Forest Ecology and Management 296 (2013) 74–80

Contents lists available at SciVerse ScienceDirect

Forest Ecology and Management

journal homepage: www.elsevier .com/ locate/ foreco

Spatial patterns of seedling-adult associations in a temperate forest community

Isabel Martínez a,⇑, Fernando González Taboada b, Thorsten Wiegand a, José Ramón Obeso b,c

a UFZ, Helmholtz Centre for Environmental Research – UFZ, Department of Ecological Modelling, Permoserstr. 15, 04318 Leipzig, Germanyb Área de Ecología., Dpto. Biología de Organismos y Sistemas de la Universidad de Oviedo, C/Valentín Andrés Álvarez s/n, E33071 Oviedo, Spainc Research Unit of Biodiversity (CSIC, UO, PA), University of Oviedo, 33071 Oviedo, Spain

a r t i c l e i n f o a b s t r a c t

Article history:Received 29 September 2012Received in revised form 7 January 2013Accepted 8 February 2013Available online 22 March 2013

Keywords:Biotic interactionsFagus sylvaticaFleshy-fruited treesMark correlation functionSeedling emergenceSpatial point patterns

0378-1127/$ - see front matter � 2013 Elsevier B.V. Ahttp://dx.doi.org/10.1016/j.foreco.2013.02.005

⇑ Corresponding author. Present address: EstaciónCSIC), C/Américo Vespucio s/n, E-41092 Sevilla, Spain+34 954 62 1125.

E-mail address: isamcano@gmail.com (I. Martínez

The spatial patterns of seedling recruitment were examined in a temperate deciduous forest stand of NWSpain. The emergence and survival of individual seedlings were sampled during two recruitment seasonsfor the five dominant tree species (Corylus avellana, Crataegus monogyna, Fagus sylvatica, Ilex aquifoliumand Taxus baccata). Point pattern analyses based on the mark correlation functions and the independentmarking null model were used to explore the relationship between seedling density and the location ofindividual adults of the same and of different species. Overall, we found that negative or null patterns ofassociation dominated at intermediate to large scales in our study site. Surprisingly, there were almost nopositive associations at small scales, except for some pairs of fleshy-fruited species. At the same time, themassive recruitment of F. sylvatica following a mast event was accompanied by positive associations atlarger scales. Spatial changes in seedling abundance were demonstrated to depend not only on the dis-tribution of conspecific adult trees, but to lay a spatial signature of the location of adults from other spe-cies. The temporal persistence of some of these patterns and changes associated to varying productionhighlight the need for a community approach to study tree recruitment.

� 2013 Elsevier B.V. All rights reserved.

1. Introduction

The spatial arrangement of recruits relative to their parent treesis a challenging question in plant ecology, with important conse-quences for both population and community dynamics (Crawley,1997; Murrell et al., 2001). The main framework for this is pro-vided by the so called Janzen–Connell hypothesis (Janzen, 1970;Connell, 1971), which predicts that specialized predators, includ-ing pathogens, will reduce seedling density beneath parent trees,thereby enhancing the recruitment of other species. This hypothe-sis has been criticized both in empirical and theoretical grounds(Hubbell, 1980; Clark and Clark, 1984; Condit et al., 1992; Barotet al., 1999; Hyatt et al., 2003; Freckleton and Lewis, 2006). How-ever, the Janzen–Connell hypothesis has also received broad sup-port, which could explain contrary results by chance eventslinked to predator saturation (Schupp, 1992; Burkey, 1994; Hulmeand Benkman, 2002) or plant–soil feedbacks (Bever, 1994; Packerand Clay, 2000, 2003).

Most of the studies focusing on the Janzen–Connell hypothesishave ignored spatiotemporal effects on plant recruitment dynam-ics. Indeed, only a few studies have examined these effects and,

ll rights reserved.

Biológica de Doñana (EBD-. Tel.: +34 954 466 700; fax:

).

in most cases, they have studied only one species or they have beenconducted under laboratory conditions (but see Augspurger, 1983;Cintra, 1997; Peters, 2003; Miriti, 2007; Queenborough et al.,2007). Here, we examined in a temperate forest not only the spatialrelationship between adults and seedlings within species, but alsothe relationships with heterospecifics (i.e., community level pat-terns of association). The adoption of this framework is needed be-cause of the variable importance of different factors during therecruitment season (e.g., different climatic conditions favoringseedling emergence and survival in a species specific manner),and because of different patterns of temporal overlap among spe-cies (e.g., intra- and inter-specific competition and facilitation,storage effects, preemption of space, etc. Chesson, 2000). Also, acommunity approach is important because of the potential interac-tion among different tree species in terms of the temporal variabil-ity in seed production (e.g., masting years and other cycles of lessamplitude), and the changing response of both dispersers and pre-dators driven by selection of resources with varying availabilities(Martínez and González Taboada, 2009).

Different ecological processes may leave a particular signatureon the spatial arrangement of individuals and therefore, studyingspatial associations may help to reveal the importance of underly-ing mechanisms (McIntire and Fajardo, 2009). A powerful tool tocharacterize the spatial structure of plant communities is pointpattern analysis, which studies the spatial distribution of themapped position of individuals within a given study region

I. Martínez et al. / Forest Ecology and Management 296 (2013) 74–80 75

(Wiegand and Moloney, 2004; Perry et al., 2006; Illian et al., 2008;Law et al., 2009). Although mainly used to study static patterns indetail, point pattern analysis can be also applied to study theirtemporal variation (Zang and Skarpe, 1995; Wiegand et al., 1998;Riginos et al., 2005; Felinks and Wiegand, 2009). Contrasting theobserved patterns to null models derived from specific hypothesescan help to identify the relative importance of different mecha-nisms (Wiegand and Moloney, 2004). However, causal relation-ships need to be carefully stated because different processes mayresult in the same spatial pattern (Levin, 1992; Moloney, 1993; Jel-tsch et al., 1999; Wiegand and Moloney, 2004). Finally, results ofspatial point pattern analysis can suggest hypotheses about theunderlying processes which can then be proven experimentally(Levin, 1992; Silvertown and Wilson, 1994; Crawley, 1997).

In this study, we take advantage of a detailed sampling of spa-tial changes in seedling density during two consecutive recruit-ment seasons to gain understanding about the recruitmentprocess in a temperate forest of the Cantabrian range (NW Spain).Previous work demonstrated a predominance of small scale posi-tive associations between adult individuals of different species inthis study site (Martínez et al., 2010). Here, we used point-patternanalysis to describe the spatial association of seedlings and adultsat the community level. This allowed us to find out if similar posi-tive associations occurred also between adults and seedlings, andto derive hypotheses on the underlying mechanisms and their rel-ative importance. We further examined potential changes in pat-terns of association with time, which are expected due to thepresence of a mast event of the main canopy species Fagus sylvaticaL. In particular, we were interested in answering the followingquestions: (1) Is there any spatial relationship between seedlingsand either conspecific or heterospecific adults? (2) Do these rela-tionships remain constant within and between seasons? (3) Isseedling recruitment enhanced or inhibited under the canopy ofheterospecific trees? (4) Are spatial patterns of recruitment enoughto explain the relationship between adults?

2. Materials and methods

2.1. Study area

The study was conducted during two consecutive recruitmentseasons (May–September in 2004 and 2005) at the Teixeu site(43�1704900N, 5�3002500W, 1000 m a.s.l., Asturias province, NWSpain). The Teixeu site is located at the Northern edge of a temper-ate deciduous beech forest (F. sylvatica), and its name refers to thepresence of yew Taxus baccata L., which is the other main canopyspecies. The forest understory is composed of fleshy-fruited species(hawthorn Crataegus monogyna Jacq., holly Ilex aquifolium L. androwans Sorbus sp.), and hazel Corylus avellana L. The study area islocated on a steep North-east facing slope (average incline is25.6�) which is bounded in the west by the mountain ridge, inthe east by flatter pasture areas, and in the North by a rocky areawhere no trees grow. Soil conditions are nearly homogeneous inthe plot, with a poor development caused mainly by the steepslopes and limestone outcrops. The climate of the region is Atlan-tic, with annual temperature of 9.4 �C (mean, 1970–2009) andrainfall distributed throughout the year (1609 mm per year, mean1970–2009). Snow is abundant from December up to early April.

From the early twenty century until now, the site has been usedfor extensive farming with seasonal occurrence of cattle Box taurusand goats Capra hicus. Additionally, red deer Cervus elaphus andwild boar Sus scrofa are abundant. Consumption and trampling ofseedlings and saplings by these species have been detected at thestudy site, and all but S. scrofa browse also on leaves and buds fromlower branches of adult and juvenile trees. No major logging event

or fire has occurred in the study region, although some firewoodcollection has occurred sporadically. Other perturbations like slopeeffects or strong storms are less frequent at the site, although theyare important in forest renewal because of the creation of gaps.

2.2. Plant species

We focused on the five most abundant tree species at the studysite (Martínez et al., 2010); I. aquifolium (Aquifoliaceae, 73.1 stems/ha); C. avellana (Corylaceae, 66.2); F. sylvatica (Fagaceae, 28.4); C.monogyna (Rosaceae, 24.4); and T. baccata (Taxaceae, 20.9). Hereaf-ter we will refer to these species by their genus name.

Crataegus, Ilex and Corylus are understory species, while Fagusand Taxus are late-successional, shade-tolerant species. Crataegus,Ilex and Taxus are fleshy-fruited trees (arils, not true fruits in thecase of Taxus) whose seeds are dispersed during the autumn by mi-grant bird frugivores, almost exclusively thrushes (Turdus spp.,Turdidae), (Snow and Snow, 1988; Martínez et al., 2008). In con-trast, Fagus and Corylus produce large dry-fruits dispersed at firstby barochory (i.e., by gravity), and after by dyszoochory as a resultof caching rodents (mainly the wood mouse Apodemus sylvaticus,yellow-necked mouse A. flavicollis; both mouse species also preyupon dispersed seeds of the fleshy-fruited species). A comparativestudy on predators’ preferences of seeds of fleshy-fruited trees inthe field showed a selection order Taxus > Ilex > Crataegus (Garcíaet al., 2005). Nevertheless, seed predation on dry-fruited speciesis more intense (Martínez and González Taboada, 2009).

Almost all seeds of Fagus and Corylus germinate the same yearthey are dispersed, but seeds of Ilex, Crataegus and Taxus need upto 4 years of dormancy to complete embryo development (Hu,1975; Beckett and Beckett, 1979; Thomas and Polwart, 2003). Inaddition, as mentioned above, herbivores could exert strong pres-sures on the following plant stages. Previous work demonstrated apredominance of small scale positive associations between adultindividuals of different species in our study site (Martínez et al.,2010). Indeed, multi-species clumps up to 2.5 m in diameter andcomprising a few individuals, frequently from Ilex and Corylus,were common in the study plot. See Martínez et al. (2010) for fur-ther details of the study plot, as well as for a detailed account ofadult intra- and inter-specific spatial patterns of association be-tween tree species. Finally, it is also interesting to note that theseforests currently have a relict character in the Cantabrian range,and that some of the species are locally endangered (Taxus and Ilex)or declining (Fagus, Jump et al., 2006; Peñuelas et al., 2007).

2.3. Data collection

In May 2004, we established 105 fixed 0.5 � 0.5 m quadratswithin the �3 ha study stand (180 m � 170 m, Fig. 1) to evaluateseedling emergence and survival. Quadrats were located in all ofthe microhabitats present to include the different types of vegeta-tion cover. This distribution ensured at the same time a good rep-resentation of different distance intervals among samplingquadrats and adult trees of each species. We distinguished amongtree species and between female and male individuals in dioeciousones. Open gaps within the forest covered by herbaceous vegeta-tion were also included. In the microhabitats corresponding to for-est areas (i.e., different canopies), quadrats were located randomlynear trees with P10 cm dbh, and spaced P5 m apart to reduce oreliminate canopy overlap among conspecific trees. For open micro-habitats, quadrats were randomly selected but spaced >2 m apartto maximize spatial variation.

Quadrats were visited three times per season, and the presenceof each emerged seedling was recorded and mapped to the nearest1 cm. For each seedling, we recorded its survival in each successivesurvey during the sampling season. Seedling age was distinguished

Fig. 1. Maps of the study stand showing the locations of seedling quadrats and adult trees (grey circles) for each species. Symbol size (black) is proportional to the number ofseedlings per quadrat.

76 I. Martínez et al. / Forest Ecology and Management 296 (2013) 74–80

on the basis of cotyledon presence–absence and stem woodiness(see Peterken and Lloyd, 1967; Thomas and Polwart, 2003). Finally,we calculated the number of seedlings of each tree species perquadrat and revision. Some plots were lost or suffered tramplingduring the sampling period. These plots were replaced in nearlythe same locations, and their seedlings were mapped and consec-utively recorded.

During autumn 2005, we recorded the location (x and y coordi-nates) of the bole of every adult tree (i.e., those individuals with aheight greater than 1 m) within the 3 ha study stand. This ensuredthat all adults surrounding each seedling quadrat in a circle of10 m radius were recorded. We gathered the coordinates of treesusing a GPS, and measured distances from trees to their neighborsusing tape. This information was then introduced and corrected ina GIS, and checked during the next visit to the field as the map grown.In the case of Ilex and Corylus, in which vegetative spread by sprout-ing is common, clumped steams growing from the same base wereconsidered a single individual. Because Ilex and Taxus are dioeciousspecies, the sex of each individual of these species was determinedand included in our database. Only the five most abundant tree spe-cies were included in the analyses; the abundance of the other treespecies present (34 stems of other four species: Fraxinus excelsior,Sorbus aucuparia, S. aria and Tilia platyphyllos) was too low to per-form a reliable point pattern analysis.

2.4. Statistical analyses

The spatial association between seedling densities and bothconspecific and heterospecific adult trees was examined usingmark correlation functions. Mark correlation functions extendspatial point pattern analysis to handle quantitative informationattached to spatial points (Stoyan and Penttinen, 2000; Illianet al., 2008; Law et al., 2009). The general approach consists in esti-mating a test function of mark values conditional on the distancebetween pairs of points. However, our data structure is slightlymore complex because we have two types of points (i.e., adult treesand quadrats) which require bivariate mark correlation functions(Raventós et al., 2011). The points represent the locations i of theadult trees and the centers j of the quadrats, and the pattern ofquadrats j is augmented by a quantitative mark mj that indicatesthe density of seedlings in this quadrat. We therefore employed atest function t(mj) that selects an adult tree i as focal point and aquadrat j at distance r away and returns the seedling density mj

in quadrat j.The resulting mark correlation function kt(r) yields the mean

density of seedlings at distance r from adult trees [ct(r)], normal-ized by the mean density ct of seedlings in the quadrats, i.e.,

kt(r) = ct(r)/ct. Technically we implemented the mark correlationfunction using the estimator

k̂tðrÞ ¼Pn

i¼1

Pqj¼1tðmjÞ � kðkxi � xjk � rÞPn

i¼1

Pqj¼1kðkxi � xjk � rÞ|fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl}

non-normalized mark correlation function

Pni¼1

Pqj¼1tðmjÞ

nq|fflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflffl}normalization constant

,ð1Þ

where n and q are the number of adult trees and quadrats, respec-tively, t(mj) is the test function that returns the density of seedlingsin quadrat j, and k(x) is a kernel function which determines whetheror not two points i and j are located at distance r. The kernel func-tion yields a value of one if the distance kxi � xjk between points iand j is inside the interval (r � dr/2, r + dr/2) and zero otherwise.Paralleling the pair correlation function, we used concentric ringsof 0.5 m width around each focal adult to estimate the test function.We isolated in this way different distance classes and avoided inte-grating small scale second-order effects to larger scales (Wiegandand Moloney, 2004). The analyses were constrained to inter-pointdistances between 0 and 30 m, allowing us to consider small scale,neighborhood effects and medium to large scale effects on seedlingrecruitment.

Eq. (1) is a so called r-mark correlation function (Illian et al.,2008) which yields the conditional mean seedlings density ofquadrats that have an adult tree at distance r. Clearly, the condi-tional mean ct(r) may differ from the unconditional mean ct if prox-imity of adult trees and/or the respective environmental conditionsin proximity of adult trees influence the seedlings dynamics in thequadrats. Thus, values of kt(r) greater than one correspond to seed-ling densities above the mean, and were considered indicative ofan enhanced recruitment and a positive association between seed-lings and adult trees. On the other hand, values below unityindicate recruitment below normal densities and a negative associ-ation between seedlings and adult trees.

However, our data structure is slightly more complex becausewe collected additionally the species mark of the adult trees andwe determined seedling densities in the quadrats separately foreach species. Thus, we can calculate ‘‘partial’’ mark correlationfunctions ks1,s2,t(r) which select only adult trees of species s1 andseedlings of species s2. This allowed us to quantify both intra-and inter-specific spatial relationships of seedlings around adults.

Deviations of the test function from unity were tested by simu-lating independent marking. With this null model, seedling densi-ties were randomly redistributed between sampling plots,breaking up any association determined by the precise location ofmarks. By keeping constant point locations, deviations from predic-tions of the null model reflected positive or negative association dueto enhanced or reduced seedling recruitment around conspecific

I. Martínez et al. / Forest Ecology and Management 296 (2013) 74–80 77

and heterospecific adult trees. The position of adult trees remainedfixed to preserve any effect associated to adult structure in our ran-dom simulations (i.e., an antecedent condition; Wiegand and Molo-ney, 2004). However note that observed positive or negativeassociations may be caused by a variety of mechanisms includingseed dispersal and predation, competition, herbivory and abioticperturbations, which are conditioned by the characteristics of eachmicrosite. Following our objectives, we were only interested in thefinal outcome of all these mechanisms, but did not aim to elucidatethe detailed role of particular mechanisms such as habitat heteroge-neity in causing the observed patterns.

To assess the significance of the test function under the nullmodel we generated approximate (two-sided) 95% simulationenvelopes by calculating for each distance r the 5th lowest andhighest values of the summary statistic from 199 Monte Carlosimulations of the null model. All analyses were done using thesoftware Programita for point pattern analysis (Wiegand andMoloney, 2004).

3. Results

During the two consecutive recruitment seasons, we followedthe fate of the 1594 seedlings present in the sampling plots. Ourcensuses revealed great variation in seedling recruitment betweenspecies and seasons (Fig. 1, Fig. S1 in Appendix). Ilex (1099) andCrataegus (385) seedlings dominated in numbers during both sea-sons, while Taxus (27) and, especially, Corylus (5) were very scarce.On the other hand, Fagus presented intermediate abundance levels(78), although it rivaled Crataegus numbers during the second sea-son, following a seed mast event. Increased seedling abundanceduring the second season was also apparent in the case of Cratae-gus and, especially, Ilex.

Fig. 2. Results of the bivariate point pattern analyses between seedling densities (differperiod (2004 and 2005). Positive association (orange) and negative association (blue) rfunction at each spatial scale (i.e., observed values higher and lower than the 95% envemethods for further details, and Figs. S2–S5 in the Appendix for the more detailed, classilegend, the reader is referred to the web version of this article.)

In general, the independent marking null model was able to re-veal distinct patterns of association between seedling densities andboth conspecific and heterospecific adults (e.g., Ilex and Crataegusseedlings showed mostly negative associations to adults and Fagusand Taxus mostly positive). Although there was an important var-iation in seed production and seedling emergence between years,no substantial variation was found in the spatial patterns betweenthe 2 years (Fig. 2; see also Figs. S2–S5 in Appendix for the moredetailed, classic graphical presentation). In the following, we detailthe main exceptions found for different species combinations atcertain scales, highlighting either positive or negative associationsfor some pairs which, in a few cases, also varied in time.

Crataegus seedlings presented in general a negative associationwith both conspecific and heterospecific adults at medium to largescales (10–25 m, Fig. S2). Negative associations with conspecificadults were apparent at distances above 20 m, with a reductionin average seedling density by a factor kt(r) � 0.75 relative to theexpectation. This effect became clearer as the recruitment seasonprogressed, especially during the second season. Additionally,small scale positive association of Crataegus seedlings around theirparents (kt(r) � 1.5, r < 2.5 m) was detected during the first season.Negative association with heterospecific adults was especially in-tense in the case of dry fruited species, Corylus and Fagus(kt(r) = 0.5 for r [10,20] m), and almost undetectable in the caseof Ilex. The negative relationship with Taxus adults presented amarked peak at 15 m, where the average density of Crataegus seed-ling felt below 50%. On the other hand, a trend towards increasedaggregation at small scales (r < 5 m) seemed to emerge as therecruitment season progressed (Fig. S2); despite it was not signif-icant in most sampling dates.

The spatial relationship between the abundance of Fagus seed-lings and either conspecific or heterospecific adult trees presented

ent seedling species in columns) and adults (species in rows) during the samplingepresent significant positive and negative values of the bivariate mark correlationlopes based on simulations of the independent marking null model). See Statisticalc graphical presentation. (For interpretation of the references to color in this figure

78 I. Martínez et al. / Forest Ecology and Management 296 (2013) 74–80

a sharp contrast between the two recruitment seasons analyzed,and also with the patterns observed for seedlings of the other spe-cies. Significant departures from the independent marking nullmodel indicated positive association between Fagus seedlings andadults of all species at intermediate and/or large scales (r > 7 m,Figs. 2 and S3). The association with conspecific adults was moreintense than with heterospecifics, but constrained at intermediatescales in the first season (r [7,15] m, kt(r) > 2). Nevertheless, lowerseedling densities preclude further inferences and cast somedoubts on these results (i.e., low seedling densities yield wide sim-ulation envelopes and erratic fluctuations in the mark correlationfunction that are difficult to interpret). During the second season,Fagus seedlings were more abundant and presented associationwith conspecific adults from small to large scales (5–30 m), witheven a doubling of background abundance at intermediate scales.

In general, the mark correlation function indicated that Ilexseedlings were significantly over dispersed relative to heterospe-cific adults at intermediate and large scales (Figs. 2 and S4). Thispattern was more apparent in the case of the dry-fruited speciesCorylus (r [15, 25] m, kt(r) � 0.7–0.8) and Fagus (r > 5 m, kt(r) upto 0.5), and presented some small changes between seasons(Fig. S4). Negative association was almost undetectable in the caseof conspecific adults, and indeed a slight trend towards positiveassociation emerged at small scales (r < 5 m; Fig. S4). A similarbut more evident pattern was detected between Ilex seedlingsand Taxus adults, with a significant positive association that re-mained constant in time at very small scales (r < 2.5 m,kt(r) > 1.5), but somewhat weaker negative association at interme-diate scales (r > 10 m, kt(r) � 0.75).

Finally, Taxus showed almost no signals of positive or negativespatial associations between seedling abundance and adult trees,with the exception of a weakly significant positive association atintermediate scales with all adult species, excluding Ilex (Figs. 2and S5). Again, low seedling densities call for caution when inter-preting results for Taxus. Analyses of intra-specific seedling-adultassociation considering only female individuals resulted in thesame patterns described above for the dioecious species Ilex andTaxus (Fig. S6).

4. Discussion

Point pattern analyses were used to explore the relationship be-tween seedling density and the location of individual adults of thesame and of different species in a temperate deciduous foreststand, examining the variation in this relationship during tworecruitment seasons. We found rich non-random spatial structuresin the seedling-adult relationships with Crataegus and Ilex seed-lings showing an inclination to negative associations and Fagusand Taxus seedlings an inclination to positive association. Negativeor null patterns of association predominated at intermediate andlarge scales in our study site. Surprisingly, there was almost no po-sitive inter-specific association at small scales (i.e., seedlings underthe canopies of the adult trees), except for some pairs of fleshy-fruited species. At the same time, the massive recruitment of Fagusfollowing a mast event was accompanied by positive associationsat larger scales. Spatial patterns in seedling abundance dependednot only on the distribution of conspecific adult trees, but werealso influenced by the location of adults from other species. Thisindicates that complex community level processes acted at thistemperate forest and that a community perspective is necessaryto understand and interpret these patterns. In the following, wediscuss several mechanisms – especially seed dispersal, betweenseedlings and adults intra- and inter-specific competition and facil-itation, and perturbation – that may help explain our findings.

Short term recruitment failure at the study site was evident insome species. Corylus was a striking example. Its seedlings wereextraordinarily rare, precluding any attempt to analyze associationpatterns. This fact contrasts with the size structure of Corylusadults – L-shaped, the same as Ilex – and their numerical abun-dance. The low seed production observed during the study period,together with high seed predation by rodents (Martínez and Gon-zález Taboada, 2009), could explain the low recruitment observedin this species. The high abundance of Corylus in the study plot canbe explained by mechanisms like storage effects, reliance on rarerecruitment events, and asexual reproduction by sprouts (Higginset al., 2000). Nevertheless, disentangling appropriately which pro-cesses are involved in this remains a subject for further researchand requires long-term data.

The relative lack of Fagus seedlings near old conspecifics withrespect to its overabundance at medium to large scales could bea consequence of the low survival of seedlings beneath old Fagustrees (as previously reported by Peltier et al., 1997; Szwagrzyket al., 2001; Rozas, 2006), and is in agreement with Janzen–Connelldynamics. Indeed, the minimum scale (5–10 m) where positiveassociations between Fagus seedlings and adults of the other spe-cies appeared was very similar to the mean radius of the canopyof individual Fagus adults, indicating a massive recruitment in can-opy gaps. In contrast to the well know advantage to grow under thecanopy of other species, this pattern illustrates the potential of Fa-gus seedlings to colonize nearby patches and to pre-empt space,especially during mast years. Inter-annual heterogeneity has beenrevealed in an analysis that contrasted seed dispersal at the samesite (Martínez and González Taboada, 2009).

On the other hand, enhanced recruitment was observed be-neath the canopy of some pairs of fleshy fruited species. Seedlingsof the understory fleshy-fruited tree species Ilex and Crataeguswere positively associated at small scales with conspecific adultsand with the old Taxus trees of the plot. This is in accordance withtheir seed dispersal mechanisms and the resultant seed depositionpatterns. After feeding fruits, birds used to perch on trees like Ilexand Taxus, which provided both food and protection from preda-tors, (Guitián and Bermejo, 2006; Martínez et al., 2008). As a result,and given the short gut passage time of thrushes, a large propor-tion of the seeds are finally delivered under heterospecific trees,resulting in multispecific seed rains (Obeso and Fernández-Calvo,2002; García et al., 2007). Beyond this effect derived from the seeddispersal template, the immediate neighborhood around adulttrees seems to represent a favorable microenvironment for seed-ling emergence and early survival of the fleshy-fruited tree species.The question which remains is why Taxus seedlings show such aweak positive association with conspecifics. One explanation maybe that significant effects were more difficult to reveal for Taxus,which had only 67 individuals and presents low seedling densities.Another explanation could be competition with adult trees. In fact,although yew is shade tolerant even under dense shade, yew estab-lishment is hindered and saplings often die or show poor growthbeneath the shade of beech or yew (reviewed by Thomas and Pol-wart, 2003). Interestingly, a similar pattern was found in the othercanopy species Fagus, i.e. a nearly neutral pattern of seedlingrecruitment beneath understory species that suggest a competitiveadvantage to monopolize space (Thomas and Polwart, 2003; Pack-ham et al., 2012).

Our forest stand provides a good opportunity to check theimportance of short time seedling recruitment patterns to explainadult distribution. A striking feature of our study site is the pres-ence of small scale multi-species clumps comprising a few adultindividuals, frequently from Ilex and Corylus (Martínez et al.,2010). In contrast, patterns of association between either Corylusor Ilex adults resulted in a negative association at medium to largescales with Crataegus and Ilex seedlings, but in a positive

I. Martínez et al. / Forest Ecology and Management 296 (2013) 74–80 79

association with Fagus seedlings. This mismatch between spatialpatterns of association between adult individuals, and betweenseedlings and adults, is indicative of the importance of processesoperating after seedling emergence. We hypothesize that this mis-match is a consequence of Corylus and Ilex sensitivity to mechani-cal damage (i.e., slope effects, snow and wind pressures, andtrampling), given that it usually consists of several thin stems(Kanno et al., 2001). This would provide a relative advantage toseedlings recruited near adults in contrast to a reduced survivalin forest gaps.

In this study, we found complex non-random spatial associationpatterns between seedling densities and both conspecific and het-erospecific adults. We found intra-specific associations but, moreimportantly, our results indicated that spatial changes in seedlingabundance might lay a spatial signature of the location of adultsfrom other species. At the same time, we identified changes in spa-tial patterning associated to varying seed production. Jointly, theseresults highlight the need for a community approach to study treerecruitment.

Acknowledgements

We thank Bea Blanco, Chus Oliveros, Xurde Sánchez and AliciaValdés for their help in the field. Two anonymous reviewers pro-vided useful suggestions that are greatly appreciated. IM was sup-ported by a PFPU grant (MEC) and by the European researchCouncil (ERC) advanced grant 233066 to TW; FGT acknowledgesfinancial support by a Severo Ochoa FICYT grant (PCTI 2006–2009, Gobierno del Principado de Asturias). This work was alsopartially supported by the project FP7-REGPOT-2010-1 program(Project 264125 EcoGenes).

Appendix A. Supplementary material

Supplementary data associated with this article can be found, inthe online version, at http://dx.doi.org/10.1016/j.foreco.2013.02.005.

References

Augspurger, C.K., 1983. Offspring recruitment around tropical trees: changes incohort distance with time. Oikos 40, 189–196.

Barot, S., Gignoux, J., Menaut, J.C., 1999. Demography of a savanna palm tree:predictions from comprehensive spatial pattern analyses. Ecology 80, 1987–2005.

Beckett, K., Beckett, G., 1979. Planting Native Trees and Shrubs. Jarrold, Norwich,UK.

Bever, J.D., 1994. Feedback between plants and their soil communities in an old fieldcommunity. Ecology 75, 1965–1977.

Burkey, T.V., 1994. Tropical tree species diversity: a test of the Janzen–Connellmodel. Oecologia 97, 533–540.

Chesson, P., 2000. Mechanisms of maintenance of species diversity. Annu. Rev. Ecol.Evol. Syst 31, 343–366.

Cintra, R., 1997. A test of the Janzen–Connell model with two common tree speciesin Amazonian forest. J. Trop. Ecol. 13, 641–658.

Clark, D.A., Clark, D.B., 1984. Spacing dynamics of a tropical rain forest tree:evaluation of the Janzen–Connell model. Am. Nat. 124, 769–788.

Condit, R., Hubbell, S.P., Foster, R.B., 1992. Recruitment of conspecific adults and themaintenance of tree and shrub diversity in a neotropical forest. Am. Nat. 140,261–286.

Connell, J.H., 1971. On the role of natural enemies in preventing competitiveexclusion in some marine animals and in rain forest trees. In: den Boer, P.J.,Gradwell, G.R. (Eds.), Dynamics of Populations. Centre for AgriculturalPublishing and Documentation, Wageningen, pp. 298–312.

Crawley, M.J., 1997. Plant Ecology. Blackwell Science, Oxford.Felinks, B., Wiegand, T., 2009. Exploring spatiotemporal patterns in early stages of

primary succession on former lignite mining sites. J. Veg. Sci. 19, 267–276.Freckleton, R.P., Lewis, O.T., 2006. Pathogens, density dependence and the

coexistence of tropical trees. Proc. R. Soc. B 273, 2909–2916.García, D., Obeso, J.R., Martínez, I., 2005. Rodent seed predation promotes

differential seedling recruitment among bird-dispersed trees in temperatesecondary forests. Oecologia 144, 435–446.

García, D., Martínez, I., Obeso, J.R., 2007. Seed transfer among bird-dispersedtrees and its consequences for post-dispersal seed fate. Basic Appl. Ecol. 8,533–543.

Guitián, J., Bermejo, T., 2006. Dynamics of plant–frugivore interactions: a long termperspective on holly–redwing relationships in Northern Spain. Acta Oecologica29, 97–103.

Higgins, S.I., Pickett, S.T.A., Bond, W.J., 2000. Predicting extinction risks for plants:environmental stochasticity can save declining populations. Trends Ecol. Evol.15, 516–520.

Hu, C.Y., 1975. In vitro culture of rudimentary embryos of eleven Ilex species. J. Am.Soc. Hortic. Sci. 100, 221–225.

Hubbell, S.P., 1980. Seed predation and the coexistence of tree species in tropicalforests. Oikos 35, 214–229.

Hulme, P., Benkman, C.W., 2002. Granivory. In: Herrera, C.M., Pellmyr, O. (Eds.),Plant–Animal Interactions. An evolutionary Approach. Blackwell Publishing,Oxford, UK, pp. 133–154.

Hyatt, L.A., Rosenberg, M.S., Howard, T.G., Bole, G., Fang, W., Anastasia, J., Brown, K.,Grella, R., Hinman, K., Kurdziel, J.P., Gurevitch, J., 2003. The distance dependenceprediction of the Janzen–Connell hypothesis: a meta-analysis. Oikos 103, 590–602.

Illian, J., Penttinen, A., Stoyan, H., Stoyan, D., 2008. Statistical Analysis and Modellingof Spatial Point Patterns. John Wiley and Sons, Chichester.

Janzen, D.H., 1970. Herbivores and the number of tree species in tropical forests.Am. Nat. 104, 501–528.

Jeltsch, F., Moloney, K., Milton, S.J., 1999. Detecting process from snapshot pattern:lessons from tree spacing in the southern Kalahari. Oikos 85, 451–466.

Jump, A.S., Hunt, J.M., Peñuelas, J., 2006. Rapid climate change-related growthdecline at the southern range edge of Fagus sylvatica. Global Change Biol. 12,2163–2174.

Kanno, H., Hara, M., Hirabuki, Y., Takehara, A., Seiwa, K., 2001. Population dynamicsof four understorey shrub species during a 7-year period in a primary beechforest. J. Veg. Sci. 12, 391–400.

Law, R., Illian, J., Burslem, D.F.R.P., Gratzer, J., Gunatilleke, C.V.S., Gunatilleke,I.A.U.N., 2009. Ecological information from spatial patterns of plants: insightsfrom point process theory. J. Ecol. 97, 616–628.

Levin, S.A., 1992. The problem of pattern and scale in ecology. Ecology 73, 1943–1967.

Martínez, I., González Taboada, F., 2009. Seed dispersal patterns in a temperateforest during a mast event: performance of alternative dispersal kernels.Oecologia 159, 389–400.

Martínez, I., García, D., Obeso, J.R., 2008. Differential seed dispersal patternsgenerated by a common assemblage of vertebrate frugivores in three fleshy-fruited trees. Ecoscience 15 (2), 189–199.

Martínez, I., Wiegand, T., González Taboada, F., Obeso, J.R., 2010. Spatial associationsamong tree species in a temperate forest community in North-western Spain.For. Ecol. Manage. 260, 456–465.

McIntire, E.J.B., Fajardo, A., 2009. Beyond description: the active and effective wayto infer processes from spatial patterns. Ecology 90, 46–56.

Miriti, M.N., 2007. Twenty years of changes in spatial association and communitystructure among desert perennials. Ecology 88, 1177–1190.

Moloney, K.A., 1993. Determining process through pattern: reality or fantasy. In:Levin, S.A., Powell, T.M., Steele, J.H. (Eds.), Lecture Notes in Biomathematics, vol.96. Springer-Verlag, US, pp. 61–69.

Murrell, D.J., Purves, D.W., Law, R., 2001. Uniting patterns and process in ecology.Trends Ecol. Evol. 16, 529–530.

Obeso, J.R., Fernández-Calvo, I.C., 2002. Fruit removal, pyrene dispersal, post-dispersal predation and seedling establishment of a bird-dispersed tree. PlantEcol. 165, 223–233.

Packer, A., Clay, K., 2000. Soil pathogens and spatial patterns of seedling mortality ina temperate tree. Nature 404, 278–281.

Packer, A., Clay, K., 2003. Soil pathogens and Prunus serotina seedlings and saplinggrowth near conspecific trees. Ecology 84, 108–119.

Packham, J.R., Thomas, P.A., Atkinson, M.D., Degen, T., 2012. Biological flora of theBritish Isles: Fagus sylvatica. J. Ecol. 100, 1557–1608.

Peltier, A., Touzet, M.-C., Armengaud, C., Ponge, J.-F., 1997. Establishment of Fagussylvatica and Fraxinus excelsior in an old-growth beech forest. J. Veg. Sci. 8, 13–20.

Peñuelas, J., Ogaya, R., Boada, M., Jump, A.S., 2007. Migration, invasion anddecline: changes in recruitment and forest structure in a warming-linkedshift of European beech forest in Catalonia (NE Spain). Ecography 30, 829–837.

Perry, G.L.W., Miller, B.P., Enright, N.J., 2006. A comparison of methods for thestatistical analysis of spatial point patterns in plant ecology. Plant Ecol. 187, 59–82.

Peterken, G.F., Lloyd, P.S., 1967. Ilex aquifolium L. J. Ecol. 55, 841–858.Peters, H.A., 2003. Neighbour-regulated mortality: the influence of positive and

negative density dependence on tree populations in species-rich tropicalforests. Ecol. Lett. 6, 757–765.

Queenborough, S.A., Burslem, D.F.R.P., Garwood, N.C., Valencia, R., 2007.Neighbourhood and community interactions determine the spatial pattern oftropical tree seedling survival. Ecology 88, 2248–2258.

Raventós, J., Mujica, E., Wiegand, T., Bonet, A., 2011. Analyzing the spatial structureof Broughtonia cubensis (Orchidaceae) populations in the dry forests ofGuanahacabibes. Cuba. Biotropica 43, 173–182.

Riginos, C., Milton, S.J., Wiegand, T., 2005. Context-dependent interactions betweenadult shrubs and seedlings in a semi-arid shrubland. J. Veg. Sci. 16, 331–340.

80 I. Martínez et al. / Forest Ecology and Management 296 (2013) 74–80

Rozas, V., 2006. Structural heterogeneity and tree spatial patterns in an old-growthdeciduous lowland forest in Cantabria, Northern Spain. Plant Ecol. 185, 57–72.

Schupp, E.W., 1992. The Janzen–Connell model for tropical tree diversity:population implications and the importance of spatial scale. Am. Nat. 140,526–530.

Silvertown, J., Wilson, J.B., 1994. Community structure in a desert perennialcommunity. Ecology 75, 409–417.

Snow, B., Snow, D., 1988. Birds and Berries. Waterhouses T and AD Poyser, Calton,UK.

Stoyan, D., Penttinen, A., 2000. Recent applications of point process methods inforestry statistics. Statist. Sci. 15, 61–78.

Szwagrzyk, J., Szewczyk, J., Bodziarczyk, J., 2001. Dynamics of seedling banks inbeech forest: results of a 10-year study on germination, growth and survival.For. Ecol. Manage. 141, 237–250.

Thomas, P.A., Polwart, A., 2003. Taxus baccata L. J. Ecol. 91, 489–524.Wiegand, T., Moloney, K.A., 2004. Rings, circles, and null-models for point pattern

analysis in ecology. Oikos 104, 209–229.Wiegand, T., Moloney, K.A., Milton, S.J., 1998. Population dynamics, disturbance,

and pattern evolution: identifying the fundamental scales of organization in amodel ecosystem. Am. Nat. 152, 321–337.

Zang, W., Skarpe, C., 1995. Small-scale species dynamics in semi-arid steppevegetation in Inner Mongolia. J. Veg. Sci. 6, 583–592.