Category: Post 4: Sampling Strategies

Used the area-based method. The systematic sampling technique took 12 hr 37 min, random took 12 hr 43 min, and haphazard took 12 hr 28 min making it the fastest. Note that the different in time between the 3 different techniques is only a range of 15 minutes. Easter hemlock and sweet birch were the two most common tree species. The systematic sampling gave the lowest percent error (13.2%, 21.7%) making it the most accurate of the 3 samples.

Striped maple and white pine were the two rarest species. The haphazard sampling gave the most accurate experimental density have the lowest percent error (76.0%, 1.2%) from the 3 samples. In general the accuracy declined for the rare species and the vary low percent error for the white pine in the haphazard sample may just be luck.

Overall, the haphazard sample gave the more constant and lowest total percent error of all 4 tree species, also this took the shortest amount of time as stated in the beginning.

Note the majority of percent errors were very high which may be due to the sample size ( n=24) being too small. To experimentally collect a more accurate findings the sample size should increase to a large value, i.e. n=50.

I used the distance based method. The systematic sampling technique took the least amount of time, at 4 hrs, 5 mins, followed by haphazard at 4 hrs, 26 mins, and finally random which took 4 hrs, and 40 mins. Eastern hemlock and yellow birch were the two most common trees for each sampling technique. The sampling error was lowest for random (-1.7%, 29.8%) which would make it the most accurate, followed by haphazard (-6.5%, 31.7%) and systematic which had the highest error (-22.2%, 55%).

Striped maple and white pine were the two least common species for each sampling technique. Again, random (11.9%, -100%) was the most accurate, however systematic (60.6%, 123.8%) was more accurate than haphazard (118.3%, 124.7%) for the least common species. In general, the accuracy declined for rare species.

The percent error calculations are all quite large aside from the ones obtained in random sampling, and I believe this is due to the fact that 24 is too small a sample size. Generally as your sample size increases, your margins of error decrease (Statsoft, 2018) so I think increasing the sample size would yield more accurate results.

Systematic		Sample time: 4 hrs, 5 mins
		Actual Density	Data Density	% Error
Common	Eastern Hemlock	469.9	365.8	-22.2%
	Yellow Birch	108.9	168.8	55%
Rare	Striped Maple	17.5	28.1	60.6%
	White Pine	8.4	18.8	123.8%
Random		Sample time: 4 hrs, 40 mins
Common	Eastern Hemlock	469.9	461.8	-1.7%
	Yellow Birch	108.9	141.4	29.8%
Rare	White Pine	8.4	9.4	11.9%
	Striped Maple	17.5	0	-100%
Haphazard		Sample time: 4 hrs, 26 mins
Common	Eastern Hemlock	469.9	439.7	-6.5%
	Yellow Birch	108.9	143.4	31.7%
Rare	Striped Maple	17.5	38.2	118.3%
	White Pine	8.4	19.1	127.4%

 

StatSoft. (2018). Designing and Experiment – Power Analysis. Retrieved February 2, 2018 from: http://www.statsoft.com/Textbook/Power-Analysis#power_doe3

In the online forest sampling tutorial given, I have chosen to do 1. Random sampling using area, 2. Systematic sampling along a topographic gradient using distance, and 3. Haphazard sampling using area. The Haphazard method had the fastest estimated time to sample at 2h38min, compared to 12h47min for the random sampling method, and 4h7min for the systematic sampling.

According to actual data, the two most common species in the Snyder-Middleswarth Natural Area were Eastern Hemlock and Sweet Birch. Let us use tables to compare the % error of the different sampling strategies for both.

Species	Measures	Actual Data	Data for The Random Sampling Method	Data for The Systematic method	Data for The Haphazard method	% Error Random Sampling	% Error Systematic Sampling	% Error Haphazard Method
Eastern Hemlock	Density	469.9	354.2	479.0	380	24.6%	1.94%	19.13%
	Frequency	73%	71%	70.8%	80%	2.7%	3.01%	9.6%
	Dominance	33.3	19.8	35.5	39.6	40.5%	6.61%	18.92%
	Relative Density	50.6	44.0	54.2	43.2	13%	7.11%	14.62%
	Relative Frequency	33.8	32.1	37.0	33.3	5.1%	9.47%	1.48%
	Relative Dominance	44.4	45.6	53.6	54.7	2.7%	20.72%	23.2%
	Importance Value	42.9	40.6	48.2	43.7	5.4%	10%	1.86%
	Morisita Index	1.89	2.33	1.05	1.35	23.3%	44.44%	28.57%
Sweet Birch	Density	117.5	41.7	64.5	60.6	64.51%	45.11%	48.43%
	Frequency	43.0%	25%	29.2%	20.0%	41.86%	32.09%	53.49%
	Dominance	20.2	5.1	11.3	8.4	74.75%	44.06%	58.42%
	Relative Density	12.7	5.2	7.3	6.8	59.05%	42.52%	46.46%
	Relative Frequency	19.9	11.3	15.2	8.3	43.21%	23.62%	58.29%
	Relative Dominance	26.9	11.8	17.1	11.6	56.36%	36.43%	56.88%
	Importance Value	19.8	9.4	13.2	8.9	52.53%	33.33%	55.05%
	Morisita Index	2.27	3.20	0.00	5.00	40.97%	100%	120.26%

 

 

 

 

 

 

 

 

Then, let us do the same thing for the two most rare species; Striped Maple and White Pine.

Species	Measures	Actual Data	Data for The Random Sampling Method	Data for The Systematic method	Data for The Haphazard method	% Error Random Sampling	% Error Systematic Sampling	% Error Haphazard Method
Striped Maple	Density	17.5	0.0	18.4	60.0	NA	5.14%	242.86%
	Frequency	6.0%	0.0%	4.2%	20.0%	NA	30%	233.33%
	Dominance	0.7	0.0	0.6	3.6	NA	14.29%	414.29%
	Relative Density	1.9	0.0	2.1	6.8	NA	10.53%	257.89%
	Relative Frequency	2.8	0.0	2.2	8.3	NA	21.43%	196.43%
	Relative Dominance	0.9	0.0	1.0	5.0	NA	11.11%	455.55%
	Importance Value	1.8	0.0	1.7	6.7	NA	5.56%	272.22%
	Morisita Index	17.00	NA	24.00	5.00	NA	41.18%	70.59%
White Pine	Density	8.4	8.3	0.0	20.0	1.19%	NA	138.09%
	Frequency	4.0%	4.0%	0.0%	20.0%	0%	NA	400%
	Dominance	0.9	1.1	0.0	0.4	22.22%	NA	55.55%
	Relative Density	0.9	1.0	0.0	2.3	11.11%	NA	155.55%
	Relative Frequency	1.9	1.8	0.0	8.3	5.26%	NA	336.84%
	Relative Dominance	1.2	2.5	0.0	0.6	108.33%	NA	50%
	Importance Value	1.3	1.8	0.0	3.7	184.62%	NA	184.62%
	Morisita Index	16.13	24.00	NA	NA	48.79%	NA	NA

 

For the Shannon-Weiner diversity index (not shown in above tables), the most accurate measure was the one given by the random sampling method using area which was giving the exact same figure as actual data: 1.5. However, looking at the % error for the two most common and two rarest species, accuracy greatly varies within the three sampling strategies depending on the measure and the species concerned. For the Sweet birch, the % error was extremely high for all three methods, and in all measures. As for the striped maple, the systematic method was the most accurate, given that the random sampling method did not account for any tree of that species, while the % error of the haphazard method was considerably higher than for the systematic sampling. Finally, the random sampling method was the most accurate for the white pine species. Its percentage error was noticeably lower than in the systematic sampling, and the haphazard method did not provide any data for the white pine. Before doing this tutorial, I was expecting that accuracy would increase in the same direction as species abundance, so I was quite surprised to see how far off were the results for the sweet birch species measures. After doing this tutorial, I realized that for an area as wide as the Snyder-Middleswarth Natural Area, it would have been preferable to use more than 24 samples for better accuracy.

H. Zulfiqar

For this experiment, I selected the Mohn Hill community. It had an abundance of unique species that I sampled across the 84 sampling opportunities. Red maple, white oak, and chestnut oaks were found at the highest density regardless of the sampling technique used. Many of the rare species were missed in the samples, while other rare species were measured inaccurately. Definitely, more accuracy of measurements were achieved in more common species as compared to those that are more rarely located.

The most efficient sampling method was definitely haphazard sampling (t = 4hr 24min) – taking an hour less than both the random and systematic sampling strategies (t = 5hr 6min and t = 5hr 4min, respectively). For most common tree species with frequencies >10% the densities were best estimated by random sampling (table 1). Rare species were difficult to measure with accuracy using any of the three sampling methods. 24 samples were not enough to accurately capture the number of species in this habitat.

Table 1. Percent error between the density of common and rare species (the two most common species were red maple and white oak, while my two rarest species were sweet birch and American basswood) for each of the three sampling methods used (systematic, random and haphazard).

 

	Systematic Sampling (% error)	Random Sampling (% error)	Haphazard Sampling (% error)
Red Maple	8.8%	0.1%	19.0%
White Oak	3.2%	4.4%	39.6%
Sweet Birch	n/a	558.3%	n/a
American Basswood	230.0%	n/a	n/a

 

Blog Post 4: Sampling Strategies

• Three sampling techniques that were used in the Virtual forest tutorial are Systematic, Random and Haphazard with 30 samples taken from each.
• The results show that systematic has the fastest estimated sampling time by over an hour and haphazard had the slowest sampling time.
• From the data below, the accuracy did in fact change with species abundance. Overall, the systematic sampling strategy was the fastest and the most accurate.  The comparison % errors are low across the different samples with the samples spreading out over a large area to allow for more accurate comparisons.

Systematic: 5 hrs, 5 min

Random: 6 hrs, 21 min

Haphazard: 13 hrs, 35 min

Comparison of % error between 2 common & 2 rare tree species:

Eastern Hemlock (Common), Sweet Birch (Common), Striped Maple (Rare), White Pine (Rare);

Systematic: 2.2%, 13%, 27%, 12%

Random: 19%, 7%, 11%, 12%

Haphazard: 15.9%, 4.8%, 51.9%, 49.7%

 

 

Blog post 4

 

For the virtual tree sampling experiment, three methods were used: systematic sampling along a topographic gradient, random sampling, and haphazard sampling. The fastest sampling technique was the systematic sampling along a topographic gradient which took 12 hours, and five minutes to complete. The tables present the two most common and rare species found during the different sampling techniques. It was also found that the systematic sampling technique along a topographic gradient had the highest accuracy for the sampling of common species. The random sampling of the tree species (Table 2.) had the highest accuracy and precision for the rare species in the tests. The haphazard sampling technique (Table 3.) had the lowest accuracy overall for both common and rare species. The overall accuracy was not affected by the species abundance in the studies. The results from the second field study (Table 2) had the highest accuracy overall for all three of the studies.

 

Table 1. Systematic sampling of tree species along a topographic gradient at Snyder-Middleswarth Natural Area.

 

species	# of individuals	density (stems/ha)	Percent error
Eastern Hemlock	141	96	25.80%
Yellow Birch	47	108.9	42.10%
Striped Maple	4	16	8.57%
White Pine	4	16	90.50%

 

Table 2. Random sampling of tree species at Snyder-Middleswarth Natural Area

 

species	# of individuals	density (stems/ha)	Percent error
Eastern Hemlock	152	633.3	34.70%
Yellow Birch	35	145.8	33.98%
Striped Maple	4	16.7	4.57%
White Pine	2	8.3	1.19%

 

Table 3. Haphazard sampling of tree species at Snyder-Middleswarth Natural Area

 

species	# of individuals	density (stems/ha)	Percent error
Eastern Hemlock	155	645.8	37.43%
Sweet Birch	48	200	70.21%
Striped Maple	2	8.3	52.57%
White Pine	1	4.2	50.00%

 

The results of the three area based sampling strategies used in the virtual forest tutorial are summarized below.

 

Systematic Sampling (12 hours, 7 minutes):

 

Most Common Species	Data Densities Actual Densities	% Error
Eastern Hemlock	504.2 469.9	7.3
Sweet Birch	112.5 117.5	-4.3
Rarest Species
Chestnut Oak	66.7 87.5	-23.8
Red Maple	137.5 118.9	15.6

 

Random Sampling (12 hours, 42 minutes):

Most Common Species	Data Densities Actual Densities	% Error
Eastern Hemlock	504.2 469.9	7.3
Sweet Birch	137.5 117.5	17
Rarest Species
Striped Maple	41.7 17.5	138.3
White Pine	20.8 8.4	147.6

 

Haphazard (12 hours, 59 minutes):

Most Common Species	Data Densities Actual Densities	% Error
Eastern Hemlock	540 469.9	14.9
Sweet Birch	108 117.5	-8.1
Rarest Species
Red Maple	41.7 118.9	-64.9
Striped Maple	8 17.5	-54.3

Systematic sampling was the most accurate for both common and rare species. % error was quite higher for the rarer species for all three methods.

The most efficient technique timewise was systematic, followed by random, followed by haphazard. The time differences weren’t’ too dramatic, with a spread of only 52 minutes between the most and least efficient.

Considering how close to the actual data the sample was in the systematic method, it would seem to be a sufficient sample size to have a solid understanding of the species numbers and abundance of the common species. However, the rarer species would seem to require further sampling to get more representative data.