Input data may be either 'dependent' or 'independent'. Dependent data consists of species counts by a 'primary' observer and species counts by a 'secondary' observer. These counts are collected at multiple 'stops' along a sampling route. At each stop, the primary observer records the number of birds of each species he/she hears. The secondary observer notes which birds were observed by the primary observer and records only birds which were not observed by the primary observer.
Observers switch duties (primary/secondary) at each stop (collection point).
"Independent" data consists of counts of species observed by both observers and counts of species detected by each observer which were not detected by the other observer.
To use program DOBSERV, an input file must be prepared. This file must be an ASCII 'text' file (no word-processing or spreadsheet format codes). It can easily be created from Excel (or other spreadsheet program) by exporting to a 'comma-separated-variable' file. If a word-processor is used, it could be created by saving as a 'MSDOS text' or 'ASCII text' file.
The following is a subset of a sample data file:
OBS1= 6,OBS2= 1,DATE=06/30/98,ROUTE=46040 6,AMRO, 5, 0 6,GRCA, 2, 0 6,HOSP, 2, 0 6,MODO, 0, 1 6,HOFI, 2, 0 6,SOSP, 1, 0 1,NOMO, 1, 1 1,AMRO, 3, 0 1,NOCA, 1, 0The first line of the input file is ignored by program DOBSERV and could contain comments or a description of the data in the file. In the example it tells the date, route and observers.
All succeeding lines must contain 4 (for dependent data) or 5 (for independent data) items, separated by commas. The example is dependent data, so there are 4 items.
The first item is the code signifying which observer is the 'primary' observer. (This is ignored for independent data.) In the first 6 observations, observer number 6 was the 'primary' observer, and in the next 3 observations, observer number 1 was the 'primary' observer.
The second item denotes which species was observed. This should be a 4- character abbreviation (for formatting purposes). (AMRO=American Robin)
The third item is the number of birds of the species which were detected by the 'primary' observer (regardless of whether the secondary observer detected them). In the example, the first observation indicates that the primary observer detected 5 AMRO's.
For independent data, the third item would be the number of birds of the species which were detected by the first observer which were not detected by the second observer.
The fourth item is the number of birds of the species which were detected by the 'secondary' observer which were not detected by the primary observer. In the example, no additional AMRO's were detected by the secondary observer, but 1 MODO (fourth observation) was detected that was not detected by the primary observer.
For independent data, the fourth item would be the number of birds of the species which were detected by the second observer which were not detected by the first observer.
For independent data, the fifth item would be the number of birds of the species which were detected by both observers.
Once the data are entered in Excel, click 'Save as', change the file type to 'comma-separated-variable (csv)', give it a meaningful name and click 'Save'. Then close Excel.
A form will appear with an abstract describing the methods and a single button labeled '>>Step 1 - Convert data to SURVIV input'.
Click this button and a dialog window will appear asking for the name of the input file. Select the dependent sample data file, 'sample.csv' and click the 'Open' button. (A reminder of the sample filenames appears on the DOBSERV form.)
After selecting the file, a dialog box appears asking if this is an independent or dependent data file. Click 'No' for dependent data.
Next, the program displays a list of species with four buttons to the right. After each species abbreviation, you'll see a number in parenthesis() and a number in square braces[]. The number in parenthesis is the total number of observations of that species for both observers. The number in square braces is a 'group number' which you can assign. The purpose of the 'group number' is to allow you to combine detection probabilities across groups of species. So, in addition to the model with a single detection probability for all species, p(.), and the model with species-specific detection probabilities, DOBSERV will run a model with the detection probabilities grouped as specified.
Usually, there isn't enough data for every species to produce 'good' estimates by species. For those species where there is not enough data, we can estimate N by assuming the detection probability for those species is the same as some other species. To do this, we need to assign the 'rare' species to a group number, which will be the same group number as a similar species which has enough data. You can do this (assign groups to species) at this point, or later (when you find out that the analysis didn't work due to insufficient data).
By default, DOBSERV will group species with low numbers of observations (less that a cutoff value (eg., 10)) for you automatically.
Next, click the button labeled '>>Step 2 - Run SURVIV with input file'. Another window should appear briefly, then disappear. Click '>>Step 3 - Pick Best Model'.
Summary statistics from several models will appear with 'radio buttons' to the left. The top model is selected by default. Click '>>Step 4 - Compute estimates of N'.
Looking at the output, you might notice that DOBSERV did not compute standard errors. This is due to a problem with insufficient data for the 'best model' which was selected in step 3. Even though data for species with low observation counts were pooled, standard errors could not be computed when there are no observations for the 2nd observer.
To correct this problem, we need to find which species had no detections by the 2nd observer. Click the 'view' menu, and select 'View SURVIV output - estimates of p'. Look at each COHORT to see how many were detected by only observer #1. You'll notice that for PUMA, it says 'COHORT=13; 13:...', meaning that of the 13 observed, 13 were detected by observer #1. For CAGO, it says 'COHORT=20; 0:..' meaning that of the 20 observations, zero were seen by observer #2.
click the 'step' menu to go back to step 2. Scroll the species list down until you see the species names 'CAGO' and 'PUMA'. Click on 'CAGO' in the list, then click the button labeled 'Force species into group x for all models'. Click 'PUMA', then 'Force species...'. The list will be modified to indicate these selections (x at far right). This tells the program to pool those species with the other low-detection species, even though they have more detections than the specified cutoff value (10).
Next, repeat steps 3 and 4. The output of the N's will appear after the previous output (without SE's). You can view the full SURVIV output by going to the 'view' menu.
Click 'File', and 'Quit' to exit the program.