You are on page 1of 9

This article appeared in a journal published by Elsevier.

The attached copy is furnished to the author for internal non-commercial research and education use, including for instruction at the authors institution and sharing with colleagues. Other uses, including reproduction and distribution, or selling or licensing copies, or posting to personal, institutional or third party websites are prohibited. In most cases authors are permitted to post their version of the article (e.g. in Word or Tex form) to their personal website or institutional repository. Authors requiring further information regarding Elseviers archiving and manuscript policies are encouraged to visit: http://www.elsevier.com/authorsrights

Author's personal copy

Forest Ecology and Management 302 (2013) 5461

Contents lists available at SciVerse ScienceDirect

Forest Ecology and Management


journal homepage: www.elsevier.com/locate/foreco

Patterns of stocks of aboveground tree biomass, dynamics, and their determinants in secondary Andean forests
Miguel A. Pea a,b, Alvaro Duque c,
a

Doctorado en Ecologa, Universidad Nacional de Colombia Sede Medelln, Calle 59A # 63-20, Medelln, Colombia Instituto de Hidrologa, Meteorologa y Estudios Ambientales de Colombia (IDEAM), Bogot D.C., Colombia c Departamento de Ciencias Forestales, Universidad Nacional de Colombia Sede Medelln, Calle 59A # 63-20, Medelln, Colombia
b

a r t i c l e

i n f o

a b s t r a c t
In this study, we aimed to identify the main factors that affect the dynamics and rates of aboveground biomass (AGB) accumulation in secondary forests located in the interandean valleys of Colombia. We used two censuses performed on trees with a diameter at breast height of P10 cm in 10 1-ha plots to answer the following questions: (1) Does prior use of the land for raising crops (instead of pastureland for cattle ranching) promote an increase in the rate of AGB accumulation? (2) To what extent did climate, soil characteristics, the functional traits of resident species, and/or stand age affect the stocks and dynamics of the AGB in these secondary forests? We did not nd any evidence that prior land use for cattle ranching had a negative effect on the tree AGB-accumulation rate, when compared with land previously used for raising crops. Greater AGB stocks at an intermediate fertility point may be associated with a higher abundance of tall, short-lived, fast-growth pioneer species. Tree mortality rates in secondary forests were primarily determined by the forests composition and the variation in the amount of rainfall between sites. Soil fertility, which is known to accelerate the growth rate of plant species, was signicantly and positively associated with the rate of AGB recruitment. We did not nd any evidence for a negative correlation between wood density and growth, as has been reported for primary tropical forests. The net AGB change (%) was primarily associated with the low soil-fertility characteristics of sandy well-drained soils that could, in the presence of steep slopes, increase forest dynamics. Land restoration that restores the natural cycle of secondary forests could assist in the mitigation of global warming by promoting key environmental functions, such as carbon sequestration. 2013 Elsevier B.V. All rights reserved.

Article history: Received 30 December 2012 Received in revised form 17 March 2013 Accepted 19 March 2013

Keywords: Facilitation Functional traits Soil fertility Variance partitioning Vegetation dynamics Global change

1. Introduction Secondary succession can be dened as the long-term, directional change in both forest structure and species composition that follows a large-scale disturbance event (Chazdon, 2008). Fostering the management and development of secondary forests in tropical regions is a promising approach for the reduction of greenhouse gas concentrations and thus could be an important means of mitigating climate change (Brown and Lugo, 1990; Chazdon and Coe, 1999; Grace, 2004; Sierra et al., 2007; Silver et al., 2000). However, the underlying causes of the high variability in the capability of secondary forests to recover aboveground biomass (AGB) are not fully understood. Variation in the rate of AGB accumulation have commonly been associated with land use history, soil fertility, climatic characteristics, and the structure and degree of forest con Corresponding author. Tel.: +57 4 4309089; fax: +57 4 4309079.

E-mail addresses: mapenah@unal.edu.co (M.A. Pea), ajduque09@gmail.com, ajduque@unal.edu.co (A. Duque). 0378-1127/$ - see front matter 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.foreco.2013.03.025

servation within the landscape matrix (Chazdon, 2008; Guariguata and Ostertag, 2001; Kauffman et al., 2009; Silver et al., 2000). Prior use of land for raising crops rather than for pasturing cattle seems to speed up the rate of growth and the yield of the natural succession (Ferguson et al., 2003; Zimmerman et al., 2007). Likewise, soil fertility and higher temperatures may facilitate the recovery of AGB, and recovery may also accelerate with proximity to seed sources (Chazdon, 2008; Guariguata and Ostertag, 2001; Lu et al., 2002). The use of chronosequences to characterize the patterns of succession in stands of different ages is one of the most widely employed tools for assessing the rate of AGB accumulation in secondary forests (Letcher and Chazdon, 2009). However, the linear approach of changes in the AGB accumulation rate assessed using chronosequences does not enable researchers to identify the ecological processes that cause the observed, non-linear, incremental changes in the AGB over time (Chazdon, 2008; Chazdon et al., 2007, 2010; Silver et al., 2000). Because a fundamental premise of functional ecology concerns the level of relationship

Author's personal copy

M.A. Pea, A. Duque / Forest Ecology and Management 302 (2013) 5461

55

between readily observed plant traits and unobserved characteristics (i.e., growth rate, survival or death), a signicant portion of the non-linear component of the AGB-accumulation rate is expected to be controlled by changes in species performance due to functional species turnover over time (Chazdon et al., 2010; Oliver and Larson, 1990; Rees et al., 2001). Hence, the study of vegetation dynamics should help identify the ecological lters imposed by the interaction between environmental gradients and species ecological requirements throughout the successional process (Chazdon et al., 2007, 2010; Wright et al., 2010). This study aimed to identify the main factors that affect variation in the rate of AGB accumulation in secondary forests located in the interandean valleys and mountains of Colombia. We used two censuses carried out on 10 1-ha plots in forests whose ages ranged between 15 and 28 years to answer the following questions. (1) Does a prior use of land for raising crops (in comparison to pastureland for cattle ranching) promote an increase in the rate of AGB accumulation? If so, places formerly employed for raising crops should show higher AGB stocks at a similar age. (2) To what extent do climate, soil characteristics, species functional traits, and/or stand-age determine the stocks and dynamics of the AGB in these secondary forests? We expect that (i) older forests on richer soils may store more AGB; (ii) mortality rates may increase with stand-age and soil fertility; (iii) the AGB recruitment rates may decrease with standage and increase with soil fertility; (iv) the AGB rate of growth of trees may be higher in richer soils; (v) the net change in AGB (%) may be positively correlated with soil fertility and negatively correlated with stand-age. In all cases, we assumed that the presence of larger trees and higher proportional amounts of shade-tolerant species would directly correlate with either gains or losses in the AGB. The functional approach described here offers an alternative hypothesis to explain the rate of accumulation and dynamics of the AGB in interandean secondary forests. 2. Methods 2.1. Study area The study area is located in the northwest region of the Colombian Andes between 5330 13.700 and 6230 31.000 North and 74510 000 and 75020 20.300 West. The study sites were all established around dams employed for power production by the ISAGEN Company. Three different sites, hereinafter referred to as San Rafael (SR, which is dened by EA1, EA2 and EA3), San Carlos (SC, which is dened by EA4 and EA5) and Miel (NC, which is dened by NC1 to NC5) were used. Both SR and SC are part of the eastern region of Antioquia (EA), while Miel is located in the northern part of the Caldas region (NC) (Fig. 1). The total area of secondary forests around the dams currently protected by the Company is approximately 10,000 ha. The study area included an altitudinal gradient of 525 m asl to 1330 m asl and experienced annual precipitation ranging from 3000 mm to 5500 mm (Table 1). The topography and geology of the area are highly variable because of the presence of alluvial and volcanic sediments, which creates a regional mosaic in the soils and in the geomorphological properties of the region (Malagn, 2003). This study was conducted based on information from 10 1-ha permanent plots that were subjected to census in 2007 and 2010. Each plot was divided into 25 subquadrats of 400 m2 (20 m 20 m) each. The ve plots located in NC were square (100 m 100 m), while the ve plots established in EA were rectangular (20 m 500 m) (Table 1). During the rst census (2007), all trees with diameter at breast height P 10 cm (DBH; 1.30 m) were tagged, mapped, tallied, and collected. Trees with buttresses and aerial roots were measured 50 cm above these structures (Condit, 1998). In all cases, the DBH point of measurement was

painted to permit subsequent measurement at the same site on the trunk. Multiple stems were separately recorded but, except for the largest, were dened as sprouts. During the second census (2010), we registered DBH growth, recruitment, and mortality (Condit, 1998). We also measured the total height of 40% of the total number of individuals within each plot. For individuals with heights lower than 15 m, we used a berglass, telescopic pole (Hastings Measuring Sticks 3JF-108823); otherwise, a laser range-nder hypsometer (Nikon 550) was employed. The age of the plots since abandonment (stand-age) in 2010 varied between 15 and 28 years (Table 1). The plots in EA were formerly mostly used as cattle pasture, while the former dominant land use for the plots in NC was crop cultivation. Information regarding stand age and land use was obtained from the former owners of the land containing the plots and from the staff of the ISAGEN Company. Voucher collections were performed for each unique species in each plot. We collected vouchers in all cases where there was any doubt about a plants similarity to another individual already collected within the same plot. Taxonomic identications were made by comparing specimens with herbaria material and, for some groups, with the help of specialists. All of the samples are preserved at the University of Antioquias Herbarium (HUA). Plants that could not be identied as to species were considered to be a separate species, based on the morphology of their vegetative characteristics, and were treated as a species thereafter. 2.2. Environmental variables and species functional traits We measured the geographical position of the starting point of each plot (the 0,0 coordinate) using a GPS and measured the altitude at this point using an altimeter. We used a clinometer to calculate the slope of each quadrant of each plot. The annual mean precipitation was obtained from the local weather stations near each site, and altitude (m asl) was employed as a surrogate of temperature. After removing the organic horizon, we collected soil samples from each 20 m 20 m quadrant from the A horizon (mineral soil) at ve points. At each quadrant, we mixed the ve samples, and a 500 g soil sample was taken. We used the average of the data from the 25 quadrants for all analysis performed at the 1-ha scale. The chemical properties and texture of the collected soil samples were analyzed at the Biogeochemical Analysis Laboratory at the Universidad Nacional de Colombia in Medelln. Exchangeable Ca, Mg and K were extracted with 1 M ammonium acetate and analyzed using atomic absorption. The available P was extracted with L-ascorbic acid and analyzed using a UVVIS spectrophotometer. The soil pH was measured in water with one part soil to two parts water. Exchangeable Al was extracted using 1 M KCl solution and determined by titration with 0.1 M NaOH. The cation-exchange capacity (CEC) was calculated as the sum of the concentrations of bases, and organic matter (OM) content was determined according to Walkley and Blacks volumetric method. Soil texture (sand, silt and clay) was determined using the hydrometer method. Before conducting any analysis of the data, we excluded all remnant individuals and species that came from prior crops, such as Citrus sinensis and Theobroma cacao. Likewise, we excluded all trees that were formerly left for shading cattle; these were represented by a total of 7 individuals in all plots. Based on the height measurements made in all plots, we developed a single generic exponential function (H = boDBHb1) to estimate the total height of all remaining individuals in all plots without eld measurements. We selected wood density (WD) and height (maximum and average) as the key functional traits (Wright et al., 2010) to help explain AGB accumulation rates and species-assemblage performance within stands of different ages in the forest successional path. WD (g cm3) has been positively associated with plant

Author's personal copy

56

M.A. Pea, A. Duque / Forest Ecology and Management 302 (2013) 5461

Fig. 1. Geographical location of the 10 1-ha plots established for the study of forest secondary succession in the Colombian interandean valleys. Names and characteristics of the plots are described in Table 1.

Table 1 Geographical and environmental description of the 10 1-ha plots surveyed in secondary forests in the interandean valleys of Colombia. NC: plots located at the Miel (north of Caldas, land formerly used for crops); EA: plots located in eastern Antioquia (land formerly used for pasture). AGB: aboveground biomass. Site Miel Miel Miel Miel Miel San Rafael San Rafael San Rafael San Carlos San Carlos Plot code (Stand age) NC1 (15) NC2 (15) NC3 (25) NC4 (25) NC5 (25) EA1(18) EA2(18) EA3(18) EA4(28) EA5(28) Coordinates 5330 48.100 N 74540 15.300 W 5330 13.700 N 74550 38.800 W 5330 57.400 N 74550 38.800 W 5330 41.900 N 74540 48.700 W 5330 14.600 N 74530 35.000 W 6220 22.300 N 75010 30.500 W 6230 31.000 N 75020 20.300 W 06210 31.6300 N 75000 27.01400 W 6120 16.200 N 74510 05.400 W 6130 15.300 N 74510 000 W Altitude (m asl) 600 525 625 570 530 1300 1280 1330 955 790 Precipitation (mm) 5500 5500 5500 5500 5500 3540 3540 3540 3000 3000 Individuals 2007 (2010) 688 (743) 707 (723) 671 (673) 713 (709) 702 (681) 542 (554) 686 (723) 658 (677) 676 (695) 845 (928) Species 2007 (2010) 105 (117) 57 (68) 83 (86) 83 (87) 65 (70) 55 (64) 67 (79) 44 (56) 126 (142) 120 (140) AGB (Mg ha1) 2007 (2010) 105.5 (115.1) 74.3 (85.5) 81.5 (87.3) 133.9 (133.4) 112.5 (115.1) 85.2 (94.0) 87.1 (93.9) 86.7 (96.1) 96.1 (105.0) 108.9 (123.3)

Author's personal copy

M.A. Pea, A. Duque / Forest Ecology and Management 302 (2013) 5461

57

defense and longevity versus growth, while height has been associated with competitive ability and light capture (Westoby, 1998). The maximum height (Hmax) of the individuals in each plot was calculated as the average of the 10 tallest trees, and average height (Hmean) was dened as the mean height of the trees found in each plot. We assigned the WD values available in the literature (Chave et al., 2006; Zanne et al., 2009) to each species found in each plot. In cases in which we could not assign a WD value at the species level, we used the average value at the genus or family level. For individuals without a botanical identication, we used the average WD value of all other individuals found in the plot. Average WD (WDmean) was dened as the mean of the trees found in each plot. All of the WD data were pooled and divided into two extreme functional groups, which were dened by the rst (WD0.25) and fourth quartiles (WD0.75), respectively. Then, for each plot, we calculated the total number of individuals with a WD0.25 below 0.45 g cm3 and the number of individuals with a WD0.75 equal or higher than 0.61 g cm3. We performed a one-way analysis of variance (ANOVA) to test for signicant differences in soil properties and functional traits between regions. When signicant differences were found, we carried out a Tukeys honestly signicant difference test. Prior to calculations, values in percentage form were arc-sin transformed. The remaining variables, except for pH, were log-transformed. 2.3. Above-ground biomass stocks We used the following allometric equation developed to estimate the AGB of secondary forests established in the surroundings of a dam in the region of Porce, Antioquia (Sierra et al., 2007): lnAGB 2:232 2:2422 lnDBH. The Porce region is separated from EA and NC, respectively, by approximately 80 and 200 km. This allometric equation was built based on 152 trees that were harvested in a geographic area covering approximately 4400 ha, with an altitudinal range between 900 and 1500 m asl (J.I. del Valle pers. comm. February 15 2013). 2.4. Above-ground biomass dynamics The annual AGB mortality rate (m) between 2007 and 2010 was estimated as m = (log (AGB 2007/(AGB 2007D))/time) 100. The annual AGB recruitment rate (r) was estimated as r = (log ((AGB 2007D + R)/(AGB 2007D))/time) 100. Relative growth rate (g) between 2007 and 2010 was estimated as g = (log ((AGB 2010 R)/(AGB 2007D))/time) 100. The net AGB change (n) in each plot was then estimated as n = (log (AGB 2010/AGB 2007)/ time) 100, where AGB 2007 and AGB 2010 are the AGB in each census, D is the AGB of dead trees, and R is the AGB of new recruits (Valencia et al., 2009). 2.5. Does a prior use of land for raising crops promote an increase in the rate of AGB accumulation? We used ANOVA to test for signicant differences in the AGB stocks, m, r, g, and n between regions. When signicant differences were found, we carried out a Tukeys honestly signicant difference test. 2.6. To what extent do climate, soil characteristics, species functional traits, and/or stand-age determine the stocks and dynamics of the AGB in these secondary forests? Multiple regression analysis based on ordinary least squares (OLSs) was performed to analyze the inuence of climate, physical and chemical properties of soils, species functional traits, and stand age on the AGB stocks in 2010 in the 10 1-ha plots. The climatic

variables were represented by the estimated annual, mean precipitation and the altitudinal position of each plot. The altitudinal position was used as a proxy for temperature variation. The physical properties of soils were assessed by the square root of the arc-sintransformed percentages of the slope of the terrain, and the clay, loam, and sand contents. The chemical properties of soils (soil fertility) were evaluated by means of the soil pH and the log-transformed Ca, Mg, K, Al, CEC, and P content and the square root of the arc-sin-transformed percentages of organic matter (OM). The square root of the arc-sin-transformed slope was included to characterize the form of the terrain. The log-transformed WDmean, Hmax, Hmean, WD0.25, and WD0.75 were employed to evaluate functionally driven processes at the forest stand scale. The log-transformed stand-age was used to evaluate the inuence of forest age after abandonment on the AGB accumulation rate. Each set of climatic, edaphic, functional and stand-age variables were independently subjected to the stepwise procedure, based on the Akaike Information Criterion (AIC), to select the signicant variables that best explained the variation of the AGB stocks (Burnham and Anderson, 1998). Soil chemistry variability was reduced by means of Principal Component Analysis (PCA) and divided into two independent datasets. Each soil fertility data set was composed of the variables with weights P 0.65 in the rst two PCA axes. The quadratic terms of each of the selected edaphic variables were also included in the independent stepwise selection procedure. We then re-applied the stepwise procedure based on AIC to the all-joined, independent set of variables previously selected as signicant from each climatic, edaphic, functional and stand-age set of variables to dene the overall best (and most parsimonious) explanatory model (Burnham and Anderson, 1998). All variables with a signicance level (a) 6 0.05 were retained in the nal models. We used this stepwise procedure because we had a larger number of explanatory variables than plots. This analysis allowed for a further partitioning of the AGB variation explained by the different sets of variables. With all variables remaining in the nal model, we performed a variance partitioning analysis (Legendre and Legendre, 1998) to evaluate the extent to which the selected climatic, edaphic, functional or stand-age factors explained the pattern of the AGB variation observed in the 1-ha plots. To correct for statistical bias, we adjusted the R2 values to account for the number of plots and explanatory variables (Peres-Neto et al., 2006). To quantitate the extent to which climate, soil, species functional traits, and stand-age controlled the pattern of AGB dynamics (m, r, g, and n) in the 1-ha plots, we applied a stepwise procedure based on the OLS and AIC to select the most signicant explanatory variables in the same way as described above, but we used m, r, g, and n instead of the AGB stocks as the dependent variables. As before, a variance-partitioning analysis based on the adjusted R2 value was carried out on the most parsimonious model selected for each dependent variable (Peres-Neto et al., 2006). In this case, the functional factors WDmean, Hmax, Hmean, WD0.25, and WD0.75 were calculated on the dead (for m), recruited (r) or live (g) individuals in each plot, and the functional factors were then log-transformed as well. All statistical analyses were performed using the R package version 2.15 (R Development Core Team, 2012). Variance partitioning was performed with Vegan 17.2 in R (Oksanen, 2010).

3. Results 3.1. Environmental and functional variation between regions Between regions, the only two environmental variables that showed signicant differences were P concentrations and the slope of the terrain (Table 2). The soil fertility variation did not show any

Author's personal copy

58

M.A. Pea, A. Duque / Forest Ecology and Management 302 (2013) 5461

Table 2 Mean and standard deviation of the explanatory variables by region. NC: north of Caldas (Miel); EA: eastern Antioquia (San Carlos and San Rafael). See Table 1 for symbols and details. Letters indicate signicant differences ( P 6 0.05, P 6 0.01, P 6 0.001) between means using the Tukeys honestly signicant difference test. Variable Sand (%) Silt (%) Clay (%) pH OM (%) Al (meq/100 g de suelo) Ca (meq/100 g de suelo) Mg (meq/100 g de suelo) K (meq/100 g de suelo) CEC P (ppm) Slope (%) WD (g cm3) Hmax (m) Hmean (m) WD0.25 WD0.75 AGB (Mg ha1) Mortality rate (%) Recruitment rate (%) Growth rate (%) Net change (%) NC 57.74 3.01 14.26 2.03 28.00 3.22 4.39 0.17 6.94 0.90 2.71 0.61 0.39 0.31 0.21 0.15 0.14 0.05 3.44 0.95 3.40 0.57b 60.29 6.22b 0.51 0.02 24.79 1.61 14.50 0.48 237.20 65.14 199.40 60.17 107.24 20.48 4.32 0.94 1.31 0.65 5.60 1.51 2.40 2.14 OA 55.18 8.35 14.46 3.13 30.37 5.71 4.39 0.19 5.61 1.37 2.80 0.37 0.28 0.10 0.14 0.04 0.11 0.02 3.32 0.40 2.36 0.38a 41.35 13.71a 0.53 0.02 22.69 2.14 14.23 0.28 173.40 29.05 200.80 115.09 102.47 12.50 2.32 0.47 1.32 0.10 4.44 0.47 3.28 0.58 F value 0.42 0.02 0.65 0.00 3.29 0.08 0.69 0.90 0.74 0.08 11.70 7.91 2.79 3.08 1.23 4.00 0.00 0.20 18.10 0.00 2.71 0.78

3.3. Aboveground biomass changes Overall, the studied forests showed an average annual AGB mortality rate of 3.32 1.27% y1 between 2007 and 2010. We found signicant differences in the average mortality rate between regions (Table 2). A high percentage (86.3%) of the AGB mortality rate variation in the entire region was primarily determined by the climatic variability and the functional properties of species (Table 3, Fig. 3c and d). The average annual AGB recruitment rate was 1.31 0.44% y1. We found no signicant differences in the average AGB recruitment between regions (Table 2). Nonetheless, 88.1% of the variation in the AGB recruitment rate for the entire study area was primarily determined by the soil fertility and the functional properties of species (Table 3, Fig. 3e and f). The average AGB growth rate was 5.02 1.22% y1. We found no signicant differences in the average AGB growth between regions (Table 2). None of the selected variables could signicantly explain the relative variation of the AGB growth in the 10 plots. The average annual net AGB change was 2.84 1.55% y1. At the plot level, all plots except the NC4 plot gained AGB (Fig. 4). We found no signicant differences in the average net AGB change between regions (Table 2). Nonetheless, 56.6% of the net AGB change variation in the study region was primarily determined by stand-age and soil fertility (Table 3, Fig. 3g and h). 4. Discussion 4.1. Inuence of prior land use on the aboveground biomass accumulation rate In this study, we did not nd any evidence that the prior use of land for cattle ranching had any negative effect on the tree AGB accumulation rate in comparison with land formerly used for raising crops (Ferguson et al., 2003; Zimmerman et al., 2007). In the wet lowlands of the Colombian Andean mountains, local factors were shown to overcome the inuence of both stand-age and regional features in determining the rate of forest recovery after major disturbances. For example, one of the youngest plots (NC1) had a higher AGB than an older plot in the same region (NC3), and the AGB of NC1 was even higher than that of the oldest plots considered in this study (EA4). The high AGB found in the NC1 plot appeared to be a consequence of the presence of a larger number of remaining resident trees, such as Ficus insipida, rather than of the type of land use before abandonment. Although the remaining trees were excluded from the AGB assessment in the plots, their presence in the forests may have facilitated the establishment of woody and shade-tolerant species (Schlawin and Zahawi, 2008). This phenomenon, which forms the basis of the nucleation model (Peterson and Carson, 2008), promotes a quicker recovery of forest structure (Guariguata and Ostertag, 2001; Kauffman et al., 2009) and helps increase local AGB stocks. However, other non-assessed processes that operates at a regional scale, such as the likely effect of the surrounding landscape on the individual forest stands (Chazdon et al., 2007), could not be discarded as a determinant force that obscured the expected negative inuence of former land uses. 4.2. Determinants of the variation in the aboveground biomass stocks

systematic trend associated with either the region or the stand-age (Fig. 2). The functional variables did not show any signicant differences between regions (Table 2). 3.2. Aboveground biomass stocks The secondary forests in the interandean valleys and mountains of Colombia showed an average AGB of 104.85 16.20 Mg ha1 in 2010. We found no signicant differences in the average AGB stock between regions (Table 2). Overall, 82.1% of the variation in AGB stock over the entire region studied was determined by the soil fertility and the functional properties of the species present (Table 3, Fig. 3a and b).

pH

PCA 2

Mg Ca K

OM

CEC

NC15 NC25 EA18 EA28 3 2 1

Al

PCA 1

Fig. 2. Principal Component Analysis that illustrates the soil fertility variation in 10 1-ha plots located in the Colombian interandean valleys. Names and characteristics of the plots are described in Table 1.

Consistent with our rst expectation of the second research question, the only environmental factor that was shown to be signicant in explaining the differences in the AGB stocks present in the secondary forests evaluated in this study was the content of Ca, which, due to its signicant correlation with Mg and K content, represents soil fertility (Fig. 2). Nonetheless, the

Author's personal copy

M.A. Pea, A. Duque / Forest Ecology and Management 302 (2013) 5461

59

Table 3 Final regression models dening the determinants of aboveground biomass stocks and dynamics in the interandean valleys of Colombia. AGB: aboveground biomass stocks by 2010; m: AGB annual mortality rate; r: AGB annual recruitment rate; n: AGB annual net change. Model ln(AGB) = 4.92 + 0.48 ln(Ca) + 0.27 ln(Ca ) + 3.63 ln(Hmean) m = 27.78 + 2.66 ln(Precipitation) + 2.60 ln(WD0.25) 20.12 ln(Hmean) p r = 32.75 + 0.58 ln(Ca) + 2.26 sin1 ( (OM)) 7.28 ln(WDmean) 14.85 ln(Hmean) p n = 13.58 6.50 ln(stand-age) + 12.32 sin1 ( (OM))
2

Adjusted R2 (%) 82.05 86.32 88.12 56.56

P-value 0.004 0.002 0.004 0.022

AIC 21.61 18.09 4.30 33.27

(a)
Predicted 4.7 4.9
NC15 NC25 EA18 EA28

(b)
Functional traits 0.522 0.164 Soils 0.463

4.5

Residuals = 0.179

(c)
5 Predicted 3 4

4.2

4.4

4.6

4.8

5.0

5.2

(d)
Functional 0.306 traits 0.281 Climate 0.276

Residuals = 0.137

(e)
Predicted 1.0 1.5 2.0

(f)
Functional traits 0.852 0.356 Soils 0.385

Residuals = 0.119

(g)
Predicted 3 4 5

(h)
Stand age 0.671 0.471 Soils 0.365

Residuals = 0.434

2 3 Observed

Fig. 3. Observed values versus values predicted by the regression models and their respective variance partitioning (Venn diagram) for the AGB stocks in 2010 (a and b); AGB annual mortality rate (c and d); AGB annual recruitment rate (e and f); and AGB annual net change (g and h). The observed negative values in the shared component of the variance partitioning means that the set of variables together explain the variation better than the sum of the individual effects of each variable (Legendre and Legendre, 1998).

relationship between the AGB and soil fertility was quadratic rather than monotonic, implying the existence of an optimum point at an intermediate fertility that maximizes the AGB throughout the successional path. In addition to soil fertility, and as was expected, tree height was positively correlated with increased AGB stock. Therefore, greater AGB stocks at an intermediate fertility point may be associated with a higher abundance of tall, short-lived, fast-growth pioneer species. Because short- and medium-lived pioneer species seem to be more efcient in allocating resources after the abandonment of pastures or elds than long-lived, slow-growth, shade-tolerant species (Chapin et al., 1986; Rees et al., 2001), a differential species response to the spatial variation in the availability of nutrients in

soils may be a key determinant of forest recovery and the associated increase in the AGB accumulation rate in tropical, secondary forests (Chazdon et al., 2010; Rees et al., 2001; Saldarriaga et al., 1988). The lack of relevance of the altitudinal range covered by the forest plots included in this study might be due to the lack of changes in temperature substantial enough to affect the AGB accumulation rate. Likewise, the plot stand-age range considered may have been too small to differentiate AGB stocks. Therefore, the absence of differences in AGB stocks between plots of different ages illustrates the inappropriateness of using chronosequences for assessing and comparing AGB accumulation rates in environmentally heterogeneous ecosystems (Walker et al., 2010).

Author's personal copy

60

M.A. Pea, A. Duque / Forest Ecology and Management 302 (2013) 5461

Mortality Recruitment Growth Net change

NC1

NC2

NC3

NC4

NC5

EA1

EA2

EA3

EA4

EA5

In fact, in this study, tree mortality rates in secondary forests s composition and the were primarily determined by the forest rainfall variation between sites. With respect to compositional traits, our regression model showed a positive relationship between mortality rates and the number of species with a low WD but a negative relationship between canopy height and mortality rates. These two functional factors perfectly matched the aforementioned process of canopy closure, which is related to the increase in mean tree height independent of stand-age, and results in high mortality of seedlings, shrubs, and small trees of shadeintolerant species with a low WD (Capers et al., 2005). With respect to rainfall variation, a likely higher rate of gap formation caused by the combination of high annual rainfall and steeper slopes in NC than in EA, offers a logical explanation for the higher mortality rates found in the NC region. Therefore, in forests 15 30 years after abandonment, the continuous shift in the relative abundance of species toward a proportional increase in shade-tolerant species (Fig. S2) appears to be the most important process determining the AGB mortality rates of pioneer species at this early stage of the natural succession in the interandean valleys of Colombia.

Annual AGB dynamic (%)

Plot

Fig. 4. Dynamic and net changes in the AGB (% y1) in 10 permanent plots located in secondary forests in the interandean valleys of Colombia.

Because the sample size of this study was relatively small (10 plots), we performed the same analysis at the quadrant scale (20 m 20 m) to determine the extent to which our analysis depended on sample and plot size. Overall, the total amount of AGB variation explained by the quadrant-based model decreased to 55.0% in comparison to the 82.1% that was explained by using the 1-ha plots. The nal regression model based on the quadrants indicated that edaphic, climatic, and functional variables were signicant (Fig. S1a). Nonetheless, when we conducted variation partitioning, the species functional traits accounted for 89% of the total explained variation of the AGB stocks, while none of the other individual factors accounted for more than 6% of the total variation (Fig. S1b). The main difference in the results obtained with the 1ha regression model and the quadrant-based regression model was a decrease in the relative importance of soil fertility and an increase in the relative importance of climatic variables as factors that control the AGB variation at ner spatial scales. Because species functional traits remained the most important factors that determined AGB stocks at both scales, our results emphasize the dominant role of individuals and taller species with high WD in determining the increase in AGB during the secondary succession process (Chazdon, 2008; Finegan, 1996; Rees et al., 2001). 4.3. Patterns and determinants of forest aboveground biomass changes 4.3.1. Aboveground biomass mortality rates The second expectation of the second research question, which posits an increase in the mortality rate with increased stand-age and soil fertility, was fully rejected by our data. The expected increase in mortality with age had been associated with canopy closure at older ages and the beginning of the self-thinning phase, which promotes enhanced stem exclusion and a consequent increase in tree mortality (Chazdon, 2008; Oliver and Larson, 1990). However, in this study, differences in the speed of canopy closure were determined by the local conditions rather than by the time elapsed since abandonment. Likewise, the lack of the importance of soil-fertility for mortality rates was mainly because before canopy closure occurs during the early stages of natural succession, exclusive competition may have not yet occurred and the inuence of soil fertility on tree mortality was thus irrelevant.

4.3.2. Aboveground biomass recruitment rates Our third expectation of the second research question, that of a negative correlation between AGB recruitment rate and stand-age and a positive correlation between AGB recruitment and soil fertility, was partially borne out. Because the AGB recruitment rate did not differ signicantly with stand-age, the rst part of our expectation was not supported. Because AGB recruitment is promoted by the availability of space and light at the forest oor, the longer canopy closure takes, the higher the AGB recruitment rate. In contrast, soil fertility, which is known to accelerate the growth rate of plant species, was signicantly and positively associated with the rate of AGB recruitment (Rees et al., 2001). Nonetheless, variance partitioning indicated that the functional composition of the forests was the major determinant of the rate of AGB increase (Fig. 3f). As shown by the negative relationship between AGB recruitment and average WD and Hmean found in the regression model, AGB recruitment was higher in forests dominated by pioneer species of smaller sizes than in forests already subjected to stem exclusion by self-thinning and a shift in functional diversity through the permanent inclusion of new shade-tolerant species (Fig. S2) (Chazdon, 2008; Finegan, 1996).

4.3.3. Aboveground biomass growth rates The fourth expectation of the second research question, that the AGB rate of growth of the trees may be higher in richer soils, was not supported by our data. The even AGB rate of growth recorded across plots emphasizes the inability of local edaphic factors to control the relative increase in living mass of secondary forests that have not passed the stem exclusion phase (sensu Chazdon, 2008). Moreover, the lack of signicant differences between plots in the mean AGB rate of growth recorded for WD0.25 individuals (5.7%) or for WD0.75 individuals (4.7%) (t test = 1.39, d.f. = 12.12, P > 0.05) did not provide any evidence for the existence of a negative coordination between wood density and growth, as has been reported for primary tropical forests (Wright et al., 2010). The very similar AGB growth rates observed in most plots for tree species with different traits may have been caused by a consistent level of competition for light among individuals racing to reach the forest canopy. Therefore, at this stage of development of the secondary succession, an even availability of light should neutralize any advantage of fast-growth pioneer species relative to slow-growth shade-tolerant ones.

Author's personal copy

M.A. Pea, A. Duque / Forest Ecology and Management 302 (2013) 5461

61

4.3.4. Aboveground biomass net change Our fth prediction of the second research question, that net AGB change (%) would correlate positively with soil fertility and negatively with stand-age, was partially conrmed by the regression model. The OM contents were positively correlated with sand (R = 0.6), but negative and signicantly correlated with soil fertility. Therefore, the net AGB change was primarily associated with the low soil-fertility associated with sandy well-drained soils that, in presence of steep slopes, could increase forest dynamics. The decrease in the rate of net AGB associated with stand-age could be explained by a lower proportional increase in the AGB (%) due to a higher a AGB mortality rate in the older plots of NC. In EA, in contrast, the net AGB change was not related to stand-age and m, r, and g were quite uniform among plots of different age (Fig. 4). Our study focused on the accumulation rate of AGB in secondary forests after large-scale disturbances within a relatively short period of the natural succession (1530 years) in an environmentally complex geographic region. Overall, local edaphic conditions and functional diversity have been shown to play key roles in inuencing the rate and dynamics of tree AGB accumulation. This result emphasizes the fact that land restoration that restores the natural cycle of secondary forests could assist in the mitigation of global warming by promoting key environmental functions, such as carbon sequestration. Acknowledgments The authors of this study extend our special thanks to all the staff members of the Equipo Ambiental of ISAGEN, especially to Julian Carmona, Huber Vanegas, Claudia lvarez, and Ana Gmez. We are very much indebted to the students, forest engineers and eld assistants who participated in the eldwork and the processing of information. The authors acknowledge the valuable collaboration and comments of Juan Saldarriaga, Natalia Norden and three anonymous reviewers. English editing of the rst versin submitted was covered by the rea curricular en bosques y conservacin ambiental de la Universidad Nacional de Colombia. This work was made possible by funding received from ISAGEN (Interinstitutional Cooperation Agreement # 46-3127). Alvaro Duque was partially supported by the Convocatoria Nacional de Investigacin y de creacin artstica de la Universidad Nacional de Colombia 2010 2012. Alvaro Duque (through the Contract No. IDEAM II-CPS-001 de 2013) and Miguel A. Pea received funding from the project Consolidacin de un Sistema de Monitoreo de Bosques y Carbono, como soporte a la Poltica Ambiental y de Manejo en Colombia, which is supported by the Gordon and Betty Moore Foundation. Appendix A. Supplementary material Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.foreco.2013.03. 025. References
Brown, S., Lugo, A.E., 1990. Tropical secondary forest. J. Trop. Ecol. 6, 132. Burnham, K.P., Anderson, D.R., 1998. Model selection and inference: a practical information-theoretic approach. Springer-Verlag, New York. Capers, R., Chazdon, R.L., Brenes, A.R., Vilchez, B., 2005. Successional dynamics of woody seedling communities in wet tropical secondary forest. J. Ecol. 93, 1071 1084. Chapin, E.S., Vitousek, P.M., Van Cleeve, K., 1986. The nature of nutrient limitation in plant communities. Am. Nat. 127, 4858. Chave, J., Muller-Landau, H.C., Baker, T.R., Easdale, T.A., ter Steege, H., Webb, C.O., 2006. Regional and phylogenetic variation of wood density across 2456 Neotropical tree species. Ecol. Appl. 16, 23562367.

Chazdon, R.L., 2008. Chance and determinism in tropical forest succession. In: Carson, W.P., Schnitzer, S.A. (Eds.), Tropical Forest Community Ecology. WileyBlackwell, Oxford, pp. 384408. Chazdon, R.L., Coe, F.G., 1999. Ethnobotany of woody species in second-growth, oldgrowth, and selectively logged forests of northeastern Costa Rica. Conserv. Biol. 13, 13121322. Chazdon, R.L., Letcher, S.G., Breugel, M., Martinez, M., Bongers, F., Finegan, B., 2007. Rates of change in tree communities of secondary tropical forests following major disturbances. Philos. Trans. R Soc. Lond. B 362, 273289. Chazdon, R.L., Finegan, B., Capers, R., Salgado-Negret, B., Casanoves, F., Boukili, V., Norden, N., 2010. Composition and dynamics of functional groups of trees during tropical forest succession in Northeastern Costa Rica. Biotropica 42, 31 40. Condit, R., 1998. Tropical Forest Census Plot. Springer-Verlag, R.G. Landes Company, Berlin, Georgetown, Texas. Ferguson, B.G., Vandermeer, J., Morales, H., Grith, D.M., 2003. Post-agricultural succession in El Peten, Guatemala. Conserv. Biol. 17, 818828. Finegan, B., 1996. Pattern and process in neotropical secondary forests: the rst 100 years of succession. Trends Ecol. Evol. 11, 119124. Grace, J., 2004. Understanding and managing the global carbon cycle. J. Ecol. 92, 189202. Guariguata, M.R., Ostertag, R., 2001. Neotropical secondary forest succession: changes in structural and functional characteristics. Forest Ecol. Manage. 148, 185206. Kauffman, J.B., Hughes, R.F., Heider, C., 2009. Carbon pool and biomass dynamics associated with deforestation, land use, and agricultural abandonment in the neotropics. Ecol. Appl. 19, 12111222. Legendre, P., Legendre, L., 1998. Numerical Ecology. Elsevier Science, Amsterdam. Letcher, S.G., Chazdon, R.L., 2009. Rapid recovery of biomass, species richness, and species composition in a forest chronosequence in the northeastern Costa Rica. Biotropica 41, 608617. Lu, D., Moran, E., Mausel, P., 2002. Linking Amazonian secondary succession forest growth to soil properties. Land Degrad. Develop. 13, 331343. Malagn, D., 2003. Ensayo sobre tipologa de suelos colombianos, nfasis en gnesis y aspectos ambientales. Rev. Acad. Colomb. Cienc. 27, 319341. Oksanen, J., 2010. Vegan: community ecology package project. Oliver, C.D., Larson, B.C., 1990. Forest Stand Dynamics. McGraw-Hill, New York. Peres-Neto, P.R., Legendre, P., Dray, S., Borcard, D., 2006. Variation partitioning of species data matrices: estimation and comparison of fractions. Ecology 87, 26142625. Peterson, C.J., Carson, W.P., 2008. Processes constraining woody species succession on abandoned pastures in the tropics: on the relevance of temperate models of succession. In: Carson, W.P., Schnitzer, S.A. (Eds.), Tropical Forest Community Ecology. Wiley-Blackwell, Oxford, pp. 367383. R Development Core Team, 2012. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. ISBN 3900051-07-0, URL <http://www.R-project.org/>. Rees, M., Condit, R., Crawley, M., Pacala, S., Tilman, D., 2001. Long term studies of vegetation dynamics. Science 293, 650655. Saldarriaga, J.G., West, D.C., Tharp, M.L., Uhl, C., 1988. Long-term chronosequence of forest succession in the upper ro Negro of Colombia and Venezuela. J. Ecol. 76, 938958. Schlawin, J., Zahawi, R., 2008. Nucleating succession in recovering neotropical wet forests: the legacy of remnant trees. J. Veg. Sci. 19, 485492. Sierra, C.A., del Valle, J.I., Orrego, S.A., Moreno, F.H., Harmon, M.E., Zapata, M., Colorado, G.J., Herrera, M.A., Lara, W., Restrepo, D.E., Berrouet, L.M., Loaiza, L.M., Benjumea, J.F., 2007. Total carbon stocks in a tropical forest landscape of the Porce region, Colombia. Forest Ecol. Manage. 243, 299309. Silver, W.L., Ostertag, R., Lugo, A.E., 2000. The potential for carbon sequestration through reforestation of abandoned tropical agricultural and pasture lands. Restor. Ecol. 8, 394407. Valencia, R., Condit, R., Muller-Landau, H.C., Hernandez, C., Navarrete, H., 2009. Dissecting biomass dynamics in a large Amazonian forest plot. J. Trop. Ecol. 25, 473482. Walker, L.R., Wardle, D.A., Bardgett, R.D., Clarkson, B.D., 2010. The use of chronosequences in the studies of ecological succession and soil development. J. Ecol. 98, 725736. Westoby, M., 1998. A leaf-height-seed (LHS) plant ecology strategy scheme. Plant Soil. 199, 213227. Wright, S.J., Kitajima, K., Kraft, N.J.B., Reich, P.B., Wright, I.J., Bunker, D.E., Condit, R., Dalling, J.W., Davies, S.J., Daz, S., Engelbrecht, B.M.J., Harms, K.E., Hubbell, S.P., Marks, C.O., Ruiz-Jaen, M.C., Salvador, C.M., Zanne, A.E., 2010. Functional traits and the growthmortality trade-off in tropical trees. Ecology 91, 36643674. Zanne, A.E., Lpez-Gonzlez, G., Coomes, D.A., Ilic, J., Jansen, S., Lewis, S.L., Miller, R.B., Swenson, N.G., Wiemann, M.C., Chave, J., 2009. Data from: Towards a worldwide wood economics spectrum. Dryad. Digital Repos. http://dx.doi.org/ 10.5061/dryad.234. Zimmerman, J.K., Aide, T.M., Lugo, A.E., 2007. Implications of land use history for natural forest regeneration and restoration strategies in Puerto Rico. In: Cramer, V., Hobbs, R. (Eds.), Old Fields: Dynamics and Restoration of Abandoned Farmland. Island Press, Washington, pp. 5174.