Supplementary Material s16

Supplementary Material N. Duarte, L. H. Pardo, M. J. Robin-Abbott. Susceptibility of Forests in the Northeastern U.S. to Nitrogen and Sulfur Deposition: critical load exceedance and forest health

Methods Site Description We included 4057 plots from national and regional forest health surveys in this analysis (Figure 1, Table 1). These included the plots from the national USDA Forest Service Forest Inventory and Analysis (FIA) program, including Forest Health Monitoring (FHM) program. The FIA program’s Forest Monitoring Survey consists of a three phase sampling approach used to track status and trends in forest extent, cover, growth, mortality, removals, and overall health (http://www.fia.fs.fed.us). Phase 1 consists of remote sensing to identify where the forested land is (not used in this study). Phase 2 (P2) consists of collecting basic forest structure data for one sample site per 2428 hectares of forest. Phase 3 (P3) (originally referred to as Forest Health Monitoring) consists of a subset of P2 plots which are measured for a broader suite of parameters (Coulston et al. 2005; USDA-FS 2006). The FIA program’s Forest Monitoring Survey is described in detail at: http://www.fia.fs.fed.us; further details on sites can be found in Duarte et al. (2011a).Due to the grid layout of the FIA plots, not all of the plots are forested; only plots with some forested area were included in this analysis.

Other forest health surveys in New England states, including the North American Maple Program (NAMP), the Vermont Hardwood Health Survey (HHS), and the Vermont Monitoring Cooperative Forest Health Plots (VMC-FH) were included in this assessment. Data from the National Resource Inventory (NRI) soil pits and from county soil surveys (Natural Resource Conservation Service (NRCS) Soil Survey Staff 2003) and additional research sites in Vermont were also included (Table 2).

The VMC-FH plots, located on Mount Mansfield and at Lye Brook Wilderness Area in Vermont (http://sal.snr.uvm.edu/vmc/), are laid out and are sampled according to FIA P3 plot protocol (see http://www.fia.fs.fed.us.). The NAMP plots (0.01 ha) are located either in managed sugarbushes or in sugar maple-dominated northern hardwood stands. The Vermont HHS sites (1 ha) were sampled as a ground survey carried out as a compliment to aerial infrared photography

1 conducted by the Vermont Department of Forestry, Parks, and Recreation (http://www.vtfpr.org/protection/for_protect.cfm). Access to geographic coordinates for FIA and FHM plots is strictly limited by law in order to protect the privacy of private landowners who permit sampling on their land (http://www.fia.fs.fed.us/library/papers-presentations/). Therefore, we worked closely with a FIA GIS Specialist to overlay the geographic coordinates of the individual plots with digitized county soil survey maps (Soil Survey Geographic (SSURGO) Database; http://soils.usda.gov/survey/geography/ssurgo/) in order to estimate the missing soil input data. Similarly, we worked with the FIA GIS Specialist to model the missing climate data. The methods employed were developed during the Pilot Phase of the NEG/ECP Forest Mapping project (Duarte et al. 2011b). In order to display the results spatially for FIA plots while protecting landowner privacy, we used publically available “fuzzed and swapped” co-ordinates that were switched with those for a similar plot within the county (swapped) and altered by approximately 0.5 miles (fuzzed). These changes should not alter general patterns at the regional scale. According to FIA protocol, the publicly available coordinates for all New York FIA plots are the geographic center of the county within which the plot is located (county centroid). Therefore it was not possible to show spatial patterns within the county. Instead, we plotted the mean of the county for the entire county, in order to give a general idea of the larger-scale, regional spatial patterns. Because there are no publicly available co-ordinates for these FHM plots, they are not included in the maps, but are included in the figures.

Substrate Type/Clay Content Method We applied the substrate type/clay content method described by Sverdrup et al. (1990) to estimate soil mineral weathering rates. The substrate type/clay content method uses three categories of soil substrate: acidic, intermediate, and basic. Acidic soil substrates include granites, gneiss, sandstones, and felsic rocks; intermediate soil substrates in clude diorite, granodiorite, conglomerates and most sedimentary rocks other than sand stone; basic soil substrates in clude mafic rocks, sedimentary rocks with low carbonate content, and carbonate rocks. Information about the substr ate from which the soil was formed, available from soil series descriptions, was required to appropriately categorize the soil series.

The following equations are used in the substrate type/clay content method: 2 We = 56.7 * Clay – 0.32 * Clay for acidic substrates (1) 2 We = 500 + 53.6 * Clay – 0.18 * Clay for intermediate substrates (2)

2 We = 500 + 59.2 * Clay for basic substrates (3)

-1 -1 Where: We = empirical mineral weathering rate for 1 m soil (eq ha y ) Clay = average clay percent in the mineral soil (%)

This empirical mineral weathering rate is then corrected for the air temperature: ((A/(2.6+273))-(A/(273+Tm))) Wc = We * e (4)

-1 -1 -1 Where: Wc = weathering rate corrected for air temp. (eq ha m y ) A = Arthenius constant (3600º K) Tm = Mean annual air temperature (ºC)

Lastly, the weathering rate is corrected for the actual depth of the mineral soil, through the B horizon:

W = Wc * depth * .01 (5)

Where W = the estimated mineral weathering rate (eq ha-1 y-1) Depth = the depth of the mineral soil (cm)

More details on nutrient removal calculations The Tree Chemistry Database, used to identify nutrient content for nutrient removal calculations can be accessed via http://www.treesearch.fs.fed.us/pubs/9464. In order to determine the nutrient removal via harvesting, we assumed saw timber harvest. If, in fact, a whole tree harvest were used, significantly more biomass would be removed, lowering the

CL Smax and increasing CL nutN. By definition, in using any mean or median scenario, the most sensitive sites are overlooked. Thus, this type of analysis is best used to regions where more sensitive sites are likely to lie so that further more intensive studies can be conducted in those areas. It is also useful for identifying which sites are very unlikely, either by virtue of soil type or deposition input, to be susceptible to atmospheric deposition. Finally, certain sites may deviate from expectations. For example, a high elevation site on Mt Mansfield, Vermont had deeper soils than would be expected, resulting in a higher critical load and lower exceedance. In contrast, poor soils at low elevation would lead to a lower critical load than expected.

Figure S.1. NSSC mean scenario for sites in New England and New York a. NSSC mean mineral weathering rates for sites in New England and New York b. Mean critical loads for acidity (S+N) for sites in New England and New York FIA. c. Mean exceedance for acidity (S+N) for sites in New England and New York FIA.

4 Figure S.2. Growth versus exceedance by species for FIA (P2 plots) Only species with significant correlations using Spearman’s rank correlation analysis (α =0.05) are shown. See Table 4 for statistical results.