I was thinking about this again this morning and I had an alternate idea.

To think of the problem another way, I want to be able to link an animal to an infusate (which defines a set of tracers without concentrations). That is one link. When I have an animal, I want to essentially add an annotation to the InfusateTracer table (but I cannot put it directly in that table). It is as if I want to add a concentration field to the InfusateTracer table, so to do it without changing that table, I just have to link to it... So that's what I should do instead of independently linking the animal table to the tracer table to define the concentrations, I should link the animal table to the InfusateTracer table using a linking table (because it is a many to many relationship). In other words, I am annotating the links between infusate and tracer. The relationships would look like this:

class Animal(Model):
    name = CharField()
    infusion_rate = FloatField()
    infusate = ForeignKey(Infusate)

class Infusate(Model):
    name = CharField()
    tracers = ManyToManyForeignKey(Tracer)

class Tracer(Model):
    name = CharField()

class InfusateTracer(Model):
    infusate = ForeignKey(Infusate)
    tracer = ForeignKey(Tracer)

class AnimalInfusateTracer(Model):
    animal = ForeignKey(Animal)
    infusate_tracer = ForeignKey(InfusateTracer)
    concentration = FloatField()

I think this is the answer because I don't have to worry about synchronizing two sets of multiple independent links from infusate to tracer and from animal to tracer (so that they each link to the same set).

When I am loading data, all I have to worry about is looking for infusatetracer records That have the correct infusate and correct tracer!