Effect of artificial shelters of dead branches on acorn survival and dispersal in shrub-lacking Mediterranean dehesas

Published Date
November 2015, Volume 46, Issue 5, pp 965–978

Title 

Effect of artificial shelters of dead branches on acorn survival and dispersal in shrub-lacking Mediterranean dehesas

• Author 

• María José Leiva, 
• María Vera

Abstract

Acorn predation is a major bottleneck for oak self-regeneration in grazed Mediterranean oak savannahs (dehesas), which lack of shrub cover. Establishing artificial shelters of dead branches (SODBs) could decrease acorn predation by protecting the seeds against domestic ungulates. However, rodents are also common in dehesas and their dual role as acorn predators and dispersers could be affected by the SODBs. Two experiments were conducted to assess this role of the SODBs on acorn survival and dispersal. They were carried out in six dehesa stands with different levels of rodent activity (i.e., high and low). The results revealed a positive effect of the SODBs on the acorns safety in those stands with low level of rodent activity. Here, the SODBs acted as effective physical barriers against livestock, and the acorn survival was significantly higher (3- to 7-fold) beneath the SODBs than in the Open treatment. Conversely, in those stands with high level of rodent activity the SODBs significantly increased acorn predation by rodents (acorn survival was 1.6- to 9-fold lower beneath the SODBs than in the Open treatment). In addition, the SODBs acted as preferred sink for dispersed acorns, but these acorns were completely predated by mice. These results suggest that establishing SODBs could be a suitable and inexpensive method to improve oak regeneration in dehesas and in other disturbed ecosystems, especially whether this method is combined with other simple techniques to protect acorns against predation by rodents.

Introduction

The difficulties of many oak species to regenerate from seedlings in Mediterranean and other seasonally dry environments have been broadly documented (e.g., Di Castri et al. 1981; Herrera et al. 1994; Pausas et al. 2004; Badano et al. 2009; Rodríguez-Calcerrada et al. 2010; Leiva et al. 2013). Among the most important causes considered for this low regeneration are high acorn predation, seedling browsing and trampling by ungulates, and high early seedling mortality due to severe summer drought (Espelta et al. 1995; Castro et al. 2002; Pulido and Díaz 2005; Smit et al. 2009). All these factors can be summarized as a lack of “safe sites” for acorns and seedlings (sensu Harper 1977). In the Mediterranean area of the Iberian Peninsula oaks typically form a genuine, human-derived, savannah-like ecosystem type locally called “dehesa”. This ecosystem type also suffers due to the lack of oak self-regeneration.

The dehesas are composed of an overstorey of scattered oaks (mainly holm-oak, Quercus ilex subsp ballota L., and cork oak, Q. suber L.) and an annual grassland layer. Shrubs are generally absent in the understorey, or are at very low density. These ecosystems are the result of a long history of human transformation of the Mediterranean forest through clearing, browsing and ploughing (Grove and Rackham 2001; Costa et al. 2006; Martín-Vicente and Fenández-Ales 2006). However the dehesas are of great value because of the high level of plant and animal biodiversity they support, and the important environmental services that they provide (Doctor-Cabrera 2003; Costa et al. 2006; Marañón et al. 2012), and are to be preserved under the EU Habitat Directive and the UNESCO MaB programme. The dehesas are mainly devoted to extensive livestock rearing. Here the guild of acorn predators/dispersers is very important for the maintenance of these ecosystems. Domestic (i.e., cattle, sheep, goats, Iberian pigs) together with wild ungulates (wild boar, red deer) consume the acorns, and also browse and trample the seedlings. Wild middle-sized mammals (European rabbits, Oryctolagus cuniculus L, Eurasian badgers, Meles meles L, Mediterranean hares, Lepus granatensis R) are other common acorn predators in dehesas. In addition, small rodents, typically Algerian mouse (Mus spretus) and wood mouse (Apodemus sylvaticus) inhabit the dehesas (Leiva and Fernández-Alés 2003; Muñoz et al. 2009). These small rodents are likewise important acorn consumers (Siscart et al. 1999; Leiva and Fernández-Alés 2003), but they also cache and disperse acorns (e.g., Xiao et al. 2005; Gómez et al. 2008; Schupp 1990; Hulme 2002). However, the net role of small rodents as acorn dispersers versus predators is controversial (Vander Wall 1990). Many studies consider that seed harvesting by small rodents is mostly equivalent to predation (e.g., Schupp 1990; Hulme 2002; Mendoza and Dirzo 2007), while other studies argue that the benefit gained from the relatively few surviving cached seeds is greater than the cost of having most seeds consumed (e.g.,Vander Wall 2001; Vander Wall and Longland 2004; Vander Wall et al. 2005). Thus, the role of these small rodents in the dehesas needs to be better assessed.

The ultimate cause of low oak regeneration from seedling in the dehesas seems to be the simplified structure of its understory. The open grassland does not act as a “safe site” (sensuHarper 1977) for acorns and seedlings as the acorns are highly exposed to ungulates predation (Pulido and Díaz 2005; Bonal and Muñoz 2007), the seedlings are not protected against browsing and trampling, and the open grassland does not provide the shaded conditions required for seedling survival. In the Mediterranean and in other seasonally dry environments, shrub cover enhances seedling survival during summer through decreasing the high level of sun irradiation, temperature and drought (e.g., Pulido and Díaz 2005; Gómez-Aparicio et al. 2006; Smit et al. 2009; Leiva et al. 2013). In addition, shrubby cover is considered essential shelter for small rodents, the potential acorn dispersers, to avoid predators (Jensen et al. 2003; Lagos et al. 1995; Torre and Díaz 2004 but see Muñoz et al. 2009), and represents an attractive micro-site for the acorns being dispersed by rodents (Den Ouden et al. 2005). Thus, lack of shrub cover in grazed dehesas likely prevents dispersal activity of small rodents.

For these reasons, greater structural heterogeneity in the understorey of grazed Mediterranean dehesas and other shrub-lacking ecosystems likely may help improve crucial phases of oak regeneration. Artificial shelters of dead branches could mimic the role of shrub cover, by decreasing acorn predation by the broad guild of acorn consumers and acting as a sink for dispersed acorns. It has been shown that these shelters also provide shaded conditions essential for seedlings to survive summer drought at early stages. For example residual logs, branches and other coarse woody debris improved pine seedling growth and water status in burned Mediterranean areas (Marañón-Jiménez et al. 2013). The establishment of these artificial shelters could be a fast, relatively inexpensive way to reach some level of structural heterogeneity in the dehesas’ understorey. Their use could be an alternative or complementary method to the slow, spontaneous process of shrub recovery mediated by extensive grazing abandonment. In the Mediterranean savannah-like ecosystems of central Spain, it takes more than 16 years without grazing to reach a sufficient level of shrub encroachment to allow seedlings recruitment (Ramirez and Diaz 2008).

The aims of this study are: (1) to assess the role of artificial shelters of dead oak branches on acorn predation, in comparison with unprotected situations and with standard fences against livestock, (2) to detect potential interactions between the role of the artificial shelters and the level of rodent activity at stand level; and (3) to analyze whether the shelters act as a preferred sink for dispersed acorns and whether these acorns have greater probability to escape predation than the acorns dispersed to the open grasslands.

Methods

Study area

The study was conducted in representative “dehesa” stands located in two Protected Areas (Natural Parks) in south-western Spain: “Doñana” and “Sierra Norte de Sevilla” (DÑ and SN hereafter). Both areas are 60 km apart. The former is in an inland plain close to the Atlantic coastline, at a mean altitude of 39 m asl, while the second is in a mid-mountain area, at a mean altitude of 260 m asl. Climate is Mediterranean, with wet, mild winters and long, dry summers. Mean annual temperature is 17 °C in DÑ, and ranges from 13.4 to 17.6 °C in SN. Mean annual rainfall is 590 mm in DÑ, and varies between 600 and 750 mm along altitudinal gradients in SN. Soils are very sandy in DÑ, derived from quarcitic sands which come from marine regression, while predominant soils in SN are acid, derived from slate rocks.

Six dehesa stands were included in this study, two in DÑ and four in SN. The stands were ca 5 ha in size and were at a minimal distance of 1.5 km from each other. The specific characteristics of different stands are presented in Table 1. In all stands oak (pure holm-oak or mixed holm-oak and cork oak populations) density was within the range of dehesas in south western Spain (i.e., 30–42 trees·ha−1), and the understory was occupied by a short annual grassland layer with no shrubs. They were grazed by different types of livestock and inhabited by medium sized and small wild mammals that predate on acorns (rabbit, hare, badger, wood mouse). In addition, the stands were chosen according to the following criteria: the oaks bore abundant acorn crop at the beginning of the study, dead branches from oak pruning were available, and there was evidence of differential rodent activity among them.

Table 1

Stand characteristics: Names (DÑ, Doñana; SN, Sierra Norte de Sevilla; H, high rodent activity; L, low rodent activity)

Stands	Location	Oak density (trees ha−1)	Livestock type (stocking rate)	Medium-sized mammals
DÑL	37° 15′ 31″N 6° 19′ 51″W	34	Cattle (40 heads in100 ha) Horses (10 heads in100 ha)	Rabbits Hares
DÑH	37° 14′ 33″N 6° 19′ 37″ W	30	Goats (500 heads in 100 ha, Grazing 6 h day−1)	Rabbits Badgers
SN1L	37° 40′ 04′N 5° 56′ 44″W	42	Cattle (45 heads in 80 ha) Iberian pigs (30 heads in 80 ha)	Rabbits Hares
SN2L	37° 40′ 31′’N 5° 56′ 07″W	32	Cattle (40 heads in 90 ha)	Rabbits
SN3H	37° 39′ 43″N 5° 56′ 01″W	39	Cattle (30 heads in 150 ha) Iberian pigs (60 heads in 150 ha)	Rabbits Hares
SN4H	37° 39′ 05″N 5° 57′ 26″W	31	Cattle (50 heads in 100 ha) Iberian pig (90 heads in 100 ha)	Rabbits

Livestock type (grazing information provided by shepherds and staff). Presence of medium-sized mammal (according to shepherd information, personal observations and presence of scats and clues)

Rodent activity assessment

Rodent activity was indirectly assessed by careful prospection of rodent signs and traces in the oak trunks (i.e., scats, gnawed seeds and sawdust). Five 200-m length lineal transects were established in each stand, and the oaks occurring 10 m left and 10 m right were examined and recorded.

Experimental design

Artificial shelters of dead branches (SODBs hereafter) were built using branches from oak pruning of 5–6 cm in diameter × ca 1.5 m in length. These SODBs were hollow, conical structures of ca. 1 m2 base area × 75 cm height. They were used in two experiments, one to assess the amount of acorns that escaped predation (acorn retention experiment) and the other to assess their potential role as a sink for dispersed acorns (dispersal experiment).

Acorn retention experiment

The experiment included the six aforementioned dehesa stands, the half of them exhibited high rodent activity and the other half low rodent activity (see Results section). At each stand seven treatment blocks were established (each block was placed beneath the canopy border of a randomly selected oak tree). The blocks were at a minimal distance of 40 m from each other. In each block a single replicate of each of the following five treatments was applied: (1) “Open”, unrestricted to any kind of acorn predators; (2) “Fenced”, small sized (1 m2 base area × 1.20 m height) standard fences to exclude livestock but to allow mid- and small-sized mammals through an open strip (30 cm-tall) at the bottom; (3) “Fenced and Caged”, cubic shaped cages (30 × 30 × 15 cm) of wire-mesh (5 × 5 cm-cells) to exclude predators other than rodents. The cages were placed inside fences (as describe above) to avoid livestock trampling; (4) Protected by a “SODB” (as described above) which were expected to exclude ungulates but not mid- and small-sized mammals; (5) Protected by “SODB and Caged”, cages (as described above) were installed beneath a SODB to exclude predators other than rodents. This makes seven replicates per treatment per stand and a total of 210 sampling points of acorn retention. At each sampling point we offered 25 mature and apparently sound acorns spread in 30 × 30 cm supply plots. The acorns were collected from the ground behind the canopy of several (i.e., 6–7) holm-oak trees per stand and were pooled and homogenized by size. The experiment began in late October 2013 and ended in mid-April 2014, during the natural acorn-fall period. During this period the supply plots were monitored monthly, and the amount of retained acorns (i.e., whole, not gnawed acorns) was recorded. A total of 5250 acorns were used in this experiment.

Digital camera traps (Wildview Extreme 3) were used to record qualitative information on rodents visiting the plots. Cameras were placed in the “Fenced and Caged” treatment, in three blocks per stand. They were installed for four one-night periods during the experiment.

Acorn dispersal experiment

The dispersal experiment was conducted in the three “high rodent activity” stands (i.e., DÑH, SN3H and SN4H, see Results section) with seven replicated experimental units per stand. The experimental units consisted in a “Fenced and Caged” supply plot (as described above) and a “SODB” in the proximity (i.e., 1.5 m apart). The SODBs were not supplied with acorns in order to assess their potential role as a sink for dispersed acorns. As in the previous case experimental units were placed beneath the canopy border of a randomly selected oak and the replicates in the same stands were at a mean distance of 40 m from each other. Lots of 25 marked acorns were added to the supply plots (“Fenced and Caged”) what makes a total of 525 acorns in the three stands. The acorns were collected using the same method described above and were individually marked by 2 × 14 cm tagged and coloured plastic flags (following Perea et al. 2011a) attached to the acorns with a staple. The use of staples to attach the flags instead of commonly used wire (Li and Zhang 2003; Xiao et al. 2004, 2006) simplified the labelling process (F. Pulido pers. comm.) and added very little weight (i.e., 44 ± 1 mg) to the acorns. Digital camera traps as described above were also placed in three supply plots per stand.

This experiment began in late October and ended in mid-April. Censuses on acorn fate began 4 days after initiating the experiment and were repeated weekly during the first month and fortnightly afterward. When a dispersed acorn was later located by us, information was recorded on: distance to the source plot which was measured by a laser distance metre (Leica DISTO DM8), microhabitat type (SODB, spontaneous dead branches lying on the ground, open grassland), acorn position (soil surface, buried into the soil or litter) and predation degree (entire acorn; scarcely predated: ≤20 % consumed; moderately predated: 20–60 % consumed; highly predated: 60–90 % consumed; completely consumed: empty coat).

To assess whether the distribution of dispersed acorns was random with respect to the abundance of different microhabitats the cover of the SODB, dead branches and open grassland was estimated within a circle of radius 8.5 m, centered in the supply plot. This distance encompasses the mean dispersal distance that was found in this experiment (see Results section). The cover of the SODB was estimated from its base area after measuring maximum and minimum diameters. The cover of spontaneous dead branches lying on the ground was estimated after measuring their individual length and width. The grassland cover was estimated as the difference among the area of the 8.5 m-radius circle (i.e., 227 m2) and the area covered by the other microhabitats.

Data analyses

The final number of whole acorns in the supply plots was used as response variable in the acorn retention experiment, while gnawed acorns and broken acorn parts were not taken into account in the data analyses. A GLM—Logistic model was used to analyze the acorn retention as affected by different factors (i.e., rodent activity, stand identity and treatment against predators). Rodent activity (two categories) and stand (six categories) were included as random variables, with stand nested within rodent activity, while treatment against predators (five categories) was nested within stand and rodent activity. The SODB-Caged treatment was used as reference category in the model. The response variable was fit to a binomial probability distribution with logit as the link function (Quinn and Keough 2002).

In the dispersal experiment all acorns that were relocated by us out of the supply plots were considered as “dispersed”, regardless of whether they were whole or had experienced any degree of consumption.χ2 tests were used to analyse differences in the distribution of dispersed acorns among different microhabitats. Many acorns were repeatedly relocated by us at different times, and exhibited progressive increase in consumption level. Thus, relocated acorns were classified in different temporal consumption categories: highly or totally consumed from the first relocation, wholly or scarcely consumed at early relocations but totally consumed at the end, and wholly or scarcely consumed at early relocations and lost afterwards. The χ2 test was used to analyze whether the frequency of these temporal consumption categories was significantly different among different microhabitats.

Results

Rodent activity assessment and records from camera traps

Most oaks that exhibited rodent traces were those with damaged trunks (i.e., exhibiting holes and rotten zones), while very few apparently sound oaks exhibited rodent signs. According to the abundance of the former oaks category the stands were classified as follows: “low rodent activity” (L), those with <5 % oaks exhibiting rodent signs, and “high rodent activity” (H), those with ≥25 % of oaks exhibiting rodent signs. Three of the stands, namely DÑL, SN1L and SN2L, were in the low rodent activity class while the other stands, namely DÑH, SN3H and SN4H, were in the high rodent activity class.

Twelve images from camera traps revealed the presence of mice (Mus spretus and Apodemus sylvaticus) visiting the plots (“Fenced and Caged” treatment) in all stands. A direct observation event of Mus spretus and several observations of rabbits also took place in the DÑH and SN1L stands, respectively. All observations were done in the early morning.

Acorn retention

The three explanatory variables included in the model were highly significant (Table 2). A positive and significant model paramenter (β = 1.51, P < 0.001) for the low rodent activity category of stands indicated an overall increase in acorn retention under these conditions comparatively to the high rodent activity category. Different stands within each category of rodent activity also varied in acorn retention. Under low rodent activity, the DÑL stand experienced the highest overall acorn retention. The model parameter for this stand was positive and highly significant (β = 2.39, P < 0.001), indicating higher acorn retention in this stand than in the SN2L stand, the reference category. However the same was not true for the stand SN1L whit non-significant model parameter (β = 0.39, P = 0.82), indicating no difference comparatively to the SN2L stand. Under high rodent activity, the DÑH stand experienced the lowest overall acorn retention. The model parameter for this stand was negative and significant (β = −2.29, P < 0.01), indicating lower amount of retained acorns comparatively to the SN4H, the reference category. However the same was not true for the stand SN3H, whit non-significant model parameter (β = 0.06, P = 0.61), indicating no difference comparatively to the SN4H stand.

Table 2

Summary of the GLM Logistic model testing the factors affecting acorn retention

Source	df	Wald X2	P value
Rodent activity	1	291.6	0.0001
Stand	4	381.5	0.0001
Treatment	24	419.7	0.0001

Log likelihood = 38032.3; AIC = −76004.5; deviance = 882.1

Under normal field conditions (i.e., Open treatment), 20.6–23 acorns (i.e., 82 to 92 % of supplied acorns) were loss in average from different stands (Fig. 1). The effect of treatments against predators on acorn retention exhibited an inverse pattern in stands with low and high rodent activity, respectively (Fig. 1). Under low rodent activity (i.e., stands DÑL, SN1L and SN2L), the acorn retention in the Open treatment was much lower than in the other treatments (Fig. 1a, c, e). The model parameters for this treatment category were negative and significant in the three stands (Table 3), indicating a decrease in acorn retention relatively to the SODB-Caged treatment (i.e., reference category). In addition, in two of these stands (SN1L and SN2L) the acorn retention was lower in the Fenced treatment where mid-sized mammals were allowed than in other exclosures (Fig. 1c, e). In these stands the model parameters were negative and significant for the Fenced treatment (Table 3) but not for the SODB treatment where mid-sized mammals where allowed too. Oppositely, under high rodent activity all non-SODB treatments (i.e., Open, Fenced and Fenced-Caged) retained more acorns than the SODB treatments, both Caged and not-Caged (Fig. 1b, d, f), and the model parameters were positive and significant for the former treatments in the three stands (Table 3). No difference in acorn retention among both SODB treatments (Fig. 1) was indicated by lack of significant model parameters for the SODB category (Table 3).

Fig. 1

Acorn retention (number of acorns that escaped predation) in different treatments (average values ± se bars) in each stand

Table 3

Model parameters (β) for categories of exclusion against predators (treatments) in different stands and significance level according to Wald χ2

Treatments	Low rodent activity stands			High rodent activity stands
	DÑL	SN1L	SN2L	DÑH	SN3H	SN4H
Open	−3.91***	−2.03***	−1.65***	2.35**	1.13***	1.13***
Fenced	ns	−1.19***	−0.85**	1.54*	1.07***	1.10***
Fenced-caged	ns	ns	ns	2.10**	1.19***	1.07***
SODB	ns	ns	ns	ns	ns	ns
SODB-caged	–	–	–	–	–	–

*** P < 0.001; ** P < 0.01; * P < 0.05

Acorn dispersal

Of the 175 acorns supplied in each stand a total of 156, 152 and 154 acorns were dispersed in stands DÑH, SN3H and SN4H, respectively (i.e., 87–89 % of total), and the rest remained in the supply plots. Of these remaining acorns, 5, 3 and 2 acorns, respectively, were consumed “in situ”, and 14, 20 and 19 acorns, respectively, remained unconsumed (i.e., 8–11.4 % of total). Of the acorns that were dispersed, a total of 85, 89 and 96 were relocated in stands DÑH, SN3H and SN4H, respectively (i.e., 48–55 % of total) and the rest were not found (i.e., lost acorns).

The majority of relocated acorns were not cached although several cached acorns were also found in all stands (12, 5 and 7 acorns in DÑH, SN3H and SN4H, respectively). Most of these acorns were individually cached, although 7 acorns were found in a common larder in the DÑH stand (Table 4).

Table 4

Observed and expected (in brackets) numbers of acornsa in different categories, dispersed to different microhabitats in the high rodent activity stands

Acorns in different consumption categories	SOBD	Open grassland	Spontaneous branches
Stand DÑH (χ2 = 20.95; P < 0.001)
Totally or highly consumed from the beginning	20 (11.3)	4 (13.0)	7 (6.5)
Entire at the beginning and completely predated at end	6 (10.9)	177g (12.7)	71i (6.3)
Entire at the beginning and loosed afterward	5 (8.7)	154i (10.2)	4 (5.1)
Stand SN1H (χ2 = 12.85; P < 0.025)
Completely or highly predated from the beginning	35 (28.6)	4 (9.7)	12 (12.6)
Entire at the beginning and completely predated at end	4 (7.3)	4 (2.5)	52i (3.2)
Entire at the beginning and loosed afterward	11 (14.0)	93i (4.7)	5 (6.2)
Stand SN2H (χ2 = 20.38; P < 0.001)
Completely or highly predated from the beginning	27 (17.9)	6 (13.3)	4 (8.7)
Entire at the beginning and completely predated at end	11 (15.9)	122i (11.8)	10 (7.8)
Entire at the beginning and loosed afterward	5 (12.5)	144i (9.3)	71i (6.5)

aSome categories include an amount of acorns that were found in caches (superscript numbers indicate this amount), either individually (i) or in a group (g). The rest of the acorns were found on soil surface

χ2 values and significance of differences in each stand is reported

Average dispersal distances for relocated acorns were 4.8 ± 3, 6.5 ± 4 and 5.7 ± 4 m in the DÑH, SN3H and SN4H stands, respectively. However, four acorns in total were dispersed much farther, at 45, 70, 88 and 122 m, respectively.

Acorns were not randomly moved by mice to different microhabitats in any of the stands (χ2> 134, P < 0.001; Table 3), with the proportion of acorns moved to the SODBs and to the spontaneous dead branches lying on the soil being higher than expected and the proportion of acorns moved to the open grassland being lower than expected.

Acorns in different temporal consumption categories (Table 4) were not randomly distributed among different microhabitats in any stand (χ2 > 12, P < 0.03). Of the acorns that were moved to a SODB, those highly consumed or completely consumed from the first relocation event were more frequent than expected, while those that were entire at the beginning and either totally consumed or lost afterwards were less frequent than expected. Conversely, of the acorns moved to the open grasslands those that were entire at the beginning and either totally consumed or lost afterward were more frequent than expected, while those that were completely or highly consumed from the beginning were less frequent than expected. Finally, there were no differences among observed and expected frequencies for the acorns that were moved to spontaneous branches.

Discussion

Effect of shelters of dead branches on acorn predation

The results of this study revealed that, in grazed Mediterranean dehesas, the potential benefits of artificial shelters of dead branches for oak regeneration were dependent on the level of rodent activity in the stands. Rodent activity needed to be low to gain a positive effect of the SODBs on the acorns safety. Although we have measured “acorn retention”, which does not take into account those acorns potentially dispersed by mice, this measurement can be considered as an effective estimator of the amount of acorns that escaped predation, as the final fate of dispersed acorns was predation in all cases. When rodent activity was low (i.e., L-stands), livestock was of high importance for predation with regard to the few acorns that escaped predation in the Open plots comparatively to the Fenced and SODB treatments. Thus, the shelters of dead branches acted as effective physical barriers against livestock, avoiding acorn consumption by these ungulates. Similar results were found by Leverkus et al. (2013) in the case of wild ungulates (boars) in burned Mediterranean areas. In that case, wild boars predated fewer acorns where fallen burnt logs were left on ground than where trunks were cut and the branches were masticated. Moreover, contrary to our expectations, the SODBs were not apparently accessed by rabbits and hares, indicating an additional positive effect of the artificial shelters against acorn consumption by this guild of predators. The mid-sized acorn-predators (i.e., rabbit, hare and badger) are locally abundant in many Mediterranean dehesas (Crawley and Long 1995).

In opposition, when rodent activity was high (i.e., H-stands), the shelters of dead branches increased acorn predation by rodents comparatively to the non-SODBs treatments. These results are consistent with studies which state that rodent behaviour may be affected by the provision of shelter where they can safely forage and feed (e.g., Díaz 1992; Hulme 1994; Den Ouden et al. 2005). Also, Pulido et al. (2013) found a high removal rate by rodents of the acorns placed under piles of dead branches in dehesas in central Spain, while Leverkus et al. (2013) found a high acorn predation by rodents in burned areas with fallen burnt logs lying on ground.

Effect of shelters of dead branches on acorn dispersal

Results of the dispersal experiment carried out in the high rodent activity stands were consistent with the results of the predation experiment conducted in the same stands. Thus, the proportion of acorns unhandled by rodents in the supply plots varied from 8 to 11.4 % in the dispersal experiment and from 7.6 to 16 % in the predation experiment (Fenced and Caged treatment, Fig. 1). On the other hand, acorns that were dispersed in our study stands represented 87–89 % of the offer which fit within the range provided for acorn dispersal by rodents in other Mediterranean oak ecosystems (i.e., 96.6 and 65.7 % in central and south-western Spain, respectively) (Muñoz and Bonal 2007; Gómez et al. 2008).

The non-random overrepresentation of the acorns that were moved by rodents to the artificial shelters and the spontaneous branches lying on the ground comparatively to the open grassland (Table 5) indicated that these microhabitats acted as preferred sinks for the dispersed acorns. The same rodent behaviour has been reported in other oak ecosystems regarding shrub cover, a preferred destination for seeds dispersed by mice (Den Ouden et al. 2005; Perea et al. 2011a). In spite of this, the SODBs did not act as “safe sites” for acorns (sensu Harper 1977) as most of the acorns moved to the SOBDs were consumed by rodents significantly more quickly than those acorns moved to the open grasslands. These results are consistent with the findings from Perea et al. (2011a), which found that seeds that were dispersed to shrub cover had a higher probability of being consumed than the seeds that were dispersed to open microsites. No dispersed acorns survived until the end of the experiment in any of the stands and microhabitats included in this study, in agreement with studies which consider that seed harvesting by small rodents is mostly equivalent to predation (e.g., Schupp 1990; Hulme 2002; Mendoza and Dirzo 2007).

Table 5

Observed and expected (in brackets) proportions of acorns dispersed to different microhabitats in the high rodent activity stands

	SOBD	Open grassland	Spontaneous branches
DÑH (N = 85) χ2 = 134.8; P < 0.0001	0.36 (0.10)	0.42 (0.83)	0.21 (0.07)
SN3H (N = 89) χ2 = 220.8; P < 0.0001	0.56 (0.16)	0.19 (0.76)	0.25 (0.08)
SN4H (N = 96) χ2 = 165.8; P < 0.0001	0.45 (0.14)	0.33 (0.78)	0.22 (0.08)

(N): Total number of acorns included in the analysis. Expected values are based on cover of each microhabitat type. Values of χ2 tests and significance level are provided

Our results also show that in most cases the consumption of an individual acorn by mice was a progressive process. As far as we know there is little information on this progressive consumption suffered by the individual acorns. However, these results should be taken into account facing acorn fate. It has been shown that partial endosperm consumption of non-vital acorn parts, which is frequent in the acorns found in the field (Mancilla-Leyton et al. 2012), does not prevent germination and seedling establishment under controlled conditions (Perea et al. 2011b). Nevertheless, according to this study, partial acorn consumption by rodents was just the first stage of a complete predation process with not opportunity for seedling establishment.

Management implications

Piles of dead branches—an inexpensive and abundant material from oak pruning or self-shedding—can contribute to decrease acorn predation by ungulates in the Mediterranean dehesas and other disturbed forests, thus enhancing oak regeneration in these ecosystems. Other potential benefit of these artificial shelters is preventing trampling and browsing on seedlings by ungulates. This browser avoidance effect was also provided by coarse wood debris in old-growth temperate forests (Smit et al. 2012) and by shrubs in Mediterranean forests (Pulido and Díaz 2005; Smit et al. 2008, 2009). Thus, the SODBs could substitute the lack of shrub cover in dehesas and other grazed ecosystems or in ecosystems affected by other disturbances. Although the benefits of the shelters of dead branches on acorn safety were limited to stands with a low rodent activity in this study, they could be extended to sites with higher rodent activity, whether this method is combined with other techniques to prevent predation by rodents. For instance, acorns included in small individual capsules (Castro et al. 2015) or seeds treated with mice repellents were efficiently protected against predation by rodents (Willoughby et al. 2011 but see Leverkus et al. 2013).

High variability between years in rodent’s density and activity is characteristic of Mediterranean dehesas and other oak forests (e.g., Muñoz et al. 2009). It likely represents temporal windows for successful use of shelters of dead branches to improve oak regeneration. In addition, the SODBs provide shaded conditions which are essential to avoid seedling desiccation during summer in the Mediterranean environment (Herrera et al. 1994; Badano et al. 2009; Pausas et al. 2004; Leiva et al. 2013; Marañón-Jiménez et al. 2013). Finally, the SODBs method could be selectively applied in suitable dehesa stands. As far as this study indicates these stands would be those with a low proportion of damaged oaks, which exhibited low rodent activity. However, a more detailed study on this relationship would be necessary.

Acknowledgements

Our special thanks go to the management of the Doñana and Sierra Norte de Sevilla Natural Parks, Dehesa de Gatos S. L. and Navalagrulla state for their valuable help, support and facilities. We also thank to José María Fedriani for useful advices on rodent activity assessment and to Fernando Pulido for advices on acorns tagging. Frank Balao helped us with statistical analysis. Benjamín Ojeda provided very valuable collaboration in different stages of the experiments development, Frank Blanco-Velazquez and JC Muñoz-Reinoso helped us with camera traps installation and pictures monitoring. Juan Manuel Mancilla-Leyton and Angel Martín-Vicente provided useful support in the field and laboratory. Two anonymous reviewers helped improve the quality of preliminary version of the manuscript. Rachel Baron revised the English version of the manuscript.

References

1. Badano EI, Pérez D, Vergada CH (2009) Love of nurse plants is not enough for restoring oak forests in a seasonally dry tropical environment. Restor Ecol 17:571–576CrossRef
2. Bonal R, Muñoz A (2007) Multitrophic effects of ungulate intraguild predation on acorn weevils. Oecologia 152:533–540CrossRefPubMed
3. Castro J, Zamora R, Hódar JA, Gómez JM (2002) Use of shrubs as nurse plants: a new technique for reforestation in Mediterranean mountains. Restor Ecol 10:297–305CrossRef
4. Castro J, Leverkus AB, Fuster F (2015) A new device to foster oak forest restoration via seed sowing. DOI, New For. doi:10.1007/s11056-015-9478-4
5. Costa JC, Martín-Vicente A, Fernández-Alés R, Estirado-Oliet M (2006) Dehesas de Andalucía Caracterización Ambiental. Consejería de Medio Ambiente-Junta de Andalucía, Sevilla
6. Crawley MJ, Long CR (1995) Alternate bearing, predator satiation and seedling recruitment in Quercus robur L. J Ecol 83:683–696CrossRef
7. Den Ouden J, Jansen PA, Smit R (2005) Jays, Mice and Oaks: Predation and dispersal of Quercus robur and Q. petraea in North-western Europe. In: Forget PM, Lambert J, VanderWall SB (eds) Seed fate, Predation, dispersal and seedling establishment. CABI, Wallingford, pp 223–240CrossRef
8. Di Castri F, Goodall DW, Specht RL (1981) Mediterranean-type shrublands. Elsevier, Amsterdam
9. Díaz M (1992) Rodent seed predation in cereal crop areas of central Spain: effects of physiognomy, food availability, and predation risk. Ecography 15:77–85CrossRef
10. Doctor-Cabrera AM (2003) Territorio e Incendios Forestales. Consejería de Medio Ambiente-Junta de Andalucía, Sevilla
11. Espelta JM, Riba M, Retana J (1995) Pattern of seedling recruitment in west Mediterranean coppiced holm-oak (Quercus ilex L.) Mediterranean forest as influenced by canopy development. J Veg Sci 6:465–472CrossRef
12. Gómez JM, Puerta-Piñeiro C, Schupp EW (2008) Effectiveness of rodents as local seed dispersers of Holm oaks. Oecologia 155:529–537CrossRefPubMed
13. Gómez-Aparicio L, Valladares F, Zamora R (2006) Differential light responses of Mediterranean tree saplings: linking ecophysiology with regeneration niche in four co-occurring species. Tree Physiol 26:947–958CrossRefPubMed
14. Grove AT, Rackham O (2001) The nature of Mediterranean Europe. An ecological history. Yale University Press, New Haven
15. Harper JL (1977) Population biology of plants. Academic, London
16. Herrera CM, Jordano P, López-Soria L, Amat JA (1994) Recruitment of a mast-fruiting, bird-dispersed tree: bridging frugivore activity and seedling establishment. Ecol Monogr 64:315–344CrossRef
17. Hulme PE (1994) Post-dispersal seed predation in grasslands: its magnitude and sources of variation. J Ecol 82:645–652CrossRef
18. Hulme PE (2002) Seed-eaters: seed dispersal, destruction and demography. In: Levey DJ, Silva WR, Galetti M (eds) Seed dispersal and frugivory: ecology, evolution and conservation. CAB, Wallingford, pp 257–273
19. Jensen SP, Gray SJ, Hurst JL (2003) How does habitat structure affect activity and use of space among house mice? Anim Behav 66:239–250CrossRef
20. Lagos VO, Contreras LC, Meserve PL, Gutiérrez JR, Jaksic FM (1995) Effects of predation risk on space use by small mammals: a field experiment with a Neotropical rodent. Oikos 74:259–264CrossRef
21. Leiva MJ, Fernández-Alés R (2003) Postdispersive losses of acorns from Mediterranean savannah-like forests and shrublands. Forest Ecol Manag 176:265–271CrossRef
22. Leiva MJ, Mancilla-Leyton JM, Martín-Vicente A (2013) Methods to improve the recruitment of holm-oak seedlings in grazed Mediterranean savanna-like ecosystems (dehesas). Ann For Sci 70:11–20. doi:10.1007/s13595-012-0225-0CrossRef
23. Leverkus AB, Castro J, Puerta-Piñero C, Rey Benayas JM (2013) Suitability of the management of habitat complexity, acorn burial depth, and a chemical repellent for post-fire reforestation of oaks. Ecol Eng 53:15–22CrossRef
24. Li HJ, Zhang ZB (2003) Effect of rodents on acorn dispersal and survival of Liaondong oak (Quercus liatungensis Koidz.). Forest Ecol Manag 176:387–396CrossRef
25. Mancilla-Leyton JM, Cambrollé J, Martín-Vicente A (2012) The impact of the common rabbit on cork oak regeneration in SW Spain. Plant Ecol 213:1503–1510CrossRef
26. Marañón T, Ibáñez B, Anaya-Romero M, Muñoz-Rojas M, Pérez-Ramos IM (2012) Oak trees and woodlands providing ecosystem services in southern Spain. In: Roterham ID et al (eds) Trees Beyond the Wood. An exploration of concepts of woods forests and trees, Wildtrack, Sheffield, pp 369–378
27. Marañón-Jiménez S, Castro J, Querejeta JI, Fernández-Ordoño E, Allen CD (2013) Post-fire wood management alters water stress, growth, and performance of pine regeneration in a Mediterranean ecosystem. Forest Ecol Manag 308:231–239CrossRef
28. Martín-Vicente A, Fenández-Ales R (2006) Long term persistence of dehesas. Evidences from history. Agroforest Syst 67:19–28CrossRef
29. Mendoza E, Dirzo R (2007) Seed-size variation determines interspecific differential seed predation by mammals in a neotropical rain forest. Oikos 116:1841–1852CrossRef
30. Muñoz A, Bonal R (2007) Rodents change acorn dispersal behaviour in response to ungulate presence. Oikos 116:1631–1638. doi:10.1111/j.2007.0030-1299.15710.xCrossRef
31. Muñoz A, Bonal R, Díaz M (2009) Ungulates, rodents, shrubs: interactions in a diverse Mediterranean ecosystem. Basic Appl Ecol 10:151–160CrossRef
32. Pausas JG, Bladé C, Valdecantos A, Seva JP, Fuentes D, Alloza JA, Vilagrosa A, Bautista S, Cortina J, Vallejo R (2004) Pines and oaks in the restoration of Mediterranean landscapes of Spain: new perspectives for an old practice—a review. Plant Ecol 171:209–220CrossRef
33. Perea R, San Miguel A, Gil L (2011a) Acorn dispersal by rodents: the importance of re-dispersal and distance to shelter. Basic Appl Ecol 12:432–439CrossRef
34. Perea R, San Miguel A, Gil L (2011b) Leftovers in seed dispersal: ecological implications of partial seed consumption for oak regeneration. J Ecol 99:194–201CrossRef
35. Pulido FJ, Díaz M (2005) Regeneration of a Mediterranean oak: a whole cycle approach. Ecoscience 12:92–102CrossRef
36. Pulido F, Castaño FM, Peña L., Mateos A, Bonilla A, Duque J (2013) Utilización de restos de poda como refugio de la regeneración en dehesas de encina. http://www.congresoforestal.es/actas/doc/6CFE/6CFE01-104.pdf. Accessed 28 Nov 2014
37. Quinn GP, Keough MJ (2002) Experimental design and data analysis for biologist. Cambridge University Press, CambridgeCrossRef
38. Ramirez JA, Diaz M (2008) The role of temporal shrub encroachment for the maintenance of Spanish holm oak Quercus ilex dehesas. For Ecol Manag 255:1976–1983CrossRef
39. Rodríguez-Calcerrada J, Cano FJ, Valbuena-Carabaña M, Gil L, Aranda I (2010) Functional performance of oak seedlings naturally regenerated across microhabitats of distinct overstorey canopy closure. New For 39:245–259CrossRef
40. Schupp EW (1990) Annual variation in seedfall, post-dispersal predation and recruitment of a neotropical tree. Ecology 71:504–515CrossRef
41. Siscart D, Diego V, Lloret F (1999) Acorn ecology. In: Rodá F, Retana J, Gracia CA, Bellot J (eds) Ecology of Mediterranean evergreen oak forests. Springer, Berlín, pp 75–86CrossRef
42. Smit C, den Ouden J, Diaz M (2008) Facilitation of Quercus ilex recruitment by shrubs in Mediterranean open woodlands. J Veg Sci 19:193–200CrossRef
43. Smit C, Díaz M, Jansen P (2009) Establishment limitation of holm oak (Quercus ilex subsp. ballota (Desf.) Samp.) in a Mediterranean savanna—forest ecosystem. Ann For Sci 66:511–517CrossRef
44. Smit C, Kuijper DPJ, PrenticeD Wassem MJ, Cromsigt JPGM (2012) Coarse woody debris facilitates oak recruitment in Białowieza Primeval forest, Poland. For Ecol Manag 284:133–141CrossRef
45. Torre I, Díaz M (2004) Small mammal abundance in Mediterranean post-fire habitats: a role for predators? Acta Oecol 25:137–142CrossRef
46. Vander Wall SB (1990) Food hoarding in animals. University of Chicago Press, Chicago
47. Vander Wall SB (2001) The evolutionary ecology of nut dispersal. Bot Rev 67:74–117CrossRef
48. Vander Wall SB, Longland WS (2004) Diplochory: are two seed dispersers better than one? Trends Ecol Evol 19:155–161CrossRefPubMed
49. Vander Wall SB, Kuhn KM, Beck MJ (2005) Seed removal, seed predation, and secondary dispersal. Ecology 86:801–806CrossRef
50. Willoughby IH, Jinks RL, Morgan GW, Pepper H, Budd J, Mayle B (2011) The use of repellents to reduce predation of tree seed by wood mice (Apodemus sylvaticus L.) and grey squirrels (Sciurus carolinensis Gmelin). Eur J For Res 130:601–611CrossRef
51. Xiao Z, Zhang Z, Wang Y (2004) Dispersal and germination of big and small nuts of Quercus serrata in a subtropical broad-leaved evergreen forest. For Ecol Manag 195:141–150CrossRef
52. Xiao Z, Zhang Z, Wang Y (2005) Effects of seed size on dispersal distance in five rodent-dispersed fagaceous species. Acta Oecol 28:221–229CrossRef
53. Xiao Z, Jansen PA, Zhang Z (2006) Using seed-tagging methods for assessing post-dispersal seed fate in rodent-dispersed trees. For Ecol Manag 223:18–23CrossRef

For further details log on website :
http://link.springer.com/article/10.1007/s11056-015-9486-4