In a 2-hop IEEE 802.11-based wireless LAN, the distributed coordination function (DCF) tends to equally share the available capacity among the contending stations. Recently alternative capacity sharing strategies have been made possible through the QoS enhancement IEEE 802.11e. We propose a versatile infinite-state Markov reward model to study the bottleneck node in a 2-hop IEEE 802.11-based ad hoc network for different adaptive capacity sharing strategies. We use infinite-state stochastic Petri nets (iSPNs) to specify our model, from which the underlying QBD-type Markov-reward models are automatically derived. The impact of the different capacity sharing strategies is analyzed by CSRL model checking of the underlying infinite-state QBD. Our modeling approach helps in deciding under wich circumstances which adaptive capacity sharing strategy is most appropriate. Furthermore we analyze how the QoS parameters from IEEE 802.11e fit into our model.