Assessing the diversity and density of birds at pine forest in tam dao national park

MINISTRY OF AGRICULTURE AND RURAL DEVELOPMENT
VIETNAM NATIONAL FORESTRY UNIVERSITY

STUDENT THESIS
Title
ASSESSING THE DIVERSITY AND DENSITY OF BIRDS AT PINE
FOREST IN TAM DAO NATIONAL PARK

Major: Natural Resources Management
Code: D850101
Faculty: Forest Resources and Environmental Management

Student: Le Thi Hoa

Student ID: 1253090010

Class: K57 Natural Resources Management

Course: 2012 - 2016

Advanced Education Program
Developed in collaboration with Colorado State University, USA

Supervisor: Assoc. Prof. Dr. Vu Tien Thinh

HaNoi,November/2016


ACKNOWLEDGEMENTS
This thesis was impossible without the support of various professors at Forestry
University and Colorado State University.

I am grateful to my advisor Assoc. Professor Vu Tien Thinh for giving me a
constant support and enthusiasm guidance during the time of research and writing of this
thesis.
Thanks also go to Professor MacDonald for his support towards my thesis.
Finally, I would like to thank staffs in Tam Dao National Park and local people
who provide useful information and kind support.
Ha Noi, November 2016


ABSTRACT
The density and diversity of wild animal especially bird species always change, it
is necessary to assess the number and density of these species. Wildlife density estimation
is an important field that useful for managing and conserving biodiversity. However, there
is not much research on wildlife density at current time. In field survey, not all individuals
are detected, especially the one far from transect. Therefore detection probability should be
concerned before a survey is conducted. In this study, density of bird species at pine forest
in Tam Dao national park was estimated using DISTANCE program. The survey was
conducted at pine forest in Tam Dao National Park (TDNP) from July 21 to August, 2016.
During the survey 16 species belong to 8 families with 954 groups were detected. With the
density of 1.59 birds/ha, Sooty-headed Bulbul (Pycnonotus aurigaster) is the most
dominant species comparing with three remaining species. The density of Red-whiskered
bulbul (Pycnonotus jocosus), Red-vented bulbul (Pycnonotus cafer), Rufous-backed shrike
(Lanius schach) respectively are 0.278, 0.495 and 0.22 birds/ha. The detection probability
and number of bird groups which are detected by observation decrease with the increase of
distance, but in some case this trend is not satisfied. By density comparison between
distances sampling and traditional methods, distance sampling method which use detecting
probability indicate its higher accuracy result. Therefore adjusting the density by using the
detection probability is an effective method in wildlife survey and monitoring.

TABLE OF CONTENTS
ACKNOWLEDGEMENTS
ABSTRACT
TABLE OF CONTENTS
LIST OF THE DATA TABLES
LIST OF THE FIGURES
INTRODUCTION ................................................................................................................. 1
GOALS AND (SPECIFIC) OBJECTIVES ............................................................................ 4
2.1. GOALS ............................................................................................................................. 4
2.2. OBJECTIVES ..................................................................................................................... 4
METHODS ............................................................................................................................ 5
3.1. STUDY AREA: ................................................................................................................... 5
3.2. DATA COLLECTION: ......................................................................................................... 6
3.2.1. COLLECTING DATA:....................................................................................................... 6
3.2.2. DATA ANALYSIS ............................................................................................................ 8
RESULTS ............................................................................................................................ 12
4.1. SPECIES DIVERSITY AT PINE FOREST IN TDNP................................................................ 12
4.2. DENSITY OF BIRD SPECIES AT PINE FOREST: .................................................................... 14
4.2.1. DESCRIBING SURVEY DATA ......................................................................................... 14
4.2.2. MODELING DETECTION PROBABILITY BY DISTANCE METHOD ...................................... 15
4.2.3. ESTIMATING DENSITY OF BIRD SPECIES ....................................................................... 20
DISCUSSION ...................................................................................................................... 22
CONCLUSION .................................................................................................................... 24
REFERENCES
APPENDIX


LIST OF THE DATA TABLES

Table 3.1: Field data sheet used to collect information ......................................................... 8

Table 4.1: Bird diversity at pine forest in TDNP ................................................................. 12
Table 4.2: Number and percentage of Observation and Hearing in detection species ........ 13
Table 4.3: Distance group division ...................................................................................... 14
Table 4.4: Red-whiskered Bulbul’ parameters .................................................................... 16
Table 4.5: Red-vented bulbul’ parameters ........................................................................... 17
Table 4.6: Sooty-headed Bulbul’ parameters ...................................................................... 18
Table 4.7: Rufous-backed shrike’ parameters ..................................................................... 20
Table 4.8: Density of each specie in each line transect ...................................................... 21


LIST OF THE FIGURES

Figure 3.1: Location of pine forest in TDNP ......................................................................... 6
Figure 3.2: Line transect method ........................................................................................... 7
Figure 3.3: Graph of four standard functions used in Distance method. ............................... 9
Figure 4.1: Detection probability functions for Red-whiskered Bulbul .............................. 15
Figure 4.2: Detection probability functions for Red-vented bulbul..................................... 17
Figure 4.3: Detection probability functions for Sooty-headed Bulbul ................................ 18
Figure 4.4: Detection probability functions for Rufous-backed shrike ............................... 19


INTRODUCTION
Bird is very diverse group of animal. Currently, there are over 10500 bird species,
classified in 27 orders. Bird play an important role in natural ecosystem. Bird are excellent
indicators of forest ecosystem health as their abundance and diversity are closely related to
habitat disturbance. It is an important chain in the food web. Bird may be insectivores,
frugivores, and nectarivores and it is also food of another animals. Some nectar-feeding
birds are important pollinators, and many frugivores play a key role in seed dispersal.
Plants and pollinating birds often co-evolve, and in some cases a flower's primary
pollinator is the only species capable of reaching its nectar. Bird are often important to

island ecology. Bird have frequently reached islands that mammals have not; on those
islands, birds may fulfill ecological roles typically played by larger animals. Birdlife
international has ranked Vietnam as one of the leading countries of density and diversity of
bird. According to many statistic data, Vietnam’s bird population is over 870 species
(Nguyen Cu, 1995).
Tam Dao National Park is a protected area in North Vietnam. It was established in
1996, succeeding from the Conservation Forest Tam Dao which was formed in 1977. Tam
Dao national park is a precious natural resource, where keep high biological diversity with
many rare and endemic plants and animals. Tam Dao forest also has many species of rare
medicinal plants as a useful sources of medicinal. Moreover, tourism in TDNP has been
becoming a remarkable economic income.
In Vietnam, study on bird species is a major field from past to current time. Before
1945, almost of bird research project were done by foreign scientist. In this period, the
most famous is two French scientists are Delacour and Jabuille. From 1945 to 1954,
because of war, all researches were interrupted. After that, bird research was started again
in 1957. The most remarkable projects belong to authors like Vo Quy (1962-1966), Tran

1


Gia Huan (1960-1961), Do Ngoc Quang (1965), Vo Quy and Anorova N.C (1967).
Generally, scientist had focused on classification. In 1971, professor Vo Quy had
summarized his research from 7 year before and published the book: “Biology of common
bird in Vietnam”. Then, when Vietnam formally became independent, “Bird Vietnam” and
“Morphology and classification” were introduced. At that moment, the book “The list of
Vietnam bird” of Vo Quy –Nguyen Cu (1995) was published. The list contains 19 orders,
81 families and 828 species of Vietnam bird (Vo Quy and Nguyen Cu, 1995). From that
biodiversity preservation in Vietnam started to develop and “Bird Vietnam” book
published by Nguyen Cu, Le Trong Trai, Karen Phillips (2000).
Fauna which includes birds in Tam Dao was studied by some French professor like

J.Delacouri (1931), Osgood (1932), Bourret (1943)…Investigation in Tam Dao has
strongly developed from 1954, include projections of University students. Forest Inventory
and Planning Institute (1990-1992), counted for 281 wild life species which include 58
mammals, 46 reptiles, 19 amphibians, and 158 bird species in Tam Dao national park. Vo
Quy and Nguyen Cu (1995), found 239 bird species in Tam Dao national park. Tam Dao
was recognized as one of the important bird areas in Vietnam, because region has the
international importance in bird conservation. The area has a large number of bird species
restricted to a biogeographic unit. In particular, the region recorded some species that have
limited distribution are found only in a few areas in Vietnam such as: Blue-naped Pitta
(Pitta nipalensis), Purple Cochoa (Cochoa purpurea), Chestnut-headed Tesia (Tesia
castaneocoronata), Pale-footed Bush Warbler (Cettia pallidipes) and Rufous-headed
Parrotbill (Paradoxornis ruficeps) (Tordoff 2002).
In field survey, the density and diversity estimation of fauna is much difficult
than flora. The biggest challenge is detect the entire individual in study area, almost
object far from transect is missed. For this reason, ecologists generally have to depend
2


on some kind of estimate of abundance or density. There exists a variety of method
exist to do this job. Each kind of wildlife is best suitable with one survey method. For
example, insects, aquatic organisms, soil organisms can be surveyed by using point, plot
survey methods; number of fishes or small mammals are detected by mark and
recapture method; etc. In case of bird, the best method used is line transect survey
method. However, it is still impossible to obtain a complete count or census of a natural
bird population. It mean that the density estimated is smaller than reality and detection
probability is smaller than 1. There are two ways to deal with this error. The first way is
survey in narrow transects, but this way is not efficient in term of time and economic.
And the rest way is using the data collected from objects in far transect to estimate the
detection probability, and then use this number to adjust the density estimate. The term
“distance sampling” refers to a suite of method that will estimate the absolute density of

biological population, based on accurate distance measurements of all objects near a
line or point (Buckland et al. 1993). The main methods are line-transect sampling and
point-transect sampling and in this survey, I used line transect method. Distance
sampling is a good method for both flora and fauna survey.
Although, up to now there are number of research about bird at pine forest
in Tam Dao national park, most of them are general study, and there is not much study
which point out the density and diversity of a particular order. Therefore, in this study,
the density and diversity of bird species at pine forest in Tam Dao national park are
assessed using the distance sampling method.

3


GOALS AND (SPECIFIC) OBJECTIVES
2.1. Goals
The goal of this research is to provide basic information on the diversity and
density of bird which can contribute to the management and conservation of biodiversity at
pine forest in Tam Dao national park.
2.2. Objectives
- To assess the diversity of bird of pine forest in Tam Dao national park.
- To estimate the density of bird of pine forest in Tam Dao national park

4


METHODS
3.1. Study area:
The study was conducted in pine forest in Tam Dao national park with coordinates
21°21’- 21°42’ degrees north latitude, 105°23’- 105°44’ degrees east longitude.
The current area of Tam Dao national park is 34,995 hectares belonging to 3

provinces are Vinh Phuc, Thai Nguyen and Tuyen Quang. Entire Tam Dao national park
border lie on Tam Dao mountain range, stretching over 80 kilometers, runs northwest –
Southeast from Son Duong District (Tuyen Quang) to Phuc Yen Town (Vinh Phuc). Tam
Dao mountain range includes many mountains over 1,300 meters, the highest peak is Tam
Dao Bac (1,592 meters).The area of Tam Dao national park is 34,995 hectares, including
26,163 hectares of forest, mostly natural moist evergreen forest with 70% coverage of the
entire area. Also, in Tam Dao national park exist some other forest types such as evergreen
subtropical moist low mountain, pygmy forest, bamboo forest, restoration forest after
mining, plantation, scrub, grass land, etc. Due to human activities such as harvesting,
cultivation or forest fire, current forest areas are mainly on 700 meter altitude. Below this
elevation, forest has been cleared and replaced by secondary shrubs, grass-shrubs, and pine
plantation forest.
Study area is pine forest. This is a plantation of Tam Dao national park with area
approximate to 2900 ha. In below map, study area is represented by violet.( Figure 3.1)

5


Source: Tam Dao national park
Figure 3.1: Location of pine forest in TDNP
3.2. Data collection:
3.2.1. Collecting data:
Line transect method
Bird survey was conducted from June to August in 2016 at pine forest in Tam Dao
national park. Data from previous survey (Vu Tien Thinh et al., 2012) was also used for
analysis.
Equipments for survey are binoculars, cameras, data sheet, and map.
Data were collected using the line transect method. I used line transect method
for surveying bird density and diversity. I choose this method because it is usually
6

designed for some cases: animals has small size or the terrain of survey area not favorable
for moving when we conduct to survey.
I set up 12 line transect at pine forest which are from the elevation of 200m-600m.
The transect is randomly distributed in study area. Birds are surveyed from early morning
(5:30) to mid-noon (11:00 to 12:00) and from 15:00 to 18:00. Because birds are most
active at this time. Each transect’s length is 500 meters. The selected transects were
located away from the status boundary at least 70m to minimize edge effects and impact of
the status change on the detecting probability. Transects were arranged at least 100 m apart
to make sure independence. ( Figure 3.2). Observers go along transects with the speed is
around 0.5km/40 minutes, observe and record the presence, number of bird by observing
and singing. And, observers also estimate the distance from object to transect. Each
transect is surveyed in six times.

Figure 3.2: Line transect method

7


Individuals are detected to be recorded as y1,y2 ,…yi in which yi is the distance
from line transect to individual i.
The collected data is recorded in data sheet as table below (Table 3.1)
Table 3.1: Field data sheet used to collect information
Site: ……………….

Observer (s): ………….

Line transect number: ….

Time start/finish: ……..

Date: …………. .….

Weather………………
Species

Date

Distance

Time Transect Coordinate

Quantity
name

(m)

Heard/Seen

3.2.2. Data analysis
 Estimating detection probability
In order to estimate the density of objectives, surveyors need to estimate detection
probability by using the investigate data. Detection probability is estimated base on
frequency distribution at different distance to transect or the observer. And, the software
which is popular used to analyze detection probability is DISTANCE 6.0 (Thomas et al,
2010).
Before estimating detection probability, surveyors must define detection probability
function. Four basic functions which are used to estimate detection probability by distance
are:

8




Uniform: ( )



Hazard-rate: ( )



Half-normal: ( )



Negative exponential: ( )

( )

With: g(y) is the detection probability of an object with distance to line
transect is y (length unit); w, ,

b are the estimated parameters;

Beside of these four functions, there are three expansion series which can be used
to change the shape of detection probability curves and simulate better the variation of
detection probability. These three expansion series are Cosince, Simple polynomials, and
Hermite polynomials series. The combination of standard functions and expansion series is

show in the function: g(y) = basic function(y) + expansion series(y)
In some cases, this combination can indicate the relationships between
detection probability and distance better.

Figure 3.3: Graph of four standard functions used in Distance method.

9


Aikaike’ Information Criterion (AIC) (Anderson, 2007) is used to determine which
function is the best for estimating detection probability. ). This standard is based on the
principle of balance between the standard deviation and variance. Function which has
lowest AIC value will be used. In addition, Distance method also provides χ2 values to
check the suitability of function with the distribution of empirical frequency. When the
best function is determined, detection probability (Pa) is calculated by divide the area of
upper function curve part by rectangle area and time with transect wide, function:
∫

( )

 Estimating density of species
The estimation of density with distance sampling cannot exactly unless five
fundamental assumptions have to be satisfied:


Objects on the line are detected with certainty. Objects directly on the line

are 100% detected g (0) =1 and the probability of detection substantially decreases with the
increase of distance. (Buckland et al. 2001).


Objects do not move. All measurements are made from the objects’ initial

location, before it was affected by the observer (Buckland et al 2001).


Measurements are accurate. All angles, distance, objects, sex and other

necessary measurements are measured with accuracy without any errors (Buckland et al.
2001).


For animal species that occur in clusters (groups), cluster sizes are recorded

without error. In some circumstances, cluster sizes may be accurately estimated close to
the line or point, but poorly estimated at larger distance. Bias from this source can be
avoided by using the regression correction for size-biased sampling that is the default for
clustered data in the software Distance (Buckland et al. 2001).

10




The sampled plots (circles or strips) are representative of the entire survey

region. This is not usually listed as an assumption, because if an appropriately randomized
design is used, the assumption holds by design. However, non-random plots are often
covered, in which case this assumption becomes important.
Although other assumptions are made, generally only the above five have any

practical significance.
With each transect whose length is L meters, there is (are) n object(s) detected. The
perpendicular distance of each object to the transect line is also recorded. When all objects
located on the line are detected with certainty, the density of objects in the survey are (D)
is estimated as:
̂

̂

(Buckland et al, 2001)

With: n – Total objects is detected in the surveys
̂ - Detection probability ≤ 1
a = 2Lw (w – width of line transect, L – total line length)
Variance of density estimation is calculated by the formula:
̂ ( ̂) = ̂² {

̂( )

+

̂( ̂ )
( ̂ )

+

̂ ̂( )
( )

}

Variance of density estimation show fluctuations the number of individuals
detected in transect (E(s) is group size).
The total number of individual in each species (population size) is calculated by
time density (̂)with area (A): ̂

̂

The variance of population size is: ̂

( ̂)

11

̂

( ̂ ).


RESULTS
4.1. Species diversity at pine forest in TDNP
I conducted to survey 12 line transects at pine forest in Tam Dao national park and
sixteen species belonging to 8 bird families were detected. In which, the researcher
identified species of consequence namely: Sooty-headed bulbul (Pycnonotus aurigaster),
Red-vented Bulbul (Pycnonotus cafer), Common tailorbird (Orthotomus sutorius) and
Ashy Drongo (Dicrurus leucophaeus) (Table 4.1)
Table 4.1: Bird diversity at pine forest in TDNP
No

Family

Latin name

Streak-breasted Scimitar

Pomatorhinus

Babbler

ruficollis

Striped tit-babbler

Macronous gularis

43

Garrulax pectoralis

38

Puff-throated babbler

Pellorneum ruficeps

21

Buff-breasted babbler

Pellorneum tickelli

38

Red-whiskered Bulbul

Pycnonotus jocosus

32

Red-vented bulbul

Pycnonotus cafer

109

Greater Necklaced

1

Laughingthrush
Timaliidae

Pycnonotidae
2

Number of

Common name

Sooty-headed Bulbul

Pycnonotus
aurigaster

individual
16

172

Puff-throated Bulbul

Alophoixus pallidus

71

3

Laniidae

Rufous-backed shrike

Lanius schach

24

4

Sylviidae

Common tailorbird

Orthotomus sutorius

141

5

Dicruridae

Ashy Drongo

Dicrurus leucophaeus

140

6

Cisticolidae

Rufescent prinia

Prinia rufescens

31

7

Corvidae

Gray Treepie

Dendrocitta formosae

16

Red-billed blue magpie
8

Paridae

Great tit

Urocissa
erythrorhyncha
Parus major

12

34
28


Table 4.2: Number and percentage of Observation and Hearing in detection species
Species

# Observation

#Hearing

Total

%Observation

%Hearing

(Group)

(Group)

number

(%)

(%)

Streak-breasted Scimitar Babbler

2

14

16

12.5

87.5

Striped tit-babbler

36

7

43

83.7

16.3

Greater Necklaced Laughingthrush

38

0

38

100

0

Puff-throated babbler

2

19

21

9.5

90.5

Buff-breasted babbler

10

28

38

26.3

73.7

Red-whiskered Bulbul

16

16

32

50

50

Red-vented bulbul

44

65

109

40.4

59.6

Sooty-headed Bulbul

152

20

172

88.4

11.6

Puff-throated Bulbul

29

42

71

40.8

59.2

Rufous-backed shrike

16

8

24

66.7

33.3

Common tailorbird

95

46

141

67.4

32.6

Ashy Drongo

86

54

140

61

39

Rufescent prinia

23

8

31

74.2

25.8

Gray Treepie

12

4

16

75

25

Red-billed blue magpie

22

12

34

64.7

35.3

Great tit

18

10

28

64.3

35.7

Total

601

353

954

63

37

13


Table 4.2 refers the number and percentage of Observation and Hearing in
detection species:
In all 16 species, almost all groups were detected through observation. In total,
there were 601 groups ( 63%) that were detected by observation, while the number of
group were discovered by sound were 353 ( 37%). And, the common points for all 16
distance data sets are the rising of Hearing proportion and decline of Observation
proportion with the increase of distance from transect.
4.2. Density of bird species at pine forest:
4.2.1. Describing survey data
During the survey, there are 16 bird species groups were detected. Because
there are a lot of species were detected but time was limited so we choose 4 bird species
including : Red-whiskered Bulbul (Pycnonotus jocosus), Red-vented bulbul (Pycnonotus
cafer), Sooty-headed Bulbul (Pycnonotus aurigaster), Rufous-backed shrike (Lanius
schach) to estimate density (distance data of each group is considered as a “distance data
set”). Almost all groups were detected under 20m of distance from transect lines. In order
to model the detection as distance increases, distance data was divided into 5 groups as the
following table:
Table 4.3: Distance group division

Distance interval( m)
No

Species
0-10

10_20

20-30

30-40

40-50

1

Red-whiskered Bulbul

1

20

3

5

3

2

Red-vented bulbul

26

36

25

9

13

3

Sooty-headed Bulbul

86

49

7

7

23

4

Rufous-backed shrike

10

3

7

2

2

123

108

42

23

41

Total

14


4.2.2. Modeling detection probability by distance method
The same trend for all four data sets is that the detection probability decreases with
the increase in distance from transect lines. This trend is clearly indicated by detection
probability histograms from distance method. By using distance method, the best suitable
functions for each species were determined. In addition, the χ2 values were also provided

to check the suitability of function with the distribution of empirical frequency. If χ2
values exceed 0.05, selected function can model the variation of the detection probability
by distance. Following are the distance data analysis by distance software including
functions’ histogram and tables of relevant parameters.
Red-whiskered Bulbul:

Figure 4.1: Detection probability functions for Red-whiskered Bulbul

15


Table 4.4: Red-whiskered Bulbul’ parameters
Density

Detection
Functional form

AICValue

CV (%)
(Birds/ha)

probability
Half-normal

101.1

0.123

47.61

0.278

Uniform

101.27

0.102

47.06

0.27

103.1

0.273

52.29

0.266

101.56

0.166

48.84

0.258

Negative

exponential
Hazard-rate

With the smallest AIC value (101.1), Half- normal function is the best fit model
which can well show the detection probability of Red-whiskered Bulbul (Pycnonotus
jocosus) density. The next orders are Uniform, Hazard-rate and Negative expotential which
have higher AIC values. And, χ2 values, degree of freedom, and p-value are 5.24, 3 and
0.68 (>0.05). These are good parameters to model the fluctuation of detection probability
by distance.
Red-vented bulbul:

16


Figure 4.2: Detection probability functions for Red-vented bulbul
Table 4.5: Red-vented bulbul’ parameters
Density

Detection
Functional form

AICValue

CV (%)
(Birds/ha)

probability
Half-normal

340.35

0.09

28.92

0.435

Uniform

339.77

0.08

28.69

0.495

341.77

0.084

28.69

0.453

341.01

0.17

30.16

0.423

Negative
exponential
Hazard-rate

With the smallest AIC value (339.77), Uniform function is the best fit model which
can well show the detection probability of Red-vented bulbul (Pycnonotus cafer) density.
The next orders are Half- normal, Hazard-rate and Negative expotential which have higher
AIC values. And, χ2 values, degree of freedom, and p-value are 1.63, 3 and 0.67 (>0.05).
These are also good parameters to model the fluctuation of detection probability by
distance.

17


Sooty-headed Bulbul:

Figure 4.3: Detection probability functions for Sooty-headed Bulbul
Table 4.6: Sooty-headed Bulbul’ parameters
Detection
Functional form

AICValue

Density
CV (%)

probability

(Birds/ha)

Half-normal

452.2

0.07

28.98

1.328

Uniform

447.13

0.08

29.11

1.325

467.7

0.1

29.5

1.41

446.61

0.09

29.3

1.59

Negative
exponential
Hazard-rate

18


With the smallest AIC value (446.61), Hazard- rate function is the best fit model
which can well show the detection probability of Sooty-headed Bulbul (Pycnonotus
aurigaster) density. The next orders are Uniform, Half- normal and Negative expotential
which have higher AIC values. And, χ2 values, degree of freedom, and p-value are 3.59, 1
and 0.36 (>0.05). These are also good parameters to model the fluctuation of detection
probability by distance.
Rufous-backed shrike:

Figure 4.4: Detection probability functions for Rufous-backed shrike

19