Best of luck to you!! It's relatively easy to set up a self-sustaining tank... but extremely difficult to make one that looks like what you want. For example, you could find your nearest pond, dig up some substrate from it, buy a whole bunch of plants, pour some pond water in, add some fish, and let the ecosystem sort itself out. Or you can remove the equipment and maintenance from an established tank and watch it crash. Eventually - several months to a year down the road - the tank will come to some type of equilibrium (or more accurately, a stable
limit cycle). The tank would exhibit all stages of the life cycle, not just plant and fish growth, but also senescence and death. At any given moment there would be a few plants fading away slowly, and some more plants growing to fill in the space. Just like any ecosystem. Algae might end up being a major element in the ecosystem too, eventually reaching some stable equilibrium point. Algae might not look good, but it certainly can have an important role in an ecosystem.
Also, the aquaria we are used to seeing are very top heavy in the food chain department. In nature, the bottom of the food chain is massive compared to the top. That is, there is a whole array of life consisting of bacteria, protozoa, algae, infusoria, planaria, plants, etc. that would be needed to sustain even a single guppy. You might have to accept that it could take a 100 gallon tank to house a single small fish in a completely closed ecosystem. Especially true if that fish is carnivorous. Not quite as bad if the fish is an algae eater, herbivorous, or omnivorous.
That said, if you loosen the restriction of
completely self-sustaining and allow a person to perform maintenance or feeding, say, on a weekly or monthly basis, you can shift the balance and perhaps have higher fish load and less algae. I'm imagining a cherry shrimp and nerite snail tank that a person might only feed on the weekends, for example. Once you start to keep fish, I think it gets harder.
I definitely think your concept is achievable - as long as you let go of any preconceptions about what it should look like and just let it evolve into what it wants to be. You'd have to set it up then just watch it evolve, and not interfere. That said, I think there are several things you could do at the outset, though, that would strongly influence the final equilibrium point and make for reasonable aesthetics:
- I strongly second bpimm's comment - the El Natural method would provide a good foundation for the ecosystem. Zapin's approach overlaps a lot with this method. Basically, a soil based substrate as described in Diana Walstad's Ecology of the Planted Aquarium would give rooted plants some long term advantages and stability and shift the tank balance from [an algae tank with a few plants] to [a planted tank with some algae].
- For plants I like stepheus's suggestion - grassy foreground plants like eleocharis or e. tenellus - for several reasons. These are rooted plants, which will make good use of the substrate and fish-poo nutrients, again shifting the balance from algae to plants. Second, they would require no maintenance (unlike stem plants) and would eventually reach an equilibrium population (unlike large swords which seem to grow without bound). Finally, these plants would probably not look too bad when they die and fade away (compared to larger plants).
- I might also consider some floating plants. I'm not sure how they would shift the balance. On one hand they can make use of atmospheric CO2 and can reduce the nutrient levels in the water column - which could shift the balance away from algae. But on the other hand, they might cover the entire water surface and prevent healthy growth from the rooted plants, leading to poor substrate health. If you did use floating plants, I might recommend adding some crypts as well, to keep the substrate healthy under reduced light.
- For critters, I would add snails. Malaysian trumpet snails may be beneficial to the substrate as Zapin said. Pond or ramshorn snails would be good for debris removal. Nerites would probably be good (I've never had them) at removing some of the tough algaes like green spot from the aquarium glass, although they don't breed in freshwater so they wouldn't be self sustaining. I would not worry about the other snails being too prolific - eventually the population will stabilize.
- Some folks have kept colonies of black worms (??) alive in their substrate, and these might be a good food source for omnivorous fish. I believe Alex ("Miss Fishy" screen name) has done this before, here's one link but I think she has others describing this as well.
- I would also add cherry shrimp. They are prolific breeders and you could be assured that a stable population would exist in the tank.
- And finally fish. Not sure if too many would fit the bill. I would say stick with very small, prolific, non-carnivorous fish. Maybe endler's livebearers, guppies, or dwarf platies. Again, being prolific and small, you'd be assured that a few would survive long enough to breed, thus sustaining a population. Some balance would probably be reached. Eventually many of the fry would become food for the adults and/or the inverts in the tank. These livebearers will also eat some algae and the infusoria growing on it, and snail eggs when hungry.
- If you came back in 3 months and saw that algae was out of control, you could add some otocinclus. But that's only if the tank is an algae factory. The otos would shift the balance back toward the plants. With luck they might breed and keep the population stable. But if there is not much algae in the tank after several months, then the oto's would likely not survive long.
One last consideration is the things that can escape the tank. Namely water (evaporation) and CO2. Evaporation top-off might only be a monthly chore with a good lid, but I'm not sure about the CO2/alkalinity chemistry. I'm really not sure on this one, but it
might be helpful if you had a small amount of calcareous material in the tank. I know that many artificial lakes and ponds require periodic additions of limestone (or is it lime?). So that might be a long term factor too.
Wow I guess I got carried away.

It's just such a cool idea and I wish you all the best.

Someday when I have a basement I might try to do this as an experiment. Please let us know how it turns out!!
