In the virtual forest tutorial, I sampled the Snyder-Middlesworth Natural Area using area based systematic, random, and haphazard sampling strategies.
The fastest strategy was systematic, with an estimated sampling time of 12 hours, 35 minutes. Random was close behind, with an estimated sampling time of 12 hours 37 minutes. The slowest was haphazard, with an estimated sampling time of 13 hours.
To assess the accuracy of the approaches, I calculated the percent error of their estimated densities for the two most common species and the two least common. The data can be found in the table below.
|
Systematic % Error |
Random % Error |
Haphazard % Error |
||
|
Eastern Hemlock (Most common)
|
6.4% |
24.1% | 45.6% | |
|
Red Maple (2nd most common)
|
24.5% |
36.9% |
24.5% |
|
|
Striped Maple (2nd least common) |
100% |
42.9% |
31.4% |
|
|
White Pine (Least common) |
233.3% |
1.19% |
52.4% |
With the systematic approach, accuracy got progressively worse as the species got less abundant. With the random and haphazard approaches, accuracy did not seem to correlate with the abundance of the species. When averaging their 4 results, random had an average percent error of 91.1%, random had 26.3%, and haphazard had 38.5%. For this reason I would say that random sampling was the most accurate sampling method overall. However, systematic was the most accurate when only looking at the most populous species, with an average percent error of 15.5%