3. Literature Documentation for AREM

User's Manual:
Avian Richness Evaluation Method (AREM)
for Lowland Wetlands of the Colorado Plateau

3.1 Documentation of Indicator Thresholds

3.1.1 Introduction

The following narrative describes the basis for the specific thresholds used for particular indicators contained in the AREM field form (Appendix B). If desired, readers wishing to gain a broader, more fundamental understanding of the indicators that AREM uses may consult Adamus (1993a).

Whenever numeric thresholds are included in an evaluation method, there is considerable risk that some users will assume that these values are soundly supported by extensive research or are appropriate to all situations. This is unfortunately not true. In reality, few if any of the specific thresholds included in AREM or other rapid evaluation methods are conclusively supported by research findings. Where they seem to be, there still remains considerable uncertainty as to whether the supporting research can be validly extrapolated to all situations where the evaluation method might be used.

Why, then, are specific thresholds used? The reason is that thresholds, by standardizing important judgements, are essential to improve a method's replicability (the tendency of different users to answer questions about the same area in the same manner). When methods are poorly replicable, their credibility and, in some cases, their legal defensibility suffers.

Despite the frequent lack of conclusive documentation for particular thresholds, the numbers used by AREM are not merely speculative. Their general order of magnitude reflects the author's inferences from broadly-accepted ecological principles as defined by current wetlands literature and expert opinion. Users wishing to substitute other threshold values may do so, provided they (a) explicitly document where they made substitutions, and (b) cite appropriate data that supports the alternative values. In particular, use of alternative thresholds may be justified to reflect differing effects of other seasons, regions, species, and ecosystem types.

3.1.2 Area Thresholds

AREM uses thresholds for habitat area (i.e., size or acreage) in several places (#'s below refer to numbered questions on the field form, Appendix B; boldface has been added below to emphasize area thresholds that are used):

Rationale for Area Thresholds. As compiled above, the area thresholds that AREM uses are 0.1 acre, 1 acre, 20 acres, and 40 acres. AREM assumes that larger wetlands and larger patches of habitat within wetlands tend to have more species, but that the exact effects depend primarily on the local pool of species and the type of habitat. The AREM species models also recognize the role of other factors in influencing the species-area relationship, such as patch width or shape (Section 3.1.3 below) and proximity to other wetlands or agricultural lands (for further information see Section 3.1.6, and optionally Section 4.4.1 in Adamus 1993a).

The larger two thresholds (20, 40 acres) are applied only to open water habitat. Reservoirs larger than 26 acres are used by the largest numbers of migrant and wintering ducks in arid western Oklahoma (Copelin 1953), and those larger than 40 acres are preferred by ducks in north Texas (Hobaugh and Teer 1981) and Quebec (DesGranges and Houde 1989). Geese most often use wetlands larger than 20 acres (Guthery and Stormer 1984). In the prairie pothole region a cumulative wetland area of 200-900 acres might be required to support the full complement of local species (Brown and Dinsmore 1986), and 10 of 25 species did not use wetlands smaller than about 2 acres. However, data from north-central Minnesota (Williams 1985) and Maine (Gibbs 1991) demonstrate the elusiveness of simplified relationships between avian richness and wetland area.

The two smaller thresholds (0.1 and 1 acre) are applied to both open water and vegetated wetland habitats. The 0.1 acre threshold approximates the theoretically smallest home range of about 67% of the 90 species that regularly breed in the Colorado Plateau region¹¹, whereas the 1 acre threshold approximates the theoretically smallest home range size of about 31% of the breeding species. Studies cited in Adamus (1993a) and Leibowitz et al. (1992) document regular bird use of wetlands as small as 1 acre, and a simulation study by Gibbs (1993) highlights the adverse potential consequences of loss of such wetlands. However, few species use isolated vegetation patches of less than about 1 acre, because such patches often have maximum dimensions no larger than 100 ft, which severely exposes their inhabitants to the elements and predators. In some arid regions of North America, few seasonal wetlands smaller than 0.1 acre and few semipermanent wetlands smaller than 2 acres persist over time (Stewart and Kantrud 1971), unless subsidized by irrigation water. Although evidence suggests that one-acre patches of forest in the eastern U.S. are too small for many breeding birds, many of the species that use Colorado Plateau lowland wetlands are "edge" species presumably accustomed to using small patches. In eastern Colorado, one riparian species (tree swallow) nested only in riparian fragments larger than 37 acres, another (black-capped chickadee) nested in fragments larger than 17 acres, several nested in fragments larger than about 3 acres (American kestrel, northern flicker, hairy woodpecker, downy woodpecker), and one (house wren) nested in fragments as small as 0.3 acre, which was the smallest measured (Gutzwiler and Anderson 1987).

In a sample of 187 wetlands located within 1000 ft of canals and ditches in the Lower Gunnison-Uncompahgre subregion, 158 (84%) of the wetlands were smaller than 1 acre. In the same general area, a survey of 800 wetlands by Rector et al. (1979) also included natural bottomland wetlands, and reported the mean wetland size as 27 acres. However, both of these studies measured the area of an entire wetland, whereas AREM requires estimation of the area of the individual habitat types within a wetland.

Width Thresholds

AREM uses thresholds for width in the following places:

Rationale for Width Thresholds. As compiled above, the width thresholds that AREM uses are 3 ft, 25 ft, 100 ft, and 500 ft. AREM assumes that wider wetlands and wider patches of habitat within wetlands usually support more species, but that exact effects depend primarily on the species and the type of habitat. The AREM species models also recognize the role of other factors that influence bird use of narrow or wide patches, such as patch area (Section 3.1.2 above), and proximity to other wetlands or agricultural lands (Section 3.1.4 below, and Section 4.4.1 in Adamus 1993a).

The AREM specifications that width of open water patches be greater than 500 ft and certainly not less than 3 ft (question #3) are based on the assumption that narrower patches would restrict the ability of some larger waterbirds to take flight. Data from other regions (Ambrose et al. 1983) suggests that, for taking flight, a few waterbird species require open water or otherwise flat areas wider than about 500 ft. Shorebirds along inland migratory routes appear to favor mudflats wider than about 100 ft, and especially wider than 1500 ft, even though large numbers can occur on lakeshore flats as narrow as 3-6 ft (Taylor and Trost 1992).

Birds nesting in vegetation patches narrower than 100 ft, and certainly those in areas narrower than 25 feet, are likely to be frequently disturbed by predators (which follow the edges of vegetation patches) and human visitors (because such narrow patches offer little visual buffer). Because specific supporting data were lacking from the Colorado Plateau, these numerical values were based mostly on the author's judgement. Studies in other regions report that zones of emergent vegetation are most likely to develop along ditches wider than about 30 ft (Linde 1969). Along areas of open water, nesting waterfowl prefer strips of emergent vegetation that are wider than 25 ft; strips wider than 70 ft are considerably better (Atlantic Waterfowl Council 1972).

3.1.4 Distance Thresholds

AREM uses thresholds for distance in the following places:

Rationale for Distance Thresholds. As compiled above, the distance thresholds that AREM uses are 300 ft, 600 ft, 1000 ft, 0.5 mi, and 3 mi. AREM assumes that wetlands with beneficial habitat features located nearby (especially within these distances) will usually support more species, and that wetlands with detrimental features located nearby will support fewer species, other factors being equal.

The 300-ft threshold is intended to approximate a maximum dimension of the theoretical home range sizes of 16 (19%) of the 86 species that breed in Colorado Plateau wetlands, calculated as described in footnote #10 and assuming circular shapes for home ranges. In other words, vegetation or other habitat outside the wetland but within 300 ft is likely to be used by several species that mainly use the wetland, and several species outside the wetland but within 300 ft will regularly use the wetland. The 300-ft threshold is further documented by the following: (a) from some limited field data, Croonquist and Brooks (1993) considered widths of 300 ft to be adequate for maintaining avian richness in wooded stream corridors in Pennsylvania, (b) in Maryland, a study of riparian birds indicated that neotropical migrant species breed mostly in corridors wider than about 330 ft (Keller et al. 1993), and (c) in Colorado, the density and type of vegetation is often altered within 295 ft (90 m) downslope from lined canals, and is sometimes altered up to 590 ft (180 m) downslope.

The 0.5- and 3-mi thresholds relate to three factors: (a) migratory corridors for shorebirds and songbirds, (b) foraging distances of larger waterbirds, and (c) dispersal distances of wetland-dependent songbirds (i.e., the distance that young birds move between wetlands where they were hatched and, the following year, wetlands where they breed).

Many migratory birds (which comprise about 81% of the region's wetland avifauna) appear to concentrate in wetlands in the vicinity of major rivers and lakes, and use these areas as corridors or "stepping stones" as they migrate long distances. Accordingly, AREM assumes that wetlands located within 0.5 mi of large water bodies would serve this purpose, but this value and the values that defines "large" were based exclusively on the author's judgement because no supporting regional data were available. In the central parts of the United States, small wetlands seem to support the most nesting waterfowl when located within about 3 mi of a large water body or a permanently or semipermanently flooded wetland to which birds can freely travel when foraging (Low 1945, Warren and Bandel 1968, Williams 1985, Dobie 1986, Baldassare and Bolen 1987). Duck broods that are forced to move long distances due to a low density of wetlands on the landscape have relatively lower survival rates (Rotella and Ratti 1992). Average dispersal distances of song sparrows (a wetland species) are 0.6 to 0.9 mi (Nice 1937, Johnston 1956). In arid irrigated areas of Texas, pheasants use areas within about 0.7 mi of wetlands (Guthery and Stormer 1984), and pheasants in Michigan, Wisconsin, and North Dakota have an average travel radius of about 0.25 mi (Sather-Blair and Linder 1980).

The 600-ft threshold for visual seclusion also is based on the author's judgement, but reflects generally the literature on sensitive species in other regions. Wintering bald eagles in some areas take flight when approached from as far away as 800-1600 ft (Stalmaster and Newman 1978, Knight and Knight 1984, McGarigal et al. 1991, Buehler et al. 1991). Many waterbirds take flight when humans on foot are seen approaching within 75 to 175 ft.

3.1.5 Other Thresholds

AREM uses several other numeric thresholds, as follows:

Rationales for the Other Thresholds. The first threshold listed above (#3, water >2 inches deep) describes water that is deep enough for most of the region's waterfowl species to use, based on the anatomy and behavior of these species. It is recognized that most waterfowl prefer greater depths (to at least 12-18 inches depth); AREM accounts for that fact indirectly by including water body area as an indicator in the waterfowl species models.

The 1-ft/s threshold used in question #4(a) describes a velocity beyond which most ducks and other birds that use water as a substrate are much less likely to occur. The threshold is based on the author's judgement and reflects an assumption that at greater velocities, waterbirds would spend so much energy maintaining their position that their feeding success would be lowered.

The 10-inch threshold used in question #4(b) describes a degree of water clarity assumed to be sufficient to (a) allow at least minimal growth of completely submersed aquatic plants and associated aquatic insects that are important to many waterfowl species, and (b) permit waterbirds that feed on subsurface foods to see these foods. This threshold is based on literature from other regions (as compiled in Kantrud 1990) which suggests that one of the submersed plants that is most-favored by Colorado Plateau waterfowl -- sago pondweed, Potamogeton pectinatus -- grows poorly when water column visibility is less than about 10 inches.

The threshold used in question #4(c) -- dry-out occurring at least one year in five -- describes a frequency of water level drawdown that is assumed sufficient to increase the productivity of open-water wetlands, partly by oxidizing organic matter and associated nutrients located in wetland sediments, and facilitating seed germination. Abundant evidence from other regions (Knighton 1985, Fredrickson and Taylor 1982) supports the assumption that wetlands which otherwise are permanently flooded support greater secondary production if they periodically dry out or exchange waters with major rivers or lakes. The specific threshold (one year in five) reflects literature on marsh management that recommends drawdown frequencies of once every 3-5 years (range, 1-10 years)(Harris and Marshall 1963, Linde 1969, Payne 1992).

The threshold of "three" trees used in question #7 describes a condition assumed to be minimally acceptable to most riparian species. Birds (particularly perching raptors) occasionally visit wetlands with only one or two trees, but a threshold of three was set to allow for minimal resource turnover (i.e., long-term replacement of trees lost from blowdowns). The threshold is based entirely on the author's judgement.

The threshold of 50% canopy closure to differentiate dense from open stands of woody vegetation (#8, 12) also is based entirely on the author's judgement, because regional data relating avian richness to canopy closure in lowland riparian areas were not available.

AREM uses a threshold of >12 inches diameter to define large trees (#9) and a threshold of >5 inches diameter (#10) to define dead trees and limbs useful to cavity-nesting wildlife, which contribute importantly to avian richness. In northeastern Colorado, cottonwoods having cavities larger than 1 inch are used (or excavated) by downy woodpecker, black-capped chickadee, and house wren; those larger than 3-4 inches are used by wood duck, American kestrel, western screech owl, and northern flicker. Most cavities are found in limbs of greater than 6-12 inches diameter (Sedgwick and Knopf 1986).

The threshold of 1 ft for vegetation height, used to differentiate short from tall stands of herbaceous vegetation (#14), is based on the assumption that many birds which dwell mostly at ground level (e.g., shorebirds, waterfowl) cannot routinely see above much taller vegetation. These species often prefer wetlands with vegetation <1 ft so that approaching predators can be detected and avoided in time. Some other species (e.g., bitterns) seldom use stands of vegetation shorter than about 20 inches, and especially not shorter than 1 ft. Studies in Arizona (Ohmart et al. 1985) demonstrated the importance of herbaceous vegetation height for predicting the birds that use irrigated wetlands.

A threshold of 60% for cover types (#15, 16) within an area of specified size is used to differentiate landscapes that are functionally (from a bird perspective) dominated by the specified types of land cover from ones that are not. Recognizing that different species respond to land cover at different scales, AREM requires that users consider whether this proportion is attained at distances of 0.5 mi and/or 3 mi from the evaluated wetland. The proportion (60%) was derived from "percolation theory" by O'Neill et al. (1988) and reflects a theoretical probability that an unspecified organism will be able to move freely across a landscape among patches of its favored habitat. The applicability of this assumption to birds in the Colorado Plateau region is untested.

The thresholds for visual seclusion (#17) and predation potential (#18) are based entirely on the author's judgement, because regional data relating avian richness to these types of disturbances were not available. They are perhaps the most subjective of the thresholds included in AREM.

The 5-ft threshold for height of banks suitable for bank-nesting birds is based primarily on the author's observations and assumes that birds nesting in banks much lower than this would be more severely harassed by mammalian predators.

3.2 Documentation of Weighting Factors

AREM currently provides users with the option of weighting species according to any of six characteristics. These are described below. Weights are currently assigned on a 1-10 ordinal scale, but there are no empirical data to support particular numerical values, so users can choose other values and scales.

3.2.1 Species Dependency on Wetland/Riparian Habitat

The WEIGHTS database of AREM defines the following categories:

"Highly Dependent" means that surface water is the species' primary substrate: e.g., common goldeneye. "Intermediate" means the species occurs only where water/wetland is present: e.g., spotted sandpiper. "Dependent" means the species also uses uplands, but uses wetlands frequently: e.g., warbling vireo. AREM users can see which category each species has been assigned to, and change the designation if they prefer, by loading the EDITAREM program (see p. 42), selecting the WEIGHTS database, and editing values in the WETDEPEND column. Current assignments of species to particular categories were based entirely on the author's interpretation of literature and experience with the species.

3.2.2 Relative Abundance

The WEIGHTS database of AREM defines the following categories:

"Uncommon" species are ones that can be found in suitable habitat at the appropriate season at a rate of about 1-10 per day, whereas "fairly common" species are encountered at a rate of 10-25 per day, "common" species at a rate of 25-100 per day, and abundant species at a rate of >100 per day. These definitions are from Dexter and Lavad (1992), and separate abundance codes have often been assigned for different seasonal periods. Species that are not likely to be encountered daily, even at an appropriate season and in suitable habitat, are not included in AREM. However, if there are known instances of a regularly-present species having such a low detection rate solely because of its characteristic secretiveness or obscurity, then it should nonetheless be factored into calculations of avian richness. AREM users can see which category each species has been assigned to, and change the designation if justified, by loading the EDITAREM program, selecting the WEIGHTS database, and editing values in the column corresponding to the desired subregion and season. Current assignments of species to particular categories were based mainly on literature (cited on p. 70 of Adamus 1993a), but abundance categories of species breeding in the Grand Valley subregion were modified after data from the author's field studies had been analyzed, to reflect insights gained from the data. There is considerable potential for improving this information, and thus the assignment of appropriate weights, through collection and analysis of additional field data (see p. 42 for discussion of this). Information on relative abundance in the WEIGHTS database is identical to information in the REGIONS database.

3.2.3 Taxonomic Uniqueness

AREM assigns more weight to wetlands having species that are, relatively speaking, taxonomically unrelated. For example, a wetland whose (say) five species are each from a different taxonomic order will be calculated to have greater taxonomic diversity than a wetland whose five species are from taxonomically different families within a single order. That is because within the hierarchy of taxonomic classification, orders are considered to be generally more distinct than families. This is important because animal communities that are taxonomically diverse are often genetically and functionally diverse, i.e., gene pools contain considerable variability, species sometimes fill a wide variety of trophic roles, and food webs can be complex, presumably leading to greater adaptability and sustainability of populations (Vane-Wright et al. 1991).

AREM calculates the taxonomic uniqueness of a wetland in the following manner. After predicting which species should occur in a wetland, AREM calls up the TAXINF database and sorts all species present in the evaluated wetland according to the values in the TAXINF database, proceeding from left (beginning with the "Order" column) to right (ending with the "Species" column). This sequence reflects decreasing taxonomic uniqueness. Then, the program assigns the maximum value to each species as follows:

After assigning weights to all species, AREM sums them to give the final "taxonomic uniqueness" score for the "weighted habitat." Information in the TAXINF database was derived from DeSante and Pyle (1986) and Ehrlich et al. (1988).

3.2.4 Neotropical Migrant Status

The WEIGHTS database of AREM defines the following categories:

"Type A" species breed only in the United States and/or Canada and migrate to the Neotropics. "Type B" species also breed in the Neotropics. Other regional species generally do not migrate to Neotropics. AREM users can see which category each species has been assigned to, and change the designation if they prefer, by loading the EDITAREM program, selecting the WEIGHTS database, and editing values in the NEOTROPIC column. Current assignments of species to particular categories were based entirely on information compiled by Carter and Barker (1993).

3.2.5 Official Conservation Designations

The WEIGHTS database of AREM assigns a weight of "10" to any species that has any of the following official conservation designations:

Endangered (federally listed)
Threatened (federally listed)
Candidate for Threatened or Endangered list
G3 ("rare/uncommon globally but not imperiled")
G4 ("not rare; apparently secure but cause for longterm concern")
S1 (sensitive due to possibly declining populations and/or rarity statewide)
S2 (sensitive due to possibly declining populations and limited in-state distribution)
S3 (sensitive but fairly common statewide)
"Watch List" (apparently sensitive but information lacking)

Species not officially designated as belonging to any of the above categories are assigned a weight of "1." AREM users can see which category each species has been assigned to, and change the designation if warranted, by loading the EDITAREM program, selecting the WEIGHTS database, and editing values in the corresponding columns (ENDANGERED, THREATENED, CANDIDATE, G3, G4, S1, S2, S3, WATCHLIST). Species whose official designations apply only to breeding populations, and which do not breed in the study region, were not included unless their nonbreeding populations also had been officially assigned special status. Information on official designations of species came from the state natural heritage programs in Colorado, Utah, and Wyoming.

3.2.6 Hunted Status

The WEIGHTS database of AREM defines just two categories:

The "Hunted" species category includes all wetland species, whether resident or migratory, that can be harvested legally in the study region. AREM users can see which category each species has been assigned to, and change the designation if warranted, by loading the EDITAREM program, selecting the WEIGHTS database, and editing values in the corresponding column (HUNTED).

¹¹ The theoretical home range size estimate was derived using the allometric equation of Schoener (1968) as demonstrated by Van Horne and Wiens (1991). The equation is: A = 98.6M^1.15

where A is the predicted home range area (in ha) and M is the body mass (in g). From Dunning (1984, 1993), I obtained average body mass data for 86 of the 102 species that breed regularly in lowland wetland/riparian areas of the region. The 26 breeding species categorized as "highly dependent" or "intermediate" with regard to use of wetlands were projected to have a mean home range of 5.53 acres (range 0.04-42 acres), whereas the mean for the 60 species categorized simply as "dependent" was projected as 1.03 acres (range 0.01-14 acres). The accuracy of the allometric formula is difficult to gauge, but an attempt was made to determine its comparability. When I compared (for 40 of the species) the predicted home range areas with areas inferred from Breeding Bird Census (BBC) data (by dividing plot area by number of pairs, for a small number of plots from wetland habitats nationwide), I found all but 4 of 40 the allometric-based ranges to be smaller than BBC-based ranges. Areas predicted by the allometric equation were an order of magnitude smaller than BBC-based areas for 17 (43%) of the 40 species. This could mean that a threshold somewhat larger than one acre might be appropriate to represent home range sizes of Colorado Plateau breeding bird species.

Back Table of Contents Forward