Jet opening angles of long gamma-ray bursts (lGRBs) appear to evolve in cosmic time, with lGRBs at higher redshifts being on average more narrowly beamed than those at lower redshifts. We examine the nature of this anti-correlation in the context of collimation by the progenitor stellar envelope. First, we show that the data indicate a strong correlation between gamma-ray luminosity and jet opening angle, and suggest this is a natural selection effect -only the most luminous GRBs are able to successfully launch jets with large opening angles. Then, by considering progenitor properties expected to evolve through cosmic time, we show that denser stars lead to more collimated jets, and argue that the observed anti-correlation between opening angle and redshift can be accounted for if lGRB massive star progenitors at high redshifts have higher average density compared to those at lower redshifts. This may be viable for an evolving IMF, and under the assumption that average density scales directly with mass, this relationship is consistent with the form of the IMF characteristic mass evolution suggested in the literature. The jet angle-redshift anti-correlation may also be explained if the lGRB progenitor population is dominated by massive stars at high redshift, while lower redshift lGRBs allow for a greater diversity of progenitor systems. Overall, however, we find both the jet angle-redshift anti-correlation and jet angle-luminosity correlation are consistent with the conditions of jet launch through, and collimation by, the envelope of a massive star progenitor.In this paper, we consider how different progenitor systems connect to GRB observables over cosmic time. We are motivated by the results of Lloyd-Ronning et al. (2019b) who found that certain intrinsic long gamma-ray burst (lGRB) properties appear to evolve