Liver tissue and stomach contents--The concentration of each element in liver and stomach contents was plotted vs. total organic carbon content of the sediment in order to highlight those sites that contained fish with elevated tissue concentrations (Figs. 55a-k). Total organic carbon was chosen for the abscissa because it is one physical-chemical factor which may control the bioavailability of metals to individual fish.

Few correlations were found between mean liver and stomach contents concentration (Table 9), possibly due to greater temporal variation in the concentration of elements in stomach contents than that which may occur for liver. However, Figures 55 a-k show that at some sites, high liver concentration (upper figure) corresponds to high stomach contents concentration (lower figure). For several elements a site producing high concentrations in fish liver can also be found to have fish with that element at high concentration in stomach contents. Unfortunately, there were only a few samples for stomach contents and not every stomach contents concentration had a corresponding liver concentration, although we present data for most species and sites (Table 1 shows which species and sites have stomach contents data). These figures also help to explain the contradictory occurrence of high element concentrations in reference site (Bodega Bay, Dana Point, and Nisqually Reach) fish livers. It appears that the prey items in the stomachs of some fish can have high concentrations of certain elements which may in turn produce high liver concentrations. This is further supported by the high lability for some elements at reference sites as seen in the extractable-metals study (Fig. 49a-k).

Another interesting pattern is that some of the highest concentrations for the elements antimony, cadmium, mercury, and silver in liver occur at sites with the lowest TOC levels (0 - 0.5%). This leads to the hypothesis that elements at sites with lower sediment TOC may have greater bioavailability to fish than elements at sites with high TOC. No such pattern occurs for the metals copper, lead, tin, and zinc. For the essential metals copper, selenium, and zinc, regulation by the animals may occur which would mask any such uptake patterns. For the nonessential metals lead and tin, it may be that the levels are not sufficiently high to produce such a pattern of high liver concentration at low TOC content. Additionally, there may be relatively few sites that are contaminated with these metals, or the bioavailability of these metals is very low due to strong complexation by the sediment. It should be kept in mind that essential elements such as copper, selenium, and zinc are regulated by fish and there may be a fine line between normal and toxic concentrations.

For many of the elements (antimony, arsenic, cadmium, copper, lead, mercury, nickel, selenium, silver, and zinc) it is obvious that some of the highest liver concentrations are found at the reference sites in Washington and California. This is generally not true for elements in flathead sole liver from Alaska when the reference site (Lutak Inlet) is compared to the urban sites. In general, the stomach contents concentrations, which is one measure of exposure, supports the high liver concentrations. For the elements antimony, arsenic, cadmium, lead, mercury, nickel, selenium, and silver, high liver concentrations have a correspondingly high stomach concentration at some of the West Coast reference sites. For example, silver (Fig. 55i) was very high in the liver of many fish from Dana Point and Bodega Bay. Although we had only a few measurements of stomach contents from these sites, we did find very high levels in one Dana Point fish, indicating at least the possibility that high concentrations can exist in the stomach contents. This would imply that the prey items, which were generally sediment dwelling invertebrates, were accumulating high concentrations of these elements that may then pass to the fish predators. It should also be kept in mind that while the mean concentration for most elements in reference sediments was low, there were often a few stations from these sites which contained high concentrations. While these results are not conclusive, the data allow us to see the possibilities and formulate hypotheses for future work.

This pattern of increased liver concentration at low TOC sites may explain the observation that for some elements, fish livers from reference sites are often as high or higher than some urban sites. Most of our reference sites are characterized by low TOC content (Fig. 45). Any increase above natural background geochemical concentrations at these low TOC sites may allow an element to be more available on a unit weight basis compared to those sites with finer sediments and higher TOC content. This may be highly relevant for the sediment-dwelling invertebrates which selectively feed on the organic-carbon rich fine fraction of these coarse-grained sediments. This is a simplistic observation which ignores many controlling ligands for metals, such as acid volatile sulfides (AVS), oxides of manganese and iron, and other inorganic phases. It is possible that TOC correlates to one or several of these other ligands and may explain the observed pattern without being the causal factor. Acid volatile sulfides may have been a better variable to use for the abscissa, but we have no measurements for this parameter. Total organic carbon may be correlated to AVS content because sediments high in TOC are often anoxic, and AVS is associated with only anoxic sediments. The reference sites generally contain coarse sediment particles (sand) which contain little TOC and are hence probably fully oxic. Sediments with low levels of AVS and TOC, in addition to coarse particles which contain low levels of iron and manganese oxides, may produce higher bioavailability on a per gram basis when compared to finer-grained sediment.

Significant correlations were observed between sediment and stomach contents for several elements when all species were considered as a group (Table 9). When individual species or species-groups were considered, some of these associations became stronger, suggesting that these species were more likely than others to take up and accumulate elements from dietary sources. For example, mercury in all species correlated at r = 0.57, but when only English sole was considered, the correlation was even stronger (r = 0.92) (however the sample size was small: n = 6 site-year occurrences). Similar patterns can be seen for other elements and a few for stomach contents and sediment are shown in Figs. 56a-c. The general pattern of mercury in all species (Fig. 56a) shows a moderate but significant trend (r = 0.57) with several of the fish displaying high stomach contents concentrations at those sites with high sediment concentrations. Chromium (Fig. 56b) displayed a highly significant association (r = 0.84) which appears to indicate that prey of sand bass and white croakers contain high levels of this metal. This pattern was also strong in English sole and white croaker (Table 9) but weak in all other species. Lead in sand bass and croakers (Fig. 56c) showed a similar trend (r = 0.65) between the concentration in stomach contents and that found in sediment.

The correlation of sediment and stomach contents is always suspect because of the possibility that sediment may contaminate the sample and lead to erroneous conclusions about the source. We visually inspected our samples and determined that they were essentially free of sediment; hence, we believe that the dietary source to fish was mainly animal prey and not sediment. If sediment had contaminated our samples, we believe that many elements, not just a few, would have shown high correlation coefficients. Additionally, for elements such as mercury, the stomach contents concentration is often the same or higher than the sediment concentration which means that large quantities of sediment would be required to contaminate a sample in order to achieve these observed concentrations. Although small amounts of sediment in the samples may have eluded our detection, their presence would be acceptable because sediment is a possible dietary source of elements for these species and its contribution would be important.

Correlations of elements in liver tissue, stomach contents, and sediment--Correlation analysis of the sediment and liver tissue data was performed in order to determine if any associations existed. Our main hypothesis was that high concentrations of elements in sediment would be reflected in the concentrations measured in the liver tissue of fish from the different sites.

We examined the association between elements in fish liver and sediment with all available data. The results presented in Table 10 are the correlations of all mean log₁₀ concentrations of elements in liver and sediment for each site over years. Each sediment and liver concentration was matched by location (site), species, and year. Some of the correlations in Table 10 were deemed spurious because a few points determined the high correlation coefficient which was not representative of the entire sample. Two of these correlations are shown in Figures 57a and b. We also tried correlating liver concentration with sediment concentration that had been normalized to organic carbon and found no improvement in describing relationships.

A few elements showed a consistent pattern over species. Mercury in liver and sediment was significantly correlated for all species which appears to be primarily due to the flatfish (English sole, flathead sole, hornyhead turbot, and starry flounder (minus two individual outliers) (Fig. 58a). This relationship can be improved when certain species combinations are chosen. In Figure 58b, the correlation of mercury in liver and sediment was 0.67 when we plotted species from predominantly low mercury sites (all flathead sole) and species from sites which were predominantly high in mercury (all white croaker). We were not able to assess the prey composition of each species; hence, we cannot derive any conclusions about dietary input to elucidate this particular association. Another example is zinc in flatfish which appears to have inversely correlated liver and sediment concentrations (Fig. 58c).

One reason for the lack of correlation for most elements may be due to the high variability between tissue samples from a given species. It is not uncommon to find that the concentration of an element in three independent tissue or sediment samples from a site vary by a factor of 2 to 5. Even for one species at a given site in a given year, the coefficient of variation can exceed 50-100% or more. For example, in 1986 English sole from Bodega Bay (a reference site) exhibited a mean and standard deviation (sd) concentration in liver of silver, mercury, and cadmium of 1.3 (1.2), 1.0 (0.6) and 2.3 (1.8) µg/g, respectively. An example of the variation for both liver and sediment concentrations at various sites for one element and species can be seen in Figure 59.

Because of such large variability, attempting to correlate sediment and tissue concentrations is usually futile. Due to the fact that concentrations of elements in sediments are patchy and localized and because fish are usually very mobile, we would expect such variation to be normal. In the future, we recommend analysis of infaunal invertebrates which are generally much less mobile than fish and may reflect the elevated levels of metals and elements found in urban areas of our coastline. These data, in conjunction with concentrations in fish and stomach contents, may help to pinpoint areas of contamination. Assessment of infaunal invertebrates and their bioaccumulation of toxic elements will help us to determine food-chain transfer from prey organisms to fish and possible human consumers.

We also explored the concentrations of elements in liver, stomach contents, and sediment from all available data. The results are presented in GT2 plots (Figs. 60 a-l). The data are log₁₀ means of all liver, stomach contents, and sediment concentrations for a species at all sites where it occurs, over all years collected. Each bar in the GT2 plot is the mean concentration of that element in its respective compartment (e.g., BSLV is barred sand bass liver) and the vertical whiskers are the comparison interval. (In the floating bar-plots of Figures 3 through 43, the box is the standard deviation.) Nonoverlapping whiskers indicate significantly different means at the alpha = 0.05 level. Numbers above the bars are sample sizes. The concentration of elements from each sediment station (A, B, and C) and individual fish analyzed at a given site were used to construct the GT2 plots. With these GT2 plots, comparisons can be made between like or dissimilar compartments within a figure (e.g., liver between different species or stomach contents vs. liver for one species). Unfortunately, there was no one-to-one correspondence between liver and stomach contents samples (e.g., stomach contents were composited and livers were from individuals). Also, because we have very little information regarding the nature of the stomach contents from these samples, we cannot make any inferences about trophic transfer. Some species (e.g., English sole) feed predominantly on invertebrates and others ingest fish. Some species (e.g., white croaker) will take both kinds of prey which may lead to highly variable stomach contents concentrations. Another cause of variability is the mobility of the fish which allows them to feed over a large area that may have a patchy distribution of sediment contaminants. On a temporal scale, stomach contents concentrations can be much more variable than liver concentrations because fish feed over a wide area with known chemical heterogeneity, whereas the liver integrates over the whole exposure area.

In general, by looking at the concentrations in all individuals of a species we are ignoring some geographical patterns, including urban vs. reference sites, but we are increasing the sample size over that seen in Figures 9 through 43 which gives us more power to discriminate patterns. Also, by looking at a species, we are integrating some geographical patterns; for example, all flathead sole samples were from Alaska and barred sand bass were generally from southern California (San Diego).

When individual elements are considered, some interesting trends can be seen. One clear trend for many elements (e.g., antimony, chromium, lead, nickel) is that the liver concentration is lowest of the three compartments, followed by stomach contents, and then sediment concentration which is the highest of the three. Other elements (copper and zinc) show basically no pattern, although a few (e.g., copper in starry flounder and zinc in hornyhead turbot and starry flounder) show a trend of increasing concentration from sediment to stomach contents to liver. Still other elements (arsenic, cadmium, mercury, selenium, and silver) were quite low in sediment and displayed an increasing trend in tissue (stomach contents to liver), indicating bioaccumulation.

Antimony. When all samples for antimony are considered, white croaker and barred sand bass liver were generally lower (not significantly) than the other species (Fig. 60a) and sediment was generally uniform. However, white croaker liver was significantly lower in concentration than English sole, but only slightly. None of the stomach contents concentrations were significantly different when compared among species, and there was a strong tendency for antimony to decrease from sediment to liver.

Arsenic. Arsenic, a nonessential element, exhibited few patterns (Fig. 60b). The most noteworthy is that flathead sole had significantly higher concentrations of arsenic in their liver than all other species except English sole, even though the sediment concentration associated with flathead sole was not particularly high. Interestingly, all flathead sole samples were from Alaska, which is noted for high arsenic levels in fish (Fig. 14) although sediment concentrations were not elevated (Fig. 3). There seems to be a trend of increasing concentration (significantly higher) from sediment to liver for flathead sole and English sole for arsenic which may indicate biomagnification, (assuming two or more trophic levels are involved).

Cadmium. The data for cadmium are highly variable and few patterns can be discerned (Fig. 60c). When means are examined, it appears that there was a general increase of concentration from sediment to liver for all species. This pattern is strongest for English sole, starry flounder, and white croaker, and there was a significant difference between liver and sediment concentrations for four of the species (English sole, hornyhead turbot, starry flounder, and white croaker) indicating bioaccumulation. When all species are compared, none show significant differences in liver cadmium.

Chromium. The data for chromium are relatively consistent for liver, stomach contents, and sediment concentrations (Fig. 60d). Examination of sediment and liver concentrations shows no difference among species. Because chromium is considered to be an uncommon contaminant, it is not surprising to find such uniformity.

Copper. Copper concentrations are generally uniform but show some pattern of increasing from sediment to tissue in starry flounder and white croaker (Fig. 60e). Also noteworthy is barred sand bass which exhibited significantly reduced copper concentrations in the liver, a possible species specific characteristic. This is supported by the low concentrations in fish from urban and nonurban sites. Interestingly, barred sand bass occurred at sites with the highest sediment copper. This observation is counterintuitive and will be pursued further. White croaker and starry flounder liver also seems to be higher in copper than that found in the other species, especially English sole and flathead sole, which appears to correlate with sediment concentrations.

Lead. Lead was always lower in liver tissue than stomach contents or sediment (Fig. 60f). For most species there was a strong pattern of decreasing concentration from sediment to stomach contents and liver; however, none of the species were significantly different. The concentration in liver appears uniform across all species with the highest variability in flathead sole, which were all from Alaska. Alaskan sites display some of the highest and lowest lead concentrations in both fish liver and sediment concentration (Figs. 4 and 16).

Mercury. The concentration of mercury in liver, stomach contents, and sediment was relatively uniform across sediment samples, but significantly elevated for some sites (Fig. 60g). Mercury was considerably (and significantly) higher in barred sand bass and white croaker liver than in flathead sole, English sole, and starry flounder, which seems to be associated with sites that have high sediment concentration. Barred sand bass and white croaker are species collected from sites (such as Long Beach, San Diego, Hunter's Point) that are considered contaminated with mercury (Table 5). Flathead sole, from the Alaskan sites which contained very little mercury in the sediment, displayed the lowest mercury concentrations in liver. Barred sand bass, which may prey on fish, are found at sites with high mercury concentrations in the sediment and are therefore more likely to bioaccumulate mercury than species which prey predominantly on organisms lower in the food web (e.g., invertebrates). The other species listed above are all flatfish and feed predominantly on invertebrates. It is interesting to note that mercury in fish and invertebrates is predominantly in the organic form (Bloom 1992), which allows it to be bioaccumulated to a greater extent than the inorganic forms.

Nickel. Nickel concentrations were always lower in fish liver and considerably higher in sediment (Fig. 60h). There appears to be no difference among species except that English sole and white croaker liver show slightly higher levels of nickel than the other four species. The sediment concentrations for nickel were not greatly different among sites and the lowest nickel concentrations in liver are associated with the lowest sediment concentrations.

Selenium. There was a very strong gradient of selenium concentration from the sediment to stomach contents and liver (Fig. 60i). There was essentially no difference among sediment concentrations as seen by the comparison intervals except at sites for flathead sole which were significantly higher than sediment found at sites where English sole and barred sand bass were collected. Also, concentrations of selenium in white croaker were significantly higher than those found in starry flounder and hornyhead turbot, although the sediment concentration for sites where these species were collected was not high. Because selenium is an essential element, and the concentrations measured in liver were in the normal range for fish livers, there appears to be no contaminant-related concentrations. However, the range in selenium concentration that determines healthy from toxic may be narrow and, of course, if levels were high enough to be toxic, those individuals may not have been collected.

Silver. This metal was significantly lower in the liver of barred sand bass and almost so for flathead sole when compared to sediment (Fig. 60j). Conversely, when hornyhead turbot and white croaker were examined, silver concentrations in their livers were higher (but not significantly) than that found in sediment. Silver in English sole and starry flounder liver was essentially the same as in sediment. There was high variability in silver concentrations found in fish liver, ranging almost tenfold between barred sand bass and hornyhead turbot. Silver in liver of white croaker and hornyhead turbot was significantly higher than that found in barred sand bass, English sole, flathead sole, and starry flounder. Both flathead sole and English sole had low liver concentrations which corresponded to low sediment concentrations. Conversely, barred sand bass liver silver content was low but the sediment concentration was higher than most other sediment groups.

Tin. There was no pattern among liver or sediment tin concentrations, although English sole and starry flounder displayed a slight elevation in concentration over white croaker and barred sand bass (Fig. 60k). The significantly lower tin concentration in stomach contents of flathead sole from Alaska may reflect lower levels of organotins, which we suspect are very low in invertebrate prey at these sites. Because organotins (particularly TBT) can bioaccumulate in tissue, sites with slight differences in total tin, but large differences in organotins, may contain invertebrates with widely different body burdens of tin.

Zinc. When all sites are considered, it appears that sediment zinc concentrations varied only slightly among sites (Fig. 60l). The same is true for zinc in liver; however, hornyhead turbot were significantly higher than barred sand bass. Because zinc is an essential element and regulated by fish, we are not sure if the differences seen in liver were due to exposure differences or inherent species variability