Jump to main content.

Linking Human Disturbance To Biological Change

Like any scientific inquiry, understanding how human disturbance influences, alters, and degrades biological processes is an iterative process. Hypotheses are proposed on the basis of current understanding; theories are modified in the wake of results; and new insights drive subsequent hypothesis testing. As knowledge accrues, connections between predictions and results tighten, and new studies present fewer surprises. A paradigm forms.

The development of biological monitoring tools has followed this type of scientific process. After 20 years of testing, consistent patterns have emerged (see literature review in Barbour et al., 1999; Karr and Chu, 1999; Karr et al., 2000). The same biological measures tend to correlate with human disturbance in very different geographic settings, e.g., the number of taxa decline, tolerant taxa dominate, and taxa with unique habitat requirements disappear. Yet when results closely match expectations, concerns arise regarding "circular reasoning." An argument based on circular reasoning is one in which the conclusion is embedded in the premise, as for example, in the statement, "decline in mayfly taxa richness is a good indicator of biological disturbance because we find many types of mayflies at undisturbed places." Without a direct causal link, the concern is that metrics may be selected as indicators of human disturbance simply because they are correlated with human disturbance (Suter, 2001). That is, biological indicators may be unrelated to specific biological or societal values.

Because biological systems are complex and human disturbance is multidimensional, single causes and mechanisms of impairment are difficult to isolate; as a result, much of the evidence for human degradation of natural resources is correlative. In such situations, although the path to causality is blocked by the inability to perform controlled experiments and use statistical inference, logical argument (or weight of evidence) constructed according to a recognized set of rules can be used instead (Beyers, 1998). In fact, this approach typically yields a stronger case because researchers consider alternative explanations explicitly, rather than assuming they do not operate. Results from the Mid-Atlantic illustrate how a causal argument can be constructed to support the idea that human disturbance causes biological change.

Addressing concerns about circular reasoning

The data for MAIA were not collected with the intention of demonstrating a causal link between human disturbance and biological degradation; however, the expectation of a cause and effect relationship is implicit in EMAP's survey design and project goals. The process used to test and select metrics did, however, support the type of structured logical argument reviewed by Beyers (1998) for establishing a causal connection between human influence and biological change. In fact, results from the MAIA pilot supported six of Beyers' ten criteria (Chart 2).

Researchers in the field of epidemiology face a similar challenge in defining causality when linking a specific disease with its infectious agent. Rules developed within that field can be applied in an environmental context because the situations are parallel (Beyers, 1998). In epidemiology, patients develop a disease; an investigator cannot randomly infect patients with a variety of infectious agents to see which one causes the disease. Analogously for environmental studies, human development cannot be "applied" at will to see how a place will respond. Furthermore, treatments cannot be replicated in either case. Each patient is unique in terms of medical history and life style as is each watershed unique in terms of its geological formation, hydrological structure, size, and climate (Hurlbert, 1984; Heffner et al., 1996).

Chart 2. Ten criteria for constructing causal arguments (modified after Beyers, 1998).

  1. Strength: a large proportion of sampling units are affected in exposed areas compared with reference areas
  2. Consistency: the association has been observed at other times and places
  3. Specificity: the effect is diagnostic of exposure
  4. Temporality: exposure must precede the effect in time
  5. Dose-response: the intensity of the observed effect is related to the intensity of the exposure
  6. Plausibility: a plausible mechanism links cause and effect
  7. Evidence: a valid experiment provides strong evidence of causation
  8. Analogy: similar stressors cause similar effects
  9. Coherence: the causal hypothesis does not conflict with current knowledge
  10. Exposure: indicators of exposure must be found in affected organisms

Six of Beyers' ten criteria were relevant for constructing a logical argument for the causal connection between human disturbance and biological decline in Mid-Atlantic streams.

Figure 4 shows multimetric index values for fish, invertebrates, and diatoms.

Figure 4. Multimetric index values for fish, invertebrates, and diatoms as a function of types of human disturbance. All index values were higher at reference sites. Diatom index values were lower for sites with high nutrients. Diatom index values were higher than invertebrate index values for sites with acid deposition.

The remaining four criteria (numbers 3, 6, 7 and 8 in Chart 2) require that an observed effect (metric value) be diagnostic of exposure (human disturbance); a plausible mechanism of action exists to link cause and effect; controlled experiments support causation; and analogous responses are associated with similar stressors. A broader survey of the literature could add further examples under these headings to the overall argument for causality. For example, Yoder and DeShon (2002) used metrics to diagnose disturbance type, and Richardson and Kiffney (2000) present experimental evidence for the effect of heavy metals on invertebrates. Nonetheless, the examples above serve to illustrate how the logical construction of an argument for causality represents an alternative to doubts regarding circular reasoning.

Table 1. Spearman's correlation of three multimetric indexes (based on fish, invertebrates, and diatoms) with selected measures of human disturbance. All correlation coefficients were significant; only values > 0.3 (or < -0.3) are shown. Measures of human disturbance were related to (1) nutrients (total nitrogen [N], total phosphorus [P], ammonia [NH4]); (2) acidity (acid neutralizing capacity [ANC] and sulfate [SO4]); (3) sediment (turbidity [Turb], percentage of sand and fine sediments [%S_F], and pebble size corrected for stream power [PbSz]); and (4) channel and riparian condition (riparian vegetation [RVeg], sum of all disturbance types within the riparian area weighted by proximity to the stream [RDist], and average of measures from a rapid habitat protocol [RBP]). Measures of general disturbance included chloride (CL); Bryce et al.'s [1999] disturbance categories derived from watershed and riparian measures of disturbance (Bryce); and the sum of urban, agricultural, and mining land use within the watershed (%Dist).

Measure
Fish Invertebrate Diatom
N
-0.45 -0.32 -0.54
P
    -0.61
NH4
-0.33 -0.36 -0.32
ANC
  -0.33 -0.53
SO4
-0.34    
Turb
  -0.33 -0.39
%S_F
  -0.54 -0.39
PbSz   0.43 0.33
RVeg
    0.30
RDist
  -0.35  
RBP
  0.42 0.36
CL
-0.45 -0.31 -0.55
Bryce
-0.39 -0.57 -0.54
%Dist
 -0.33 -0.40 -0.54
Total 6 11 12

<< previous page
 next page >>

 

Biological Indicators | Aquatic Biodiversity | Statistical Primer

Local Navigation

Jump to main content.