Response:

Background concentrations of nutrients in streams (independent of human activities) are determined in part by parent geology, soils, physiography, natural vegetation, and climate. Natural sources of P are rocks and soil, and sedimentary rocks, especially marine calcareous formations, tend to have the highest concentrations of P, while crystalline rocks have the lowest concentrations. Therefore, one might expect that ecoregions, which take into account most of the above factors, would have better theoretical explanatory power for reference nutrient conditions than geology alone.

Land use explains much of current observed ambient nutrient concentrations, but it is problematic for defining nutrient expectations. Reference expectations are what we expect under natural conditions, with no human activities that discharge nutrients to waterbodies. Land use is comprised of many of the activities that are sources of nutrients (agriculture, urban, suburban, commercial, etc.). Reference sites should be defined to have the fewest and smallest anthropogenic sources of nutrients in a region.

There have been recent several analyses of using ecoregions as strata for nutrient criteria. Rohm et al. (2002) examined ambient nutrient concentrations throughout the lower 48 states and showed distinct differences among the 14 aggregated nutrient ecoregions. They also showed that least disturbed reference sites had substantially lower N and P concentrations in part of nutrient region XI (Central and Eastern Forested Uplands).

Smith et al. (2003) used the USGS SPARROW model to extrapolate distributions of natural background nutrient yields and concentrations (ideal reference) in all 14 nutrient ecoregions. Their results also showed sharp differences among the nutrient ecoregions, and further suggested that most of EPA’s estimates of TN quartiles are low.

Dodds and Oakes (2004) used regression analysis to project (extrapolate) nutrient concentration for reference or background in the absence of satisfactory reference sites. They used regressions of nutrient concentrations on land use, and extrapolated back to the absence of urban or agricultural land for a Kansas dataset, and also throughout the lower 48 states. They also showed differences in both baseline (reference) concentrations as well as the effect of land use among the aggregated nutrient ecoregions.

Finally, Pan et al. (2000) examined benthic diatom species composition among level 3 ecoregions within a single aggregated nutrient ecoregion (XI, Central and Eastern Forested Uplands). In reference sites, there was no detectable effect of level 3 ecoregion on the diatom species composition, however, expected periphyton metrics may vary within ecoregions if significant hydrogeomorphologic variation exists among streams. These results suggest that the aggregation of the nutrient ecoregions is adequate for the Appalachian highlands, and further stratification to level 3 is not necessary.

In summary, the aggregated Level 3 ecoregions for nutrients appear to be sufficient for defining nutrient expectations. It is quite possible that more detailed examinations may reveal the need to go to a finer stratification in some regions and for some waterbody types. For example, southeastern lakes can be stratified by water color and pH, and both reference expectation and response to enrichment differ among the 4 lake types (e.g., Shannon and Brezonik 1976).

References

Dodds, W.K., and R.M. Oakes. 2004. A technique for establishing reference nutrient concentrations across watersheds affected by humans. Limnol. Oceanogr.: Methods 2: 333-341

Pan, Y., R.J. Stevenson, B.H. Hill, and A.T. Herlihy. 2000. Ecoregions and bentic diatom assemblages in Mid-Atlantic Highlands streams, USA. J. N. Am. Benthol. Soc. 19: 518-540.

Rohm, C.M., J.M. Omernik, A.J. Woods, and J.L. Stoddard. Regional characteristics of nutrient concentrations in streams and their application to nutrient criteria development. J. Am. Water Resour. Assoc. 38: 213-239.

Shannon, E.E., and P. L. Brezonik. 1972. Limnological characteristics of north and central Florida lakes. Limnol. Oceanogr. 17: 97-112.

Smith, R.A., R.B. Alexander, and G.E. Schwarz. 2003. Natural background concentrations of nutrients in streams and rivers of the conterminous United States. Environ. Sci. Tech. 37: 2039-2047.

Review:

Stevenson et al. (2006) show that very different relationships should occur between stream algae among regions with different hydrologic regimes. However, the effect on nutrient criteria development may not be as clear. Stevenson et al. (2006) found great differences in periphyton biomass relationships with nutrients in the hydrologically unpredictable and predictable streams of the unglaciated landscape of Kentucky and glaciated landscape of Michigan, respectively. Invertebrate biomasses were an order of magnitude lower in KY than MI streams, probably due to flood and drought stress (Riseng et al. 2004). Grazers apparently regulated diatoms much more than the filamentous green algae Cladophora, because much lower diatom biomasses were observed in MI than KY, where grazer density was very low. Diatoms were affected relatively little by nutrient concentrations, compared to Cladophora. Diatom biomass was only constrained at the very lowest nutrient concentrations. Predicted Cladophora cover, however, increased with TP concentration and increased more in Michigan than Kentucky. This is likely due to the lower frequency of spates in Michigan than Kentucky that would disrupt Cladophora accrual. This may not effect nutrient criteria development based on the potential for nuisance Cladophora cover, however, because Cladophora cover can develop to extensive growths in moderate nutrients even in Kentucky, if conditions are stable for sufficiently long. So peak biomass of Cladophora is probably similar in both ecoregions at the same nutrient concentrations, except that these peaks do not occur as often in hydrologically variable streams as stable streams. Given the potential effects of nuisance algal blooms on invertebrates and fish, these differences in frequency may not be important. For example, a nuisance bloom every 3 years of similar magnitude as a consistent bloom every year may have relatively similar effects on fauna that have either generation times or population cycles close to those periods.

In some cases, aggregate Level 3 ecoregions may not be sufficient for refining nutrient reference conditions based on natural potential of streams. Stream size, hydrology related to slope, and landscape conditions may have important effects on expected natural nutrient concentrations in streams, rivers, and lakes. As nutrient ecoregions seem to explain some variation in expected nutrient concentrations in streams, recent work in Michigan done by Stevenson, Soranno, Cheruvelil and Rollins indicate that site specific attributes explain great variation in natural TP of expected in streams and lakes of Michigan. This work is currently not published and is currently being refined. Regression models were developed to explain variation in expected, near natural concentrations in a stream or lake based on hydrogeomorphology of the waterbody and landscape attributes. A regression model predicting over 50% of the variation in TP among streams was constructed that indicates stream slope, % agriculture and urban land use in the watershed, % wetlands in the watershed and riparian zone, stream width, and stream sinuosity affect TP conditions. The lake model included lake depth, dissolved organic carbon, and % agricultural and urban land use in the watershed. Thus, expected (reference) nutrient conditions in streams and lakes can be established for each stream specifically. Then, based on expected variation associated with natural conditions and effects of increasing nutrient concentrations, specific nutrient criteria can be established for each stream or lake, based on its natural potential and effects of nutrient enrichment on that system.

Please add the caveat to the paragraph about the Pan et al (2000) observations that expected periphyton metrics may vary within ecoregions if significant hydrogeomorphologic variation exists among streams. Recent work that I have seen (Hawkins and Cao or vice versa) indicates that it is necessary to account for this variation to detect changes in some regions.

Riseng, C. M., M. J. Wiley, and R. J. Stevenson. 2004. Hydrologic disturbance and nutrient effects on benthic community structure in midwestern US streams: a covariance structure analysis. Journal of the North American Benthological Society 23:309-326.

Stevenson, R. J., S. T. Rier, C. M. Riseng, R. E. Schultz, and M. J. Wiley. 2006. Comparing effects of nutrients on algal biomass in streams in 2 regions with different disturbance regimes and with applications for developing nutrient criteria. Hydrobiologia 561:140-165.

Update:

If you look at geographic scales finer than your target ecoregion, and your data are good, you will find effects of the finer-scale geographic differences on nutrient expectations. The question is, is the smaller-scale spatial variability substantial enough to make a difference in setting nutrient criteria? The answer depends on the magnitude of the natural spatial variation compared to the magnitude of nutrient problems and an agency's ability to regulate them.

With respect to the first reviewer’s findings, I think that nutrient concentrations influenced by urban and agricultural land use do not meet requirements for reference condition, because reference implies urban and ag land use as close to zero as possible. Further, we expect ecological differences in algae responses based on habitat, hydrology, light, etc.; but these are responses, not reference nutrient concentrations. I concur with the first reviewer’s clarification of the Pan et al. work.

Related to Criteria and Standards Development #1
Related to Data Gathering and Assessment #6

Response:

This is precisely the reason that the EPA has taken a nutrient ecoregion approach. While the exact boundaries of ecoregions may not work well for baseline nutrient criteria (Wickham et al 2005), a spatially explicit sampling regime should pick naturally high nutrient areas if a set of reference sites can be identified. An adequate reference site would have little or no anthropogenic activity in the watershed and no point-sources of nutrients. If such a set of reference sites cannot be found, then it is possible to sample across a range of sites, use the percentage agriculture and urban to create a statistical model that can be used to extrapolate back to the reference condition (Dodds and Oakes 2004). The more difficult question is where should the nutrient criteria be set relative to the median reference level? The national EPA guidance document recommends setting this at the upper 75% level of the reference distribution (Buck et al 2000). Obviously, it is not reasonable (or desirable if it could be achieved) to set nutrient criteria at levels below which they occur naturally in the environment.

Buck, S., G. Denton, W. Dodds, J. Fisher, D. Flemer, D. Hart, A. Parker, S. Porter, S. Rector, A. Steinman, J. Stevenson, J. Stoner, D. Tillman, S. Wang, V. Watson, and E. Welch United States Environmental Protection Agency. 2000. Nutrient Criteria Technical Guidance Manual, Rivers and Streams. EPA-822-B-00-002.

Dodds, W.K. and R. M. Oakes 2004. A technique for establishing reference nutrient concentrations across watersheds affected by humans. Limnology and Oceanography Methods 2: 333-341.

Wickham, J.D., K. H. Riitters, T. G. Wade, and K. B. Jones 2005. Evaluating the relative roles of ecological regions and land-cover composition for guiding establishment of nutrient criteria. Landscape Ecology 20: 791-798.

Review:

While a glossary of these terms is provided in Appendix A, the following information may also be useful:

• Ecoregions: Examples of U.S. ecoregions are provided in the draft as Figures 3.1 and 3.2. Table 2 (Chapter 3) also provides additional information about alternative landscape and wetland classification schemes related to ecoregions.
  
• Baseline: A set of data or studies used to determine the reference - or natural - condition.
  
• Median: The central tendency of a population in which half of the observations are larger and half are smaller, i.e., the fiftieth percentile.
  
• Spatially-Explicit Sampling Scheme: A method for collecting data that incorporates the spatial variation associated with landscape features (hydrologic, geologic, biologic, land use, etc.) at the regional and local level.
  
• Extrapolate: To predict beyond the range of experience or observations. Needed when reference conditions are not available.
  

Review:

I agree with the first reviewer. It is critical to develop and test the relevant nutrient ecoregions for your area. In the regionalization, it will be necessary to demonstrate that the elevated concentrations are truly natural, as may be the case with phosphatic soils. In some agricultural areas, decades of nitrogen fertilization have raised soil and groundwater nitrate levels, and these are clearly not natural, even though they are the current "background" concentrations.