Environmental Management

, Volume 51, Issue 3, pp 750–758

Effects of Logged and Unlogged Forest Patches on Avifaunal Diversity

Authors

    • Department of Environmental Sciences, Faculty of Natural ResourcesUniversity of Tehran
  • Mohammad Kaboli
    • Department of Environmental Sciences, Faculty of Natural ResourcesUniversity of Tehran
  • Mahmoud Karami
    • Department of Environmental Sciences, Faculty of Natural ResourcesUniversity of Tehran
  • Vahid Etemad
    • Department of Environmental Sciences, Faculty of Natural ResourcesUniversity of Tehran
  • Saeedeh Baniasadi
    • Department of Environmental Sciences, Faculty of Natural ResourcesUniversity of Tehran
Article

DOI: 10.1007/s00267-012-9979-2

Cite this article as:
Ghadiri Khanaposhtani, M., Kaboli, M., Karami, M. et al. Environmental Management (2013) 51: 750. doi:10.1007/s00267-012-9979-2

Abstract

In the Hyrcanian forests of northern Iran, reduced-impact silviculture systems, (single-tree and group-tree selection) were applied over a large area, which generated different local habitat structures. The aim of this study was to assess the differences between treated and untreated areas of forest and their effect on avian richness, abundance and diversity (R.A.D). Birds were surveyed during the breeding season in 2009 by 100-point counts, equally distributed in the treated and untreated area. Avian R.A.D was significantly different and higher in the untreated area. Generally, forestry practices cause noticeable changes in canopy percentage, tree composition, snags and shrub number. Treated forest habitats in the area of study had a much more developed understory, fewer snags and fewer large diameter trees. The results highlighted the importance of forest maturity and showed that preventing silvicultural disturbances may not be the best solution for conserving and enhancing biodiversity. Rather, methods such as selective cutting seem an appropriate and sustainable way of forest management. It is suggested that forests should be managed to conserve structural elements which create favorable habitat for bird species, preventing future species losses due to logging practices.

Keywords

Forest structureBird communityRichnessAbundanceDiversitySilvicultural treatment

Introduction

Human exploitation of forest ecosystems has greatly affected forest structure and landscape in many regions of the world. Several studies have shown that changes in vegetation structure and composition following timber harvesting in forest habitats can result in dramatic changes in composition of avian communities and in the abundance of many bird species (Thompson and others 1999; Augenfeld and others 2008; Tozer and others 2010). The main effects of regular wood extraction include habitat fragmentation, stand-age gradients, changes in tree species composition, foliage height profile, decrease in large canopy trees, and changes in the amount of snags and downed woody debris. All the above factors can influence bird communities to a large extent (MacArthur and MacArthur 1961; Laiolo and others 2003). Thus, for the purpose of forest management and avian conservation, the identification of critical structural elements and their relationships to bird species richness and abundance is essential (Diaz and others 2005; Ghadiri and others 2012a).

Hyrcanian deciduous forest (Caspian forest) covers a narrow strip along the south of the Caspian Sea. Due to its geographic isolation, these forests are composed of highly important fauna and flora. The Hyrcanian forest is the only remaining part of broad leaved deciduous forests belonging to the third geological era, the Tertiary period (Salehi Shanjani and others 2002). Today, protection of this unique forest ecosystem is essential not only for conservation purposes such as preserving biodiversity, adapting to expected climate changes, and avoiding desertification, but also for the purpose of sustainable forest management.

Recently, low-impact forest management methods, including single and group tree selection and small clearcuts have been proposed as alternatives to large clearcuts in this area (Etemad 1994). Since the timber harvesting methods diverse in magnitude, intensity, frequency and pattern of disturbance, so habitat structures alteration are also different under different kind of these logging methods. Previously, there had not been a comprehensive study on the effect of alternative silvicultural methods on avian community in this area.

In this study, we determined how the structural changes in forests, resulting from logging, affect local species richness, abundance and diversity (R.A.D) of forest birds. Our goals included: (i) quantifying the difference in structure and floristic composition between logged and unlogged forests and (ii) elucidating the relationships between particular structural components of forests that create bird habitat (e.g., understory, large trees, snags, logs) and bird species abundance. The results could be used to identify critical relationships between forest structure and avian composition.

Method

Study Area

Field work was conducted in the Kheyrud Forest Research Station of University of Tehran (KFR), an 8,000 hectare forest which is a portion of Caspian Hyrcanian mixed Forest, located approximately 7 km east of Nowshahr, Mazandaran province (36°40′–36°27′ N, 51°43′–51°22′ E), within the Alborz mountain range, northern Iran (Fig. 1). The elevation in the region ranges from 50 to 2,200 m above sea level, composed of mostly broadleaved woods. The annual amount of precipitation and average temperature is 1,300 mm and 17 °C, respectively, and the climate of the area, according to the Emberge climate system, is wet with cold winters (Etemad 2002).
https://static-content.springer.com/image/art%3A10.1007%2Fs00267-012-9979-2/MediaObjects/267_2012_9979_Fig1_HTML.gif
Fig. 1

Study site location within the Forest Research Station of University of Tehran

Dominant plant communities include Parrotio-Carpinetum, Parrotio-Buxetum, Tilio-Buxcetum, Querco-Carpinetum, Fageto-Carpinetum, Fagetum Type and Fagetum mixed (Etemad 1994; Sarmadiyan and Jafari 2001). KFR is divided into 7 districts; however, only the second district, Namkhane, was surveyed. In order to minimize the forest types/structural variation on dependent variables (i.e., avian richness, abundance and diversity), the sampling attempts were restricted to Namkhane with Fagetum and Fageto-Carpinetum communities, comprising over 992 ha and ranging from 700 to 1,500 m a.s.l. Shariati (2009) did not find any significant difference in birds’ richness and abundance in this zone. The stands of Namkhane district are dominated by Hornbeam (Carpinus betulus) and Beech (Fagus orientalis), and due to the timber harvesting the number of hornbeam has increased (Etemad 1994; Etemad 2002).

Namkhane district is divided into 27 units, five of which weren’t surveyed due to inaccessibility caused by the steep slope. The average area of these units ranges from 22.0 to 71.9 ha. Between 1995 and 2009, parts of this district were treated using the following silvicultural systems: clear cut in small areas (0.5–1 ha), retention cut, and mostly single and group tree selection. These are referred to as “treated area”. The other parts of Namkhane forest, which served as “untreated forest,” has never been harvested and is representative of the regional forest vegetation (Fig. 1).

Sampling

Bird Census

Bird surveys were conducted during the breeding season, April to May 2009, by skilled observers from dawn to approximately 10:00 AM using point counts (as suggested by Johnson 2007). In each point count a single visit was done. Unlimited distance count was chosen because of difficulty of estimating distance in a forest with diverse vegetation structure (Tozer and others 2010). The survey was stopped during rain or strong wind because of reduction in detectability (Bibby and others 1992; Mitchell and others 2001). At each point, all bird species heard and/or seen during a 10-min period were recorded. Overflying birds were recorded but not included in the analysis (Laiolo 2002).

Sampling points were at least 250 m apart to minimize the probability of sampling the same bird more than once (Laiolo 2002; Diaz 2006) and to avoid spatial autocorrelation.

Habitat Variable

Previous studies of bird distributions in KHR forests (Shariati 2009) were used to identify potential structural elements or properties of forests that could be of importance for bird habitat. In the fall of 2008, the treated and untreated area of Namkhane district was precisely mapped. Sampling plots were located randomly across the two types, but kept at least 250 m apart. Thus, a total of 100 point counts were obtained: 50 in the untreated forest and 50 in the treated part. When the study commenced, treatments had been applied for 14 years (including the year of study).

A 17.8 m radius plot was placed within the center of each census point to sample vegetation structure and environmental variables. The structure of vegetation was quantified visually. In each plot, dominant factors of vegetation structure were the tree species and the diameter at breast height (DBH-M) of all living trees (more than 5 cm), and the availability of large trees (larger than 30 cm DBH) (e.g., Diaz 2006). Other factors representing vegetation structure included number of canopy layers (canopy N), tree number (tree N), shrub number (shrub N), percentage of upper and lower shrub layer (2.5–7 and 1–2.5 m, respectively), percentage of herb layer (≤1 m), medium tree height (height-M), tree volume (T-volume), and tree number in different DBH classes (5–9, 10–29, and 30–59 cm, DBH >60 cm). Dead tree number (dead tree N) was recorded and for all woody debris, including snags and downed woody debris (DWD) with diameter >5 cm in the plot. The DBH and the length of entire woody debris were noted.

Mean wood volume was obtained using Kuchler (1967) equation number 1:
$$ V = \frac{{d^{2} \times \pi }}{4} \times h \times 0.9 $$
(1)
where V is the volume, d is tree DBH and h is the tree height.
The complexity of the vertical distribution of foliage weight was measured with tree height distribution (THD) which was divided into three layers (<10 m, 10–20 m, >20 m) to characterize the vertical structure of the forest (Kuuluvainen and others 1996). In equation number two THD was quantified with the Shannon_wiener formula:
$$ {\text{THD}} = H = \sum\limits_{i = 1}^{n} {{\rm P}i} \log_{e} {\rm P}i $$
(2)
where Pi is the proportion of trees in the ith height layer (Kuuluvainen and others 1996).

Data Analysis

Through data analysis different factors related to avian community were measured and some appropriate analyses for comparing different habitat structure elements were used.

Spatial Autocorrelation

At first the spatial autocorrelation was investigated in the original data (avifaunal richness, abundance and composition). Spatial correlograms were constructed using Moran’s I spatial autocorrelation coefficient (Moran 1950; Legendre and Legendre 1998) at 10 distance class. The significance of correlogram was considered after Bonferroni correction for multiple tests.

Birds R.A.D

Bird richness (R) was counted as the number of different bird species in every point count, and bird abundance (A) was measured as the number of all individuals found per point count.

Bird diversity (D) per plot was calculated using Shannon and Wiener (1949) index in equation number 3:
$$ H = \sum\limits_{i = 1}^{s} {\;{\rm P}i} \log_{2} {\rm P}_{i} $$
(3)
where H shows index of species diversity, s number of species and Pi shows the proportion of total sample belonging to the ith species.

T Test

Differences in species R.A.D and differences in habitat structure among the two forest stands were compared for each habitat variable using t test, but prior to that, normality and homoscedasticity of variables were explored by testing for Kolmogrov-Smirnov and Levene test, respectively. Non-normal data were transformed into log-transformed data (log(x + 1)).

Discriminate Function Analysis (DFA)

Discriminate function analysis (DFA), is a multivariate analysis, was used to predict membership in naturally occurring groups. A general view of forest characteristic variable, as independent variables in treated and untreated stand, were used in DFA to show the placement of all variables toward each other. In DFA, the independent variables were the predictors and the dependent variables, coded 1 (as a symbol of treated area) and 2 (untreated area), were the groups.

General Linear Models (GLMs)

General linear models (GLMs) are used frequently to identify the main habitat factors that explain the habitat selection of a species in a certain area (Buckland and Elston 1993). Habitat selection among avifauna operates through a series of behavioural decisions at several spatial scales, which explains why studying their distributional pattern is difficult. Indeed habitat selection is a hierarchical process in which individuals first select the general area in which to live, then within this area they select among the available patches, finally choosing a nest site (Morrison and others 1992; Pribil and Picman 1997).

Before analysis of DFA and GLM, Pearson correlation coefficient was used and some correlated variables like tree volume, which was correlated with both tree DBH and tree height, were omitted.

Multivariate analysis was undertaken using ADE4 (Laiolo and others 2003; Kaboli and others 2006) and t tests and other related analysis were conducted using SPSS 17.0 software. GLM was conducted by R 2.8.1.

Results

Spatial Autocorrelation

The final correlograms of spatial autocorrelation showed that response variables in connected sampling units weren’t similar for randomly associated pairs of units. Therefore, there was no need for constructing an autocorrelation term (Autocor) which expresses the potential influence of neighboring sampling units (Fig. 2).
https://static-content.springer.com/image/art%3A10.1007%2Fs00267-012-9979-2/MediaObjects/267_2012_9979_Fig2_HTML.gif
Fig. 2

Spatial correlations for response variables [Richness (a), Abundance (b) and the first axis of bird gradient (c)] white circles represent insignificant Moran’s I values after Bonferroni correlation with 250 m distance

Bird Diversity

During the survey, 21 bird species were identified in 100 census plots. The same species were recorded in the plots of both stands and are listed in Appendix A. Common Buzzard (Buteo buteo) was not involved in analysis.

Results of t tests showed that bird species R.A.D were significantly different and were higher in untreated stands than in treated stands (p ≤ 0.05; Table 1). Figure 3 shows the mean abundance and the standard error of birds between treated and untreated sites in study area.
Table 1

Mean, SD and sample size of bird species R.A.D in treated and untreated stands with significant differences tested by means of t test

 

Treated stand

Untreated stand

t

P value

Mean

SD

N

Mean

SD

n

Species richness

3.09

0.03

49

4.69

0.03

49

3.47

<0.001

Species abundance

11.06

0.29

49

14.52

0.40

49

2.90

<0.001

Species diversity

1.59

0.05

49

2.96

0.30

49

5.52

0.04

https://static-content.springer.com/image/art%3A10.1007%2Fs00267-012-9979-2/MediaObjects/267_2012_9979_Fig3_HTML.gif
Fig. 3

The mean abundance and the standard error of birds in study area between treated and untreated forest area

Stand Structure

Forest stands differed markedly in structural components according to the result of t tests (Tables 2, 3). Forest treatments appear to have contributed to lower vegetation cover and a reduced canopy layer, as is apparent from the results (Table 2) which indicate denser vegetation and canopy in untreated stands. Harvest effect was also reflected in a reduced basal area and tree number (Table 3). Treated forest canopy was less dense, with cut trees creating canopy openings which facilitated the proliferation of understory layers. Moreover, shrub number and upper shrub cover were considerably higher in the treated stands (Tables 2, 3).
Table 2

Mean, SD and sample size of important attributes of vegetation layer in treated and untreated forests in Kheyrud forest with significant differences tested by means of t tests

Variables

Treated stand

Untreated stand

t

P value

Mean

SD

n

Mean

SD

N

Total vegetation cover

102.92

4.16

49

118.09

4.17

49

2.57

0.01

Canopy layer

50.08

2.27

49

68.14

2.28

49

5.66

0.01

Upper shrub layer

3.66

0.35

49

2.59

0.36

49

0.54

0.04

Low shrub layer

3.01

0.24

49

2.78

0.27

49

0.62

ns

Herb layer

27.56

2.17

49

26.29

2.33

49

0.35

ns

Medium tree height

18.72

3.07

49

20.25

2.65

49

0.37

<0.001

ns Not significant

Table 3

Mean, SD and sample size of tree variables in treated and untreated forests in Kheyrud forest with significant differences tested by means of t tests

Variables

Treated stand

Untreated stand

t

P value

Mean

SD

N

Mean

SD

N

Tree number

39.06

2.31

49

48.07

2.52

49

2.63

0.01

Number of large treesa

21.16

1.38

49

26.40

1.26

49

2.79

<0.001

Mean DBH (cm)

0.36

0.01

49

0.38

0.11

49

1.35

ns

Shrub number

77.56

7.89

49

58.32

6.13

49

2.00

0.04

Dead tree number

4.64

0.78

49

7.54

0.70

49

2.58

0.01

Snag number

4.32

0.42

49

6.66

0.37

49

4.17

0.01

DWD number

2.60

0.32

49

3.10

0.30

49

1.35

ns

THDb

18.87

0.49

49

18.16

0.35

49

1.14

ns

Tree species richness

3.92

1.84

49

3.62

1.24

49

0.92

ns

ns Not significant

aTrees with DBH >30 cm

bTree height distribution

Eleven forest structure variables (canopy N, tree N, shrub N, height-M, DBH-M, dead tree N, Tru5-9, Tru10-29, Tru30-59, Tru >60, percentage of herb layer), were applied for DFA and GLM. The result of DFA analysis shows that treated and untreated forests partly overlap. This is largely due to the type of harvesting which was mostly single and group selection that has less effect on the forest compared to other older methods like clearcutting. Polygons in the untreated area had more trees and taller trees with larger DBH, characteristics more associated with late successional or mature forests. On the other hand, the treated area polygon was more related to shrub number and small young trees which are characteristics of early successional stands (Fig. 4).
https://static-content.springer.com/image/art%3A10.1007%2Fs00267-012-9979-2/MediaObjects/267_2012_9979_Fig4_HTML.gif
Fig. 4

Location of point counts, treated points (line), and untreated points (bold line) in two first planes of the DFA of 11 habitat variables measured in the field. Each convex polygon corresponds to the scatter of point-census in which the habitat variables were recorded (See “Habitat variable” sub-section for abbreviation)

Forest structure variables were considered in GLM analysis together with birds’ R.A.D. Table 4 shows the results of a linear model identifying the habitat attributes that better explain the variations in bird R.A.D. Variables most influencing bird species richness were tree number and number of tree with DBH 10–29 and 30–59. Bird abundance was mainly determined by the increase in trees’ DBH and number of trees with DBH classes 10–29 and >60 cm. Bird species diversity is mostly correlated with number of dead trees, tree number, and number of trees with DBH classes 10–29 and 30–59 cm.
Table 4

Model selection results for predicting the relation among dependence variables and independence variables in untreated stand

Dependence variable

df

R2

Model

AICc

P level

  

Abundance

49

0.12

DBH-M+ Tru10- 29 + Tru >60

245.72

0.02

  

Richness

49

0.24

Tree N+ Tru 10-29 + tru 30-59

171.20

0.01

  

Diversity

49

0.34

Tree N+ Dead tree N+ Tree 10-29+ Tree 30-59

107.33

0.00

  

Discussion

During this study, the effect of forest logging on avian communities in Hyrcanian forest, northern Iran, was surveyed. There was a significant difference in birds’ R.A.D between treated and untreated sites and R.A.D was higher in untreated sites.

Although the forest management was short term, there were noticeable differences in vegetation structure as well as differences in avian community factors. In the study area forestry practices have caused significant changes in canopy percentage, tree composition, lower vegetation layer covers, shrub number and dead wood. All of these changes can influence biological diversity and have been recognized as indicators of sustainable forestry programs. As with other studies (Laiolo and others 2003; Doyon and others 2005; Diaz 2006) alteration of vegetation characteristic due to harvesting is an inevitable fact.

Significant differences in bird species richness and abundance were reported in a study by Diaz (2006) in forests in the southern part of the Sierra de Guadarrama, and also by Vergara and Schlatter (2006), in mature Nothofagus forests on Tierra del Fuego Island. However, our result contradict the findings of Kilgo (2005) and Zurita and Zuleta (2008) that birds abundance difference were insignificant between logged and unlogged area.

Effects of Forest Structure on Avian Communities

Forest structure is known as a determinant factor for avian richness and complexity (Doyon and others 2005; Vergara and Schlatter 2006; Ghadiri and others 2012a) and alteration in vegetation structure complexity and floristic composition are quite often related to changes in associated bird communities (see Shochat and others 2001; Laiolo 2002; Machtans and Latour 2003; Diaz 2006).

Intact natural forests are indeed more complex and diverse with dense canopies and greater density of trees than treated forests (Kuuluvainen and others 1996).

In the study area, reduction in tree layers through tree harvesting allowed more light to reach the forest floor, which resembles early successional stages of forest habitats and can result in a reduction of the number of birds such as Black Woodpecker (Dryocopus martius) (Ghadiri and others 2012b), Green Woodpecker (Picus viridis) and Nuthatch (Sitta europaea) which depend on more complex environments (Holmes and Pitt 2007; Zurita and Zuleta 2008). However, the proliferation of understory layers, particularly the upper and lower shrub layers, which was the result of canopy opening are important variables for Syrian Woodpecker (Dendrocopos syriacus), Red-breasted Flycatcher (Ficedula parva) and Chaffinch (Fringilla coelebs), whose numbers increased in the treated stands (Doyon and others 2005; Holmes and Pitt 2007).

Effects of Tree Species on Avian Communities

Change in tree composition is one outcome of forest harvesting. Beech and Hornbeam were the commoner tree species in the study area. Forestry practices carried out in the area promoted noticeable changes in tree species composition leading to a domination by Hornbeam, which is an opportunist species (t = 2.17, p = 0.02). This is contradictory to the management goal of decreasing the number of opportunist species, including Hornbeam and Row wood (Parrotia persica). Harvesting should ideally increase the chance of growing valuable species such as Beech. Perhaps this will be addressed in future managerial decisions.

There was no significant difference in tree height diversity (THD) which is an indication of vertical forest structure. However tree number, especially large trees, were significantly higher in the untreated areathan in the treated area. The significance of large trees is that they benefit birds by offering nesting sites for cavity nesters such as Black Woodpecker and Green Woodpecker, and may support abundant resources like arthropods in bark and dead woody tissues for species like Wren (Troglodytes troglodytes) and Gold crest (Regulus regulus) as well (Nadkarni and Matelson 1989; Sillett 1994). In treated forests, these important resources are diminished and this could explain the low abundance of birds reported. This result resembles those previously report-ed in forest bird communities, where tree characteristics affect bird richness and abundance (Diaz 2006; Thinh 2006 and references therein).

Many studies in European forests have shown that bird species richness, density, and biodiversity increase with forest age. In fact, stand-age is positively correlated with tree volume and in turn, with productivity of forest areas; thus, its effect on bird communities (especially on interior species) can be very strong. This can be a subject for further investigation in future research.

Dead Wood

Dead wood, which is divided into snags and downed woody debris, is a good resource for many species as a breeding, roosting and foraging habitat. The first group in the study area to utilize deadwood, especially snags with more than 25 cm DBH, are woodpeckers, especially Black Woodpecker and Green Woodpecker as primary cavity nesters, followed by the Nuthatch as a weak primary cavity nester and finally followed by secondary cavity nesters like Coal Tit (Parus ater). Deadwood resources were significantly more abundant in untreated stands than treated stands (also reported in a study by Lohr and others 2002; Tozer and others 2010).

Implications for Conservation or Management

The results highlight the importance of mature large trees. These trees provide more food availability for avifauna than younger trees, as well as more breeding sites for birds nesting in tree cavities. Since mature forests have more diverse strata than younger ones, they enhance bird communities in many forest types (as suggested by Shochat and others 2001; Laiolo and others 2003).

The present study supports this view, as characteristics typical of mature forests, including height of trees, average diameter of the thickest trunks and number of trees of thick or medium trunk, favor not all but most of bird species in the native community.

Conclusion

It is concluded that alterations in forest elements and composition modifies bird use of the space. In this study, noticeable differences in habitat structure were observed between the treated and untreated stands in terms of lower vegetation layer cover, tree characteristics, and availability of coarse woody debris. Forest structure was the most important factor for predicting avian assemblages and it can be easily affected by different silvicultural systems. Several studies have revealed that where species diversity is reduced, the remaining species will spread out their foraging niches and reduce their overlap; however, our data collection was not designed to study such effect. Future research should consider more details which further explain the change in pattern of bird distribution.

Acknowledgments

The authors wish to thank A. Ebrahimpour and M. Shariati for their help with bird surveys, B. Hosseinzade and M. Tohidifar for their constructive comments and also I. Beheshti Tabar and A.C. Miller for their help in preparing the final version of the manuscript. The research was supported by the Department of Environmental Sciences, Tehran University, Iran.

Copyright information

© Springer Science+Business Media New York 2012