It has been traditionally hypothesized that the heavy quark (charm c and bottom b) fragmentation is universal across different collision systems, based on the notion that hadronization as a soft process should occur at the characteristic nonperturbative quantum chromodynamics (QCD) scale, ΛQCD. However, this universality hypothesis has recently been challenged by the observation that the c- and b-baryon production relative to their meson counterparts in minimum bias proton-proton (pp) collisions at the CERN Large Hadron Collider (LHC) energies is significantly enhanced as compared to the electron-positron (e+e−) collisions. The conception of nonuniversality is unambiguously reinforced by the latest measurement of the charged-particle multiplicity dependence of the b-baryon–to–meson yield ratio, Λb/B, by the LHCb experiment in s=13TeVpp collisions at the LHC, evolving continuously from the saturation value in minimum bias pp collisions toward the small value in e+e− collisions as the system size gradually reduces. We address the multiplicity dependence of b-baryon production in the canonical statistical hadronization model with input b-hadron spectrum augmented with many hitherto unobserved states from quark model predictions. We demonstrate that the decreasing trend of the Λb/B toward low multiplicities can be quantitatively understood from the canonical suppression on the yield of Λb, as caused by the requirement of strict conservation of baryon number in sufficiently small systems. We have therefore proposed a plausible scenario for understanding the origin of the nonuniversality of heavy quark fragmentation in elementary collisions. Published by the American Physical Society 2024