Printer Friendly

Factors affecting species' risk of extinction: an empirical analysis of ESA and NatureServe listings.

I. INTRODUCTION

There are two principal sources of data on ecologically imperiled species in the United States: (1) listings of threatened and endangered species under the Endangered Species Act (ESA, 1973), and (2) listings of species considered by NatureServe to be at risk of extinction. The language of the ESA admits no influence on the determination of listings other than scientific necessity:
 The Secretary shall make determinations required by subsection (a)(1)
 of this section solely on the basis of the best scientific and
 commercial data available to him after conducting a review of the
 status of the species and after taking into account those efforts, if
 any, being made by a State or foreign nation, to protect such species,
 whether by predator control, protection of habitat and food supply, or
 other conservation practices, within any area under its jurisdiction;
 or on the high seas. (U.S. Code, Title 16, Chapter 35, Section
 1533(b)(1)(A), emphasis added)


Listings of species at-risk of extinction by NatureServe (2002) are based on:
 ... a consistent and rigorous methodology for assessing extinction
 risk that is based on evaluation of multiple factors. Evaluation
 criteria include: the number and condition of populations and
 individuals; the area or range occupied by the species; population
 trends (that is, whether numbers are increasing, stable or declining);
 and known threats. Biologists assess each species against these
 multiple risk factors based on the best available scientific
 information and assign the appropriate conservation status rank.


In theory, then, listings under the ESA and NatureServe should be consistent. In fact, there is substantial discrepancy between the two indices, with the NatureServe listings containing many more species considered ecologically imperiled than have obtained recognition and protection under the ESA. In part, this discrepancy may reflect financial constraints imposed on the U.S. Fish and Wildlife Service, which is charged with implementing the ESA. In addition, it has been argued that implementation of the ESA has been influenced by economic considerations, as translated through the political process. For example, Bean (1991) and Mehmood and Zhang (2001) have argued that economic factors played an important role in determining congressional votes on amendments to the ESA. Moreover, there is evidence that how fast species proposed for listing actually get listed (Bean, 1991; Ando, 1999, 2001, 2003), the types of species that get listed (Metrick and Weitzman, 1996; Weitzman and Metrick, 1998; Ando, 2003), and the geographic distribution of listings (Rawls and Laband, 2004) are subject to the influence of economic interests through the political process. Complicating matters is the fact that implementation of the ESA is also subject to judicial intervention (Associated Press, 2004).

These criticisms of how the ESA has been implemented may create doubts about whether ESA listings are based significantly on scientific criteria. In this article, the authors provide empirical evidence in support of scientific foundations for ESA listings. They do so by analyzing factors that influence species' ecological imperilment across states, using two different measures of species imperilment: the fraction of all species in a state identified by NatureServe as being "at-risk" of extinction, and the fraction of species in a state listed under the U.S. Fish & Wildlife Service's Endangered Species Act (ESA). At issue is whether the factors that influence listings compiled by NatureServe also influence ESA listings.

One of the authors' principal findings is that both measures of species' ecological imperilment are influenced strongly by the fraction of species found only in each state. These endemic species exist in relatively small, ecologically distinct niches and are characterized by small populations that are consistent with being ecologically vulnerable. Another consistent finding is that aquatic species are especially subject to being listed as ecologically imperiled. However, there are notable differences between the two explanatory models, especially with respect to the impact of human population growth and the imperilment of plant species. In the next section, the authors identify factors that influence species' ecological viability. They then introduce their empirical model and data, followed by presentation and discussion of their empirical findings.

A. Factors that Influence Species' Ecological Viability

Biodiversity and Endemic Species. For a given rate of naturally occurring extinctions at any given point in time, the number of ecologically fragile species per fixed geographic area will be greater in areas inhabited by relatively large numbers of species than in areas not supporting much biodiversity (Wilson, 1988). One way of making apt comparisons across desired units of analysis is to identify the number of fragile species in each unit, with an explicit control variable for total biodiversity in each unit. Alternatively, one can divide the number of fragile species in the focus area by the total number of species found in the focus area. This approach, which the authors adopt, permits comparison across units to be made in terms of the proportion of species that are imperiled.

In addition, due to wider ranges of moisture, temperature, and geophysical attributes, some states have greater numbers of unique ecological niches than others, which support plant and animal species found nowhere else. These endemic species are inherently more likely than species with wider ranges of habitat to be characterized by low populations and to be vulnerable to relatively sudden changes in environmental conditions. Ceteris paribus, the authors expect the fraction of ecologically imperiled species in a state to be positively impacted by the fraction of species endemic to that state.

Human Population Growth. The health and/or ecological viability of plant and animal species may also be impacted (for better or worse) by anthropogenic activity. It seems likely that the severity of any deleterious impacts is positively related to the significance and extent of man's activity. Human activity (e.g., residential and commercial building) directly pressures populations of target plants and animals. As a result, as human populations increase, they "crowd out" other species. If human population-induced crowding-out of other species occurs to the point of jeopardizing the viability of those other species, the number of ecologically fragile species should be greater in states characterized by higher rates of growth in the population of Homo sapiens than in states with lower population growth rates of Homo sapiens, ceteris paribus.

Species Type. The impact of mankind on non-human life forms is not uniform across species. Some species thrive in the presence of humans; others appear to be quite human-intolerant. To account for the diversity of species in each state, the authors include variables measuring the portion of a state's species that NatureServe classifies as: (1) terrestrial vertebrates, (2) aquatic vertebrates, (3) vascular plants, or (4) invertebrates. As the fraction of all species that are terrestrial increases, the fraction of species that are ecologically imperiled might be expected to climb, since man is also a terrestrial being. Thus, the crowding-out effect referred to previously may be, in the main, especially germane to other terrestrial species. However, many terrestrial species are mobile, at least within some geographic range, and thus able to escape the immediate impact of human presence/activity. This suggests that although terrestrial species are in direct competition with Homo sapiens for space, the crowding-out effects implied by a steadily increasing human population may not be significantly detrimental to most terrestrial species.

On the other hand, even though humans are not overtly aquatic creatures, the effects of man's presence and activities are felt acutely by aquatic species. This is because people directly alter the natural composition of water by using it as a coolant and as a repository for waste materials, and because water filters out land and air pollution, exposing aquatic life to high concentrations of these pollutants. Further, aquatic life cannot easily escape some, if not most, of these pollutants, since they travel in suspension and move wherever water moves. Many terrestrial animals can actively avoid mankind's toxic residuals; aquatic life cannot. Thus, one might hypothesize that aquatic life may be more imperiled than terrestrial life by man's activities. This is an unresolved empirical question. Furthermore, because plants live in fixed locations and cannot migrate easily, they are, in theory, especially susceptible to exogenous disturbance of their environment. Again, this suggests that plants may be differentially adversely affected by man's presence as compared with mobile terrestrial life. Whether invertebrates are more or less imperiled than other taxonomic groups by man's activities is difficult to determine a priori. In the absence of prior expectation about the strength of species-class effects, the authors look to the coefficient estimates to gain information. However, for the purposes of this investigation, the issue of interest is whether ESA listings and NatureServe listings are both influenced by similar ecological considerations. Thus, the authors will be looking for evidence that the influence of species-class effects (if any) is reflected in both ESA and NatureServe listings.

Hunting, Fishing, Farming, and Forest Cover. Hunters and fishermen have asserted that they promote the ecological viability of the species they hunt because their license revenues fund research on species reproduction, health, management, habitat enhancement, and so on. But there is an associated reason to believe that hunting and fishing promote the ecological viability of game species: private landowners have an economic incentive to set aside and/or develop habitat conditions that game species desire. By proving to prospective hunters and fishermen that desirable game species are abundant on their property, private land owners can profit by selling hunting and/or fishing rights. As a side effect of enhancing the ecological well-being of the target species, private land owners' preservation and development of habitat for game species also expands habitat for certain non-game species that share ecosystems with game species. But the aggregate net effect of hunting and fishing on species imperilment is not clear. Human disturbance of critical micro-habitat, albeit unintended, still may harm the affected species. In addition, by encouraging the ecological vitality of target species, there may be numerous indirect effects (some positive, some negative) on other species.

The incidence of hunting and fishing (or agriculture) in a state also may serve as a proxy for the fraction of the population that is rural. While a specific link between percent of rural population and fraction of ecologically imperiled species has not, to the authors' knowledge, been established, one anecdotal claim is that rural dwellers have a greater appreciation for nature than do urban dwellers. If so, then states with a relatively greater proportion of agricultural activity than others may have fewer ecologically fragile species because the relatively large rural population avoids activities that degrade the environment. It is also possible that farming and ranching promote more diverse habitats for species than exist in urban areas. If so, this also would suggest an inverse relationship between measures of agricultural activity and measured species imperilment.

As the amount (percent) of forest cover in a state increases, the authors generally would expect to observe fewer ecologically threatened species. This is because for those species that are forest-dependent, more forest implies less resource stress. Having said this, it should be pointed out that not all forest cover is created equal. While not much is known yet about the ecological consequences of intensively managed forests, there are fears that monoculture forests support far less species diversity than do natural forests (Carlton, 2004). It is also possible that NatureServe or the FWS perceive threats (due perhaps to imminent forest loss) to be great for species dwelling in forests. Furthermore, some environmental groups have pressured the FWS to list animal species in order to protect old-growth forests, which would indicate a positive relation between listings and forest cover. Thus, the sign of the forest variable is difficult to predict a priori.

B. Political Variables

Because of the previously documented concern that ESA listings may be subject to political influences, the authors consider several political variables in their models. First, they include a variable to measure the environmental rating of each state's congressional delegation. The authors use the League of Conservation Voters (LCV) rating on a scale of 0 to 100, with 100 being the highest environmental score, averaged across the members of each state's delegation to the U.S. House of Representatives, over the 1973-2000 period. It is tempting to declare that states with higher LCV scores will be characterized by more ESA listings. However, the relationship may not be this simple. It may be that states with relatively large numbers of voters who are sensitive to environment issues elect representatives who support pro-environment legislation and also have adopted state-specific and local measures to protect the environment. This means that a relatively high LCV score for a state's congressional delegation could be related negatively to that state's relative incidence of ecologically fragile species.

The second political variable the authors use is the percent of the 1973-2000 period during which a state had a member on the Interior Subcommittee of the House of Representatives' Appropriations Committee. Since this subcommittee provides budgetary oversight for the U.S. Fish & Wildlife Service (which controls the listing process pursuant to the ESA), the authors expect that states with more representation on this subcommittee are better able to influence listings under the ESA. For similar reasons, a reviewer suggested that the authors include variables that reflect each state's representation on the Environment and Natural Resources Subcommittee of the House Merchant Marine and Fisheries Committee, and on the Clean Water, Fisheries, and Wildlife Subcommittee of the Senate Environment and Public Works Committee. (1) Both of these subcommittees are critical to legislative re-authorization of the ESA. This reviewer also suggested including a variable that measures each state's representation in the White House, through either the presidency or vice presidency. The fact that the president can withhold his signature from legislatively approved bills provides the president (and surely also the vice president) opportunities to make suggestions to members of the above-listed subcommittees when potential listings are of special interest to their home states.

As pointed out by Ando (1999, 2001), the net effect of political representation on these committees (or in the White House) is ambiguous. No doubt, economic development interests occasionally collide, politically, with environmental interests. But politicians represent all interests, so there is no a priori way of discerning whether representation on the aforementioned committee will be, on balance, reflected in more or fewer listed species.

C. Dependent Variables

There are at least two sources of information on the number of ecologically imperiled species per state: the U.S. Fish and Wildlife Service, which lists threatened and endangered species under mandate from the Endangered Species Act (1973), and NatureServe. NatureServe was formerly the statistical arm of the Nature Conservancy, and since 1999 has operated as "... the country's leading source of biological information for conservation planners, government agencies and land managers" (Stevens, 2000). In cooperation with natural heritage program members in all 50 states, NatureServe has compiled, and maintains, a detailed database of over 21,000 plant and animal species in the United States, including nearly 16,200 vascular plants, approximately 2,550 native vertebrate animal species (including mammals, birds, reptiles, amphibians, and freshwater fishes), and a wide spectrum of invertebrates (including "all 2,600 species in the following groups: freshwater mussels, freshwater snails, crayfishes, large branchiopods, butterflies and skippers, underwing moths, tiger beetles, and dragonflies and damselflies"). Since the number of ecologically imperiled species is likely to be related to the overall number of species in existence in a given state, the authors model species imperilment as the fraction of all known (according to NatureServe) species in a state that are identified by NatureServe as at risk of extinction or on the ESA list.

NatureServe's "at-risk" species are defined as the number of "a state's plants and animals that are at risk of extinction due to rarity or other factors." This measure includes species with a conservation status of extinct, imperiled, or vulnerable (corresponding to Global Heritage Conservation Ranks of GX, GH, G1-G3). If both "at-risk" and ESA listing variables constitute unbiased measures of species' ecological imperilment, then the authors would expect the effects of the explanatory variables to be similar for both models. However, if the coefficients are different, then the authors might reasonably question whether both measures of species imperilment truly are unbiased.

II. MODEL AND DATA

The specific model the authors estimate is:

(1) Percent Imperiled Species[.sub.i] = [a.sub.1]PctEndemic[.sub.i] + [a.sub.2]PctTerrestrial[.sub.i] + [a.sub.3]PctAquatic[.sub.i] + [a.sub.4]PctPlant[.sub.i] + [a.sub.5]PctInvert[.sub.i] + [a.sub.6]PopGrowth[%.sub.i] + [a.sub.7]PctForestland[.sub.i] + [a.sub.8]PctFarmland[.sub.i] + [a.sub.9]PctHuntFish[.sub.i] + [a.sub.10]LC[V.sub.i] + [a.sub.11]Housefws[%.sub.i] + [a.sub.12]Senatefws[%.sub.i] + [a.sub.13]Presvp[%.sub.i] + [a.sub.14]ISHAC[%.sub.i] + [[epsilon].sub.i].

The authors introduce two measures of Imperiled Species:

PctAt-Risk[.sub.i] = the number of species in state i identified by NatureServe as ecologically "at risk" in 2000 divided by the total number of vascular plant and animal species catalogued by NatureServe that are found in state i, and

PctListed[.sub.i] = the number of species in state i listed as threatened or endangered by the U.S. Fish & Wildlife Service under the auspices of the Endangered Species Act, divided by the total number of vascular plant and animal species catalogued by NatureServe that are found in state i.

Definitions of the explanatory variables are as follows:

PctEndemic[.sub.i] = the number of species endemic to state i, divided by the total number of plant and animal species catalogued by NatureServe that are found in state i.

PctTerrestrial[.sub.i] = the number of vertebrate terrestrial species found in state i, divided by the total number of plant and animal species catalogued by NatureServe that are found in state i.

PctAquatic[.sub.i] = the number of vertebrate aquatic species found in state i, divided by the total number of plant and animal species catalogued by NatureServe that are found in state i.

PctPlant[.sub.i] = the number of plant species found in state i, divided by the total number of plant and animal species catalogued by NatureServe that are found in state i.

PctInvert[.sub.i] = the number of invertebrate species found in state i, divided by the total number of plant and animal species catalogued by NatureServe that are found in state i.

PopGrowth[%.sub.i] = average annual population growth rate in state i from 1973-2000.

PctForestland[.sub.i] = the proportion of state i's area that was characterized by forest cover, averaged over 1977, 1987, and 1997.

PctFarmland[.sub.i] = the proportion of state i's area that was devoted to agricultural (farming and ranching) production, averaged over 1973-2000.

PctHuntFish[.sub.i] = the proportion of state i's population that engaged in hunting/fishing in 1991.

LC[V.sub.i] = the average rating by the League of Conservation Voters (on a scale of 0-100) of state i's delegation to the House of Representatives during 1973-2000.

Housefws[.sub.i] = the proportion of time from 1973 to 2000 that state i was represented on the Environment and Natural Resources Subcommittee of the House Merchant Marine and Fisheries Committee, which is responsible for making periodic recommendations regarding re-authorization of the ESA to the full House of Representatives.

[FIGURE 1 OMITTED]

Senatefws[.sub.i] = the proportion of time from 1973 to 2000 that state i was represented on the Clean Water, Fisheries, and Wildlife Subcommittee of the Senate Environment and Public Works Committee, which is responsible for making periodic recommendations regarding re-authorization of the ESA to the full Senate.

Presvp[%.sub.i] = the proportion of time from 1973 to 2000 that state i was represented in the White House, either by the president or vice president.

ISHAC[%.sub.i] = the proportion of time from 1973 to 2000 that state i was represented on the Interior Subcommittee of the House Appropriations Committee, which is responsible for budgetary oversight for the U.S. Fish and Wildlife Service--the agency charged with making ESA listing decisions.

[[epsilon].sub.i] = is the error term.

Equation (1) was estimated (for each measure of species imperilment) using spatial autocorrelation correction. Spatial econometric techniques have been devised to examine relationships among nearby entities. Perhaps the most fitting description of the nature of the problem is Tobler's (1979) first law of geography: "everything is related to everything else, but near things are more related than distant things." For example, if species face relatively high risks in one state, species in neighboring states are likely to be affected by spillover threats. Figures 1 and 2 show the distribution of these two measures of species imperilment.

The two types of spatial regression models that have been employed most are spatial lag models and spatial error models. The spatial lag model (also known as the mixed regressive-spatial autoregressive model) is postulated as:

(2) y = [rho]Wy + X[beta] + [epsilon]

where [rho] is the coefficient of the spatially lagged dependent variable, W is a spatial weights matrix (to be discussed below), X is an N by K matrix, [beta] is a K by 1 vector of parameters associated with the exogenous variables X, and [epsilon] is a normally distributed disturbance term with a diagonal covariance matrix.

The spatial error model (also known as the linear regression model with a spatial autoregressive disturbance) is postulated as:

(3) y = X[beta] + [epsilon]

[epsilon] = [lambda]W[epsilon] + [mu]

where [lambda] is the autoregressive coefficient, W is a spatial weights matrix, and [mu] is a well-behaved (i.e., homoskedastic and uncorrelated) disturbance term (Anselin, 1988, pp. 34-35).

[FIGURE 2 OMITTED]

The consequences of ignoring spatial correlations are serious. Ignoring spatial lag dependence will yield biased and inconsistent ordinary least squares (OLS) estimators. Ignoring spatial error dependence will yield unbiased but inefficient OLS estimators, and the OLS standard errors will be biased. Likelihood ratio and Lagrange multiplier test statistics have been developed to determine which model best fits the data (Anselin, 1988, pp. 58-59).

Several potential spatial weights matrices have been employed by researchers. The authors employ a binary contiguity matrix of Moran (1948) and Geary (1954). If two states share a common border, they are treated as neighbors and a 1 is assigned to the weights matrix; if they do not share a common border, a value of 0 is assigned. A contiguity matrix is N by N. For the authors' example of 49 U.S. states, the contiguity matrix has 2,401 cells of zeros or ones. (Note that Alaska has no neighboring states. Also note that Hawaii is not included, because of its unique island characteristics.) Sample statistics for the data are reported in Table 1.

III. FINDINGS

The question of interest is whether factors that significantly influence NatureServe listings also significantly influence ESA listings. If so, this would suggest that, indeed, ESA listings do have a legitimate scientific foundation. If the U.S. Fish and Wildlife Service is a little more cautious than NatureServe, the implicit intercept term (percent terrestrial, percent aquatic, percent plant, and percent invertebrate sum to 1) in the ESA listings equation might be smaller, but the coefficients on the remaining explanatory variables in the two models should show a general pattern of consistency. In certain respects, this is exactly what the authors find. However, they do find notable differences between the two estimated models.

The estimated models, shown in Tables 2 and 3, have extremely high fit for cross-section analyses--the independent variables explain approximately 85-95% of the variation in the number of "at-risk" (ESA-listed) species per state. The spatial correlation tests, shown at the bottom of the tables, indicate that the spatial lag models are appropriate. Only spatial lag results are shown. The significance of the endemic variable is the reason for this strong fit. In the "at-risk" model, the coefficient on the endemic species variable consistently is somewhat greater than 1.0, meaning that the percent of "at-risk" species increases by more than 1% for every additional 1% increase in endemic species. Because of their unique characteristics and often small populations, endemic species are more likely to be considered "at-risk." Additional species that rely on these endemic species may be "at-risk" as well. In the ESA listings model, the endemic coefficient is only about one-fifth as large. This is evidence that the U.S. Fish and Wildlife Service is more conservative (more constrained may be more accurate) in listing species than is NatureServe.

The authors find no evidence of a statistically significant relationship between the percentage of terrestrial species and either measure of species imperilment. On the other hand, they find strong, consistent evidence of a positive relationship between the percentage of aquatic species in a state and species imperilment, for both measures. The authors note that the estimated impact of PctAquatic is three times greater for at-risk listings than for ESA listings. They also find a positive and significant relationship between the percentage of plants and the percent of species listed as being at risk, but this variable is not significant in the ESA listing model. These latter two results provide some evidence in support of the claim by Metrick and Weitzman (1996) that there may be a bias in the ESA listing process that favors large, "warm fuzzy" animals--species that are more likely to be adversely impacted by man's hunting/fishing activity--over ugly, cold-blooded, immobile animals and plants, even though the latter may be more critically imperiled than the former. Although the authors' results are not directly comparable with Metrick and Weitzman (1996) (because they did not consider plants, and because the authors group species into terrestrial and aquatic types), there is evidence that plants do not receive as much attention from the FWS as from NatureServe. Nonetheless, the endemic and species categories explain a sizable portion of the variation across states in both dependent variables and provide evidence that ESA listing outcomes have, in some measure, a legitimate scientific basis.

The authors find a significant, positive impact of population growth on at-risk species, but no evidence that population growth influences ESA listings. This provides additional indication that ESA listings may be determined, in part, by factors other than scientific necessity, notwithstanding the language of the Act.

The authors find consistent evidence that both the percentage of "at-risk" species and the percentage of ESA listings are inversely related to the percentage of the state population that engages in hunting/fishing activity (PctHuntfish). In turn, the percentage of the state population that engages in hunting/fishing activity is highly correlated with the percentage of land in agriculture (PctFarmland), so the authors do not include both variables in any of their estimated models. A reviewer suggested that both variables may reflect the proportion of the state population that is non-urban or, perhaps more accurately, has direct contact with and appreciation for wildlife. It is also possible that agricultural land may provide some habitat for species. The negative and significant estimated co-efficient on PctFarmland is consistent with both of these interpretations. As for many of the other variables, the authors note that the coefficients are larger for PctHuntfish and PctFarmland in the at-risk model.

Both measures of species imperilment are related in a positive and significant manner to the percent of forest cover in a state. (Since PctForestland is highly [negatively] correlated with PctFarmland, the authors do not include both variables in any models.) As mentioned earlier, the impact of forests is difficult to predict a priori. If only the ESA listing variable were positively (and significantly) related to forest cover, perhaps the finding could be attributed to the desire of environmental groups to use the ESA to protect old-growth forests. However, since the data from both NatureServe and the FWS suggest a positive relationship between species imperilment and extent of forest cover, additional investigation is warranted.

Turning to their measures of political influence, the authors generally find little evidence that political considerations influence either "at-risk" listings (an expected result) or ESA listings (an unexpected result). Across numerous model estimations in addition to the ones reported in Tables 2 and 3, the authors fail to observe a significant relationship between ESA listings and the extent of state representation on the Interior Subcommittee of the House Appropriations Committee; the Environment and Natural Resource Subcommittee of the House Merchant Marine and Fisheries Committee; the Clean Water, Fisheries, and Wildlife Subcommittee of the Senate Environment and Public Works Committee; or the presidency or vice presidency. (2)

Finally, the authors find that a state's League of Conservation Voters rating, as averaged across its congressional delegation, consistently demonstrates a significant inverse relationship with the percentage of species in a state that are at risk, but only one model shows a statistically significant relationship with the percentage of species listed as threatened or endangered under the ESA. The LCV rating serves as an indirect measure of how environmentally "concerned" the voters in a state are, at least relative to voters in other states. Such concern might reasonably be expected to translate into more care for the environment, broadly speaking. Thus, the authors might expect to observe an inverse relationship between LCV scores and the measures of species imperilment. However, one should not be too quick to jump to this conclusion. It is also plausible that some representatives are shrewd enough to receive high LCV ratings for the policies that impact the nation at large, but are adept at shielding their own constituents from the economic development harm of an ESA listing. In other words, these politicians' "concern" for the environment manifests itself, in terms of environmental regulations and policies, more at the national level than in their own state (Hussein and Laband, 2005). This would indeed yield the authors' finding of an inverse relationship between LCV scores and "at-risk" listings, but less robust relationship between LCV scores and ESA listings by state. But this is the only evidence that the authors are able to present (and it is inferential evidence at best) that pressure may be brought to bear by politicians in ways that influence the distribution of listings under the ESA. It is also possible that liberal states (which generally have high LCV scores) tend to be in the north and east, where there are fewer endangered species. (3)

IV. DISCUSSION

Although listings under the Endangered Species Act are supposed to be determined strictly on the basis of scientific evidence pertaining to species' ecological circumstances, previous investigators have argued that both the timing of listings and the types of species listed are influenced by non-scientific considerations. This raises a question about how much scientific underpinning there is to ESA listings. The authors investigated this question empirically by directly comparing models that estimate the impact of ecological and political factors on ESA listings and NatureServe listings of species at risk of extinction. In several areas, the authors find that scientific factors play a significant role in ESA listings. First, their findings reveal that both indices of species' ecological imperilment are influenced strongly (positively) and consistently by at least three ecological considerations: (1) the fraction of species in a state that are endemic, (2) the percentage of aquatic species in a state, and (3) the percentage of forest cover in a state. Second, the authors find consistent evidence that two anthropogenic factors, the percentage of the over-age-16 population that engages in hunting and fishing and the percentage of farmland in a state, negatively influence both listings of imperiled species. Both of these variables arguably affect the ecological viability of certain plant and animal populations.

But the authors do find areas where scientific factors are not significant for ESA listings but are for NatureServe's "at-risk" listings. First, differences in the size of the estimated coefficients on the variables in the NatureServe listings model (larger) as compared with the ESA listings model (smaller) are consistent with the observation that NatureServe lists many more species as imperiled than the ESA does. Second, NatureServe listings show a statistically significant impact of plants, whereas ESA listings do not. The authors also find that human population growth has a positive impact on NatureServe listings, but not on ESA listings. Perhaps the FWS may be pressured by citizens in rapidly growing areas to not list species under the ESA. In conclusion, why these differences between ESA and NatureServe listings exist remains a legitimate topic of inquiry.

REFERENCES

Ando, A. W. "Waiting to be Protected Under the Endangered Species Act: The Political Economy of Regulatory Delay." Journal of Law and Economics, 42. April 1999, 29-60.

______. "Economies of Scope in Endangered Species Protection: Evidence From Interest-Group Behavior." Journal of Environmental Economics and Management, 41, 2001, 312-32.

______. "Do Interest Groups Compete? An Application to Endangered Species." Public Choice, 114, January 2003, 137-59.

Anselin, Luc. Spatial Econometrics: Methods and Models. Dordrecht: Kluwer, 1988.

Associated Press. "Group May Sue to Block Species Protection." CNN.com at www.cnn.com/2004/LAW/11/16/endangered.species.ap/index.html. Accessed November 16, 2004.

Bean, M. "Looking Back Over the First Fifteen Years," in Balancing on the Brink of Extinction: The Endangered Species Act and Lessons for the Future, edited by K. Kohm. Washington, D.C.: Island Press, 1991.

Brownson, C. B. Congressional Staff Directory. Mount Vernon, VA., various years.

Carlton, J. "Forest Fire: In the Sierras a Raging Debate over Clear-Cutting." Wall Street Journal. May 27, 2004: A1, A10.

Geary, R. "The Contiguity Ratio and Statistical Mapping." The Incorporated Statistician, 5, 1954, 115-45.

Hussain, Anwar and David N. Laband. "The Tragedy of the Political Commons: Evidence From U.S. Senate Roll Call Votes on Environmental Legislation." Public Choice, 124, 2005, 353-64.

League of Conservation Voters. National Environmental Scorecard, various years.

Mehmood, S. R., and D. Zhang. "A Roll Call Analysis of the Endangered Species Act Amendments." American Journal of Agricultural Economics, 83(3), August 2001, 501-12.

Metrick, A., and M. L. Weitzman. "Patterns of Behavior in Endangered Species Protection." Land Economics, 72, 1996, 1-16.

Moran, P. "The Interpretation of Statistical Maps." Journal of the Royal Statistical Society B, 10, 1948, 243-51.

NatureServe. States of the Union: Ranking America's Biodiversity. Arlington, VA, April 2002.

Rawls, R. P., and D. N. Laband. "A Public Choice Analysis of Endangered Species Listings." Public Choice, 121(3-4), October 2004, 263-77.

Stevens, W. K. "U.S. found to be a leader in its diversity of wildlife." New York Times, March 16, 2000, sec. A. p. 18.

Tobler, W. "Cellular Geography," in Philosophy in Geography, edited by S. Gale and G. Olsson. Dordrecht: Reidel, 1979, 379-86.

U.S. Department of Agriculture. Forest Resources of the United States. 1997, edited by W. Brad Smith, John S. Vissage. David R. Darr, and Raymond M. Sheffield. Washington, D.C.: U.S. Forest Service, 2001.

U.S. Department of Agriculture, National Agricultural Statistics Service. 2002 Census, Percent of land covered by farms, 1973-2000. http://www.nass.usda.gov:81/ipedb/.

U.S. Department of Commerce, Bureau of the Census. 1991 National Survey of Fishing, Hunting, and Wildlife-Associated Recreation, Table 58, http://www.census.gov/prod/1/gen/interior/.

U.S. Fish and Wildlife Service. Threatened and Endangered Species, http://ecos.fws.gov/tess_public/TESSWebpage.

Weitzman, M. L., and A. Metrick. "Conflicts and Choices in Biodiversity Preservation." Journal of Economic Perspectives, 12(3), 1998, 21-34.

Wilson, E. O. "The Current State of Biological Diversity," in Biodiversity, edited by Edward O. Wilson and Francis M. Peter. Washington, D.C.: National Academy Press, 1988.

DAVID N. LABAND and MICHAEL NIESWIADOMY*

*This is a revision of an article presented at the Western Economic Association International 79th annual conference, Vancouver, July 1, 2004, in a session organized by David Laband, Auburn University. This research was supported by a McIntire-Stennis grant to the first author through the School of Forestry and Wildlife Sciences at Auburn University. The authors appreciate the assistance provided by Karen Anderson and Tim Hall of the U.S. Fish and Wildlife Service and the helpful comments of Wade Martin and two reviewers. The authors are solely responsible for remaining errors.

Laband: Professor of Forest Economics & Policy, Auburn University, Auburn, AL. E-mail labandn@auburn.edu

Nieswiadomy: Professor of Economics, University of North Texas, Denton, TX. E-mail miken@unt.edu

ABBREVIATIONS

FWS: U.S. Fish and Wildlife Service

ESA: Endangered Species Act

LCV: League of Conservation Voters

OLS: Ordinary Least Squares

1. The exact names of these subcommittees changed occasionally over the 28-year period that the authors analyzed.

2. It should also be noted that ESA listings surely are influenced by court cases. This might be responsible for the divergence in numbers of ESA listings versus NatureServe listings. However, it is not at all clear that federal court cases significantly affect the state-by-state distribution of ESA listings.

3. The authors thank a reviewer for this comment. Also, this reviewer suggested that the empirical results for ESA listings might be influenced by a few states with large numbers of listed species. To check this, the authors re-estimated all of their models without California, Florida, and Texas (states with large numbers of endemic species and large numbers of ESA-listed species) to see whether their empirical results were different from the ones reported in Tables 2 and 3. However, the results reported in Tables 2 and 3 are robust to both samples of states.
TABLE 1 Sample Statistics

VARIABLE MEAN ST.DEV. MIN MAX

Pctlisted (a) 0.010 0.007 0.004 0.044
PctAt-risk (b) 0.07 0.05 0.02 0.29
PctEndemic (b) 0.01 0.03 0.00 0.19
PctTerrestrial (b) 0.14 0.02 0.10 0.22
PctAquatic (b) 0.05 0.02 0.01 0.09
PctPlant (b) 0.69 0.05 0.60 0.81
PctInvert (b) 0.12 0.03 0.04 0.18
Popgrow% (c) 1.13 0.94 0.00 4.80
PctForestland (d) 41.75 24.34 1.53 89.66
PctFarmland (e) 42.65 25.56 0.31 95.28
PctHuntfish (f) 0.25 0.074 0.130 0.41
Senatefws% (g) 18.62 19.89 0.00 83.33
Housefws% (g) 32.73 36.30 0.00 100.00
Presvp (b) 4.16 9.98 0.00 42.86
ISHAC% (i) 0.16 0.25 0.00 0.86
LCV (j) 46.57 17.95 13.40 84.60

Sources:
(a) Listings as a proportion of total species. U.S. Fish & Wildlife
Service: http://ecos.fws.gov/servlet/TESSWebpageUsaLists?state=all.
(b) Proportion of total species. NatureServe (2002).
(c) Percent per annum over 1973-2000 period. Data:
http://www.census.gov/popest/archives/index.html.
(d) Percent of land area covered by forests (public and private)
averaged over 1977, 1987, 1997 period. Data from USDA (2001).
(e) Percent of land covered by farms, 1973-2000. Data:
http://www.nass.usda.gov:81/ipedb/.
(f) Proportion of age 16 and older in hunting and fishing. 1991 National
Survey of Fishing, Hunting, and Wildlife-Associated Recreation.
http://www.census.gov/prod/1/gen/interior/, Table 58.
(g) Percent of time that a state was represented on the House (or
Senate) subcommittee with oversight of U.S. Fish and Wildlife Service
over the 1973-2000 period. Membership found in Brownson (annual
1973-2000), Congressional Staff Directory.
(h) Percent of time that a state had a president or vice president in
office, 1973-2000. Source: see note g above.
(i) Proportion of years during 1973-2000 when state i was represented on
the Interior Subcommittee of the House Appropriations Committee. Source:
see note g above.
(j) Average League of Conservation Voters 1973-2000 rating (from 0 to
100) of state i's House of Representatives' delegates.
http://www.lcv.org/scorecard.

TABLE 2 At-Risk Species Regression Estimation Results--49 States

 Estimated Coefficients (standard errors)
Explanatory Variable Model 1 Model 2

W_Atrisk 0.2980*** (0.752) 0.2825*** (0.0719)
PctEndemic 1.0414*** (0.0733) 1.0693*** (0.0754)
PctTerrestrial -0.0956 (0.0862) -0.0831 (0.0834)
PctAquatic 0.2993*** (0.1179) 0.3142*** (0.1064)
PctPlant 0.0622** (0.0280) 0.0646*** (0.0273)
PctInvert -0.0001 (0.0714) -0.0193 (0.0713)
Popgrow% 0.0102*** (0.0028) 0.0105*** (0.0028)
Senatefws% -3.2E-6 (0.0001)
Housefws% -6.7E-5 (5.5E-5)
Presvp%
ISHAC%
LCV -0.0004*** (0.0001) -0.0004*** (0.0001)
PctHuntfish -0.0414 (0.0294) -0.0472* (0.0277)
PctForestland 0.0002** (8.76E-5) 0.0002** (9.07E-5)
PctFarmland
[R.sup.2] 0.9465 0.9481
N 49 49
Likelihood statistic 150.374 151.086
Spatial lag LR test stat. 13.17*** 12.19***

 Estimated Coefficients (standard errors)
Explanatory Variable Model 3 Model 4

W_Atrisk 0.3282*** (0.0740) 0.3404*** (0.0713)
PctEndemic 1.1173*** (0.0826) 1.0709*** (0.0692)
PctTerrestrial -0.0909 (0.0889) -0.0285 (0.0879)
PctAquatic 0.4044*** (0.1068) 0.3734*** (0.1099)
PctPlant 0.0499* (0.0234) 0.0387* (0.0237)
PctInvert 0.0235 (0.0691) 0.0362 (0.0690)
Popgrow% 0.0092** (0.0029) 0.0090*** (0.0029)
Senatefws%
Housefws%
Presvp% -0.0002 (0.0002)
ISHAC% 0.0104 (0.0070)
LCV -0.0003** (0.0001) -0.0003*** (0.0001)
PctHuntfish
PctForestland
PctFarmland -0.0002** (0.0001) -0.0003*** (0.0001)
[R.sup.2] 0.9485 0.9495
N 49 49
Likelihood statistic 151.463 152.006
Spatial lag LR test stat. 14.64*** 17.52***

***Statistically significant at 0.01 level; **Statistically significant
at 0.05 level; *Statistically significant at 0.10 level.

TABLE 3 ESA Listings Regression Estimation Results--49 States

 Estimated Coefficients (standard errors)
Explanatory Variable Model 1 Model 2

W_Atrisk 0.3516*** (0.1123) 0.3385*** (0.116)
PctEndemic 0.2037*** (0.0159) 0.2039*** (0.0171)
PctTerrestrial 0.0221 (0.0185) 0.0181 (0.019)
PctAquatic 0.1105*** (0.0263) 0.0932*** (0.0255)
PctPlant 0.0008 (0.0060) 0.0011 (0.0062)
PctInvert -0.0083 (0.0155) -0.0050 (0.0165)
Popgrow% -0.0004 (0.0006) -0.0004 (0.0006)
Senatefws% 3.98E-5 (2.16E-5)
Housefws% -2.89E-8 (1.27E-5)
Presvp%
ISHAC%
LCV -5.24E-5* (3.12E-5) -3.83E-5 (3.19E-5)
PctHuntfish -0.0159** (0.0067) -0.0121* (0.0066)
PctForestland 4.98E-5** (1.96E-5) 5.0E-5** (2.11E-5)
PctFarmland
[R.sup.2] 0.8738 0.8655
N 49 49
Likelihood statistic 224.000 222.370
Spatial lag statistic 8.265*** 7.175***

 Estimated Coefficients (standard errors)
Explanatory Variable Model 3 Model 4

W_Atrisk 0.4444*** (0.1108) 0.4108*** (0.1119)
PctEndemic 0.2036*** (0.0195) 0.2080*** (0.0167)
PctTerrestrial 0.0139 (0.0206) 0.0183 (0.0211)
PctAquatic 0.0959*** (0.0268) 0.1029*** (0.0254)
PctPlant -0.0011 (0.0054) -0.0027 (0.0056)
PctInvert -0.0015 (0.0167) 0.0006 (0.0056)
Popgrow% -0.0007 (0.0007) -0.0005 (0.0007)
Senatefws%
Housefws%
Presvp% 2.19E-5 (5.03E-5)
ISHAC% 0.0017 (0.0017)
LCV -7.53E-6 (2.8E-5) -8.91E-6 (2.78E-5)
PctHuntfish
PctForestland
PctFarmland -3.91E-5* (2.03E-5) -4.22E-5** (2.03E-5)
[R.sup.2] 0.8491 0.8531
N 49 49
Likelihood statistic 220.197 220.621
Spatial lag statistic 11.053*** 10.621***

***Statistically significant at 0.01 level; **Statistically significant
at 0.05 level; *Statistically significant at 0.10 level.
COPYRIGHT 2006 Western Economic Association International
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2006 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Title Annotation:Endangered Species Act
Author:Laband, David N.; Nieswiadomy, Michael
Publication:Contemporary Economic Policy
Geographic Code:1USA
Date:Jan 1, 2006
Words:7171
Previous Article:Information, wildlife valuation, conservation: experiments and policy.
Next Article:State adoption of environmental audit initiatives.
Topics:

Terms of use | Privacy policy | Copyright © 2022 Farlex, Inc. | Feedback | For webmasters |