Introduction

The production of water resources is one of the most important assets of forest ecosystems, particularly when they depend on the sustainability and economic development of a society. Runoff produced in the watersheds ofthe tributaries ofthe rivers Tunal-San Pedro, Santiago Bayacora and Sauceda at the hydrological region No. 11, which eventually drain into the Mezquital river are stored in dams for irrigation in the area known as Valle del Guadiana. In the case of River Tunal, this comes from the top of the particular properties Culebras and Cuevecillas of the municipality Pueblo Nuevo in the Sierra Madre Occidental.

The Tunal River is of great economic importance because its waters are stored in the dam Guadalupe Victoria and released into the Mezquital river to irrigate the Valle de Guadiana; a similar case occurs with Sauceda River (also a tributary of the Mezquital River), which originates in the Sierra de Cacaria and whose volumes are regulated for the purpose of irrigating the dam Peña de Águila. The irrigated surface runoff-water in the Valle del Guadiana is close to 15 000 ha, out of which 3000 ha are irrigated with water from the dam Santiago Bayacora, 9000 ha with Guadalupe Victoria and 3 000 ha with the dam Peña de Águila(^{SEMARNAT, 2009}). This irrigated area fluctuates over time as a function of the climate variability that identify this hydrologic region (^{Gómez et al, 2012}).

The limited availability of climate and hydrometric data in this region, severely limit a detailed hydroclimatic historical behaviour analysis and its possible performance over time. This information is of basic importance to plan the use of water in the Valle del Guadiana region where the necessary agricultural goods are produced to sustain the population of the city of Durango and whose aquifer fed by those tributaries, extracting water used for the purpose of human and industrial consumption, and just like other aquifers in northern Mexico have problems of overfishing and water quality (CNA, 2009; ^{Chávez Soto, 2010}; ^{INEGI, 2012}).

The aim of this study was to use the annual growth of some conifers present in the area where the runoffs feed the dam Valle del Guadiana as an indirect method to determine the historical fluctuations of precipitation and registered flows in nearby hydrometric stations to the storage dams.

Materials and methods

Methodology

In order to quantify the hydroclimatic variation in the runoff producing basins conforming the dams that supply the Valle del Guadiana; growth nuclei of tree species were collected in several places of mixed coniferous forest in the village of Pueblo Nuevo, Durango. Chronologies previously developed in the region were used for comparative purposes to define the extent of the influence of weather events and the potential impact of weather patterns in the radial increment of the main tree species (Table 1, Figure 1).

Table 1 General characteristics of the study sites. 

Figure 1 Geographical distribution of sampling sites and sites with existing dendrochronological series in the upper basin of the river Mezquital. 

At each site, ancient trees were selected and were obtained from three to four radial increments with a drill Pressler 0.5 mm internal diameter and lengths of 35-50 cm and cross sections of dead trees to facilitate and achieve dated chronologies extend over time. Samples were dated by conventional dendrochronological technics (^{Stokes and Smiley, 1968}; ^{Fritts, 1976}). Each individual growth was divided into growth bands of early-wood (EW), late-wood (LW) and ring-whole (RW) and measured with Velmex system (^{Robinson and Evans, 1980}). Co-dating, climate response quality and accuracy of measurement of each ring was observed with the program COFECHA (^{Holmes, 1983}; ^{Grissino-Mayer, 2001}).

Biological trends unrelated to climate were removed withArstan program (^{Cook and Holmes, 1986}), generating a standardized chronology for that locality (^{Fritts, 1976}), of which three versions ofit are made, Standard, the Residual andArstan (^{Cook, 1987}). In orderto analyse the common climate response between chronologies, simple correlations were used, and through response function analysis, the climate response was determined of the chronologies with annual, and seasonal precipitation and with runoff volumes recorded in hydrometeorological stations distributed in adjacent sites of the capturing dams. Also, the influence of phenomena of general circulation was determined, which explain much of the hydroclimatic variability observed at the regional level (^{Méndez and Magaña, 2010}). The association with these circulatory phenomena was performed by comparing the representative rates of dendrochronological índices ofthe upperbasin of the Mezquital river with indices of ENSO, a phenomenon that significantly impacts northern Mexico (^{Stahle et al, 1998}; ^{Cleaveland et al., 2003}; ^{Villanueva et al, 2009 Cerano et al., 2011}).

The developed dendroclimatic reconstructions were verified as far as possible with previous reconstructions and historical documents related to the production of food and other social and economic impacts described for the region (^{Endfield and Fernández, 2006}).

Results

Extension of chronologys

The dendrochronological series developed in this study ranged in length; Molinillos (MOL), was the shortest and covers the period 1945-2012, (68 years), Sierra del Nayar (SNA) to 1810-2012 (203 years), Cuevecillas-Culebras (CUC), 1770 to 2012 (243 years). The chronology for the site Bayas (BAY) (^{González-Elizondo etal., 2005}) ranged from 1655-2001 (347 years) and the San Pedro-Mezquital (RSPM) river, 1113-2011 (899 years) (^{Stahle, 2011}, Com. Pers.).

Correlations between sets, considering the Sierra del Nayar (SNA) were significant (r >0.5, p <0.01), especially for those located within the upper watershed of the Mezquital River (CUC, BAY) and at sites with similar elevations in the Western Sierra Madre (Figure 2). In order to analyse the hydroclimatic fl uctuations affecting the availability of water in Valle del Guadiana were considered only those chronologies from the upper basin of the Mezquital River, which drains directly into this valley.

Figure 2 Association between dendrochronological series in the Sierra Madre Occidental, based on the Sierra del Nayar (SNA). 

Molinillos

Within this Property, there were no ancient individuals found, due to previous uses ofthe species reported since 1986 (^{Smart Wood, 2002}), so the chronologies EW, LW and RW of Pinus durangensis only covered the period from 1945 to 2012. The response function analysis showed a significant association between the rate ofearly-wood and the integrated regional precipitation by the weather stations Charco Verde (23.967° north latitude, 105,450° west longitude, 2163 m of elevation), Llano grande (23,883° north latitude, 105.2° west longitude, 2406 m ofelevation), Navios Viejos (23.833 north latitude and 105.0° west longitude, 2492 m of elevation) and "Pueblo Nuevo" (23.383° north latitude, 105.367° west longitude, 1586 m of elevation) for the period 1963-1990, with a correlation of0.741 (p< 0.00), a situation that implies that the annual growth of early-wood of P durangensis on this site, explains 55% of variation in precipitation for this region, relevant information to determine the fluctuations in rainfall variability and possible trends in the short and medium term.

With respect to the late-wood, we found a significant relationship between the rate oflate-wood and accumulated precipitation for the period from June to September, a time where it occurs between 75-80% ofannual precipitation and therefore, where most of the runoff is generated, which are captured in the dams located in the Valle del Guadiana. The correlation observed between both variables was 0.785 for the period 1963 to 1990 (p< 0.00, n= 28), which accounts for about 62% of the variation in precipitation in the age quality of year.

In order to analyse the relationship between dendrochronological indices and runoff recorded in neighbouring hydrometric stations at the dams, the rate of early-wood of the dendrochronological series ofPinus durangensis was compared with the standard volumetric data ofthe hydrometric station "El Pueblito", located 750 m downstream of the dam Guadalupe Victoria and is used to irrigate an area of 9 000 ha in the Valle del Guadiana. The relationship obtained between EWI and the annual volume recorded was 0.72, for 1965-2000 (p< 0.000), implying that early-wood explains 52% of the volumes released into the dam.

Although, individual dendrochronological series were developed for the sites Sierra del Nayar and Cuevecillas-Culebras by their proximity (no more than 20 km apart in a straight line) and their common climate response, the measurement series of Psedudotsuga menziesii developed separately were integrated into a single database to generate a new chronology (CUC-SNA), covering the period 1770-2012 (243 years) which is more representative of the dominant climatic conditions the upper basin of the Mezquital river.

The regional series ofearly-wood responded significantly (r= 0.72, 1961-1991, n= 31 years, p< 0.00) to the seasonal rainfall from January to May, which covers part of the winter-spring season. Even when rainfall occurs during this period represents only 10 to 14% ofthe annual total, this precipitation is ofgreat importance to the radial increment ofconifers in the northern and central Mexico (^{Cleaveland et al, 2003}; ^{Villanueva et al, 2005}) and southwestern United States (^{Griffin et al, 2011}) and provides information on hydroclimatic variability that occurs in this period and is influenced by phenomena of general circulation (^{Stahle et al, 1998}).

The precipitation record period available (1961-1991) was divided into two sub-periods with calibration and verification purposes. Since both sub-periods were significant, the total period of available data was used to generate the reconstruction model, which consists of a simple regression equation, which is:

Where: Y_t is the reconstructed precipitation for January-May (mm); Xt: early-wood index of regional chronology.

The reconstructed precipitation showed high annual and multi-year variation, dry periods are observed in the decades 1790-1800, 1820, 1860-1890, 1910, 1950, 1970 and 2000. Wet periods were less frequent and were found in the 1830's, 1850, 1920, 1940, 1980 and 2010 (Figure 3).

Figure 3 Reconstruction of seasonal precipitation from January to May for the upper basin of El Mezquital with a regional chronology of early-wood of Pseudotsuga menziesii. 

The regional series of late-wood Pseudotsuga menziesii showed a positive association (r= 0.605, 1963-2003, p< 0.01) with summer precipitation; however, a better response was obtained with the series of late-wood of the site Sierra del Nayar, with a correlation of 0.74 (p< 0.01) for the period 1965-1996 (n= 32), indicating that the late-wood of Pseudotsuga menziesii on this site explains 55% of the variation in precipitation in this part of the basin in the period from July to September, rainfall which in turn is related to the degree of availability of runoff.

The period of available precipitation records (1965-1996) was subdivided into two periods for calibration (1981 1996) and verification (1965-1980), as both sub-periods showed significant correlations with precipitation of 0.78 (p< 0.01 ) and 0.68 (p< 0.01), respectively; the total period of records used to generate a linear model for the purpose of reconstruction, which was as follows:

Where: Yt: is the reconstructed precipitation for the period from July to September (mm); X_t: index of late-wood ofthe SNA chronology.

The reconstruction ofprecipitation from July to September showed severe droughts in the decades of1810-1820, 1870, 1910, 1930, 1950, 1970 and 2000 and, wet periods in 1830, 1900, 1940, 1960 and 1980 (Figure 4).

Figure 4 Reconstructed precipitation of the seasonal period July-September with a regional chronology of late-wood of Pseudotsuga menziesii in the upper basin of the Mezquital river.  

Much of the annual and multi-year variation in the upper watershed of the Mezquital River can be explained by the impact of circulatory phenomena, such as ENSO, which is a standardized index based on pressure differences recorded between the islands of Tahiti and Darwin, Australia. This index measures the pressure fluctuations occurring on a large scale between the tropical western and east Pacific. The negative phase of ENSO is associated with a low pressure in Tahiti and high pressure in Darwin if this condition continues for long time, coinciding with a warming ofwaters in the eastern tropical Pacific, phase known as "Niño", while reverse conditions occur during the "Niña" phase (^{Ropelewski and Halpert, 1987}).

The dendrochronological series RWI, EWI and LWI representative of the upper watershed of the Mezquital River were compared with reconstructed indices of ENSO of the winter period for northern Mexico (^{Stahle et al., 1998}), annual rates of this phenomenon with means of zero reported by ^{Allan et al. (1996)} and tropical rain index (TRI), parameter that estimates the variability of ENSO based the rainfall anomalies recorded in the Central Pacific region and might be more stable than the South Oscillation Index (SOI) (^{Wright, 1979}). We found a highly significant association between RW and EW indices and ENSO indices reconstructed in the winter period, with correlations of-0.52 (p< 0.000, 1770-1977, n= 208, p< 0.000) for both indices. This association increased for RW and EW chronologies, when compared with the annual ENSO indices, obtaining high correlation values in the order of r= -0.58 (p< 0.01, 1876-1996, n= 121) and r= -0.59 (p< 0.01, 1876 to 1996, n= 121), respectively (Figure 5).

Figure 5 Relationship between the regional series of indices of early-wood for the upper basin of the Mezquital river and the annual indices of the Southern Oscillation Index (SOI). 

Discussion

The comparison of the dendrochronological series developed in the upper watershed of the Mezquital River, because of its closeness between them, about 20 km apart from each other, showed similar trends for certain periods, a situation attributed to the impact of common weather events that affected simultaneously all the study area. This common behaviour was evident for the decades 1820, 1860, 1890, 1910, 1920, 1950, 1980 and 2000, where the values of the indices fell below the average value (1), showing limiting conditions for the development trees, a situation that is usually associated with severe drought (^{Fritts, 1991}).

The opposite happened for the decades 1810, 1830, 1900, 1970 and 1990, where the rates exceeded the average value, suggesting the presence of favourable conditions for growth such as wet periods. The common behaviour among chronologies derived from multiple correlation analysis and principal component has been widely used to integrate regional chronologies in central and northern Mexico and explain the impact of climatic variables and circulatory phenomena of wide impact (^{Villanueva et al., 2005}; ^{Stahle et al., 2008}; ^{Cerano et al., 2012}). Only this way we can explain the extent of extreme hydroclimatic phenomena that can affect a watershed hydrologic or an entire country.

The dendrochronological series of Pinus durangensis for ring-whole, early and late-wood were supported by a sample of exceptional size (more than 80 radios after 1960) and generated dendrochronological values considered as excellent, as were the intercorrelation between series (0.68) and average sensitivity (0.385), exceeding the values ofthese parameters for many dendrochronological species in Mexico used for this purpose (^{Constante et al., 2010}; ^{Speers, 2010}).

The sensitivity of the species and the high interannual variation was reflected in the excellent response to climatic variables chronology, where both the series of ring-whole and early-wood, responded significantly to the seasonal rainfall from February to August, where presents more than 80% of the annual precipitation occurring in the upper basin of the Mezquital river. This response exceeds the chronologies of Pseudotsuga menziesii, Taxodium mucronatum and other conifers, which normally respond to winter-spring precipitation (October or November of the previous year to April, May or June ofthe current year's growth) (^{Cleaveland et al, 2003}; ^{Villanueva et al, 2005}; ^{Cerano et al, 2011}; ^{Irby et al., 2013}).

This situation is likely explained by the accumulation of photosynthates in the winter and spring required for radial growth of the tree at the beginning of the growing season (^{Pallardy, 2008}). An additional explanation is that the rain of the winter period, being of low intensity, accumulates in the soil profile, where it remains available for the radial growth of the tree at the beginning of the growing season (^{Fritts, 1976}).

The series late-wood responded significantly to cumulative rainfall from June to September, which is basically during the summer season where the maximum runoff occur and are finally captured in dams, ofwhich the agricultural area Valle del Guadiana feeds. This series of late-wood also showed a significant correlation with the annual runoff volume of the hydrometric station "El Pueblito", which records the volume in the dam Guadalupe Victoria, irrigating an area of 9000 ha in the Valle del Guadiana.

Significant associations were detected between the chronologies of early, late-wood and ring-whole of Pinus duranguensis, seasonal rainfall and recorded drained volumes, note the high potential of the species for the development of paleoclimatic reconstructions, useful to analyse the hydroclimatic variability in several seasonal periods. In this study, due to the absence oflong-lived trees, timing generated was of short extension (1945 to 2012); however, the dominance of this species in different stands ofmixed forest in the region does not rule out the possibility ofextending and developing a dendrochronological network with this species, from which it can generate a broader historical overview of hydroclimatic variability in different seasons of the year and that will generate information that contributes to implement a plan for the sustainable use of surface and groundwater in the water resources of Valle del Guadiana.

The chronologies EW, LW and RW of Pseudogtsuga menziesii in the upper watershed of the Mezquital River of the sites Cuevecillas-Culebras and Sierra del Nayar climatically have similar behaviour and are integrated into common chronology (CUC-SNA), increasing their potential for explaining the hydroclimatic variability of the site. Just like other chronologies of the previously developed species in the northern and central Mexico, this series showed a response to seasonal winter-spring precipitation, particularly during the period from January to May. The advantage of using this chronology, rather than that of Pinus duranguensis do not consider its best climate response, but rather in extent, covering the last 243 years (1770-2012). The variation observed in this reconstruction was reflected in the presence of dry periods detected in the decades of 1790-1800, 1820, 1860-1890, 1910, 1950, 1970 and 2000, wet periods were less frequent and were observed in the decades 1830, 1850, 1920, 1940, 1980 and 2010.

Dendroclimatic reconstructions of winter-spring previously developed in the region, show prolonged droughts in decades similar with slight variations from the starting year or termination of a dry or wet period (^{Cleaveland et al., 2003}; ^{González et al., 2005}; ^{Arrocena, 2013}). The outstanding wet period is noted for the 1830 where the conditions of high precipitation occurred for several consecutive years (^{Cleaveland et al., 2003}), a situation that was not only limited to this region, getting spread in much of western North America (^{Fye et al, 2003}).

Summer rainfall is the most important in this region of Mexico, because ofits implications in seasonal agriculture and the production of runoff to fill the dams for irrigation purposes; in this context, the reconstruction of rainfall in the July-September period, which constitutes about 85% of the annual precipitation in this basin is of great importance for water management in the "Valle del Guadiana". The reconstruction of summer precipitation indicated dry periods in the decades of 1810-1820, 1870, 1910, 1930, 1950, 1970 and 2000 and, wet periods in 1830, 1900, 1940, 1960 and 1980. On an annual basis, with extreme drought in the years 1859, 1867, 1879, 1887, 1894, 1902, 1908 1911, 1916-1918, 1922-1929, 1951, 1976-1982, 1989, 1991, 1999, 2011 and 2012. These dry years seriously impacted in the recorded volumes in the hydrometric station "El Pueblito" which began its registration in 1953; This way, the period 1976-1982 showed minimal values, as 1989, 1991, 1999, 2011 and 2012 (CNA, 2009). At period level, input ranges have fluctuated over time; thus, in the period 1962-1980 were recorded 136.0 Hm³ (cubic hectometers), from 1980 to 1995, 143.9 Hm³ and from 1995-2010, 114.2 Hm³ (^{Gómez et al, 2012}), volumes related with the variability in precipitation for the upper watershed of the Mezquital River, situation magnified in the recent decades, probably influenced by anthropogenic activities (^{Stahle et al, 2011}).

The historical impact of drought in this region has not been documented at all; however, because many ofthe dry periods of high intensity categorized simultaneously hit other regions ofthe country, it is considered that, the effect would have been similar; thus, the drought of the decades 1780-1790, 1800, 1820, 1890,1910,1950 and 1990lead to shortage of food, livestock mortality and epidemics in Chihuahua (^{Endfield and Fernández, 2006}), decreasing the surface irrigated in the lagoons region (^{Villanueva et al., 2005}). The drought of1785-1786, known as the "Year of hunger" severely impacted the production of basic grains in the centre of Mexico and produced widespread of famine (^{Florescano, 1980}), but also appeared in northern Mexico, probably more intense, a situation that is deduced by lower values of dendrochronological indices compared with others to central Mexico.

A high percentage of annual precipitation in the Sierra Madre Occidental occurs during the summer, influenced by the North American Monsoon or Mexican Monsoon (^{Gochis et al, 2005}); however, the multi-annual variability or low frequency in this region is influenced by atmospheric patterns of general circulation, ENSO case (^{Sathle et al, 1998}; ^{Cleaveland et al, 2003}), even though, during the warm phase, produces intense droughts in central Mexico and humid in the north, while in its cool phase the opposite happens, this dipolar behaviour has been determined in dendroclimatic studies (^{Therrell et al, 2002}) and on studies of fire frequency (^{Yocom et al., 2010}). The relationship found in this study between the indices of early-wood and SOI are the highest recorded in northern Mexico, which makes clear that, the extratropical connection of ENSO is very important in this region, given its predictability at short term, can help to support management plans of water resources in this region.

Conclusions

The dendrochronological network developed for the upper watershed of the Mezquital river, which is the source of water producing for the Valle del Guadiana, revealed the hydroclimatic variability that has enacts this basin during the last 240 years. The dendrochronological series showed significant associations between them, indicating that tree species are subject to common weather events. One of the components of this network, which is the chronology of Pinus durangensis indicated that, the dendrochronological series EW and RW responded significantly to the seasonal rainfall from February to August; while the LW to the precipitation and runoff occurred in the summer.

The limiting factor for reconstruction of this species was the extension of the chronology, which only covered the last 68 years (1945-2012). Pinus durangensis is one of the most exploited species of major economic importance in northern Mexico and finding long-lived stands of trees is quite dificult, but a closer exploration of the stands where it is distributed will allow to develop extensive series, helping to generate a more detailed historical analysis of the hydroclimatic variability of watersheds where it develops.

The generation ofchronologies with climatically sensitive species and with higher longevity, such as Pseudotsuga menziesii allowed making a reconstruction of the winter-spring precipitation (January-May) and summer (July-September), the latter related to the runoffs stored in the dams, water used in the agriculture and livestock sector. The understanding of the hydroclimatic variability in the upper basin of the Mezquital river is essential for the institutions responsible for water management in the region that require more expertise to integrate a plan for sustainable water management in the Valle del Guadina.

This hydroclimatic variability at annual and multi-anaual level has a high influence of circulatory atmospheric phenomena, such as ENSO, which showed one of the strongest signals detected in northern Mexico so far. The detailed study of the impact of this phenomenon in the region, through the construction of a more complete dendrochronological network is important for improving actions for predicting water availability in the Valle del Guadiana, since with the current technological advances available is possible determining the behaviour of the circulatory phenomenon in advance.