In this paper, we consider contention-based ad-hoc networks, where links use mini-slots to contend for the channel (i.e., channel probing), and followed by data transmission over a longer duration, as in CSMA. We investigate channel-aware distributed scheduling under the multipacket reception (MPR) model, which involves a joint process of channel probing and distributed scheduling. In this study, we formulate channelaware distributed scheduling as a team game, with the objective being to maximize the overall network throughput. We use optimal stopping theory to tackle this problem, and show that the optimal policy for distributed scheduling has a threshold structure. Observing that the network throughput depends heavily on the contention probabilities of links, we then generalize the study to jointly optimize the rate thresholds and the contention probabilities, and propose a two-stage algorithm for computing the optimal threshold-probability pair by invoking fractional optimization and convex programming.