of 12
1
oŽ.:ȋͬͭͮͯͰͱͲͳʹ͵Ȍ
Scientific Reports
| (2024) 14:6189
|
https://doi.org/10.1038/s41598-024-56362-1
www.nature.com/scientificreports
An emergent constraint
on the thermal sensitivity
of photosynthesis and greenness
in the high latitude northern
forests
Junjie Liu
1,2
*
& Paul O. Wennberg
2
*
Despite the general consensus that the warming over the high latitudes northern forests (HLNF) has
led to enhanced photosynthetic activity and contributed to the greening trend, isolating the impact
of temperature increase on photosynthesis and greenness has been difficult due to the concurring
influence of the
CO
2
fertilization effect. Here, using an ensemble of simulations from biogeochemical
models that have contributed to the Trends in Net Land Atmosphere Carbon Exchange project
(TRENDY), we identify an emergent relationship between the simulation of the climate-driven
temporal changes in both gross primary productivity (GPP) and greenness (Leaf Area Index, LAI) and
the model’s spatial sensitivity of these quantities to growing-season (GS) temperature. Combined
with spatially-resolved observations of LAI and GPP, we estimate that GS-LAI and GS-GPP increase
by 17.0 ± 2.4% and 24.0
± 3.0% per degree of warming, respectively. The observationally-derived
sensitivities of LAI and GPP to temperature are about 40% and 71% higher, respectively, than the
mean of the ensemble of simulations from TRENDY, primarily due to the model underestimation of
the sensitivity of light use efficiency to temperature. We estimate that the regional mean GS-GPP
increased 28.2
± 5.1% between 1983–1986 and 2013–2016, much larger than the 5.8
± 1.4% increase
from the
CO
2
fertilization effect implied by Wenzel et
al. This suggests that warming, not
CO
2
fertilization, is primarily responsible for the observed dramatic changes in the HLNF biosphere over
the last century.
Keywords
Thermal sensitivity, Photosynthesis, High latitude northern forests, Emergent constraint
Temperature over the northern hemisphere (NH) high latitudes (>
50° N) has been increasing at more than
twice the rate of the rest of the globe. Given the continuing increases in greenhouse gases, this trend is unlikely
to slow in the foreseeable future (IPCC AR6). Concurrently, ground and satellite observations have illustrated
dramatic changes in terrestrial biosphere activity: longer growing
season
1
,
2
, a greening trend over the majority of
the
region
3
,
4
7
, and an increase in carbon uptake from the atmosphere leading to an enhancement in the atmos-
pheric
CO
2
seasonal cycle
amplitude
8
,
9
12
. These trends alter photosynthesis, a process that converts light energy
into chemical energy through electrochemistry fixing atmospheric carbon into organic carbon compounds
through
carboxylation
13
. The maximum rate of both electron transport and carboxylation increases exponen-
tially with temperature before reaching an optimal
temperature
14
,
15
. The mean growing season temperature over
land
> 50° N is between 5 and 18 °C (Fig. S1), generally lower than the optimal growth temperature for most
plants even accounting for acclimation and
adaptation
16
,
17
. Thus, the increase in temperature has been proposed
as a mechanism that drives the increase of photosynthesis over the
region
3
,
8
,
9
. However, diagnosing the extent
to which temperature enhances plants growth is complicated by the co-occurring increase in
CO
2
that enhances
photosynthesis by increasing the difference in the rate of transport of
CO
2
and water through the stomata and
increasing the efficiency of the carboxylating enzyme in
C
3
plants
18
,
19
, the so called “
CO
2
fertilization
” effect
20
.
The
CO
2
fertilization and warming effect on photosynthesis are relatively well-understood at leaf and canopy
scale
15
, but there remains significant uncertainty in predictions of how these changes are altering the global
OPEN
1
Jet Propulsion Laboratory, California Institute of Technology, Pasadena, USA.
2
California Institute of Technology,
Pasadena, USA.
*
email: junjie.liu@jpl.nasa.gov
; wennberg@caltech.edu
2
Vol:.(1234567890)
Scientific Reports
| (2024) 14:6189 |
https://doi.org/10.1038/s41598-024-56362-1
www.nature.com/scientificreports/
carbon cycle. Consequently, the current state-of-science terrestrial biogeochemical models (TBMs) show a large
range of the response of photosynthesis to climate change and
CO
2
increase
21
,
24
. For the HLNF, the uncertainty
in the simulated net carbon uptake is close to 100%
22
.
A few past studies derived empirical emergent relationships between observables and model simulations of
the carbon-climate feedback factor
γ
over the
tropics
23
25
, the photosynthesis-concentration feedback factor
β
GPP
over the northern hemisphere (NH) mid to high
latitudes
27
, and then used observations to constrain
γ
and
β
GPP
over these regions. Winkler et al.
26
applied a similar concept to constrain the combined
γ
GPP
and
β
GPP
effect over
the northern high latitudes. However, the quantitative impact of temperature increases on plant growth over
the high latitudes (the
γ
GPP
effect) is still elusive and is anticipated to be distinct from that in the tropics due to
the large climatological
differences
27
. Over the warm tropics, increasing temperature can reduce the terrestrial
biosphere net carbon uptake from the atmosphere, causing positive carbon-climate
feedbacks
23
, while over the
high latitudes, the warming trend generally invigorates plants growth, enhancing
CO
2
uptake and thereby acting
as a negative feedback to the
climate
27
. The projections of future climate change critically depend on the under
-
standing of these carbon-climate feedbacks. Here, we denote the photosynthesis-CO
2
-concentration feedback
factor as
β
GPP
and photosynthesis-climate feedback as
γ
GPP
to distinguish them from the carbon-climate feedback
factor
γ
that describes changes of carbon pools due to
climate
28
.
In this study, we expose an emergent relationship in the high latitude northern forest (HLNF) between the
spatially-derived
sensitivity of photosynthesis and greenness to temperature and the
temporal
changes driven
by the changing climate from an ensemble of TBMs. Using the observed
spatial
sensitivity of both photosyn-
thesis and greenness, we constrain how photosynthesis and greenness are responding to warming
GPP
)
that
has occurred over the past decades (Fig. S1). Our approach is complementary to the tropical carbon-climate
feedback factor proposed by Cox et al.
23
and Sullivan et al.
24
, the
β
GPP
effect on GPP over temperate and boreal
forest derived by Wenzel et al.
29
, and the combined
γ
GPP
and
β
GPP
effect over the high latitudes by Winkler et al.
26
.
The emergent relationship derived here is based on our earlier
study
10
which introduced the use of the
spatial
sensitivity of photosynthetic activity to temperature to infer historical
temporal
changes in photosynthesis. As
the spatial gradient in
CO
2
is both transient in nature (due to atmospheric transport) and always small (generally
less than 15 parts per million (ppm)), the
CO
2
effect on the
spatial
sensitivity of GPP and greenness to tempera-
ture is negligible. The spatial sensitivity of GPP and greenness to temperature reflects the potential equilibrium
sensitivity of vegetation to warming that includes the effect of real-world physiological and ecological
adaption
24
,
thus it is suitable to infer long-term sensitivities of vegetation to warming.
Results
Spatially-derived sensitivity of GPP and greenness to temperature
The emergent constraint on the warming effect on plants growth derived in this study builds upon the relation-
ships between
spatial
sensitivity of GPP and greenness to temperature and their corresponding
temporally-derived
sensitivities to temperature. Thus, we first quantify the spatial sensitivity of GPP and greenness to temperature
(Fig. S1 and sections “
Materials and methods
”, “
Workflow to derive the observation-constrained photosynthesis/
greenness: climate feedback factors
”). As shown in Liu et al.
10
(Fig. S5), the observed spatially-derived sensitivity
of greenness to temperature over the HLNF is time-invariant. We anticipate that if the
spatially-derived
sensitivity
of GPP and greenness to temperature from models is also time-invariant, then the simulated
temporal
changes
of GPP and greenness caused by temperature would be similar to that predicted by the corresponding spatial
sensitivity to temperature. Thus, we first evaluated the temporal consistency of the simulated spatial sensitivities
of GPP and greenness to temperature in current terrestrial biogeochemical models. We examined both Leaf Area
Index (LAI) and GPP from an ensemble of TBMs from the Trends in Net Land Atmosphere Carbon Exchange
project (TRENDY) v6, and simulations where only
CO
2
was varied (S1) and simulations where both
CO
2
and
climate were varied (S2). The differences between these two runs reflect the impact of climate change only. LAI
is generally defined as one-half of the total green leaf area per unit horizontal ground surface area with unit
of
m
2
/m
2
30
; GPP is a function of both the absorption of photosynthetically active radiation (APAR) (related to
greenness) and light use efficiency (LUE), which is a function of many factors including environmental drivers,
e.g.,
temperature
31
. APAR is a product of photosynthetic radiation (PAR) and the fraction of absorbed PAR
(fPAR) by plants. As fPAR and LAI are interchangeable through Beer’s law
approximation
32
, we estimate fPAR
from the LAI reported from the TRENDY models to disentangle the contributions of both greenness and LUE
to the sensitivity of GPP to temperature.
Since the S1 runs are driven by a 20 years repeating climatology, we calculated 20 years mean GPP and LAI
from the TRENDY models starting at 1901. To increase the sample size, we subsampled these into 10-year
overlap (e.g., 1901–1920, 1911–1930, etc.), which results in 10 groups each for S1 and S2 runs for each model.
We selected grid cells (>
50° N) with at least 40% tree cover fraction (Table S1), and then fitted the correlation
between growing season mean temperature (GS-T) and growing season GPP and LAI for each group from each
model using an exponential fit (section “
Materials and methods
”). We selected 40% as a forest threshold to
remove grid cells with dominant grassland and cropland vegetation types in both model runs and observations,
since water availability could be dominant climate driver over these vegetation
types
33
.
An exponential, rather than linear relationship, is used following the general description of the dependence
of both electron transport and carboxylation on temperature in cold ecosystems, such as those analyzed
here
14
,
15
(section “
Materials and methods
”). Farquar et al.
15
shows that the carbon assimilation rate follows a nonlinear
curve before reaching an optimal temperature. The same study shows that the temperature dependence of the
kinetic properties of rubisco carbonxylase rate follows an exponential relationship. Furthermore, using a linear
model the limiting behavior at low temperature is pathological: the implied photosynthetic rate would be negative
even at temperatures above freezing (Fig. S3). The exponential fitting also allows us to calculate the Q10 values