Challenge to all Trout Camps

[pcray1231]

I have a basic understanding of statistics. It looks like what your saying is there was collinearity in the regression model?

Yes.

It would seem like the fact that zero sympatric streams had a barrier but yet gradient was not drastically different would support a causal link to barriers.

I'm not sure I see the same? Brook trout streams were smallest and highest gradient. Brown trout streams were largest and lowest gradient. Sympatric streams fell in the middle on both accounts.

I think you are going by Table 1, and comparing sympatric to brook trout with barriers. Correct? It's a fair point, that none of the sympatric streams had barriers. But also look at brook trout streams with barriers vs. brook trout streams without barriers. They are, on average, even more brookie style than brook trout streams without barriers.

Looking at figure 4 is fairly telling. The "tell" would be if there were black circles in the lower left, but no black triangles. i.e. streams expected to be brown trout but were brookies instead. There are 2 black circles kind of down there. But right among them are several black triangles (brook trout streams without barriers). You can kind of drawn a diagonal line from upper left to lower right that separates brook trout from brown trout water based on size and gradient. That's your base "rule", the major correlation here. There are a few exceptions. Those exceptions can important, but more of the brookie favoriting exceptions, i.e. brookie streams to the lower left of the line, actually don't have barriers. So it's not showing a barrier effect. If you could have 1 factor that explained most of the exceptions, then you could say, yeah, that might be matter. But that's not the case here.

In this study, barriers were primarily located on waters that would be expected to be brookies anyway. In the exceptions, where brookies are present when other factors say you could have browns, more of them don't have barriers than those that do. I'm not saying I'm against barriers. I think there are situations where they are likely beneficial. We're on the same team. Just saying the data in this study can't really be used to prop it up. If anything it tells you that they aren't needed EVERYWHERE. Which is something I think we'd have agreed upon as well.

When the benefit of something like a barrier, where being beneficial is the exception rather than the rule, it's going to be real hard to find it in statistics. Case studies are a better approach. Find a stream that SCREAMS browns, but has a barrier and is brookies due to it. Or make one, by taking a sympatric stream with a sizable % of browns, erect a barrier, remove the browns, and monitor the brookie population.

 

[silverfox]

The beauty of science and peer reviewed research is asking questions and testing others' hypothesis. This is how we learn. The mistake is in accepting all published papers as the gospel truth without critical thinking and analyzation.

I agree. However, this implies that people have blindly adopted published papers without thinking critically about the findings, which is a pretty strong accusation.

 

[lycoflyfisher]

Silverfox, no direct accusation. However you can find examples of such action in all facets of the scientific community especially when politics and big money is considered. Examples all over from medical fields (pharmacy), chemical research, climate issues, etc. Not to mentioned the studied and well intentioned use of biological controls for invasive species that have gotten out of hand.

Science is rarely settled especially in the real world outside of controlled laboratory settings.

 

[Fish Sticks]

So comparing brook trout streams with no barrier and sympatric streams like you said, if barriers did not have some causal effect on the fact that 100% of sympatric trout streams did not have one I would expect gradient and stream size to “come along for the ride” as you mentioned before that barriers did. It appears that 11.3 to 11.9 for gradient and 5.0 to 5.2 for size are very similar and the barrier effect is highlighted by this it would seem.

 

[FarmerDave]

I agree and this obviously was peer reviewed before it made it to our forum. I think “peer” is the operative term and I would be interested hear that review process from the fisheries scientists that added it to the EVTJV science and data page as well in addition to ours.

I agree as well, but if statistical experts were not included in the peer group, than...

One does not have to also be a stream biologist as well as a statistical expert to review or evaluate something like this from a statistical stanfpoint. In fact I could argue that it could be better if the statistical expert was totally independent, and not a stream biologist in order to avoid existing personal beliefs from creeping in.

I haven't finished reading the study, but I'd say that pcray explained it from a statisitcal standpoint extremely well in 317. Wouldn't it make more sense to argue his point, rather than whether or not it was peer reviewed or whether he is qualified?

 

[Fish Sticks]

I agree as well, but if statistical experts were not included in the peer group, than...

One does not have to also be a stream biologist as well as a statistical expert to review or evaluate something like this from a statistical stanfpoint. In fact I could argue that it could be better if the statistical expert was totally independent, and not a stream biologist in order to avoid existing personal beliefs from creeping in.

I haven't finished reading the study, but I'd say that pcray explained it from a statisitcal standpoint extremely well in 317. Wouldn't it make more sense to argue his point, rather than whether or not it was peer reviewed or whether he is qualified?

It actually probably was a statistician who went over the study designs and ran the data. Thats generally how these things work, if your at a University you have an academic in statistics check the model you select or suggest a better one.

 

[FarmerDave]

It actually probably was a statistician who went over the study designs and ran the data. Thats generally how these things work, if your at a University you have an academic in statistics check the model you select or suggest a better one.

Probably?

That word does not compute and really isn't an argument at all.

Now please try to splain how Pat's analysis is wrong from a statistical standpoint. I can't. He splained his point very well IMHO and didn't even disagree with the basic conclusion, Just that barriers won't make a difference in some/many cases. However, they might in others.

 

[Fish Sticks]

I agree as well, but if statistical experts were not included in the peer group, than...

One does not have to also be a stream biologist as well as a statistical expert to review or evaluate something like this from a statistical stanfpoint. In fact I could argue that it could be better if the statistical expert was totally independent, and not a stream biologist in order to avoid existing personal beliefs from creeping in.

I haven't finished reading the study, but I'd say that pcray explained it from a statisitcal standpoint extremely well in 317. Wouldn't it make more sense to argue his point, rather than whether or not it was peer reviewed or whether he is qualified?

To summarize my second to last post:

if there was an inseparable effect between stream size, gradient, and barrier or not on the out come of allopatric brook trout presence or total absence, my point is why does the allopatric brook trout stream group with no barrier and sympatric stream group have almost the same size and gradient?

If stream size and gradient were really what was driving species present you would expect to see that trend continue in groups with no barrier such as the two compared above.

 

[Fish Sticks]

Probably?

That word does not compute.

Yes it meaning most likely because when your at a university there are people who are there to study only statistics. It is extremely common place for those studying fields outside of stats to submit their models and data to these departments and working with the data can actually become course work for those studying statistics. I used the word pribably because I don’t know what the situation is at Allegheny university despite this being the norm

 

[FarmerDave]

I agree. However, this implies that people have blindly adopted published papers without thinking critically about the findings, which is a pretty strong accusation.

Not really.

Are you suggesting nobody ever does this?

What's worse is criticizing someone for questioning it.

 

[silverfox]

Not really.

Are you suggesting nobody ever does this?

What's worse is criticizing someone for questioning it.

No. In context, I interpreted it differently.

 

[FarmerDave]

View attachment 1641227974

So comparing brook trout streams with no barrier and sympatric streams like you said, if barriers did not have some causal effect on the fact that 100% of sympatric trout streams did not have one I would expect gradient and stream size to “come along for the ride” as you mentioned before that barriers did. It appears that 11.3 to 11.9 for gradient and 5.0 to 5.2 for size are very similar and the barrier effect is highlighted by this it would seem.

keep in mind, I still haven't read the whole thing. Anyway...

I don't see your logic. Especially when it comes to your last sentence. Barrier is not even in that equation so it doesn't seem that way to me.

I can come up with several comments and questions.

As Pat said, the streams with barriers are all very small and high gradient where brook trout have the advantage. So they would likely be brook trout streams anyway. Can we agree on that and move on?

Now lets look at the Brown trout streams. Assuming this means all brown trout and no brookies. Notice those are significantly larger and with lower gradient where brown trout have a clear advantage. Makes sense, right?

So, we are comparing just columns 2 and 4 which are similar in size and gradient, so lets lump them together for a minute. What exactly does Sympatric mean. Does it mean at least 1 of each? Do we know the composition of the biomass? Note that over half of the streams in this size class are Brook Trout only, the rest have some brown trout. Care to bet which species is more common on those 13 Sympatric streams? This tells me that this size class of stream also favors brook trout with all other things being similar. The study does not prove this, but there are strong indications.

So, do I think barriers on those sympatric streams would be beneficial to brook trout? Probably, but the study does not prove that. Just because the tiniest streams have barriers and only brook trout does not suggest (to me) that the reason is the barriers. The left column becomes basically irrelevant in that comparison. One only needs to stay at a Holiday Inn Express to see that.

 

[FarmerDave]

No. In context, I interpreted it differently.

I didn't see it as finger pointing, but since he didn't "like" any of my posts, maybe you are right. 😁

 

[pcray1231]

View attachment 1641227974

So comparing brook trout streams with no barrier and sympatric streams like you said, if barriers did not have some causal effect on the fact that 100% of sympatric trout streams did not have one I would expect gradient and stream size to “come along for the ride” as you mentioned before that barriers did. It appears that 11.3 to 11.9 for gradient and 5.0 to 5.2 for size are very similar and the barrier effect is highlighted by this it would seem.

If you compare brown trout streams to any other category, the differences are striking. Lower gradient, larger streams.

Sympatric streams and "brook trout without barriers" are very similar. I agree with you. But neither category has barriers, so in no way does that say anything about the effectiveness of barriers. Maybe the biggest thing I see is elevation, which may lead into geologic formations underlying. But, it wasn't enough to be statistically significant, so.... The only other point I'll say is that these might have a lot of overlap. It's possible/likely that some of the "brook trout without barriers" are actually sympatric, they just didn't find that odd brown in the mix during the study. Also possible/likely that some of the sympatric streams are brook trout dominated, but they did happen to find the occasional brown. Several of the streams in these two groups might switch categories if you were to re-do the study, but probably not all.

Only 1 category here has barriers. Only 1 category can tell you anything about barriers. If you compare brook trout streams WITH barriers to any other category, including brook trout without barriers, brook trout streams with barriers are significantly higher gradient and smaller. They lean more towards brookies based on size and gradient, and therefore the barrier doesn't tell you much.

Again, it also does not tell you they DON'T work. It says nothing either way. Inconclusive.

 

[Fish Sticks]

keep in mind, I still haven't read the whole thing. Anyway...

I don't see your logic. Especially when it comes to your last sentence. Barrier is not even in that equation so it doesn't seem that way to me.

I can come up with several comments and questions.

As Pat said, the streams with barriers are all very small and high gradient where brook trout have the advantage. So they would likely be brook trout streams anyway. Can we agree on that and move on?

Now lets look at the Brown trout streams. Assuming this means all brown trout and no brookies. Notice those are significantly larger and with lower gradient where brown trout have a clear advantage. Makes sense, right?

So, we are comparing just columns 2 and 4 which are similar in size and gradient, so lets lump them together for a minute. What exactly does Sympatric mean. Does it mean at least 1 of each? Do we know the composition of the biomass? Note that over half of the streams in this size class are Brook Trout only, the rest have some brown trout. Care to bet which species is more common on those 13 Sympatric streams? This tells me that this size class of stream also favors brook trout with all other things being similar. The study does not prove this, but there are strong indications.

So, do I think barriers on those sympatric streams would be beneficial to brook trout? Probably, but the study does not prove that. Just because the tiniest streams have barriers and only brook trout does not suggest (to me) that the reason is the barriers. The left column becomes basically irrelevant in that comparison. One only needs to stay at a Holiday Inn Express to see that.

That was my point that the barrier was not in the equation for sympatric vs. brook trout with no barrier. The point I was trying to make was that if smaller high gradient was soley driving brook trout presence then when you take barrier out of the equation a stream with brown encroachment should have a different characteristic of gradient or size than one without brown encroachment but those two above groups are almost exactly the same in size and gradient. So ergo something else is driving the fact that there brown trout are able to encroach in those sympatric streams if size and gradient are essentially equal.

Then once you see that size and gradient had seemingly no determination of allopatry vs. sympatry between those teo groups the fact that half of allopatric brook trout streams had barriers, half didn’t, but zero sympatric streams had barriers seems telling of a barrier effect.

 

[Fish Sticks]

If you compare brown trout streams to any other category, the differences are striking. Lower gradient, larger streams.

Sympatric streams and "brook trout without barriers" are very similar. I agree with you. But neither category has barriers, so in no way does that say anything about the effectiveness of barriers. Maybe the biggest thing I see is elevation, which may lead into geologic formations underlying. But, it wasn't enough to be statistically significant, so.... The only other point I'll say is that these might have a lot of overlap. It's possible/likely that some of the "brook trout without barriers" are actually sympatric, they just didn't find that odd brown in the mix during the study. Also possible/likely that some of the sympatric streams are brook trout dominated, but they did happen to find the occasional brown. Several of the streams in these two groups might switch categories if you were to re-do the study, but probably not all.

Only 1 category here has barriers. Only 1 category can tell you anything about barriers. If you compare brook trout streams WITH barriers to any other category, including brook trout without barriers, brook trout streams with barriers are significantly higher gradient and smaller. They lean more towards brookies based on size and gradient, and therefore the barrier doesn't tell you much.

Again, it also does not tell you they DON'T work. It says nothing either way. Inconclusive.

The other way that you could look at the 91% brown trout streams did not have a barrier is how did the brown trout get above a 4m barrier in the other 9%?

It was likely people or the barrier was constructed post invasion.


I am not denying that stream characteristics are important in species presence(not that scientists know how they all interact with each other to predict on a stream by stream
Basis yet). I am just saying that almost all streams with brown trout had no barriers and that not all streams that had brook trout only had different size and gradient from sympatric streams with brown trout brown trout which would makes the barrier the elephant in the room. Yes the pure brown trout streams are bigger lower gradient but I think the mesh point for native brook trout and invasive brown trout where there are sympatric populations is where we would likely be talking about future barriers in alot of cases so whats going on their likely more relevant to informing management

 

[pcray1231]

That was my point that the barrier was not in the equation for sympatric vs. brook trout with no barrier. The point I was trying to make was that if smaller high gradient was soley driving brook trout presence then when you take barrier out of the equation a stream with brown encroachment should have a different characteristic of gradient or size than one without brown encroachment but those two above groups are almost exactly the same in size and gradient. So ergo something else is driving the fact that there brown trout are able to encroach in those sympatric streams if size and gradient are essentially equal.

Then once you see that size and gradient had seemingly no determination of allopatry vs. sympatry between those teo groups the fact that half of allopatric brook trout streams had barriers, half didn’t, but zero sympatric streams had barriers seems telling of a barrier effect.

But there is a general, obvious, overwhelming trend towards high gradient and smaller favoring brookies.

The "brook trout without barriers" and sympatric fall in the gray area in between based on averages. Yes, other factors are at play that weren't really captured. I mentioned elevation, but it's a small signal and doesn't overcome the noise. Doesn't mean it not at play, but you can't prove it statistically. And accuracy of determining sympatric vs. allopatric brookies likely adds noise as well, that is not an easy thing to do with any certainty, I'm quite confident at least a couple of the "brook trout without barriers" category actually could be sympatric. Plenty of other potential sources of noise that can hide a signal, right down to some guy with a backpack of brown trout fry 80 years ago had to take a dump the day he stocked stream X, and thus turned around earlier, but had a girl with him when he went to stream Y and went all the way up and made a day of it.

But the brook trout with barriers were the smallest and highest gradient group. Based on the overriding trend, the most likely to be brookies.

Statistical methods are by and large getting at averages. That's what they do, they compare averages of a population and determine what the strongest correlation coefficients are. And barriers are lost in the shuffle. Again, I am NOT telling you they don't work, I think they probably do in some circumstances. Just that the data collected in this particular study doesn't support making a call either way.

And if the situations where barriers are beneficial are the exceptions rather than the rule, a statistical approach of comparing averages isn't going to work. You can use them to make the base expectation, and then you analyze the exceptions. Is there a stream that absolutely screams brown trout stream based on all other factors, but has a barrier and is brookies? Statistics will only tell you it's an outlier, you have to figure out why. And even going there, in this study, there are 2 streams with barriers that are brookies only and would otherwise be expected to be browns or sympatric. Great, would love to read more about those 2 streams. But, there are more than 2 streams without barriers that are brookies only and would otherwise be expected to be browns. Maybe the 2 exceptions that have barriers, that's the reason, and the ones that don't have barriers stay brookies for some other reason, I dunno. But a straight up statistical crunch of the numbers does not show it.

 

[FarmerDave]

That was my point that the barrier was not in the equation for sympatric vs. brook trout with no barrier. The point I was trying to make was that if smaller high gradient was soley driving brook trout presence then when you take barrier out of the equation a stream with brown encroachment should have a different characteristic of gradient or size than one without brown encroachment but those two above groups are almost exactly the same in size and gradient. So ergo something else is driving the fact that there brown trout are able to encroach in those sympatric streams if size and gradient are essentially equal.

Then once you see that size and gradient had seemingly no determination of allopatry vs. sympatry between those teo groups the fact that half of allopatric brook trout streams had barriers, half didn’t, but zero sympatric streams had barriers seems telling of a barrier effect.

I still don't see your logic, but IMO pcray's logic is very sound.

 

[Fish Sticks]

But there is a general, obvious, overwhelming trend towards high gradient and smaller favoring brookies.

The "brook trout without barriers" and sympatric fall in the gray area in between based on averages. Yes, other factors are at play that weren't really captured. I mentioned elevation, but it's a small signal and doesn't overcome the noise. Doesn't mean it not at play, but you can't prove it statistically. And accuracy of determining sympatric vs. allopatric brookies likely adds noise as well, that is not an easy thing to do with any certainty, I'm quite confident at least a couple of the "brook trout without barriers" category actually could be sympatric. Plenty of other potential sources of noise that can hide a signal, right down to some guy with a backpack of brown trout fry 80 years ago had to take a dump the day he stocked stream X, and thus turned around earlier, but had a girl with him when he went to stream Y and went all the way up and made a day of it.

But the brook trout with barriers were the smallest and highest gradient group. Based on the overriding trend, the most likely to be brookies.

Statistical methods are by and large getting at averages. That's what they do, they compare averages of a population and determine what the strongest correlation coefficients are. And barriers are lost in the shuffle. Again, I am NOT telling you they don't work, I think they probably do in some circumstances. Just that the data collected in this particular study doesn't support making a call either way.

And if the situations where barriers are beneficial are the exceptions rather than the rule, a statistical approach of comparing averages isn't going to work. You can use them to make the base expectation, and then you analyze the exceptions. Is there a stream that absolutely screams brown trout stream based on all other factors, but has a barrier and is brookies? Statistics will only tell you it's an outlier, you have to figure out why. And even going there, in this study, there are 2 streams with barriers that are brookies only and would otherwise be expected to be browns or sympatric. Great, would love to read more about those 2 streams. But, there are more than 2 streams without barriers that are brookies only and would otherwise be expected to be browns. Maybe the 2 exceptions that have barriers, that's the reason, and the ones that don't have barriers stay brookies for some other reason, I dunno. But a straight up statistical crunch of the numbers does not show it.

I still don't see the logic.

Gradient and stream size does not uniformly predict species presence in this study because there are streams of the same size and gradient that are sympatric brook/brown and ones that just have brook trout so there must be another factor driving species composition in those streams besides gradient and stream size. Thats obviously is not proof of causation since neither group had a barrier but again it highlights something besides size and gradient are at work there. I would really like to hear the authors take on this.

 

[FarmerDave]

I disagree, and I think I already addressed that and so did pcray, and did a better job if it.

28 streams of similar size.

• 15 of them are exclusively brook trout.
• 13 had some browns, but we don't know how many.
• 0 were brown trout exclusive.

That couple with this...

• Larger streams with lower gradient were mostly brown trout exclusive.
• Smaller streams with higher gradient were brook trout exclusive. These also happen to have barriers.

My conclusions can logically only be as follow.

1. Smaller streams with high gradient favors brook trout. But I admit I already knew that. I grew up in NWPA.
2. Slightly larger streams (but still small) with somewhat less gradient still favor brook trout, but does not exclude browns.
3. Largest streams with low gradient favor browns.

As much as one may want to say this study supports the idea that barriers made a difference in any of the sample streams in this study, they did not, or at least the study did not prove this. This is not to say that they don't work outside of the sample streams in this study.