Dan Cayan1,2 Mary Tyree1 and Mike Dettinger2,11 Scripps Institution of Oceanography - UCSD, 9500 Gilman Drive - 0224, La Jolla, CA 92093-0224
2 U.S. Geological Survey
Monthly counts of female Cx. tarsalis from scores of Sacramento/Yolo, Sutter/Yuba, and Kern MVCD samples in the Central Valley of California are investigated to determine how they relate to climate variability over the last 2-4 decades. There has been considerable interannual variability in Cx. tarsalis abundance, part of which is correlated across the entire region. Summer Cx. tarsalis fluctuations have correlated positively with moisture accumulation in the region, as indicated by precipitation, streamflow and snow water content in the Sacramento-San Joaquin watersheds. Prediction of summer Cx. tarsalis abundance from prior winter moisture variables may provide a useful look ahead at several months lead-time.
As part of a collaborative NOAA Office of Global Programs grant, we are examining temporal changes in monthly mosquito abundance and their links to short term climate variations. Seasonal climate variations are substantial in California (Cayan and Riddle, 1993; Stahle et al., 2001; Knowles, 2002) and they have large regional footprints, especially in winter (Cayan and Peterson, 1989; Cayan 1996; Dettinger et al., 1998). Seasonal climate forecasts are only modestly successful but, certain conditions, such as large El Nino events, may provide useful skill for predicting temperature and precipitation over California (Namias, 1978; Cayan et al., 1999; Dettinger et al., 1999; Gershunov et al., 2000). The objectives of this study are to quantify relationships of vector counts with climate and, to develop and assess long-range (seasonal) forecasts. Our rationale is based on the premise that if climate contributes significantly to variations in mosquito abundance, then measurements of mosquitos across the California region should vary in a similar fashion and be linked to changes in temperature and water availability.
Time series of monthly mosquito abundance (Cx. tarsalis females counts [/trap-day] were supplied by C. Barker of the Arbovirus Field Station in Bakersfield, California, from New Jersey light trap records maintained by the Kern Mosquito and Vector Control District [MVCD] in the southern San Joaquin Valley, and by Sacramento/Yolo and Sutter/Yuba MVCDs in the Sacramento Valley of California. We have focused on female Cx. tarsalis because of its central involvement in the transmission of encephalitis viruses. The three data sets considered here cover the periods 1973-2000, 1986-2000, and 1954-2000, respectively. Because many of the mosquito traps were unevenly sampled thru time or provided fewer than 10 years of data, for several of the analyses to follow, a subset of 28 mosquito trap series were selected from the original 54, 62 and 49 mosquito trap histories from Kern, Sacramento/Yolo and Sutter/Yuba Mosquito Districts collections, respectively. For some of the analyses, the monthly mosquito abundance data were log transformed to provide more normally distributed time series.
In our search for climatic links, monthly precipitation from the NOAA San Joaquin and Sacramento climate divisions, monthly flows from the U. S. Geological Survey Kern River gage, and April 1 snow water content from selected California Department of Water Resources Cooperative Snow Surveys snow courses in the southern and central Sierra Nevada were employed. All of these climate data were available from 1954-2000.
Abundance and Seasonality
The overall monthly averages of Cx. tarsalis females at representative mosquito traps from the three Districts varied considerably across the region, with nearly an order of magnitude difference between individual traps (Figure 1, right).
Month of climatological maximum (left) and annual mean log10 (monthly counts) of Cx. tarsalis females captured in New Jersey light traps of the Sutter/Yuba, Sacramento/Yolo and Kern MVCDs. Years of record as shown on map.
Cx. tarsalis abundance peaked in the summer between June and September in almost all traps (Figure 1, left and figure 2, upper). Interestingly, the mosquito series from the southernmost traps, those in Kern County and the southern portion of Sacramento/Yolo Counties, had peak abundance in later summer (August and September), while the northernmost traps, those from the Sutter/Yuba District and the western and northern locations from the Sacramento/Yolo District, had peak abundance in June and July.
Monthly mean (upper) and coefficient of variation (standard deviation/mean) of log transformed counts of Cx. tarsalis females at selected light traps.
Monthly and Interannual Variability
Inspection of the Cx. tarsalis abundances reveals relatively large monthly and interannual variability and, in many cases, remarkable trends over the duration of the sampling period. Many of these trends are towards decreasing abundances with time; perhaps because environmental conditions and light levels at the trap sites have changed. Therefore, the linear trend was removed to yield (detrended) residual anomalies for the subsequent correlation analyses.
Considering the variability of individual months, the relative variability (coefficients of variate of the log transformed monthly counts) tend to be greatest in the transition months before and after the summer peak period (Figure 2, lower). This coefficient of variation measure ranged from 0.2 to nearly 1.0.
By correlating the anomalous, detrended monthly mosquito counts at one selected reference series from each of the three MVCDs, with those from each of the other mosquito trap series, an idea of the typical spatial span of higher and lower than normal mosquito abundances can be determined. Monthly anomalies were calculated by subtracting the monthly long-term average (detrended) value from the value (detrended) for each particular month. This exercise, performed for the early summer (May-July) and late summer (August-October) months, shows modest coherence among nearby and relatively distant mosquito traps, especially in late summer.
Short Period Climate Linkages
Because part of the mosquito life cycle is keyed to water availability, tests for linkages between anomalous monthly moisture-related climate variations and mosquito abundances were conducted. For the Kern MVCD series, we used winter precipitation, spring snow accumulation and water year streamflow from representative gages. The correlations among these series and monthly-detrended mosquito abundances were computed for a subset of the mosquito series containing 10 or more years of data. Using April 1 snow water content as an index, these correlations were computed with the April-October monthly mosquito abundance data. Correlations that were significant, different from zero, were nearly all positive. Strongest correlations from the Kern dataset occurred in late spring and early summer (April - June); they are relatively weak in July - October (Figure 3).
Correlations between April 1 snow water contents at various elevation zones in the Kern River watershed and Kern MVCD. April - June female Cx. tarsalis (upper) and July - September female Cx. tarsalis (lower).
In contrast, correlations of winter (December-February) precipitation with Sacramento/Yolo and Sutter/Yuba detrended mosquito abundance are weak (or even negative) in April and May, but increase in summer, with maximum positive correlations in August and September (Figure 4). From temperature correlations, we see that warmer temperatures in winter are associated weakly with higher AMJ female Cx. tarsalis abundances at Kern MVCD, but this connection is not as strong as were the connections for winter and spring precipitation.
Correlations between December - February Sacramento climate division precipitation and selected Sacramento/Yolo MVCD monthly female Cx. tarsalis counts.
Towards Seasonal Forecasting
Importantly, it appears that for several of the female Cx. tarsalis series, there is a positive correlation with winter precipitation (or spring snow accumulation) and summer abundances, meaning that prior season moisture indices may be useful predictors of summer mosquito abundance. As a preliminary test of this, linear regression was developed for predicting the average of Kern MVCD abundance series from April 1 snow water. This model, when fitted to the early (1973-1990) record, captured 40% of the year-to-year variance of April-June mosquito abundance; when applied to the remaining 10 years of independent data (1991-2000), predicted 29% of the variance of April-June mosquito abundances (Figure 5).
Comparison of observed and "forecasted" April-June female Cx. tarsalis counts, based on 1973-1990 linear regression with April 1 snow water content in Kern River watershed; Cx. tarsalis counts are averages from Kern MVCD traps 8004 and 8007.
Further work will relate other regional moisture variables (streamflow, soil moisture) and broader scale climate measures (atmospheric circulation, tropical Pacific El Nino/Southern Oscillation conditions, and Pacific sea surface temperature) to mosquito abundances. Mosquito trap series are being compiled from additional MVCDs and will be assembled in electronic format, so these will also be examined. The potential for useful, seasonal predictions also needs to be more fully explored and developed.
Cayan, D.R., K.T. Redmond, and L.G. Riddle, 1999: ENSO and hydrologic extremes in the Western United States. J. Clim., 12(9), 2881-2893.
Cayan, D.R., 1996: Interannual climate variability and snowpack in the western United States. J. Clim., 9(5), 928-948.
Cayan, D.R. and L. Riddle, 1993: Atmospheric circulation and precipitation in the Sierra Nevada. Proceedings of the AWRA 28th Annual Conference and Symposium, Reno, Nevada, Nov 1-5, 1992, Managing Water Resources During Global Change, R. Herrmann, Ed, 711-720.
Cayan, D.R. and D.H. Peterson, 1989: The influence of North Pacific atmospheric circulation on streamflow in the West. Geophys. Monograph 55: Aspects of Climate Variability in the Pacific and the Western Americas, December 1989, 375-397.
Dettinger, M.D., D.R. Cayan, and K.T. Redmond, 1999: United States streamflow probabilities based on forecasted La Nina, winter-spring 2000. Experimental Long-Lead Forecast Bulletin, 8(4), 57-61.
Dettinger, M.D., D.R. Cayan, H.F. Diaz, and D.M. Meko, 1998: North-south precipitation patterns in western North America on interannual-to-decadal time scales. J. Clim., 11(12), 3095-3111.
Gershunov, A., T.P. Barnett, D.R. Cayan, T. Tubbs, and L. Goddard, 2000: Predicting and Downscaling ENSO Impacts on Intraseasonal Precipitation Statistics in California: The 1997/98 Event. J. Hydromet, 1, 201-210.
Knowles, N., 2002: Natural and human effects in the San Francisco Bay-Delta Estuary and watershed. Water Resources Research, (in press).
Namias, J., 1978: Multiple causes of the North American abnormal winter 1976-77. Mon. Wea. Rev., 106, 279-295.
Stahle, D.W., M.D. Therrell, M.K. Cleaveland, D.R. Cayan, M.D. Dettinger, and N. Knowles, 2001: Ancient Blue Oaks reveal human impact on San Francisco Bay salinity. EOS, Transactions, American Geophysical Union, 82(12), 141, 144-145.