<p dir="ltr">Understanding forest carbon sequestration is crucial for predicting and managing the carbon cycle, yet we lack evidence for whether, when and how the carbon sink in tropical forests recovering from land use change is nutrient limited. Here we show how the tropical forest recovery rate responds to experimental nutrient manipulation over a secondary succession gradient in a naturally recovering Central American landscape. Nutrient limitation of aboveground biomass accumulation shifts from strong nitrogen limitation in young forests to no evidence of nitrogen or phosphorus limitation in older secondary or mature forests. Nitrogen addition increases aboveground biomass accumulation by 95% in recently abandoned pasture and 48% in 10-year-old forests. Conversely, we observe no influence of nitrogen on older forests and no evidence of phosphorus limitation at any stage. If our findings of nitrogen limitation extend to young tropical forests globally, nitrogen could prevent the sequestration of 0.69 (0.47-0.84) Gt CO2 each year.<br>The data and code included in this deposit contains the data files and code necessary to create figures and statistical analyses for this study.</p><p dir="ltr"><b>File list:</b></p><p dir="ltr">Tang_etal_readme.docx</p><p dir="ltr">Tang_etal_metadata_public.pdf - contains metadata for all data files</p><p dir="ltr">Tang_etal_AguaSalud_Soil_classification.pdf</p><p dir="ltr">Tang_etal_Gigante_Soil_classification.pdf</p><p dir="ltr">Tang_etal_biomass_dynamics.csv</p><p dir="ltr">Tang_etal_biomass_recovery.csv</p><p dir="ltr">Tang_etal_P_values_adjustment.csv</p><p dir="ltr">Tang_etal_soil_data.csv</p><p dir="ltr">Tang_etal_total_P_0-30_calculations.xlsx</p><p dir="ltr">Tang_etal_biomass_code.Rmd</p><p dir="ltr">Tang_etal_soil_data_analysis.Rmd</p>
Funding
Facilitating the tropical forest carbon sink: The evolution and function of symbiotic N2 fixation
Natural Environment Research Council
Find out more...Carbon Mitigation Initiative, bp
How did the evolution of plants, microbial symbionts and terrestrial nutrient cycles change Earth's long-term climate?
Natural Environment Research Council
Find out more...BIOmes of Brasil - Resilience, rEcovery, and Diversity: BIO-RED
Natural Environment Research Council
Find out more...British Council #275556724
Leverhulme Trust
Lang Assael Family Innovation Fund, Cary Institute of Ecosystem Studies
Millbrook Garden Club
Chinese Scholarship Council-University of Leeds joint scholarship
Stanley Motta
Frank and Kirstin Levinson
Hoch Family
Andrew W. Mellon Foundation
Scholarly Studies Program of the Smithsonian Institution
U Trust
Heising-Simons Foundation
Geographic description
Agua Salud (9o13’N, 79o47’W, 330 meters above sea level) and Gigante (9o06’31’’N, 79o50’37’’W, 60 meters above sea level) are located in the Republic of Panama.Time period
1997-2019Methodology
Experimental design
The Gigante fertilization experiment (where the ‘600-year forest’ is located) was established in mature forest of at least 600 years post human disturbance in 1997. It consists of four nutrient addition treatments (control, nitrogen, phosphorus, and nitrogen plus phosphorus) with each replicated four times (1 forest age × 4 treatments × 4 replicates) (Wright et al., 2018). The area of each plot is 0.16 ha (40×40 m). The Agua Salud Project fertilization experiment consists of experimental plots at three different successional stages, very young secondary forests established immediately after the abandonment of cattle pastures (named “0-year forests”), and two middle-age secondary forests (named “10-year forests” and “30-year forests”). The experiment began in 2015 with plot establishment of four replicates per treatment for the 10- and 30-year forests and the establishment of four replicate former pasture plots that were clear of trees as 0-year forests. In 2016, we established a fifth replicate set of plots for all forest ages and began fertilization treatments with the same nutrient addition treatments as the mature forests. In both sites, within each replicate, we blocked the control, nitrogen, phosphorus, and nitrogen plus phosphorus plots within sites on the landscape to minimize the effects of small-scale variations in climate, soils, and surrounding forest tree diversity and seed source. Plots are located on the landscape to eliminate the possibility of flow of nutrients from one plot to another based on slope, and the minimum distance between plots is 40 m. The area of fertilized plots is 0.16 ha (40×40 m) and the control plots are 0.1 ha and 20x50 m since they do not need a fertilization buffer. In every Agua Salud fertilization plot, trees are monitored only within the inner 0.1 ha leaving a buffer zone on four sides that was fertilized.
Fertilizer was added as coated urea ((NH2)2CO) and triple superphosphate (Ca(H2PO4)2·H2O) in nitrogen and phosphorus-treated plots, respectively. Annual doses were 125 kg N·ha-1·yr-1 and 50 kg P·ha-1·yr-1, and fertilizers were added by hand in four equal doses at regular intervals beginning approximately two weeks after the beginning of regular, heavy rains (approximately 15-30 May, 1-15 July, 1-15 September, and 15-30 October) (Wright et al., 2011).
We also measured the annual rainfall when the Agua Salud fertilization experiment was established. We found changes in rainfall does not affect biomass accumulation and its dynamics.
Forest inventory
We monitored all 76 plots since the start of the nutrient fertilization (i.e. 2015 in Agua Salud forests and 1997 in Gigante forest). All free-standing woody plants (trees, palms, and lianas – hereon referred to as trees since 75% of plants were trees) within the plots were identified, but the monitoring protocols differed slightly between the two sites. In Agua Salud, in the center 0.1 ha of plots all stems of trees and palms with diameters ≥ 5 cm and all lianas with diameter ≥ 1 cm were measured, as well as 50% of all tree and palm stems with diameter between 1 and 5 cm. All trees of the same size cutoffs were measured in the control plots. In Gigante, trees with diameter ≥ 10 cm were measured in the whole plot, and trees with diameter between 1 and 10 cm were measured in the central 20×30m subplot. We use the nested design to catch the dynamics of small trees. For the large trees, diameters were measured above any buttresses or other deformities of the lower trunk (Wright et al., 2018, van Breugel et al., 2013). All diameters were measured at 1.3 meters height.
In Agua Salud, plots were censused prior to fertilization in 2015, with the exception of the plots established in 2016. Plots were then censused yearly after fertilization from 2016 to 2019. In Gigante, plots were censused every five years from 1997 to 2018; however, we focused on the census prior to fertilization in 1997 and the censuses after fertilization from 2003 to 2018. We divided our analysis into two census intervals to capture (1) the most recent forest dynamics that would be comparable to the climatic conditions as Agua Salud experiment (2013-2018) and (2) the forest characteristics prior to fertilization and the subsequent post-fertilization dynamics (the 1997 to 2013 censuses).
Biomass at the plot scale
We first estimated the biomass (kg stem-1; referred to as biomass) of all recorded stems in Agua Salud and Gigante plots. Biomass was calculated by applying different allometric functions to estimate the aboveground biomass of each stem of each tree, liana and palm. For trees, we estimated the biomass of each stem using the following allometric function (Chave et al., 2014):
AGB = exp [-1.803 – 0.976E + 0.976 ln (WD) + 2.673 ln (DBH) – 0.0299 [ln(DBH)2].
where AGB represents aboveground biomass (kg stem-1), E is the local climatic index (Rutishauser et al., 2020), and WD is wood density (g cm-3). DBH represents the diameter at 1.3 meter (cm). The climate index, E=0.05645985 near our study site, represents the effect of environment on tree height allometry (Rutishauser et al., 2020). Species-specific wood density (in g cm-3) was estimated from the most common species in Agua Salud and Gigante (S. Joseph Wright unpublished data and ref. (Rutishauser et al., 2020)).
For lianas, the aboveground biomass of each stem was calculated using a liana-specific allometric equation (Schnitzer et al., 2006, Lai et al., 2017):
AGB = exp [-0.999 + 2.682 * ln (DBH)].
For palms, we calculated the above-ground biomass using a palm-specific allometric equation (Rutishauser et al., 2020, Goodman et al., 2013):
AGB = 0.0417565 * (DBH) 2.7483.
We summed the biomass of all stems one centimeter and above in each plot and scaled to on hectare to get the plot scale per hectare biomass (Mg ha-1). To account for differences in the methods of inventorying all stems five centimeters and above and stems one to five centimeters in half of the plot at Agua Salud, we cloned the biomass of stems one to five centimeters to double the biomass of stems in that size class. At Gigante where we inventoried all stems greater than ten centimeters and stems one to ten centimeters in 37.5% of the plot area, we summed the biomass of stems one to ten centimeters and multiplied by 2.667.
Forest dynamics at the plot scale
We calculated the mean annual net change of biomass for each plot and census interval (between 2015 and 2016, 2016 and 2017, 2017 and 2018, and 2018 and 2019 for Agua Salud plots, and between 1997 and 2013, and 2013 and 2018 for Gigante plots; Mg ha-1 year-1).
We then calculated plot scale biomass gains and losses for each census interval in Agua Salud plots (the same intervals as described above), and for longer intervals in the Gigante plots (between 1997 and 2003, between 2003 and 2008, between 2008 and 2013, between 2013 and 2018; Mg ha-1 year-1). For Gigante, we then took the mean of dynamics between 1997 and 2003, between 2003 and 2008, and between 2008 and 2013 to get the dynamics for the interval 1997 to 2013. We then took the mean dynamics of these census intervals to represent the full fertilization period. Growth was calculated as the gains of the trees recorded in the first census year that survived until the next census, divided by the time between the two censuses. Recruitment was calculated as the total aboveground biomass gain of trees which were recorded in the later census but not in the previous, divided by the time between the two censuses. Mortality was calculated as the aboveground biomass loss because of the loss of trees recorded in the initial census by the second census, divided by the period between the two censuses. Due to potential systematic error of overestimating growth and underestimating recruitment because of the cutoff of minimum tree size, we combined growth and recruitment to one measure of biomass gain and considered mortality as biomass loss.
Finally, for forest ages where we found significant effects of nutrient, we converted the net biomass change due to nutrients into carbon dioxide equivalent. We took the difference in biomass sequestered between the forests with versus without the added nutrient into CO2 by multiplying by 0.47 (the carbon:biomass ratio (Martin et al., 2011)) and 3.66 (the number of CO2 molecules per carbon). We then took the mean of the values at 0 and 10 years to calculate the mean carbon dioxide accumulation due to nutrients over the first ten years of forest recovery (tons CO2 ha-1 year-1).Location of field samples or specimens
Soil samples are archived at the Smithsonian Tropical Research Institute.Taxonomic species or groups
Tropical wet forests.Additional bibliographic references
see Tang_etal_metadata_public.docxData Sharing Statement
The Cary Institute of Ecosystem Studies furnishes code and data under the following conditions: The code or data have received quality assurance scrutiny, and, although we are confident of their accuracy, Cary Institute will not be held liable for errors in the code or data. Code and data are subject to change resulting from updates in data screening or models used. To cite code or data, click on the Cite button on this page.
