bs_bs_banner

Austral Ecology (2013) ••, ••–••

Grass allometry and estimation of above-ground biomass

in tropical alpine tussock grasslands
IMMACULADA OLIVERAS,1* MAARTEN VAN DER EYNDEN,2,3 YADVINDER MALHI,1
NELSON CAHUANA,4 CARLOS MENOR,5 FLOR ZAMORA4† AND
TORBJØRN HAUGAASEN2
1
Environmental Change Institute, School of Geography and the Environment, University of Oxford,
South Parks Road, Oxford OX13QY, UK (Email: [email protected]); 2Department of
Ecology and Natural Resource Management, Norwegian University of Life Sciences, Ås, and 3The
Government of Norway’s International Climate and Forest Initiative, Norwegian Ministry of the
Environment, Oslo, Norway; 4Universidad Santo Antonio Abad del Cusco, Cusco, Peru; and
5
Departamento de Geología y Geoquímica, Facultad de Ciencias, Universidad Autónoma de Madrid,
Madrid, Spain

Abstract The puna/páramo grasslands span across the highest altitudes of the tropical Andes, and their ecosys-
tem dynamics are still poorly understood. In this study we examined the above-ground biomass and developed
species specific and multispecies power-law allometric equations for four tussock grass species in Peruvian high
altitude grasslands, considering maximum height (hmax), elliptical crown area and elliptical basal area. Although
these predictors are commonly used among allometric literature, they have not previously been used for estimating
puna grassland biomass. Total above-ground biomass was estimated to be of 6.7 ± 0.2 Mg ha−1 (3.35 ± 0.1 Mg C
ha−1). All allometric relationships fitted to similar power-law models, with basal area and crown area as the most
influential predictors, although the fit improved when tussock maximum height was included in the model.
Multispecies allometries gave better fits than the other species-specific equations, but the best equation should be
used depending on the species composition of the target grassland. These allometric equations provide an useful
approach for measuring above-ground biomass and productivity in high-altitude Andean grasslands, where destruc-
tive sampling can be challenging and difficult because of the remoteness of the area. These equations can be also
applicable for establishing above-ground reference levels before the adoption of carbon compensation mechanisms
or grassland management policies, as well as for measuring the impact of land use changes in Andean ecosystems.

Key words: above-ground productivity, Andes, Calamagrostis spp., carbon stock, puna.

INTRODUCTION Asner et al. 2012), but other tropical non-forest eco-

systems can also contain substantial amounts of
Estimation of carbon stocks in tropical ecosystems and carbon (e.g. Scurlock & Hall 1998; Scurlock et al.
emissions of greenhouse gases from tropical deforesta- 2002; Gibbon et al. 2010).
tion and forest degradation have received increasing The Andean Neo-tropical tussock grasslands
attention in recent years (e.g. Maniatis & Mollicone (termed as puna or páramo depending on their species
2010; Asner 2011). Progress in developing a forest composition) span the entire tropical Andes, and rep-
carbon compensation mechanism for tropical coun- resent the main ecosystem at altitudes above approxi-
tries under the United Nations Framework Conven- mately 3500 m a.s.l. However, despite their extensive
tion on Climate Change (UNFCCC), commonly distribution, we lack integrative descriptions of their
referred to as Reduced Emissions from Deforestation physiognomy and functioning (Baldassini et al. 2012).
and Forest Degradation – REDD+ (e.g. UNFCCC The Andean high-altitude grasslands have been
2010), has sparked further research on establishing exposed to human activity for millennia (White &
reliable methods for carbon estimation in tropical eco- Maldonado 1991), but anthropogenic pressures have
systems (Phillips et al. 1998; Malhi & Phillips 2004; significantly increased over the last decades (Luteyn
Saatchi et al. 2007; Asner 2009). Tropical forests have 1992, in Ramsay & Oxley 1996), mainly from grazing
received the most attention (e.g. Saatchi et al. 2011; and burning (Bustamante & Bittencourt 2007). Apart
from the direct impact on the ecosystem diversity and
*Corresponding author. degradation, grazing and burning have also direct
†
Deceased. implications for carbon stocks. In an overview of
Accepted for publication September 2013. twelve different Andean grassland biomass studies
© 2013 The Authors doi:10.1111/aec.12098
Austral Ecology © 2013 Ecological Society of Australia
2 I . O L I V E R A S ET AL.

(including puna/páramo) provided by Hofstede et al. study focuses on the puna grasslands of the south-
(1995), above-ground biomass estimations ranged eastern Andes of Peru, which are under increasing
from 205 g m−2 to 2829–3553 g m−2, depending on the pressure from fire (Oliveras et al. 2013) and grazing.
degree of grazing or burning and pools sampled, which The study was conducted on relatively undisturbed
translates into 1.0 to 14.1–17.8 Mg C ha−1. Ramsay & puna grassland in the buffer area of Manu National
Oxley (2001) estimated standing biomass of 794 g m−2 Park: the area had not been burned in 6 years and
and 837 g m−2 at two Ecuadorian páramo grassland there was no grazing activity. The aims of this study
sites, which translates to 4.0–4.2 Mg C ha−1, and were to (1) estimate above-ground biomass and
Gibbon et al. (2010) estimated 6.5 Mg C ha−1 in their biomass carbon content in a puna ecosystem, (2)
grazed Peruvian puna sites. Although the different develop allometric equations for dominant grasses and
methodologies used in the various studies make com- (3) explore the suitability of allometric equations as a
parisons difficult, it is recognized that high-altitude reliable method for estimating above-ground produc-
grasslands can play an important role in climate tivity in tropical montane tussock grasslands.
change mitigation and may be a suitable area for
carbon compensation mechanisms because of their
extensive distribution and ability to sequester carbon
METHODS
(Conant 2010).
Quantification of above-ground biomass and above-
ground productivity can be achieved by three broad Study site
methods: (i) through intensive biomass destructive
sampling; (ii) through allometric equations (Brown The study was carried out in the south-eastern Peruvian
1997; Navar et al. 2004; Chave et al. 2005; Feldspauch Andes, at approximately 3300 m a.s.l. in the south-western
et al. 2011), and (iii) through remote sensing tech- buffer area of Manu National Park (13°18′S, 71°58′W)
(Fig. 1). At this altitude, a typical puna vegetation type is
niques (e.g. Saatchi et al. 2007; Asner 2009). Destruc-
observed, dominated mainly by tussock-forming grasses.
tive methods may be a feasible method in open Dominant species include Calamagrostis spp., and Festuca
grasslands, but the remoteness of some areas and dolichophylla J. Presl. (Gibbon et al. 2010). Average annual
limited resource availability for transportation and rainfall ranges from 1900 to 2500 mm, with a wet season
processing the samples often make them very costly spanning from October to April (Gibbon et al. 2010). Mean
and logistically challenging. Remote sensing methods annual temperature is approximately 11°C at 3600 m a.s.l.
require targeted field-based measurements for calibra- (Gibbon et al. 2010). Puna soils are largely composed of an
tion and validation. Allometric equations require an organic-rich A-layer, stony B/C-layers, and no Oh-layer
initial extensive destructive biomass sampling, but (Gibbon et al. 2010; Zimmermann et al. 2010).
they can later be used as a consistent and non-
destructive method for estimating above-ground
biomass and productivity. Currently there are reliable Data collection
allometric equations for different types of tropical
forests (e.g. Chave et al. 2005; Feldspauch et al. 2011). We established eight transects of 30 m length and 2 m width
However, few equations have been developed for non- in a puna area of the Wayqecha Biological Reserve in July
forest tropical and subtropical ecosystems (Johnson 2010. Each transect consisted of eight 4 m2 plots evenly
et al. 1998; Navar et al. 2004). One of the first steps distributed along the transect. At each transect, eight plots of
2 × 2 m were delineated, each separated by two metres. At
towards an integrative assessment of above-ground
each plot, we identified all plants to the genus or species level.
productivity in non-forest tropical and subtropical For each tussock we measured the longest basal diameter
ecosystems would be the development of reliable (d1), the basal diameter perpendicular to d1 (d2), the height as
allometric equations for major ecosystems, which encountered in the field (h), the maximum height (stretched
could be adopted as a systematic method for measur- by hand; hmax), the longest tussock crown diameter (dc1) and
ing, reporting and verifying carbon compensation the diameter perpendicular to dc1 (dc2).The crown- and basal
schemes. With this regard, multispecies models allow diameters were used to estimate elliptical crown areas (CA)
for wider applicability than species-specific models, and basal areas (BA), following Johnson et al. (1998). The
but their suitability is easily more limited because of highest vegetative tiller was defined as plant height, excluding
the inherent variability introduced in the models when reproductive tillers that may surpass vegetative tillers (whose
biomass is negligible, Guevara et al. 2002). All plants were
considering different species together.
hand clipped at ground level. Following Ramsay and Oxley
The tropical montane cloud forest – puna transition
(2001), fixed dead material (i.e. dead leaves still attached to
zone is experiencing increased deforestation and deg- the tussocks, Waren et al. 1994) was harvested, but ground
radation from fire and grazing pressure (Bustamante & litter was not. Most dead material remained attached to the
Bittencourt 2007), but comprehensive studies on the tussocks and the ground litter content was little compared
carbon stocks and productivity of these grasslands and with the attached dead material. All plants were fresh
the impacts of disturbance on them are scarce. This weighed, bagged and subsequently oven dried at 70°C to

doi:10.1111/aec.12098 © 2013 The Authors

Austral Ecology © 2013 Ecological Society of Australia
 G R A S S B I O M A S S A L L O M E T RY I N A N D E A N G R A S S L A N D S 3

Fig. 1. Location of the study area (black dot), next to the southern Andean corner of Manu National Park (in grey), Peru
(outline).

constant weight, and weighed to the nearest 0.1 g. Dried Table 1. Regression models tested to estimate above-
vegetative biomass was assumed to contain 50% carbon, ground biomass (AGB)
following the same criteria as previous studies in the same
area (Gibbon et al. 2010). Model Equation

I AGB = a + b(BA)
II AGB = a(BA)b
Regression models III AGB = a + b(hmax)
IV AGB = ahmaxb
We developed species-specific allometric equations for V AGB = a + b(CA)
the four dominant tussock grasses: Calamagrostis spp. VI AGB = a (CA)b
(n = 1334), Festuca dolichophylla (n = 223), Scirpus rigidus VII AGB= a + b(CA) + c(hmax)
(n = 638) and Juncus balticus (n = 177). These have growth VIII AGB = a(CA)b · (hmax)c
forms (i.e. tussocks) that make them suitable for developing IX AGB = a + b(BA) + c(CA)
allometric equations and constituted the majority (84.9%) of X AGB = a(BA)b · (CA)c
the total biomass at the study site. In addition, multispecies XI AGB = a + b(BA) + c(hmax) + d(CA)
equations were developed, combining data from the four XII AGB = a(BA)b · (hmax)c · (CA)d
aforementioned species. The remaining species were rosette
BA = basal area (cm2), CA = canopy area (cm2),
or cushion plants, and were therefore excluded from the
hmax = plant maximum height (cm).
analysis.
We tested 12 models (Table 1) based on linear relation-
ships (y = ax + b or Y = ax1 + bx2 + cx3 for multiple explana-
tory variables) and non-linear regressions according to the differently in power-law and in log-transformed models
power function (y = axb or y = ax1b · x2c for multiple explana- (Gingerich 2000; Kerkhoff & Enquist 2009; Xiao et al.
tory variables). One of the most common approaches to 2011). We therefore explored the error structure and dis-
fit power-law relationships is by log-transforming both tribution of power-law models of untransformed data
sides of the equation, yielding to a linear relationship, (non-linear relationships, NLR) and log-transformed linear
log(y) = log(a) + b · log(y). However, stochasticity manifests regressions (LR) following Xiao et al. (2011) using the smart

© 2013 The Authors doi:10.1111/aec.12098

Austral Ecology © 2013 Ecological Society of Australia
4 I . O L I V E R A S ET AL.

library of R (Warton et al. 2006). Data were not corrected for 84.9% of the total above-ground biomass. The sites
logarithmic bias (Baskerville 1972). The response variable had a high species richness, with a total number
was plant above-ground biomass (g), and explanatory vari- of 21 herbaceous species, 12 shrub species (being
ables were basal area (cm2), crown area (cm2), and maximum Senecio rhizomatosus, and Ageratina cuzcoensis the most
height (cm). We chose maximum height because it was a
common) and 5 fern species (belonging to Licopodium
better predictor than height encountered in the field in all
and Blecnum genera).
models (data not shown).
The best statistical model was selected by comparing the
squared coefficient of regression (r2), the residual standard
error (RSE) of residuals, and the Akaike’s information crite- Allometric equations
rion (AIC) for all models. While higher values of r2 show
better fit, least squares approximation minimizes RSE. Simi- Power-law relationships (models II, IV, VI, VIII, X
larly, the model with the lowest AIC indicates the best model, and XII, Table 1) were explored by applying the
that is, the one which offers the best fit whilst allowing for the logarithmic transformation of data and fitting a linear
number of parameters (Akaike 1974). relationship, as the analysis of the error distribu-
Models were developed and calibrated with 80% of the tions supported the assumption of multiplicative log-
data (Calamagrostis spp., n = 1085; Festuca dolichophylla, normal error better than non-linear regression error
n = 183; Scirpus rigidus, n = 508; and Juncus balticus,
(Appendix S1). Linear regressions did not show good
n = 140). The remaining 20% of data were used to validate
fits for any species-specific (Appendices S2–S5) or
the models by applying prediction using the best fitted
models (Calamagrostis spp., n = 249; Festuca dolichophylla, multispecies models (Table 2). Models improved
n = 40; Scirpus rigidus, n = 130; and Juncus balticus, n = 37). when the power function was used, either with a single
All statistical analyses were carried out using the R software or with multiple estimators (Table 2 and Appendices
2.13.2 (R Development Core Team 2012). S2–S5). The analysis showed that models based on
basal area and crown area were good estimators of
plant above-ground biomass in both species-specific
(Appendices S2–S5) and in multispecies regressions
RESULTS
(Table 2), but adding maximum height to the models
always improved them. However, maximum height
Above-ground biomass and carbon stock itself was a poor predictor of above-ground biomass.
The power-law relationship based on basal area and
The above-ground vegetation was estimated to contain crown area (model X) and the power-law relationship
on average 6.7 ± 0.2 (standard error) Mg ha−1dry based on basal area, maximum height and crown area
biomass, which translates into 3.35 ± 0.1 Mg C ha−1. (model XII), gave the best fits for all species and for
Calamagrostis spp., Scirpus rigidus, Festuca doli- the multispecies analysis (Table 2, Appendices S2–S5)
chophylla, and Juncus balticus contributed 68.4%, and we therefore tested the consistency of these
7.8%, 7.7%, and 1.0% to the total biomass, models by comparing 20% of the measured above-
respectively. The four focal species thus comprise ground biomass data (that had been not used to derive

Table 2. Regression models for four puna grass species combined (Calamagrostis sp., Scirpus rigidus, Festuca dolichophylla and
Juncus balticus)

Coefficient

Regression a b c d r2 RSE AIC

I −7.518 0.011 0.708 65.87 21 948

II 0.061 0.756 0.734 0.334 1 262
III −105.9 3.000 0.249 105.6 23 797
IV −3.045 2.554 0.517 0.452 2 448
V −43.94 0.082 0.678 69.16 22 139
VI −2.532 1.316 0.686 0.364 1 607
VII −38.29 0.084 −0.132 0.678 69.15 22 139
VIII −3.245 1.021 0.921 0.718 0.345 1 394
IX −35.51 0.007 0.043 0.783 56.75 21 365
X 5.5e-3 0.478 0.675 0.817 0.278 548
XI −29.62 0.006 −0.138 0.045 0.784 56.72 21 364
XII 1.0e-3 0.480 0.935 0.373 0.851 0.251 150

The best-fit parameters are reported for each model, squared coefficient of regression (r2), residual standard error (RSE),
AIC = Akaike’s Information Criterion, degrees of freedom = 1956.

doi:10.1111/aec.12098 © 2013 The Authors

Austral Ecology © 2013 Ecological Society of Australia
 G R A S S B I O M A S S A L L O M E T RY I N A N D E A N G R A S S L A N D S 5

0.5 1.0 1.5 2.0 2.5

Model X, multispecies Model XII, multispecies

(0.97 ± 0.02)
y = (1.12 ± 0.03)x y = (1.13 ± 0.02)x
(0.97 ± 0.01)
2
r = 0.815 RSS = 0.274 2
r = 0.857 RSS = 0.241

2.5

2.0
AGB observed (g)

1.5

1.0

0.5

0.5 1.0 1.5 2.0 2.5

AGB predicted (g)

Fig. 2. Multispecies regressions between predicted above-ground biomass (AGB predicted) with Models X and XII and the
above-ground biomass collected in the field (AGB observed). Each cross corresponds to an individual tussock. Coefficient values
are expressed with the standard error (± SE). The black line corresponds to the model regression, and the dashed line
corresponds to the 1:1. RSS corresponded to residual standard error.

the models) with the estimated above-ground biomass and XII for Scirpus rigidus (Model X: r2 = 0.806,
by the models. Model XII gave a slightly better fit than RSE = 0.222, AIC = −82.1; Model XII: r2 = −0.842,
model X for all species, probably because of the inclu- RSE = 0.200, AIC = −186, Appendix S3), Festuca
sion of maximum height (Fig. 2, Appendices S6,S7). dolichophylla (Model X: r2 = 0.784, RSE = 0.234,
The multispecies analysis confirmed models X and AIC = −7.4; Model XII: r2 = 0.816, RSE = 0.218,
XII as the best models (Table 2), and the validation AIC = −33, Appendix S4) and Juncus balticus (Model
identified model XII as the best one for the X: r2 = 0.766, RSE = 0.242, AIC = −4.57; Model XII:
multispecies equation (Fig. 2). Comparing this analy- r2 = 0.778, RSE = 0.238, AIC = 0.70, Appendix S5).
sis to the species-specific models, the multispecies Models X and XII for Calamagrostis spp. had a higher
models X and XII (Table 2) had higher regression coefficient of regression and lower RSE and AIC
coefficients and higher RSE and AIC than models X (Model X: r2 = 0.841, RSE = 0.246, AIC = 37.8;
© 2013 The Authors doi:10.1111/aec.12098
Austral Ecology © 2013 Ecological Society of Australia
6 I . O L I V E R A S ET AL.

Model XII: r2 = 0.855, RSE = 0.235, AIC = −59.7, This study shows that basal area, maximum height
Appendix S2) than the multispecies models X and XII and crown area provide a reliable basis for above-
(Table 2), being consistent with validation data ground biomass estimation in tussock grasslands, using
(Fig. 2, Appendices S6,S7). allometric equations. Basal area and crown area were
the most influential predictors in our models, but
adding maximum height always improved the fit of the
DISCUSSION models. Other studies in tussock grasslands (Andariese
& Covington 1986; Johnson et al. 1998; Nafus et al.
The above-ground vegetation was estimated to contain 2009) have also found basal area to be the main
3.35 Mg C ha−1. This value is similar to results from explanatory variable. However, results differ with
Ecuadorian páramo grasslands, where carbon densi- regards to inclusion of other variables. Johnson et al.
ties of 4.0 and 4.2 Mg C ha−1 were found at 3750 and (1998), Andariese and Covington (1986) and Nafus
4000 m a.s.l., respectively (Ramsay & Oxley 2001). et al. (2009) reported small or no improvement when
However, our results contrast significantly with height was added to their models. Guevara et al. (2002)
Gibbon et al. (2010) who estimated above-ground reported that adding height improved the model for
carbon density at 6.5 Mg C ha−1 for their puna sites. some species investigated. Nevertheless, equations only
This difference may be explained by the fact that based on maximum height were not good predictors of
Gibbon et al. (2010) also quantified the contribution above-ground biomass, nor were equations based on
of litter and moss to the carbon pool. The litter con- crown area. Basal area was shown to be the main
tribution in the study area was, however, negligible determinant of goodness of fit. This parameter may be
because most dead material remained attached to the time consuming to measure in field, but is still a useful
tussock (up to 70%, Hofstede et al. 1995, Appendix tool to allow for a relatively fast assessment of above-
S8). ground biomass, especially in such remote ecosystems.
Several other biomass studies have been carried out The Calamagrostis spp. allometric equations gave
in a selection of high-altitude tropical and temperate better fits than the multispecies model, but the other
grasslands (Hofstede et al. 1995), but differences in species-specific equations had similar estimation
sampling techniques make comparisons difficult. results to the multispecies model. Therefore, the equa-
More studies with a standardized methodology should tions developed for Calamagrostis spp. based on BA
be conducted in order to enhance the capability of and CA (models X and XII) would be the most appro-
cross-site comparisons, which will ultimately lead to a priate in grassland communities where Calamagrostis
better understanding of the spatial distribution of spp. is the dominant tussock grass. However, for
carbon content in high altitude tropical grassland species-rich tussock grassland communities, the
ecosystems. Another source of variation among studies multispecies models based on basal area and crown
is seasonality and time of biomass sampling. In the area (models X and XII) would give a more solid
current study, we included live and dead above-ground prediction of above-ground biomass. Nafus et al.
biomass to avoid any seasonal effects, although season- (2009) also developed multispecies grassland equa-
ality in the study area is almost non-existent, because tions in a study conducted in Arizona. Independent of
radiation and temperature remain fairly constant the variables included, the successful combination of
throughout the year and relative humidity is still above different species into one model may have important
85% in the driest months (July–August, Girardin et al. implications for future estimates of above-ground
2013). carbon stocks in tussock grasslands using allometric
The current study only reported for above-ground equations, as it eliminates the need of species identifi-
carbon stocks because the goal of the study was to cation and destructive sampling. However, it must be
provide estimation of the above-ground biomass as underlined that multispecies models are only an
well as sound allometric equations that allow for non- advantage in general above-ground biomass, produc-
destructive estimations of above-ground biomass. The tivity and carbon stock measurements. In ecological
contribution of below-ground biomass is thought to be studies, it would be of interest to know the species
very important in open high-altitude grasslands, with composition of the investigated area, especially in such
similar carbon stocks above-ground and below-ground highly diverse ecosystems. It is also important to note
(Oliveras, unpubl. data, 2012). that, although these allometries may be applicable in
Almost 85% of the vegetation was composed of four most tropical Andean grasslands with similar species
tussock grassland species, of which Calamagrostis spp. composition (thus dominated by Calamagrostis sp.,
accounted for most of the biomass. However, 38 dif- Juncus sp., Scirpus sp. and Festuca sp. tussocks), vali-
ferent species of grasses, shrubs and ferns were iden- dation studies for other areas would be very valuable to
tified, showing the high diversity of these high-altitude test its wider applicability. This is especially important
grasslands, as has also been found in other studies in for areas with different environmental factors or dif-
puna and páramo (Luteyn 1999). ferent species composition.
doi:10.1111/aec.12098 © 2013 The Authors
Austral Ecology © 2013 Ecological Society of Australia
 G R A S S B I O M A S S A L L O M E T RY I N A N D E A N G R A S S L A N D S 7

The current study provides a reliable method for September 2012.] Available from URL: http://www.fao.org/
non-destructive biomass and productivity estimation in docrep/w4095e/w4095e00.htm
Bustamante J. A. & Bittencourt M. D. (2007) Ecological zoning
high-altitude tussock grasslands. Most allometric equa-
of an Andean grasslands (puna) at the Manu Biosphere
tions have been developed for tropical forests, but grass- Reserve, Peru. Int. J. Environ. Sustain. Dev. 6, 357–72.
lands can also store large amounts of carbon and have as Chave J., Andalo C., Brown S. et al. (2005) Tree allometry and
much productivity as forest ecosystems (Scurlock & improved estimation of carbon stocks and balance in tropi-
Hall 1998). The applicability of these results may be cal forests. Oecologia 145, 87–99.
useful in remote and often challenging puna grasslands Conant R. T. (2010) Challenges and Opportunities for Carbon
of the Tropical Andes, where destructive sampling can Sequestration in Grassland System – A Technical Report on
Grassland Management and Climate Change Mitigation. Food
be of limited applicability due to logistical constraints.
and Agriculture Organization of the United Nations, Rome.
These equations can be also applicable to measure the Feldspauch T. R., Banin L., Phillips O. L. et al. (2011) Height-
productivity of grassland systems and for establishing diameter allometry of tropical forest trees. Biogeosciences 8,
reference levels before the adoption of carbon compen- 1081–106.
sation mechanisms or grassland management policies, Gibbon A., Silman M. R., Malhi Y. et al. (2010) Ecosystem
as well as for measuring the impact of land use changes carbon storage across the grassland–forest transition in the
in Andean ecosystems. high Andes of Manu National Park, Peru. Ecosystems 13,
1097–111.
Gingerich P. D. (2000) Arithmetic or geometric normality of
biological variation: an empirical test of theory. J.Theor. Biol.
ACKNOWLEDGEMENTS 204, 201–21.
Girardin C. A. J., Farfan-Rios W., Garcia K. et al. (2013) Spatial
patterns of above-ground structure, biomass and composi-
We would like to express our gratitude to the Depart-
tion in a network of six Andean elevation transects. Plant
ment of Ecology and Natural Resource Management Ecol. Div. doi: 10.1080/17550874.2013.820222.
at the Norwegian University of Life Sciences, the Guevara J. C., Gonnet J. M. & Estevez O. R. (2002) Biomass
Norwegian State Educational Loan Fund and Andreas estimation for native perennial grasses in the plain of
and K. Ludvig Endresens Legat for funding the Mendoza, Argentina. J. Arid Environ. 20, 613–19.
project. We would also like to thank Guissela Romani, Hofstede R. G. M., Mondragon Castillo M. X. & Rocha Osorito
Efrain Choque, Sandra Almeyda, Jose Kala, Jose C. M. (1995) Biomass of grazed, burned, and undisturbed
páramo grasslands, Colombia. I. Aboveground Vegetation.
Antonio and Walter Huaraca Huasco for assistance
Arct. Alp. Res. 27, 1–12.
with the fieldwork. The fieldwork was conducted with Johnson P. S., Johnson C. L. & West N. E. (1998) Estimation of
the generous assistance from ACCA (Asociación para phytomass for ungrazed crested wheatgrass plants using
la Conservación de la Cuenca Amazónica). allometric equations. J.Range Manage. 41, 421–5.
Kerkhoff A. J. & Enquist B. J. (2009) Multiplicative by nature:
why logarithmic transformation is necessary in allometry.
J. Theor. Biol. 257, 519–21.
REFERENCES Luteyn J. L. (1992) Páramos: why study them? In: Páramo:
An Andean Ecosystem under Human Influence (eds Balslev &
Akaike H. (1974) A new look at the statistical model Luteyn) pp. 1–14. Academic Press, London.
identification. IEEE Trans. Automat. Contr. 19, 719–23. Luteyn J. L. (1999) Páramos: A Checklist of Plant Diversity,
Andariese S. W. & Covington W. W. (1986) Biomass estimation Geographical Distribution and Botanical Literature. Memoirs
for four common grass species in Northern Arizona of the New York Botanical Garden. New York Botanical
ponderosa pine. J. Range Manage. 39, 472–3. Garden, New York.
Asner G. P. (2009) Tropical forest carbon assessment: integrat- Malhi Y. & Phillips O. L. (2004) Tropical forests and global
ing satellite and airborne mapping approaches. Environ. Res. atmospheric change: a synthesis. Phil. Trans. R. Soc. B Biol.
Lett. 4, 034009. Sci. 359, 549–55.
Asner G. P. (2011) Painting the world REDD: addressing scien- Maniatis D. & Mollicone D. (2010) Options for sampling and
tific barriers to monitoring emissions from tropical forests. stratification for national forest inventories to implement
Environ. Res. Lett. 6, 021002. REDD+ under the UNFCCC. Carbon Balance Manage. 5,
Asner G. P., Mascaro J., Muller-Landau H. C. et al. (2012) A art 9, 14.
universal airborne LiDAR approach for tropical forest Nafus A. M., McClaran M. P., Archer S. R. & Throop H. L.
carbon mapping. Oecologia 168, 1147–60. (2009) Multispecies allometric models predict grass
Baldassini P., Volante J. N., Califano L. M. & Paruelo J. M. biomass in semidesert rangeland. Rangel. Ecol. Manage. 62,
(2012) Regional characterization of the structure and 68–72.
productivity of the vegetation of the Puna using MODIS Navar J., Mendez E., Najera A., Graciano J., Dale V. & Parresol
images. Ecología Austral 22, 22–32. B. (2004) Biomass equations for shrubs species of
Baskerville G. (1972) Use of logarithmic regression in the esti- Tamaulipan thorn scrub of North-eastern Mexico. J. Arid
mation of plant biomass. Can. J. For. Res. 2, 49–53. Environ. 59, 657–74.
Brown S. (1997) Estimating biomass and biomass change Oliveras I., Malhi Y., Salinas N. et al. (2013) Changes in forest
of tropical forests: a primer. FAO For. Pap. 134 [Cited 24 structure and composition after fire in tropical montane

© 2013 The Authors doi:10.1111/aec.12098

Austral Ecology © 2013 Ecological Society of Australia
8 I . O L I V E R A S ET AL.

cloud forests near the Andean treeline. Plant Ecol. Div. analysis of vegetation records from 1945 to 1994. Aust. J.
doi: 10.1080/17550874.2013.816800. Bot. 42, 607–39.
Phillips O. L., Malhi Y., Higuchi N. et al. (1998) Changes in the Warton D. I., Wright I. J., Falster D. S. & Westoby M. (2006)
carbon balance of tropical forests: evidence from long-term Bivariate line-fitting methods for allometry. Biol. Rev. 81,
plots. Science 282, 439–42. 259–91.
R Development Core Team (2012) R: A Language and Environ- White S. & Maldonado F. (1991) The use and conservation of
ment for Statistical Computing. R Foundation for Statistical natural resources in the Andes of southern Ecuador. Mt. Res.
Computing, Vienna. ISBN 3-900051-07-0, [Cited 28 Dev. 11, 37–55.
August 2012.] Available from URL: http://www.r-project Xiao X., White E., Hooten M. B. & Durham S. L. (2011) On
.org the use of log-transformation vs. nonlinear regression for
Ramsay P. M. & Oxley E. R. B. (1996) Fire temperatures and analyzing biological power laws. Ecology 92, 1887–94.
post fire plant community dynamics in Ecuadorian grass Zimmermann M., Meir P., Silman M. R. et al. (2010) No dif-
páramo. Vegetatio 124, 129–44. ferences in soil carbon stocks across the tree line in the
Ramsay P. M. & Oxley E. R. B. (2001) An assessment of Peruvian Andes. Ecosystems 13, 62–74.
aboveground net primary productivity in Andean grasslands
of central Ecuador. Mt. Res. Dev. 21, 161–7.
Saatchi S., Harris N. L., Brown S. et al. (2011) Benchmark map
of forest carbon stocks in tropical regions across three SUPPORTING INFORMATION
continents. Proc. Natl Acad. Sci. USA 108, 9899–904.
Saatchi S., Houghton R. A., Dos Santos Alvalá R. C., Soares
J. V. & Yu Y. (2007) Distribution of aboveground live Additional Supporting Information may be found in
biomass in the Amazon basin. Global Change Biol. 13, 816– the online version of this article at the publisher’s
37. web-site:
Scurlock J. M. O. & Hall D. O. (1998) The global carbon sink: a
grassland perspective. Global Change Biol. 4, 229–33. Appendix S1. Akaike’s Information Criterion coeffi-
Scurlock J. M. O., Johnson K. & Olson R. J. (2002) Estimating cients on untransformed data.
net primary productivity from grassland biomass dynamics Appendix S2. Regression models for Calamagrostis
measurements. Global Change Biol. 8, 736–53. sp.
UNFCCC (United Nations Framework Convention on Climate Appendix S3. Regression models for Scirpus rigidus.
Change) (2010) Outcome of the work of the Ad Hoc
Appendix S4. Regression models for Festuca
Working Group on Long-term Cooperative Action under
the Convention: policy approaches and positive incentives dolichophylla.
on issues relating to reducing emissions from deforestation Appendix S5. Regression models for Juncus balticus.
and forest degradation in developing countries; and the Appendix S6. Species-specific regressions between
role of conservation, sustainable management of forests predicted above-ground biomass with model X and
and enhancement of forest carbon stocks in developing the above-ground biomass collected in the field.
countries. UNFCCC COP 16, Cancun. [Cited 23 July Appendix S7. Species-specific regressions between
2013.] Available from URL: http://unfccc.int/resource/docs/
2010/cop16/eng/07a01.pdf#page=2
predicted above-ground biomass with Model XII and
Waren C. H. A., Paps W. A. & Williams R. J. (1994) Long-term the above-ground biomass collected in the field.
vegetation change in relation to cattle grazing in subalpine Appendix S8. Photo of one of the Wayqecha
grassland and heathland on the Bogong High Palins: an transects.

doi:10.1111/aec.12098 © 2013 The Authors

Austral Ecology © 2013 Ecological Society of Australia