Catch-Effort Surveys

Seems to be a lot of interest/concerns about catch-effort surveys. Instead of hijacking another thread, let's start a new thread.

Since I am a positive person, let's start on the good. What good can be gleaned from them?

It's a tool that can be used to monitor population trends.

I got one from the Iowa DNR earlier this week

These surveys are one tool that can shed light on trends in wildlife. The general idea is that given equal effort, you catch more animals when there are more animals on the landscape.
A simple example would be if you set 20 mouse traps in your barn for a week each year. The first year you catch 50 mice. The next year you set the same amount of mousetraps for the same amount of time and you catch 30. One explanation is there the mouse population declined. Of course, there are a bunch of assumptions made – such as your skill level didn’t dramatically increase, or the mice got harder to catch (maybe more spilled grain around). For any one trapper, it is impossible to meet these assumptions. That is why these surveys involve a huge number of people. If you randomly pick 1,000 people, or a few thousand people, and ask for this information, these assumptions are less important.

Here is Wisconsin, about 1,500 people got a survey asking about how many beaver traps were set and how many beaver were caught in those traps. This is what we call a catch-per-unit effort survey. We randomly pick 1,500 names to receive the survey. Some of those people don’t trap beaver, or don’t trap during the time period we are talking about. Some of those people have great access and have trapped forever and are really good trappers. Others are not as experienced and not as good. But, if we randomly pick 1,500 again next year, it is likely that we will again have a similar mix of skill and situations. The information for any one trapper isn’t that useful. If, for example, Nimsy has great access one year and then he gets kicked off the following year, that may explain some of his decreased success rates and it may not have anything to do with the overall beaver population trend. But if we have a huge sample of trappers, like hundreds or thousands of trappers, these individual situations sort of even out.

Some of the critiques are:
What about weather? The longer the surveys have been done, the easier it is to account for this. After a few years, you can start to account for the variation that weather has.

What about different levels of skill? This is why it is really important to survey a lot of trappers. If you randomly pick 1,000 (just an example) trappers each year, it is reasonable to assume that each year you are going to get some good trappers, some not good trappers and some in between. With a really large sample size of trappers, you should get a sample that is fairly representative of trappers as a whole. There will always be some outliers. Jim Bridger may have a great success rate while I have a terrible success rate, but with a large group this evens out.

What about animal vulnerability over time? Some animals are easier to catch early in the season or late in the season. Typically, the information is looked at in small chunks of time to account for this. 1,000 coyote traps set in the breading season is not the same as 1,000 traps set in late October. So, we have to account for seasonal vulnerability.

Some strengths are:
These surveys may be one of the only sources of information. Some species simply don’t have a lot of information. It’s really hard, and expensive, to do marking studies and such for many animals. These surveys can provide some information on species that are otherwise hard to monitor.

These surveys can cover a large expanse. As trappers, we often think about the population on the few townships we are trapping on. Agencies are thinking about much larger areas. These surveys allow for information across a much larger area than field surveys can allow.

These surveys are cheap. If you need data, these surveys are a cost-efficient way to get a lot of information. For the cost of one good GPS collar, we can survey 1,000 trappers.


These surveys are not perfect, but they are useful. Like any data source, there are some flaws. They are designed to look at small scale trends, like a county level, and are not really for year to year changes. They are for long term trends over large areas.

Here is some detailed information if interested.
Roberts_Cornell
Chapter 1 talks about it in detail and chapter 3 talks about it as it relates to management


Happy to answer questions – just be patient as I am about to sight in the rifle for a long weekend of hunting.

-Nathan

What assumptions are you making in your example for beaver trappers in WI? Do we assume that skilled and unskilled fill out the survey at the same rate? What if skilled trappers were twice as likely not to fill out the survey? What if price impacted a certain segment of the population to respond differently on the survey. Would it make a difference in the results?

How do you test to make sure your don't have false assumptions impacting your results. I guess in this case, one sign would be significantly different response rates on surveys from one year to the next?

Nathan, I recieved one of the surveys this year. I work seasonal and I don't get much time in Nov. why wouldn't they do it over a season instead of a month? Just curious because next week is deer season and a lot of people don't go trapping during it so the survey is only good for three weeks in my situation at most {and a lot of others}.

Not saying not to do these, but if you just want trend, just send a questionnaire asking each trapper if he is seeing more or less sign. Ask them about all the animals on their lines.

" As trappers, each and every one of you, if you're indeed out in the field on a regular basis (I'm out there every day), can be taught to do your own surveys and can keep a pretty well-tuned eye on what the predators, as well as the prey species, are doing in terms of population trends."

Source: Gulo

Let's take beaver. If there are half the active houses on my line as opposed to the year before and I trap the active houses my catch rate is not going to change. Or is it?

The major assumption that has to be valid for catch-effort surveys to be valid for making comparisons across time and space is detection or capture probability have to be equal across all participants, all years, and all locations. This assumption is all but impossible to meet when dealing with uncontrolled systems.

Some of the factors that can affect detection include but are not limited to population sizes, behavior or target animals, behavior of animals in response to other competitors or prey, movement, populations of incidental non-target species, weather, food availability, vulnerability to methods, trap happy or trap shy animals, etc. Other factors affect detection through the trapper including but not limited to experience, effort (# nights, # traps, # properties, area covered, etc.), what a trapper targets, fur prices, gas prices, if the trapper is fur trapping vs. nuisance trapping, etc.

Some of these assumptions can be relaxed if additional information is collected but there is little chance that all will be exactly the same everything. However, if a catch-effort survey is set up right, detection probability can be estimated and accounted for among years and locations, thus allowing for more accurate comparisons.

How important is it to account for detection probability? Several studies show false negatives (misses) as low as 5%) can greatly change perceived population trends or relationships. So, if you miss 1 animal per 20 traps, you are missing enough to falsely affect perceived population trends. If you gang set and get a stubborn coyote, you miss on multiple traps which increases the bias. If a grinner is in your trap and the coyote passes by but can't be caught, your catch effort will also be biased low although the coyote was there, which is also a false negative.

With all this said, catch-effort surveys which are indices are starting to phase out and be replaced by occupancy estimates which tend to estimate and account for these biases from these controllable and non-controllable factors that affect detection. These newer methods were first used with wildlife in the early 1980's but only recently started receiving more attention with nearly all articles in major wildlife journals including detection estimation. Despite ample examples of how catch effort designs are often inaccurate and biased, many biologist are still resistant to changing to better methods, even if it would require only minor adjustments in the methodologies. I was one of them at one time. However, using data provided by trappers will likely always have substantial bias making it limited in use simply because it is doesn't include misses.

Did someone say survey?

The assumptions we make are that when we survey a randomly selected 1500 ppl that those ppl are just as random as the year before. We also assume that biases from reporting (such as successful ppl being more likely to report than non-successful) are relatively consistent. In other words- if some subset is more / less likely to report for some reason, than that is true across years. It’s not perfect, but it is an index with value. Again, that is why these are done with huge sample sizes and a random draw.

On the beaver example- we ask about years experience trapping and account for that. We have an idea of the distribution of years experience of trappers as a whole and can compare that to the responses we get from ppl who fill out the survey. The main concern is making sure that if only rookies or only old timers fill out the survey, that those trends hold true the next year.

We do ask ppl, about 2000 a year, if they think populations are increasing, decreasing, stable etc. Its hard to put a number to those other than “x% believe the population is greater than the year before”. It’s useful and another source of information that is looked at.

We picked November because last year we asked when trappers do the majority of their beaver trapping. November was the top selection. You generally get more precise data from a short period of time than a long period because ppl are less likely to have trouble remembering what they did 4 months ago. Again, not perfect, but useful.

The assumptions of detection get a little trickier. If gear changes (ban on body grips for example), this causes problems. We look at it by type of trap to help account for this. The same is true for weather and time. I can get into the statistical analysis of it, but for now I’ll just say we account for it.

I think detection prob is more of an issue for things like sign surveys. I’ve published papers on this too (river otter). I would struggle with how to interpret it for abundance issues. An example might be a Wisconsin marsh. We could have a marsh with 1,000 huts on it. A decade later it has 100 huts. However, it is still occupied and if we looked at occupancy only we would say it had a high prob of occupancy a decade ago and still has a high prob of occupancy despite a 90% decline.

I’m not arguing catch-effort is perfect, but it provides a lot of useful information for low costs. It provides another way for trappers to be involved in gathering data and informing decisions too. It is one source of information that is combined with others to help understand what is happening across a large landscape. We also use age/at-harvest models, harvest-independent surveys like otter flights, winter tracking, cameras etc, opinion surveys, and make-recapture /gps/telemetry studies. Combined, these information sources help me and others when we develop management recommendations, answer policy makers, and fight never-ending litigation.

If I missed any points, let me know. Again, be patient as it is the busiest time of the year. Be safe out there.
Nathan

K.I.S.S. I try to apply this to problem solving.

I remember one time the area biologist wanted to spend all kind of money doing Supe surveys and other expensive stuff ( which was never going to happen) and the old retired biologist on the Board of Game asked, "Can't you just hire a super cub and fly around for a day and see whats going on out there?"

Sounds like Spraker...

Wish we had more guys like that.......

Jack

Hope you found some time to get out in the woods.

So when one survey tells you that majority of trappers do trapping in November, how do you ensure you are not biasing your survey by eliminating a subsegment of the population by then only asking about trapping in November?

While the majority might be out in the beginning, many of the big beaver guys I know are out in the spring. Now when you send them a report, they tell you they didn't trap beaver?

Probably more of a gauge of engagement level of the sportsmen than anything else.