1: \begin{abstract}

2: Representations of branching Markov processes and their

3: measure-valued limits in terms of

4: countable systems of particles are constructed for

5: models with spatially varying birth and death rates.

6: Each particle has a location and a ``level,'' but unlike

7: earlier constructions, the levels change with time. In

8: fact, death of a particle occurs only when the level of

9: the particle crosses a specified level $r$, or for the limiting

10: models, hits infinity. For branching Markov processes,

11: at each time $t$, conditioned on the state of the process,

12: the levels are independent and uniformly distributed on

13: $[0,r]$.

14: For the limiting measure-valued process, at each time $t$, the joint

15: distribution of locations and levels is conditionally

16: Poisson distributed with mean measure $K(t)\times\varLambda$,

17: where $\varLambda$ denotes Lebesgue measure, and $K$ is the

18: desired measure-valued process.

19: 

20: The representation simplifies or gives alternative proofs

21: for a variety of calculations and results

22: including conditioning on extinction or

23: nonextinction, Harris's convergence theorem for

24: supercritical branching processes, and diffusion

25: approximations for processes in random environments.

26: \end{abstract}