Study area

Cerro El Potosí is one of the highest peaks of the Eastern Sierra Madre. Geographically, it is located in the Galeana municipality, in the central-southern part of the state of Nuevo León, at the extreme geographical coordinates 24º50’35” to 24º53’16” N and 100º13’16” to 100º15’12” W. Because its altitude gradient ranges between 2 000 m, at the base of the mountain, and 3 721 m at its peak, it has various types of climate, depending on the altitude and exposure; BS₁kw”(e) semi-arid climates are predominant at the western exposure, with greater humidity at the eastern and northern exposures. According to ^{García-Arévalo and González-Elizondo (1991)}, the climate of the mountain is E(T)H(e), cold and extreme (^{García, 2004}), changing at lower altitudes, where the temperatures range between 7 and 14 ^°C, with annual precipitations of up to 400 mm (^{IMTA, 2013}).

The hydrology of CPO is very significant, as this mountain is the source of the San Fernando river, which drains the runoffs of the San Fernando-Soto La Marina hydrological region that discharge into the Gulf of Mexico.

The vegetation is highly diverse; Pinus cembroides Zucc. and Pinus greggii Engelm. ex Parl. grow in the lower part of the mountain, followed by a mixed conifer forest composed of various pine species at a higher altitude, (Pinus ayacahuite Ehrenb. ex Schltdl., Pinus pseudostrobus Lindl.), Pseudotsuga menziesii (Mirb.) Franco and Abies vejarii Martínez. Pinus hartwegii Lindl. grows above 3 000 masl, sometimes associated with Pinus culminicola (^{Beaman and Andresen, 1966}; Risking and Patterson, 1975).

Representative tree species of the various altitude floors of CPO were selected for dendrochronological purposes and in order to determine the influence that high and low frequency climate variations exert on their development. A Pinus culminicola population established on a shallow soil at a rocky site was located at the top of the mountain, at the coordinates 24^o51’37.5” N and 100^o13’56.8” W and at an altitude of 3 200 m. The associated species was Pinus hartwegii. However, in other sites almost pure populations of P. culminicola were found, but with high mortality of individuals due to the 1998 fire; for this reason, 50 trees were selected, and cross sections of both living and dead specimens were collected.

For Pinus hartwegii, two to three growth cores were obtained from 50 living trees in several locations near the P. culminicola populations; the geographical location of one of the sampled sites is 24^o52’21.9” N, 100^o13’56.4” W, at an altitude of 3 715 masl. As for Pseudotsuga menziesii, the growth cores corresponded to specimens from the site located at 24^o49’58.8” N and 100^o04’01.2” W, at 3 200 masl. As for the lower stratum, growth cores were collected from a site located at the coordinates 25^o30’35.5” N and 100^o29’00.120” W, and an altitude of 1 980 masl (Figure 1).

Figure 1 Geographical location of the species sampled for dendrochronological purposes at Cerro El Potosí, Nuevo León. 

The samples were mounted, polished and dated using conventional dendrochronological techniques (^{Stokes and Smiley, 1968}). Due to the irregular radial growth of P. culminicola, several radii were taken in each cross section in order to determine those with the least irregular growth. The total, early-wood and late-wood ring growth series were measured using a VELMEX measuring system at a resolution of 0.001 mm, and were then verified using the COFECHA software (^{Holmes, 1983}; ^{Grissino-Mayer, 2001}). Subsequently, the measurement series were standardized using the ARSTAN software (^{Cook, 1987}), firstly, with a negative exponential curve or a negative slope straight line, and secondly, with a flexible frequency curve with a 0.5 response and an N/2 wavelength, N being the length of the time series (^{Cook and Peters, 1981}).

The statistics generated with the COFECHA and ARSTAN programs, such as the intercorrelation between series, medium sensitivity, signal/noise ratio and Expressed Population Signal (EPS) were compared among all the species involved. The dendrochronological series for each species were contrasted in the common period, through simple correlations and principal components, in order to generate a dendrochronological series that would be representative of the dominant climate conditions of the altitude gradient. A seasonal rainfall reconstruction was developed based on the response function between average climate records of the stations close to CPO, which were integrated into a characteristic climate series of the region (Table 1), and the dendrochronological series that is representative of the hill, and its interannual and multiannual variability was analyzed, as well as its similarity to other precipitation reconstructions for the region.

The influence of atmospheric circulation phenomena was determined by comparing the indexes of the dendrochronological series against the values of the Southern Oscillation Indexes (SOI) (^{Allan et al., 1996}), Tropical Rain Index (TRI), an estimate of the variability of El Niño Southern Oscillation (^{Wright, 1979}); Pacific Decadal Oscillation (PDO) (^{Mantua et al., 1997}), Atlantic Multidecadal Oscillation (AMO) (^{Van Oldenborgh et al., 2009}), and the Palmer Drought Severity Index (PDSI), reconstructed for northeastern Mexico (^{Stahle et al., 2016}).

The late-wood dendrochronological series was compared to the presence of hurricanes that have impacted CPO or that have been detected within a radius of 50, 100 and 200 km. The effect was considered to be positive when the late-wood index was above 1.0, while there was no impact in those cases when the index was average (i.e. equal to 1.0) or below 1.0.

Dendrochronological series

A total ring chronology of 254 years (1757-2010) was generated for the 75 sampled Pinus culminicola trees, with two sections per tree (150 cross sections). 58 series were used for a 590-year chronology (1420-2009) of Pinus hartwegii; this is the most extensive chronology, as this is the longest-living species at CPO. At intermediate altitudes, of approximately 3 200 m, early-wood, late-wood and total ring chronologies of Pseudotsuga menziesii, stretching between 1845 and 2010 were based on 190 dated series. The chronology of Pinus greggii, a species that is representative of the lower part of the mountain, and of which early-wood, late-wood and total ring series were obtained, covered the years from 1850 to 2010.

Certain dendrochronological parameters, such as the intercorrelation between series, which show common values ranging between 0.55 and 0.65; the mean sensitivity (0.15 to 0.65), first-order autocorrelation (0.6 to 0.8), and signal/noise ratio (the higher the value, the better) for the generated chronologies exhibited values within intervals considered as normal (^{Speer, 2010}); the Expressed Population Signal (EPS) was also included, for climate reconstruction purposes (Table 2).

Despite the differences detected in the dendrochronological indexes for the compared species at the various altitudes of CPO, a similar behavior was detected between taxa, particularly in the decades 1860, 1890, 1900, 1920, 1950, 1970 and 2000, with indexes below the unit related to droughts, and with values above the unit associated to humid periods in the decades 1840, 1870, 1900, 1940 and 1960 (Figure 2).

Figure 2 Comparison between the ring-width indexes of Pinus culminicola Andresen & Beaman, Pinus hartwegii Lindl., Pseudotsuga menziesii (Mirb.) Franco and Pinus greggii Engelm. ex Parl. present at the altitude gradient of Cerro El Potosí in Nuevo León. 

When comparing the behavior of the indexes in terms of the variation in growth at the various altitudes of the mountain, the correlation for the overall ring series between Pinus culminicola and P. hartwegii during the common period from 1850 to 2010 was 0.68 (p<0.01). The value for this association diminished to 0.38 (p<0.01) when contrasted with Pseudotsuga menziesii, present at 2 500 masl, and to 0.17 (p<0.03) with Pinus greggii, located at 1 980 masl.

Relationship between growth rings and climate variables

The response of the representative growth series of CPO and the climate station integrated with the aforementioned stations (Table 1) shows that the overall standard ring chronology responded significantly to the January-September accumulated precipitation for the 1976-2004 period, with a correlation of 0.78 (p<0.01), which amounts to 61 % of the variance in seasonal accumulated precipitation in winter, spring and summer.

The precipitation records were calibrated for the total period of available records (1976-2004), as well as by 14-year sub-periods (1976-1989, 1990-2004). Based on this analysis, the ring width indexes were determined to account for 48 to 79 % of the variance in precipitation, while the total period accounts for 59 % of this variation. Therefore, the total period of available data was utilized for reconstruction purposes. The comparison between the parameters derived from the regression of the estimated precipitation sub-periods and the rainfall data recorded for the same sub-periods validates the robustness of the model. The reconstruction models surpassed the statistical validity tests when compared to independent precipitation values (Table 3).

The calibration model derived from the regression analysis of the 1976-2004 period was:

Where:

Y _t = Estimated precipitation of the January-September period (mm)

X _t = Total ring index for a particular year

The reconstructed precipitation exhibited a high interannual and multiannual variation in the 311 reconstructed years (Figure 3).

The dotted line represents the average of the January-September accumulated precipitation records; the gray line represents the interannual precipitation values, and the dark line is a flexible curve at decade level for highlighting low-frequency (multiannual) events

Figure 3 January-September reconstructed precipitation for CPO. The reconstructed period was adjusted to an EPS>0.85.  

Influence of general atmospheric circulation phenomena

In an extensive region such as northeastern Mexico, the common response between growth ring series distributed at various sites of the Eastern Sierra Madre is due to the influence of general atmospheric circulation phenomena like ENSO, AMO, and PDO, which, directly or through tele-connections, affect the interannual and multiannual climate variation at a regional level.

The analysis of the impact of ENSO on CPO determined a close but negative relationship with the reconstructed (November-February) SOI, as well as with the instrumental SOI for May-December. This relationship between the chronology and the ENSO indexes improved when the January-April TRI (r= 0.53, p<0.0001, n= 103 years) was utilized (Figure 4). This suggests that the growth of the trees in that region may account for 27 % of the variability of the precipitation anomalies produced by ENSO during the winter-spring seasons.

Figure 4 Relationship between the total ring indexes of the regional chronology on CPO and the tropical rain index (TRI). 

The relationship with the PDO was also significant, though for a longer period, which covered the months from March to December; a similar situation was determined based on the PDSI records for the summer (June-September), which exhibited a significant correlation (r=0.316, p<0.0007). No significant relationship with AMO was found to exist (Table 4).

Dendrochronological indexes and the influence of hurricanes

During the summer, one of the main sources of humidity at Cerro El Potosí are hurricanes. This influence seems to be most important in the formation of late wood of those species occurring on CPO. 40 % of the 24 cyclonic events recorded between 1887 and 2010 within a radius of 50 to 200 km, with CPO at the center, produced a higher annual ring width index (>1.0), while in the series of late Pseudotsuga menziesii and Pinus greggii wood, there was an increase above the average (>1.0) in at least 50 % of these events. The cyclones that occurred at a radius of 200 km in the years 1921, 1928, 1975 and 1999 had no impact on the ring-width index, which was below average (<1.0) for all taxa; therefore, although these phenomena may have been present in the region, they probably did not affect CPO.