2.1.1 Spatial Design Overview

Spatial Designs

Spatial Designs

Spatial designs describe how sampling effort is to be allocated across a study area.  The most appropriate spatial design for you depends on your monitoring requirements, monitoring design characteristics, and monitoring constraints as identified by you in your answers to the status and trend basic questions.   In general, the following types of spatial designs are available:

Census

The census spatial design describes the location of all the sites comprising the domain of interest.  In some cases, a single site might be used to estimate the total number of fish in a population, by the establishment of a counting facility located strategically where all fish will pass and be counted.  In other cases, the census might consist of counting fish throughout the population’s domain occupied (or potentially occupied), e.g., at all reaches where the species occurs.  In any case, a census implies that all elements will be enumerated.  In some cases, it is feasible to conduct a census in a part of the population’s domain, but not all.  In these cases, the term “restricted census” applies—part of the population’s domain can be censused; part will be sampled using another type of design.

Model-based

A model-based spatial design relies on selection of sites based on the need to estimate parameters or coefficients of a model that will be used to make the population estimates. Such models typically include one or more independent variables or covariates such as environmental conditions or habitat quality. Sites are generally selected along the important gradients governing the model parameters.  A simple model might be a relationship between a population’s growth rate and temperature.  Sites might be selected at locations covering a thermal gradient over the range of the population’s thermal tolerance.  Then the model would be used to estimate productivity across all sites in the domain.  A restricted model-based spatial design refers to situations in which the selection of locations in part of the domain is guided by the candidate model, and locations in other parts are selected by other methods..

Survey

The term survey in the current context implies the use of a randomization rule in the selection of locations across the domain of interest and the caveat that all locations have a positive chance of being selected.  Approaches available to achieve these criteria for monitoring natural resources include: simple random sampling, systematic sampling, and grts (Stevens and Olsen, 2004) based sampling.  A restricted survey design implies that part of the domain will be sampled by application of a survey, and other parts by application of one of the other spatial designs.  There are six sub-categories of survey-based designs:

  • Non-stratified Independent Random Survey – a sampling technique where a group of subjects (a sample) for study is selected from a larger group (a population).  Each individual is chosen entirely by chance and each member of the population has an equal chance of being included in the sample.  A non-stratified sample does not take separate samples from strata or sub-groups of a population.
  • Stratified or Variable Probability Independent Random Survey – a sampling technique where a group of subjects (a sample) for study is selected from a larger group (a population).  Each individual is chosen entirely by chance and each member of the population has a known, but possibly non-equal, chance of being included in the sample.  A stratified sample is obtained by taking samples from each stratum or sub-group of a population.
  • Non-stratified Generalized Random-Tessellation Stratified (GRTS) Survey – produces a probability sample with design-based variance estimators.  It provides a spatially balanced, random sample, allows for unequal probability sampling, and can provide an over-sample of sample sites to accommodate field implementation issues.  A non-stratified sample does not take separate samples from strata or sub-groups of a population.
  • Stratified or Variable Probability GRTS Survey – produces a probability sample with design-based variance estimators.  It provides a spatially balanced, random sample. Allows for unequal probability sampling, and can provide an over-sample of sample sites to accommodate field implementation issues.  A stratified sample is obtained by taking samples from each stratum or sub-group of a population.
  • Non-stratified Systematic Survey – a method of selecting sample members from a larger population according to a random starting point and a fixed, periodic interval.  Typically, every “nth” member is selected from the total population for inclusion in the sample population.  Systematic sampling is still thought of as being random, as long as the periodic interval is determined beforehand and the starting point is random.  A non-stratified sample does not take separate samples from strata or sub-groups of a population.
  • Stratified Systematic – a method of selecting sample members from a larger population according to a random starting point and a fixed, periodic interval.  Typically, every “nth” member is selected from the total population for inclusion in the sample population.  Systematic sampling is still thought of as being random, as long as the periodic interval is determined beforehand and the starting point is random.  A stratified sample is obtained by taking samples from each stratum or sub-group of a population.

Opportunistic

An opportunistic-based design is where you will only be able to sample at sites that are selected based on ease of access or other subjective criteria.

In some instances, categories may be combined to produce hybrid designs.  For example, part of your domain may be sampled by counting fish as they pass over a weir (census).  The remaining portion of the domain may be best monitored by a survey.

Each of these spatial designs has strengths and weaknesses.  In general, the chance of making poor inferences is highest for opportunistic spatial designs and lowest for census designs.  Conversely, opportunistic spatial designs will generally be less expensive to implement that census designs.