The role of pacemaker properties in vertebrate respiratory rhythm generation is not well understood. To address this question from a comparative perspective, brain stems from adult turtles were isolated in vitro, and respiratory motor bursts were recorded on hypoglossal (XII) nerve rootlets. The goal was to test whether burst frequency could be altered by conditions known to alter respiratory pacemaker neuron activity in mammals (e.g., increased bath KCl or blockade of specific inward currents). While bathed in artificial cerebrospinal fluid (aCSF), respiratory burst frequency was not correlated with changes in bath KCl (0.5-10.0 mM). Riluzole (50 microM; persistent Na(+) channel blocker) increased burst frequency by 31 +/- 5% (P < 0.05) and decreased burst amplitude by 42 +/- 4% (P < 0.05). In contrast, flufenamic acid (FFA, 20-500 microM; Ca(2+)-activated cation channel blocker) reduced and abolished burst frequency in a dose- and time-dependent manner (P < 0.05). During synaptic inhibition blockade with bicuculline (50 microM; GABA(A) channel blocker) and strychnine (50 muM; glycine receptor blocker), rhythmic motor activity persisted, and burst frequency was directly correlated with extracellular KCl (0.5-10.0 mM; P = 0.005). During synaptic inhibition blockade, riluzole (50 microM) did not alter burst frequency, whereas FFA (100 microM) abolished burst frequency (P < 0.05). These data are most consistent with the hypothesis that turtle respiratory rhythm generation requires Ca(2+)-activated cation channels but not pacemaker neurons, which thereby favors the group-pacemaker model. During synaptic inhibition blockade, however, the rhythm generator appears to be transformed into a pacemaker-driven network that requires Ca(2+)-activated cation channels.