MONITOR and the power of monitoring programmes
Contents
1. erroneously Because MONITOR uses random numbers for its simulations the results are a little different each time so your results will be a bit different Nevertheless it is clear that this monitoring programme will not give us the power of 2 0 90 that we want for an increase of 3 per year Press any key to move to the next page of output Power to Detect Negative Trends The power to detect negative trends is lower than the power for the corresponding positive trends That is because declining populations mean smaller samples as time goes on and less chance of showing a significant result whereas increasing population means larger sample sizes The power to detect a decline of 3 per year is far below our requirement of 0 90 Press any key again to see the last two pages of output graphs showing the power to detect trends You can save the results in a text file if you wish and open it with Notepad later You can then cut and paste results to another document Press F6 to open the Save Results dialogue box Enter the file name including the extension eg if you want it to be BEARS TXT you will have to type the TXT as well You will also be asked for a description to appear at the top of the file The file will be saved in the same folder as MONI TOR4 EXE Page 4of 5 E How does MONITOR do it For each trend line MONITOR does this 1 For each survey it calculates the mean result expected if the trend is actually occurr2. trees to forage From a particular observation point it is possible to get repeatable counts of the number of bears travelling to and from the grove on any given day Not all the bears in the sanctuary visit the grove and some are occasionally double counted so the daily counts from the observation point represent an index of the bear population s size not a true census of the population During a single season 15 separate day long counts of the bears were made which yielded a average of 15 6 bears observed per day and an accompanying standard deviation of 3 6 bears The following question arose among individuals concerned about the bear population s status would allocating the time of one park warden to count bears on 3 different days during peak of masting periods each season produce data useful for monitoring Dachigam s bear population Specifically would this intensity of monitoring effort exerted over a 10 year period be sufficient to detect annual trends positive and negative of at least 3 in the bear population at a probability of gt 0 90 Users Manual p 22 Let s recap the main points o there is only one spot where we can do the surveys we intend to do surveys every year for 10 years each survey will consist of 3 counts on separate days during our pilot study we recorded an average of 15 6 bears per day with SD 3 6 if there is an increase or decrease of 3 per annum we want to be 90 certain of picking it up and no
3. ie the first year Note that the default is 0 for MONITOR the first survey is Occasion 0 not Occasion 1 Our 10 surveys are Occasions 0 1 2 3 4 5 6 7 8 and 9 which are the defaults so just press Tab ten times You do not have to do one survey each year If you decided to do a 9 year programme with surveys every 2 years you would have 5 surveys and the Occasions would be 0 2 4 6 and 8 Or you could do a five year programme with surveys every 6 months 0 0 5 1 1 5 2 2 5 3 3 5 4 4 5 The surveys do not have to be equally spaced you might want to do surveys in years 1 2 and 3 and then in years 8 9 and 10 The time units do not have to be years if you are monitoring aphids on your roses months or weeks would be more appropriate for bacteria hours But the time units must be the same as those used for the trend so if the trend you are looking for is 3 per year the units for survey occasions must be in years Trends can be linear or exponential see graphs from the Manual For the bears we expect trends to be linear 70 r 70r Linear Trends Exponential Trends 60 60 o 50 o 50 F 40 a 40 E E s 30 30 lt 2 20 30 10 10 Q i fi i i J 0 4 8 12 16 20 0 Time Time Under TRENDS Type Linear pressing the space bar to change between Linear and Exponential Significance Level 0 05 Number of Tails 2 because the trend m
4. MONITOR and the power of monitoring programmes This introduction includes a brief explanation of the concept of power in the context of monitoring biological populations how to install and start MONITOR setting up an analysis in MONITOR an explanation of what MONITOR does to produce its results getting finished 0 OF Ov O On the way we will use MONITOR to design a monitoring programme for bears in Dachigam WS Kashmir Unlike most biological statistical software MONITOR is a study design tool it will not analyse your data A The power of monitoring programmes Here we are concerned with programmes designed to detect changes increases or decreases in animal populations over a period of time usually several years We can rarely produce an exact count of the population so our conclusion about the change is inferred from a series of samples There is always the risk of making an incorrect inference and two kinds of error can arise Type I We conclude that there is a change in the population when in reality there is no change Type II We conclude that there is no change when in reality the population has increased or decreased Tests of statistical significance are designed to avoid Type I errors or at least to reduce their occurrence to an acceptable level usually 0 05 or 5 of studies This method asks How likely are we to get these results if there is in reality no change Only if this likeliho
5. ay be upwards or downwards Page 30f 5 Constant Added is only needed for exponential trends and Trend Variation only applies if there is more than one plot See the Manual if you need to use these Rounding Whole as the counts of bears ona given day will be whole numbers Trend Coverage Complete so MONITOR will calculate results for all trends from 10 to 10 Replication 5000 the number of random trials that MONITOR will use to get the results Back in 1995 when this program was written computers were slow and running a complete set of trends with 5000 replications might take hours With modern computers we don t need to run individual trends or to be parsimonious with the number of replications Running a complete set of trends with 5000 replications took 35 seconds on my notebook W hen data entry is complete press F2 to run the simulation A box opens with a progress report and closes when the simulation finishes D The output from MONITOR Press F3 to see the first page of results Power to Detect Positive Trends Power to Detect Positive Trends 1 666 1 606 When I ran this power to detect a 3 increase was shown as 0 629 That means that out of the 5000 simulations that MONITOR ran for a real increase of 3 per year the change was detected for only 3145 62 9 For the remaining 1855 simulations 37 1 the results were not significant and the null hypothesis of no change would have been accepted
6. ing For example for a 3 decline the mean for the survey in the second year will be 15 6 3 15 13 It assumes that observations are normally distributed so it pulls 3 random values from a normal distribution with mean 15 13 and SD 3 6 to represent the 3 observations for that survey remember that we planned to count bears on 3 days each year Then it moves on to the next survey and does the same 2 When it has simulated results for one set of surveys it fits a straight line to the results and tests to see if the result is significantly different from a flat no change line 3 Then it repeats steps 1 and 2 many times depending on the Replication setting 5000 in our example and calculates the proportion that were significantly different from the no change result That proportion is the power for that trend the proportion of times that a statistically significant decline was reported when a 3 per year decline was actually occurring Notice that we do not expect to be able to say for sure how big the trend is just that there is a statistically significant change in the number of bears coming to the grove of mast fruiting trees MONITOR does the same thing for a trend of 0 That is it checks to see how often a significant change is found even though no change is occurring Yes that happens And the frequency should be about the same as the significance level if no change is occurring the data will show a statisticall
7. od is very small lt 5 do we reject the no change hypothesis and conclude that a change has occurred Power is the probability of rejecting the no change hypothesis when it is false A study with low power will probably conclude there is no change even when a change occurs Clearly that is likely to be a waste of time and money So we need to design programmes with high power MONITOR will calculate the power of a monitoring programme before you start so that you can make sure it has sufficient power or you decide not to bother B Getting started The situation with the MONITOR software is not clear as I write Please follow the link on wcsmalaysia org stats for the latest source to download the program and the User s Manual What follows refers to MONITOR version 4 1 Create a folder for MONITOR and download the file MONI TOR4 EXE to this folder As with any exe file it s wise to run a virus check with an up to date virus scanner after downloading it MONITOR does not install itself on Windows systems to run it you simply click on the exe file It is useful to have a desktop icon to start the program right click on MONI TOR4 EXE and select Send to desktop create a shortcut MONITOR is an ancient DOS program whichsometimes has problems with file names and long file paths used in current versions of Windows for example if it is located in the folder C Documents and Settings Your Name My Documents Da
8. t concluding that there is no significant change 0000 C Entering information in MONITOR 5 x j eee MONITOR eee Power Analysis for Population Monitoring Programs PLOTS TRENDS Number Monitored Type Counts Plot Survey Significance Level Initial Values Number of Tails Constant Added SURVEYS Trend Variation Number Conducted Rounding Occasions Trend Coverage Replications Whole Complete 66 Enter number of replications for simulating the monitoring program lt Fi gt Help lt F2 gt Run lt F3 gt View Results lt F4 gt Save Inputs lt F5 gt Load Inputs lt F6 gt Save Results lt F gt Go DOS lt F8 gt Run Batch lt F gt Converter lt F1 gt Quit Page 2o0f 5 Enter the information in the appropriate fields You can move between fields with the tab key or by clicking on the field Under PLOTS Number monitored 1 as there is only one plot to be monitored the grove of masting trees Counts Plot Survey 3 as counts will be made 3 times per survey ie per year Click on Initial Values and a box will open Enter these values Mean Initial Value 15 6 and Standard Deviation 3 6 these are the values we got during our pilot survey W eight 1 not used for this analysis as there is only one plot Under SURVEYS Number conducted 10 one per year for 10 years Note that the minimum number of surveys is three Click on Occasions and a box will open for Survey 1
9. ta analysis Monitor If this happens move the folder to C Monitor 2 Open MONITOR by clicking on the desktop shortcut or MONITOR4 EXE A splash screen appears press any key Updated 15 Oct 05 Page 1 of 5 MONITOR has a limited amount of help which you can access with the F1 key There is also a good users manual by James Gibbs www esf edu efb gibbs I m including large chunks of the manual in what follows A weakness in the manual is the description of how to enter information into the program do not attempt to understand that until you have worked through an example B Example the bears of Dachigam The following is an example of an application of MONITOR to guide the design of a monitoring program consisting of a single plot monitored over time The application focuses on a population of Himalayan black bears Selenarctos thibetanus in the Dachigam Wildlife Sanctuary Kashmir India The bears are a conspicuous inhabitant of the sanctuary but little is known of the population s status or trends Initiation of a small scale monitoring program for this population would be useful given the multitude of threats now facing the population Throughout most of the year black bears in Dachigam are scattered throughout the sanctuary and are very difficult to count At certain periods of the year however during the peak fruiting period for local mast bearing trees most bears in the sanctuary travel to a large central grove of masting
10. y significant increase or decrease about 5 of the time if 0 05 is your chosen significance level erroneously F Designing an effective monitoring programme So the proposed monitoring programme 3 days per year for 10 years does not have enough power How should we modify it to get the power we want In this case we can t change the number of plots but we can increase the number of days spent doing counts each year Maybe 5 would be enough In MONITOR change the value for Counts per Plot per Survey to 5 and run the program again Does that give us the power of 0 90 that we want for both increases and decreases If not experiment with other values There is a summary of results on page 24 of the manual In other situations we may be able to change other parameters to increase the power Maybe we could do more plots or quadrats or transects or do surveys at 6 monthly intervals instead of once a year Perhaps we can improve and standardize our methods to reduce the SD After thinking carefully we may decide to change the significance level All of these changes we can feed into MONITOR to see if they yield the power we want The manual has two other examples which you may like to work through G Getting finished That s very easy Press F10 and then Y or Esc to quit MONITOR Page 5o0f 5
Download Pdf Manuals
Related Search