Can supersymmetry transformations involve a composite sector?

Question

Recently I have revisited (vixra) the idea of building the needed scalar fields of a supersymmetric theory as composites of the fermionic sector. I made no claims about the fermionic sector itself: it could be truly elementary, it could be a composite of the bosonic sector, or of itself, or of a more fundamental theory. For instance, an open superstring theory with Chen-Paton factors labeling the end of strings could be said to be a theory of supersymmety between two composite sectors.

So the question is, under what conditions can the scalar sector be composite? Can the fermionic sector still be elementary, or is it forced to become composite too?

Is it possible for one sector to be composite of the other, or must the subcomponents be essentially new objects?

Comments

Thanks to a reference of @40277, I see that in 1974 the supersymmetric string model was still called the "MFM", Meson-Fermion Model, by Sherk and Schwarz, in http://ccdb5fs.kek.jp/cgi-bin/img/allpdf?196800234  Also it was referred as "Fermion Meson dual model" by Corrigan and Olive in 1972, or Brink and Fairlie in 1974, 

---

Discussion regarding the relevance of the name "Meson Fermion model" here:
http://physicsoverflow.org/23109/discussion-regarding-meson-fermion-model-question-composite . The question is whether Schwarz meant "Meson quark model", and implied a supersymmetry of this sort, between constituents and bound states.

---

I wonder if this paper could be related? Induced N = 2 composite supersymmetry in (2+1) dimensions. From the abstract:

Starting from N = 1 scalar supermultiplets in (2+1) dimensions, we build explicitly the composite superpartners which define an N = 2 superalgebra induced by the initial N = 1 supersymmetry.