Two simple algorithms for refining mammalian receptor selection in ecological risk assessments


  • Lawrence V. Tannenbaum

    Corresponding author
    1. U.S. Army Center for Health Promotion and Preventive Medicine, MCHB-TS-REH, Building 1675, Aberdeen Proving Ground, Maryland 21010–5403
    Current affiliation:
    1. The current address of J.D. Farrington is P.O. Box 117,Wilbraham, MA 01095, USA.
    • U.S. Army Center for Health Promotion and Preventive Medicine, MCHB-TS-REH, Building 1675, Aberdeen Proving Ground, Maryland 21010–5403
    Search for more papers by this author


Although guidelines exist for selecting appropriate ecological receptors for risk assessments at contaminated sites, it can be demonstrated that many of the mammals commonly evaluated are spatially irrelevant. Terrestrial risk assessments could be simplified and made more efficient, though, if mammals that are initially considered were screened for their spatial relevance. This article presents 2 simple algorithms that each demonstrate that most mammalian receptors are not spatially relevant for the overwhelming majority of hazardous waste and other contaminated sites. The algorithms use readily available and curiously overlooked spatial distribution (e.g., animal density) information and suggest that contaminated sites need to be 80 to 100 acres in size to justify the inclusion of most mammalian receptors. Given that hazardous waste sites are generally much smaller than this, many ecological risk assessments (ERA) could reasonably dispense with incorporating mammals entirely. An awareness on the part of decision makers and risk managers of the nonsuitability of many mammalian receptors evaluated in terrestrial ERAs could significantly impact the perceived need to monitor or remediate sites. This article also examines the anticipated challenges of regulators and other decision makers when entertaining the notion of a spatial relevance screen for mammals.


Several commonly acknowledged criteria for receptor selection in terrestrial ecological risk assessments (ERAs) include species having a realistic and significant potential for exposure and species likely to represent the greatest level of exposure (Weston Solutions 2004). The commonplace evaluation in ERAs of certain mammals, such as the white-tailed deer (Odocoileus virginianus) and the red fox (Vulpes vulpes), however, often constitute cases of the most minimal of exposures and, arguably, are not reasonable cases to consider. For example, where these species are considered for evaluation in assessments for sites or areas of concern that are only 2 or 3 acres in size, a cursory review of home ranges for these species reveals that only an infinitesimal portion of the animal's time would ever be spent at the site. With the likelihood of the mammal contacting the contaminated land parcel in question being so minimal, it would be appropriate at this early point of the risk assessment process to remove such species from further consideration. In stark contrast, however, such receptors are typically evaluated in a multi-tiered system that begins with the assumption that the animal spends 100% of its time at the site. Classically, such initial tier assessments indicate a significant potential for the receptor to be at risk. In the next tier assessment, however, the species' true home range (that is, area use) is 1st considered and, when factored in, it obliterates the initial, high potential-for-risk concerns.

The sequence described above prompts the question of what the ERA gained by initially identifying a seeming environmental concern based on a forced and unrealistic assumption, only to find that no such legitimate concern existed when realism was applied. Following a sequence such as the one above can negatively influence ERAs in 2 distinct ways. First, ERAs are made unnecessarily complicated because they include layers of exposure and toxicological information for a series of receptors that are ecologically irrelevant to the site either because too few animals are present at the site to constitute a concern or because animals will not spend sufficient time contacting the site to matter toxicologically. Second, a negative influence of potentially greater consequence is that routinely including fixed lists of mammalian receptors in ERAs can potentially blind risk assessors and managers from recognizing that appropriate receptor lists at managed sites probably involve fewer animals than stakeholders may realize. In some cases, mammalian assessments at sites may be unnecessary altogether.

This article offers 2 approaches for introducing improvements to mammalian receptor selection in terrestrial ERAs. One focuses on animal density, and the other considers the interplay of animal density and animal home range. Both approaches are predicated on essential questions about receptor protection that ecological risk assessors may feel uncomfortable asking, and that environmental regulators may feel uncomfortable answering.


A question that ecological risk assessors and environmental regulators would probably prefer to avoid is “How many individuals of a particular (nonprotected or nonspecial status) species constitute a concern at a contaminated site?” The question brings to the forefront a notion that, if given due consideration and not summarily dismissed, can have vast ramifications on what is perceived to be the universe of situations requiring ERAs. The notion that some minimal number of animal losses (including loss of biological function or fitness, not only animal deaths) is acceptable at a contaminated site is relevant to the ERA process because, as is clear from U.S. Environmental Protection Agency (USEPA) guidance and numerous other sources (Emlen 1989; Suter 1992; Bartell 1996; USEPA 1999), the goal of an ERA is to protect the population rather than the individual. If it were known, for example, based on our knowledge of species-specific animal distribution patterns, that a contaminated site had just 1 deer or 1 coyote present, the justification would be lacking for either conducting a risk assessment or cleaning up a site to benefit that lone species representative (assuming that it could be demonstrated that the animal was at risk).

En route to developing an algorithm for refining mammalian receptor selection, it is rather imperative that risk assessors understand how many individuals (i.e., how many more than 1) constitute a population of interest within an ERA context. Two individuals would be an absolute minimum, although that might not guarantee the presence of both a male and a female necessary to satisfy the condition that a population have the capability to breed. A mere 2 or 3 individuals might not ensure reproductive capability either because some individuals might be too old to be reproductively functional. Here, a highly conservative figure of 5 animals is suggested for the size of a population that matters in a risk assessment sense. It is important to note that it is most unlikely that a site with such a minimal population would actually follow the conventional ERA process. That process would entail holding a scoping meeting, producing a series of draft and final planning documents (to include work plans, sampling and analysis plans, and health and safety plans), securing a contract laboratory for the analytical work, mobilizing in the field for data collection, and producing a remedial investigation report that contains a risk assessment. Nevertheless, using this minimal population number, a density-based algorithm can be developed to determine the size of the area that would be expected to contain 5 animals (hereafter referred to as the 5-animal area or 5AA).

Table 1 shows the range of densities (animals per acre) reported in the literature for 11 mammals commonly considered in terrestrial ERAs (USEPA 1993; CH2M Hill 2001), excluding the small rodent/insectivore grouping (discussed later). Table 1 also provides the computed 5AAs (and for discussion purposes, the 4-animal areas, 4AA, and 10-animal areas, 10AA) based on the highest literature-recorded densities and for average reported densities where such data were available. The conservatism of the 5AAs based on the highest densities derives from the density ranges often spanning 2 or more orders of magnitude, the reported densities reflecting singular studies usually conducted in a given season, and the densities generally referring to only 1 sex of the species. Assuming that 5 animals, as the number of individuals to be concerned with, is truly conservative, a contaminated site that is any smaller than a given species' 5AA means that the site is spatially irrelevant within a risk assessment context for that species. Therefore, irrespective of the tier or phase of an ERA, such a receptor should not be evaluated.

For example, Table 1 suggests that, for 5AAs based on maximum-reported densities, there is no point in assessing the risk to mule deer (Odocoileus hemionus) at a site that is any smaller than 25 acres. This conclusion is reached without consideration given to a site's pattern of contamination but, rather, by recognizing that there simply are not enough mule deer to be concerned with at a site that is any smaller than 25 acres. Table 1 suggests that about half of the listed species where 5AAs are based on maximum literature-reported densities and all but 1 of the species where 5AAs are based on average densities will be inapplicable to ERAs at managed sites. Interestingly, many sites in which ERAs have already been conducted are smaller in size than most 5AAs and smaller than approximately half of the (maximum density-based) 10AAs shown in Table 1.

Unfortunately, there is no way to conveniently illustrate this point because program management information systems such as the USEPA's Superfund Program's Comprehensive Environmental Response, Compensation, and Liability Information System database do not track the size of land parcels that are assessed for ecological risk, which are often dramatically smaller than the overall listed acreage of the “site.” Despite this limitation, it is, nevertheless, noteworthy and instructive that 59.5% of National Priority List sites, in their totality, are smaller than 20 acres (USEPA 1989a). Further, sites are often subdivided into operable units (for example, into a groundwater unit or a soil unit for their manageability). The terrestrial portions of many operable units are notably smaller than this 20-acre figure. It is also important to note that although some sites may be enormous in size, such as Department of the Defense installations that cover thousands of acres, these are often compartmentalized so that terrestrial ERAs occur at areas that are orders of magnitude smaller than the overall site.


Another question that ecological risk assessors and environmental regulators would probably prefer to avoid is “What minimal area of contaminated soil is toxicologically inconsequential to a given (nonprotected or nonspecial status) receptor?” Although a stakeholder (e.g., a regulator), feeling uncomfortable with leaving known contamination behind (Tannenbaum 2003a), might feel compelled to remediate a contaminated parcel of any size, in actuality, a relatively large contaminated parcel could be spatially meaningless to a site's mammalian receptors, precluding the need for a risk assessment or the consideration of a remedial strategy. For example, given the immense home ranges of red fox (V. vulpes) and coyote (Canis latrans), a contaminated 1-foot-square area of soil should, in theory, not pose a concern to either species. Such a parcel would constitute an infinitesimally small portion of a receptor's home range and, due to chance alone, the area would be so infrequently contacted by the receptor that chemical exposures would be inconsequential. Admittedly, some consideration must be given to the contamination and the size of a contaminated land parcel that could be inconsequential to ecological receptors. Ecological risk assessors could use allowable contaminated area (ACA) information to identify truly site-relevant mammalian species and to recognize those instances when mammalian risk assessments are unnecessary.

One way to arrive at an approximate figure for a mammal's ACA is by adapting the established health-risk assessment guideline (USEPA 1989b) used to determine which of a site's detected chemicals should be retained within a risk assessment. Typically, a minimum detection frequency of 5% is needed to justify retaining a chemical. In theory, we should be able to impose the condition that an animal also needs to be present at a given site at least 5% of its time for it to be appropriately included in an ERA's receptors-of-concern list. Thus, if an animal's home range is sufficiently large relative to a contaminated site (as is often the case), so that the animal is away from the site 95% or more of its time, the animal, like the infrequently detected chemical, would not satisfy a threshold requirement for site presence.

Table Table 1.. Minimum site sizes of concern based on maximum and average literature-reported animal densities
  Site sizes of concern based on maximum reported densities (acres)Site sizes of concern based on average reported densities (acres)
SpeciesRange of reported densities (animals/acre)a4AA5AA10AA4AA5AA10AA
  1. a From Burt and Grossenheider 1980; Damuth 1987; Chapman and Feldhamer 1992; USEPA 1993; CH2M Hill 2001.

  2. b Data gap = average animal-density data not available.

  3. c Based on data of 9 NE U.S. states.

  4. d The average reported species density taken from the literature source that provides that statistic (Damuth 1987), is smaller than the maximum reported density from other sources.

Black-tailed jackrabbit Lepus californicus0.081–140.290.360.7280100200
Coyote Canis latrans0.0006–0.009444.4555.51,1114,0005,00010,000
Eastern cottontail rabbit Sylvilagus floridanus0.27–
Kit fox Vulpes macrotis0.0009–0.0014,0005,00010,000– data gapb
Long-tail weasel Mustela frenata0.31–2.81.431.793.58133.3166.7333.3
Mink Mustela vison0.004–0.04100125250– data gap –
Mule deer Odocoileus hemionus0.002–0.220255098.8123.5247
Muskrat Ondatra zibethicus1.05–300.130.170.34– data gap –
Raccoon Procyon lotor0.002–0.616.68.216.4308384768
Red fox Vulpes vulpes0.019–0.03133.3166.7333.36678331,666
White-tailed deer Odocoileus virginianus0.01–0.04c10012525080100d200

A potential problem with this approach is that it appears to be sanctioning the dismissal of large tracts of land in cavalier fashion. Further, the problem is exacerbated if the ACA is expressed in acres, as follows. If a “nominal” ACA (note: calculated ACAs are also described and illustrated below) of 5% of a white-tailed deer's home range is selected as a reasonable first choice (meaning that 95% of the time, the deer is not present in this space), those with a vested interest in a site's ecological management might not balk at the figure. However, 5% of a white-tailed deer's average home range (of 650 acres) equates with a 32.5-acre area, and stakeholders would be unlikely to consent to ignoring such a vast property based on a parallel to the risk assessment convention of selecting chemicals of concern with a frequency-of-detection screen. A more tenable, nominal ACA of 2% of an animal's home range (i.e., the animal is not present at the contaminated site 98% of the time) is, therefore, selected as a starting point for this algorithm. Figure 1, which illustrates what 1, 2, 3, and 4% of an animal's home range looks like pictorially, is supplied to defend the selection of 2% as a reasonable (nominal) ACA.

An approach for determining a maximal ACA that is more practical and defensible than an arbitrarily chosen small percentage of the home range, such as the nominal 2% figure discussed above, involves the integration of elements of the foregoing density-based population assessment approach and reliable home range information. In the density-based algorithm, the 5AA (i.e., the area anticipated to hold 5 representatives of a species) is that area that is of 1st ecological relevance. That area is sufficiently large to justify either an endeavor to assess risk or to implement a site cleanup to ensure species protection should unacceptable risk be demonstrated. The next-smallest density-based area of interest, the 4AA (Table 1), is de facto that area that is inconsequential in risk assessment terms because it houses too few species' representatives to merit consideration. Hence, the 4AA, as an absolute geographical area in acres, is a simplistic expression of the ACA.

A more utilitarian ACA can be articulated, however, by expressing it in the form of a ratio of the 4AA divided by the species' home range. Further, in the interest of maintaining conservatism while determining the largest possible ACA, species' minimum home ranges can be used in this ACA calculation. This choice reflects a regulator's preference for maximizing the chance that a potential exposure of concern will be recognized. For example, a regulator would much prefer to construct a receptor's area use factor (the ratio of the site size to the home range) by using the receptor's minimum home range in the ratio's denominator.

Figure Figure 1..

Graphical representation of 1, 2, 3, and 4% of a mammal's home range (inner figures, from smallest to largest) under theoretical conditions; (A) assumes the home range is symmetrical (B) assumes more realistic conditions in which the home range is irregular in shape.

For the mammals shown in Table 1, minimum home ranges were secured in 2 ways: the smallest home ranges reported in the open literature were selected and equations (for herbivores, carnivores, or omnivores) relating animal body weight to home range were employed (Harestad and Bunnell 1979). In the latter case, the lowest-reported, adult body weight of each mammal was used to generate minimum home ranges because body weight and home range are positively correlated. Table 2 presents the home ranges arrived at when using the 2 approaches. As Table 2 shows, reported adult body weights, in almost every case, vary by at least 2-fold and often higher. By electing to use the lowest-recorded, adult body weights in the body weight/home range equations, the resulting estimated, minimum home range for each animal is truly conservative. Table 2 also shows that in only 2 instances (black-tailed jackrabbit and white-tailed deer) was the calculated, minimum home range smaller than the smallest literature-reported home range.

Figure 2 presents calculated ACAs for 11 mammals by overlapping 4AAs (based on maximum, literature-reported densities and average, literature-reported densities where available information allowed) and minimum home ranges. All of the ACAs exceed the originally suggested nominal figure of 2%, with the black-tailed jackrabbit's ACA slightly above this value (3.9%) and the ACAs for 3 species (kit fox, red fox, and muskrat) far in excess of the home range (values larger than 100%). To illustrate this last point, the ACA of 476%, based on the conservatively derived, minimum home range of the kit fox, means that an area almost 5 times as large as the kit fox's home range can be contaminated and yet not require this species to be assessed for health impacts. Where 4AAs based on average (as opposed to maximum) animal densities are considered and where, to maintain conservatism, minimum home ranges are still assumed, derived ACAs (the yellow areas of Figure 2) range from a minimum of 57% of the home range (eastern cottontail rabbit) to more than 10 times the home range size (black-tailed jackrabbit and long-tail weasel).

Table Table 2.. Minimal home range determination for mammals commonly evaluated in ecological risk assessments
SpeciesRange of reported adult body weights (kg)Calculated home range based on lowest reported body weight (acres)aRange of reported home ranges (acres)b
  1. a From Harestad and Bunnell 1979.

  2. b From Burt and Grossenheider 1980; Chapman and Feldhamer 1992; USEPA 1993; CH2M Hill 2001.

Black-tailed jackrabbit Lepus californicus1.3–3.17.4140–49.9
Coyote Canis latrans9.8–15.872,8043,532–16,796
Eastern cottontail rabbit Sylvilagus floridanus0.7–1.83.943–20
Kit fox Vulpes macrotis1.5–35,670840–1,330
Long-tail weasel Mustela frenata0.198–0.34036112.35–298.9
Mink Mustela vison0.567–1.3621,510640–2,470
Muskrat Ondatra zibethicus0.5––0.42
Mule deer Odocoileus hemionus70–15043290–600
Raccoon Procyon lotor4.2–8.83,14039–2,560
Red fox Vulpes vulpes3–714,553123.5–7,410
White-tailed deer Odocoileus virginianus22.5–180135.8319–1,280
Figure Figure 2..

Calculated allowable contaminated areas (ACA), which is the ratio of the 4-animal area (4AA) to the home range, for the larger mammals commonly evaluated in ecological risk assessments. Home ranges are represented as white or open areas. 4AAs based on maximum species density are shown in green and 4AAs based on average species density are shown in yellow. The ACAs (percentages) and corresponding acreages (in parentheses) also are shown. For explanation of values that are greater than 100%, see text.


The foregoing demonstrates rather dramatically that almost all of the commonly evaluated, larger, nonprotected-status mammals are inappropriate for selection in ERAs. The reasons include that contaminated sites are typically too small, that animals are sparsely distributed, and that home ranges of most species are larger relative to the size of the contaminated sites. The density-based algorithm alerts the risk assessor to the reality of the site of interest not containing enough species representatives to justify a risk assessment undertaking. If it is reasonably argued that 10 species representatives, and not 5, constitute the prerequisite for conducting an ERA, then the situation is exacerbated. Not only does the site-size requirement for a terrestrial ERA double, but there are more cases in which the area containing the requisite number of animals exceeds the normative size of sites at which ERAs are formally conducted.

The density-based algorithm for receptor selection is not intended to serve as an iterative screening tool, in which risk assessors consider several densities, beginning with biased (high) values that make it appear, initially, that there is legitimacy in evaluating a given receptor. The ranges of reported densities for the mammals shown in Table 1 underscore that even in the exaggerated case (that is, where the largest-known density is used), not enough animals will be encountered at a typical site to warrant their inclusion in an ERA. Because mammalian receptor selection at contaminated sites was not meant to be an iterative process, subject to the same pitfalls of iterative hazard quotient (HQ) calculation in ERAs (Tannenbaum, Johnson, et al. 2003), only an average species density value should be used for determining the need for a mammal's evaluation. Of note, the 5AAs and 10AAs in Table 1 based on average literature-reported densities are, with only 1 exception, 1 or more orders of magnitude larger than those based on maximum densities. Also of note, the nonavailability of average density information for 3 species listed in Table 1 does not impact on the finding that the only mammals potentially appropriate for study are the Eastern cottontail rabbit and the muskrat (discussed further below).

The hybrid algorithm that constructs the ratio of the 4AA to the home range, aside from distilling the concept of an ACA, corroborates the discoveries of the density-based algorithm. Owing to the nature of terrestrial mammals to migrate or forage over large areas and compounded by species having notably low densities, the prototypical Superfund-type site is generally without need of a mammalian risk assessment. Aside from the fact that an animal's home range may exceed the size of a site, and often by a great deal (as is the case, for example, of the kit fox), the hybrid algorithm illustrates that a significant portion (or even multiples) of a mammal's home range can be contaminated and not matter in a Superfund context, again, because too few species' representatives are contained within such areas. Where a species has a linear home range, as in the case of the mink, the nonsuitability of the mammal is all the more dramatic. The data suggest that of the mammals most often considered in risk assessments, only the Eastern cottontail rabbit and the muskrat may be relevant for evaluation in terrestrial ERAs. Despite having respective ACAs of 57% (average density-based) and 155% (maximum density-based), both species' home ranges are sufficiently small (approximately 3 acres) so as to almost always be subsumed by contaminated sites of interest. It should be noted, however, that muskrats spend most of their lives in or near marshes, lakes, or streams; feed mostly on aquatic vegetation; and are truly not terrestrial species (CH2M Hill 2001). For other than the Eastern cottontail rabbit, the 4AAs calculated using the hybrid algorithm provide an indication of the minimal site size that can reasonably warrant an ERA for mammals. The smallest, average, density-based 4AAs shown in Table 1 (occurring twice in the table) suggest that no land parcel smaller than 80 acres would necessitate a mammalian evaluation. This is a substantially larger figure than the de minimis 1.5- to 2-acre thresholds for sites requiring an ERA, endorsed by a growing number of states (MA DEP 1996; PADEP 1997; TNRCC 2001; WAC 2001). Curiously, although these states recognize the concept of a contaminated land parcel being inconsequential within an ERA context, the basis for the designations are not articulated. Presumably their thresholds reflect animal density and animal home range considerations similar to those described above for birds and mammals, the 2 classes of terrestrial receptors routinely assessed in ERAs (Tannenbaum, Bazar, et al. 2003). Also, it is noteworthy that recently several wide-ranging receptors were excluded from consideration in a risk assessment because the site constituted an area that was less than a stated 10% of the foraging range (Parsons 2004). In that case, the basis for the decision was not provided, although it would appear that a deciding factor was an insufficient degree of contact with the affected site. The ACAs computed in this article, though, defend the areas that can be considered inconsequential on the basis of an insufficient animal presence.

The spatial analyses above, identifying limited opportunities for commonly evaluated larger mammals to be rightfully included in ERAs, corroborate the observations that terrestrial receptors at hazardous waste sites have not been known to display signs of toxicological stress or impact in the past. It has been argued that the majority of ecological receptors are unlikely to ever display signs of chemical-induced stress or impact (Tannenbaum 2002, 2003b, 2005). Previously suggested reasons for the absence of demonstrated effects in the field include the receptor's biological adaptation to its contaminated environment over time (Tannenbaum 2003b), chemical sequestration in the microsites of the soil particle (and subsequent reduced chemical bioavailability; Alexander 1995), and enhanced chemical breakdown in the environment (Alexander 2000), the last 2 occurring in aged soils. The foregoing treatment may suggest that the commonplace phenomenon of mammals not being spatially relevant at contaminated sites, should be added to this list. In practical terms, how could one know or suspect that a site might be harmful or limiting to a receptor, when even under the most ideal of conditions, the receptor is rarely in direct contact with the site, as in the case of red fox being assessed at a 10-acre site (see further discussion below)?

It would appear that the size of the contaminated parcels that regulatory agencies oversee is, in large part, responsible for the situation in which commonly evaluated mammals are found to be inappropriate for study. Probably the best example of this comes from the Resource Conservation and Recovery Act (RCRA) Program's sites, where Superfund ERA guidance (USEPA 1997, 1998) is commonly applied. RCRA solid waste management units (SWMUs) are typically process plants, factory buildings, and the like, and are frequently in the range of less than 1 to 3 acres in size. Commonly, 10 or 20 SWMUs at a listed site (such as a Department of Defense installation) are individually assessed in a risk assessment report. With 5AAs and minimal home ranges (Tables 1 and 2) far exceeding the size of the individual SWMUs, most, if not all, of the ecological assessments conducted for individual SWMUs are rather evidently unnecessary. The preferred approach may be to combine SWMUs to form a considerably larger land parcel, one in which a significant portion of a mammal's time would be spent. However, using a site-wide assessment approach would necessitate the inclusion of noncontaminated parcels among SWMUs. With this approach, mammals that might now be spatially relevant would probably not be found to be at risk because the exposure-point concentrations that drive estimations of potential for risk (i.e., HQs) would be significantly diluted.

Numerous species that fall within the small rodent/insectivore grouping (mice, rats, voles, and shrews) can be expected to be present in virtually all terrestrial habitats throughout the United States. With home ranges that are often dwarfed by the smallest of hazardous waste sites, and with sufficient densities per acre to matter ecologically (relative to the approach adopted in this article), it would appear that representatives of this grouping should routinely be evaluated in terrestrial ERAs. Their nonconsideration in this article, however, reflects 2 realities, 1 programmatic and 1 ecological. Programmatically, crude as it may sound, no remedial actions to date have ever proceeded on the basis of an estimated potential for risk to a small rodent species, and it is unrealistic to think that cleanups in the future will proceed on their behalf (Tannenbaum 2005). Small rodents (just like earthworms), although commonly evaluated, highly soil-exposed, likely to bioconcentrate certain compounds in soil, and a mainstay of many a larger receptors' diets, simply do not trigger considerations for site remediation in the risk-management psyche. This is not to say, however, that their evaluation (on the desktop, in the laboratory, or in the field) does not provide enhancements to ERAs. Recent indications suggest that small rodents are better integrated into ecological assessments when they serve as the test species for a presumptive, reproduction-based, field-truthing method that can shed light on the health of larger terrestrial receptors (Tannenbaum, Bazar, et al. 2003; Tannenbaum 2005). Ecologically, although small rodents and insectivores may be spatially relevant at contaminated sites, their mammalian predators (i.e., species for which a cleanup could realistically proceed) are commonly not. Further, it is unlikely that there exist sufficiently large contaminated sites to give rise to the case in which a predator's diet would consist exclusively of contaminated rodents. Last, the concern that a site's contaminated soil has all but obliterated the rodent population (a condition yet to be shown) to the point where mammalian predators are stressed due to a lack in diet is not defensible. Recalling that nearly all contaminated sites are historically contaminated (Tannenbaum 2002, 2003b, 2005), suggests that predators have long since discovered diet-item insufficiencies and have modified their foraging habits to survive.

Challenges to algorithm application

The prospect of the regulatory community embracing a notion of screening sites for spatial relevance and dispensing with assessments for larger mammals should species “screen out,” is not likely to be readily adopted. Anticipated challenges would likely reflect both an unwillingness to approach ERAs for contaminated sites from other than the status quo and a mindset that overlooks reasonableness as risk assessment's essential tenet (USEPA 1989b). Regulators, for example, can be anticipated to continue insisting that a red fox be evaluated at a 10-acre site, arguing that the fox might spend nearly 100% of its time there. As is clear from the literature, though, the smallest known home range for the red fox is 123.5 acres (Table 2), and therefore, to assume that the fox is only found at the 10-acre site amounts to betraying the fox's essential biology. An equally unreasonable request would be for the fox to be evaluated as spending 85% of its time at the contaminated 10-acre site, on the grounds that the site offers far superior fox habitat than does the site's surrounding acreage. Although the fox, like many other mammals (e.g., deer), may not use its home range uniformly, it would be unrealistic to assume that its time apportionment is so heavily skewed that 85% of its time would be spent at the site, with the remaining 15% of its time spent contacting the other 113.5 acres of its home range. Furthermore, it is quite unrealistic to assume that in each and every instance where a site stands to be assessed, that coincident with the contamination profile, the site happens to have strikingly higher quality habitat than occurs on adjacent noncontaminated property. Even if it would be true that the contaminated site's habitat was of notably superior quality, this condition would still not claim the fox's exclusive site-fidelity. Realistically, other than being tethered to a stake or fenced in, the total number of hours or days that a red fox would spend at a 10-acre site in the course of a year would be so minimal as to constitute only a case of acute exposure, which is of little importance in ERAs (Suter 1992; Tannenbaum 2003b).

At the root of such insistence that wide-ranging mammals be evaluated in ERAs as a matter of course may be the regulator's uneasiness and nonfamiliarity with the possibility of a site legitimately not having a single mammal that is appropriate for study (Tannenbaum 2003a). Until the larger ERA community acknowledges that there is no requirement that at least 1 mammal be assessed at every terrestrial site and accepts the defensible position that mammals do not have to be evaluated in many cases, unrealistic assumptions will continue to dot the risk assessment landscape.

It is curious that readily available spatial-distribution information for commonly evaluated mammals has been overlooked or misapplied for so long, especially when that information exists within the very same documents that are routinely accessed for other ERA needs (USEPA 1993; CH2M Hill 2001). Electing to ignore the availability of such spatial information (that could well indicate a mammal's nonsuitability in an ERA), gives the impression that the information is highly suspect and was only placed in the guidance documents to make them appear more comprehensive. One would think that, if there were lesser trust associated with the spatial information, the guidance would provide necessary caveats regarding their use, but such is not found. Why body weight and normalized ingestion rate information found in the guidance can be, and is, readily used, but animal density and home range information is applied (if at all) only after HQs have been computed, would appear to be a scheme for ensuring that ERAs proceed to at least a screening-level HQ. If the spatial information were to be applied initially in an earlier screening step though, most mammalian receptors would not merit evaluation, a finding that, as mentioned above, regulators would prefer to avoid. Electing to not use readily available spatial-distribution information allows the risk assessor to complete an evaluation but at the expense of being inconsistent in approach. If mule deer, for example, with a reported range in adult body weight of 70 to 150 kg, had been selected as a receptor to evaluate, running a food-chain model with an assumed adult body weight outside of this range would be an unacceptable practice. The results of such modeling would speak only to hypothetical cases that ERA's should not be considering, such as “What would the risk from contaminant exposure be to mule deer at a given site, if their adult body weights were to change over time, notably achieving values significantly lower or higher than what is presently described in the literature?” Thus, selected values for life history and wildlife exposure parameters, necessary for food-chain modeling are always drawn from the universe of values that the parameters can actually assume in nature. Spatial-distribution information used in ERAs should be treated no differently than wildlife-exposure parameter information. To assume an inappropriately high density for a given species at a site when actual animal-density information indicates otherwise amounts to an unsubstantiated case and one that an ERA should not consider. Similarly, to assume that 1 or more species' representatives contact a site frequently, when natural home range information indicates that such would not occur, is to forego an opportunity to simplify an ERA by avoiding extraneous considerations.

One last anticipated challenge would focus on the mammals that are commonly assessed in ERAs, arguing that these are only surrogates for a much-expanded list of mammals that could potentially populate a contaminated site. The challenge would contend that a spatial-relevance screening of the commonly evaluated mammals is insufficient and that a level of uncertainty would surround the anticipated finding that few, if any, mammal species merit consideration.

In light of the availability of quality spatial-distribution information, the best response to this challenge is to recommend a species-by-species review of as many mammal species as are believed to be present at contaminated sites. In this manner, uncertainty concerns that derive from screening only surrogate species would be reduced.

Ecological risk assessment implications

Since the inception of ERAs for contaminated sites some 2 decades ago, the regulator's approach to demonstrating that sites are either safe or risk-posing for terrestrial ecological receptors has been with the use of the HQ method. Although it might be intuitively obvious that a receptor (e.g., a mammal) would spend only a fraction of a percent of its time at a site, regulators would prefer to generate a HQ that attempts to express the ratio of the receptor's anticipated chemical intake to a (no-effect or effect-based) reference dose for the same chemical. Where the HQ is below 1.0, the conclusion that the site is safe for the receptor is a correct one, however, the basis for the conclusion is incorrect. An example to demonstrate this would be where a wide-ranging mammal is assessed at a considerably larger site than the 10-acre site discussed above (such as a 35-acre site) and where the highest of numerous chemical-specific HQs is 0.02. Masked entirely by the HQs is the fact that the site would support less than a requisite number of species' representatives (argued above to be set, conservatively, at 5 animals) to legitimize a HQ-based assessment or any other kind of assessment. The starker reality is that, at most, only 1 species' representative of any of the mammals considered in this article (excluding the Eastern cottontail rabbit) would be anticipated to be present at the site. Thus, the hypothetical, 35-acre site is safe for the many mammals, not because the HQ was below 1.0 but because the site did not have a question about mammal health that needed to be resolved in the 1st place. This situation is akin to the commonly encountered case of an incomplete exposure pathway in human-health risk assessments. If a site is currently zoned for industrial use and, in fact, has industrial activity ongoing, it would be incorrect to assess a residential land-use scenario there for the current time frame. The calculated HQ of 0.02 in the hypothetical case can easily lead to further site misrepresentation. By virtue of a risk assessment having produced HQs, the site remains open at any time for HQ revisitation, as in the case when what is thought to be a more defensible, toxicity reference value (the denominator of a HQ ratio) becomes available. Site misrepresentation is most erroneous and potentially even more damaging to the ERA process when HQs greater than 1.0 are produced for spatially irrelevant mammals. If for example, a Lowest-Observed-Adverse-Effect-Level-based HQ of 75 were computed in an advanced-tier ERA for the red fox at the 35-acre site, it would likely be interpreted to mean that a cleanup must proceed because it appears that fox, on a daily basis, are consuming a chemical with toxic properties at multiples of an effect-level dose. However, even if the hypothetical site were to have the highest-quality habitat for fox, deer, and nearly all other mammals evaluated in this article, there would be no need for a HQ exercise because, at most, there would be 1 species' representative present and not a population.


The algorithms presented above suggest that available spatial-distribution information for mammals can be harnessed to identify those nonprotected status species that are site-relevant and to highlight that, in the usual case, ERAs are marked by the inclusion of mammals that are not ecologically relevant. Inroads to applying spatial-distribution information can only begin if there is a willingness to acknowledge 2 notions, namely, that a certain number of species' representatives can have unhealthful exposures to a contaminated site and that a sizable land parcel can be contaminated and still be ecologically inconsequential. Arriving at that number of animals and the land parcel size can be easily accomplished because data are available to support such determinations. Applying algorithms, such as those presented here, can significantly reduce the need to assess, monitor, or remediate sites on behalf of mammals. Seemingly, algorithms similar to those provided here can be developed to also screen birds for their relevance in ERAs for contaminated sites.


The author thanks Robyn Lee for her technical assistance. The following individuals carefully reviewed the manuscript and provided helpful contributions: Dennis Druck, Lia Gaizick, David Reed, and Keith Williams. The technical assistance with graphics of Joyce Kopatch and the editing of Joyce Woods are very much appreciated.

Disclaimer—The opinions or assertions contained herein are the views of the author and are not to be construed as official or as reflecting the views of the Department of the Army or the Department of Defense.