years), a secondary felling should be applied, aiming to release the most promising regeneration patches. Subsequent removal cuttings should aim to gradually uncover the new cohort of regenerated trees. These removal cuttings should elapse a maximum of 5 years, to prevent the negative effect due to competition or damage from harvested mature trees on the young saplings. At the end of the period, 5-10 extra mature trees per ha are to be retained.

The regulation of light through operational forestry practices influences significantly the germination, survival and growth of the P. pinaster . Shelterwood method applied by light interventions (reduction of basal area between 25 and 50 % at maximum) favors the seed germination generating seedlings protected by residual trees. In this way, seedlings will grow to generate stands compatible with the provision of ecosystem services (resin and wood), while obtaining other environmental services. A reduction of 50% of the basal area should be applied when economic conditions restrict operations, thus larger amount of wood must be removed to make feasible the intervention. As for other Mediterranean pines, the role of tree canopy protection will be more important as the climatic conditions become more stressful (especially extreme dry and hot summers). When an adequate amount of viable seedlings is obtained (after regeneration felling or taking advantaging of advance regeneration) a final felling should be applied to avoid damages to new crop trees. In some special cases, a reduced number of trees from the previous cohort can remain in the stand to enhance biodiversity values.

The management proposals for natural regeneration of P. halepensis on absence of fire should aim to mimic small perturbations ( Prévosto et al ., 2012 ). Beltrán et al . (2011) propose even-aged structures following shelterwood regeneration system with 80-150 years rotation. The authors suggest the application of shelterwood method in two phases: seeding cutting and final cutting. Moreover, they propose optionally the regeneration by clearcutting with reserves, retaining uniformly spaced ca. 50 trees/ha after the clearcut. In addition, they point out the importance of some treatments to help regeneration: clearing, with the partial or total removal of the understory vegetation ( Serrada, 2003 ) and slash management to facilitate the germination of the seeds and establishment of the saplings. Soil preparation techniques as scarification and even prescribed burning are also recommended to favor the contact of the seeds with the mineral soil and to promote regeneration of P. halepensis stands after cutting ( Prévosto & Ripert, 2008 ).

Many different bottlenecks have been identified to limit natural regeneration of the Iberian pine species, except for P. halepensis . The main part of the limiting subprocesses are climate-mediated, thus joint occurrence of favourable conditions for natural regeneration occurs on intermittent events. Under changing climate scenarios we expect an increase of the time lapse between two successful regeneration events. Consequently, new silvicultural schemes aimed to promote natural regeneration should rely on taking profit of all the successful regeneration events, favouring advanced regeneration, ensuring mid-shaded optimal conditions for seed germination and seedling survival, preventing the creation of big gaps, and programing fellings in those mast years where abundance of seeds allow predator satiation. Intensive clearcutting and uniform shelterwood systems are mainly substituted by more flexible schedules, including group shelterwood systems. This would result in stand age structures closer to that of semiregular forest, which mimics the natural dynamics of regeneration of the species, but could be more vulnerable to forest fires, a main risk in Mediterranean forests.

Author wish to thank all the persons and institutions involved in the installation, maintenance and monitoring of the different trials, mainly: G. Madrigal, M. Conde, E. Garriga, F. J. Gordo, Ayto. El Portillo and Servicio Territorial de Medio Ambiente de Valladolid ( P. pinea ); A. Bachiller, I. Cañellas, J. Donés and Centro de Montes de Valsaín ( P. sylvestris ); J.A. García Abarca and J. L. Serrano Cuenca ( P. nigra ); Oficina Técnica de Prevención Municipal de Incendios Forestales de la Diputación de Barcelona-OTPMIF ( P. halepensis ); I. Ruano, E. Rodríguez-García, C. Ordóñez, and Servicio Territorial de Medio Ambiente de Segovia ( P. pinaster ).