To explain aspects of the quantum-to-classical transition, quantum Darwinism explores the fact that, due to interactions between a quantum open system and its surrounding environment, information about the system can be spread redundantly to the environment. Here we recall that there are in the literature two distinct and non-equivalent ways to make this statement precise and quantitative. We first point out the difference with some simple but illustrative examples. We then consider a model where Darwinism can be seen from both perspectives. Moreover, the non-Markovianity of our model can be varied with a parameter. In a recent work [F. Galve et al., Sci. Reps. 6, 19607 (2016)], the authors concluded that quantum Darwinism can be hindered by non-Markovianity. We depart from their analysis and argue that, from both perspectives to quantum Darwinism, there is no clear relationship between non-Markovianity and quantum Darwinism in our model. FIG. 1: Three qualitatively distinct possibilities for a partial information plot (PIP). The solid line (a) corresponds to a case where a small fraction of the environment already has average mutual information close to H(S), being the signature of quantum Darwinism. The linear profile (b) can be seen (approximately) in our model for small interaction times (see Sec. V). In profile (c), f δ is close to 0.5, so the redundancy is very small. This kind of profile can be obtained from random pure states drawn according to the Haar measure [13].