Life Sciences · Ecology · Evolution · Research / Principles and Exploration
Quadrat sampling — principles and investigation
When it is difficult to count the number of organisms in a large area, the total number is estimated by examining samples in a small quadrat. Change the quadrat, number of samples, and arrangement method, and compare the estimates with the actual map. We examine how increasing the sample and selecting a representative sample each affect the error.
How does it work?
If you repeat random sampling, large samples tend to reduce the average error. Selection bias in selecting clusters does not disappear even if the number of samples is increased.
Estimated number of individuals = average density (individuals/m²) × 10000 m²
Try this investigation
- Grasslands with mixed colonies are fixed to map seed 137. To place directly, drag the scene or select it with the position slider and press the Add Sample button.
- Record sample numbers 5 and 20 from the same map, area, and sampling seed. For each condition, we also calculate the average absolute error of 100 separate random samples.
- If you only expand the area and cover the entire area, check if the estimated value is the same as the actual value. We also explore the bias that occurs when only clusters are selected through direct placement.
Interpreting your results
What is the equation for estimating the total population?
The average density is the total number of individuals in the sample/(number of samples × quadrat). Multiply this by the total area of 10000 m².
Assumptions and limitations
It is a 100×100 m virtual grassland with 600 fixed objects. We mixed 240 uniformly distributed individuals and 360 individuals from three clusters. To equalize the probability of edge inclusion, the grassland is divided into equal-area grids and then the grids are restored and extracted. Direct placement also fits into this grid, and repeated grids can be counted again. We omit the actual plant size, species classification, seasonal changes, and omission of detection. The average absolute error over 100 trials is a separate random sampling value and does not necessarily decrease from a single result.