I am planning a functional response experiment with an invasive fish species and a native comparator. I am using 2 prey types, at 6 densities, resulting in 24 individual treatments (2 fish species*2 prey species*6 densities). All treatments are being replicated 3 times. We only have 18 fish of each species, and we want to make a comparisons in consumption among the fish species and the prey species. To avoid time confounds, we have randomised all treatments for a single replicate, and are planning to repeat this list 3 times for the 3 replicates. In between each replicate we allow the fish 3 days of recovery to minimise the effect of a learned response towards certain prey types. But because we use the same numbers of fish of each species, every individual fish has an equal chance of being selected, which should "evens out" the effect of re-using fish.
My question boils down to this: Is it better to fully randomise each replicate such that fish will inevitably be re-used between replicates but in a random way, or to run all replicates of a single fish species/prey type combination at once and run species/prey types sequentially, which introduces time as a confound and reduces our ability to make comparisons between the consumption of the fish towards the 2 prey species?
Our problem with ensuring individual fish are not re-used is that we cannot identify individual fish, nor do we have the facilities to separate them by e.g. which prey type/densities they have already been exposed to, so it becomes difficult to tell which fish have previously been used in which treatment.
You are correct that an individual fish experiencing the same prey type and density three times consecutively would be pseudoreplication, but the likelihood of this happening is equal among all individual fish and between each replicate fish are placed back into a holding tank and reselected at random 3 days later.
A further question would be:
As we only require 12 fish of each species for each replicate (2 prey types*6 densities), would using the same 12 fish be better than using a pool of 18? This makes certain that fish will be re-used, but also ensures that every fish receives the same amount of recovery time and is fed with standard food between replicates on the same schedule. If we use a general pool of 18 fish per species and select 12 at random each time, we reduce the likelihood that fish are being re-used but we also bring in the possibility that fish which have been used before have an advantage over those who have not by being more acclimated to the experimental conditions.
The literature is replete with examples of learning, or at least conditioning, when re-using fish. The only way you can control for - or eliminate the possibility of - learning/conditioning is to use novel fish. If you can get more fish, I recommend doing so to eliminate any possibility of learning. If that is not possible, I'm leaning toward the 12-fish approach you describe in your response to Watz and using the other 6 fish for learning/conditioning experiments. Run a few trials to determine if learning/conditioning occurs, then either quantify the effect or use whatever the results are as an interpretational aid in understanding functional response.
I have done this sort of thing when examining the effects of zebra mussels on foraging. Because of the learning issues, we did not reuse fish....we just went out to the lake and got more. But we were limited by availability when working with sturgeon.
So, I have 3 suggestions:
1. Simplify the design...perhaps by cutting back on the prey densities. This will increase your replication and therefore your power.
3. Run your analysis on fake data...you may run out of degrees of freedom depending on the stat you are using.
3. You may not need as many tanks as you think to systematically plan reuse. Set up and run a block of treatments....."used" fish go in the used tank #1. Randomize, set up, and run the next block. Fish go to used tank #2 etc....and by the time you get back to tank 1...it will be time for reuse.
And ask Jamie Dick for help. He'd be well on top of this sort of thing ;)....and say hi for me.
Thank you for your answers, in the end we were able to sample more fish (despite it being a 5 hour drive each way) such that re-use of fish became a non-issue.
- Badges
- Science topic
- Similar topics
- Biological Science
- Zoology
- Ichthyology