Simulation for Sites and Replicates for Occupancy

“How many camera traps should I put?”

“How many days should I leave my camera traps for?”

Usually when people are doing occupancy estimates, these are the questions that many have. I had one. One way to answer this is doing simulations. So, I tried.

I tried 1000 simulations for three conditions.

1. Low occupancy (0.20) with high detection (0.80).

2. Moderate occupancy (0.50) with moderate detection (0.50).

3. High occupancy (0.80) with low detection (0.20).

For each of these conditions, I had three study designs. First, 100 sites with 20 replicates. Second, 20 sites with 100 replicates in each. And third, 70 sites with 30 replicates.

High sites with low replicates seem to be the best design when our real occupancy is low and detection is high.

Even when both occupancy and detection moderate, higher sites with lower replicates seem to be the best option.

And even when occupancy is high and detection is low, higher sites and lower replicates seem to work the best. However, the detection CI is more narrower for 70 sites with 30 replicates.

Conditions	N= 100, k =20		N=20, k =100		N=70, k = 30
	Occupancy CI	Detection CI	Occupancy CI	Detection CI	Occupancy CI	Detection CI
Occupancy – 0.8 Detection – 0.2	0.716-0.873(0.8)	0.18-0.221 (0.199)	0.65-0.95 (0.797)	0.181-0.221 (0.2)	0.715-0.901 (0.801)	0.181-0.219 (0.2)
Occupancy – 0.2 Detection – 0.8	0.13-0.28 (0.199)	0.76-0.841 (0.801)	0.05 – 0.35 (0.201)	0.752-0.85 (0.796)	0.114-0.286 (0.198)	0.762-0.841 (0.801)
Occupancy – 0.5 Detection – 0.5	0.4-0.59 (0.498)	0.472-0.534 (0.5)	0.3-0.7 (0.502)	0.468-0.532 (0.5)	0.386-0.614 (0.499)	0.47-0.529 (0.5)

If you want, you can use your own values for occupancy, detection, sites and replicates, and even the number of simulations that you want to run. Here is the link to the full code in GitHub

https://github.com/Ljeevan9211/Simulations-in-R/blob/master/First_trial_Sample_Required_Occupancy.R.

This is my first trial with writing simulations all by myself. It would be hugely helpful for me if you comment or message me if you find anything weird or have any improvements.