Oceans and Deserts
According to the World Atlas of Biodiversity by Groombridge et al (excerpts available on Google Books) the active biomass on land is 550 billion tons of carbon, and 99.9% of that is plant material. The ocean is only 10 billion tons, mostly as animals. The reasons for this surprising result is trees, which incorporate a vast amount of biomass as woody structure, mostly inaccessable to predators, whereas the ocean's plants are mostly plankton, which is quickly eaten by the animals.
Krill and fish are obvious predators of plankton. But viruses are far more deadly. An average cubic centimeter of sea water contains 1 million single celled organisms (including plankton), and 10 million viruses (according to a TED talk by Craig Venter). Meanwhile, near infrared (half of sunlight) penetrates the ocean by only a few centimeters. Only half of blue light penetrates more than a meter in nutrient-rich waters. So the ocean is a surprisingly hostile environment for dense life, in contrast to the land.
The total surface area of the earth is 5.1E14 m2, of which 1.5E14 m2 is land and 3.6E14 m2 is ocean. This results in an average biomass density of 3900 g/m2 for land, and 28 g/m2 for the oceans. Both the land and the ocean are diverse, with rain-forest jungle contrasting with deserts on land, and upwellings and coastal shelves contrasting with unmixed water over most of the ocean. But even alkaline deserts may have a biomass density of 2000 g/m2 according to Here Comes the Sun: Solar Thermal in the Mojave Desert—Carbon. Reduction or Loss of Sequestration? by Campbell et. al, with uptake of 100gC/m2, comparable to grassland and temperate forest.
Solar photovoltaic farms in the desert will interrupt this sequestration, by blocking the sunlight driving the plants. If they are "power tower" heat engines, they will also use scarce water. The oceans are much more "deserted", and a much better place for solar energy collection, ecologically speaking. The ocean is an especially appropriate place for rectennas for space solar power, especially if the occasional sea-bird can be steered away from the beam, perhaps using mechanical "predators" to scare them away.
Biomass and Carbonate sequestration
Land plants tend to sequester carbon for decades, reversably. Their roots ("the rhizosphere") create acids and a zone of high CO2 near the rock at the bottom of the soil, which break down the rock, releasing phosphates and other nutrients, while converting rock to carbonate minerals. Carbonation of rocks is the main long term CO2 removal process, while sequestration in wood merely stores it for later release.
Single cell ocean plants do not directly sequester much carbon at all, because their life cycle is less than a week. Many of these plant cells gradually sink, some are eaten by viruses, and some are eaten by krill, coral, and other primary animal feeders. Some of the higher level feeders make calcium carbonate, which sequesters carbon and sends it to the sea floor, a primary form of carbon sequestration in the ocean.
Rocks are the primary reservoir of carbon on the planet, and the deep ocean is a reservoir for CO2. Long term management of CO2 involves creating carbonate rock, primarily by biological processes. CO2 in the biologically active ocean above the thermocline acidifies it, making the formation of carbonate shells difficult.
The principal limit on ocean productivity is nutrients. The ocean is divided by a horizontal region called the thermocline; below the thermocline, it is salty and dense and nutrient rich and dark. Above the thermocline, less salty and dense, nutrient starved and light. It may be possible to increase primary productivity and the production of shells through judicious artificially created upwelling in some patches of the ocean. This must be done for the long term, providing stable conditions for higher level communities time to establish and regulate themselves.
Injudicious dumping of nutrients into oceans, such as humans create with phosphate fertilizer runoff from rivers, does not establish these stable communities. With pulse nutrients, algae grow explosively, doubling many times per week, while the higher level feeders take weeks or months to establish themselves. Because all plants both produce oxygen and consume it, the consumption of phosphates is quickly followed by the consumption of oxygen from the water, making it lethal for fish.
Perhaps with enough energy, we can remove the phosphates from river water, recycling them for agricultural use, or releasing them slowly enough into the ocean communities to give them time to adapt. With better modelling and management of fertilizer use, perhaps by using crops engineered to thrive on slow release of fertilizer and automated mechanisms that release it gradually throughout the growing season, rather than in pulse application, the waste runoff can be greatly reduced, and the remaining waste flow can be designed to feed stable and established communities.
If artificial species can be developed that are resistant to viruses, mechanically stronger, and better at capturing sunlight and CO2 and nitrogen, provided with phosphates and steered into "farms", they could greatly multiply the biological activity of the oceans. That is a very big if. We don't know nearly enough biology to do this, much less to do it safely, but it does point towards future opportunities to sequester CO2 and produce more food and carbon fuel. Perhaps, 100 years from now, the "CO2" crisis will be a lack of it, rather than an overabundance. While we should work towards expeditious restoration of a healthy 300ppm CO2 , we must not panic and unleash mechanisms that we cannot control.
Why does this matter to Server Sky? Server Sky will put infrastructure in orbit, not on the ground. That is a better place to put solar cells. Someday, "Power Sky" will beam power as microwaves to earth. While it may be possible to construct receiving antennas with crops underneath, that will probably add more expense then the agricultural revenue justifies. Weeds and scrub may grow under the rectennas, but may interfere with operations, and be eradicated with weed killer by some operators. So, it is likely the rectennas will result in even less land area for nature. If it is possible, the rectennas should be built floating on the water, at a safe distance from land.
Server sky can also provide the computational infrastructure to support nutrient cycle modelling, and support robotic fertilization and plant management. Rather than an expensive human making a single pass application of fertilizer once a year, slow robots can roam the fields continuously, applying fertilizer or mechanical pest management as needed, perhaps as often as once a week. A solar powered robot running 80 hours a week, covering a swath a meter wide and moving at half a meter per second, can tend to 14 hectares a week. A human on a large tractor can tend to more in less time, but the salary and fuel and cost of capital make single-pass application and annual tending of crops necessary.