For　data　collection　in　large-scale　wireless　sensor　networks　(WSNs),　dynamic　clustering　provides　a　scalable　and　energy-efficient　solution,　which　uses　cluster　head　(CH)　rotation　and　cluster　range　assignment　algorithms　to　balance　the　energy　consumption.　Nevertheless,　most　existing　works　consider　the　clustering　and　routing　as　two　isolated　issues,　which　is　harmful　to　the　connectivity　and　energy　efficiency　of　the　network.　In　this　paper,　we　provide　a　detailed　analysis　on　the　relations　between　clustering　and　routing,　and　then　propose　a　joint　clustering　and　routing　(JCR)　protocol　for　reliable　and　efficient　data　collection　in　large-scale　WSN.　JCR　adopts　the　backoff　timer　and　gradient　routing　to　generate　connected　and　efficient　intercluster　topology　with　the　constraint　of　maximum　transmission　range.　The　relations　between　clustering　and　routing　in　JCR　are　further　exploited　by　theoretical　and　numerical　analyses.　The　results　show　that　the　multihop　routing　in　JCR　may　lead　to　the　unbalanced　CH　selection.　Then,　the　solution　is　provided　to　optimize　the　network　lifetime　by　considering　the　gradient　of　one-hop　neighbor　nodes　in　the　setting　of　backoff　timer.　Theoretical　analysis　and　simulation　results　prove　the　connectivity　and　efficiency　of　the　network　topology　generated　by　JCR.