Within the field of conservation ecology, the concept of a "keystone species" is pretty central. It is based on ideas from ecology about inter-species interactions while drawing upon the visualization of a bridge's key-stone, representing the fundamentally important pinnacle stone of an arched structure. Appropriately - for the imagery - keystone species tend to be near or at the top of the "food chain"*; oftentimes being predators. The idea is that keystone species act as a limiting pressure on other species, diminishing the possibility of explosive growth (and eventual collapse) of a species in the system (as well as potential negative outcomes from such a boom and bust).
For example, wolves are considered to be a keystone species in many of the North American ecosystems, acting - alongside other large predators - as mediators in population growth among prey species, such as deer, by - to put it bluntly - killing them, which means that there are fewer fawns come springtime than if there were no wolves. This reduction from the potential deer population means that the existing deer will not likely be in such strong competition with each other to find food (and that their food sources will not be over-exploited). Conversely, if you remove wolves and other predators (and ban hunting or trapping), the deer population will boom for a few years, and then - as they outstrip their food source - will suffer massive starvation-induced die-offs. In short, it's the wolf population that keeps the ecosystem in "balance".
There is emerging a complementary concept of "cornerstone species", in which species that are low on the "food chain"* provide an essential and foundational role in the ecosystem. Although there is come question - at least in my mind - of whether this is yet another example of terrestrial systems appearing to be top-down mediated, while marine systems seem to be bottom-up mediated, I think that this concept has its own merits in understanding the functioning of ecosystems (both for pure science as well as for conservation goals).
In the linked press release, the researchers "analyzed the impact of removing seaweed and sessile animals, such as mussels and barnacles, from the rocky shores of Northeastern’s Marine Science Center in Nahant, Mass. The experiments were designed to mimic naturally occurring changes in biodiversity on rocky shores." They found that the removal of these relatively rare species caused "major declines in the abundance and diversity of animals, such as snails, crabs and other mobile animals".
It seems to me that we are finding that - not too surprisingly - that ecosystems are interactive and controlled by various links throughout, in linear and non-linear manners. If we can recognize this point about how ecosystems function, then it seems to me that our collective actions on the planet are randomly removing portions of the system without any real idea about what consequences will come about. If our ecosystems are like structures, we may well end up removing both the keystones that hold up the ceilings, but also the cornerstones that firm up the walls. This is why ecology should be studied as an applied science, to understand how our actions will have an impact upon everything else that is connected to them (and onward out from that connection, ultimately back to us).
We understand only dimly the course of impacts that removing a top predator has on an ecosystem. We understand, too, only dimly the impacts of removing a key producer has on an ecosystem. However, the total, linked effects of the removal of so many different things from so many different parts of so many different ecosystems as well as the introduction and cross-movement of so many other parts of ecosystems across into novel locations remains - as a whole - invisible. In other words, as much as we know of the direct impacts that wolves have on North American ecosystems, we remain ignorant of the indirect ecosystem impacts, and we know next to nothing about the direct and indirect impacts of species loss and species invasions across the globe.
Without keystones or cornerstones, a structure would have to be greatly simplified, making it prone to collapse at any perturbation. While I recognize that buildings are not ecosystems (nor vice versa), I think that the analogy holds; a greatly simplified ecosystem will be neither robust nor resilient, and our position within that ecosystem will become evermore imperiled. (And this will be in addition to habitat loss due to changed climate and sea level rise, with some projections of the latter as being up to 70ft.)
* "Food chain" is a term that has some scientific problems associated with it, since the recycling processes of decomposition don't fit well within the one-way concept. Interactions are better described along the lines of a "food web". However, that conceptualization also has problems associated with it.