We reviewed a mass balance model developed in 2001 that guided establishment of the phosphorus total maximum daily load (TMDL) for Upper Klamath and Agency Lakes, Oregon. The purpose of the review was to evaluate the strengths and weaknesses of the model and to determine whether improvements could be made using information derived from studies since the model was first developed. The new data have contributed to the understanding of processes in the lakes, particularly internal loading of phosphorus from sediment, and include measurements of diffusive fluxes of phosphorus from the bottom sediments, groundwater advection, desorption from iron oxides at high pH in a laboratory setting, and estimates of fluxes of phosphorus bound to iron and aluminum oxides. None of these processes in isolation, however, is large enough to account for the episodically high values of whole-lake internal loading calculated from a mass balance, which can range from 10 to 20 milligrams per square meter per day for short periods.

The possible role of benthic invertebrates in lake sediments in the internal loading of phosphorus in the lake has become apparent since the development of the TMDL model. Benthic invertebrates can increase diffusive fluxes several-fold through bioturbation and biodiffusion, and, if the invertebrates are bottom feeders, they can recycle phosphorus to the water column through metabolic excretion. These organisms have high densities (1,822–62,178 individuals per square meter) in Upper Klamath Lake. Conversion of the mean density of tubificid worms (Oligochaeta) and chironomid midges (Diptera), two of the dominant taxa, to an areal flux rate based on laboratory measurements of metabolic excretion of two abundant species suggested that excretion by benthic invertebrates is at least as important as any of the other identified processes for internal loading to the water column.

Data from sediment cores collected around Upper Klamath Lake since the development of the TMDL model also contributed to this review. Cores were sequentially extracted to determine the distribution of phosphorus associated with several matrices in the sediment (freely exchangeable, metal-oxides, acid-soluble minerals, and residual). The concentrations of phosphorus in these fractions varied around the lake in patterns that reflect transport processes in the lake and the ultimate deposition of organic and inorganic forms of phosphorus from the water column. Both organic and inorganic phosphorus had higher concentrations in the northern part of the lake, in and just west of Goose Bay. At the time that these cores were collected, prior to restoration of the Williamson River Delta, this area was close to the shoreline of the lake and east of the Williamson River mouth. This contrasts with erosional inputs, which, in addition to being high to the east of the pre-restoration Williamson River mouth, were higher in the middle of the lake than at the northern end. Organic forms of phosphorus had particularly high concentrations in the northern bays. When these cores were used to calculate a new estimate of the whole-lake-averaged concentration of total phosphorus in the top 10 centimeters of the lake sediments, the estimate was about one-third of the best estimate available when the TMDL model was developed.