Apologies for not getting more information out sooner but I had some technical difficulties posting updates last week (internet rather website in origin). The short statement posted last week was a test that seems to have worked...
These last few weeks I've been continuing to work on the analysis of the screw trap data and have started to look at the electrofishing data. In particular, I've been looking running statistical analyses of the numbers of fish recorded as having damage and the samples taken from the river in the April 2014 electrofishing survey.
For a larger image of this graph click here
This graph appears to show that the mean number of damaged fish in each year is different. In fact, if you look at the error bars on the graph you can see that they all overlap except those on the 2012 bar. So the only clear relationship we can take from here is that the samples collected in 2012 show a larger number of damaged fish than any other year.
If you examine deeper; a t-test for different means shows that all years are different except 2010/ 11 and 2009/ 13. Again the graph would seem to bear this out as the actual bars for 2010/ 11 and 2009/ 13 certainly group close together when compared to the other graphs. However, assuming that the damage data is not normally distributed, a chi2
analysis of the data suggests that none of the years are significantly different to each other. However, chi2
becomes unreliable when there are large numbers of zeroes in the dataset (as there are here) so these results must be viewed with caution. Indeed the next step for the analysis will be to test for normalcy in the data and re-test with a different non-parametric test such as the Mann-Whitney U-test.
Unfortunately, while we can to a degree quantify these relationships, we cannot explain them. Because of limitations in the data collected we cannot say why 2012 showed so much more damage than any other year. Lacking data on predator numbers or movements we cannot say if there were more adult or juvenile predators that year (or both), nor can we cannot link the damage to a particular species. Flow data does not correlate with damage. In fact, for the six years of the study four show no correlation between flow and the percentage of damaged fish, one shows a weak negative correlation and one a strong negative correlation suggesting that the fish are not being damaged by the riverine environment (discussed previously on the 21.08.2014).
While we may never be able to explain the increase in 2012, future work using juveniles density surveys, predator surveys, flow and substrate studies, may allow us to identify an explanation for the observed relationship. With this data in hand it may be possible to compare the identified contributing factors the historical data to explain patterns seen in the the earlier study years.
I've also been looking at the Electrofishing data. Starting with the 2014 electrofishing data, I've been able to identify one site where the ratio of the juveniles length to weight is significantly different to the other sites. The lowest site of the survey, where the River Taodail merges with the Carron, gave rise to juveniles that were skinnier than the other fish in the river (confirmed with a Mann-Whitney U-test). Is this due to the low position in the river? The influence of the Taodail? Does the ratio held true year on year? Is the ratio associated with a genetic difference in the fish? Or was it a one-off due to something specific to this year? Future electrofishing surveys should continue to takes weights of juveniles captured and include this site again, perhaps even traveling up the Taodail to see if the fish in the Taodail are morphologically/ genetically different from those in the Carron. Perhaps comparative habitat surveys may help shed some light on these differences.
Finally, initial observations of this same data seemed to show that the fish being caught in the Coulags River were, on average, both bigger and heavier than those elsewhere in the river. This needs a bit more analysis as while the Coulags has some of the longest and heaviest fish in the survey sample, the ratio between length and weight remained similar to the other sites (except the Taodail site obviously). This might suggest that the weight of the fish is increasing in line with the length of the fish (more research needs to be done to see if there is an accepted model relating weight gain to length), therefore the ratio will remain the same regardless of the actual size/ weight of the fish. Or it might indicate that the Coulags produces bigger fish. Point to not here is that the survey was performed in April and there may be something about the Coulags that allows it to produce larger fish at that point of the year.
Future work for me will be to look at the length and weight data to see if there is any significant differences there (rather than in the ratios). future work for the project may be to look at the Coulags to try and determine why it produces bigger fish, or whether the growth rate is different here and the fish all end up the same size eventually.