26 DWYER, NIXON, OVIATT, PEREZ, AND SMAYDA 



Defining boundaries, inputs, and outputs is obviously easier for 

 some ecosystems. For example, the Hubbard Brook watersheds 

 (Waide etal., 1974) have easily measurable inputs, and all material 

 output flows through a weir where it can be monitored. Defining 

 system boundaries in Narragansett Bay is much more difficult, and 

 defining fluxes across an operationally established estuarine— oceanic 

 boundary is a major research task in itself (Kremer and Nixon, 

 1977). 



Our spectral analysis of Narragansett Bay is constrained by the 

 availability of suitable data. The analysis is a detailed examination 

 of the in-system and closed feedback-loop dynamics (Child and 

 Shugart, 1972; Shinners, 1972) of the primary producer compart- 

 ment of the Narragansett Bay ecosystem. We make no claims that the 

 frequency responses we calculated represent those of the whole 

 system. If Patten's (1975) linearization proposal is correct, however, 

 the many nonlinear submechanisms governing this compartment 

 (e.g., pronounced high-frequency phytoplankton species succession 

 and saturation kinetics for light and nutrients) will be minimized in a 

 tendency toward linearization of the frequency response of primary 

 production. 



One input to primary production, surface ammonia, is an 

 integrating compartment in the ecosystem. It is taken up preferen- 

 tially by phytoplankton in Narragansett Bay and is limiting most of 

 the year. Through feedback mechanisms, ammonia concentration is 

 affected by primary producers. Ammonia concentration is also 

 modified by an exogenous input (i.e., river runoff), which carries 

 ammonia from numerous sewage treatment plants, and by wind- 

 driven water turbulence, which modifies nutrient regeneration by 

 benthic sediments. Exact analysis of this problem requires a detailed 

 signal -theory approach, which was not attempted for this analysis. 

 Thus we must make the simplifying assumption that the effect on 

 ammonia concentration of primary production fed back through the 

 food web is small compared v^th the direct effect of ammonia on 

 primary producers. This assumption is supported by measurements 

 of zooplankton ammonia excretion, which show that throughout 

 most of the year only a small portion of the primary production in 

 the bay is supported by nitrogen that has been rapidly regenerated 

 through the herbivores (Vargo, 1976). The assumption is even more 

 nearly correct for the microcosm experiments, in which high levels of 

 ammonia were added artificially. Ammonia levels in control micro- 

 cosms (modified only by internal feedback mechanisms) remained 

 much lower than those in microcosms receiving an artificial input 

 (Fig. 2). 



