Forest Ecology and Management 319 (2014) 44–50

Contents lists available at ScienceDirect

Forest Ecology and Management

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

How do edge effect and tree species diversity change bird diversity andavian nest survival in Germany’s largest deciduous forest?

http://dx.doi.org/10.1016/j.foreco.2014.02.0040378-1127/� 2014 Elsevier B.V. All rights reserved.

⇑ Corresponding author. Tel.: +49 551 3922257; fax: +49 551 398806.E-mail address: pbatary@gmail.com (P. Batáry).

Péter Batáry ⇑, Stefanie Fronczek, Claudia Normann, Christoph Scherber, Teja TscharntkeAgroecology, Georg-August University, Grisebachstr. 6, D-37077 Göttingen, Germany

a r t i c l e i n f o

Article history:Received 27 November 2013Received in revised form 30 January 2014Accepted 5 February 2014Available online 24 February 2014

Keywords:Artificial nestBeechBreeding birdsForest edgeMayfield survival rateNest predation

a b s t r a c t

Habitat fragmentation is a major driver of species loss. Here we test the hypotheses that high tree diver-sity in a large deciduous forest enhances bird diversity and nest survival. We further expect that forestedges support higher bird diversity when different habitat types adjoin, whereas nest predation is nothigher, because the large forest area mitigates potential edge effects. We studied how edge-centre differ-ences and tree diversity (beech-dominated vs. tree-species rich) affect the bird community and survivalrates of ground breeding birds’ nests based on an artificial nest predation experiment in the HainichNational Park, Germany. We surveyed birds three times during the breeding season. We selected six for-est stands with low tree diversity (i.e. dominated by beech) and six with high tree diversity (i.e. tree-spe-cies rich). Each forest stand contained four bird survey plots (plot 1: 0–30 m, plot 2: 60–90 m, plot 3:120–150 m and plot 4: 180–210 m distant from edge; altogether 48 bird survey plots). Additionally eachplot corner contained one artificial ground nest baited with one Blue-breasted Quail egg and one plasti-cine egg for eight days of exposure in the middle of the breeding season. Bird abundance and diversitywere higher in the first 30 m of the forest. Bird diversity, including ground breeding birds, was alsoenhanced by higher percentages of bushes, which can provide enhanced food supply, perches as wellas sheltering. Nest predation showed no edge effect, supporting the idea that small area of forest frag-ments causes more important negative effects than the edge in large forest remnants. Predation rateswere higher in tree-species rich stands compared to beech-dominated stands, probably due to greaterdiversity and density of mammalian predators. Edge effects shaped the bird community compositionand positively affected abundances of tree and shrub breeding birds, but did not affect ground breedersand the nest predation of ground nests. Shrub breeders accumulating in forest edges might, however, suf-fer more from nest predation in forest fragments. In conclusion, bird diversity and avian egg predationwere affected by both forest edges and tree diversity in surprisingly different ways.

� 2014 Elsevier B.V. All rights reserved.

1. Introduction

As a consequence of habitat fragmentation, the global amountof habitat edges increases more and more. Road constructions,housing development, agricultural intensification and forest log-ging have been shown to lead to an elevated number of artificiallycreated habitat edges (Fahrig, 2003). Edge effects occur at the bor-der of two adjacent ecosystems interacting with each other (Mur-cia, 1995), and these effects are less pronounced in large fragmentsdue to their smaller perimeter-area ratio (Helzer and Jelinski,1999). The effects of edges can be due to (1) changing abiotic con-ditions such as radiation influx, temperature, wind and humidity,(2) direct biotic effects resulting in changed abundance and

distribution of a species caused by the physiological tolerance ofspecies, or (3) indirect biological effects such as predator–prey-interactions, parasitism or competition (Murcia, 1995).

Edges can have a positive effect on bird life because of increasedabundance and species richness at forest edges (Johnston, 1947).However, higher nest predation risk and nest parasitism rates nearforest edges have been shown to influence bird reproductive suc-cess (Robinson et al., 1995), with a strong negative effect on birdpopulation densities (Fretwell, 1972). According to this, Gatesand Gysel (1978) suggested that edges function as ‘‘ecologicaltraps’’ for nesting birds. After their initial study, nest predation inrelation to edge effect was studied all over the world in differenthabitat types. Results of edge effects on nest predation at forestsites have been variable so far. In a meta-analysis, however, Batáryand Báldi (2004) showed that there is a generally higher nestpredation rate in habitat edges compared to habitat interiors. So

P. Batáry et al. / Forest Ecology and Management 319 (2014) 44–50 45

far, there has been no consensus on the question of how far an edgeeffect can penetrate into a habitat patch (e.g. >200 m Wilcove et al.,1986; <150 m in Laurence, 2000). Most studies found an edge ef-fect penetrating not farther than 150 m into the habitat, presum-ably about 50 m from the edge (Batáry and Báldi, 2004), butKeyser et al. (1998) found edge effects only in small, not largeforests.

The number of tree species in a forest may have additional ef-fects on nest survival of ground breeding birds as well as on birddiversity (Salek et al., 2010). Generally, there should be a positiverelationship between habitat heterogeneity, such as by increasedtree diversity, and the diversity of associated species such as birds(Balaz and Balazova, 2012). The assumption is that heterogeneousenvironments provide a greater number of habitat niches and envi-ronmental resources leading to increased species diversity (Vivian-Smith, 1997). In forests dominated by one species, such as beech(Fagus sylvatica), undergrowth vegetation is less species-rich thanin mixed stands (for herb layer from the same study area see Voc-kenhuber et al., 2011).

Artificial nests are a common tool to study edge effects on nestpredation (see e.g. Svobodová et al., 2012), although their reliabil-ity has been long debated (Wilson et al., 1998). Parental nest pro-tection as well as conspicuous behaviour, e.g. parent birds leavingthe nest, may modify nest predation (Berg, 1996), or even deter-mine nest site selection (Tryjanowski et al., 2000). Since parentbirds are not present at artificial nests, artificial nests may not cor-rectly estimate true predation rates (Major and Kendal, 1996).However, for comparative purposes (e.g. comparing different hab-itats or time periods), artificial nests are a timesaving, non-invasivemethod that can represent trends in the nest predation rates (Batá-ry and Báldi, 2005).

In the present study we analysed how forest edges and treediversity affect (i) the bird communities and (ii) survival rates ofground breeding birds’ nests based on an artificial nest predationexperiment. Data collection took place in the largest, connecteddeciduous non-managed forest in Germany (Vockenhuber et al.,2011), so that the forest centre should be minimally affected bythe edge. We tested the following main hypotheses: (1) Edge ef-fects: predation rates in artificial ground nests are higher at forestedges than in forest interiors. Further, both species richness andabundance will be higher at forest edges than in forest interiors, be-cause of higher microhabitat heterogeneity and availability of nest-ing sites and food resources. (2) Effects of tree diversity: birdspecies richness and abundance are higher in tree-species rich for-est stands due to the availability of more niches and food resourcesthan in the beech-dominated stands. In addition, we expected high-er predation rates on artificial ground nests in species-rich foreststands due to higher predator diversity. (3) Interactions betweenedge effects and tree diversity: edge effects on nest predation andbird abundance will be more pronounced in forest stands withlow tree diversity (hereafter termed ‘‘beech-dominated’’ stands),where the contrast between the more heterogeneous forest edgeand the more uniform forest interior is greater than at forest standswith high tree diversity (hereafter called ‘‘species rich’’ stands). (4)Finally, examining three major breeding types (tree breeders, shrubbreeders and ground breeders), we expected the strongest edge andtree diversity effects on the abundance of shrub breeders due totheir special nesting site requirements.

2. Methods

2.1. Study area and sampling design

The study was conducted in the Hainich National Park in thefederal state of Thuringia, Germany. With a size of about

16.000 hectares, the Hainich is the largest continuous deciduousforest area in Germany (Sobek et al., 2009; Vockenhuber et al.,2011). The National Park (covering 7.500 ha) is located on a moun-tain range of shell limestone reaching a maximum elevation of494 m a.s.l. (mean annual temperature: 7.5 �C, mean annual rain-fall: 640 mm; Mölder et al., 2006). Although the forests of the Hai-nich had been used to extract timber over centuries, the standsremained semi-natural compared with other German forests, andforest management has been stopped at least 50y ago.

Before mid-19th century the forest stands had been irregularlyused as coppice-with-standards for many years. Later on, manage-ment was changed to a multiple aged forest system called ‘‘Plenterforest’’ (Schmidt et al., 2009). Plenter forests are high forests thatare permanently regenerated by selective cutting of mature trees.In 1964, the study site became a military training area, which en-abled the development of a large area of near-natural forests. Final-ly, in 1997, the area was turned into a National Park, and recentlybecame part of the UNESCO world natural heritage (Mölder et al.,2006). Today almost 90% of the total area of the National Park isnot managed at all, only a few marginal areas are still used as apasture.

The most abundant tree species in the area is the commonbeech (F. sylvatica); however, in given stands tree diversity is highwith numerous deciduous tree species such as ash (Fraxinus spp.),lime (Tilia spp.) and maple (Acer spp.; Knohl et al., 2008). Around70% of the area is covered by woodruff beech forest (Galio odorat-i-Fagetum).

Within the Hainich National Park, we selected twelve study for-est stands for data collection. Six stands were dominated by beech(mean beech dominance: 81.4%, mean tree species richness: 3.1)and the other six stands contained more tree species (mean beechdominance: 25.1%, mean tree species richness: 6.4). All standswere distributed homogenously along the northern, eastern andsouthern borders of the National Park, which is surrounded by ara-ble land or grassland. The distance between forest stands was atleast one kilometre to ensure independence of replicates (seeAppendix A.1in Supporting information).

At each stand, a transect was laid out starting at the forest edgewith a total length of 210 m. The length of transects was limited tothis distance in order to meet the criteria that each transect shouldbe situated in a forest part that has similar tree diversity. Forestedges consisted of dense shrubs and saplings such as blackthorn(Prunus spinosa) and common ash (Fraxinus excelsior) reaching awidth of around 5 m. Each forest stand contained four bird surveyplots (plot 1: 0–30 m; plot 2: 60–90 m; plot 3: 120–150 m; plot 4:180–210 m) with a plot size of 30 � 60 m (width � length) (Fig. A.1in Supporting information). This plot design was chosen to ensurea sufficient number of non-overlapping counting points withinstands.

2.2. Bird survey

Birds were surveyed using a point-counting method with threebird survey rounds covering the whole breeding season: end ofApril, end of May and middle of June. Surveys started at sunrisewithin 4 h, restricted to good weather conditions (no rain orstorm). For each plot, sampling took 5 min in which all birds wereregistered via listening and sighting while standing still in the mid-dle of each plot. Sampling was limited to 5 min, which was enoughtime to register even inconspicuous bird species, but not too longto risk double counting. We took special attention not to countthe same bird individual more than once. Birds flying through werenot counted. For later analyses, bird species were classified accord-ing to their nesting site as tree-, shrub- and ground breeders(Table A.1 in Supporting information; Bezzel, 1993). Within oneday we surveyed four forest stands. The order in which the forest

46 P. Batáry et al. / Forest Ecology and Management 319 (2014) 44–50

stands were sampled was changed in consecutive censuses for thefollowing survey round. In total, 48 plots were surveyed threetimes.

2.3. Nest predation experiment

For the nest predation experiment, we used artificial groundnests. Each corner of the bird survey plot contained one groundnest; hence the minimum distance between neighbouring nestswas about 30 m (Fig. A.1). In total 192 artificial ground nests wereexposed.

The artificial nests mimicked the nests of common groundbreeders in the study area, such as Willow Warbler (Phylloscopustrochilus), Common Chiffchaff (Phylloscopus collybita), Wood War-bler (Phylloscopus sibilatrix), Wren (Troglodytes troglodytes) andEuropean Robin (Erithacus rubecula). We built a small hollow onthe ground for each artificial nest under saplings, near deadwoodor trunks. In each nest, one Blue-breasted Quail egg (Coturnix chin-ensis) was placed, which was similar in size (length is �22 mm) tothe eggs of local ground breeding bird species (12–15 � 15–20 mm; Bezzel, 1993). In addition, each artificial ground nest wasbaited with one white plasticine egg (15 � 20 mm) in order toroughly determine the predator guild (see next paragraph). Bam-boo sticks with red adhesive tape were placed 50 cm away fromthe artificial nests to mark the place for recovery at nest controls.

Nest exposition started at the beginning of May (from the 5th tothe 18th of May), which is the main breeding season of the fourground breeding species mentioned (Bezzel, 1993). Nests werecontrolled after 4 and collected after 8 days of exposure. A nestwas considered as preyed upon only if the quail egg was damagedor lost. We decided to do this, since plasticine eggs might overes-timate the nest predation by attracting small mammals (Purgeret al., 2012). Marks left on the eggs (mostly on plasticine eggs)were assigned to four predator guilds (mammalian or avian, smallor large; Appendix A.2 in Supporting information). Missing eggswere categorized as depredated by an unknown predator.

2.4. Vegetation survey

Vegetation surveys of the understory (bushes and tree saplings)were performed during nest predation controls. We estimated thepercentage of understory vegetation for each bird survey plot insteps of five percentages from 0% to 100%. Additionally, the coverof the understory was estimated in a radius of 10 m around thenests using the same percentage classes.

2.5. Statistics

The maximum abundance out of the three bird survey roundswas calculated by taking the highest abundance value for each birdspecies in each plot for each stand. We used linear mixed effectsmodels to test the effects of particular environmental variableson bird species richness, total abundance and abundance of birdgroups (tree, shrub or ground breeder). The fixed environmentalvariables were tree diversity (beech-dominated or tree-speciesrich), distance from forest edge (continuous variable), bush per-centage per plot (continuous variable) and their two-way interac-tions. Forest stand was used as a random effect term to take intoaccount that four bird survey plots were sampled within the samestand (R command of full model as an example with all bird rich-ness: lme(species_richness�(tree_diversity + plot + bush_percent-age)^2, random = �1|stand, data = bird, method = ‘‘ML’’). To avoidheteroscedasticity and to optimize model fit, we applied differentvariance functions. Then the minimal adequate model (MAM)was selected via stepwise model selection by the second orderAkaike’s Information Criterion (AICc) using the stepAICc function

(modified by CS, see URL: http://wwwuser.gwdg.de/~cscherb1/stepAICc.txt) of the MASS package (Venables and Ripley, 2002) ofR (R Development Core Team, 2013). Final models were refittedby restricted maximum likelihood. The normality of model residu-als was assessed using normal quantile–quantile plots. Calcula-tions were made using the nlme package (Pinheiro et al., 2013)in R.

In order to test whether tree diversity (beech-dominated ortree-species rich), distance from forest edge and bush percentageper plot affected the community composition of birds, we addition-ally performed partial redundancy analyses (RDA). The species ma-trix was constrained by the predictor variables tree diversity(beech-dominated or tree-species rich), distance (factor) or bushpercentage (continuous variable). For a better characterization ofstudy plot with different distance, in the ordination analyses, dis-tance was treated as a factor. Prior to the analyses, the species ma-trix was transformed with the Hellinger-transformation (Legendreand Gallagher, 2001). This transformation allows the use of Euclid-ean-based ordination methods with community composition datacontaining many zeros, i.e. characterised by long gradients. Pseu-do-F values with the corresponding P values were calculated bypermutation tests based on 999 permutations. Calculations wereperformed using the vegan package for R (Oksanen et al., 2013).

Prior to the analysis of the nest predation data, nest days werecalculated following Mayfield (1961) and Hazler (2004). The mid-point of the two controls was assumed as time when predationevents occurred; this resulted in 2, 6 or 8 days of survival (i.e. nestdays). In addition, data were pooled for each plot for each stand.For the nest days and the predation events this was done by takingthe sum of numbers, while for bush percentage 10 m around thenests the mean number was taken. We fitted a generalized linearmixed model with binomial errors to analyse the effects of envi-ronmental variables on daily nest survival rates. The binomialdenominator consisted of the nest fate (0 = survive; 1–4 = fail)and the numerator consisted of nest days (between 8 and 32 daysof survival). We included the following fixed effects variables: treediversity (beech-dominated or tree-species rich), distance fromedge (continuous variable), bush percentage 10 m around the nest(continuous variable) and all two-way interactions between thesefactors. Forest stand was used as random factor to correct for spa-tial autocorrelation. Non-significant variables (P > 0.1) were dis-carded using a manual stepwise backward selection procedureuntil reaching the Minimal Adequate Model. Calculations wereperformed using the glmmPQL function in the MASS package forR (Venables and Ripley, 2002).

3. Results

3.1. Bird diversity

Altogether we registered 539 individuals representing 33 spe-cies during the three bird survey rounds. The results of the linearmixed model showed a higher bird abundance as well as speciesrichness at the forest edge compared to the forest interior (Table 1;Fig. 1a). In addition, bush percentage had a positive effect on birdspecies richness (Table 1; Fig. 2). Bush percentage did not changewith distance from forest edge considering all study plots (one-way ANOVA, F3,44 = 0.04; P = 0.991).

Most species were tree breeders (19) with 293 individuals, fol-lowed by shrub breeders (9/144) and ground breeders (5/102). Dis-tance from forest edge had a negative effect on bird abundances oftree breeders and shrub breeders (Table 1; Fig. 1b and c). Finally,ground breeders’ abundance increased with increasing bush per-centage (Table 1; Fig. 1d).

Table 1Results of linear mixed effects models testing the effects of tree diversity (beech-dominated vs. tree-species rich forest stands), distance from forest edge and bush cover (% bushand tree sapling cover in the 30 � 60 m bird survey plot) on bird species richness and abundance.

Variable df F P

All bird richness Tree diversity (T) – – –Distance (D) 34 34.25 <0.001Bush% (B) 34 16.33 <0.001

All bird abundance Tree diversity (T) – – –Distance (D) 35 14.12 <0.001Bush% (B) – – –

Tree breeder abundance Tree diversity (T) – – –Distance (D) 35 5.54 0.024Bush% (B) – – –

Shrub breeder abundance Tree diversity (T) – – –Distance (D) 34 16.98 <0.001Bush% (B) 34 2.95 0.095

Ground breeder abundance Tree diversity (T) 10 0.60 0.457Bush% (B) 34 4.15 0.049T � B 34 3.98 0.054

Fig. 1. Mean (±SEM) bird abundance of each plot (30 � 60 m) from forest edge to forest interior in beech-dominated and tree-species rich forest stands: (a) total birdabundance; (b) tree breeder abundance; (c) shrub breeder abundance; and (d) ground breeder abundance. Although distance of bird survey plots was used as a continuousvariable in mixed effects models, barplots are used here for a better presentation of results.

P. Batáry et al. / Forest Ecology and Management 319 (2014) 44–50 47

In the ordination analysis, distance from forest edge explained asignificant part of the variation in the bird species matrix (11.12%,pseudo-F3,42 = 1.84, P = 0.001; Fig. 3). Tree diversity and bush per-centage did not have a significant effect on bird community com-position (2.81%, pseudo- F1,42 = 1.41, P = 0.105; 2.34%, pseudo-F1,42 = 1.18, P = 0.264 respectively).

3.2. Nest predation

From the 192 artificial ground nests 70 were depredated(36.5%). We found that nest survival was significantly lower intree-species rich stands compared with beech-dominated stands(t = 2.30, P = 0.044, df = 10), whereas it decreased only marginally

Fig. 2. Relationship between bird species richness and bush percentage per birdsurvey plot. Solid line represents the regression line from the general linear mixedeffects model (Table 1).

Fig. 3. RDA ordination biplot with bird species (closed dots) and four distancesfrom forest edge (open dots): Plot 1 (0–30 m); Plot 2 (60–90 m); Plot 3 (120–150 m); Plot 4 (180–210 m). For visibility, only bird species with the highestfraction of variance fitted by the two first factorial axes are indicated (Cyacae:Cyanistes caeruleus, Denmaj: Dendrocopos major, Erirub: Erithacus rubecula, Fricoe:Fringilla coelebs, Phycol: Phylloscopus collybita, Physib: Phylloscopus sibilatrix, Phytro:Phylloscopus trochilus, Regign: Regulus ignicapillus, Sylatr: Sylvia atricapilla, Turmer:Turdus merula, Trotro: Troglodytes troglodytes).

0.75

0.80

0.85

0.90

0.95

1.00

0-30 60-90 120-150 180-210

Dai

ly n

est s

urvi

val r

ate

Distance from forest edge [m]

Beech dominated sites

Tree species rich sites

Fig. 4. Mean nest survival days with ±SEM from forest edge to forest interior inbeech-dominated and tree-species rich forest stands.

48 P. Batáry et al. / Forest Ecology and Management 319 (2014) 44–50

with increasing distance from forest edges (t = 1.80, P = 0.081,df = 35; Fig. 4). In most cases we could not identify the predators(56%), since either both eggs disappeared or egg shell remainingdid not contain any identifiable predator marks. In the remainingcases, mostly mammalian predators prey on the nests, especiallylarge mammals (27%), but also small mammals (19%). Predationby birds was quite rare (3%).

4. Discussion

4.1. Bird diversity

Our results show positive edge responses of birds with higherspecies richness and abundances (including tree and shrub breed-ers) at the forest edges compared to forest interiors, supporting ourfirst hypothesis. This positive avian edge response has often beenreported in the literature (e.g. Villard, 1998; Sisk and Battin,2002; Salek et al., 2010). The higher species abundance at edgesis probably due to the increased attractiveness of edges for for-est-edge specialists and habitat generalists (Luck et al., 1999; Ber-ry, 2001). This was also reflected in the ordination analysis. Thefirst RDA axis separated the first plot, i.e. the forest edges fromthe forest interiors (plots 2–4). At forest edge typical shrub breederspecies were present like Blackbird (Turdus merula) and EurasianBlackcap (Sylvia atricapilla), whereas in the forest interior bird spe-cies like Wren (T. troglodytes), European Robin (E. rubecula) andCommon Firecrest (Regulus ignicapillus) occurred.

An explanation for the positive edge response may be thatedges concentrated nesting and food resources due to their bushystructure, while dense bushes were generally absent or rare in theforest interiors. This could explain why shrub breeders had a high-er abundance at forest edge, while bush percentage had no effecton them. Furthermore, the drier and brighter conditions of edgesand their shelter from wind attract many insects (e.g. Scherberet al., in press) contributing to an increased abundance of insectiv-orous species, which might have found more foraging opportuni-ties than in the surrounding fields or forest interior.

Ground breeders, however, showed no edge response, but werepositively affected by bush percentage. Since bush percentage wasequally distributed along transects, this would explain why forground breeders no edge effect occurred. In general, there is anassumption that nest concealment is unlikely to reduce nest preda-tion of ground nests, since they are more likely depredated bymammals using olfactory cues (Evans, 2004). Nevertheless, bushesare an important component of forests that can enhance habitatquality for birds, for instance by sheltering their nests from envi-ronmental factors like rain, wind and heat, serving as perch sitesand providing food resources. This could also explain why bird spe-cies richness was positively affected by bush percentage (Cherka-oui et al., 2009).

P. Batáry et al. / Forest Ecology and Management 319 (2014) 44–50 49

Concerning our second hypothesis, there were no significantdifferences between beech-dominated stands and tree-species richstands regarding species richness and abundance as well as forbreeding categories. In contrast to our study, DeGraaf et al.(1998) found that forest structure was the most important factor(over forest cover-type or stand-size class) in determining the birdcommunities of US forests. In our case, however, the contrast inforest structure might have been lower between tree-species richand beech-dominated stands, which resulted in similar bird com-munity composition too.

We did not find support for our third hypothesis that edge effectis more pronounced in beech-dominated than in tree-species richstands. Although regarding bush percentage there were no differ-ences along transects as well as edge structure was similar in bothtree diversity types. Therefore the difference between edge andinterior is more or less the same and not as supposed with moreheterogeneous edges in beech-dominated stands compared to for-est interior.

Despite our extensive bird surveys, we recorded only compara-tively few bird species. A possible reason may be that the land-scape-scale amount of suitable habitat for many forest specialistspecies is probably lower than the minimum threshold for persis-tence (Angelstam et al., 2004), although we studied the largestdeciduous forest area (16.000 ha) in Germany. Additionally, theforest of the Hainich National Park has not been managed duringthe last 50 years, but the current forest composition and structureis still determined by the former management actions. Therefore,our discussion and conclusion is rather valid for those typicaland widespread European situations, where forests bear only sim-plified structure and impoverished composition (but see We-sołowski and Tomiałojc, 2005).

4.2. Nest predation experiment

In the second part of our study we investigated the effects ofdistance to forest edges and tree species diversity on daily nest sur-vival rates of artificial ground nests. In contrast to our hypothesiswe did not find an elevated nest predation closer to the forestedges. This could be due to the compensating influence of the largeforest area, which can reduce the edge effect on avian nest preda-tion. Keyser et al. (1998) found that nest predation strongly in-creased with reduced fragment size, but without additionalimportance of edge effects. Since our study was conducted in onecontinuous large forest area lower amount of edge zones are pres-ent, mitigating edge effects. Moreover, mammalian predators fromadjoining fields penetrate in forest edges, which dominate smallfragments, searching for food, while in large forests with an intact‘‘core’’ area, interior forest predators avoid edges affecting only for-est interiors. Predation events by forest interior predators and ma-trix predators may not overlap at forest edge (that would lead toincreased ground nest failure), but interior as well as edge preda-tion risks can be similar (Stephens et al., 2004). However, to testthat at forest edge different nest predators occur compared to for-est interior, further experiments need to be conducted identifyingnest predators via video recording or similar tracking tools (Ste-phens et al., 2004; Weatherhead and Blouin-Demers, 2004).

In addition, the Hainich National Park provides great alternativefood supply along the whole studied transect for large mammalianpredators, like high abundance of mice and voles for red fox (Vulpesvulpes), mustelids or the alien raccoon (Procyon lotor) as well asseeds and nuts for red squirrel (Sciurus vulgaris), which could re-duce predation pressure on ground breeders as well as in edges.Also the abundant wild boars (Sus scrofa) that occur everywherefrom forest interior to forest edge might have contributed to miss-ing edge effects. While rummaging with their snouts through the

litter layer searching for food, they are destroying ground breedingnests.

We confirmed the hypothesis that predation rates are higher intree-species rich stands compared to beech-dominated stands.Higher nest predation rates at tree-species rich stands were prob-ably due to greater diversity and density of mammalian predators.Since several nests were depredated by large mammals (mostprobably by wild boars), we can suppose that the many missedeggs are also attributable to large mammalian predators in the Hai-nich National Park.

5. Conclusions

Bird diversity and avian egg predation were affected by bothforest edges and tree diversity, but in very different ways. Edge ef-fect shaped the distribution of bird community composition, andpositively affected abundances of tree and shrub breeder birds,however, did not affect the ground breeders and also the nest pre-dation of ground nests. Shrub breeders accumulating in forestedges might, however, suffer more from nest predation.

Birds were attracted by forest edges with a higher abundance inthe first 30 m of the forest, so that large forest areas should be ofinterest for conservation of forest-interior and edge-sensitive birds,which need an intact ‘‘core’’ zone. In addition, bird diversity, in par-ticular abundance of ground breeding birds, was enhanced by hab-itat structure with high bush percentage that provides enhancedfood supply, perches as well as sheltering. Correspondingly, con-servation plans focusing on large unfragmented habitats with agreat horizontal forest structure should have in general positive ef-fects on bird populations like represented in the Hainich NationalPark.

Furthermore, results of our nest predation experiment showedno edge effect, leading to the assumption that forest size is animportant determining factor. In large forests the amount of edgesis relatively low and influences on edges from the centre is high,therefore edge effects appeared to be weaker compared to smallfragments with the same total amount of forest. Additionally wefound that in our large temperate forest large mammals, such aswild boars appeared to be main predators of ground nest with min-or impact of nest concealment on nest success. To prove this di-rectly, camera studies would be needed on real nests, whichwould enable a conservation plan for controlling the wild boarpopulation within the national park.

Acknowledgements

This study was supported by the DFG (German Research Foun-dation) within the framework of the Research Training GroupGraduiertenkolleg 1086. During the preparation of the paper PBwas supported by the DFG (BA 4438/1-1). Finally, we are gratefulto the staff of the Hainich National Park, who gave valuable help.

Appendix A. Supplementary materials

Supplementary data associated with this article can be found, inthe online version, at http://dx.doi.org/10.1016/j.for-eco.2014.02.004. These data include Google maps of the mostimportant areas described in this article.

References

Angelstam, P., Roberge, J.-M., Lõhmus, A., Bergmanis, M., Brazaitis, G., Dönz-Breuss,M., Edenius, L., Kosinski, Z., Kurlavicius, P., Larmanis, V., L�ukins, M., Mikusinski,G., Racinskis, E., Strazds, M., Tryjanowski, P., 2004. Habitat suitability indexmodelling as a conservation tool – a review of habitat parameters for forestbirds in the Baltic Sea region. Ecol. Bull. 51, 427–453.

50 P. Batáry et al. / Forest Ecology and Management 319 (2014) 44–50

Balaz, M., Balazova, M., 2012. Diversity and abundance of bird communities in threemountain forest stands: effect of the habitat heterogeneity. Pol. J. Ecol. 60, 629–634.

Batáry, P., Báldi, A., 2004. Evidence of an edge effect on avian nest success. Conserv.Biol. 18, 389–400.

Batáry, P., Báldi, A., 2005. Factors affecting the survival of real and artificial GreatReed Warbler’s nests. Biologia 60, 215–219.

Berg, A., 1996. Predation on artificial, solitary and aggregated wader nests onfarmland. Oecologia 107, 343–346.

Berry, L., 2001. Edge effects on the distribution and abundance of birds in a southernVictorian forest. Wildlife Res. 28, 239–245.

Bezzel, E., 1993. Kompendium der Vögel Mitteleuropas – Passeres Singvögel, firsted. AULA-Verlag, Wiesbaden.

Cherkaoui, I., Selmi, S., Boukhriss, J., Hamid, R.-I., Mohammed, D., 2009. Factorsaffecting bird richness in a fragmented cork oak forest in Morocco. Acta. Oecol.35, 197–205.

DeGraaf, R.M., Hestbeck, J.B., Yamaski, M., 1998. Associations between breeding birdabundance and stand structure in the White Mountains, New Hampshire andMaine, USA. Forest Ecol. Manage. 103, 217–233.

Evans, K.L., 2004. The potential for interactions between predation and habitatchange to cause population declines of farmland birds. Ibis 146, 1–13.

Fahrig, L., 2003. Effects of habitat fragmentation on biodiversity. Annu. Rev. Ecol.Evol. Syst. 34, 487–515.

Fretwell, S.D., 1972. Populations in a Seasonal Environment, first ed. PrincetonUniversity Press, Princeton.

Gates, J.E., Gysel, L.W., 1978. Avian nest dispersion and fledgling success in field-forest ecotones. Ecology 59, 871–883.

Hazler, K.R., 2004. Mayfield logistic regression: a practical approach for analysis ofnest survival. Auk 121, 707–716.

Helzer, C.J., Jelinski, D.E., 1999. The relative importance of patch area and perimeter-area ratio to grassland breeding birds. Ecol. Appl. 9, 1448–1458.

Johnston, V.R., 1947. Breeding birds of the forest edge in Illinois. Condor 49, 45–53.Keyser, A.J., Hill, G.E., Soehren, E.C., 1998. Effects of forest fragment size, nest

density and proximity to edge on the risk of predation to ground-nestingpasserine birds. Conserv. Biol. 12, 986–994.

Knohl, A., Søe, A.R.B., Kutsch, W.L., Göckede, M., Buchmann, N., 2008. Representativeestimation of soil and ecosystem respiration in an old beech forest. Plant Soil302, 189–202.

Laurence, W.F., 2000. Do edge effects occur over large spatial scales? Trends Ecol.Evol. 15, 134–135.

Legendre, P., Gallagher, E., 2001. Ecologically meaningful transformations forordination of species data. Oecologia 129, 271–280.

Luck, G.W., Possingham, H.P., Paton, D.C., 1999. Bird responses at inherent andinduced edges in the Murray Mallee, South Australia. 1. Differences inabundance and diversity. Emu 99, 157–169.

Major, R., Kendal, D., 1996. The contribution of artificial nest experiments tounderstanding avian reproductive success: a review of methods andconclusions. Ibis 138, 298–307.

Mayfield, H., 1961. Nesting success calculated from exposure. Wilson Bull. 73, 255–261.

Mölder, A., Bernhardt-Römermann, M., Schmidt, W., 2006. Forest ecosystemresearch in Hainich National Park (Thuringia): first results on flora andvegetation in stands with contrasting tree species diversity. Waldökologie 3,83–99.

Murcia, C., 1995. Edge effects in fragmented forests: implications for conservation.Trends Ecol. Evol. 10, 58–62.

Oksanen, J., Blanchet, G.F., Kindt, R., Legendre, P., Minchin, P.R., O’Hara, R.B.,Simpson, G.L., Sólymos, P., Stevens, M.H.H., Wagner, H., 2013. vegan:Community Ecology Package. R package version 2.0-9.

Pinheiro, J., Bates, D., DebRoy, S., Sarkar, D., the R Development Core Team, 2013.nlme: Linear and nonlinear mixed effects models. R package version 3.1-111.

Purger, J.J., Kurucz, K., Csuka, S., Batáry, P., 2012. Do different plasticine eggs inartificial ground nests influence nest survival? Acta Zool. Acad. Sci. H. 58, 369–378.

R Development Core Team, 2013. R version 3.0.1: A Language and Environment forStatistical Computing. R Foundation for Statistical Computing, Vienna, Austria.

Robinson, S.K., Thompson III, F.R., Donovan, T.M., Whitehead, D.R., Faaborg, J., 1995.Regional forest fragmentation and the nesting success of migratory birds.Science 267, 1987–1990.

Salek, M., Svobodová, J., Zasadil, P., 2010. Edge effect of low-traffic forest roads onbird communities in secondary production forests in central Europe. LandscapeEcol. 25, 1113–1124.

Scherber, C., Vockenhuber, E., Stark, A., Meyer, H., Tscharntke, T. in press. Effects oftree and herb biodiversity on Diptera, a hyperdiverse insect order. Oecologia,(doi: http://dx.doi.org/10.1007/s00442-013-2865-7).

Schmidt, I., Leuschner, C., Mölder, A., Schmidt, W., 2009. Structure and speciescomposition of the seed bank in monospecific and tree-species rich temperatebroad-leaved forests. Forest Ecol. Manage. 257, 695–702.

Sisk, T.D., Battin, J., 2002. Habitat edges and avian ecology: geographic patterns andinsights for western landscapes. Stud. Avian Biol. 25, 30–48.

Sobek, S., Scherber, C., Steffan-Dewenter, I., Tscharntke, T., 2009. Sapling herbivory,invertebrate herbivores and predators across a natural tree diversity gradient inGermany’s largest connected deciduous forest. Oecologia 160, 279–288.

Stephens, S.E., Koons, D.N., Rotella, J.J., Willey, D.W., 2004. Effects of habitatfragmentation on avian nesting success: a review of the evidence at multiplespatial scales. Biol. Conserv. 115, 101–110.

Svobodová, J., Koubová, M., Mrštny, L., Albrecht, T., Kreisinger, J., 2012. Temporalvariation in nest predation risk along habitat edges between grassland andsecondary forest in Central Europe. Eur. J. Wildl. Res. 58, 315–323.

Tryjanowski, P., Kuzniak, S., Diehl, B., 2000. Breeding success of the Red-backedShrike (Lanius collurio) in relation to nest site. Ornis Fenn. 77, 137–141.

Venables, W.N., Ripley, R.D., 2002. Modern Applied Statistics with S. Springer, NewYork.

Villard, M.A., 1998. On forest-interior species, edge avoidance, area sensitivity, anddogmas in avian conservation. Auk 115, 801–805.

Vivian-Smith, G., 1997. Microtopographic heterogeneity and floristic diversity inexperimental wetland communities. J. Ecol. 85, 71–82.

Vockenhuber, E.A., Scherber, C., Langenbruch, C., Meissner, M., Seidel, D.,Tscharntke, T., 2011. Tree diversity and environmental context predict herbspecies richness and cover in Germany’s largest connected deciduous forest.Perspect. Plant Ecol. Evol. Syst. 13, 111–119.

Weatherhead, P.J., Blouin-Demers, G., 2004. Understanding avian predation: whyornithologists should study snakes. J. Avian Biol. 35, 185–190.

Wesołowski, T., Tomiałojc, L., 2005. Nest sites, nest depredation, and productivity ofavian broods in a primeval temperate forest: do the generalisations hold? J.Avian Biol. 36, 361–367.

Wilcove, D.S., McLellan, C.H., Dobson, A.P., 1986. Habitat fragmentation in thetemperate zone. In: Soulé, M.E. (Ed.), Conservation Biology: The Science ofScarcity and Diversity. Sinauer Associates, Sunderland, pp. 237–256.

Wilson, G.R., Brittingham, M.C., Goodrich, L.J., 1998. How well do artificial nestsestimate success of real nests? Condor 100, 357–364.