Dendrochronological series are reliable sources of information to analyze past hydrological and climatological variation that provides useful information for the management of water resources within basins. We analyzed dendrochronological series obtained from the upper Mayo River Basin using principal components analysis to determine a common climatic signal. Although the complete series extended for over 350 years, the representative period common to all series was from 1750 to 2014 (265 years) with an expressed population signal of over 0.85. Climate data (precipitation and temperature) were collected from the North American Land Data Assimilation System 2 model of the Land Data Assimilation System and hydrometrics records were obtained from the National Commission of Water in Mexico. The results of the response function showed an association of mean monthly temperature with the ring width series for the months of December of the previous year, May and October of the year of growth, and seasonally from January to July (r = -0.75, n = 36, p < 0.05). A significant response to rainfall of earlywood growth was observed for June, November, and December of the previous year, January and February of the growth year, and seasonally, from October of previous year to May of the current growth year (r = 0.70, n = 35, p < 0.05). Significant association also was found between earlywood and the accumulated runoff from October of the previous year to May of the growth year, which was used for reconstruction of the runoff flow between 1750 and 2014, which showed evidence of decadal drought. Significant correlation was found between the reconstructed runoff series, and the Southern Oscillation index (r = -0.42, n = 228, p < 0.05), but not significant with the Pacific Decadal Oscillation index (r = 0.16, n = 115, p < 0.05). We also observed significant (p < 0.05) associations with the drought indices Palmer Drought Severity Index and Standardized Precipitation Evapotranspiration Index (r = 0.56, r = 0.41, respectively). Our study demonstrated the potential of assimilated data for dendroclimatic reconstructions and the feasibility of generating hydroclimatic information of extreme events that have not been recorded in the available climatic and hydrological instrumental records.
A serious problem for studying regional hydroclimatic variation is the limited time period of records, usually not exceeding 70 years of data, and the lack of adequate climatic and hydrometric information (
The understanding of the historical behavior of hydroclimatic variability is essential to determine trends and fluctuations through time, which in turn can provide information for the design of optimal and integrated natural resource management plans and forecasts, such as determining water volumes for agriculture, anticipating the frequency of wildfires due to temperature fluctuations, or developing climate models with a high predictive capability. All this knowledge is essential to implement strategies for adaptation to climate variability and for the mitigation of its impacts (
Tree growth rings are one of several existing proxies to infer or reconstruct historical patterns of environmental variables because their growth is influenced by the environmental conditions occurring during their formation, they have annual resolution, and their temporal coverage is several centuries, which allows a better understanding of ancestral climate (
The objective in this study was to develop a response function between dendrochronological series from the upper MRB and climate variables (precipitation and temperature) from the assimilated data in the NLDAS-2 model, and to reconstruct historical runoff volumes to determine past water availability, the occurrence of extreme hydroclimatic events, and the influence of general atmospheric circulatory patterns.
The Mayo River Basin (MRB), located in the southern portion of Sonora, Mexico, is ecologically and economically important due to its orographic, biological, and hydrological characteristics. It is a biological corridor that harbors a rich biodiversity which varies with the altitudinal gradient. It is also a zone of surface water and groundwater production that supplies the Mayo River 038 Irrigation District, which irrigates 96,914 ha (
The MRB has an area of 27,541 km2 extending within the Southern Sonora region (from 28° 26′ 01.21″ to 26° 39′ 12.9″ N, and from 109° 56′ 23.05″ to 107° 57′ 25.42″ W), and forms part of the Hydrological Region 9 (
The predominant climate in the basin varies from arid to semi-arid and from warm to temperate dry (
The vegetation communities in the MRB are distributed along the altitudinal gradient between the coast and the mountains and include coastal dunes, mangrove thickets, mesquite scrub, riparian vegetation, tropical low deciduous forest, induced grassland, and in the higher elevations, oak and pine forests (
The runoff volumes were obtained from the San Bernardo hydrometric station, located upstream from the Adolfo Ruiz Cortines Dam; there are more hydrometric stations inside the MRB watershed, but the San Bernardo hydrometric station was selected because it has complete data from 1960 to 2014, and the location is near the natural flow of the Mayo River for more accurate reconstruction.
The dendrochronological series for the MRB were downloaded from the International Tree-Ring Data Bank (http://gis.ncdc.noaa.gov/kml/paleo_tree.kmz -
To obtain a regional series representative of the dominant climate conditions in the basin, we used principal components analysis (PCA) for the four dendrochronological series (
The mean temperature (°C) and precipitation (mm per month) values were downloaded from the NLDAS-2 model as National Oceanic and Atmospheric Administration meteorological data in gridded binary format, which is based on a numerical meteorological prediction model of the Global Forecast System. The spatial resolution of the data is 0.125° and covers the years 1979 to the present. However, the spatial extent of the data is limited and only covers a part of northern Mexico that includes the MRB.
The climate variables of the NLDAS-2 model were downloaded in gridded binary format and the monthly data were extracted using ArcGIS® ver. 10.3 (ESRI, Redlands, CA, USA). The mean monthly precipitation and temperature values were overlaid onto the layer with the study area and the pixels converted to points (vector) to generate a grid and extract the data.
The association of the variables runoff, precipitation, and temperature was established by simple correlations (
The analysis of drought severity was made using the median hydrological reconstructions of the Mayo River (50% probability of occurrence), considering dry years as those with runoff values below the median. From the hydrological perspective, there are two parameters for categorizing droughts, their duration and their severity (
The duration of drought (
where
where
where
The resulting reconstruction was associated with general atmospheric circulation phenomena by comparing it to the El Niño-Southern Oscillation (ENSO) index through the Southern Oscillation Index (SOI -
The reconstructed dendrochronological series were correlated with drought indices to understand how historic data influenced meteorological variables in the series. The indices used were obtained from the North American Drought Atlas self-calibrated Palmer Drought Severity Index (scPDSI, instrumental summer calibration -
Significant associations of global atmospheric circulation phenomena were determined by spectral wavelet coherence analysis using the biwavelet library of the “dplR” package in R (
The dendrochronological series available for the MRB were generated from the species
Considering the common period (years where all series data overlap in time: 1644-2002,
According to the matrix correlation analysis of the four series, the highest and more significant (p < 0.05) associations were found between the sites Bisaloachic (BIS) and Tutuaca (TUT - r = 0.83) separated by 10 km, followed by that between Cebadillas de Ocampo (CDO) and BIS (r = 0.63), separated by 62 km, while the lowest correlation was found between the Barrancas del Cobre (COB) and CDO (r = 0.36) series, separated by 50 km (
To obtain a climatic series that was representative of the MRB over the four dendrochronological series (BIS, COB, CDO, and TUT), PCA was used for the 359-year common period (Fig. S1 in Supplementary material). The PCA showed the stdRW value of the COB series in the first component (PC1) explained 66.21% of the total variance (p < 0.05), and the BIS, CDO, and TUT series in the second component (PC2) accounted for 14.25% of the variance. This result indicates that the COB series responds to a different aspect of climate, thus it is useful to represent the MRB climate variability. Therefore, the COB series was correlated with the climatic and hydrometric data for the analysis of the response function with precipitation and temperature and used for the reconstruction of runoff.
The response function graph (
The correlation between stdRW in site COB and the monthly and seasonal records of mean temperature of the NLDAS-2 model was significant for all months, but weak in December of the previous year, and October and December of the year of growth. The best correlation was for the January-July seasonal period (r = -0.75, n = 36, p < 0.05), which explained 56.2% of the total seasonal variation. A negative correlation suggests that at a higher temperature there was less annual radial growth of
This response is common to many forest tree species, in particular those susceptible of water stress (
A significant association was observed between the accumulated monthly precipitation and the standard earlywood chronology of the COB site for the months of June, November, and December of the previous year and February of the year of growth (
The significant response to winter precipitation with earlywood growth is influenced by the warm phase of the ENSO (
As it is characteristic in northern Mexico, the winter precipitation in the MRB is one fourth of the total annual volume. The winter season has a high interannual variability, although the presence of north winds favors winter rain and lowers the temperature in the Sonora region by increasing the incursion of polar air masses (
The data of runoff volume (in thousands of square meters) recorded by the San Bernardo hydrometric station was normalized through a log-normal transformation and compared with the earlywood series (stdEW) to improve its correlation (
This seasonal association at the end of the North American Monsoon System (NAMS -
To validate the resulting reconstruction model, the recorded runoff data were divided into two sub-periods, one used for calibration (1961-1987), and another one for verification (1988-2014) using the “verify” subroutine of the Dendrochronological Program library (
where
Because the reconstruction was made by normalizing the runoff volumes (the antilogarithm was used to transform the values to 103 m3), a reconstruction that showed extraordinary values present from 1664 to 1749 may not reflect true conditions because of the reduced number of growth cores in that period. Therefore, the reconstruction was limited to the 265-year period between 1750 and 2014, during which the expressed population signal value was 0.85 (
The reconstruction showed high interannual and multiannual variation and periods of extraordinary drought were observed with flow volumes under one half of the mean standard deviation of the median (
The spectral analysis showed significant (p < 0.05) frequencies of 2.1, 2.3, and 2.8 years in the earlywood series of the COB site (
Among the associations with different indices of general atmospheric circulation (ENSO, PDO) and drought severity (PDSI, SPEI -
A significant association between drought severity indices and the winter flow volumes was observed for the analyzed periods. Because the correlations found between PDSI (r = 0.56, n = 253, p < 0.05) and SPEI (r = 0.41, n = 59, p < 0.05; an index associated with drought based on precipitation and evapotranspiration -
The results of the wavelet coherence analysis between the reconstruction of winter runoff volumes and the ENSO series index during the 1750-1978 period (
The drought events recorded in the MRB are coincident with episodes recorded for most of northern and central Mexico, having variable intensity between regions, which conforms to the idea that climate is strongly modulated by common climate phenomena with a periodicity of droughts of approximately 50-60 years (
The decade of the 1950s had several years of below average observed and reconstructed runoff, and was characterized by several strong La Niña events. This decade presented a negative phase of PDO (
The relationship between PDO and reconstructed runoff presented a lesser but still significant correlation than the ENSO, which means the PDO affects the runoff and the precipitation in the MRB. As seen in other studies, the positive warm phase of PDO increases the winter rainfall in northern Mexico and southwestern USA (
The NAMS is the atmospheric circulation phenomenon that favors summer rains between the months of June and July, sometimes extending to September, generating a convergence of warm and humid air from the Pacific Ocean and the Gulf of Mexico. The arrival of these converging currents propagate about 75% of the annual precipitation on the windward slope of the Sierra Madre Occidental, but the interannual variability of precipitation is of lesser magnitude (
In the MRB, the contribution of rainfall from the influence of tropical cyclones and hurricanes generated in the Pacific Ocean - with maximum activity recorded in September and October - appears to be the main cause of the reconstructed high volumes of runoff during years in which tropical cyclones occurred, such as 1971 (Storm Katrina), 1977 (Hurricane Doreen), 1979 (Typhoon Tip), and 2001 (Hurricane Juliette -
The results from our study determined associations between earlywood growth response and records of precipitation and temperature. We observed significant correlation between precipitation from October to April and the standardized earlywood series in COB; that is, earlywood growth responds to the water from winter rainfall that is stored in the soil. Also, we found a significant negative correlation between the standardized ring width and the mean monthly temperature between January and July, meaning that higher mean monthly temperatures inhibit annual radial growth. This result is also shown in other studies around the world where the rise in temperature due to global warming reduces growth. The reconstruction of seasonal runoff from October to May in the MRB spanned the years 1750-2014 and indicated the occurrence of extraordinary drought that had not been recorded in the instrumental records, which cover the last 55 years only.
The climate change trend might exacerbate the availability of water in the MRB, which would have severe social and economic impacts. it is therefore advisable to establish management plans that take into consideration the occurrence of long severe droughts not evidenced by meteorological records but observed in historical reconstructions. Broad-impact climate phenomena like ENSO and NAMS have a strong influence on water availability in northern Mexico. Although we found no significant correlation between the dendrochronological series and the NAMS, summer precipitation produces the highest runoffs in this region of Mexico and is responsible for dam filling. While the winter runoff accounts for only 25% of the total annual volume, winter rain is highly significant in the radial growth of trees and allows the analysis of the significant influence of the association with the ENSO, considered as better defined in North America.
The use of modern technologies to access climate information represents a high-resolution alternative to obtain climate variable series through the use of assimilated data, as in the specific case of the NLDAS-2. Using such technology, we were able to establish that the earlywood series are significantly correlated to the winter-spring seasonal precipitation, making it possible to model and predict volumes of biomass that could be produced by forests.
The use of assimilated data is still in the initial stages in Mexico and its broader application could, in the short term, allow the production of better prediction models using the available dendrochronological series, and the development of more reliable hydroclimatological reconstructions, which can help to achieve a more sustainable management of forest resources and of the runoff volumes generated in basins, which are essential for food production in this region of northern Mexico.
This study was carried out through funds provided by CONACyT (
Geographic location of Mayo river basin, dendrochronological series (COB: Barrancas del Cobre; CDO: Cebadillas de Ocampo; TUT: Tutuaca; BIS: Bisaloachic), and hydrometric station San Bernardo in Sonora, Mexico.
(a) Dendrochronological series of stdRW. (b) Correlation matrix between the four dendrochronological series.
Climatic response between the dendrochronological serie COB standard version and the precipitation and temperature records. The values out of the horizontal dotted lines are significant (95%).
(a) Observed records at the hydrometric station in San Bernardo
(a) Spectral analysis of early wood series. (b) Wavelet coherence analysis between seasonal runoff volume reconstructed and ENSO index.
Dendrochronological series in the Mayo River Basin. (RW): RW ring width; (EW): earlywood; (LW): latewood.
No. | Site | Code | Period (no. of years) | Species | Chronologytype |
---|---|---|---|---|---|
1 | Barrancas del Cobre | COB | 1644-2014 (371) |
|
RW, EW, LW |
2 | Bisaloachic | BIS | 1537-2002 (466) |
|
RW, EW, LW |
3 | Cebadillas de Ocampo | CDO | 1588-2002 (415) |
|
RW, EW, LW |
4 | Tutuaca | TUT | 1534-2012 (479) |
|
RW, EW, LW |
Validation of the seasonal runoff volume reconstruction model by comparison of the observed dataset and the reconstructed dataset based on tree-ring series. (1): No significance test is available for this statistics, but any positive result indicates that the reconstruction contributes unique paleoclimatic information (
Statistics | Period | |
---|---|---|
1961-1987 | 1988-2014 | |
Pearson’s correlation | 0.75 * | 0.73 * |
Reduction of error (1) | 0.81 | -0.07 |
3.11 * | 3.84 * | |
Non-parametric signs test | 7 * | 5 * |
First negative difference (2) | 1 * | 7 * |
Hydrological drought events for the runoff volume reconstruction in the Mayo River Basin.
Period | Duration(No. years) | Severity(× 103 m3) | RelativeSeverity | Type ofDrought |
---|---|---|---|---|
1752 - 1755 | 4 | -437,602.87 | 0.39 | Intense |
1840 - 1843 | 4 | -295,973.61 | 0.27 | Intense |
1891 - 1894 | 4 | -438,377.06 | 0.40 | Intense |
1909 - 1912 | 4 | -335,209.08 | 0.30 | Intense |
2011 - 2014 | 4 | -397,945.93 | 0.36 | Intense |
1772 - 1776 | 5 | -499,780.22 | 0.45 | Intense |
1953 - 1957 | 5 | -507,516.29 | 0.46 | Intense |
1959 - 1965 | 7 | -421,794.46 | 0.38 | Intense |
1757 - 1768 | 12 | -1,109,510.97 | 1.00 | Extraordinary |
Association between general atmospheric circulation phenomena, drought index and reconstructed runoff volume. (SOI): Southern Oscillation Index reconstructed (
Index | Period (years) | |
---|---|---|
SOI | 1750-1977 (228) | -0.42 (0.000*) |
PDO | 1900-2014 (115) | 0.16 (0.073 ns) |
PDSI | 1750-2002 (253) | 0.56 (0.000*) |
SPEI | 1951-2009 (59) | 0.41 (0.001*) |
Fig. S1 - Principal Component Analysis with the four dendrochronological series of stdRW in the Mayo river basin.
Fig. S2 - (a) Response function analysis between the tree-ring chronology COB stdEW and monthly runoff. (b) Relationship between Log-Normal seasonal runoff October-May and stdEW series of COB.
Fig S3 - Relationship between reconstructed runoff and (a) SOI rec, (b) PDO, (c) PDSI, (d) SPEI.