WestslopeCutthroat Trout   

    The environmental variables used in this analysis were chosen based on previous studies of hybridization in trout (1, 2, 3). I aslo included variables that I thought might be good indicators of HI in westslope cutthroat trout (i.e. temperature). The adjusted R^2 values for both models were very similar and could explain 63-64% of the observed hybridization index. Based on these analyses, it  appears that the all variables model can explain just as much of the HI pattern as the best fit model. The main issue that differentiates these two models is statistical significance and ultimately the reliability of the proposed model. The six factors of the best fit model showed the greatest statistical reliability and together, produced a higher R^2 value than the all variables model. When analyzed individually, there were no variables that could produce a R^2 value greater than 64.5%.

    Some of the variables that I expected to greatly influence hybridization levels such as rainbow trout stocking, were excluded from the best fit model. Possible reasons for why the coefficients were not statistically significant and the possible impact of each variable of on HI are discussed in further detail below.

Samples Size

    Probably the biggest issue in my analysis is the number of sample sites and high variability in the environmental factors tested . The sites used for this project are a subset of a larger data set of 131 sample sites. Only 49 had complete data for all the variables tested. This resulted in roughly 8-9 sites per hybridization level,  which might be too low of a sample set given the strong variablility of the factor data. 

   Stocking of rainbow trout into the study system is the reason why we see hybridization in westslope cutthroat trout populations. The rainbow trout trout have to get there in order to hybridize. It was, however, excluded from the best fit model and showed a relatively week correlation coefficient.  There are a number of reasons why this might be.

    The first problem may be a result of the time frame of stocking events in which I based my analyses. I chose to use stocking data over two, ten-year intervals, (from 1950-1960 and 1990-2000). I chose to include data over a ten year interval because stocking locations would vary from year to year and data from a single year may not adequately represent the overall stocking activity.  

    I chose to run the analyses over two different time intervals because I thought it would be interesting to see which had a greater impact on the hybridization levels we see today, stocking that happened about 50 years ago or more recently 20 years ago. The first stocking records began in the 1950's, this is why I used it as a measure for earliest stocking events. This does not mean that the rivers were not being stocked prior to 1950, just that no one was recording it. As a more recent indicator of stocking activity impact on HI, I chose to use data from the 1990's because I wanted to ensure enough time had passed between the stocking event and when HI measures were taken. This was to ensure that the fish could experience several reproductive cycles, the average age to maturity is about 3 years in trout. It may be that the intervals I chose to test are still not telling the whole story of how rainbow trout stocking influences HI.

    As mentioned above, there was a lot of interannual variability in stocking sites which may explain the weak relationship from my analysis. Other things to consider are the number of fish stocked per location, how often it was stocked per year and the overall fish density in the system. The more fish there are, the higher the likelihood of hybridization. Unfortunately, my analyses did not test for these variables. 

    Another factor to consider is the proximity of stocking sites to my sample sites, this is overall what I was testing for.  Both maps that I created actually show that the rainbow trout stocking sites are quite  far from the majority of my sample sites.  A previous study of rainbow trout movement in streams estimated that they stay within a 10 km radius from the stocking location. If this is true, then the rainbow trout stocking data is much too far from my sample sites to have any significant effect on hybridization level. In the real world though, straying events are quite common. This is where individuals in a habitat migrate to other areas to live and reproduce. All it really takes is a handful of rainbow trout to genetically distrupt a population of westslope cutthroat trout. 

    I also alluded to the fact that my data is limited to recorded introductions of rainbow trout. It is possible that rainbow trout  were being  introduced into the system before records were kept. It is also possible that fisherman may have stocked the lakes themselves to enhance the fishing in the stream system. Only recently are we becoming aware of the impacts of introducing foreign or invasive species into an ecosystem. As a result, laws prohibiting the introduction of non-native species are relatively recent. It is also difficult to monitor and enforce these laws. One thing is certain, rainbow trout are getting there somehow and hybridizing.

Distance to nearest sample site with pure rainbow trout

    As fish are prone to stray, it makes good sense that proximity to hybridized sites would  impact  hybridization levels in adjacent streams.This factor was found to be a good predictor as a part of the best fit model. Much like stocking rainbow trout in to a system, the presence of rainbow trout nearby would increase the likelihood of hybridization events. 

    Some possible problems with this theory is that my analysis measures a euclidean distance, and not stream distance. Thus, my measure does not adequately depict straying  unless the streams were perfectly linear and intersect.  With my measures, it could happen that two locations with the same HI could be close to one another, but never intersect. At this basic level, proximity to hybrized sites would not be a causal relationship.  Trout do not walk across land, but there are some fish that do!

    To help plead my case, though, human activity can help move fish across a terrestrial landscape. Fishermen can release fish caught in one fishing location to a nearby stream that and hybridization could subsequently ensue. This might sound a bit far fetched, but it may happen more than you would think. The government imposes limits on the number of fish caught per season. If a fisherman catches a fish in one location, keeps it alive in his bucket and moves to a nearby location to continue fishing, he could reel in more fish than allowed by law, so he has to throw the fish back. What if the fish he catches in the new location is better than the one caught at the previous site? Well he would likely throw back the fish he caught at the previous site and thus be introducing a possibly hybridized fish into the system.

Powerlines, railroads, pipelines and access roads

    Previous studies on hybridization in trout have shown that human activity can promote inbreeding (1, 2). My analyses placed three of these variables in the best fit model. Proximity to pipelines did not reliably explain the pattern of hybrization. This may be because they are relatively benign. Once they have been laid down they can sit there for decades before maintenence is necessary. Powerlines would likely need to have regularly scheduled maintenance, railroads are used frequently and so are access roads. This results in continual disturbance of the system.

    Human activity, causes habitat degradation. Compared to domesticated rainbow trout, westslope cutthroat trout have a very narrow set of variables that they can tolerate in their habitat. This is particularly true for spawning habitat. If the necessary habitat has been destroyed, westslope cutthroat trout are more likely to spawn in the same places ast rainbow trout and this would increase the likelihood of interbreeding. 

    Human activity can also increase the mortality of nearby fish. Only the resilient survive. Because westslope cutthroat trout are so sensitive, they are likely more suceptible than rainbow trout to these impacts. This can create a system where there are a lot of rainbow trout and whatever westslope cutthroat trout do survive, end up mating with rainbow trout simply because westslope cutthroat are too few in number. 

Water Temperature

    This variable was chosen for my analysis as recent work (3) suggests that temperature may regulate hybridization levels. My analysis could not use this variable as a predictor of hybridization. This is likely due to the sampling issues mentioned above and strong variation in measured values. This variation could be due to the fact that measurements were taken at various times during the day (water temperature fluctuates daily), in different months (sampling took place between July and August) and that measurements were taken in different years (temperature varies from year to year).

    Westslope cutthroat trout are very sensitive to changes in temperature and do best in very cold water. Rainbow trout can tolerate a wider range of temperatures and have been selectively bred to grow well in warm temperatures. Thus, we would expect higher levels of hybrization in warmer temperatures as pure westslope cutthroat trout may die due to competition or thermal stress. This positive relationship is reflected by the coefficient sign in the all variables model. 

Mean and Maximum Depth

    Both of these variables show strong positive relationships with hybridization level in the all variables model. The deeper the water, the stronger the hybridization. Only maximum depth showed a statistically significant coefficient, and ended up in the 'best fit' model. It is unclear why a deeper stream would show stronger hybridization, but westslope cutthroat trout tend to spawn in shallower waters. In the absence of shallow water, they may spawn in the same places as rainbow trout and increase the likelihood of interbreeding.

Elevation

    Elevation was one of the two stream environment variables in the best fit model. This variable has also been found in previous studies to be a good predictor of hybridization levels (1, 2, 3) . As elevation increases, hybridization levels decrease.

    Amongst scientist, studying hybridization in trout, it is widely believed that this relationship this the result of rainbow trout stocking. Overall, rainbow trout tend to be stocked at lower elevations and over time they might eventually spread to higher 'ground'. It was hypothesized that it was only a matter of time before we would see hybridization at higher elevations.

Physical Barriers

    Physical barriers prevent hybrid individuals from entering and leaving a stream. This restricts the spread of hybridization (2, 3).  The presence of barriers have long been implicated as the only factor preventing rainbow trout from getting into a local system and hybridizing. The all variables OLS did not demostrate a strong relationship and the coefficient was not significant. I think the problem here is that my sample sites were too far from one another to really tell the whole story.  To get a real sense of the impact of a barrier, an analysis would have to be done at a finer scale, above and below the barrier. Unfortunately the sample sites used in this project were too far apart to really tease out this thiis relationship.

Stream Order

    Strahler stream order showed a relatively strong positive relationship with hybridization level, and previous studies have also found this measure to be a good predictor (1, 2).  In my study, the coefficient probability of stream order was not significant and it was kept out of the best fit model. A larger number of sample sites would have probably teased out this relationship and produced a statistically significant coefficient probability.

    The reason why the relationship is positive is that a high stream order means that the sample site has a complex branching pattern and that lots of smaller tributaries feed into the stream. Hybridization levels could be high just due simply to the fact that there are likely to be more hybrids if multiple sources are converging at a site, versus a sample site that is 1st order (has no source stream sources).

    Possible issues with this analysis can also be owed to the fact that stream order, a measure of stream branching, was determined by DEM rasters. Elevation data may not adequately depict a stream network. For example, based simply on elevation, a dry valley can be classified as a stream and may ultimately intersect a stream. This would create error or uncertainty in the determination of the stream order.

(Click on images to see full maps)

Maps of Human Impact Variables

Maps of Stream Environment Variables