In large ad hoc networks with nodes equipped with multiple antennas, the
tradeoff between multiplexing and diversity gains in each point-to-point
link impacts the overall network capacity, albeit in a different way than
the widely used Zheng-Tse tradeoff. A maximum network capacity balances
between diversity and multiplexing, the former allowing increased node
density and the latter allowing increased point-to-point throughput. Adding
spatial modes necessarily reduces the optimal contention density as a cost
for increasing the link throughput. However, this tradeoff is profitable in
the large antenna and large path loss regimes.  As specific examples, we
consider the transmission capacity with N antennas performing spatial
multiplexing with equal power allocation, and then the case where the
stream corresponding to the worst eigenvalue is eliminated for the N X N
and 2 X N cases. We also develop an optimal power allocation for the
spatial modes which greatly improves the performance relative to equal
power allocation.