Journal of Hydrology 462–463 (2012) 53–66
Contents lists available at ScienceDirect
Journal of Hydrology journal homepage: www.elsevier.com/locate/jhydrol
Water balances of old-growth and regenerating montane cloud forests in central Veracruz, Mexico L.E. Muñoz-Villers a,b,⇑, F. Holwerda c,d, M. Gómez-Cárdenas d,e, M. Equihua a, H. Asbjornsen d, L.A. Bruijnzeel c, B.E. Marín-Castro f, C. Tobón g a
Depto. de Ecología Aplicada, Instituto de Ecología, A.C., Xalapa, Veracruz, Mexico Department of Forest Engineering, Resources and Management, Oregon State University, Corvallis, OR, USA Department of Hydrology and Geo-Environmental Sciences, Faculty of Earth and Life Sciences, VU University, Amsterdam, The Netherlands d Department of Natural Resource Ecology and Management, Iowa State University, Ames, IA, USA e Instituto Nacional de Investigaciones Forestales, Agrícolas y Pecuarias, Oaxaca, Oax., Mexico f Posgrado en Ciencias, Instituto de Ecología, A.C., Xalapa, Veracruz, Mexico g Universidad Nacional de Colombia, Medellín, Colombia b c
a r t i c l e
i n f o
Article history: Available online 4 March 2011 Keywords: Cloud water interception Evapotranspiration Streamﬂow Volcanic soils Leakage Catchments
s u m m a r y This paper compares the water budgets of two adjacent micro-catchments covered by mature (MAT) and 20-year-old secondary (SEC) lower montane cloud forests, respectively, in central Veracruz, Mexico over a 2-year period. Rainfall (P) and streamﬂow (Q) were measured continuously, whereas dry canopy evaporation (transpiration Et), wet canopy evaporation (rainfall interception I), and cloud water interception (CWI) were quantiﬁed using a combination of ﬁeld measurements and modeling. Mean annual P was 3467 mm, of which typically 80% fell during the wet season (May–October). Fog interception occurred exclusively during the dry season (November–April), and was 62% of annual P for both forests. Rainfall interception loss was dominated by post-event evaporation of intercepted water rather than by within-event evaporation. Therefore, the higher overall I (i.e. including CWI) by the MAT (16% of P vs. 8% for the SEC) reﬂects a higher canopy storage capacity, related in turn to higher leaf area index and greater epiphyte biomass. Annual Et totals derived from sapﬂow measurements were nearly equal for the MAT and SEC (790 mm each). Total annual water yield calculated as P minus (Et + I) was somewhat higher for the SEC (2441 mm) than for the MAT (2077 mm), and mainly reﬂects the difference in I. Mean annual Q was also higher for the SEC (1527 mm) than for the MAT (1338 mm), and consisted mostly of baseﬂow (90%). Baseﬂow recession rates were nearly equal between the two forests, as were stormﬂow coefﬁcients (4% and 5% for MAT and SEC, respectively). The very low runoff response to rainfall is attributed to the high inﬁltration and water retention capacities of the volcanic soils throughout the 2 m deep proﬁle. The water budget results indicate that 875 and 700 mm year1 leave the SEC and MAT as deep groundwater leakage, which is considered plausible given the fractured geology in the study area. It is concluded that 20 years of natural regeneration following cloud forest disturbance in central-eastern Mexico is capable of producing near-original hydrological behavior. Ó 2011 Elsevier B.V. All rights reserved.
1. Introduction Although tropical montane cloud forests (TMCF) cover only 1.4% (215,000 km2) of the world’s tropical forests (Scatena et al., 2010), their location in mountain regions exposed to frequent fog and high rainfall, together with the associated low evaporative losses, typically results in much higher water yields compared to noncloud-affected forests (Zadroga, 1981; Bruijnzeel, 2001). However, ⇑ Corresponding author at: Department of Forest Engineering, Resources and Management, Oregon State University, Corvallis, OR, USA. Tel./fax: +1 15 417378719. E-mail address: [email protected]
(L.E. Muñoz-Villers). 0022-1694/$ - see front matter Ó 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.jhydrol.2011.01.062
much of the original TMCF have disappeared, and many of the remaining cloud forests are threatened by local and regional climatic warming effects such as a rising cloud base or reduced overall cloud incidence (Still et al., 1999; Lawton et al., 2001; Karmalkar et al., 2008; Mulligan, 2010). Knowledge of TMCF hydrological functioning has increased considerably in the last 15 years (recently summarized by Bruijnzeel et al., 2010), with most published work focusing on single processes, mostly rainfall and cloud water interception and, to a lesser extent, evaporation. However, studies of the impact of TMCF conversion to pasture or cropping, or of cloud forest regeneration on streamﬂow are almost non-existent (Bruijnzeel, 2005, 2006; cf. Ingwersen, 1985).
L.E. Muñoz-Villers et al. / Journal of Hydrology 462–463 (2012) 53–66
The effect of cloud forest conversion or regrowth on water yield is likely to reﬂect a trade-off between the loss of the extra water formerly gained via cloud water interception and the difference in water use (rainfall interception plus transpiration) between the old and new vegetation (Zadroga, 1981; Ingwersen, 1985; Bruijnzeel, 2005). A possible negative effect on streamﬂow is likely to be most pronounced during the dry season, when inputs of cloud water are typically of greater importance (Harr, 1982; Holder, 2003; Garcia-Santos, 2007). The effect may be further enhanced by advanced degeneration of the soil’s inﬁltration capacity after forest clearing due to compaction by cattle or machinery. Under such conditions, recharge of soil and groundwater reserves during the rainy season may become impaired to the extent that dry season ﬂows are reduced (Bruijnzeel, 2004). To date, only two studies have addressed these questions in some detail. According to Ingwersen (1985), partial clearing of Douglas ﬁr forest in the Paciﬁc Northwest of the US, an area subject to heavy fog incidence (Harr, 1982), caused a temporary decrease in baseﬂows during the dry summer months. Interestingly, the effect gradually disappeared after 5–6 years when the regenerating trees were apparently effective at capturing sufﬁcient amounts of cloud water (Ingwersen, 1985). Conversely, under the very wet conditions prevailing in northern Costa Rica, a process-based modeling approach (Schellekens, 2006) indicated that cloud forest conversion to pasture did not lead to signiﬁcant decreases or increases in annual and seasonal water yields because the ‘loss’ of the former fog water input was compensated by lower water use of the pasture (Bruijnzeel, 2006). However, to properly assess the effects of land use and climate change on TMCF hydrology, additional information is required for different cloud forest settings around the world, particularly for areas experiencing a well-developed dry season, as potentially negative impacts might be greatest under these conditions (Ingwersen, 1985; Bruijnzeel, 2005). Similarly, although secondary forests are currently more widespread than old-growth forests in many tropical environments (Uhl et al., 1988; Xu et al., 1999; Fox et al., 2000), this is not reﬂected in the volume of hydrological research conducted in regenerating tropical forests (e.g. Fritsch, 1992, 1993; Giambelluca, 2002; Bruijnzeel, 2004; Hölscher et al., 2005). The rate of forest (and soil) recovery after abandonment of pasture or agricultural land is strongly inﬂuenced by previous land management (Uhl et al., 1988; Hölscher et al., 2005; Zimmermann et al., 2006; Aide et al., 2010). As such, different stages of cloud forest development reﬂecting different land use histories may well exhibit different site water balances (cf. Ingwersen, 1985) but, again, information is extremely scarce (cf. Takahashi et al., 2010). In Mexico, TMCFs cover less than 1% of the country’s territory, but they harbor 10% of the country’s plant biodiversity (Rzedowski, 1996). Although many have been disturbed or converted to other land uses (mostly pasture, coffee plantations, and agricultural land: Muñoz-Villers and López-Blanco, 2007), the hydrological effects of these changes are essentially unknown. This type of information is urgently needed as the recognition of the value of TMCFs as local providers of high-quality water continues to increase throughout Mexico, as evidenced by a series of nationwide conservation initiatives that have emerged in the last decade (MuñozPiña et al., 2008). The present study aims to partly ﬁll this gap in knowledge by quantifying the hydrological effects of montane cloud forest disturbance and regeneration in a seasonally dry setting in central Veracruz State (central-eastern Mexico), using a combination of hydrological (rainfall and cloud water interception, and streamﬂow dynamics) and ecophysiological (tree water uptake) ﬁeld measurements and modeling. The central research questions of the investigation were: (1) How does the hydrological behavior of a 20-year-old naturally regenerating cloud forest compare to that of an old-growth cloud forest? (2) What are the key
variables controlling the water ﬂuxes in this TMCF ecosystem? and (3) How does the seasonal climate affect the water balances of these forests?
2. Study area The research was carried out in two adjacent headwater catchments covered by mature (hereafter called MAT; 24.6 ha) and regenerating (hereafter called SEC; 11.9 ha) lower montane cloud forests (LMCF), locally called bosque mesóﬁlo de montaña (García-Franco et al., 2008). The micro-catchments are located at 97°020 W and 19°290 N at an average altitude of 2170 m a.s.l. on the eastern (windward) slopes of the Cofre de Perote volcano, and around the La Cortadura Forest Reserve of the municipality of Coatepec (central Veracruz State, Mexico; Fig. 1). The MAT and SEC catchments are part of the 38 km2 basin of the Los Gavilanes river, which forms the principal water supply for the city of Coatepec and surrounding villages (García-Coll et al., 2004). Site characteristics are listed in Table 1. The forest catchments have steep slopes and deeply incised valleys; slopes ranging from 20° to 45° cover more than half the area of each catchment (Muñoz-Villers, 2008). The streams draining the catchments are perennial. The volcanic soils were classiﬁed as Umbric Andosols (Campos, 2008; FAO-UNESCO, 1997). Deep (2.5 m) and multi-layered soil proﬁles characterize the middle and upper portions of the hillslopes, whereas much shallower soils (0.6 m) are found closer to the streams (Marín-Castro, 2010). The soils are characterized by a silt loam texture, relatively low bulk densities, high porosity, high organic matter content, and high water retention capacity (Geris, 2007; Table 1). The volcanic soils are overlying a semi-permeable, moderately weathered andesitic breccia, underlain, in turn, by weathered and fractured andesitic to basaltic rocks from the Oligocene–Neogene period (D. Geissert personal communication). According to the Köppen classiﬁcation modiﬁed by Garcia (1988), the climate between 2000 and 3000 m elevation in this region is ‘temperate humid with abundant rains during the summer C(m)(w)’, with average temperatures and annual rainfall between 12 and 18 °C and 2000 and 3000 mm, respectively. The climate at this latitude is inﬂuenced by the trade winds and the subtropical high pressure belt (Metcalfe, 1987). In the northern hemisphere winter, the proximity of the subtropical high leads to stable, dry weather but the occasional passage of cold fronts produces light rains and/or drizzle for 1–3 days per event (Baez et al., 1997). With the northward movement of the Inter-Tropical Convergence Zone in the summer, the region comes under the inﬂuence of the easterly trade winds that bring humid conditions with frequent showers and thunderstorms (Baez et al., 1997). Hence, the climate can be divided into two seasons: a relatively dry season (November– April) and a wet season (May–October). A more detailed description of the local weather will be given below. The MAT forest is a well-conserved, old-growth lower montane cloud forest, with an average canopy height of about 27 m, and emergent trees up to 40 m (García-Franco et al., 2008). The forest is composed of deciduous (Quercus ocoteoifolia, Quercus corrugata, Clethra macrophylla) and evergreen broadleaf (Miconia glaberrima, Parathesis melanosticta, and Alchornea latifolia) species; the forest is also rich in vascular epiphytes of the family Bromeliaceae (García-Franco et al., 2008). The leaf area index (LAI) was estimated at 6.3 from measurements of annual leaf litter production and leaf speciﬁc area. Approximately half of the SEC forest is dominated by deciduous alder stands (Alnus jorullensis), with an average canopy height of about 20 m and an LAI of 5.2; this forest regenerated following a wild-ﬁre occurring 20 years ago (Gómez-Cárdenas, in preparation). Aerial photo analysis showed that the other half of the SEC forest regenerated from pasture land that was abandoned
L.E. Muñoz-Villers et al. / Journal of Hydrology 462–463 (2012) 53–66
Fig. 1. Location of the study area in central Veracruz, Mexico, and maps for the old-growth (MAT) and secondary (SEC) montane cloud forest catchments showing the instrumentation sites. Sources: Topographic data from the Instituto Nacional de Estadística, Geografía e Informática (INEGI)(1993) (1:250 000 scale: Mexico) and INEGI (2000) (1:50 000 scale: Mexico). Micro-catchment boundaries from Muñoz-Villers (2008)
Table 1 Topographic, soil physical, and vegetation characteristics of the mature (MAT) and secondary (SEC) forest catchments at La Cortadura. PAW is the Plant Available Water, LAI is leaf area index. Where available, the standard deviation (SD) is given.
Area (ha) Mean slope (°) Mean slope length (m) Length of river channel (km) Mean slope of river channel (°) Aspect Mean soil bulk density (qb)a (g cm3) Mean soil porositya PAWa (cm3 cm3) LAIb (m2 m2) Mean canopy height (m) Stem densityb (stems ha1) Basal areab (m2 ha1) Sapwood areab (m2 ha1)
24.6 33 123 1.2 20 NW-SE 0.49 ± 0.13 0.69 ± 0.03 0.08 ± 0.03 6.3 27 ± 10c 2967 48 15
11.9 31 105 0.7 17 W-E 0.45 ± 0.06 0.74 ± 0 0.08 ± 0.02 5.2 20d 4191 32 14
a Geris (2007); the average and SD of the values of the A, B, and C horizons (up to 1 m depth). b Gómez-Cárdenas, in preparation. c García-Franco et al. (2008). d Unpublished data F. Holwerda.
P þ CWI ¼ ET þ Q þ DðS þ GÞ þ L
where P is the rainfall, CWI the input by cloud water interception, ET the evapotranspiration made up of rainfall interception (I) and transpiration (Et) (neglecting evaporation from the forest ﬂoor), Q the streamﬂow, DS and DG the changes in soil moisture and groundwater storage over the period of analysis, and L the leakage into or out of the catchment (all expressed in mm). The water balances presented in this study cover 2 years (November 2006–October 2008). Except for the changes in soil and groundwater storage and catchment leakage, all terms in Eq. (1) were obtained through a combination of ﬁeld measurements and modeling, as described in the following sections. An estimate of DS was obtained by comparing the precipitation totals of the 2 weeks preceding the beginning and end of the water balance period (Schellekens et al., 2000), whereas the change in groundwater storage (DG) was calculated as the difference between the average streamﬂow during the ﬁrst and last 10 days of the water balance period divided by the natural logarithm of the recession constant (k; see below)(Waterloo et al., 2007). The leakage term (L) was estimated as the residual. 3.2. Climate variables
around the same time (L.E. Muñoz-Villers, unpublished data; L. Martínez, personal communication). Other important species in the SEC include Quercus corrugata, Clethra macrophylla, Alchornea latifolia, Miconia glaberrima and Prunus tetradenia. 3. Methods 3.1. Catchment water balance The general water balance equation for a catchment in the cloud forest belt reads (Frumau et al., 2006):
A weather station was installed at 2128 m a.s.l. in between the MAT and SEC catchments (Fig. 1). The ground surface upwind of the station consisted of 0.5 m tall grass and had a 30° slope with an E-SE aspect. Measurements of incoming solar radiation (Sin, W m2) were made at 3 m using a Kipp & Zonen CM2 pyranometer. Temperature (T, °C) and relative humidity (RH, %) were measured at 2 m using a combined T/RH sensor (HMP45, Vaisala). Dry- (Td) and wet-bulb (Tw) temperatures were measured at 2 m using custom-built thermocouples (VU University, Amsterdam). The wetbulb thermocouple consisted of a dry-bulb thermocouple ﬁtted with a cotton wick, which was kept by a constant supply of distilled water from a reservoir. Actual vapor pressure (VPA, kPa)
L.E. Muñoz-Villers et al. / Journal of Hydrology 462–463 (2012) 53–66
was calculated from Td and Tw, after which RH was calculated using VPA and the saturation vapor pressure (SVP, kPa) for ambient temperature. Gaps in the thermocouple-derived RH data were ﬁlled using data from the humidity probe. Wind speed (U, m s1) was measured at 2.5 m using a cup anemometer (A100R, Vector Instruments) and wind direction (Udir, degrees) at 3 m using a wind vane (W200P, Vector Instruments). Measurements were made every 30 s and average data were stored every 10-min using a Campbell Scientiﬁc Ltd. 21X data logger. Data of this station were available for the entire water balance period (November 2006–October 2008). As part of a follow-up project, a weather station similar to the one at La Cortadura was installed at 2400 m a.s.l. in April 2008 (about 600 m northwest of the top of the MAT catchment; cf. Fig. 1). 3.3. Rainfall and cloud water Because of the lack of sufﬁciently large clearings in the La Cortadura forest and because of difﬁculties with obtaining site permissions in privately owned areas, rainfall (P) was measured at only one site in the SEC catchment (station SECP) and at two sites in the MAT catchment (BP1, MATP) (Fig. 1). An additional two rain gauges were installed at the weather stations (VPco and VPtg; Fig. 1). Information about the measurement period covered by each gauge is given in Table 2. The rain gauges were of the type ARG100 (Environmental Measurements), Casella CEL, and RG2 M (Onset) (all with a resolution of 0.2 mm). The data of the stand-alone rain gauges were stored using custom-built (VU University, Amsterdam) and HOBO pendant event (Onset) data loggers. Signals from the gauges at the weather stations were measured by the Campbell loggers. All tipping bucket devices (including those for throughfall and stemﬂow measurements, see below) were dynamically calibrated to account for the variable error associated with the loss of water during bucket rotation (Calder and Kidd, 1978). Fog and wind-driven drizzle were measured using a passive Juvik-type fog collector, consisting of a louvered cylindrical aluminum shade screen of 49.5 cm height and 25.6 cm diameter (0.127 m2 cross-sectional area) installed at 2.5 m in the La Cortadura weather station (modiﬁed from Juvik and Ekern, 1978; Frumau et al., 2010). Apart from having a larger collecting surface than the original Juvik-type collector, a further modiﬁcation to the gauge consisted of an extra funnel being installed on top of the cylindrical screen to separately measure the vertical (top funnel) and horizontal (bottom funnel) components of incoming precipitation (termed P and HP, respectively) (Frumau et al., 2010; see Holwerda et al. (2010) for further details). Please note that the vertical precipitation component was termed P as it is essentially equal to rainfall. As described in detail in Holwerda et al. (2010), the fog gauge data were used to separate precipitation events into events with and without fog. Precipitation events were identiﬁed on the basis of the P record and were separated by a dry period of at least three hours. In short, the precipitation captured by the fog gauge screen (HP) consisted of inclined rain, as well as wind-driven drizzle and fog. By comparing actual HP measurements with calculated amounts of inclined rainfall intercepted by the fog gauge,
Holwerda et al. (2010) showed that HP was composed primarily of inclined rain during the wet season, whereas numerous events with additional inputs from fog and/or wind-driven drizzle occurred during the dry season. Because it was not possible to separate the contributions of fog and drizzle to HP, these were taken together and referred to as cloud water (CW) (cf. Scholl et al., 2007). The dry season events were then separated into events with and without CW using a threshold value for the ratio of HP to total precipitation (HP + P) calculated from the wet season data, which included conditions of inclined rain only (see Holwerda et al. (2010) for details).
3.4. Rainfall and cloud water interception Rainfall interception (I) was determined as the difference between rainfall and throughfall (TF) plus stemﬂow (SF) using data from events without cloud water occurrence. TF was measured using 4 m long by 0.3 m wide V-shaped, stainless steel troughs (VU University, Amsterdam) installed at an angle of 15–20° to facilitate drainage, and equipped with 2 cm high risers to reduce splash losses. TF volumes were measured using a custom-built 50 ml capacity tipping bucket with logger system (VU University, Amsterdam). The TF plots in the MAT and SEC forests were located on 35–40° and 20–25° south-facing slopes, respectively (Fig. 1). Both plots were located within 100 m from a rain gauge (Fig. 1). Three and four troughs were used in the SEC and MAT, respectively, representing total collection surfaces of about 3.5 and 4.6 m2, respectively. Measurements at the SEC were made between September 2006 and June 2007, and at the MAT between April 2007 and April 2008. Stemﬂow (SF) was not measured in the MAT but in the SEC, SF was measured on ten alder trees covering the range in observed diameter classes (10–24 cm; see Holwerda et al. (2010) for further details). In order to obtain interception (I) estimates for the entire water balance period (including for events with cloud water occurrence; see Holwerda et al. (2010) for details), the single-storm form of the Liu S. model (Liu, 2001) was used:
# " ð1 pÞ E E P 1 þ P I ¼ C m 1 exp Cm R ð1 pÞR
where Cm is the canopy storage capacity (mm), p the canopy gap fraction, and E=R the ratio of the mean evaporation rate from the wet canopy (mm h1) to the mean rainfall intensity (mm h1) calculated over all hours for which P > 0.5 mm (Gash, 1979). The evaporation rate from the wet canopy (E) was calculated using the Penman–Monteith (P–M) equation (Monteith, 1965) with the surface resistance set to zero (e.g. Rutter et al., 1972). Because the calculated values of E for the two forests were almost identical, an average value was used throughout (see Holwerda et al. (2010) for details). Likewise, because of minimal differences between the respective rain gauges (see below), R values for the two forests were calculated from the P data of the La Cortadura weather station.
Table 2 Summary of the statistical comparison between daily rainfall (P) totals measured at various locations in and around La Cortadura; see Fig. 1 for the locations of rain gauges MATP, SECP, VPco, and VPtg. The ﬁrst value is the signed-rank test statistic Z calculated following Hirsch et al. (1993). Daily rainfall values are not signiﬁcantly different if Z < 1.96 (0.05 signiﬁcance level). The second value is the percent difference between P totals calculated as 100(Y X)/X. Also shown is the period over which data were available for comparison. X/Y
BP1 MATP SECP VPco
3.63/+3% November/06–October/08 1.03/+2% April/07–April/08
1.57/2% November/06–October/08 1.42/3% April/07–April/08 5.26/5% November/06–October/08
0.14/1% April/08–October/08 – 0.58/2% April/08–October/08 0.73/+3% April/08–October/08
L.E. Muñoz-Villers et al. / Journal of Hydrology 462–463 (2012) 53–66
The canopy parameters (Cm and p) and the ratio E=R were derived separately for the wet and dry seasons. This was done for two reasons: (i) rainfall intensities were higher during the wet season (see below); and (ii) both forests were dominated by species that were (semi-) deciduous during the dry season. The canopy parameters were obtained via a non-linear least square ﬁt of the Liu S. interception model (Eq. (2)) to the data. For dry season events, only events without cloud water (CW) occurrence were used in the parameterization of the interception model. For dry season events with CW occurrence, cloud water interception by the forest canopy (CWI) was calculated as TF + SF + modeled I minus P. To estimate CWI for the 2-year water balance period, linear regression equations between CWI and horizontal precipitation (HP) as measured at the La Cortadura weather station were used.
conductance was calculated using the Jarvis (1976) model, including the response functions for solar radiation (Sin), vapor pressure deﬁcit (VPD), and temperature (T) (e.g. Stewart, 1988; Dolman et al., 1991; Harris et al., 2004). The model parameters were obtained using values of daily mean canopy conductance, calculated in turn by inserting the sapﬂow-based daily Et estimates in the rearranged P-M equation (e.g. Shuttleworth, 1988), and using non-linear least squares optimization. Model parameters were derived separately for the wet and dry seasons. A soil moisture function was not included because measured changes in soil water tension during the period of sapﬂow measurement were too small (approximate range of 30 to 220 hPa) to have an effect on canopy conductance.
Discharge (Q) was measured using 90° and 53.8° V-notch weirs at the MAT and SEC catchment outlets, respectively (Fig. 1). Waterlevels were recorded every 2-min using Schlumberger LT F15/M5 Diver sensors in combination with F5/M1.5 baro-divers to compensate for atmospheric pressure (both having ±1.5 mm accuracy). Water-levels were read from a staff gauge every 2 weeks to verify the Diver measurements. Bedload sediment accumulated behind the weirs was removed regularly to ensure reliable measurements. Water-levels were converted to Q (L s1) using the experimental stage-discharge relationships for these weirs (Kindsvater and Carter, 1957), calibrated with ﬁeld volumetric and salt dilution measurements of discharge (see Muñoz-Villers (2008) for details). Continuous streamﬂow data were available for the entire water balance period. Streamﬂow (Q) was separated into baseﬂow (Qb) and quickﬂow (Qd) following the general approach of Hewlett and Hibbert (1967). The separation was performed using a slope constant of 0.030 mm h1 for both the MAT and SEC as obtained from a manual analysis of about 150 rainfall events (Muñoz-Villers, 2008). Master recession curves were constructed from dry season streamﬂows using the matching strip method (Toebes and Strang, 1964). The master recession curves were described using linear reservoir theory (Chapman, 1999):
Sap ﬂux density (Fd, kg m2 s1) was measured for the following dominant species in the MAT (Q. corrugata, Q. ocoteifolia, Ternstroemia sylvatica, Hedyosmum mexicanum, Litsea sp., Alchornea latifolia, Clethra macrophylla) and SEC (A. jorullensis, Alchornea latifolia, Clethra macrophylla) forests using a combination of thermal dissipation probes (Granier, 1985) and heat pulse sensors (Cohen et al., 1981) on a total of 50 individual trees (see Gómez-Cárdenas (in preparation) for a detailed description). Due to the higher species diversity and stand heterogeneity of the mature forest, a greater number of individual trees were monitored for sap ﬂux in the MAT (n = 37) compared to the SEC (n = 13). The number of individuals sampled per species ranged between 4 and 12, and approximated their relative abundance and range of diameter at breast height (dbh) size classes (from 10 to >35 cm) in each stand. A brief description of the methods for sap ﬂow measurements follows. Continuous measurements were made on selected individuals from January–November 2007. Fd was calculated using the empirical relation developed by Granier (1985). Several corrections were made to account for potential sources of error, including the vertical gradient of external temperature (Lundblad et al., 2001), possible overestimations due to sensor insertion into non-conductive xylem (Clearwater et al., 1999; Lu et al., 2004), wounding (O’Grady, 2000), and by comparing data from the thermal dissipation probes with the more precise heat pulse method (Gómez-Cárdenas, in preparation). Stand transpiration was estimated by multiplying average Fd by mean sapwood area per unit forest ﬂoor (As, m2 m2) by species, and then summing across all (measured) species for each cover type to obtain total Et for each forest, which included an estimate for transpiration by shrubs (Gómez-Cárdenas, submitted for publication). Sapwood area was calculated from measured dbh using species-speciﬁc allometric equations (cf. Wullschleger et al., 2001), whereas mean sapwood area per unit ground area (As) was determined from a survey of 30 and 11 10 10 m2 plots in the MAT and SEC, respectively (Gómez-Cárdenas, submitted for publication). To estimate Et for the entire water balance period, the P-M equation (Monteith, 1965) was applied using a daily calculation time step (cf. Kumagai et al., 2008). In doing so it was assumed that the temperature, humidity, and wind speed measurements made at the La Cortadura weather station were representative of the conditions at 2 m above the forests. The parameters of the P-M equation were calculated following the FAO guidelines for calculating crop reference evapotranspiration ET0 (Allen et al., 1998), with the following exceptions with regard to the surface parameters: (i) for albedo, a ﬁxed value of 0.12 was used for both forests; (ii) aerodynamic conductance was calculated using an average canopy height of 27 m for the MAT and 20 m for the SEC; and (iii) canopy
Q t ¼ Q 0 expðt=sÞ ¼ Q 0 k
where Q0 and Qt are the ﬂows (mm d ) at time 0 and t (days), respectively, s the turnover time of the groundwater storage (days), and k is the recession constant. The initial discharge value Q0 and recession constant k were obtained from linear regression analysis using log-transformed discharge data. 3.7. Soil hydrological measurements Steady-state surface inﬁltration capacities (fc, mm h1) were measured using a constant-head, single-ring inﬁltrometer (17 cm diameter; modiﬁed version of Gómez-Tagle et al., 2008), inserted down to 5 cm depth. In each catchment, 17 measurements were made along a transect running from the stream channel to the ridge on a south-facing slope (Fig. 1; Marín-Castro, 2010). The time required to reach steady-state conditions varied from 15 to 50 min (Marín-Castro, 2010). Based on the inﬁltration data, top-soil saturated hydraulic conductivities (Kfs, mm h1) were estimated using the Wu2 method (Wu et al., 1999) and employing a value of 0.12 cm1 for the alpha parameter as suggested by Elrick and Reynolds (1992) for loamy soils. Before each inﬁltration test, samples of the soil at 6 cm depth were taken for the determination of bulk density (qb, g cm3), texture, and organic matter content (SOM, %) at the soil laboratory of the Instituto de Ecología AC in nearby Xalapa (see Marín-Castro (2010) for details).
L.E. Muñoz-Villers et al. / Journal of Hydrology 462–463 (2012) 53–66
3.8. Error analysis An error analysis was performed to provide error bounds for each of the water balance components. For those components for which a standard error (SE) could not be calculated from the data, an error was estimated. The standard error (SE) of the leakage term (L; Eq. (1)) was estimated by adding the errors in the respective water balance components quadratically (cf. Calder, 1993).
3.8.1. Rainfall In addition to the systematic underestimation of rainfall due to tipping bucket rotation (see ‘‘Section 3’’), other sources of error in the measurement of P include: (1) underestimation because of wind effects (Nespor and Sevruk, 1999); (2) the interaction between inclined rainfall and sloping ground (Sharon, 1980); (3) spatial variability; and (4) wetting and evaporation losses from the gauge The wind-induced error was assumed negligible, ﬁrstly because wind speeds in the study area were low year-round (1.4 m s1 on average; Holwerda et al., 2010), and secondly because the rain gauges used at the most exposed sites (i.e. the weather stations) were especially designed to reduce the wind error (type ARG100; Environmental Measurements). Corrections for the effect of slope were also not made because of the low wind speeds. Since wetting and evaporation losses are usually small, these were also neglected. Removing these sources of error leaves spatial variability as the likely dominant error source. Hence, the standard error of the mean P for the four rainfall stations (Fig. 1, Table 2) between April and October 2008 was taken as the SE of P for the MAT and SEC catchments.
3.8.2. Rainfall and cloud water interception The standard errors of the rainfall interception loss were estimated by adding the respective errors in P, TF, and SF, and the modeling error quadratically. The SE of the TF was calculated from the measured TF totals whereas the SE of the SF was calculated from the SF totals measured at the ten alder trees in the SEC. The modeling error was represented by the relative predictive error (RPE; Liu, 2001). The uncertainty in the cloud water interception estimates were calculated by adding the errors in P, TF, SF, and I, and the modeling error quadratically. For the modeling error, the standard error of the regression of cloud water interception on horizontal precipitation was taken.
3.8.3. Transpiration Uncertainties in transpiration (Et) estimates result from: (i) errors in the measurement of sap velocity; (ii) errors in up-scaling sap ﬂow rates in individual trees to the stand level (Et); and (iii) modeling error. Because measurement errors were minimized during data processing (see above), these were not further considered in the present analysis. Scaling sap ﬂow measurements to the stand level includes two types of errors (e.g. Granier et al., 2000): (i) error in the average sap velocity Fd due to variability amongst trees; and (ii) uncertainty in the scaling factor, i.e. mean sapwood area per unit area of forest ﬂoor (As). The SE of average Fd for each species was calculated from sap velocities measured in individual trees. The SE of average As was represented by the variability encountered within the 30 and 11 10 10 m2 forest inventory plots in the MAT and SEC, respectively (see ‘‘Section 3’’). The errors in species-speciﬁc Et were calculated by multiplying Et by the quadratic sum of the relative SEs of Fd and As, and their quadratic sum was taken to represent the total scaling error. The root mean square error (RMSE) between modeled and observed Et was taken as the modeling error. The total error in Et was taken as the quadratic sum of the scaling and modeling errors.
3.8.4. Streamﬂow and soil and groundwater storage Uncertainties in the amount of streamﬂow are mainly associated with errors in water-level and discharge measurements. Regarding the former, the pressure transducer sensors used are highly stable and accurate, and have a maximum drift of ±0.1%. Hence the error in measured water-levels was likely to be very small and therefore assumed negligible. Therefore, the SE of the streamﬂow was determined from the standard deviation (SD) of the discharge values calculated using the different weir calibration factors as obtained from volumetric and salt dilution discharge measurements. For the soil- and groundwater storage, the calculated values themselves were taken as their SE (i.e. the errors were assumed to be 100%). 4. Results 4.1. General climate conditions Daily and monthly mean values of solar radiation (Sin), temperature (T), and vapor pressure deﬁcit (VPD) are shown in Fig. 2. Daily mean Sin ranged from about 50 W m2 to 350 W m2, with the highest values occurring during the late dry season (March, April) and early wet season (May). Interestingly, the day-to-day variation in Sin was much greater during the dry season than during the wet season. The same was observed for T and VPD. This greater variability reﬂects the alternation of warm and dry (high pressure) and cool and wet (cold front) weather conditions that is typical for the dry season in the study area. The daily mean T normally varied between 10 and 20 °C, although values as low as 5 °C did occur during the dry season. Daily mean VPD ranged from near-zero on rainy and cloudy days (dry and wet seasons) to values of about 1.5 kPa on dry and sunny days in the dry season. The annual
Fig. 2. (a) Daily (dots) and monthly (line) mean solar radiation (Sin) measured at the La Cortadura weather station between November 2006 and October 2008. Light gray areas indicate dry season periods. (b) Daily and monthly mean temperature (T). (c) Daily and monthly mean vapor pressure deﬁcit (VPD); also shown is the daily mean actual vapor pressure (VPA, dashed line).
L.E. Muñoz-Villers et al. / Journal of Hydrology 462–463 (2012) 53–66
Table 3 Annual totals of streamﬂow (Q) plus monthly mean, minimum, and maximum Q in mm for the wet and dry seasons as measured at the mature (MAT) and secondary (SEC) cloud forests in central Veracruz, Mexico. MAT
Dry season Qmin
147 ± 87 238 ± 149
36 ± 16 28 ± 17
164 ± 91 258 ± 145
47 ± 23 41 ± 19
Table 4 Descriptive statistics of the surface inﬁltration rate (fc, mm h1), soil bulk density (qb, g cm3), and soil organic matter content (SOM, %) measured at 6 cm depth in the mature (MAT) and secondary (SEC) cloud forests of La Cortadura (n = 17 for each parameter). Also shown are the statistics of the calculated saturated hydraulic conductivities (Kfs, mm h1). CV is the coefﬁcient of variation (SD/mean). MAT
Mean (±SD) Median CV Min Max
1352 (988) 1002 0.73 148 3440
0.25 (0.17) 0.18 0.69 0.13 0.89
49.5 (16.8) 48.7 0.34 2.1 76.1
777 (931) 450 1.20 60 3304
1140 (1004) 749 0.88 36 3444
0.45 (0.11) 0.44 0.25 0.28 0.74
21.7 (7.7) 22.4 0.35 11.4 40.6
615 (690) 478 1.12 18 2236
pattern of monthly mean actual vapor pressure (VPA) clearly shows the dominance of dry continental air during the dry season vs. moist maritime tropical air during the wet season.
4.2. Rainfall and cloud water The signed-rank test (Hirsch et al., 1993) was used to examine differences between rainfall inputs measured at various locations in and around the study catchments (Table 2). For only three out of the nine possible combinations, daily P totals between sites were signiﬁcantly different. Furthermore, differences between rain gauges were 63% in seven of the nine cases, whereas the maximum difference between rain gauge pairs was only 5%. P measured at the weather station located above the MAT and SEC catchments (VPtg, 2400 m a.s.l.) was also very similar to that measured at the La Cortadura station (2128 m a.s.l.; Fig. 1), suggesting that P did not vary much along the elevation gradient. P measured at the SECP site was considered representative of the input to the SEC catchment, whereas P measured at station BP1 and, when possible, the average of BP1 and MATP were taken as the input to the MAT catchment. Mean annual P inputs to the MAT and SEC catchments were 3427 ± 332 (SD) mm and 3506 ± 300 (SD) mm, respectively (Table 5). Average P was 3243 mm during the ﬁrst year and 3690 mm during the second year. Rainfall showed a clear seasonal
Table 5 Average annual water balance components plus calculated/estimated standard errors (SE) for the old-growth (MAT) and regenerating (SEC) cloud forests at La Cortadura: rainfall (P), calculated cloud water interception (CWI), calculated rainfall interception loss (I), predicted transpiration (Et), streamﬂow (Q), changes in soil (DS) and groundwater (DG) storage, and leakage (L).
P CWI I Et Q DS DG L
3 427 ± (32 mm; <1%) 50 ± (51 mm; 102%) 613 ± (56 mm; 9%) 787 ± (126 mm; 16%) 1 338 ± (107 mm; 8%) 26 ± (26 mm; 100%) 11 ± (11 mm; 100%) 702 ± (223 mm; 29%)
3 506 ± (33 mm; <1%) 38 ± (46 mm; 122%) 315 ± (208 mm; 66%) 788 ± (166 mm; 21%) 1 527 ± (46 mm; 3%) 34 ± (34 mm; 100%) 4 ± (4 mm; 100%) 876 ± (350 mm; 39%)
pattern (Figs. 3a and 5a). Monthly P during the wet season (May– October) typically exceeded 400 mm, while it was about 100 mm or less during the dry season (November–April). With 862 mm vs. 507 mm, the average 2006/2007 dry season P at the two catchments was 70% higher than that of the 2007/2008 dry season (Fig. 5a). Likewise, the average 2008 wet season P (3183 mm) was 34% higher than that recorded for the 2007 wet season (2381 mm; Fig. 5a). As stated earlier, horizontal precipitation (HP) measured by the fog gauge during the wet season was composed primarily of inclined rain. During the dry season, there were numerous events with fog and drizzle (‘cloud water’ CW), as also indicated by the much higher value of the event-based ratio of HP to total precipitation (HP + P; Fig. 3a). Event rainfall characteristics are summarized in Fig. 3b. The distribution of event P intensity (calculated as total event P divided by event duration) was positively skewed, as indicated by the differences between monthly mean and median values. Mean intensity was 4.5 mm h1 for the wet season and 2.2 mm h1 for the dry season. Event intensity was highest at the start of the wet season (May, June), and gradually decreased towards the end of the wet season (October). The lowest intensities were observed in the middle of the dry season (December, January). On average, the total number of dry season events was 73, with a mean size and duration of 9.2 mm and 5.9 h, respectively. Wet season events were much more numerous (174 on average), and had higher P depths (14.4 mm) and shorter duration (3.7 h) compared to dry season events. 4.3. Rainfall and cloud water interception Fig. 4 shows rainfall vs. interception loss (I), calculated as the difference between P and the sum of throughfall (TF) and stemﬂow (SF), plotted separately for wet season events and dry season events with and without cloud water (CW). For both the wet season and dry season events without CW, there was almost no further increase in I in either forest once the canopy had been saturated by P, indicating that evaporation from the wetted canopy (E) was very low compared to corresponding rainfall rates (R). This observation agrees with calculated average E values, which were very low for both the wet (0.06 mm h1) and dry (0.05 mm h1)
L.E. Muñoz-Villers et al. / Journal of Hydrology 462–463 (2012) 53–66
Fig. 3. (a) Monthly rainfall (P) and horizontal precipitation (HP) measured at the La Cortadura weather station between November 2006 and October 2008 (right yaxis), and monthly mean values of the event-based ratio of HP to the sum of HP and P and the 25th and 75th percentiles of the HP/(HP + P) distribution (dark gray area) (left y-axis). Light gray areas indicate dry season periods. (b) Monthly mean and median event rainfall intensity (R) and the 25th and 75th percentiles of the event rainfall intensity distribution (dark gray area) (left y-axis), and monthly mean event size (right y-axis). The event rainfall intensity was calculated as total event rainfall divided by event duration.
Fig. 5. (a) Monthly rainfall (P) and calculated cloud water interception (CWI) at the mature (MAT) and secondary (SEC) cloud forests of La Cortadura between November 2006 and October 2008; gray areas indicate dry season periods. (b) Monthly calculated rainfall interception loss (I) and CWI in percent of P. (c) Monthly measured (Et) and predicted (Et P-M) transpiration. (d) Monthly streamﬂow (Q).
Fig. 4. Rainfall (P) versus measured interception loss (I) in the MAT (left panels) and SEC (right panels) cloud forests of La Cortadura for the wet (upper panels) and dry (lower panels) seasons. Closed circles represent events without cloud water (CW), whereas open circles represent events with CW. Lines represent the ﬁtted Liu (2001) rainfall interception model.
seasons, and which made up only a very small fraction (0.01–0.02) of the corresponding average R of 4.95 and 2.51 mm h1, respectively. The P-I relationships derived for the respective forests also show that the canopy interception capacity (Cm) was higher for the MAT than for the SEC, whereas Cm was higher during the wet season (full foliage) than during the dry season (semi-deciduous conditions) in either forest. For the MAT, values of Cm, obtained via a non-linear least square ﬁt of the Liu (2001) interception model (Eq. (2)) to the data (Fig. 4), were 3.88 and 1.92 mm for the wet and dry seasons, respectively. Corresponding values for the SEC were 1.92 and 1.09 mm, respectively.
As a result of the higher Cm of the MAT, total I (in percent of P) for the wet season was higher for the MAT (18%) than for the SEC (10%). Interestingly, despite decreased Cm due to the partial loss of foliage, relative I from the MAT (21%) and SEC (10%) forests for dry season events without CW remained very similar to the wet season values. This is largely explained by the smaller event sizes occurring during the dry season (Fig. 3b and 4), which increased the percentage of P intercepted. For dry season events with CW, observed values of I at the MAT (9%) and SEC (zero) forests were much lower than for dry season events without CW (Fig. 4), indicating the occurrence of cloud water interception (CWI) by the canopy. Fig. 4 further compares observed values of I with those calculated using the Liu S. model. With the relative predictive error RPE (Liu, 2001) ranging between 0.3% and 3.8%, observed and modeled I season totals matched closely for each of the two forests, both for wet season and dry season events without CW. Although the totals compared well, agreement for individual events was much less (Fig. 4). However, the errors appeared to be random in nature and were not related to rainfall amount. Once the Liu S. interception model had been calibrated for dry season events without CW, the model was used to predict I associated with dry season events with CW, so as to enable estimation of CWI from the wet canopy water balance. For both forests, CWI values derived in this way correlated reasonably well (r2 > 0.60) with corresponding HP amounts measured with the fog collector. The
L.E. Muñoz-Villers et al. / Journal of Hydrology 462–463 (2012) 53–66
linear regression equation between CWI (y) and HP (x) was y = 0.19x + 0.09 (n = 30) for the MAT and y = 0.11x + 0.29 for the SEC (n = 50) (both in mm event1). Next, the Liu S. interception model and the CWI-HP relationships were used to estimate rainfall and cloud water interception, respectively, for the 2-year water balance period (November 2006– October 2008). Compared to P, calculated inputs by CWI were very low (Fig. 5a, Table 5). Annual CWI totals derived for the 2-year period were on average 50 ± 19 mm for the MAT and 38 ± 14 mm for the SEC, i.e. only 1.5% and 1.1% of the corresponding P of 3427 and 3506 mm, respectively (Table 5). Expressed as a percentage of dry season P, CWI was 8% for the MAT and 6% for the SEC. Hence, CWI calculated for the MAT was about 1.3 times higher than that derived for the SEC. The calculated monthly I values, expressed as a percentage of P, are shown in Fig. 5b. Because of the smaller average event size during the dry season (Fig. 3b), the resulting values were somewhat higher for the dry season (21% and 11% for the MAT and SEC forests, respectively) than for the wet season (17% and 8%, respectively). However, due to additional inputs by CWI in the dry season (Fig. 5b), overall interception losses (i.e. including inputs by CWI) from the MAT (14%) and SEC (6%) forests were slightly smaller fractions of P during the dry season than during the wet season. The total overall interception losses for the 2-year water balance period were 16% for the MAT and 8% for the SEC. 4.4. Transpiration The old-growth forest had a lower number of individuals as compared to the regenerating forest (2965 vs. 4190 individuals ha1, Table 1), but the average size of the trees in the MAT was larger than in the SEC, as reﬂected by higher average dbh (20.3 vs. 16.8 cm) and basal area (48 vs. 32 m2 ha1). Interestingly, total sapwood areas were very similar for the two forests at 15 and 14 m2 ha1, respectively). Mean daily stand-scale transpiration Et for the MAT and SEC between January and October 2007 was nearly equal (2.35 ± 1.08 and 2.34 ± 1.17 mm d1, respectively). Nevertheless, patterns in monthly Et varied slightly per season (Fig. 5c), with Et in the SEC being somewhat lower than that in the MAT during the dry season months of January–April and somewhat higher during the wet season months of May–August. The Jarvis (1976) canopy conductance model explained about 50% of the variance in daily mean canopy conductance during the dry season for both forests. However, the degree of variance explained was much less for the wet season (36% and 28% for the MAT and SEC, respectively). Nevertheless, daily Et estimates (y, mm) using the Penman–Monteith equation and the canopy conductance model correlated fairly well with observed values (x, mm):
2007, caused by a strong decrease in Sin that month). The mean annual Et calculated with the P–M equation was essentially equal for the two forests (787 mm and 788 mm for the SEC and MAT, respectively; Table 5), and just slightly lower than the mean annual reference evapotranspiration ET0 of 855 mm (Allen et al., 1998). The estimated mean annual net radiation (Rn) (see ‘‘Section 3’’), expressed in units of evaporation equivalent, was 1333 mm, giving an Et/Rn ratio of 0.59 for both forests.
4.5. Streamﬂow The mean annual streamﬂow (Q) during the 2-year water balance period was higher for the SEC (1527 ± 378 mm) than for the MAT (1338 ± 357 mm) (Fig. 5d, Table 3). Expressed as a percentage of corresponding P, Q was 44% for the SEC and 39% for the MAT. Baseﬂow (Qb) accounted for 90% and 88% of the total Q in the MAT and SEC, respectively. Runoff generated during rainfall events (quickﬂow, Qd) was slightly higher in the SEC (11% and 13% of annual Q in 2006/2007 and 2007/2008, respectively) than in the MAT (10% and 9%, respectively); expressed as a percentage of P, mean Qd was 5% for the SEC and 4% for the MAT. The seasonality observed in P was clearly reﬂected in the streamﬂow patterns. However, an approximately 2-month delay in Q as compared to P was observed during the ﬁrst half of the wet season, associated with the wetting-up of the catchments (Fig. 5d). The SEC showed higher mean monthly Q values as compared to the MAT (Table 3, Fig. 5d). Expressed as a percentage of annual streamﬂow, wet season Q was 85% and 82% in the MAT and the SEC, respectively (Table 3). Mean monthly Q values for the 2008 wet season were 1.62 (MAT) and 1.57 (SEC) times higher than those for 2007 (Table 3, Fig. 5d), in agreement with the higher P (+34%) recorded in the 2008 wet season (Fig. 5a). The 2006/2007 dry season was relatively wet (Fig. 5a). As a result, the corresponding seasonal ﬂow contribution to total annual Q of that year (20% and 22% for the MAT and SEC, respectively) was roughly twice that for the dry season in the following year (10% and 14%, respectively) (Table 3, Fig. 5d). The highest Q values were observed at the height of the wet season (July–September) and the lowest at the end of the dry season (March–April) (Fig. 5d). Master recession curves with durations of 76 and 49 days for the MAT and SEC, respectively, are shown in Fig. 6. Both curves show a departure from linearity towards the end of the recession, indicating catchment leakage (L). Hence, the baseﬂow recession parameters Q0 and k (Eq. (3)) were obtained from that portion of the master recession curve where log (Qt) vs. t showed linear behavior. The recession constants (k) obtained for the MAT (0.95) and SEC (0.96) were both high and nearly equal. Consequently, both forests showed very similar baseﬂow recessions (0.12 and 0.11 mm d1 for the MAT and SEC, respectively) (Fig. 6).
Dry season : y ¼ 0:81x þ 0:26; r 2 ¼ 0:74ðMATÞ; y ¼ 0:77x þ 0:39; r 2 ¼ 0:74ðSECÞ; Wet season : y ¼ 0:87x þ 0:33; r 2 ¼ 0:84ðMATÞ; y ¼ 0:79x þ 0:52; r 2 ¼ 0:70ðSECÞ: There was also a fair agreement between the monthly measured and predicted Et, especially in the case of the old-growth forest (Fig. 5c). The RMSEs of calculated monthly Et amounted to 4 and 5 mm for the MAT and SEC, respectively, corresponding to 5% and 7% of the measured average monthly Et. Transpiration was governed mainly by solar radiation (Figs. 2a and 5c) (note the drop in measured and calculated Et in May
Fig. 6. Master recession curves, as constructed for the old-growth (MAT) and regenerating (SEC) cloud forests at La Cortadura, with ﬁtted recession equations of the form: Qt = Q0kt (see text for further explanation). Note that data are plotted on a semi-logarithmic scale.
L.E. Muñoz-Villers et al. / Journal of Hydrology 462–463 (2012) 53–66
4.6. Soil hydro-physical properties The dominant texture of the soils in the old-growth and regenerating forests was silty clay loam and silt loam, respectively. The mean bulk density (qb) of the soil at 6 cm depth in the MAT was very low (0.25 g cm3) and nearly half the value found in the SEC (0.45 g cm3) (Table 4), whereas mean soil organic matter content (SOM) was 2.2 times higher in the MAT (50%) as compared to the SEC (22%). The mean surface inﬁltration rate (fc) measured in the old-growth forest (1352 ± 998 mm h1) was slightly higher than that measured in the secondary forest (1140 ± 1004 mm h1; Table 4). The near-surface saturated hydraulic conductivity (Kfs) derived from the inﬁltration measurements was 777 ± 931 mm h1 for the MAT and 615 ± 690 mm h1 for the SEC. An ANOVA test using log-transformed Kfs values showed no signiﬁcant differences between the means of the two forests at the 0.05 level (p = 0.59). 4.7. Water balance and error analysis The average annual water balance components and their standard errors are presented in Table 5. Although variation in daily P was large, the 7-month totals (April–October 2008) measured at the various rain gauge locations (Fig. 1) were very similar, resulting in an overall SE value for P of less than 1%. The error in calculated rainfall interception (I) was much higher for the regenerating (66%) than for the mature (9%) cloud forest. This was partly due to the greater coefﬁcient of variation (CV) of the mean TF measured in the SEC (13%; based on three troughs) as compared to the MAT (3%; based on four troughs); and partly because of the lower value of I of the SEC compared to the MAT which increased the relative size of the error. Although the error in the SF estimate for the SEC was relatively large (29%), it did not contribute much to the error in I because SF was small compared to TF and P (4% of P). The uncertainty in the CWI estimates was large for both the SEC (122%) and MAT (102%) forests. The overall uncertainty in calculated stand-scale transpiration Et was 16% for the MAT and 21% for the SEC. Errors associated with the up-scaling (16% of Et for the MAT and 19% for the SEC) accounted for most of the uncertainty in overall Et estimates. Modeling errors were relatively small at 5% of Et (MAT) and 7% (SEC). The overall SE of the streamﬂow was relatively small (8% and 3% for the MAT and SEC, respectively). Total annual water yield (P + CWI-ET) for the regenerating forest (2441 mm) was 18% (364 mm) higher than that for the old-growth forest (2077 mm). This is largely explained by the difference in I (298 mm lower for the SEC), and by the difference in mean annual P (79 mm higher for the SEC). Annual leakage (L) estimated as the residual in Eq. (1) was higher for the SEC (876 mm) than for the MAT (702 mm; Table 5). Although the uncertainties in the L estimates are relatively large (29% and 39% for the MAT and SEC, respectively), the results strongly suggest that approximately one quarter of the total annual water input (P + CWI) leaves the oldgrowth (20%) and regenerating forest (25%) catchments as deep groundwater leakage. 5. Discussion 5.1. Precipitation Precipitation patterns showed marked seasonal variation (Fig. 3). The wet season (May–October) was characterized by high monthly rainfall inputs and relatively high rainfall intensities, as a result of the increased humidity (Fig. 2c) and atmospheric instability. In the dry season (November–April), precipitation was mainly produced by cold fronts, and corresponding rainfall inputs and
intensities were much lower. Although observations at this cloud forest site are only short-term (2006–2008), it would seem that the mean annual precipitation (MAP) lies well above the average of 2000 ± 830 mm reported by Jarvis and Mulligan (2010) for 560 cloud forest sites world-wide. Rainfall seasonality at La Cortadura (about 0.50, calculated using the index of Markham, 1970) is also somewhat higher than the average value of 0.32 ± 0.18 derived for all UNEP-WCMC-listed cloud forests (Jarvis and Mulligan, 2010), reﬂecting the fact that the study forests receive higher rainfall in a shorter period as compared to many other cloud forests. A further difference with other cloud forests is that the bulk of the annual rainfall appears to be associated with convective rainfall events (Fig. 3), with an average intensity (4.5 mm h1) that is comparable to values normally reported for lowland rain forest sites (4.0–5.2 mm h1; Lloyd et al., 1988; Dykes, 1997; Wallace and McJannet, 2008). It is only during the dry season that the average rainfall intensity (2.2 mm h1) resembles values found at sites with comparable types of cloud forests (1.1–2.1 mm h1; Clark et al., 1998; Hölscher et al., 2004; Fleischbein et al., 2005). As discussed in detail in Holwerda et al. (2010), the passage of cold fronts during the dry season were responsible for almost all cloud water inputs in the La Cortadura cloud forest. Average cloud water interception totals by the old-growth and secondary forest was estimated at about 50 and 38 mm year1, which was only 8% and 6% of the dry season rainfall and less than 2% of the annual rainfall. Holwerda et al. (2010) suggest low fog occurrence and low wind speeds (1.4 m s1 on average) as the most probable reasons for the low cloud water inputs observed at this site. Recent visibility measurements conﬁrm the low occurrence of fog at this site (about 10% and 20% of the time during the wet and dry seasons, respectively; unpublished data, F. Holwerda). The visibility data also show that fog events typically last only about 2 h during the wet season, which is too short to produce canopy drip. Although fog events associated with cold fronts during the winter are usually of longer duration (4 h on average), these events are accompanied by drizzle in most cases, so that the ’’cloud water’’ intercepted by the canopy is most likely a mixture of the two (Holwerda et al., 2010; cf. Scholl et al., 2007). The observed lack of frequent fog at La Cortadura is in line with the previous ﬁnding that rainfall at this site is mainly convective rather than orographic. This places the study forest at the very low end of the range of cloud forests with respect to cloud water incidence (Bruijnzeel and Proctor, 1995; Bruijnzeel, 2005). 5.2. Evapotranspiration Partly because of the low cloud water inputs, the overall rainfall interception loss (i.e. including inputs by cloud water interception) from the old-growth forest was relatively high at 16% of the annual rainfall, and compares well with values found in other lower montane cloud forests little affected by fog (19%; Bruijnzeel et al., 2010; see also discussion in Holwerda et al., 2010). The relatively high interception loss of the old-growth forest also resulted from its high canopy water storage capacity (3.9 mm in the wet season). As such, post-event evaporation of intercepted water stored in the canopy dominated overall interception loss because evaporation from the wet canopy was very low compared to corresponding rainfall rates (Fig. 4; Holwerda et al., 2010). The secondary cloud forest exhibited an overall interception loss (8%) that was half the value derived for the mature forest, primarily due to a lower canopy water storage capacity (1.9 mm in the wet season, Fig. 4). Holwerda et al. (2010) attributed the higher canopy water storage capacity of the mature forest to a higher leaf area index (Table 1) and greater epiphyte biomass. Very similar results were obtained by Ponette-González et al. (2009), who measured net precipitation in various shaded coffee plantations and cloud forest stands
L.E. Muñoz-Villers et al. / Journal of Hydrology 462–463 (2012) 53–66
between 1100 and 1600 m a.s.l. in the same area. They found that interception loss increased with leaf area index and epiphyte loading (cf. Fleischbein et al., 2005). However, it can not be ruled out that the difference in canopy water storage capacity of the mature and secondary forests was partly related to differences in speciﬁc water storage capacities (i.e. the amount of water retained per unit leaf surface area) of the dominant tree species in the respective forests, which are, in turn, related to differences in leaf morphology (Aston, 1979; Keim et al., 2006; Oesker et al., 2010). Although canopy water storage capacities of both cloud forests were markedly lower during the dry season (most likely because of partial leaf shedding), both model calculations and measurements showed an increase in percentage rainfall interception loss (Fig. 5b). This can be explained by occurrence of smaller event sizes during the dry season. However, the increase in relative rainfall interception loss during this time of year was partly compensated by inputs of cloud water, so that the overall (percentage) interception loss remained almost constant throughout the year (Fig. 5b). At 787 and 788 mm per year, the transpiration totals for the secondary and mature cloud forests derived with the Penman– Monteith equation were remarkably similar. Research in eastern Amazonia showed that very young secondary vegetation (2– 4 years old) may evaporate as much as mature forest (Hölscher et al., 1997; Giambelluca, 2002; Sommer et al., 2002). Bruijnzeel (2004) presented further evidence of the rapid recovery of water use by regenerating forest in the humid tropics, and related this to high growth rates. Due to lower radiation levels, lower temperatures, and generally more adverse soil conditions, it is expected that forest regeneration will take longer in cloud forest environments, particularly in upper montane and elﬁn cloud forests (see Bruijnzeel et al. (2010) for further discussion). The presently studied 20-year old secondary forest had a higher stem density but lower basal area compared to the old-growth forest (Table 1). A modeling study of forest regeneration in the Veracruzan cloud forest belt also showed a peak in stem number (due to high light and space availability) during the ﬁrst decades (Rüger et al., 2010). The same study further suggested that it would take 80–90 years for the total stem number and total basal area to reach their steady state (Rüger et al., 2010). Interestingly, although stem number and basal area in the secondary and mature forest at La Cortadura were very different, estimates of total sapwood area were almost equal (14 and 15 m2 ha1, respectively; Table 1). This may well be the reason that annual transpiration losses from the two forests were essential equal (Table 5; see Gómez-Cárdenas (submitted for publication) for further discussion). As indicated by the high correlations between monthly radiation inputs and monthly measured totals of transpiration (r2 = 0.81 and r2 = 0.86 for the secondary and mature cloud forests, respectively) and the relatively good ﬁt between modeled and measured transpiration, seasonal changes in transpiration were mainly explained by changes in radiation (Figs. 2a and 5b; cf. McJannet et al., 2007; García-Santos et al., 2009). Nevertheless, there was some unexplained variation (especially for the secondary forest). Considering the semi-deciduous nature of these forests, it is most likely that such variations reﬂected changes in leaf area and/or leaf physiology (cf. Giambelluca et al., 2009). At almost 800 mm per year, the total transpiration of the La Cortadura cloud forests is rather high compared to values reported for other lower montane cloud forests (620 ± 98 mm; Bruijnzeel et al., 2010). This is in line with the previous ﬁnding that fog occurrence at this site is rather low, and thus does not lead to signiﬁcant reductions in annual transpiration. A further indication for this is the high fraction of net radiation used by the studied forests for transpiration (0.59). Such a high value is more typical for lowland evergreen rain forest not affected by fog and low cloud (0.61 ± 0.08) than for lower montane cloud forest (0.48 ± 0.14;
Bruijnzeel et al., 2010). In short, total water use (overall rainfall interception + transpiration) of the studied old-growth forest (1350 mm year1) falls at the very high end of the range observed for (mature) lower montane cloud forests elsewhere (1188 ± 239 mm year1; Bruijnzeel et al., 2010), being the result of small direct (cloud water interception) and indirect (transpiration suppression) effects of fog on the forest water balance. The present results also indicate that transpiration by regenerating cloud forest relatively quickly (<20 years in this case) resembles that of oldgrowth forest, but that return to pre-disturbance rainfall interception levels takes longer (>20 years in this case) because of the slower recovery of the original canopy water storage capacity. In particular, recovery of epiphyte biomass (a major component of cloud forest canopies) after disturbance is notoriously slow (Hölscher et al., 2010). 5.3. Streamﬂow Annual water yield (deﬁned as P + CWI minus ET) from the secondary forest catchment (2441 mm) was 18% higher than that from the mature forest catchment (2077 mm). This difference mainly reﬂects the difference in rainfall interception losses, and, to a lesser extent, differences in rainfall inputs (Table 5). Mean annual streamﬂow measured over the 2-year study period was 44% of the rainfall for the SEC (1527 mm) and 39% for the MAT (1338 mm). Streamﬂow clearly followed the seasonal variability of the rainfall. The above-average rainfall recorded during the 2006/2007 dry season was likely caused by an El Niño-Southern Oscillation (ENSO) episode (Ponette-González et al., 2009; Holwerda et al., 2010) and this generated higher streamﬂow amounts than observed during the more ‘‘normal’’ 2007/2008 dry season (Table 3, Fig. 5d). Likewise, total rainfall during the 2008 rainy season was about 40% higher than that of 2007, and this produced an increase of about 60% in the monthly streamﬂow values. Streamﬂow consisted largely (90%) of baseﬂow. The high recession constants (k = 0.95 and 0.96 for the MAT and SEC, respectively) indicate that the streams draining the two catchments exhibit very slow recession rates (Nathan and McMahon, 1990; Tallaksen, 1995). This, in turn, reﬂects the high water retention capacity of the volcanic soils (Geris, 2007), as well as their considerable depth (c.f. Marín-Castro, 2010), and the high water storage capacity of the groundwater systems. The nearly equal values for groundwater outﬂow rates (s = 20 and 23 days for the MAT and SEC, respectively) further suggest very similar discharge characteristics for the respective groundwater reservoirs. The occurrence of leakage (L) was observed in the master recession curves (Fig. 6; cf. Chapman, 1999). L was also quantiﬁed by solving the water balance equation (Eq. (1)). The derived values (1.9 and 2.4 mm day1 for the MAT and SEC, respectively) were rather high, but considered plausible in view of the fractured bedrock of this volcanic terrain, as observed in bedrock cores taken in the MAT catchment (L. Muñoz-Villers and C. Gabrielli, unpublished data). The saturated hydraulic conductivity of the breccia underlying the volcanic soil in the mature forest was determined at 1 mm d1 (Karlsen, 2010), which would explain about half of the observed value of L. However, the cited value does not represent water ﬂow along faults and root channels, and therefore, it may be expected to be lower than the water-budget-based values. Gonggrijp (1941) and Bruijnzeel (2006) also inferred very high values of catchment leakage (up to 9 mm d1) from the water budgets of montane headwater catchments in similarly volcanic terrain in West Java, Indonesia, and Costa Rica, respectively. Clearly, neglecting the leakage component when calculating evapotranspiration (ET) as total rainfall minus streamﬂow (e.g. Fleischbein et al., 2006), may lead to potentially large overestimations of ET depending on the nature of the geological substrate.
L.E. Muñoz-Villers et al. / Journal of Hydrology 462–463 (2012) 53–66
The rainfall–stormﬂow runoff ratios (Qd/P) obtained for the oldgrowth (4%) and the secondary forest (5%) catchments were nearly equal, suggesting any effects of differences in rainfall interception (lower in SEC), rainfall inputs (slightly higher in SEC) and soil physical attributes to be negligible. Comparing the presently found Qd/P fractions with those reported for other steep, humid tropical montane catchments shows that values for areas underlain by deep volcanic substrates are typically <5% (Bruijnzeel, 1983, 2006; Lorup, 1998), whereas higher values (>10%) are typical of catchments that either have a much shallower hydrologically active layer (e.g. Schellekens et al., 2004; Goller et al., 2005) or signiﬁcant contributions by return ﬂow and saturation overland ﬂow in depressions (Bonell, 2005). Hence, the La Cortadura catchments fall at the very low end of the runoff generation spectrum for steep, humid forest environments. This ﬁnding can be attributed largely to the extremely high porosities and top-soil inﬁltration rates of the volcanic soils of the study catchments (cf. Tables 1 and 4; Marín-Castro, 2010). No signiﬁcant differences were found in this regard between the two forests. The average maximum 10-min rainfall intensity calculated from all 2007 and 2008 wet season rainfall events was 21 ± 25 mm h1, with the values of the 25th, 75th, and 90th percentiles being 4, 30, and 56 mm, respectively (F. Holwerda, unpublished data). Hence, top-soil inﬁltration capacities in both forests would rarely, if at all, be exceeded by the prevailing rainfall intensities. In addition, observed decreases in saturated hydraulic conductivity Kfs with soil depth (Karlsen, 2010) and the steep topography suggests subsurface stormﬂow as the prevailing mechanism of runoff generation during most rainfall events. Current research in the area combines the use of hydrological observations and various tracer approaches to increase understanding of the runoff generation process in the cloud forest catchments and in a catchment converted to pasture. The nearly equal streamﬂow characteristics of the secondary and mature forest catchments suggest that 20 years of natural regeneration after disturbance is sufﬁcient to largely restore the original hydrology in this tropical montane cloud forest region with highlypermeable volcanic soils. The hydrological response of tropical secondary vegetation is practically unknown (Bruijnzeel, 2004; cf. Ziegler et al., 2004; Zimmermann et al., 2006), with the exception of the work of Fritsch (1993), who found storm runoff from zeroorder lowland catchments (<1 ha) in French Guyana to be 16% higher 4 years after logging and clearing than under undisturbed forest. Conversely, storm runoff from a logged-over catchment not subjected to clearing and burning returned to pre-disturbance levels within 3 years of regrowth. Rates of forest regrowth, and with it the rate of hydrological recovery, also depend on the duration and intensity of the land use prior to regeneration and the associated degree of soil degradation (Ziegler et al., 2004; Zimmermann et al., 2006; Günter et al., 2007). For the secondary forest under investigation, soil conditions prior to regeneration are unknown. Hence, it remains uncertain whether the full 20-year recovery period was needed to restore hydrological behavior or to what extent this was achieved before the present observations started. At any rate, the present results highlight the importance of protecting regenerating forest to restore hydrological processes and provide environmental services to society (e.g. clean and stable water supply). However, a much more popular and rapidly expanding restoration practice in Mexico is the planting of rapidly growing non-native tree species (i.e. Pinus patula) (Pagiola, 2002; Muñoz-Piña et al., 2008). The associated net hydrological impacts of this are as yet unknown.
forest in central Veracruz, Mexico over a 2-year period. Rainfall (P) and streamﬂow (Q) were measured continuously, whereas dry canopy evaporation (transpiration Et), wet canopy evaporation (rainfall interception I), and cloud water interception (CWI) were quantiﬁed using a combination of ﬁeld measurements and modeling. The main ﬁndings and conclusions of the investigation can be summarized as follows: Total water use of the old-growth forest due to overall I (560 mm year1, 16% of P) and Et (790 mm year1) was high at 1350 mm year1, being the result of small direct (CWI 62% of P) and indirect (Et suppression) effects of fog on the forest water balance. The results also indicate that Et by regenerating cloud forest (790 mm year1) relatively quickly (<20 years in this case) resembles that of old-growth forest, but that return to pre-disturbance I (8% vs. 16% of P) takes longer (>20 years in this case) because of the slower recovery of the original canopy water storage capacity. Total annual Q was somewhat higher in the regenerating (1527 mm year1, 44% of P) than in the old-growth (1338 mm year1, 39% of P) forest, mainly because of differences in I. For both forests, Q consisted mostly of baseﬂow (90%), and the very low storm runoff response to rainfall (Qd/P = 4% and 5% for the old-growth and regenerating forests, respectively) was attributed to the extremely high top-soil inﬁltration rates of the volcanic soils. The very similar streamﬂow characteristics of the two forests indicates no differences in their soil physical and hydraulic properties, suggesting that 20 years of natural regeneration is capable of producing nearoriginal hydrological behavior. These ﬁndings underscore the importance of protecting and promoting naturally regenerating forest to restore hydrological processes of ecosystems, especially as a means of providing key environmental services to society. Acknowledgements The authors would like to thank the Municipality of Coatepec, Veracruz, and the inhabitants of Loma Alta, Coatepec for allowing us to work in their land properties. We gratefully acknowledge the Instituto de Ecología, A.C., Xalapa, Ver. for their operational and logistical support, the Government of Veracruz for transporting the equipment to the ﬁeld site, and the National Forestry Commission of Mexico (CONAFOR – Gerencia Regional X Golfo Centro) for helping us with the storage of equipment. We would like to specially thank our technicians Hans Bakker and Ron Lootens (VU University), and Sergio Cruz Martínez and Adán Hernández Hernández for their great help at various stages of the project. We also thank Leónides, Eusebia, Jerónimo, Sabino and José Luis Martínez, Enrique Meza and Cristian Delfín for their ﬁeld assistance. Finally, we thank Ignacio Contreras for his unconditional support throughout the project. Funding was provided by the Consejo Nacional de Ciencia y Tecnología (CONACyT, Grants Nos. 149367, 43082, 58185), the Instituto de Ecología, A.C. (Grants Nos. GBB/10093, MEZ/10024), the Netherlands Foundation for the Advancement of Tropical Research (WOTRO, Grant No. W76-252), the National Institute of Ecology of Mexico (INE, Grant No. INE/A1-064/2007), and the National Science Foundation of the United States (NSF, Grant No. NSF/DEB0746179). References
6. Conclusions The present study compared the catchment water balances of an old-growth and a 20-year-old secondary lower montane cloud
Aide, T.M., Ruiz-Jaen, M.C., Grau, H.R., 2010. What is the state of tropical montane cloud forest restoration. In: Bruijnzeel, L.A., Scatena, F.N., Hamilton, L.S. (Eds.), Tropical Montane Cloud Forests: Science for Conservation and Management. Cambridge University Press, pp. 101–110.
L.E. Muñoz-Villers et al. / Journal of Hydrology 462–463 (2012) 53–66 Allen, R.G., Pereira, L.S., Raes, D., Smith, M., 1998. Crop evapotranspiration – guidelines for computing crop water requirements. FAO Irrigation and drainage paper 56. . Aston, A.R., 1979. Rainfall interception by 8 small trees. Journal of Hydrology 42 (3– 4), 383–396. Baez, A.P., Padilla, H., Cervantes, J., Pereyra, D., Belmont, R., 1997. Rainwater chemistry at the eastern ﬂanks of the Sierra Madre Oriental, Veracruz, Mexico. Journal of Geophysical Research – Atmospheres 102 (D19), 23329–23336. Bonell, M., 2005. Runoff generation in tropical forest. In: Bonell, M., Bruijnzeel, L.A. (Eds.), Forests, Water and People in the Humid Tropics. Cambridge University Press, Cambridge, pp. 314–406. Bruijnzeel, L.A., 1983. Hydrological and biogeochemical aspects of man-made forests in South-Central Java, Indonesia. PhD Thesis, Vrije Universiteit, Amsterdam, 256 pp. Bruijnzeel, L.A., 2001. Hydrology of tropical montane cloud forests: a reassessment. Land Use and Water Resources Research, 1.1–1.18. Bruijnzeel, L.A., 2004. Hydrological functions of tropical forests: not seeing the soil for the trees? Agriculture Ecosystems and Environment 104 (1), 185–228. Bruijnzeel, L.A., 2005. Tropical montane cloud forest: a unique hydrological case. In: Bonell, M., Bruijnzeel, L.A. (Eds.), Forests, Water and People in the Humid Tropics. Cambridge University Press, Cambridge, pp. 462–483. Bruijnzeel, L.A. (compiler), 2006. Hydrological impacts of converting tropical montane cloud forest to pasture, with initial reference to northern Costa Rica. Final Technical Report DFID-FRP Project no. R7991. VU University Amsterdam and Forestry Research Programme of the UK – Department for International Development, Aylesford, 52 pp. . Bruijnzeel, L.A., Proctor, J., 1995. Hydrology and biogeochemistry of tropical montane cloud forests: what do we really know? In: Hamilton, L.S., Juvik, J.O., Scatena, F.N. (Eds.), Tropical Montane Cloud Forests. Ecological Studies, vol. 110. Springer Verlag, New York, pp. 38–78. Bruijnzeel, L.A., Kappelle, M., Mulligan, M., Scatena, F.N., 2010. Tropical montane cloud forests: state of knowledge and sustainability perspectives in a changing world. In: Bruijnzeel, L.A., Scatena, F.N., Hamilton, L.S. (Eds.), Tropical Montane Cloud Forests: Science for Conservation and Management. Cambridge University Press, pp. 691–700. Calder, I.R., 1993. Hydrologic effects of land-use change. In: Maidment, D.R. (Ed.), Handbook of Hydrology. McGraw-Hill, Inc., New York, US, pp. 13.1–13.50. Calder, I.R., Kidd, C.H.R., 1978. A note on the dynamic calibration of tipping-bucket gauges. Journal of Hydrology 39 (3–4), 383–386. Campos, A., 2008. Evaluación de la calidad del suelo, Instituto de Ecología, A.C. – CONACyT, Xalapa, Ver., México. Chapman, T., 1999. A comparison of algorithms for stream ﬂow recession and baseﬂow separation. Hydrological Processes 13, 701–714. Clark, K.L., Nadkarni, N.M., Schaefer, D., Gholz, H.L., 1998. Atmospheric deposition and net retention of ions by the canopy in a tropical montane forest, Monteverde, Costa Rica. Journal of Tropical Ecology 14, 27–45. Clearwater, M.J., Meinzer, F.C., Andrade, J.L., Goldstein, G., Holbrook, N.M., 1999. Potential errors in measurement of nonuniform sap ﬂux using heat dissipation probes. Tree Physiology 19 (10), 681–687. Cohen, Y., Fuchs, M., Green, G.C., 1981. Improvement of the he at pulse method for determining sap ﬂow in trees. Plant Cell and Environment 4 (5), 391–397. Dolman, A.J., Gash, J.H.C., Roberts, J., Shuttleworth, W.J., 1991. Stomatal and surface conductance of tropical rainforest. Agricultural and Forest Meteorology 54, 303–318. Dykes, A.P., 1997. Rainfall interception from a lowland tropical rainforest in Brunei. Journal of Hydrology 200 (1–4), 260–279. Elrick, D.E., Reynolds, W.D., 1992. Inﬁltration from constant-head well permeameters and inﬁltrometers. In: Topp, G.C., Reynolds, W.D., Green, R.E. (Eds.), Advances in Measurement of Soil Physical Properties: Bringing Theory into Practice. Soil Science Society of America, Madison, WI, US, pp. 1–24. FAO-UNESCO, 1997. Soil Map of the World. Revised Legend, with corrections and updates. World Soil Resources Report 60, FAO, Rome. Reprinted with updates as Technical Paper 20, ISRIC, Wageningen, The Netherlands. Fleischbein, K., Wilcke, W., Goller, R., Boy, J., Valarezo, C., Zech, W., Knoblich, K., 2005. Rainfall interception in a lower montane forest in Ecuador: effects of canopy properties. Hydrological Processes 19 (7), 1355–1371. Fleischbein, K., Wilcke, W., Valarezo, C., Zech, W., Knoblich, K., 2006. Water budgets of three small catchments under montane forest in Ecuador: experimental and modeling approach. Hydrological Processes 20, 2491–2507. Fox, J., Truong, D.M., Rambo, A.T., Tuyen, N.P., Cuc, L.T., Leisz, S., 2000. Shifting cultivation: a new old paradigm for managing tropical forests. Bio Science 50, 521–528. Fritsch, J.M., 1992. Les Effects du Défrichement de la Forêt Amazonienne et de la Mise en Culture sur l’Hydrologie des Petits Bassins Versants. ORSTOM Editions, Paris, p. 392. Fritsch, J.M., 1993. The hydrological effects of clearing tropical rain forest and of the implementation of alternative land uses. In: Gladwell, J.S. (Ed.), Hydrology of Warm Humid Regions, IAHS Publications 216, pp. 53–66. Frumau, K.F.A., Bruijnzeel, L.A., Tobon, C., 2006. Hydrological measurement protocol for montane cloud forest. Annex 2, Final Technical Report DFID-FRP Project R7991. VU University, Amsterdam, 105 pp. . Frumau, K.F.A., Burkard, R., Schmid, S., Bruijnzeel, L.A., Tobon, C., Calvo, J.C., 2010. Fog gauge performance under conditions of fog and wind-driven rain. Hydrological Processes 26, 293–301. Garcia, E., 1988. Modiﬁcaciones al sistema de clasiﬁcación climática de Köppen. Offset Larios, México, D.F., México.
García-Coll, I., Martínez-Otero, A., Ramírez-Soto, A., Niño-Cruz, A., Juan-Rivas, A., Domínguez-Barrada, L., 2004. La Relación Agua-Bosque: Delimitación de zonas prioritarias para el pago de servicios hidrológicos en la Cuenca del río Gavilanes, Coatepec, Veracruz. In: Cotler, H. (Ed.), El Manejo Integral de Cuencas en México. Estudios y reﬂexiones para orientar la política ambiental. Secretaría de Medio Ambiente y Recursos Naturales, Instituto Nacional de Ecología, México D.F., México, pp. 99–115. García-Franco, J.G., Castillo-Campos, G., Mehltreter, K., Martínez, M.L., Vázquez, G., 2008. Composición ﬂorística de un bosque mesóﬁlo del centro de Veracruz, México. Boletin de la Sociedad Botánica de México 83, 37–52. García-Santos, G., 2007. An ecohydrological and soils study in a montane cloud forest in the National Park of Garajonay, La Gomera (Canary Islands, Spain). PhD Thesis. VU University, Amsterdam. . García-Santos, G., Bruijnzeel, L.A., Dolman, A.J., 2009. Modeling canopy conductance under wet and dry conditions in a subtropical cloud forest. Agricultural and Forest Meteorology 149, 1565–1572. Gash, J.H.C., 1979. An analytical model of rainfall interception by forests. Quarterly Journal of the Royal Meteorological Society 105 (443), 43–55. Geris, J., 2007. Changes in soil physical characteristics, inﬁltration and hill slope hydrological response associated with forest conversion to pasture in Central Veracruz, Mexico. MSc Thesis, VU University, Amsterdam, The Netherlands, 82 pp. Available upon request from the authors. Giambelluca, T.W., 2002. Hydrology of altered tropical forest. Hydrological Processes 16, 1665–1669. Giambelluca, T.W., Martin, R.E., Asner, G.P., Huang, M., Mudd, R.G., Nullet, M.A., DeLay, J.K., Foote, D., 2009. Evapotranspiration and energy balance of native wet montane cloud forest in Hawaii. Agricultural and Forest Meteorology 149, 230– 243. Goller, R., Wilcke, W., Leng, M.J., Tobschall, H.J., Wagner, K., Valarezo, C., Zech, W., 2005. Tracing water paths through small catchments under a tropical montane rain forest in south Ecuador by an oxygen isotope approach. Journal of Hydrology 308, 67–80. Gómez-Cárdenas, M., Asbjornsen, H., Holwerda, F., Bruijnzeel, L.A. in preparation. Whole-tree and stand transpiration in an old-growth, regenerating, and reforested Mexican montane cloud forest, submitted for publication. Gómez-Cárdenas, M., Asbjornsen, H., Holwerda, F., Bruijnzeel, L.A. submitted for publication. Sap ﬂux patterns in dominant tree species in a mature forest, pine reforestation, and naturally generating forest in a seasonal montane cloud forest zone in eastern Mexico. Gómez-Tagle, C.A., Gómez-Tagle, R.A.F., Batlle-Sales, J., Zepeda-Castro, H., GuevaraSantamaría, M.A., Maldonado-López, S., Pintor, J.E., 2008. Conductividad hidráulica saturada de campo: uso de un inﬁltrómetro de carga constante y anillo sencillo. Terra Latinoamericana 26, 287–297. Gonggrijp, L., 1941. Verdamping van een gebergtebos in West Java op 1750–2000 m hoogte boven zeeniveau (Evaporation of montane forest at 1750–2000 m a.s.l. in West Java). Tectona 34, 437–447. in Dutch. Granier, A., 1985. Une nouvelle méthode pour la mesure du ﬂux de sève brute dans le tronc des arbes. Annals of Forest Science 42 (2), 193–200. Granier, A., Biron, P., Lemoine, D., 2000. Water balance, transpiration and canopy conductance in two beech stands. Agricultural and Forest Meteorology 100, 291–308. Günter, S., Weber, M., Erreis, R., Aguirre, N., 2007. Inﬂuence of distance to forest edges on natural regeneration of abandoned pastures: a case study in the tropical mountain rain forest of Southern Ecuador. European Journal of Forest Research 126, 67–75. Harr, R.D., 1982. Fog-drip in the bull Run Municipal Watershed, Oregon. Water Resources Bulletin 18 (5), 785–789. Harris, P.P., Huntingford, C., Cox, P.M., Gash, J.H.C., Malhi, Y., 2004. Effects of soil moisture on canopy conductance of Amazonian rainforest. Agricultural and Forest Meteorology 122, 215–227. Hewlett, J.D., Hibbert, A.R., 1967. Factors affecting the response of small watersheds to precipitation in humid areas. In: Sopper, W.E., Lull, H.W. (Eds.), Forest Hydrology. Pergamon Press, pp. 275–290. Hirsch, R.M., Helsel, D.R., Cohn, T.A., Gilroy, E.J., 1993. Statistical analysis of hydrologic data. In: Maidment, D.R. (Ed.), Handbook of Hydrology. McGraw-Hill Inc., New York, US, pp. 17.1–17.55. Holder, C.D., 2003. Fog precipitation in the Sierra de las minas biosphere reserve, Guatemala. Hydrological Processes 17 (10), 2001–2010. Hölscher, D., de A. Sá, T.D., Bastos, T.X., Denich, M., Fölster, H., 1997. Evaporation from young secondary vegetation in eastern Amazonia. Journal of Hydrology 193, 293–305. Hölscher, D., Köhler, L., van Dijk, A.I.J.M., Bruijnzeel, L.A., 2004. The importance of epiphytes to total rainfall interception by a tropical montane rain forest in Costa Rica. Journal of Hydrology 292 (1–4), 308–322. Hölscher, D., Mackensen, J., Roberts, J.M., 2005. Forest recovery in the humid tropics: changes in vegetation structure, nutrient pools and the hydrological cycle. In: Forests, Water and People in the Humid Tropics. Cambridge University Press, Cambridge, pp. 598–621. Hölscher, D., Köhler, L., Kappelle, M., Leuschner, Ch., 2010. Ecology and use of oldgrowth and recovering montane oak forests in the Cordillera de Talamanca, Costa Rica. In: Bruijnzeel, L.A., Scatena, F.N., Hamilton, L.S. (Eds.), Tropical Montane Cloud Forests: Science for Conservation and Management. Cambridge University Press, pp. 610–617. Holwerda, F., Bruijnzeel, L.A., Muñoz-Villers, L.E., Equihua, M., Asbjornsen, H., 2010. Rainfall and cloud water interception in mature and secondary lower montane cloud forests of central Veracruz, Mexico. Journal of Hydrology 384, 84–96.
L.E. Muñoz-Villers et al. / Journal of Hydrology 462–463 (2012) 53–66
INEGI, 1993. Land use/cover digital data. Scale 1: 250 000. Series II. INEGI, Mexico. INEGI, 2000. Topographic-land use map. Scale 1: 50 000. INEGI, Mexico. Ingwersen, J.B., 1985. Fog drip, water yield and timber harvesting in the Bull Run Municipal Watershed, Oregon. Water Resources Bulletin 21, 469–473. Jarvis, P.G., 1976. The interpretation of the variations in leaf water potential and stomatal conductance found in canopies in the ﬁeld. Philosophical Transactions of the Royal Society of London. Series B. Biological Sciences 273, 593–610. Jarvis, A., Mulligan, M., 2010. The climate of cloud forests. In: Bruijnzeel, L.A., Scatena, F.N., Hamilton, L.S. (Eds.), Tropical Montane Cloud Forests: Science for Conservation and Management. Cambridge University Press, pp. 39–56. Juvik, J.O., Ekern, P.C., 1978. A climatology of mountain fog on Mauna Loa, Hawaii island. Water Resources Research Center. University of Hawaii, Honolulu. Karlsen, R., 2010. Stormﬂow processes in a mature tropical montane cloud forest catchment, Coatepec, Veracruz, Mexico. MSc thesis, VU University, Amsterdam, The Netherlands. Karmalkar, A.V., Bradley, R.S., Diaz, H.F., 2008. Climate change scenario for Costa Rican montane forests. Geophysical Research Letters 35 (11). Keim, R.F., Skaugset, A.E., Weiler, M., 2006. Storage of water on vegetation under simulated rainfall of varying intensity. Advances in Water Resources 29 (7), 974–986. Kindsvater, C.E., Carter, R.W.C., 1957. Discharge characteristics of rectangular thinplate weirs. Journal of the Hydraulics Division, Proceedings of the American Society of Civil Engineers, vol. 83 (HY6), Paper 1453. Kumagai, T., Tateishi, M., Shimizu, T., Otsuki, K., 2008. Transpiration and canopy conductance at two slope positions in a Japanese cedar forest watershed. Agricultural and Forest Meteorology 148, 1444–1455. Lawton, R.O., Nair, U.S., Pielke, R.A., Welch, R.M., 2001. Climatic impact of tropical lowland deforestation on nearby montane cloud forests. Science 294 (5542), 584–587. Liu, S.G., 2001. Evaluation of the Liu model for predicting rainfall interception in forests world-wide. Hydrological Processes 15 (12), 2341–2360. Lloyd, C.R., Gash, J.H.C., Shuttleworth, W.J., Marques, A.D., 1988. The measurement and modeling of rainfall interception by Amazonian rain forest. Agricultural and Forest Meteorology 43 (3–4), 277–294. Lorup, J.K., 1998. Effects on land use on the water resources in the southern Tanzanian highlands. PhD Thesis, Institute of Hydrodynamics and Water Resources, Technical University of Denmark, 201 pp. Lu, P., Urban, L., Zhao, P., 2004. Granier’s thermal dissipation probe (TDP) method for measuring sap ﬂow in trees: theory and practice. Acta Botanica Sinica 46 (6), 631–646. Lundblad, M., Lagergren, F., Lindroth, A., 2001. Evaluation of he at balance and heat dissipation methods for sapﬂow measurements in pine and spruce. Annals of Forest Science 58 (6), 625–638. Marín-Castro, B.E., 2010, Variación espacial de la conductividad hidráulica saturada en suelos de origen volcánico bajo tres usos de suelo en el centro de Veracruz, México. Tesis de Maestría. Posgrado en Ciencias, Instituto de Ecología, A.C., Xalapa, Veracruz, México. . Markham, C.G., 1970. Seasonality of Precipitation in the United States. Annals of the Association of American Geographers 60 (3), 593–597. McJannet, D., Fitch, P., Disher, M., Wallace, J., 2007. Measurements of transpiration in four tropical rainforest types of north Queensland, Australia. Hydrological Processes 21 (26), 3549–3564. Metcalfe, S.E., 1987. Historical data and climatic change in Mexico – a review. Geographical Journal 153, 211–222. Monteith, J.L., 1965. Evaporation and environment. In: The State and Movement of Water in Living Organisms. In: Proceedings of the 19th Symposium Soc. Exp. Biol., Swansea 1964. Academic Press, for The Society for Experimental Biology, UK, pp. 205–234. Mulligan, M., 2010. Modelling the tropics-wide extent and distribution of cloud forest and cloud forest loss, with implications for conservation priority. In: Bruijnzeel, L.A., Scatena, F.N., Hamilton, L.S. (Eds.), Tropical Montane Cloud Forests: Science for Conservation and Management. Cambridge University Press, pp. 14–38. Muñoz-Piña, C., Guevara, A., Torres, J.M., Braña, J., 2008. Paying for the hydrological services of Mexico’s forests: analysis, negotiations and results. Ecological Economics 65 (4), 725–736. Muñoz-Villers, L.E., 2008. Efecto del cambio en el uso de suelo sobre la dinámica hidrológica y calidad de agua en el trópico húmedo del centro de Veracruz, México. PhD Thesis, Universidad Autónoma Metropolitana, México D.F., México. . Muñoz-Villers, L.E., López-Blanco, J., 2007. Land use/cover changes using Landsat TM/ETM images in a tropical and biodiverse mountainous area of centraleastern Mexico. International Journal of Remote Sensing 29 (1), 71–93. Nathan, R.J., McMahon, T.A., 1990. Evaluation of automated techniques for baseﬂow and recession analysis. Water Resources Research 6, 1465–1473. Nespor, V., Sevruk, B., 1999. Estimation of wind-induced error of rainfall gauge measurements using a numerical simulation. Journal of Atmospheric and Oceanic Technology 16 (4), 450–464. Oesker, M., Homeier, J., Dalitz, H., Bruijnzeel, L.A., 2010. Spatial heterogeneity of throughfall quantity and quality in tropical montane forests in southern Ecuador. In: Bruijnzeel, L.A., Scatena, F.N., Hamilton, L.S. (Eds.), Tropical
Montane Cloud Forests: Science for Conservation and Management. Cambridge University Press, pp. 393–401. O’Grady, A.P., 2000. Patterns of tree and stand water use in the eucalypt open forests of northern Australia. PhD Thesis, Northern Territory University, Darwin, Australia. Pagiola, S., 2002. Paying for Water Services in Central America: Learning from Costa Rica. In: Pagiola, S., Bishop, J., Landell-Mills, N. (Eds.), Selling Forest Environmental Services: Market-based Mechanisms for Conservation and Development. Earthscan, London. Ponette-González, A.G., Weathers, K.C., Curran, L.M., 2009. Water inputs across a tropical montane landscape in Veracruz, Mexico: synergistic effects of land cover, rain and fog seasonality, and interannual precipitation variability. Global Change Biology 16 (3), 946–963. Rüger, N., Williams-Linera, G., Huth, A., 2010. Modeling the dynamics of tropical montane cloud forest in central Veracruz, Mexico. In: Bruijnzeel, L.A., Scatena, F.N., Hamilton, L.S. (Eds.), Tropical Montane Cloud Forests: Science for Conservation and Management. Cambridge University Press, pp. 584–594. Rutter, A.J., Robins, P.C., Morton, A.J., Kershaw, K.A., 1972. A predictive model of rainfall interception in forests, 1. Derivation of the model from observations in a plantation of Corscian pine. Agricultural Meteorology 9 (5–6), 367–384. Rzedowski, J., 1996. Análisis preliminar de la ﬂora vascular de los bosques mesóﬁlos de montaña de México. Acta Botanica Mexicana 35, 25–44. Scatena, F.N., Bruijnzeel, L.A., Bubb, P., Das, S., 2010. Setting the stage. In: Bruijnzeel, L.A., Scatena, F.N., Hamilton, L.S. (Eds.), Tropical Montane Cloud Forests: Science for Conservation and Management. Cambridge University Press, pp. 3–13. Schellekens, J., 2006. CQ-FLOW: a distributed hydrological model for the prediction of impacts of land-cover change, with special reference to the Rio Chiquito catchment, northwest Costa Rica. Annex 3, Final Technical Report DFID-FRP Project R7991. VU University Amsterdam. . Schellekens, J., Bruijnzeel, L.A., Scatena, F.N., Bink, N.J., Holwerda, F., 2000. Evaporation from a tropical rain forest, Luquillo Experimental Forest, eastern Puerto Rico. Water Resources Research 36 (8), 2183–2196. Schellekens, J., Scatena, F.N., Bruijnzeel, L.A., Van Dijk, A.I.J.M., Groen, M.M.A., Van Hogezand, R.J.P., 2004. Stormﬂow generation in a small rainforest catchment in the Luquillo Experimental Forest, Puerto Rico. Hydrological Processes 18 (3), 505–530. Scholl, M.A., Giambelluca, T.W., Gingerich, S.B., Nullet, M.A., Loope, L.L., 2007. Cloud water in windward and leeward mountain forests: the stable isotope signature of orographic cloud water. Water Resources Research 43 (12). Sharon, D., 1980. The distribution of hydrologically effective rainfall incident on sloping ground. Journal of Hydrology 46 (1–2), 165–188. Shuttleworth, W.J., 1988. Evaporation from Amazonian rainforest. Proceedings of the Royal Society of London. Series B, Biological Sciences 233, 321–346. Sommer, R., de Abreu Sá, T.D., Vielhauer, K., Carioca de Araújo, A., Fölster, H., Vlek, P.L.G., 2002. Transpiration and canopy conductance of secondary vegetation in the eastern Amazon. Agricultural and Forest Meteorology 112, 103–121. Stewart, J.B., 1988. Modeling surface conductance of pine forest. Agricultural and Forest Meteorology 43, 19–35. Still, C.J., Foster, P.N., Schneider, S.H., 1999. Simulating the effects of climate change on tropical montane cloud forests. Nature 398 (6728), 608–610. Takahashi, M., Giambelluca, T.W., Nullet, D.A., in press. Rainfall partitioning and cloud water interception in native and invaded forest in Hawaii’s Volcano National Park. Hydrological Processes 26. Tallaksen, L.M., 1995. A review of baseﬂow recession analysis. Journal of Hydrology 165, 349–370. Toebes, C., Strang, D.D., 1964. On recession curves 1 – recession equations. Journal of Hydrology (New Zealand) 3, 2–14. Uhl, C., Buschbacher, R., Serrão, E.A.S., 1988. Abandoned pastures in eastern Amazonia. I. Patterns of plant succession. Journal of Ecology 76 (3), 663–681. Wallace, J., McJannet, D., 2008. Modelling interception in coastal and montane rainforests in northern Queensland, Australia. Journal of Hydrology 348 (3–4), 480–495. Waterloo, M.J., Schellekens, J., Bruijnzeel, L.A., Rawaqa, T.T., 2007. Changes in catchment runoff after harvesting and burning of a Pinus caribaea plantation in Viti Levu, Fiji. Forest Ecology and Management 251, 31–34. Wu, L., Pan, L., Mitchell, J., Sanden, B., 1999. Measuring saturated hydraulic conductivity using a generalized solution for single-ring inﬁltrometers. Soil Science Society of America Journal 63 (4), 788–792. Wullschleger, S.D., Hanson, P.J., Todd, D.E., 2001. Transpiration from a multi-species deciduous forest as estimated by xylem sap ﬂux techniques. Forest Ecology and Management 143, 205–213. Xu, J., Fox, J., Xing, L., Podger, N., Leisz, S., Xihui, A., 1999. Effects of swidden cultivation, population growth, and state policies on land cover in Yunnan, China. Mountain Research and Development 19, 123–132. Zadroga, F., 1981. The hydrological Importance of a Montane Cloud Forest Area of Costa Rica. Tropical Agricultural Hydrology. John Wiley and Sons. pp. 59–73. Ziegler, A.D., Giambelluca, T.W., Tran, L.T., Vana, T.T., Nullet, M.A., Fox, J., Duc Vien, T., Pinthong, J., Maxwell, J.F., Evett, S., 2004. Hydrological consequences of landscape fragmentation in mountainous northern Vietnam: evidence of accelerated overland ﬂow generation. Journal of Hydrology 287, 124–146. Zimmermann, B., Elsenbeer, H., De Moraes, J.M., 2006. The inﬂuence of land-use changes on soil hydraulic properties: implications for runoff generation. Forest Ecology and Management 222 (1–3), 29–38.