Patch dynamics in a landscape modified by ecosystem engineers

draining of a pond and exposure of accumulatedsediments following abandonment of a site and subsequentdam failure.In the model, beaver ponds with a resident colony areconsidered as active patches (A). Typically, only theoldest pair of beaver in a colony will reproduce,producing on average 3 or 4 young annually (Jenkinsand Busher 1979). Colonies typically consist of an adultpair, yearlings, and kits, with an average size of 5.859/0.61 (SE) individuals (Svendsen 1980). Young beaverdisperse at about 2 or 3 years of age (Jenkins and Busher1979, Svendsen 1980), although if food supplies areplentiful or local population densities are high, as manyas 50% of two-year old beaver can remain at their natalcolony (van Deelen and Pletscher 1996). Dispersaldistances tend to be less than 16 km (Beer 1955, Leege1968), but distances of up to 110 km have been reported(Hibbard 1958). Mortality during dispersal tends to behigh relative to non-dispersing beaver, with mortalityrates of 40% being reported for the 1.5/2.5 year class ofbeaver in Newfoundland (Payne 1984). Birth rates percolony, site fidelity and mortality during dispersal areimportant factors in determining the model parameter n,or the per patch production rate of successful colonists.Although these factors are likely to vary across the rangeof beaver, if one assumes the figures reported above arestandard for beaver across their range, one can estimatea value for n of 0.7 successful immigrants per year.It is also possible that the creation of active patchesmay be affected by the presence of colonies of beaverthat occur in sites that do not require habitat modification.Beaver are known to build lodges on naturallyoccurring lakes and ponds as well as in the banks oflarger rivers without performing any significant engineering.In landscapes where such patches are present,the number of successful colonists that are produced bythese non-engineered sites (i) will influence the rate atwhich new active patches are formed. Although it ispossible that the number of colonists coming from nonengineeredpatches is not constant and might depend onthe number of engineered patches in a landscape, tosimplify analysis of the model, we have assumed aconstant rate of immigration.Over time, active colonies deplete the food resourcesadjacent to the pond, ponds fill with sediment, andresident beaver die eventually leading to abandonment.The rate at which this occurs (d) will be a function ofcolony size, beaver activity, the composition and abundanceof riparian zone trees or other food resources, andsediment loads of dammed streams. Knudson (1962)reported that beaver ponds may remain active for as longas 8/10 years, while in Algonquin Provincial Park,Canada, sites were occupied for an average of 5.89/0.46 (SE) years over a ten year period (Fryxell 2001).Once ponds are abandoned, they develop into wetlandswith variable hydrologic regimes and vegetation composition(Remillard et al. 1987, Johnston and Naiman1990c, McMaster and McMaster 2000, Wright et al.2002). The sites remain in this state (equivalent todegraded, D, patches) until the vegetation in the areaadjacent to the former pond site has recovered to adegree sufficient to support a new beaver colony. Therate at which this transformation from D patches to Ppatches occurs (r) depends on the successional dynamicsof the forests surrounding pond sites.In most areas, once beaver have colonized a site, thesite enters into a pattern of cyclic abandonment andrecolonization. It is relatively rare for a site that has beencolonized by beaver to revert back to a forested riparianzone (equivalent to the F state in the more complexpatch dynamic model, Ives 1942, Remillard et al. 1987,Johnston and Naiman 1990a, Pastor et al. 1993,Terwilliger and Pastor 1999), thus the parameter r islikely to be extremely low in most ecosystems affected bybeaver. The degree to which beaver prefer or avoid sitesthat have been previously colonized relative to forestedriparian zone (z) will again depend on the successionaldynamics of the riparian zone vegetation. In many areas,recently abandoned sites are dominated by species ofSalix, Populus and Alnus, preferred food species ofbeaver (Jenkins and Busher 1979). The situation whereherbivores, such as beaver, create environments favorablefor the growth of early-successional species, which areoften preferred by beaver, has been termed the retardedsuccession hypothesis (Pastor and Naiman 1992). However,it is also possible for beaver foraging to facilitate thedominance of conifers and other late-successional speciesthat are typically avoided as food sources, (theaccelerated succession hypothesis, Fryxell 2001).Furthermore, it has been shown that browsing by beavercan increase rates of phenolic glycoside production inPopulus fremontii (Martinsen et al. 1998). If suchchemical defenses against mammalian herbivory persistor browsing by beaver leads to dominance of nonpreferredspecies, sites that have been previously occupiedby beaver might be avoided.Beaver activity on the Huntington Wildlife Forest(HWF) has been surveyed annually since 1979. TheHWF is a 6000-hectare preserve located in the centralAdirondack Mountains, NY (latitude 44800?N, longitude74813?W). The topography is mountainous withelevations ranging from 457 m to 823 m. Vegetationconsists of mixed northern hardwood and coniferousforest. As part of the Adirondack Long Term MonitoringProject (ALTEMP), all active beaver sites on HWFhave been identified and mapped every fall since 1979.Although the number of colonies has fluctuated overtime (Fig. 2), the number of beaver colonies hasremained relatively constant, particularly since 1990.Although there is some variability in the numbers ofindividuals per colony, colony counts can provide auseful estimate of population sizes for beaver (BergerudOIKOS 105:2 (2004) 339