Basics of Research with Emphasis on

1
Basics of Research with Emphasis on Estimating
Abundance and Habitat
2
Objectives

• The first major failure of most conservation
  research is the lack of a clear objective
• The second major failure of research is failure
  to establish specific objective
• The third major failure is to then completely
  ignore those objectives

3
Objectives

• Action Plans provide direction, but cover
  objectives in more general terms need to refine
• For example there is a PQFAP objective for the
  Hainan hill partridge relative to forest
  management and reserves.
• Review work done by Gao Yu-ren
• Determine extant habitat
• Specific objectives
• Always keep the terms TIME and SPACE in the
  back of your mind

4
Objectives

• Best approach is to describe in terms of question
• For example
• Is purpose to describe present distribution of
  species in a country?
• Is purpose to describe present distribution of
  species in a reserve?
• Is purpose to estimate abundance of species
  in a particular locality?
• Is purpose to make comparisons of abundance
  among different reserves or among
  different types of management?

5
Objectives

• Is purpose to describe dynamics (trends) to
  assess changes in populations or distributions?
• Is the purpose to establish causal relationships
  eg. Habitat destruction, mushroom collection,
  hunting?
• Keep in mind that the purpose of a research
  project is NOT to develop conservation
  assessments of a species.
• Conservation assessments might be an outcome,
  but they are not a conclusion

6
Objectives

• Approach to our research
• Next question to ask ourselves is the type of
  approach 3 major types
• Correlational studies most common approach
• Experimentation strongest inference
• Adaptive Resource Management may be the
  future for our research direction
  (Walters 1986)

7
Sampling
What is the need?

• We sample because we can rarely do a complete
  census in fact even for humans a complete
  census often produces poorer data
• Sampling means a subset of your study population
  selection of elements of the population,
  collection of data on those, and using that data
  to draw conclusions about the population

8
Sampling

• First thing to define Your POPULATION!!
• Is it the Sichuan HP in a reserve?
• Is it the SHP in the western Sichuan?
• Is it the SHP in Sichuan?
• Is it the SHP in the China?

9
Sampling

• Second thing to define REPRESENTATION
• Does your sample represent your target
  population?
• For example

10
Sampling

• We define our target population as the population
  of quail in a 10,000 ha reserve
• Our objective is to undertake an assessment of
  the population of this quail eg What is the
  population of the quail in the reserve?
• Biologist A comes up with the following quadrat
  sampling design -

11
Sampling
10,000 ha reserve
Road
Biased sampling
12
Sampling
10,000 ha reserve
Random Sampling
13
Sampling
10,000 ha reserve
Habitat C
Habitat A
Stratified Sampling
Habitat B
14
Sampling
10,000 ha reserve
Systematic Sampling
15
Sampling

• Key is to decide design and sample number
• Sample size of ONE and ALL are both wrong because
  they are either producing no data or are a waste
  of resources

The latter in abundance monitoring is actually a
true census!!!!!
16
Unit of Measure

• Key issue to be defined at the beginning
• Commonly it is the animal
• Or the group of animals
• Or a spatial unit
• Or a unit in time
• Often very poorly defined resulting in some of
  the common research mistakes

17
Unit of Measure Some Definitions

• Census complete count e.g. number of biologists
  in this room---no error estimate
• Population Estimate Estimate of numbers of
  animals on a study area e.g. 228 orange necked
  hill partridge in Cat Tien National Park,
  VN---must have error estimate
• Density estimate Estimate of number of animals
  per unit area e.g. 15 western tragopans per km2
  in Nepal---must have error estimate

18
Unit of Measure Some Definitions

• Index
• Number of calls per point
• Number of birds observed per km of transect
• Does not usually provide a population or density
  estimate
• Can provide provide an estimate of population or
  density if combined with a quantitative technique

19
Variability in Sampling
A sample is comprised of two things?

• A measure of central tendency!!!!!!
• A measure of statistical error!!!!!!
• Statistical error is not a measure of bias!!!!!!

Do not ever present a mean or median or any
measure of central tendency without a measure of
error!!!!!!!!!!!!!!!!!
20
Scale

• Critically important Always ask if you have the
  scale right
• Best analogy Are you trying weigh elephants to
  the nearest gram and trying to weigh mice to the
  nearest kilogram
• Common mistake is to try to apply techniques
  designed for fine scale to very coarse scale
  projects

21
Concept of Detectability

• Probably most overlooked issue in research
• What is it???
• It is the relationship between our count of our
  sample units (birds) and the number that are
  actually present

22
Concept of Detectability

• Remember that good estimates of populations are
  not possible simply by increasing the sample size
  in fact extra sampling for the sake of larger
  sample size is not warranted or a good use of
  limited resources
• Any counts, or calls, or other observations must
  have some connection (hopefully measurable )with
  either absolute or relative populations
  unbiased estimates

23
Concept of Detectability

• For example
• 1,000 ha study area
• Divided into 10 100-ha plots
• We sample all 10 plots and find a mean of 12.5
  birds per plot ( 1.5 SD)
• What is our population estimate?

24
Concept of Detectability
N 125 (15)
25
Concept of Detectability

• Now suppose
• I am All Seeing and I know that our ability to
  detect birds in each plot is only 0.7.
• What then is our true population estimate?

N 179 22
26
Concept of Detectability

• Obviously we are not All Seeing !!!!!!
• However, there are numerous techniques that allow
  us to estimate detectability (b)
• For example double sampling
• For example using radio-tagged or tagged birds to
  verify
• Use of absolute estimates which include estimates
  of detectability

27
Concept of Detectability

• This is true even when we are doing INDICES
• Why?
• Because indices can even have problems if we
  cannot meet the assumption that the index does
  not vary with space and time independent of the
  population we are measuring

28
Application to Habitat Analysis

• Finding birds in the field
• In conjunction with population surveys
• Radio-telemetry

29
Variation in Detectability

• Are there more detections of SHP in Primary
  Forest because there are more birds or because
  they are easier to see?
• Can create biases in habitat analysis

30
Analysis techniques
Design I studies

• Logistic regression
• Log-linear modeling
• Simple c2 analysis (Neu et al. 1974)

31
Telemetry analysis techniques
Design II - IV studies

• Design II Animals marked, but only study area
  habitat assessed
• Design III Resource use and habitat available
  defined for each animal
• Design IV Resource use and habitat available
  defined for each observation of each animal

32
Telemetry analysis techniques
Design II - IV studies

• Scale and Resource Availability are key issues
  impacting all Designs
• Once again sampling is a key issue individuals
  radio-tagged, temporal variation in data
  collection, non-random loss of observations,
  autocorrelation

33
Telemetry analysis techniques
Defining Resource Availability

• Formal Framework provided by Johnson (1980)
• First Order Geographic range of species
• Second Order Home range location within
  distribution
• Third Order Selection of particular areas
  within a home range
• Fourth Order Microhabitat or food selection

34
Telemetry analysis techniques

• Radio-telemetry studies fit Second and Third
  Order Resources Studies
• Friedmans Test
• Johnson Method
• Compositional Analysis
• Discrete Choice Modeling
• Logistic Regression
• Log-linear Modeling
• Multiple Regression