I counted sockeye salmon at a certain site using two different methods at coming up with the number of sockeye that passed by the site. For each day (41 days in total) I counted sockeye using two methods. One was counting the total number of sockeye over the hour and the other was only counting a sub-sampled portion of the hour then expanding for the remaining hour (i.e. 10 fish in 20 minutes so 30 fish in one hour). Some numbers are negative numbers (meaning that more fish went down than up). What statistical test can use to see if there is a significant different between the two methods and how can I perform it? Should I assume the counts are normally distributed or should I assume the opposite? Parametric or non-parametric?

Thanks in advance