Warning This Web page has been archived on the Web.

Archived Content

Information identified as archived on the Web is for reference, research or recordkeeping purposes. It has not been altered or updated after the date of archiving. Web pages that are archived on the Web are not subject to the Government of Canada Web Standards. As per the Communications Policy of the Government of Canada, you can request alternate formats on the Contact Us page.

Help the Government of Canada organize its website!

Complete an anonymous 5-minute questionnaire. Start now.

Skip booklet index and go to page content

National Assessment of Pulp and Paper Environmental Effects Monitoring Data: Findings from Cycles 1 through 3

5.2 Summary of Effect Sizes

The key fish comparisons summarized in this section focus on the endpoints for which Environment Canada has developed CES -- gonad weight, liver weight and condition factor. The range and distribution of measured exposure versus reference area percent differences in Cycle 3 were quite similar to those in Cycle 2 (Fig. 2). For Figures 2 and 3, measured differences were calculated as exposure area minus reference area adjusted means, expressed as a percentage of the reference area adjusted mean. Figure 2 focuses on comparisons where exposure versus reference ANCOVA slopes were parallel (the majority of comparisons; see Fig. 3 and Environment Canada 2004a). Note also that Figure 2 provides a complete summary of the measured differences (i.e., includes both statistically significant and non-significant differences). Similar to Cycle 2, Cycle 3 condition factor showed the narrowest range in exposure versus reference area percent differences (-15% to 25%); this comparatively narrow range led to the development of a smaller CES for condition than for gonad and liver weights. Cycle 3 gonad weight percent differences ranged from approximately -70% to 70%, while liver weight differences ranged from -55% to 120%, again with a highly similar data distribution to that observed in Cycle 2. Thus, at a national level, we observed a high reproducibility between Cycles 2 and 3 in magnitude and distribution of measured differences. This is likely related to the consistency of response patterns from one cycle to the next that is discussed in section 4.3.

Distribution of measured percent differences between exposure and reference area fish in Cycles 2 and 3 for: a) condition factor, b) gonad weight and c) liver weight.

Figure 2: Distribution of measured percent differences between exposure and reference area fish in Cycles 2 and 3 for: a) condition factor, b) gonad weight and c) liver weight.

Figure 3 shows the number of comparisons that showed no statistically significant effect versus a significant effect for condition, gonad weight and liver weight. The significant effects are further divided into three categories. The first of these is a significant interaction. This occurs when the exposure versus reference area slopes are statistically different in the ANCOVA analysis; that is, when the slopes can be considered to be non-parallel. In this case, the effect of effluent exposure is to, for example, affect allocation to gonad weight differently for fish of different sizes, relative to fish in reference areas, resulting in non-parallel exposure versus reference slopes for ANCOVA regressions of gonad weight against body weight. The last two significant effect categories in Figure 3 occur only when the exposure and reference area slopes are not significantly different (no significant interaction, so effect of effluent exposure is approximately the same for fish of different sizes). The first of these last two categories consists of comparisons where the exposure versus reference area ANCOVA adjusted means were significantly different, but with an effect magnitude smaller than the CES. The second of these last two categories consists of comparisons where the exposure versus reference area adjusted means were significantly different, with an effect magnitude that exceeded the CES. Note that CES have not yet been developed for the case where ANCOVA slopes are not parallel. See Environment Canada (2004a) for further information on ANCOVA procedures and interpretation.

Summing the comparisons illustrated in Figure 3 reveals that between 40% and 60% of the comparisons for each endpoint were significant. Furthermore, 20–25 of the significant comparisons for each endpoint exceeded the CES. Thus, the number of effects that exceed the CES is relatively consistent among endpoints.

Number of exposure versus reference fish comparisons showing no significant difference a significant interaction, a significant difference in adjusted means less than CES and a significant difference in adjusted means greater than CES for a) condition factor, b) gonad weight and c) liver weight.

Figure 3: Number of exposure versus reference fish comparisons showing no significant difference a significant interaction, a significant difference in adjusted means less than CES and a significant difference in adjusted means greater than CES for a) condition factor, b) gonad weight and c) liver weight.

For fish studies, a mill will conduct focused monitoring (including magnitude and geographic extent) if any CES is exceeded, with statistical significance, in the same direction for the same sex and species for two consecutive cycles. There was this kind of exceedance of CES in Cycles 2 and 3 at approximately 15% of the mills conducting a fish survey. Of these latter mills showing consistently large effects in both cycles, approximately three-quarters exceeded the CES for condition factor, one-half for gonad weight and one-quarter for liver weight. Half of these mills exceeded the CES for two endpoints, while the other half exceeded the CES for only one endpoint. For mills expected to conduct focused monitoring, condition and liver size were usually greater in the exposure areas (a sign of nutrient enrichment), while gonad size was usually smaller in the exposure areas (a sign of disruption of resource allocation to the gonads; see section 4.3).