Can two tanks be set up to run controlled experiments?

[fandaga]

Do you think it’s possible to set up two nearly identical tanks to run meaningful controlled experiments side by side, and if so, how would you go about it? For instance, one could test experiments changing just one variable over time: such as dosing schemes, different lighting parameters, pH levels, water changes, etc. BRStv did something along these lines in 2022 with many different RedSea nano tanks to look at how starting with different initial biomes impacted the ugly phase. I personally loved this video series and am still waiting on the follow up. Granted, I know there was a lot of commentary on reef2reef about this experiment, such as not enough replicates, conditions not meaningful, corporate ties potentially leading to bias, etc. But I thought it was still very thought provoking.

I’ve been thinking about the viability of starting two 16g biocubes or 15g IM tanks with the same equipment, media and then adding identical fish and corals (identical placements on frag racks with several of the same type per tank) to monitor growth and coloration over a 6-month period. Biocubes and 15g IM tanks are reasonably affordable and there’s been a lot of success running reef tanks with them. I think with premeasured food amounts and relatively low bio-load that this could be viable to control the conditions nearly identically over a long time.

 

[tbrown]

You'd want to match flow, par, and parameters as closely as possible but yes - 2 identical tanks set up at the same time using the same equipment and salt. One would be the "control" where nothing is changed, the other would be the test.

This is basically how labs run tests.

 

[fandaga]

Yeah I actually work in a chemistry lab and have a PhD in chemistry, fwiw. Still, reefing has a lot of variables to control, particularly over a long period of time like on water top off, nutrient import and export, and weird random things going wrong like brown jelly, rtn/stn, etc. On the BRStv videos they mentioned accidentally having the lights on one tank for an extended period without realizing, and they were controlling by Apex.

I wonder if anyone has tried on these forums with two tanks and could share their advice.

 

[Randy Holmes-Farley Verified Community Expert]

Yes, one can certainly do that. I've done it to test the effect of iodine on macroalgae growth.

But it is tedious and some of the most interesting questions take substantial time to get an answer. Sadly, that difficulty is why, IMO, ideas like the utility of strontium or barium or rubidium persist despite little or no direct evidence.

 

[MnFish1]

Do you think it’s possible to set up two nearly identical tanks to run meaningful controlled experiments side by side, and if so, how would you go about it? For instance, one could test experiments changing just one variable over time: such as dosing schemes, different lighting parameters, pH levels, water changes, etc. BRStv did something along these lines in 2022 with many different RedSea nano tanks to look at how starting with different initial biomes impacted the ugly phase. I personally loved this video series and am still waiting on the follow up. Granted, I know there was a lot of commentary on reef2reef about this experiment, such as not enough replicates, conditions not meaningful, corporate ties potentially leading to bias, etc. But I thought it was still very thought provoking.

I’ve been thinking about the viability of starting two 16g biocubes or 15g IM tanks with the same equipment, media and then adding identical fish and corals (identical placements on frag racks with several of the same type per tank) to monitor growth and coloration over a 6-month period. Biocubes and 15g IM tanks are reasonably affordable and there’s been a lot of success running reef tanks with them. I think with premeasured food amounts and relatively low bio-load that this could be viable to control the conditions nearly identically over a long time.

Yes - if you use new equipment it makes it easier. (and more expensive) - I did this with 2x5 gallon tanks, HOB filters with no media (only provided for flow) identical lights and heaters (bought cheaply as a grouping at Petsmart). Live rock equally measured provided filtration.

 

[Dan_P]

Do you think it’s possible to set up two nearly identical tanks to run meaningful controlled experiments side by side, and if so, how would you go about it? For instance, one could test experiments changing just one variable over time: such as dosing schemes, different lighting parameters, pH levels, water changes, etc. BRStv did something along these lines in 2022 with many different RedSea nano tanks to look at how starting with different initial biomes impacted the ugly phase. I personally loved this video series and am still waiting on the follow up. Granted, I know there was a lot of commentary on reef2reef about this experiment, such as not enough replicates, conditions not meaningful, corporate ties potentially leading to bias, etc. But I thought it was still very thought provoking.

I’ve been thinking about the viability of starting two 16g biocubes or 15g IM tanks with the same equipment, media and then adding identical fish and corals (identical placements on frag racks with several of the same type per tank) to monitor growth and coloration over a 6-month period. Biocubes and 15g IM tanks are reasonably affordable and there’s been a lot of success running reef tanks with them. I think with premeasured food amounts and relatively low bio-load that this could be viable to control the conditions nearly identically over a long time.

If you want to do serious research, avoid the aquarium as an experimental platform. If playing the lottery is fun for you, I think running experiments in an aquarium will be fun with the occassional pay out

A hobby aquarium is a difficult experimental platform to use because of the number of variables that need to be controlled. Why study something in such a messy environment? The number of replicates needed to prove an experimental response was not normal variation makes aquarium experimentation a dubious source of knowledge when replicates are not performed. Many of the BRS experiments demonstrate this.

 

[MnFish1]

If you want to do serious research, avoid the aquarium as an experimental platform. If playing the lottery is fun for you, I think running experiments in an aquarium will be fun with the occassional pay out

A hobby aquarium is a difficult experimental platform to use because of the number of variables that need to be controlled. Why study something in such a messy environment? The number of replicates needed to prove an experimental response was not normal variation makes aquarium experimentation a dubious source of knowledge when replicates are not performed. Many of the BRS experiments demonstrate this.

True - but the environments being studied are aquaria - not test-tubes (one can use the opposite argument to yours as well) - There are times to do 'bench' research and times to do 'research more directly related to the actual thing being studied', IMHO.

 

[fandaga]

If you want to do serious research, avoid the aquarium as an experimental platform. If playing the lottery is fun for you, I think running experiments in an aquarium will be fun with the occassional pay out

A hobby aquarium is a difficult experimental platform to use because of the number of variables that need to be controlled. Why study something in such a messy environment? The number of replicates needed to prove an experimental response was not normal variation makes aquarium experimentation a dubious source of knowledge when replicates are not performed. Many of the BRS experiments demonstrate this.

So I guess to provide high confidence in a result and how many replicates are needed depends on what the hypothesis is. For instance comparing 28 ppt salinity versus 35 ppt for coral growth probably would only need two tanks to demonstrate a significant difference. But if trying to demonstrate a difference between 34 and 35 ppt, a lot of replicate tanks and significant concern on testing salinity would be required.

 

[Dan_P]

So I guess to provide high confidence in a result and how many replicates are needed depends on what the hypothesis is. For instance comparing 28 ppt salinity versus 35 ppt for coral growth probably would only need two tanks to demonstrate a significant difference. But if trying to demonstrate a difference between 34 and 35 ppt, a lot of replicate tanks and significant concern on testing salinity would be required.

Depends on the size of the variation in the measurement of the treatment and response. A very large response relative to the measurement variation will not need a large number of replicates to reject the null hypothesis.

In the case of salinity, we can do a good job measuring it. Non-destructive coral growth measurement could have substantial variation but destructive measurement could possibly have a smaller variation, for example, measuring skeleton mass change. On top of the measurement variation is growth variation. Two coral are unlikely to increase in size to the same extent even under identical conditions. Three or more specimens would be needed for each treatment to confirm growth difference between treatments. If the growth rates are large and the audience for your work are in a forgiving mood, the replicate number could be smaller :-) Making a convincing argument that the two aquaria were identical during the experiment is no trivial matter given the biotic and abiotic variables involved.

I think key to an aquarium study success is achieving a large, unambiguous response.

 

[fandaga]

Depends on the size of the variation in the measurement of the treatment and response. A very large response relative to the measurement variation will not need a large number of replicates to reject the null hypothesis.

In the case of salinity, we can do a good job measuring it. Non-destructive coral growth measurement could have substantial variation but destructive measurement could possibly have a smaller variation, for example, measuring skeleton mass change. On top of the measurement variation is growth variation. Two coral are unlikely to increase in size to the same extent even under identical conditions. Three or more specimens would be needed for each treatment to confirm growth difference between treatments. If the growth rates are large and the audience for your work are in a forgiving mood, the replicate number could be smaller :) Making a convincing argument that the two aquaria were identical during the experiment is no trivial matter given the biotic and abiotic variables involved.

I think key to an aquarium study success is achieving a large, unambiguous response.

These are some great points. Ideally statistics would need to be considered - calculating p values, ANOVA, etc.

Along these lines, there was pretty heavy statistical treatment used in the following open access Marine Biology paper looking at calcification rates for 5 different species of SPS whose growth rates were monitored in 3 different locations over only 3 months: https://link.springer.com/article/10.1007/s00227-024-04511-5

But reading that article, it was clear that their conditions were much more poorly controlled than if setting up two nearly identical aquariums. For instance, they observed roughly 60% death rates in coral frags at 2 of 3 sites due to storms and hurricanes while a third site had only 15% deaths. Also, light, salinity, temperature and nutrients were also reported to be highly variable between locations.

 

[Randy Holmes-Farley Verified Community Expert]

If you want to do serious research, avoid the aquarium as an experimental platform. If playing the lottery is fun for you, I think running experiments in an aquarium will be fun with the occassional pay out

A hobby aquarium is a difficult experimental platform to use because of the number of variables that need to be controlled. Why study something in such a messy environment? The number of replicates needed to prove an experimental response was not normal variation makes aquarium experimentation a dubious source of knowledge when replicates are not performed. Many of the BRS experiments demonstrate this.

To answer many questions, I agree. In my iodine tests, each "aquarium" was actually a petri dish in a water bath to maintain the exact same temp. Even with n=7 with added iodine and n=7 without added iodine, a 41% difference in growth did not reach statistical significance due to variability, and lighting intensity differences across the plates probably impacted the results (based on growth vs position). In an aquarium with other stuff going on, the variables are likely even more complicated.

 

[Faurek]

Yes and no, depends. To try different lighting I would have the tanks connected, but for chemistry there are too many variables. Let's put a simple lab type setup, two 5g with same pumps on the same spot and same lights at the same settings and you put 4 frags of Millie's in each, you grab water and rocks from the same source or whatever, you want to experiment growth with nitrate dosing. One tank 50% WC only and one tank same 50% WC with nitrogen dosed. If you waited for a cycle they are different no matter what, if you go straight in one gets dinos and the other doesn't, how do you access accurately the growth results?
Will it be meaningful? Still yes. If you want to start researching professionally it's not good. This stuff needs patience and a lot of time. I personally believe in learning from anecdotal but some people don't.

 

[fandaga]

Yes and no, depends. To try different lighting I would have the tanks connected, but for chemistry there are too many variables. Let's put a simple lab type setup, two 5g with same pumps on the same spot and same lights at the same settings and you put 4 frags of Millie's in each, you grab water and rocks from the same source or whatever, you want to experiment growth with nitrate dosing. One tank 50% WC only and one tank same 50% WC with nitrogen dosed. If you waited for a cycle they are different no matter what, if you go straight in one gets dinos and the other doesn't, how do you access accurately the growth results?
Will it be meaningful? Still yes. If you want to start researching professionally it's not good. This stuff needs patience and a lot of time. I personally believe in learning from anecdotal but some people don't.

I like the idea about lighting experiments in connected systems.

Regarding the tank starts. One way to do it would be to just add the same amount of media, sand and rocks from an existing established system, thereby bypassing the cycle and providing both tanks with nearly identical starting conditions. Hypothetically, an established tank could be maintained to allow for the experimental tanks to be reset periodically for each new experimental condition to be tested.

 

[Dan_P]

These are some great points. Ideally statistics would need to be considered - calculating p values, ANOVA, etc.

Along these lines, there was pretty heavy statistical treatment used in the following open access Marine Biology paper looking at calcification rates for 5 different species of SPS whose growth rates were monitored in 3 different locations over only 3 months: https://link.springer.com/article/10.1007/s00227-024-04511-5

But reading that article, it was clear that their conditions were much more poorly controlled than if setting up two nearly identical aquariums. For instance, they observed roughly 60% death rates in coral frags at 2 of 3 sites due to storms and hurricanes while a third site had only 15% deaths. Also, light, salinity, temperature and nutrients were also reported to be highly variable between locations.

This is why reading the entire paper rather than just the abstract is critical…but you know this already :-)

 

[MnFish1]

I like the idea about lighting experiments in connected systems.

Regarding the tank starts. One way to do it would be to just add the same amount of media, sand and rocks from an existing established system, thereby bypassing the cycle and providing both tanks with nearly identical starting conditions. Hypothetically, an established tank could be maintained to allow for the experimental tanks to be reset periodically for each new experimental condition to be tested.

Correct - this is not a hard experiment to do. And - if you start with the same live rock, etc - there are no other variables to control. One suggestion would be (at the end of whatever experiment you do) - reverse the lighting (or whatever you're testing) - such that each 'tank' becomes its own control. Lets say you're testing lighting in Tank A with Par 500 and Tank B with PAR 200 - at the end of 6 weeks , you have much better growth in tank A than tank B - switch the PAR in Tank A to 200 and tank B to 500 - this way you can fairly easily see whether the lighting is having a consistent effect on the Corals. The difficulty as you and many have said is measuring differences in coral - and it's best to have a plan for this before you start - and keep it consistent. Here is a link to my experiment which shows how I set up 2 aquaria to do multiple experiments on ammonia, etc. https://www.reef2reef.com/threads/e...ubbing-live-rock-affect-nitrification.886715/
Obviously there are multiple ways to design experiments - this is just an example

 

[fandaga]

So if running an experiment where the measurable is growth, what corals would you choose and how many frags of each species?

Running the lighting test seems like it would be perhaps the easiest starting point to gain confidence in running such experiments.

 

[MnFish1]

So if running an experiment where the measurable is growth, what corals would you choose and how many frags of each species?

Running the lighting test seems like it would be perhaps the easiest starting point to gain confidence in running such experiments.

I would do 3 of each. Measure parameters frequently in each tank. Something like a green slimer, an easy acoropora, but anything where you can measure height easily. Soft corals may be just by appearance. etc. The key (IMHO) is making sure everything is nearly identical or as identical as possible at the start except the variable you're testing - and make sure it continues that way throughout.

 

[Reeferbadness]

I have 2 mixed reef tanks, roughly the same size (180g and 200g) and try to keep the parameters closely matched (Alk 9.0, Calc 450, Phos .08 Nitrate 5-10ppm) but they never match exactly. Lighting is the same. I often try something in 1 tank and see if it make a difference ie AB+ or reef roids etc. It's not scientifically super sound but it gives me an opportunity to see if certain things make a difference. It's also really nice to be able to swap corals from 1 tank to another if placement or flow isn't working and i don't have space in the same tank to move a coral.

 

[fandaga]

Correct - this is not a hard experiment to do. And - if you start with the same live rock, etc - there are no other variables to control. One suggestion would be (at the end of whatever experiment you do) - reverse the lighting (or whatever you're testing) - such that each 'tank' becomes its own control. Lets say you're testing lighting in Tank A with Par 500 and Tank B with PAR 200 - at the end of 6 weeks , you have much better growth in tank A than tank B - switch the PAR in Tank A to 200 and tank B to 500 - this way you can fairly easily see whether the lighting is having a consistent effect on the Corals. The difficulty as you and many have said is measuring differences in coral - and it's best to have a plan for this before you start - and keep it consistent. Here is a link to my experiment which shows how I set up 2 aquaria to do multiple experiments on ammonia, etc. https://www.reef2reef.com/threads/e...ubbing-live-rock-affect-nitrification.886715/
Obviously there are multiple ways to design experiments - this is just an example

Great experiment! Thanks for the link. I’ll read more carefully tonight.

 

[fandaga]

I would do 3 of each. Measure parameters frequently in each tank. Something like a green slimer, an easy acoropora, but anything where you can measure height easily. Soft corals may be just by appearance. etc. The key (IMHO) is making sure everything is nearly identical or as identical as possible at the start except the variable you're testing - and make sure it continues that way throughout.

Makes sense. I was thinking duncans, monti cap, and zoas. I don’t have a ton of acro experience yet, but the green slimer sounds like a good idea.

The zoas can just count the number of polyps. Calipers and weighing the frags could be done for the others.