If no custom partitioner is defined in Hadoop then how is data partitioned before it is sent to the reducer?

The default partitioner computes a hash value for the key and assigns the partition based on this result.