RESULTS AND INTERPRETATION

Results of analyses are presented in Appendix I, and plotted versus depth in Figure 4. Summary statistics for each oxide or element are given in Table 1.

Based on preliminary sediment descriptions (Adam and others, 1989), the main sediment components of the Tule Lake core can be described in terms of a clastic component, CaCO₃, diatoms, and organic matter. Except for biogenic SiO₂ from diatom debris, estimates of these components can be obtained from the geochemical results (Figure 1). Because diatom debris is abundant in some parts of the section, the total SiO₂ values listed in Appendix I and plotted in Figure 1 are composites of nonbiogenic (clastic) SiO₂ and biogenic SiO₂ (diatom debris). In an attempt to obtain an estimate of these two different forms of SiO₂, I assumed, for reasons that will be discussed below, that the clastic fraction had average SiO₂ and Al₂O₃ contents similar to an average for Medicine Lake volcanics that cover much of the drainage basin of Tule Lake. The average SiO₂/Al₂O₃ ratio in 145 XRF analyses of this material is 3.5 (data provided by Julie Donnelly-Nolan, USGS, Menlo Park). I therefore assumed that average clastic sediment entering Tule Lake had this same average ratio, and I multiplied each value of Al₂O₃ by 3.5 to obtain an estimate of niobiogenic SiO₂. The biogenic SiO₂ content was computed by subtracting the nonbiogenic SiO₂ value from total SiO₂ for each sample.

(click for larger image- 31K) Fig. 4a (%CaCO3, %MgO, ppm Sr, %MnO, Org. C, %P2O5)	(click for larger image- 28K) Fig. 4b (%SiO2, %Al2O3, %Fe2O3, %Na2O, %K2O, %TiO2)
(click for larger image- 27K) Fig. 4c (ppm Co, ppm Cr, ppm Cu, ppm Ni, ppm Pb, ppm V, ppm Zn)	(click for larger image- 21K) Fig. 4d (ppm Ba, ppm Ce, ppm La, ppm Sc, ppm Y)
Figure 4. Lithologic summary and plots of concentrations of major-element oxides,CaCO3, and organic carbon (Org. C) in percent, and trace elements (in parts per million, ppm) versus depth in samples from Tule Lake, California.

In order to objectively examine relations among the numerous geochemical variables, and to provide a geochemical zonation of the core based on principle element associations, I ran a Q-mode factor analysis using the extended CABFAC program of Klovan and Miesch (1976). Prior to running the factor analysis, concentrations of all oxides and elements were transformed to proportions of the total range for each oxide and element. As a result of the transformation, all data were expressed on a scale of 0.0 to 1.0. After trying several different sets of reference axes in multidimensional space, I chose four orthogonal reference axes (4-factor solution) that maximize the variance of the transformed data in each dimension (varimax solution of Klovan and Miesch, 1976). The 4-factor model accounted for more then 55% of the variance in the scaled data for each variable, with an average of 70%. Basically, the 4-factor model reduced 24 measured variables (oxide and element concentrations) to four "composite" geochemical variables (factors). The intensities of the composite geochemical variables are the factor loadings. The loadings for each of the four factors for each sample are plotted versus depth in Figure 5.

The factor loadings describe the relative importance of each of the factors for each sample, but give no indication of which of the elements had the most influence on determining each of the four factors. In order to determine which elements contributed to which factor, the factor loadings for each sample were treated as composite chemical variables with attributes of one or more actual measured variables, and correlation coefficients were computed between the loadings and the 24 measured variables. Results of the correlation analysis are given in Table 2.

The correlations between factor loadings and geochemical variables (Table 2) indicate four element associations (factors) that can be used to zone the core based on the geochemical results (Adam and others, 1989). Factor I groups samples based on concentrations of Ti, nonbiogenic Si, Al, La, Zn, Ce, Cu, Ba, Ni, Co, Y, Pb, Cr, and Sc, in order of decreasing correlation coefficients in Table 2. Figure 5 shows that sediments characterized by Factor I predominate throughout most of the recovered section in Tule Lake except in some beds in the carbonate-rich interval between 70 and 120 m.

Factor II groups biosiliceous, OC-rich samples. These samples occur in several beds within the carbonate-rich interval (70-120 m), within the organic-rich interval between 130 and 160 m, and at the bottom of the core (Figure 5). This zonation is based mainly on the concentrations of biogenic SiO₂ and organic carbon (Table 2).

Factor III sediments are those that contain the highest concentrations of CaCO₃ in the interval between 70 and 120 m. These carbonate-rich sediments also contain the highest concentrations of the carbonate-related elements Sr and Mg. Notice in Table 2 that the strongest variable in this carbonate association is CaO and not CaCO₃. The CaO values are total calcium measured by XRF and include both clastic calcium and calcium from CaCO₃. The CaCO₃ values were calculated from inorganic carbon. The fact that CaO is the strongest variable in the carbonate association suggests that analytically the CaO values provide a better estimate in the variations in CaCO₃ than inorganic carbon. A scatter plot of CaCO₃ computed from both CaO and carbonate carbon (both values listed in Appendix I) is shown in Figure 6. The values calculated from CaO are usually higher because of calcium from the clastic fraction.

(click for larger image- 3.5K) Figure 6. Scatter plot of Percent CaCO3 calculated from total calcium(-Ca) and carbonate (-CC) in samples of sediment from Tule Lake, California.

Factor IV is weak, but appears to reflect variations in redox conditions within the lake. Factor IV sediments have relatively high concentrations of Fe, Mn, Sc, V, and P. CaCO₃ computed from carbonate carbon also contributes to this association (Table 2).

I interpret factor I sediments to represent a volcanic-ash association characterized by relatively high concentrations of Ti, nonbiogenic Si, Al, rare-earth elements, and trace transition elements. The raw geochemical data (Figure 4) and distribution of factor loadings (Figure 5) show that Tule Lake sediments are characterized by a predominant igneous rock component with varying amounts of biogenic silica (diatom debris) and organic matter, and, in a few beds, minor amounts of carbonate. The igneous rock component consists mainly of locally derived basic tephra with at least 12 acidic tephra layers that are widely distributed in the Pacific Northwest and provide the basis for regional correlations and tephrochronology of the Tule Lake core (Sarna-Wojicicki and others, 1988; Adam and others, 1989; Rieck and others, 1992).

In order to determine how much of the composition of the Tule Lake sediments can be explained in terms of the igneous rock component, a standard reference material is needed. An ideal reference would be the average composition of rocks in the drainage basin of Tule Lake. Lacking this information, I chose analyses of 145 samples of volcanic rocks from the Medicine Lake Highlands just south of Tule Lake (Figure 1) provided by Julie Donnelly-Nolan (Donnelly-Nolan and Nolan, 1986) as the basic igneous end member. The average of these analyses I will call MLV. Concentrations of major- and minor-element oxides were determined for all 145 MLV samples, but concentrations of only a few trace elements (Ba, Cu, Ni, Sr, Y, and Zn) were determined on a subset of the MLV samples. Because of the limited trace-element data for MLV, I also compared the Tule Lake ash to USGS standard basalt BCR-1 because there are more extensive trace-element data for this standard (Flanagan, 1969). For the acidic end member, I used USGS standard RGM-1, a rhyolite from Glass Mountain in the Medicine Lake Highland (Figure 1) (Tatlock and others, 1976). I chose sample 986 from a depth of 51.1 m (Appendix I) as a reference volcanic ash from the Tule Lake core. This sample probably represents a good average ash-rich sediment for the Tule Lake core because it comes from an interval in the core (50 to 65 m) that was deposited very slowly and contains several heterogeneous reworked tephra units (Adam and others, 1989; Rieck and others, 1992). A second reference analysis from the Tule Lake core is simply an average of analyses of all Tule Lake samples (Appendix I). These reference analyses are listed in Table 3.

Figure 7 shows log-log plots of concentrations of major- and minor-elements oxides (in percent) in Tule Lake sample 986 versus those in the MLV and RGM-1 standards, and concentrations of trace elements (in parts per million) in Tule Lake sample 986 versus those in the MLV and BCR-1 standards. Figure 7A and B shows that the major- and minor-element composition of sample 986 is closer to the basic end member (MLV) than to the acidic end member (RGM-1), particularly when considering those oxides that are least likely to be affected by weathering (SiO₂, Al₂O₃, Fe₂O₃, and TiO₂). Tule Lake sample 986 is depleted in alkaline-earth- and alkali-element oxides CaO, MgO, Na₂O, and K₂O relative to MLV, and this probably reflects removal of these elements by weathering. Tule Lake ash, represented by sample 986, is depleted in most trace elements relative to BCR-1. They are distinctly enriched in Cr, Cu, and Ni relative to BCR-1 (Figure 7C), but are somewhat depleted in Cu and Ni relative to MLV (Figure 7D). Table 3 shows that average Tule Lake sediment has a composition similar to that of ash sample 986, but is enriched in SiO₂ because of inclusion of biogenic SiO₂ from diatoms. In summary, it appears that the detrital clastic material that reached the Tule Lake basin throughout the history of the lake was derived from weathered volcanic rocks from the Medicine Lake Highlands (Figure 1). In the lake, the detrital clastic material was diluted by major amounts of biogenic SiO₂ and minor amounts of carbonate and organic matter. In other words, the sediments of Tule Lake are composed mainly of weathered volcanic ash and diatoms.

Figure 7
(click for larger image- 6.9K) Fig. 7a & 7b	(click for larger image- 6.9K) Fig. 7c & 7d
Figure 7. Log-log plots of concentrations of major- and minor-element oxides (in percent) in Tule Lake sample 986 versus those in average Medicine lake volcanics (A) and USGS standard rhyolite RGM-1 (B); and concentrations of trace elements (in parts per million) in Tule Lake sample 986 versus those in average Medicine Lake volcanics (C) and USGS standard basalt BCR-1 (D).

As explained earlier, I used the SiO₂/Al₂O₃ ratio in the MLV standard (3.5; Table 3) and the Al₂O₃ content of each sample to compute the concentration of nonbiogenic SiO₂. Percent biogenic SiO2 is then calculated by subtracting % nonbiogenic SiO₂ from total SiO₂. Using these calculations, average Tule Lake sediment contains 39.2% nonbiogenic SiO₂ and 24.8% biogenic SiO₂. Down-core variations of the computed estimates of these two forms of SiO₂ are shown in Figure 8 along with down-core variations in organic carbon repeated from Figure 4.

Q-mode factor II suggested that biogenic silica and organic matter were closely associated. Intuitively, this is what one might expect because most of the organic productivity in Tule Lake probably was from diatoms. However, based on all samples, the concentration of biogenic SiO₂ does not correlate very well with that of organic carbon (r=0.20, n=131). I used the moving correlation coefficient technique of Dean and Anderson (1974) to determine if there were any parts of the Tule Lake sequence where biogenic SiO₂ and organic carbon were more positively correlated than others (Figure 8). This technique computes correlation coefficients between two variables over a predetermined stratigraphic window that is moved down the section one point at a time. For the Tule Lake sequence I used a window that was 11-samples long so the coefficients plotted in Figure 8 represent correlations between samples 1-11, 2-12, 3-13, etc. Figure 8 shows the weak overall positive correlation between these two variables, but also shows that there are some intervals of much stronger positive correlation (r>0.4). In general, these intervals of strong positive correlation correspond to intervals of higher organic carbon content, particularly in the upper part of the sequence.

The factor analysis indicated that the sediments from Tule Lake could be described in terms of a four-component system of clastic material, carbonate, organic matter, and diatom debris. Based on the above discussion, the amounts of these components can be quantified. Assuming that the average clastic fraction has an Al₂O₃ concentration similar to that of average Medicine Lake volcanics (i.e. about 16.5; Table 3), then an average measured concentration of Al₂O₃ of 11.2% (Table 3) indicates that the average Tule Lake sediment is about 68% clastic material, derived mainly from basic igneous rocks in the drainage basin. The average concentration of organic matter would be about twice the organic carbon concentration or about 4.5%. The average concentration of CaCO₃ (computed from carbonate carbon) is 2.6% (Table 1), and the average computed biogenic SiO₂ content is 24.8%. These values add up to 99.9% and show that, on average, basic igneous rock debris makes up the bulk of Tule Lake sediment. The average loading for the clastic factor from the Q-mode analysis (Factor I) is 0.7, and the variations in Factor I loadings (Figure 5) can be used as a rough estimate of the proportion of the clastic fraction, ranging from as much as 0.9 (90%) to zero or near zero in lacustrine beds rich in carbonate, diatoms, and organic matter.

(click for larger image- 16K) Figure 8. Plots of concentrations of nonbiogenic SiO2, biogenic SiO2, and organic carbon; and moving correlation coefficients between biogenic SiO2 and organic carbon verus depth for samples from Tule Lake, California. (See text for methods of computing nonbiogenic SiO2, biogenic SiO2, and moving correlation coefficients.)

<< Methods || References Cited >>

AccessibilityFOIAPrivacyPolicies and Notices
U.S. Department of the Interior | U.S. Geological Survey URL: https://pubsdata.usgs.gov/pubs/of/1996/ofr-96-0257/results.htm Questions or Assistance: USGS Web Contact Page Last Modified: Wed Dec 7 16:02 EST 2016

---

Page designed by Joey Fisher