Search tips
Search criteria 


Logo of tropmedLink to Publisher's site
Am J Trop Med Hyg. 2010 September; 83(3): 624–632.
PMCID: PMC2929061

Interannual Variability of Human Plague Occurrence in the Western United States Explained by Tropical and North Pacific Ocean Climate Variability


Plague is a vector-borne, highly virulent zoonotic disease caused by the bacterium Yersinia pestis. It persists in nature through transmission between its hosts (wild rodents) and vectors (fleas). During epizootics, the disease expands and spills over to other host species such as humans living in or close to affected areas. Here, we investigate the effect of large-scale climate variability on the dynamics of human plague in the western United States using a 56-year time series of plague reports (1950–2005). We found that El Niño Southern Oscillation and Pacific Decadal Oscillation in combination affect the dynamics of human plague over the western United States. The underlying mechanism could involve changes in precipitation and temperatures that impact both hosts and vectors. It is suggested that snow also may play a key role, possibly through its effects on summer soil moisture, which is known to be instrumental for flea survival and development and sustained growth of vegetation for rodents.


Plague is a disease caused by the bacterium Yersinia pestis. Humans may accidentally become infected when bitten by plague-infectious flea vectors or when handling plague-infected hosts.13 Unless quickly treated with antibiotics, the disease is fatal to humans.1,2 Plague is circulating and/or persisting in many populations of small mammals and other wildlife species in diverse biotopes around the world.4 Human-plague cases occur regularly in the western United States.5 Plague was first introduced to the United States in San Francisco during the last pandemic (1899), most likely by a merchant ship arriving from Hong Kong, China.6,7 Ultimately, transmission among native rodents and their fleas resulted in an eastward spread of the disease and the establishment of discontinuous plague foci across the western United States8,9 as far east as the 100th meridian.6,10 Within 10 years, plague had spread from urban rats to wild rodents such as sciurid, sigmodontine, or microtine rodents living in surrounding rural areas.9 Climate factors are known to regulate both host and vector abundances.1114 Small mammal population dynamics are controlled by both intrinsic and extrinsic factors, where the latter includes contemporaneous and past local temperature and precipitation.11,1420 Fleas spend the majority of their life cycle in the environment and thus, are affected by surrounding conditions, notably by temperature and humidity changes.2123 During plague epizootics, the abundance of these hosts and the prevalence of Y. pestis in them, as well as the numbers of active fleas,24 are high.2527 At such times, the disease expands spatially, increasing the chances of infection for ecologically associated species or humans and their domestic animals.9 Earlier studies have consistently showed that these scenarios happen while local climatic conditions are favorable to both hosts and vectors,28,29 possibly because of a cascade of events involving above-normal precipitation and mild temperatures.29,30

The specific combination of rodent and flea species responsible for plague maintenance and amplification in the United States varies markedly across the region.9 It is, therefore, desirable to develop means of understanding and predicting the current occurrence of human plague and how this may change with possible climatic changes that are independent of the particular rodents and fleas involved. Earlier studies had concluded that there was no significant association between human-plague occurrence and large-scale climate variability,29 and this conclusion has been widely accepted in World Health Organization (WHO) reports and other sources.2,28,3136

At interannual scales, climate variability over the western United States is primarily characterized by two modes of fluctuations, with a spatial imprint extending outside this limited region: the Pacific Decadal Oscillation (PDO; decadal time periods)37 and El Niño Southern Oscillation (ENSO; 3- to 4-year time periods).38 Positive (respectively [resp.]; negative) PDO and ENSO (El Niño; resp. La Niña) phases are both associated with wetter (resp.; drier) and milder climate over the western United States. A variety of ecosystems have been shown to respond to either or both of these climate fluctuations.37,39 Recently, it was shown that the number of human cases in this region is linearly linked to PDO and the number of abnormally hot days,40 but both PDO and ENSO are responsible for the spatially coherent and predictable climate anomalies in this region.37,41,42 These large-scale climate indices are known to be efficient predictors of ecological processes in other systems, including rodent population dynamics and demographic rates.4348 In this study, we examine the associations between climate anomalies driven by PDO and ENSO and the dynamics of human plague and discuss the mechanisms that might underlie such associations. Specifically, we focus our discussion on one link in the cascade hypothesis, namely the effect of precipitation and temperature patterns (as captured by PDO and ENSO) on primary production (i.e., rodents host food availability as captured by the normalized difference vegetation index [NDVI] index) in the plague areas. Finally, we look at the possible effects of climate change on plague dynamics.

Materials and Methods


Plague data.

We used annual counts of human plague cases in the counties of exposure, available from 1950 to 2005. We considered all 105 counties (distributed across 13 states) reporting at least one plague case over the study period. No correlation exists between plague outbreak frequency and population density at the county level. The largest human-plague outbreaks typically occur in primarily rural counties with close to median population density (?3/km2). Plague counts are adjusted by county population density to produce county-level human-plague time series compatible with climate. Population density, D(c,y), for county (c) in year (y) is given as D(c,y) = population(c,y)/area(c) and is based on linearly interpolated population censuses for each county performed every 10 years by the US Census Bureau ( The adjusted plague time series (Padj) is calculated as Padj(c,y) = P(c,y)/D(c,y). The raw (P) and adjusted plague time series exhibit a strong correlation (correlation = 0.92; P value < 0.001) (Supplemental Figure S1, available at The data reveal a period of increased mean infection during 1970–1988, with an overall peak of 40 human cases around 1983.

Climate data.

Our analyses use sea surface temperature (SST) averaged over the so-called NINO3.4 region (5° S–5° N; 170–120° W) for the same time window (1950–2005). We calculated the standard deviation (SD) for this index and considered three different phases for ENSO: El Nino (index > 1 SD), neutral (−1 SD < index < 1 SD), and La Nina (index < −1 SD). The strongest ENSO signal in precipitation over the southwestern United States is observed in late winter to early spring (i.e., when local precipitation has been documented by Parmenter and others29 to be important for plague hosts and vectors). Values for the PDO index were obtained from Both ENSO and PDO are characterized by high persistence through the fall and winter until early spring. March values for these indices are representative of recent conditions as well as a measure of springtime climate forcing, which, as we shall see, is very important for summertime plague. Figure 1B and andCC depicts the time series of March PDO and March NINO3.4.

Figure 1.
(A) Total density-adjusted human-plague cases (so-called plague time series). A total of 430 cases were recorded from 1950 to 2005. (B) PDO index (March value). Red (resp. blue) bars indicate positive (resp. negative) values. (Note the major shift in ...

Extensive temperature and precipitation records are freely available (NCDC, 2003, for thousands of available meteorological stations in the United States. From these records, we chose a network of the 100 stations with the highest quality data (defined as those with the fewest missing data and the highest proximity to plague counties) (Figure 2) to build proxies for monthly temperature and precipitation regimens across the western United States.

Figure 2.
Plague distribution over the western United States; red crosses are placed at the center of the counties (delimited by grey lines) reporting human plague, and grey dots are at the 100 optimal (see Text) weather stations.

NDVI data.

The NDVI49 data cover the study area (100–125° W, 30–45° N) in a 0.25° resolution grid of monthly composite values compiled over the period 1982–1998. This is roughly the time period over which advanced very high resolution radiometer (AVHRR) data from Pathfinder are available with concomitant solar zenith angle and view zenith angle (and relative azimuth) so that a Fourier-Adjusted, Sensor and Solar Zenith Angle-Corrected, Interpolated, Reconstructed (FASIR) can be constructed (O. S. Los, personal communication). NDVI is an index of the light absorbed by chlorophyll and thereby, a derived measure of primary production. In arid zones with a simple vegetation structure, NDVI reflects variation in ground chlorophyll and vegetation coverage when snow cover is absent. For data documentation, see

Wavelet analyses.

In nature, non-stationary processes are common, and increasing evidence suggests the importance of transient dynamics in ecological processes.5053 Epidemiological time series are typically noisy, complex, and strongly non-stationary.54 Wavelet analyses provide powerful tools for analyzing such signals.5558 It is indeed well-suited to dealing with transient relationships between two signals (e.g., climate interaction with the dynamics of an epidemic). Wavelet analysis performs a time-frequency (f) decomposition of the signal and allows one to follow the evolution of the different frequency components as time progresses. We used wavelet analysis and its bivariate extensions, wavelet cross-spectrum and wavelet coherency,55 to analyze the plague data and their statistical relationships with the climatic time series (PDO and ENSO). The statistical significance of the univariate and bivariate analyses was assessed using β surrogates.59 This approach accounts for the autocorrelation structure of the original time series by assuming an underlying 1/f model for its power spectrum. Contrary to white noise and autoregressive processes, it reproduces the relative distribution of the whole range of frequencies in the signal and thus, accounts for the dominance of low frequencies often found in ecological time series.

All of the computations were done using R version 2.4 ( on the basis of the wavelet libraries developed by Cazelles and others.50,54,55


PDO and ENSO affect precipitation and primary production in the western United States.

PDO exhibits coherent positive correlations with late winter/early spring precipitation throughout the area.37,42,60,61 Consistent positive precipitation anomalies occur in late winter to spring when PDO is high (i.e., above 1 SD); this pattern is most marked when El Niño events combine with positive PDO. Conversely, a low PDO/La Niña combination results in below normal precipitation (Figure 3). Snow anomalies (measured as the amount of water precipitated on days when the daytime temperature is below freezing) are also higher during El Niño/PDO positive phases (Supplemental Figure S2, available at

Figure 3.
(A) Composite late-winter to spring precipitation anomalies for years combining significant (Left) negative (resp. positive; Right) PDO and La Niña (resp. El Niño) events as defined by below (resp. above) 1 SD for both indices. (B) Composite ...

NDVI is a good indicator of yearly food availability for small mammals living in semi-arid areas.62,63 Yearly maximum NDVI, a proxy for yearly maximum food available for small mammals,64,65 reveals above- and below-mean values between 1982 and 1998. NDVI above-mean years coincide with above-mean total precipitation (Figure 1). In the Four Corners region plateau, NDVI spatial anomalies during El Niño/positive PDO years are consistently above normal (120–150% in 1983, 1987, 1992, and 1998) and below normal during the La Niña/PDO negative year (60–80% in 1989). This pattern is less consistent in the Pacific and Northwest areas (Figure 3).

PDO and ENSO significantly explain the temporal variability of plague cases.

To explore human-plague patterns across the whole region, we used the spatial sum of all population density-scaled human cases (Figure 1A and Supplemental Figure S3, available at (a measure we refer to as the plague time series; see also plague occurrences in consecutive time windows). Plague outbreaks occur predominantly during positive phases of PDO; the number of population-adjusted human-plague occurrences is, on average, 4.4 during positive PDO phases and 0.9 during negative phases (or respective plague anomalies of 138% against 59%; P < 0.05) (Table 1). The correlation between the plague time series and the ENSO index alone is, however, generally not significant. Looking more carefully, periods of high plague occurrence seem to occur when El Niño events coincide with a positive PDO phase (e.g., 1959, 1983, 1987, and 1992). Conversely, decreases in plague cases after La Nina events are accentuated if they occur during PDO negative phases (36% versus 52%; P < 0.05) (Table 1).

Table 1
Contingency table of population-density-adjusted plague cases for southwestern US for each El Nino (La Nina) years crossed with High (Low) values of the PDO (as percentage of mean) with significance

Because the strong autocorrelation at lag one in the plague time series (r [plaguet, plaguet − 1] = 0.72; P < 0.001) might lead to an overestimation of the tested significance of the associations measured in Table 1, we carried out a wavelet analysis of the plague time series and cross-wavelet analysis of both PDO/ENSO and plague case numbers. We tested significances against 1,000 bootstrapped time series with similar frequency spectra (i.e., a dominance of low frequency; see Materials and Methods). The plague time series shows interannual cycles with periods of 7–8 years and 3–4 years that are most marked from the mid-1970s to the late 1980s after a major PDO shift (Figures 1B and and4).4). This is not inconsistent with the time periods generally associated with PDO and ENSO. To test a plague ENSO association conditional to PDO, we computed a cross-wavelet correlation and coherency. It revealed that plague and PDO are significantly coherent (P < 0.05) throughout the entire time window (Figure 4).

Figure 4.
(A) Time-frequency decomposition of the plague time series. (B) Coherency analysis between the plague time series and PDO. Thick discontinuous lines indicate significance (P < 0.1).

Three main plague regions reportedly exist in the west,9 which led us to subdivide the plague time series into three aggregated series for the southwest (30–42° N, 115–100° W), the North (42–45° N, 125–100° W), and the Pacific plague regions (30–42° N, 125–115° W) (Supplemental Figure S4 A, available at Phase analysis revealed that summed plague cases in these three regions are synchronized from around the plague outbreak in 1980–1983 until the mid-1990s (Supplemental Figure S4C, available at Time-frequency decomposition of the time series (Supplmental Figure S4B, available at also revealed that although the three regions have a dominant periodicity around 8 years, it is significant only in the Southwest (note that from around 1980–1990, all areas show a significant dominant periodicity at 3–4 years). Precipitation and NDVI anomalies (Figure 3) also revealed subtle opposite tendencies between the southwest on one hand and the north and Pacific on the other hand.

Climate change.

We explored the possible effect of climate change in the 1990s to explain discrepancies between the expected and observed number of plague cases. Indeed, the 1998 El Niño event occurring at high PDO, which lead to above-mean precipitation and NDVIs, did not result in above-normal plague cases. However, above-mean counts of hot days were recorded in 1998 at all stations (temperature > 37°C) (Figure 1F), which should be detrimental to plague28,40; warm nighttime temperatures also were observed at this time (Figure 1G) (see Discussion). Climate change has been pointed out by some as partly responsible for these trends, and nighttime temperatures have increased about 0.5°C in the west since 1990 (Figure 1G). In the case that these trends continue, some have proposed that areas of recurrent plague activity (i.e., endemic areas) will shift to higher latitudes/elevations.36,66 In our data, we indeed find a trend to higher elevation/latitude, but this trend may have started in the 1980s (Supplemental Figure S5, available at and is not significant; the use of exposure-site data for cases rather than county-level data could provide more specific conclusions to be made on the possible role of climate change on what locations will have the greatest amounts of human-plague activity in the future, but such analyses are outside the scope of the current study.


In our efforts to identify a region-wide, system-independent link between climate and the occurrence of human-plague cases in the western United States, we found that ENSO and PDO patterns, especially in combination, can be related—first, to climatic variables in the United States known to be important in determining the probability of human plague, and second, to the number of human-plague cases themselves. We showed that the years that had both high numbers of plague cases and high precipitation coincide with above-mean values of NDVI. We also noted that recent rises in temperature associated with reductions of the snow pack and its persistence correspond to anomalously low numbers of plague cases.

Significant association between large-scale climate variability and plague.

In an earlier study, Parmenter and others29 showed that local precipitation in winter and spring significantly increased the number of human-plague cases in New Mexico. These authors tried but were unable to find significant links between plague and ENSO, the dominant mode of Pacific-sector climate variability. They concluded, therefore, that local, not large-scale, climate drives plague dynamics. Several reasons may have led to this conclusion, including the use of an overaggregated index of ENSO (i.e., mean from October to May of the Southern Oscillation Index [SOI]) that is not the most relevant for North American precipitation anomalies (SST averaged over the so-called NINO3.4 region [5S–5° N; 170–120° W] is more relevant for well-documented reasons).42 Additionally, the North Pacific decadal climate variability (i.e., PDO) also exerts a direct influence on North American climate, sometimes enhancing and sometimes obscuring ENSO teleconnections.37 In this context, the clearest link is between PDO and plague levels in the west.40 The present study proposes a mechanistic explanation for this link by showing the existence of spatio-temporal coherent anomaly patterns in surface temperature and precipitation associated with PDO. This result is consistent with the dominance of PDO over ENSO regarding the storm-track position.60 Seasonal climate anomalies over North America exhibit strong variability between years characterized by the same ENSO phase and are caused largely by modulation by PDO.42 Earlier studies28,29,40 did not take this into account and were unable to find links between southwestern plague and ENSO. El Nino (resp., La Nina) signals to North American climate are strongest and most spatially coherent and predictable when El Nino (resp., La Nina) occurs during the positive (resp., negative) phase of PDO.42 Therefore, Los Niños consistently results in fairly mild winters and above-normal precipitation if occurring during the positive phase of PDO (i.e., in 1953, 1966, 1983, 1987, 1992, 1993, and 1998),37 whereas La Nina events result in coherent dry and above-normal hot climate over the western United States, especially if occurring during the negative PDO phase (i.e., in 1950, 1951, 1955, 1956, 1971, and 1974). Accordingly, we found that El Niño events occurring during positive PDO excursions are related to high plague counts and that La Niña events occurring during negative PDO consistently led to low plague values (Table 1). A notable exception is the most recent El Niño event, which was not followed by increased plague numbers (i.e., in 1998 [positive PDO]); the same was true, to a lesser extent, in 2003 (neutral PDO), but see discussion of climate change for further information.

Wavelet analysis enabled us to better determine the transient associations between PDO/ENSO and the plague time series. The coherency analysis and cross-wavelet spectrum of PDO and plague confirmed that both signals are significantly coherent over the total study period (Figure 4). Furthermore, both signals are significantly correlated from the mid-1970s at a quasi-decadal periodicity, suggesting that the low-frequency trend in the plague time series could be attributed to low-frequency changes in the western climate. The observed dominant 3- to 4-year periodicity in the plague spectrum could be linked to ENSO (Figure 4A). The time series in the three main plague regions9 (Supplemental Figure S4) studied independently also exhibits oscillations with dominant periods at around 3–4 years that are particularly synchronous during the decade when plague was most active (1980–1990). The southwestern focus is the most active among the three major plague regions in North America, with the highest counts of human cases. This area exhibits a significant dominant periodicity at 8–9 years and therefore, is primarily responsible for the coherency between the plague time series and PDO.

We know that ENSO and PDO have a strong effect, particularly in the southwestern part of our study area—an effect that can be different for the northern part of the western United States (Figure 3) ( Indeed, spatial patterns of precipitation and NDVI anomalies show that the combination of El Niño and positive PDO corresponds to wetter, greener years in the southwest, whereas the northwest and the Pacific tend to be less affected. Phase analysis (Supplemental Figure S4C), nevertheless, strongly suggests that, during the strong 1983 El Niño event combined with high PDO, all areas were similarly affected, which resulted in the highest plague years of the time series (Figure 1A). The fact that the Western and Northern areas exhibit differences with the Southwest could be explained by differences in the dominant flea species in these regions and their seasonality,9 the delayed effects of high precipitation on local rodent populations, or more simply, the small number of cases.

Potential mechanisms to explain the association between human plague and climate variability.

We have shown that the number of human-plague cases exhibits large-scale coherent positive correlations with late winter to spring precipitation (Supplemental Figure S6, available at Although yet to be proven for plague, the cascade hypothesis provides us with a possible platform from which to understand the effects of climate variability on the plague system in the United States.29,30 In assessing whether this general scheme was consistent at larger scales, we extended our precipitation analysis to NDVI anomalies. El Niño/PDO positive years inducing coherent positive precipitation anomalies corresponded to positive NDVI anomalies (Figures 1E and and3),3), suggesting that the food available for small mammals increases in the onset years of these events in the plague areas. The relationship between large increases in rodent populations and large-scale climate variability has been previously shown in many studies,6769 including those conducted in the United States.34,46,70 In some instances, lead times for rodent population increases have been quite short.17 In the case of Sin nombre virus (a type of hantavirus), which, like plague, is a rodent-associated zoonosis,34,70 these short lead times resulted in enhanced virus transmission soon after the period of enhanced precipitation. Rodent densities are known to be associated, through bottom-up processes, with increased precipitation in relatively dry areas where it may be limiting.48,71 Obviously, rodent data would be necessary to confirm the correlation at interannual scales between primary production and hosts abundance; likewise, serological or flea sampling would be necessary to confirm whether increased rodent densities led to higher plague prevalence among hosts populations—a link that may not be simple or direct.72 In the absence of these elements, we can neither accept nor reject with certainty the hypothesis that rodents were part of a trophic cascade that drove the dynamics of plague.

Plague activity also reflects flea dynamics and activity.2,9,28,29 Variations in humidity and temperature indirectly impact flea density and geographical distribution.22,73 Excessively high temperatures reduce flea survival, early-stage development, reproduction rates,21,74 fleas' ability to transmit the disease,74 and thus, overall plague activity, which is expressed in the temperature-modulated cascade hypothesis.30 Metabolic rates are higher and survival is lower when temperature increase or humidity is below a certain threshold. In fact, temperature and humidity together affect fleas.2123,7577 Rodent flea stages occurring off the host (i.e., all stages but fed fleas) are sensitive to variations of the host environment and particularly, to burrow microclimatic conditions.75,78 The macroclimate to which humans are exposed is not necessarily the best indicator of these conditions.79 In burrows, soil moisture could be a better indicator of humidity, because it affects fleas, rather than above-ground measures of humidity.80 Contrary to air (above-ground) and burrow temperatures that exhibit strong coherent correlations, burrow humidity depends more on past rainfall and soil type than on the humidity of the air outside the burrow.80,81 In this study, we have described how seasonal precipitation is instrumental in understanding the association between PDO/ENSO and human plague. Precipitation in winter and early spring mainly falls as snow in the elevated plague areas. Soil moisture is well-predicted by snow accumulation and the timing of the snow melt. The strong positive correlation between winter and early spring precipitation and plague reinforces the hypothesis that the effect of snow cover on summertime soil humidity is crucial for plague activity.

Likely decrease of plague activity in response to increased temperatures associated with climate change.

The West naturally undergoes multidecadal fluctuations between wet and dry periods associated with PDO,82 but previous studies60,61 suggest that the region has been losing snow in favor of relatively more rain and earlier snowmelt across a broad region of mountainous western North America.83 Minimum (i.e., nighttime) temperatures, increasing since 1990, are related to reduced volume and persistence of the snow pack. Although global anthropogenic signals are often difficult to separate from natural climate variability at this regional scale and early stage of climate change, it has recently and convincingly been done for this region.45 The observed regional trends of decreasing snow/rain ratio,83 earlier spring snowmelt,84 and warmer spring and summer seasons84,85 are consistent with climate-model projections. These changes, which are because of both natural and anthropogenic causes,45 influence summertime soil moisture in the western United States. Moreover, warmer temperature can dry the soil directly by enhancing evaporation, whereas dry soil, in turn, promotes warmer daytime temperature. These trends, if they continue and intensify as projected,86 will likely lead to decreased soil humidity and increased temperatures as a result of the direct and indirect (e.g., through hydrology feedbacks) effects of global warming. This could decrease flea survival and reproduction rates74,87 and therefore, reduce the number of active fleas. Global warming is expected to continue and accelerate, which almost undoubtedly will result in increasingly unfavorable conditions for fleas and decreased primary production in the spring and summer; thereby, this will lead to less-favorable conditions for large increases in rodent populations and the occurrences of plague epizootics among these animals. Based on this reasoning and the results of our study, we suggest that human plague in the southwestern United States is likely to decrease over the coming decades. Nevertheless, in places like New Mexico where human-population expansion into the wildland–urban interface is expected to continue and where humans typically contract plague during epizootics in rodents, plague exposure and risk could still increase on a local basis.

Also, this overall decrease in plague can be expected to still be punctuated by episodic outbreaks as a result of natural climate variability for decades to come. At this time, we cannot determine how ENSO and PDO variability or their influences on southwestern climate will change in response to global warming. However, we may reasonably assume that the persistence of the fall/winter PDO and ENSO into the next spring will continue to provide simple probabilistic means for projecting the severity of the next summer's plague season. This study provides information needed for developing probabilistic forecast models of seasonal plague risks. One such model would be based on simple PDO and ENSO persistence as well as long-term temperature trends. If based on the observed/extrapolated values of PDO and ENSO in the fall, the forecast model could involve long lead times, up to three seasons ahead. At shorter lead times (e.g., one season ahead), a more skillful model describing relationships between human plague and climate can be based on observed precipitation patterns (amount and type) and spring temperatures that would implicitly include natural and anthropogenic signals.

Supplementary Material

Figure S1. Raw and density-adjusted plague cases. The grey line is the raw plague cases, and the black line is the county density-adjusted plague cases. The two time series exhibit the same behavior between 1950 and 2005, with increased levels around 1980.
Figure S2. Snow anomaly in millimeters according to Pacific Decadal Oscillation (PDO) and El Niño Southern Oscillation (ENSO) phases. PDO (Left) and ENSO (Right) correspond to the snow anomalies (or precipitation in the winter when the daily temperature is below freezing) for years combining high PDO (alt. low) and El Niño (alt. la Niña) event as defined by above (alt. below) 1 standard deviation (SD) for both indices. Positive anomalies are in red, and negative anomalies are in blue. The size of the circle is proportional to the anomaly calculated as a distance to the mean.
Figure S3. Location of human plague cases every decade. For each decade that the data cover (1950–2005), crosses are placed at the center of the counties. The 1970–1990 period of high plague activity is characterized by enhanced plague activity in the entire western United States. The figure shows periods of low and high plague levels that are widespread over the area, suggesting a priori large-scale climatic influences.
Figure S4. Three subregions. The western US plague area (A) is composed of three main regions, which are (B) the Northern (42–47° N; −125 to −100° W), (C) the Western Pacific (30–42° N; −125 to −115° W), and (D) the Southwestern plague foci (30–42 °N; −115 to −100° W). (B) Time-frequency decomposition of the time series of plague in the three subregion from top to bottom (the Northern, the Western Pacific, and the Southwestern plague foci). (C) Phase analysis of the three subregions for 2- to 3-years periodicity. The Northern, Pacific, and Southwestern foci are in blue, red, and black, respectively.
Figure S5. Bow plot of the latitudes and maximum elevations of plague-infected counties per decade. Although it seems that plague occurred at higher latitude and elevation since 1990, as revealed by the mode and variance of the box plots, both are not significantly different from previous decade distribution. Point data of the location of plague must be more robust to test this particular hypothesis.
Figure S6. Local correlation between winter to spring precipitation and human plague. Positive correlation occurs primarily in the Southwest.
[Supplementary Figures]


The authors thank two anonymous reviewers for their comments on an earlier version of the manuscript. This work was funded by the Marie Curie early training site (EST) program. We thank Mike Begon and Stephen Davis for their comments on earlier versions of the manuscript.


Note: Supplemental figures are available at


Authors' addresses: Tamara Ben Ari, Centre for Ecological and Evolutionary Synthesis, Department of Biology, University of Oslo, Oslo, Norway, and Ecole Normale Superieure, Paris, France. Alexander Gershunov, Climate Research Division, Scripps Institution of Oceanography, University of California, San Diego, La Jolla, CA, E-mail: sasha/at/ Rouyer Tristan, Centre for Ecological and Evolutionary Synthesis, Department of Biology, University of Oslo, Oslo, Norway, E-mail: rouyer.tristan/at/ Bernard Cazelles, Ecole Normale Superieure, Paris, France, E-mail: cazelles/at/ Kenneth Gage, Bacterial Zoonoses Branch, Division of Vector-Borne Infectious Diseases, Centers for Disease Control and Prevention, Fort Collins, CO, E-mail: klg0/at/ Nils Chr. Stenseth, Centre for Ecological and Evolutionary Synthesis, Department of Biology, University of Oslo, Oslo, Norway, E-mail: n.c.stenseth/at/


1. Dispatch MMWR. Human Plague—Four States, 2006. Morbidity and Mortality Weekly Report. 2006. 2006. Available at. Accessed August 25. [PubMed]
2. Gage KL, Kosoy MY. Natural history of plague: perspectives from more than a century of research. Annu Rev Entomol. 2005;50:505–528. [PubMed]
3. Perry R, Fetherston J. Yersinia pestis—etiologic agent of plague. Clin Microbiol Rev. 1997;10:35–66. [PMC free article] [PubMed]
4. Tikhomirov E. In: Plague Manual: Epidemiology, Distribution, Surveillance, and Control. Dennis DT, Gage KL, Gratz NG, Poland JD, Tikhomirov E, editors. Geneva: World Health Organization; 1999. pp. 63–96. (Epidemiology and distribution of plague).
5. McNabb SJ, Jajosky RA, Hall-Baker PA, Adams D, Sharp P, Worshams C, Anderson WJ, Javier AJ, Jones GJ, Nitschke DA, Rey A, Wodajo MS. Centers for Disease Control and Prevention (CDC). Summary of notifiable diseases—United States, 2006. MMWR Morb Mortal Wkly Rep. 2008;21:1–92. [PubMed]
6. Cully JF, Carter LG, Gage KL. New records of sylvatic plague in Kansas. Journal of Wildlife Diseases. 2000;36:389–392. [PubMed]
7. Link VB. A History of Plague in the United States of America. Public Health Monograph. Washington, DC: US Public Health Service; 1955. [PubMed]
8. Adjemian JZ, Foley P, Gage KL, Foley JE. Initiation and spread of travelling waves of plague, Yersinia pestis, in the western United States. Am J Trop Med Hyg. 2007;76:365–375. [PubMed]
9. Barnes AM. Surveillance and control of Bubonic plague in the United States. Symp Zool Soc Lond. 1982;50:257–275.
10. Anderson ET. Plague in the continental United States, 1900–76. Public Health Rep. 1978;93:297–301. [PMC free article] [PubMed]
11. Stenseth NC, Samia NI, Viljugrein H, Kausrud KL, Begon M, Davis S, Leirs H, Dubyanskiy VM, Esper J, Ageyev VS, Klassovskiy NL, Pole SB, Chan KS. Plague dynamics are driven by climate variation. Proc Natl Acad Sci USA. 2006;103:13110–13115. [PubMed]
12. WHO (World Health Organization) Plague. Wkly Epidemiol Rec. 2003;78:253–260.
13. WHO (World Health Organization) Plague. Wkly Epidemiol Rec. 2005;80:138–140. [PubMed]
14. Davis S, Leirs H, Viljugrein H, Stenseth NC, De Bruyn L, Klassovskiy N, Ageyev V, Begon M. Empirical assessment of a threshold model for sylvatic plague. J R Soc Interface. 2005;4:649–657. [PMC free article] [PubMed]
15. Jaksic FM, Lima M. Myths and facts on ratadas: bamboo blooms, rainfall peaks and rodent outbreaks in South America. Aust J Ecol. 2003;28:237–251.
16. Leirs H, Verhagen R, Verheyen W, Mwanjabe P, Mbise T. Forecasting rodent outbreaks in Africa: an ecological basis for Mastomys control in Tanzania. J Appl Ecol. 1996;33:937–943.
17. Lima M, Jaksic FM. Population rate of change in the leaf-eared mouse: the role of density-dependence, seasonality and rainfall. Aust J Ecol. 1999;24:110–116.
18. Lima M, Jaksic FM. Population dynamics of three neotropical small mammals: time series models and the role of delayed density-dependence in population irruptions. Aust J Ecol. 1999;24:25–34.
19. Lima M, Keymer JE, Jaksic FM. El Nino-southern oscillation-driven rainfall variability and delayed density dependence cause rodent outbreaks in western South America: linking demography and population dynamics. Am Nat. 1999;153:476–491.
20. Stapp P, Antolin MF, Ball M. Patterns of extinction in prairie dog metapopulations: plague outbreaks follow El Nino events. Front Ecol Environ. 2004;2:235–240.
21. Krasnov BR, Khokhlova IS, Fielden LJ, Burdelova NV. Development rates of two Xenopsylla flea species in relation to air temperature and humidity. Med Vet Entomol. 2001;15:249–258. [PubMed]
22. Krasnov BR, Khokhlova IS, Fielden LJ, Burdelova NV. Effect of air temperature and humidity on the survival of pre-imaginal stages of two flea species (Siphonaptera: Pulicidae) J Med Entomol. 2001;38:629–637. [PubMed]
23. Gage KL, Burkot TR, Eisen RJ, Hayes EB. Climate and vector borne diseases. Am J Prev Med. 2008;35:436–450. [PubMed]
24. Tripp DW, Gage KL, Montenieri JA, Antolin MF. Flea abundance on black-tailed prairie dogs (Cynomys ludovicianus) increases during plague epizootics. Vector Borne Zoonotic Dis. 2009;9:313–321. [PubMed]
25. Anderson RM, May RM. Infectious Disease of Humans—Dynamics and Control. Oxford, UK: Oxford University Press; 1991.
26. Davis S, Begon M, De Bruyn L, Ageyev VS, Klassovskiy NL, Pole SB, Viljugrein H, Stenseth NC, Leirs H. Predictive thresholds for plague in Kazakhstan. Science. 2004;304:736–738. [PubMed]
27. Hudson PJ, Rizzoli AP, Grenfell BT, Heesterbeek JAP, Dobson AP. The Ecology of Wildlife Disease. Oxford, UK: Oxford University Press; 2002. (Ecology of wildlife diseases).
28. Enscore RE, Biggerstaff BJ, Brown TL, Fulgham RE, Reynolds PJ, Engelthaler DM, Levy CE, Parmenter RR, Montenieri JA, Cheek JE, Grinnell RK, Ettestad PJ, Gage KL. Modeling relationships between climate and the frequency of human plague cases in the southwestern United States. Am J Trop Med Hyg. 2002;66:186–196. [PubMed]
29. Parmenter RR, Yadav EP, Parmenter CA, Ettestad P, Gage KL. Incidence of plague associated with increased winter-spring precipitation in New Mexico. Am J Trop Med Hyg. 1999;61:814–821. [PubMed]
30. Collinge S, Johnson W, Ray C, Matchett R, Grensten J, Cully J, Jr, Gage K, Kosoy M, Loye J, Martin A. Landscape structure and plague occurrence in black-tailed prairie dogs on grasslands of the western USA. Landscape Ecol. 2005;20:941–955.
31. Duplantier J-M, Duchemin J-B, Chanteau S, Carniel E. From the recent lessons of the Malagasy foci towards a global understanding of the factors involved in plague reemergence. Vet Res. 2005;36:437–453. [PubMed]
32. Eisen RJ, Glass GE, Eisen L, Cheek J, Enscore RE, Ettestad P, Gage KL. A spatial model of shared risk for plague and hantavirus pulmonary syndrome in the southwestern United States. Am J Trop Med Hyg. 2007;77:999–1004. [PubMed]
33. Gubler DJ, Reiter P, Ebi KL, Yap W, Nasci R, Patz JA. Climate variability and change in the United States: potential impacts on vector- and rodent-borne diseases. Environ Health Perspect. 2001;109:223–233. [PMC free article] [PubMed]
34. Kolivras KN, Comrie AC. Climate and infectious disease in the southwestern United States. Prog Phys Geogr. 2004;28:387–398.
35. Kovats RS, Bouma MJ, Hajat S, Worrall E, Haines A. El Nino and health. Lancet. 2003;362:1481–1489. [PubMed]
36. Nakazawa Y, Williams R, Peterson AT, Mead P, Staples E, Gage KL. Climate change effects on plague and tularemia in the United States. Vector Borne Zoonotic Dis. 2007;7:529–540. [PubMed]
37. Mantua NJ, Hare SR. A Pacific interdecadal climate oscillation with impacts on salmon production. Bull Am Meteorol Soc. 1997;78:1069–1079.
38. McCreary JP. In: Oceans and Atmospheres. Willebrand J, Anderson DLT, editors. 1986. pp. 247–280. (Coupled ocean–atmosphere model of El Niño and the Southern Oscillation).
39. Holmgren M, Scheffer M, Ezcurra E, Gutiérrez JR, Mohren GMJ. El Niño effects on the dynamics of terrestrial ecosystems. Trends Ecol Evol. 2001;16:89–94. [PubMed]
40. Ben Ari T, Gershunov A, Gage KL, Snäll T, Ettestad P, Kausrud KL, Stenseth NC. Human plague in the USA: the importance of regional and local climate. Biol Lett. 2008;4:737–740. [PMC free article] [PubMed]
41. Diaz HF. In: El Niño: Historical and Paleoclimatic Aspects of the Southern Oscillation. Diaz HF, Markgraf V, editors. Cambridge, UK: Cambridge University Press; 1992. pp. 7–28. (Atmospheric teleconnections associated with the extreme phase of the Southern Oscillation).
42. Gershunov A, Barnett TP. Interdecadal modulation of ENSO teleconnections. Bull Am Meteorol Soc. 1998;79:2715–2725.
43. Hallett TB, Coulson T, Pilkington JG, Clutton-Brock TH, Pemberton JM, Grenfell BT. Why large-scale climate indices seem to predict ecological processes better than local weather. Nature. 2004;430:71–75. [PubMed]
44. Stenseth NC, Mysterud A. Weather packages: finding the right scale and composition of climate in ecology. J Anim Ecol. 2005;74:1195–1198.
45. Barnett TP, Pierce DW, Hidalgo HG, Bonfils C, Santer BD, Das T, Bala G, Wood AW, Nozawa T, Mirin AA, Cayan DR, Dettinger MD. Human-induced changes in the hydrology of the western United States. Science. 2008;319:1080–1083. [PubMed]
46. Stapp P. Rodent communities in active and inactive colonies of black-tailed prairie dogs in shortgrass steppe. J Mammal. 2007;88:241–249.
47. Holmgren M, Scheffer M, Ezcurra E, Gutiérrez JR, Mohren GMJ. El Niño effects on the dynamics of terrestrial ecosystems. Trends Ecol Evol. 2001;16:89–94. [PubMed]
48. Letnic M, Tamayo B, Dickman CR. The responses of mammals to La Nina (El Nino Southern Oscillation)-associated rainfall, predation, and wildfire in central Australia. J Mammal. 2005;86:689–703.
49. Forrest GH, Collatz G, Los S, Brown de Colstoun E, Landis D. ISLSCP Initiative II Data Archive. NASA; 2005.
50. Cazelles B, Chavez M, McMichael AJ, Hales S. Nonstationary influence of El Niño on the synchronous dengue epidemics in Thailand. PLoS Med. 2005;2:313–318. [PMC free article] [PubMed]
51. Cazelles B, Hales S. Infectious diseases, climate influences and nonstationary. PLoS Med. 2006;3:1212–1213.
52. Hastings A. Transient dynamics and persistence of ecological systems. Ecol Lett. 2001;4:215–220.
53. Stenseth NC, Falck W, Chan K-S, Bjørnstad ON, O'Donoghue M, Tong H, Boonstra R, Boutin S, Krebs CJ, Yoccozaj NG. From patterns to processes: phase and density dependencies in the Canadian lynx cycle. Proc Natl Acad Sci USA. 1998;95:15430–15435. [PubMed]
54. Cazelles B, Chavez M, Magny GC, Guégan J-F, Hales S. Time-dependent spectral analysis of epidemiological time-series with wavelets. J R Soc Interface. 2007;4:625–636. [PMC free article] [PubMed]
55. Cazelles B, Chavez M, Berteaux D, Ménard F, Vik J, Jenouvrier S, Stenseth N. Wavelet analysis of ecological time series. Oecologia. 2008;156:287–304. [PubMed]
56. Daubechies L. Ten Lectures on Wavelets. SIAM; 1992. Available at.
57. Lau K, Weng H. Climate signal detection using wavelet transform: how to make a time series sing. Bull Am Meteorol Soc. 1995;76:2391–2402.
58. Torrence C, Compo GP. A practical guide to wavelet analysis. Bull Am Meteorol Soc. 1998;79:61–78.
59. Rouyer T, Fromentin J-M, Stenseth NC, Cazelles B. Analysing multiple time series and extending significance testing in wavelet analysis. Mar Ecol Prog Ser. 2008;359:11–23.
60. Favre A, Gershunov A. North Pacific cyclonic and anticyclonic transients in a global warming context: possible consequences for western North American daily precipitation and temperature extremes. Clim Dynam. 2009;32:969–987.
61. Gershunov A, Cayan DR. Heavy daily precipitation frequency over the contiguous United States: sources of climatic variability and seasonal predictability. J Clim. 2003;16:2752–2765.
62. Kausrud KL, Viljugrein H, Frigessi A, Begon M, Davis S, Leirs H, Dubyanskiy V, Stenseth NC. Climatically driven synchrony of gerbil populations allows large-scale plague outbreaks. Proc R Soc Lond B Biol Sci. 2007;274:1963–1969. [PMC free article] [PubMed]
63. Yates T, Mills J, Parmenter C, Ksiazek T, Parmenter R, Vande Castle J, Calisher C, Nichol S, Abbott K, Young J, Morrison M, Beaty B, Dunnum J, Baker R, Salazar-Bravo J, Peters C. The ecology and evolutionary history of an emergent disease: hantavirus pulmonary syndrome. Bioscience. 2002;52:989–998.
64. Fuller DO. Trends in NDVI time series and their relation to rangeland and crop production in Senegal, 1987–1993. Int J Remote Sens. 1998;19:2013–2018.
65. De la Maza M, Lima M, Meserve P, Gutierrez J, Jaksic F. Primary production dynamics and climate variability: ecological consequences in semiarid Chile. Glob Change Biol Bioenergy. 2009;15:1116–1126.
66. Holt A, Salkeld D, Fritz C, Tucker J, Gong P. Spatial analysis of plague in California: niche modeling predictions of the current distribution and potential response to climate change. Int J Health Geogr. 2009;8:1–14. [PMC free article] [PubMed]
67. Jaksic FM. Ecological effects of El Nino in terrestrial ecosystems of western South America. Ecography (Cop.) 2001;24:241–250.
68. Meserve PL, Yunger JA, Gutiérrez JR, Contreras LC, Milstead WB, Lang BK, Cramer KL, Herrera S, Lagos VO, Silva SI, Tabilo EL, Torrealba M-A, Jaksic FM. Heterogeneous responses of small mammals to an El Niño Southern Oscillation Event in north central semiarid Chile and the importance of ecological scale. J Mammal. 1995;76:580–595.
69. Davalos VA, Torres MA, Mauricci CO, Laguna-Torres VA, Chinarro MP. Outbreak of bubonic plague in Jacocha, Huancabamba, Peru. Rev Soc Bras Med Trop. 2001;34:87–90. [PubMed]
70. Engelthaler DM, Mosley DG, Cheek JE, Levy CE, Komatsu KK, Ettestad P, Davis T, Tanda DT, Miller L, Frampton JW, Porter R, Bryan RT. Climatic and environmental patterns associated with hantavirus pulmonary syndrome, four corners region, United States. Emerg Infect Dis. 1999;5:87–94. [PMC free article] [PubMed]
71. Meserve PL, Milstead WB, Gutierrez JR. Results of a food addition experiment in a north-central Chile small mammal assemblage: evidence for the role of “bottom-up” factors. Oikos. 2001;94:548–556.
72. Davis S, Calvet E, Leirs H. Fluctuating rodent populations and risk to humans from rodent-borne zoonoses. Vector-Borne Zoonotic Dis. 2005;5:305–314. [PubMed]
73. Lindsay LR, Galloway TD. Seasonal activity and temporal separation of four species of fleas (Insecta: Siphonaptera) infesting Richardson's ground squirrels, Spermophilus richardsonii (Rodentia: Sciuridae), in Manitoba, Canada. Can J Zool. 1997;8:1310–1322.
74. Cavanaugh DC. Specific effect of temperature upon transmission of the plague bacillus by the oriental rat flea, Xenopsylla Cheopis. J Trop Med Hyg. 1971;20:264–273. [PubMed]
75. Krasnov B, Khokhlova I, Fielden L, Burdelova NV. The effect of substrate on surival and development of two species of desert fleas (Siphonaptera: Pulicidae) Parasite. 2002;9:135–142. [PubMed]
76. Krasnov BR, Burdelova NV, Shenbrot GI, Khokhlova IS. Annual cycles of four flea species in the central Negev desert. Med Vet Entomol. 2002;16:266–276. [PubMed]
77. Eisen RJ, Gage KL. Adaptive strategies of Yersinia pestis to persist during inter-epizootic and epizootic periods. Vet Res. 2009;40:1. doi:10.1051/vetres:2008039. [PMC free article] [PubMed]
78. Longanecker DS, Burroughs AL. Sylvatic plague studies. 9. Studies of the microclimate of the California ground squirrel burrow and its relation to seasonal changes in the flea population. Ecology. 1952;33:488–499.
79. Davis DH. Plague in South Africa: a study of the epizootic cycle in gerbils (Tatera brantsi) in the northern Orange Free State. J Hyg (Lond) 1953;51:427–449. [PMC free article] [PubMed]
80. Osacar-Jimenez JJ, Lucientes-Curdi J, Calvete-Margolles C. Abiotic factors influencing the ecology of wild rabbit fleas in north-eastern Spain. Med Vet Entomol. 2001;15:157–166. [PubMed]
81. Hall LO, Myers K. Variations in the microclimate in rabbit warrens in semi-arid New South Wales. Aust J Ecol. 1978;3:187–194.
82. Colorado River Basin Water Management Evaluating and Adjusting to Hydroclimatic Variability. Washington, DC: National Academy of Sciences; 2007.
83. Knowles N, Dettinger MD, Cayan DR. Trends in snowfall versus rainfall in the western United States. J Clim. 2006;19:4545–4559.
84. Cayan DR, Kammerdiener SA, Dettinger MD, Caprio JM, Peterson DH. Changes in the onset of spring in the western United States. Bull Am Meteorol Soc. 2001;82:399–415.
85. Alfaro EJ, Gershunov A, Cayan D. Prediction of summer maximum and minimum temperature over the central and western United States: the roles of soil moisture and sea surface temperature. J Clim. 2006;19:1407–1421.
86. IPCC The Physical Science Basis—Summary for Policymakers. Contribution of Working Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change. 2007. Available at.
87. Thomas RE. In: Biology of Disease Vectors. Beaty BJ, Marquardt WC, editors. 1996. pp. 146–159. (Fleas and the agents they transmit).

Articles from The American Journal of Tropical Medicine and Hygiene are provided here courtesy of The American Society of Tropical Medicine and Hygiene