III. Soil Characteristics

A. Background

As noted in Section II.D Geology, soils often form from the weathering and breakdown of the underlying rock or "parent material." Chemical and physical processes break down the rocks and form minerals that are characteristic of the parent material. Therefore, the natural soil on Vieques is a direct product of the island's bedrock, which is mostly granodiorite, quartz diorite, some volcanic lavas, and marine sedimentary deposits such as limestone.

What are the chemical characteristics of soils of the different geologic units on Vieques?

To compare and contrast the soils of Vieques, the soil samples were differentiated based on the underlying parent material (i.e., the rock formations or geologic units) on which they developed. ATSDR compared the constituent metals detected in the soil that developed on one geologic unit to the metals in the soil in the other geologic units. Comparisons were considered statistically significant (i.e., different) if there was less than 5% probability that the difference occurred by chance (i.e., p <0.05). Figure 3 illustrates the location of the generalized geologic units on Vieques used for this phase of the analysis (modified from Torres-Gonzalez 1989).

• Undivided marine sedimentary rocks comprise most of the LIA, a small area near Mosquito Pier, and along the southern shore from Ensenada Sombre to Bahia Corcho. These rocks--largely soft limestone--are shown as the Tl geologic unit on Figure 3. The soil developed on the Tl geologic units have significantly higher concentrations of boron (Tl mean X = 15.8 ppm, sample size (n) = 19; non-Tl X = 11.3 ppm, n = 397) and calcium (Tl X = 153,000 ppm, n = 19; non-Tl X = 20,100 ppm, n = 401) than soil of other non-Tl geologic units.
• Most of AFWTF, the EMA, and the northeast corner of the residential area are comprised of a complex assemblage of deeply weathered sandstone, siltstone, conglomerate, lava, tuffs, and breccia, largely of marine origin. These rocks are shown in Figure 3 as the Kv geologic unit. The soil developed on the Kv geologic units is significantly different (higher) than the soil of other non-Kv geologic units for boron, chromium, cobalt, iron, lead, magnesium, manganese, nickel, scandium, titanium, vanadium, yttrium, zinc, and zirconium. See the Exhibit 1 for the differences between the Kv and non-Kv geological units.

• The bedrock of Vieques is plutonic rocks (deep-seated, igneous intrusive rock), largely granodiorite and quartz diorite, abundant throughout the former NASD, the residential section, and central sections of the EMA. These rocks are shown in Figure 3 as the KTd geologic unit. The soil developed on the KTd geologic units have significantly higher concentrations of barium (KTd X = 917 ppm, n = 212; non-KTd X = 328 ppm, n = 241), strontium (KTd X = 295 ppm, n = 212; non-KTd X = 199 ppm, n = 204), and tin (KTd X = 14.2 ppm, n = 212; non-KTd X = 11.8 ppm, n = 237) than soil of other non-KTd geologic units.
• Beach and dune deposits, which are comprised of sand-sized fragments of calcite, quartz and volcanic rocks, with local concentrations of magnetite grains, are found in small pockets near the southern shore in the former NASD, near the mouth of Puerto Mosquito, and in the AFWTF. These deposits are shown in Figure 3 as the Qb geologic unit. The soil developed on the Qb geologic units have significantly higher concentrations of calcium (Qb X = 74,300 ppm, n = 3; non-Qb X = 25,700 ppm, n = 417) than soil of other non-Qb geologic units.
• Sand, silt, clay, gravel flood plain, terrace deposits, and piedmont fan deposits make up the alluvial deposits located along the north and south coasts of the former NASD, the western half of the south coast of the residential section, and in pockets along the coast of the EMA. These deposits are shown in Figure 3 as the Qa geologic unit. No significant differences appeared between the soil developed on the Qa geologic units and those found on non-Qa geologic units.
• Swamp and marsh deposits, mostly organic muck and peat, are only present in the northwest corner of the former NASD. These deposits are shown in Figure 3 as the Qs geologic unit. Because only one sample was taken from this unit, a statistical comparison could not be completed for the soil of the Qs geologic units.

Using the data collected by USGS and PRDNR in 1972 (Learned et al. 1973), ATSDR compared the chemical characteristics of the KTd geologic unit to the Tl geologic unit. This comparison is necessary for establishing the inherent chemical differences between these two units and will be useful in the discussion of the second approach ATSDR used to determine whether the soils of the LIA contain elevated levels of heavy metals (Section III.C). As shown in Exhibit 2, several metals were statistically significantly different.

B. Comparison of Vieques to the Mainland of Puerto Rico and the United States

Can the soil of Vieques be compared to the mainland of Puerto Rico?

Based on available information, only a generalized comparison can be made. While soil data are available for Vieques, the only area-wide, chemical characterization data available for the mainland of Puerto Rico are stream sediment sampling collected by USGS and PRDNR (Marsh 1992). The stream sediment samples were collected from active stream channels in drainage basins ranging in size from less than 1 square kilometer to less than 10 square kilometers. The objective of this geochemical sampling program was to characterize the distribution of commonly occurring elements and metals throughout the mainland of Puerto Rico.

Sampled stream sediments are representative of the soil and their parent material (e.g., the geologic substrates of that drainage basin) as well as other factors (e.g., human activity, etc.) that could have affected that drainage basin. The stream sediments are comprised of granular material derived from erosion, stream transportation, and deposition of the soil and drainage-related geologic deposits. The physical and chemical process responsible for the erosion, transportation, and deposition of those sediments could introduce several changes to the overall composition of the sediments as compared to the soil from which they were derived.

Ideally, ATSDR would compare the soil on Vieques to soil on the Puerto Rico mainland. But because comparable area-wide data for the mainland are not available, ATSDR determined that comparing the soil on Vieques to the sediment on the mainland of Puerto Rico would still serve to point out generalized similarities or differences between the two areas. Given the inherent differences between the two media, the following can be noted: the soil of Vieques has higher concentrations of antimony, arsenic, cadmium, calcium, manganese, molybdenum, silver, strontium, and yttrium than the stream sediment samples collected throughout the mainland (see Table 2).

However, when ATSDR compared the background sediment samples collected in the former NASD (CH2MHILL 2001) to the sediment data collected on the mainland of Puerto Rico (Marsh 1992), the levels of metals from the former NASD are lower or at the lower end of the range than those detected on the mainland, with the exception of calcium and sodium (see Table 3). Note that a small number of background sediment samples at the former NASD (n = 6) are being compared to a much larger sample set from the mainland of Puerto Rico (n = 2,852). In addition, the samples were collected and analyzed using differing procedures and methods.

How does the soil on Vieques compare to the average soil concentrations in the United States?

Vieques soil is generally comparable to the soil across the United States, especially areas underlain by igneous or volcanic rocks. A statistical comparison between Vieques and the United States was not conducted. Instead, this information is referenced to provide additional insight into the metals and other chemical concentrations detected on Vieques. The United States data represent background soil concentrations taken from uncontaminated areas across the conterminous (i.e., contiguous) United States (Shacklette and Boerngen 1984). The average concentrations for several of the metals were higher on Vieques than the average on the United States. Nevertheless, only the maximum concentrations of copper, iron, lead, tin, and zinc detected on Vieques were outside the ranges found throughout the United States (see Table 2).

As noted in Section II.H Summary of the Available Soil Sampling on Vieques, the soil samples collected by USGS and PRDNR in 1972 (Learned et al. 1973) may be as much as 4-fold higher than the true values. Therefore, any apparently "elevated" levels in comparison to background levels in the United States may be a result of this high-bias in the USGS/PRDNR data (the only island-wide soil sampling available for Vieques), rather than being truly elevated.

Regardless, as noted in Section II.D Geology, soils contain chemicals that are characteristic of the underlying rock, and the metal concentrations observed in the soil of Vieques are consistent with those observed in areas of relatively low-level mineralization (natural mineral enrichment through a variety of geologic processes) in igneous or volcanic rock.

C. Comparison of the LIA to the Remainder of Vieques

Do the soils of the LIA contain elevated levels of heavy metals?

ATSDR attempted to answer this question using two different approaches: (1) by comparing concentrations of chemicals detected at the LIA to the remainder of Vieques and (2) by comparing concentrations of chemicals detected at the LIA to background soil samples in the NASD. Both approaches yielded useful but slightly differing conclusions and each may be based upon data (high-biased) or assumptions (correlations of geologic/soil types) that may result in conclusions that are not entirely accurate.

• The first approach was to statistically compare the levels of chemicals detected in the LIA to the remainder of Vieques (see Table 4). As shown in Table 4, the soil on the LIA had significantly higher concentrations of boron and calcium than the soil from the remainder of Vieques when the soil on the LIA was compared to all soil types on Vieques.

  The LIA, however, is comprised of two geologic units--undivided marine sedimentary rocks (Tl) and a deeply weathered assemblage of largely marine sandstone, siltstone, conglomerate, lava, tuffs, and breccia (Kv) (Torres-Gonzalez 1989). As noted in Section III.A Background, chemical concentrations in soils that develop on different geologic units are inherently different. Therefore, ATSDR also compared the soil on the Tl geologic units on the LIA to soil on the Tl geologic units on the remainder of Vieques, and found that only cobalt was significantly higher on the LIA (see Table 5). In a comparison of the soil on the Kv geologic units on the LIA to the soil on the Kv geologic units on the remainder of Vieques, only calcium was significantly higher on the LIA (see Table 6).

  It should be noted that statistical comparisons could not be conducted for metals that were not detected across the entire island. In fact, antimony, arsenic, cadmium, and mercury were not detected anywhere but on the LIA. This could be because the earlier sampling conducted on the rest of Vieques (Learned et al. 1973) was not sufficiently sensitive to detect the low levels of these metals. Arsenic, for example, was detected on the LIA during more recent sampling, but was not detected during the geologic evaluation in 1972 because the detection limits were not low enough.

  This approach is based largely on the data collected by USGS and PRDNR in 1972 (Learned et al. 1973) and as noted in Section II.H Summary of the Available Soil Sampling on Vieques, these soil samples may be as much as 4-fold higher than the true values. However, regardless of the uncertainty regarding the values reported in these data, it is reasonable to expect that there is internal consistency of analytical results within this data set (i.e., values from the LIA and from the remainder of Vieques are both high-biased) and thus, ATSDR expects little impact from using these data.

• The second approach was to compare the levels of metals detected at the LIA during the 2000 soil sampling (CH2MHILL 2000a) to the background soil samples gathered in the former NASD (CH2MHILL 2001). This approach eliminates any inherent differences based on analytic and sampling methods. However, the soil samples were collected on different geologic units; therefore, in an effort to determine if the background levels for the KTd could be useful in evaluating potential contamination of the LIA by Navy training exercises, ATSDR compared the chemical make-up of the soils of the KTd with that of the soils of the Tl (see Section III.A Background). Comparing and contrasting only the soils of those two units reveals that, with the exception of barium, cobalt, iron, manganese, strontium, titanium, and vanadium that are statistically significantly different, the metals composition of the soils of the KTd and the Tl are otherwise similar. Given the qualified similarity of the soils of these two geologic units, the background values established for the KTd in the former NASD may be generally applicable as background values for the Tl in the LIA.

That established, ATSDR compared the soil samples collected from the Tl geologic unit in the LIA to the soil samples collected from the KTd geologic unit in the former NASD (see Table 7). This comparison revealed that the average concentrations for most of the chemicals in the soil from the LIA are elevated above background by 1.4 to 2.9-fold. The average arsenic concentration, however, seems to be roughly 14 times higher in the LIA than in the former NASD.

Overall, it appears that the soils of the LIA have been influenced by Navy training activities and contain elevated levels of heavy metals. Given the potential problems associated with the data and assumptions used, additional background soil sampling in the LIA would be necessary to confirm the validity of this conclusion. Although, ATSDR realizes collecting background samples from the LIA may no longer be possible. The next section, Section IV. Evaluation of The Soil Exposure Pathway, describes the methods ATSDR used to determine that despite the apparent elevation in chemical concentrations at the LIA, residents and visitors of Vieques are not being exposed to harmful levels of chemicals in the soil.

Has the LIA become more contaminated with time?

ATSDR attempted to compare soil samples collected from the LIA in 1972 (Learned et al. 1973) to soil samples collected from the LIA in 2000 (CH2MHILL 2000a), to determine if this 28-year interval of Navy training had significantly increased the level of contamination at the LIA. However, given that the data collected by USGS and PRDNR in 1972 appears to be elevated by as much as 4-fold higher than the true values (as described in Section II.H), this comparison became invalid.

D. Movement of Contamination from the LIA to the Residential Area

Is there a spatial pattern that indicates that metals are moving from the LIA to the residential area?

No. The available data do not indicate a pattern of high to low concentrations from east to west. Thus, this data set does not provide evidence indicating airborne transport of metals from the LIA to the residential area. Further analysis of this issue is being conducted by ATSDR using computer air transport models and will be presented in a separate air pathway evaluation (see Sections I and VIII).

To answer this question, ATSDR plotted on a map the locations of metal concentrations detected on Vieques. All of the studies, except the 1978 investigations (Hoffsommer and Glover 1978 and Lai 1978), identified their sampling locations either by latitude and longitude or on a figure. But only the reconnaissance geochemical survey conducted by USGS and PRDNR (Learned et al. 1973) collected samples from across the entire island. Therefore, most of the interpretations from this analysis are based on these data. As noted previously, it is reasonable to assume that there is internal consistency within the 1973 data and thus, those data are suitable for use in an evaluation of the relative spatial distribution of detected levels.

ATSDR generated chemical-specific maps for the metals that are found in munitions. None of the spatial maps showed a pattern beginning with high concentrations in the LIA and decreasing concentrations tapering off to the west of the island. Spatial maps were not generated for explosive compounds because the only detections were located at the LIA. Sampling for explosives was, however, conducted along the western border of the EMA in 1999, and none were detected (CH2MHILL and Baker 1999).

Although a spatial pattern describing a progressive east to west trend is not shown in the chemical-specific maps, other patterns are shown. For example, a western concentration of the highest levels of strontium detected is shown in Figure 6. Barium shows a similar western concentration and both chemicals seem to show an association with the areas underlain by granodiorite and quartz diorite (the KTd geologic unit).

A geologic association between the areas underlain by the undivided sedimentary rocks (the Tl geologic unit) and the occurrence of the highest detections of calcium is illustrated in Figure 7. Both vanadium and zinc show a geologic association with areas underlain by the marine sandstones and lavas (the Kv geologic unit), although the association is not a clearly developed as the calcium-Tl association.

Examination of the distribution of detected levels of the other metals does not show any clear pattern, trend, or association. This random distribution of detected levels is illustrated in Figure 8 for chromium. A similar random distribution is also exhibited by the spatial distribution of the detected levels of cobalt, copper, iron, magnesium, manganese, nickel, lead, titanium, yttrium, and zirconium.

---

1. When ATSDR compared the metal concentrations in soil samples collected from the former NASD during this effort to background (naturally occurring) soil samples recently collected from the former NASD (CH2MHILL 2001), the data collected by USGS and PRDNR was found to be higher, sometimes by 4-fold. Barium and calcium differ from this pattern for unknown reasons. Since sampling locations from both events were from areas unaffected by the Navy training activities, it would appear that the older data is artificially elevated (i.e., higher than the true values). Another indication is that for those metals analyzed and detected in all samples, the minimum values obtained during the USGS/PRDNR study are higher that the minimum values obtained during the recent background study, even though the minimum detection levels were higher than those used when the background analyses were performed. If there were no high-bias in the data reported in 1973 then several of the samples collected should have been reported as not detected. These differences are probably due, in part, to differing analytic or sampling methods. However, regardless of the reason(s) for the apparent differences, the higher values were used by ATSDR to evaluate potential health effects.
2. Please see the discussion "Can the soil of Vieques be compared to the mainland of Puerto Rico?" in Section III.B for an explanation of ATSDR's use of sediment data.
3. The laboratory that analyzed the samples noted that "a completely positive identification was not possible due to the extremely low concentrations found" (Hoffsommer and Glover 1978). They further note "if these explosives are present, the concentrations do not exceed the values reported here." ATSDR evaluated the chemical concentrations as reported to be most protective of human health; however, this should not be interpreted as validation of the laboratory study's results.
4. ATSDR utilized the available data with certain limitations. In a statistical comparison of this data to the other data collected on the LIA (Learned et al. 1973; CH2MHILL 2000a), nine of the 20 chemicals were significantly different (higher, p <0.05): ammonia, barium, cadmium, copper, lead, mercury, nickel, nitrate/nitrite, and zinc. This does not imply anything is amiss with the data; it simply means that using only the highest and second highest detections rather than the complete data set would skew any results. Therefore, these data could not be used in the statistical analyses. They were, however, used to form conclusions during the public health evaluations.

---