Minoli et al. (2022). Global crop yields can be lifted by timely adaptation of growing periods to climate change

Adaptive management of crop growing periods by adjusting sowing dates and
cultivars is one of the central aspects of crop production systems, tightly
connected to local climate. However, it is so far underrepresented in crop-
model based assessments of yields under climate change. In this study, we
integrate models of farmers

’

decision making with biophysical crop modeling

at the global scale to simulate crop calendars adaptation and its effect on crop
yields of maize, rice, sorghum, soybean and wheat. We simulate crop growing
periods and yields (1986-2099) under counterfactual management scenarios
assuming no adaptation, timely adaptation or delayed adaptation of sowing
dates and cultivars. We then compare the counterfactual growing periods
and corresponding yields at the end of the century (2080-2099). We

ﬁ

nd that

(i) with adaptation, temperature-driven sowing dates (typical at latitudes
>30°N-S) will have larger shifts than precipitation-driven sowing dates (at
latitudes <30°N-S); (ii) later-maturing cultivars will be needed, particularly at
higher latitudes; (iii) timely adaptation of growing periods would increase
actual crop yields by ~12%, reducing climate change negative impacts and
enhancing the positive CO

2

fertilization effect. Despite remaining uncertain-

ties, crop growing periods adaptation require consideration in climate change
impact assessments.

Plant phenology is the sequence of morphological and physiological
events in the annual plant growth cycle. As the onset and rate of pro-
gress of phenological events are primarily driven by temperature and
day length, the rising global temperatures due to climate change alter
the timing of plants

’

phenology. Phenological records of the recent

past in Europe show that the responses of plants to temperature
increase have determined earlier and faster phenological progress of
both wild and cultivated plants

. Faster-growing cycles associated

with higher temperatures are considered one of the main mechanisms
of climate change impacts on crop yields

sara.minoli@pik-potsdam.de

,

4

.

However, the phenology of annual crops depends also on farmers

’

decisions on sowing dates and varietal choice

5

–

7

. Annual crops have

growing-period durations (here de

ﬁ

ned as days from sowing to

maturity) of less than a year. They are thus exposed only to a fraction of
the annual weather

8

, and farmers manage crops to enhance yields and

resource use by growing them during the most favorable part of the

Received: 18 October 2021

Accepted: 25 October 2022

Check for updates

1

Potsdam Institute for Climate Impact Research (PIK), Member of the Leibniz Association, 14412 Potsdam, Germany.

2

NASA Goddard Institute for Space

Studies, New York, NY 10025, USA.

3

Columbia University, Climate School, New York, NY 10025, USA.

4

Technical University of Munich, Department of Life

Science Engineering, 85354 Freising, Germany.

5

Sustainable Intensi

ﬁ

cation Program, International Maize and Wheat Improvement Centre (CIMMYT), South

Asia Regional Of

ﬁ

ce, Khumaltar, Lalitpur 44700, Nepal.

6

Water Resources Management Group, Wageningen University & Research, 6708 PB Wageningen, the

Netherlands.

7

Centre for Crop Systems Analysis, Wageningen University & Research, 6708 PB Wageningen, the Netherlands.

e-mail:

Nature Communications

| (2022) 13:7079

1

12

3

4

5

6

78

12

3

4

5

6

7

8

year. Through plant domestication and breeding, humans have arti

ﬁ

-

cially modi

ﬁ

ed the response of cultivated varieties (cultivars) to tem-

perature and day length and by that, have expanded the area where
crop species can be grown

8

,

9

. Farmers can thus draw from a vast

assortment of cultivars that differ by their thermal time (maturity
classes), photoperiod (long-day, short-day or day-neutral) and verna-
lization requirements (vernalizing, non-vernalizing or facultative),
which are considered central for adapting cropping systems to chan-
ging climatic conditions

9

It is essential to accurately represent growing periods in crop

simulation models for estimating yield responses to climatic factors

10

.

At the global scale, calibration of crop growing periods to reported
historical crop calendars

–

has been shown to improve simulation of

observed interannual yield variability

14

. However, for climate impact

assessments, previous studies have not assessed farmers

’

adaptive

behavior and relied on observation-based static crop calendars,
thereby either assuming unchanged sowing dates and cultivar
selection

, or simply adapting crop cultivars to regain the duration of

the reference growing period that is otherwise shortened with accel-
erated phenology under climate change

17

.

Studying changes in global crop growing periods is constrained

by the lack of suf

ﬁ

cient information on management (e.g., sowing and

harvest dates, cultivar choice), the timing of phenological phases (e.g.,

ﬂ

owering, maturity), and crop development parameters (e.g., growing

degree days, base temperatures, sensitivity to photoperiod). Recent
efforts have started to explore decision-making rules that dynamically
simulate farming practices to adjust crop growing periods and have
found that climate is the primary driver of farmers

’

sowing date and

cultivar choices

18

–

, explaining the current patterns of cropping

calendars. Yet, the adaptation of such practices under climate change
and their impact on future crop yields remains poorly understood at
the global scale.

Here, we investigate how accounting for farmers

’

adaptive man-

agement affects estimates of future global crop yields under climate
change. We simulate sowing and maturity dates adaptation by com-
bining two rule-based methods

. These methods simulate farmers

’

crop-calendar selection worldwide based on agro-climatic principles
that are primarily driven by crop physiology and long-term changes in
monthly temperature, precipitation and their seasonality. We improve
these methods and jointly evaluate them against observed crop
calendars

–

by driving them with observation-based climate

Thereafter, we derive projections of location-speci

ﬁ

c sowing and

maturity dates adapted to one historical (1986

–

2005) and two future

(2060

–

2079, 2080

–

2099) climatic periods for

ﬁ

ve crops (maize, rice,

sorghum, soybean, wheat) and two water management regimes
(rainfed and irrigated). To represent cultivars adapted to these crop
calendars, we compute thermal unit requirements (TU

req

, °C day)

between sowing and maturity dates

in each period.

We use the process-based global gridded crop model LPJmL

24

25

to

simulate annual crop growing periods and yields from 1986 to 2099
under a reference scenario of the actual crop- and grid-cell-speci

ﬁ

c

historical management practices (irrigation, fertilization, tillage, and
crop residue management), assuming that no management action
would be taken in response to climate-induced impacts. We then
compare the counterfactual growing periods (with adaptation) and
corresponding yields at the end of the century (2080

–

2099) in order

to quantify how much (i) future sowing and maturity dates would shift
under different adaptation and climate scenarios and (ii) future crop
yields could be enhanced, if farmers adapted growing periods to cli-
mate change. We simulate three counterfactual sowing date and cul-
tivar scenarios where farmers are assumed to (i) continue apply all
historical management practices, including sowing dates and cultivars,
also in the future (

no adaptation

); (ii) timely adapt sowing dates and

cultivars to future climate (

timely adaptation

); (iii) adapt sowing dates

and cultivars with 20-years delay (

delayed adaptation

) (Table

).

We also test the sensitivity of yields to single measures, adapting either
sowing dates or cultivars individually, while keeping the other

ﬁ

xed at

the reference level (

sowing-day only

and

cultivar only

adaptation).

In each simulation unit, different crops are assumed to grow on

separate stands and to have only one crop cycle (sowing-to-harvest)
per year. Therefore, multi-cropping systems (e.g., rice-rice systems in
South-East Asia) are not explicitly represented. Adaptation measures
need to be climate-scenario speci

ﬁ

c and we test several General Cir-

culation Models (GCMs) to assess how robust results are against
uncertainties in climate projections (Supplementary Figs. 1

–

4). Further

details are described in

“

Methods

”

.

Results

Simulated present-day and future adapted sowing dates

At latitudes higher than roughly 30°N

–

S, growing periods are mostly

temperature-driven (Supplementary Fig. 5), with sowing dates that
depend on the start of either the warm (for spring crop types), or the
cold (for winter crop types) season. In the tropics, growing periods are
largely driven by precipitation seasonality (Supplementary Fig. 5), with
sowing dates occurring at the onset of the main rainy season (

ﬁ

rst of

the 120 consecutive wettest days in the year), except for winter wheat
that depends on temperature seasonality only, if present. For wheat,
we improved the de

ﬁ

nition of winter and spring types compared to the

original approach

, including the following combinations of sowing

seasons and vernalization requirements of the cultivars:

winter wheat

with vernalization

, sown in fall and going dormant over winter;

winter

wheat without vernalization

, sown in fall and grown over mild winters;

spring wheat without vernalization

, sown in spring. This distinction

allows to better simulate winter wheat growing seasons in warmer
climates, where temperatures are not low enough for dormancy (e.g.,
India) (Supplementary Fig. 6).

Timely adaptation

to future climate (2080

–

2099) markedly

shifts

temperature-driven

sowing

dates

in

the

extratropics

(latitudes > 30°N

–

S). Due to an earlier spring onset, spring crops

(maize, rice, sorghum and soybean) show advancing sowing dates
between

−

30 and

−

10 days, depending on the region and GCM (Fig.

a

and Supplementary Figs. 7

–

10A). An earlier spring allows for taking

advantage of a longer growing season. In contrast, adapted sowing
dates of

winter wheat with vernalization

are postponed (+10 to

+30 days) (Fig.

a and Supplementary Figs. 7

–

10A), to avoid that the

crop develops too much before going dormant, which would make it
more susceptible to frost damages and disease. Larger differences
(>+60 days) found for wheat in high latitudes (>50°N

–

S) re

ﬂ

ect the

switch from spring to winter types, meaning that sowing dates are
pushed several months forward, from spring to fall. This is the case in
regions across North America, Europe, and Asia, which in the
1986

–

2005 reference period the rule-based model

ﬁ

nds to be only

suitable for growing spring wheat because of too long and harsh
winters, become suitable for growing winter wheat under climate
change (shorter and warmer winters).

Table 1 | Growing-periods adaptation protocol

Simulation setup

Period of growing-
period calibration

Period of simulation

Reference

1986

–

2005

1986

–

2005

No adaptation

1986

–

2005

2080

–

2099

Delayed adaptation

2060

–

2079

2080

–

2099

Timely adaptation

2080

–

2099

2080

−

2099

For each simulation setup, the time interval of growing-periods calibration and crop-model
simulation are speci

ﬁ

ed. Sowing dates, maturity dates, and resulting cultivar thermal unit

requirements (TUreq) are derived in each grid cell under the climate conditions of the respective
period. Growing-period calibrations and crop-model simulations for each period are performed
for four GCMs (HadGEM2-ES, GFDL-ESM2M, IPSL-CM5A-LR, MIROC5), respectively. See

“

Methods

”

for more details.

Article

https://doi.org/10.1038/s41467-022-34411-5

Nature Communications

| (2022) 13:7079

2

Sowing dates of

winter wheat without vernalization

remain mostly

unchanged or even advanced, as no direct frost damage is expected
over mild winters (without considering possible spring frost damage at
anthesis). Other factors, i.e., diseases, might affect sowing date deci-
sions, which however, are not considered in our purely climate-driven
approach. With few exceptions, precipitation-driven sowing dates
adapted to future climate show little changes (±10 days) across all
GCMs compared to the reference scenario. This indicates that, despite
the substantial changes in rainfall (Supplementary Fig. 4), the average
timing of the onset of the main rainy season is projected to not change
much in these particular climate scenarios (Fig.

a and Supplementary

Figs. 7

–

10A). Yet, future approaches would need to address also

changes in climate and growing periods of interannual variability.

Future maturity dates and cultivars adaptation

Changes between future adapted and non-adapted maturity dates
re

ﬂ

ect two different processes. On the one hand, non-adapted crops

are subject to an overall shortening of the growing-period durations
compared to the historical ones (Supplementary Fig. 11). This is a
consequence of increased temperatures, which drive faster phenolo-
gical development. On the other hand, the adaptation rule seeks the
most suitable maturity dates adapted to future climate, aiming to
extend the growing period while avoiding hot and/or dry periods that
could hamper grain

ﬁ

lling and reduce

ﬁ

nal yields. For example,

maturity dates of spring crops occur at the end of the summer, so that
the reproductive phase (grain

ﬁ

lling) takes place after temperatures

have reached their seasonal peak. Speci

ﬁ

cally, if temperatures remain

within an optimal range throughout the year (Supplementary Table 1),
grain

ﬁ

lling starts in the warmest month of the year. If instead, tem-

peratures exceed crop-speci

ﬁ

c optima, longer growing periods are

preferred, to place grain

ﬁ

lling after the hottest period has passed.

Therefore,

timely adapted

maturity dates are generally later than with

no adaptation

(Fig.

B and Supplementary Figs. 7

–

10B). Yet, earlier

maturity dates are found in regions with water limitations, where
maturity is anticipated to avoid terminal water stress.

Winter wheat, on the other hand, matures in the warmest month

of the year, if monthly average temperatures remain below 25 °C (see
Supplementary Table 1). However, if temperatures exceed 25 °C, wheat

matures on average on the day this threshold is exceeded, in order to
escape terminal heat stress (note that daily temperatures are derived
by linear interpolation of monthly values, see

“

Methods

”

). As a result,

we

ﬁ

nd that

timely adaptation

at mid-to-high latitudes, where tem-

peratures remain within an optimal range under future climate, leads
to later wheat maturity. In contrast, ideal maturity dates are advanced
across mid-latitude regions where monthly temperatures will
exceed 25 °C. This acts in the same direction as the phenological
shortening of the growing period that occurs under the non-adapted
scenario (Supplementary Fig. 11), resulting in small maturity dates
differences (<10 days) between the two scenarios (Fig.

b and Sup-

plementary Figs. 7

–

10B). At very high latitudes (e.g., Canada, Russia),

where a switch from spring to winter wheat is expected to occur,
maturity dates are also advanced, as wheat sown in fall generally
matures earlier.

Cultivars adapted to future climate generally require accumulat-

ing more thermal units to reach maturity than the non-adapted culti-
vars. The largest shifts are found in the extratropics, where the growing
seasons extend in both directions, with earlier sowing and later
maturity (except for wheat) (Supplementary Fig. 12). Tropical regions,
do not indicate many variations in sowing dates, but adaptation of
later-maturing cultivars allow for regaining some of the growing-
period length that is

“

lost

”

due to accelerated phenology (Supple-

mentary Fig. 11).

Yield responses to crop-calendar adaptation

Global crop yields on current cropland, simulated by the process-
based model LPJmL, bene

ﬁ

ts from phenological adaptation under

future climate (RCP6.0; 2080

–

2099). For all crops combined,

global yields would be 12% (GCM range 9

–

15%) higher compared

to a scenario without adaptation (Fig.

). The largest yield gains

are found for maize (+17%) and rice (+17%), and the smallest for
wheat (+7%). The bene

ﬁ

ts of adaptation re

ﬂ

ect the warming level

of the climate scenario, being generally larger under the warmer
(HadGEM2-ES) than under the cooler (GFDL-ESM2M) climate
scenario

26

with some variations across crops, due to different

spatial patterns of warming (Supplementary Fig. 3) and cultiva-
tion area (Supplementary Fig. 13).

Fig. 1 | Crop-calendars adaptation.

Differences (days) in simulated average sowing

(

a

) and maturity (

b

) dates between

timely adaptation

and

no adaptation

scenarios

for the same climate period (2080

–

2099, RCP6.0). Crop calendars are climate-

scenario (GCM) speci

ﬁ

c, here the mean changes across four GCMs are shown.

Results for individual GCMs and all crops are reported in the Supplementary
Information.

Article

https://doi.org/10.1038/s41467-022-34411-5

Nature Communications

| (2022) 13:7079

3

Although adaptation is key for lifting yields of all considered crops

under a warming climate, it is especially important for C

4

crops.

Indeed, the increasing atmospheric CO

2

underlying the climate change

scenarios considered here is able to over-compensate the negative
impacts due to warming (

−

12%), leading to overall higher global crop

yields at the end of the century (+14%) relative to the historical period,
even without adaptation (Supplementary Fig. 14). As expected, this
effect is more pronounced for C

3

crops (rice +24%, soybean +24% and

wheat +22%) than for C

4

(maize

−

2% and sorghum 0%), which are far

less responsive to CO

2

increase

27

. Furthermore, adaptation proved to

be nearly insensitive to the assumptions on the level of atmospheric
CO

2

in our simulations, as the responses to adaptation are similar

under

elevated

(Fig.

) and

static

CO

2

(Supplementary Fig. 15). This

makes adaptation effects independent of projected CO

2

responses,

which are still a large source of uncertainty in crop models

. Another

interesting aspect is the combined effects of increasing atmospheric
CO

2

, climate change and adaptation on yield variability, which we

quantify by the coef

ﬁ

cient of variation (Supplementary Fig. 16). There

is a general expectation that the interannual variability in crop yields
will increase under climate change

28

. Our simulations support this

thesis only under the (unrealistic) scenario without CO

2

increase (

no

adaptation without CO

2

), whereas we

ﬁ

nd even slight declines in yield

variability if CO

2

is included (

no adaptation

). For rice, sorghum and

wheat, the two adaptation scenarios (

timely adaptation

and

delayed

adaptation

) show a decreased variability compared to the

no adapta-

tion

scenarios, because of negligible changes in standard deviation

along with an increase in mean yield. On the other hand, yields of
maize and soybean show increase in both mean and standard devia-
tion, resulting in an overall higher coef

ﬁ

cient of variation under

adaptation scenarios, compared with

no adaptation

. However, also for

these two crops neither the coef

ﬁ

cient of variation nor the standard

deviation of yields increase above those of the reference period.

The highest increase in yields is obtained when both sowing and

cultivar adaptation are combined (except for wheat). Yet, the latter
shows more important crop responses than the change in sowing
dates (Fig.

). This effect is explained by larger changes of maturity

dates (Fig.

), together with higher biomass accumulation rates (yield

gain per day of growth) towards the end of the growing season.
Moreover, we

ﬁ

nd that a

timely

adaptive response would be more

effective than a

delayed

response (Fig.

), which could be caused by

e.g., breeding programs lagging behind rapidly changing climatic

conditions (

30

). Particularly, a timely adaptive response matters the

most for maize, rice, and sorghum that show differences at the global
level between

timely

and

delayed

adaptation scenarios of +4%, +4%,

and +3%, respectively.

For each crop, a certain share of the cropland did not show an

advantage for yields from adaptation (yields under

timely adaptation

≤

yields under

no adaptation

). In these cases, as we only test one single

adaptation measure here (changed growing periods), and we cannot
distinguish if the negative effect is driven by inadequate assessment of
future growing seasons or by the inability of LPJmL to respond to the
adverse conditions that the growing season algorithm tries to avoid, we
assume that farmers would rather continue applying historical crop
management. Given that disadvantageous practices are unlikely to
spread, we assume that, in such cases, farmers would stay with the only
other option they have in our model setup that is to stay with the old
growing period. Therefore, we do not consider the yield losses that
would result from ineffective measures in those grid cells. The extent of
this area is about 8% (7

–

9% across GCMs) for maize, 7% (6

–

8%) for rice,

9% (4

–

13%) for sorghum, 17% (9

–

31%) for soybean, 42% (38

–

45%)

for wheat.

Bene

ﬁ

ts from growing-periods adaptation vary by region

and crop

The bene

ﬁ

t of adapting sowing date and maturity type to climate

change varies across regions and crops (Fig.

). Largest yield gains

(>30%) are most pronounced in high-latitude temperate regions,
where low temperatures are currently a limiting factor to crop growth.
Warming produces the dual effect of stimulating crop growth (e.g.,
temperatures closer to optimum for photosynthesis) and of extending
the time window of favorable conditions for crop growth. Adaptation
of growing periods to the longer favorable season is particularly sui-
table in these regions. Climate change might thus help expand crop
yields if sowing and cultivars are adapted simultaneously.

Adaptation has considerable effects (10

–

30%) throughout the

tropics, as well as in many other main breadbasket regions, including
the Corn Belt in the USA (maize and soybean), South-East Asia (rainfed
rice), the Sahel in Africa (sorghum), the North China Plains (irrigated
rice), where these crops indicate important potential bene

ﬁ

ts from

adaptation (Fig.

b

–

e).

The overall small effectiveness of adaptation on total crop yields

across mid-latitude regions (Fig.

a) is caused by largely ineffective

Fig. 2 | Bene

ﬁ

ts of sowing and cultivar adaptation on global crop yields

(2080

–

2099, RCP6.0).

Bene

ﬁ

ts on global yields are reported for all crops aggre-

gated and for each individual crop, along with the uncertainty under different
climate scenarios. The four adaptation scenarios indicate different levels of adap-
tation (adapt.):

timely adaptation

, sowing dates and cultivars adapted as the climate

is changing (2080

−

2099);

cultivar adaptation

, sowing

ﬁ

xed at the reference level,

only cultivars adapted as in timely adaptation;

sowing date adaptation

, only sowing

dates adapted as in timely adaptation, cultivars

ﬁ

xed at the reference level;

delayed

adaptation

both sowing dates and cultivar adapted but with 20-years delay, to

2060

−

2079 climate. The global yield of an individual crop is computed as the area-

weighted mean yield across all grid cells growing that crop. In grid cells where
adaptation of growing periods returned either no bene

ﬁ

t or maladaptation (yield

difference is equal or larger zero) yield losses were considered equal zero. Bars
represent the mean across GCMs (

n

= 4 GCMs), whiskers display the range across

GCMs, and gray symbols refer to individual GCMs.

Article

https://doi.org/10.1038/s41467-022-34411-5

Nature Communications

| (2022) 13:7079

4

adaptation of wheat, a crop that comprises a large part of total pro-
duction in these areas (e.g., central USA, Mediterranean, Northern
India, and Central Asia) (Supplementary Fig. 13). The pattern is con-
sistent across GCMs (Supplementary Figs. 17

–

20), and it could indicate

that either the crop-calendar adaptation rules cannot always

ﬁ

nd

better suitable growing periods or that the crop model LPJmL is
insensitive to the adverse weather conditions that the adaptation rule
tries to avoid. The latter is more plausible as, while the crop-calendar
model rules for sowing and maturity dates consider terminal heat
stress or over-winter crop damages and tries to select a shorter season
in order to avoid them, the LPJmL crop model misses adequate
representation of these important processes. Further efforts are
required to better assess winter-wheat adaptation in these regions.

Discussion

Current farming practices such as sowing date and cultivar choices are
the result of gradual adjustments and optimization of the cropping
systems to local conditions and incremental climatic change

29

–

32

. When

adaptation in management is not considered, climate change impacts
on crops might be overestimated, as previously shown by regional
studies. For instance, temperature increases have been found to
explain only half of the observed negative trend in heading date
(timing of in

ﬂ

orescence emergence) of winter wheat in Germany, while

recent faster-maturing cultivars account for the rest

33

. Similar adap-

tations have been reported for winter wheat in China

34

. In the USA,

maize yields have increased despite temperature increases, partially
owing to earlier sowing and longer-season cultivars

35

. Experimental

trials show that improved wheat cultivars with delayed anthesis and
increased grain-

ﬁ

lling rate

36

, or late-maturing cultivars with early

sowing

37

, can increase yields under the currently warming climate. At

the global scale, the scarcity of information on crop management and
its temporal evolution challenges the implementation of adaptation

scenarios in crop models

28

, so that previous assessments had to rely on

observed crop calendars

17

.

Here, we present an agro-climatic rule-based approach to project

farmers

’

decision-making rules on sowing dates and cultivar selection

into future scenarios and to assess the resulting effects on yields of

ﬁ

ve

major staple crops on current global cropland. Typically, regional and
local studies

ﬁ

nd similar mechanisms of growing-period adaptation. In

Europe, temperature-driven sowing of maize has been projected to
occur earlier under future climate

,

,

. Similarly, the tendency of

farmers to select later-maturing varieties with warming climate to
counteract the shortening of growing periods under increasing tem-
peratures and to take advantage of a longer growing season (unless
there is the risk of terminal water stress) has been shown for Europe, the
USA, and China

. Yet, some adaptation decisions remain exclu-

ded from our approach, which does not consider neither multiple-
cropping systems nor crop rotations. In these systems, the end of the

ﬁ

rst growing season often determines the beginning of the second,

adding complexity that is not captured by our climate-only driven
approach.

Climate change is expected to largely impact crop yields. The

most recent global-scale estimates based on crop-model ensemble
results (including LPJmL) show that, without accounting for adaptation
under high-warming scenarios (RCP8.5), maize global yields might
decline by

−

24%, wheat yields increase by +17%, whereas rice (+2%) and

soybean (

−

2%) yields are only marginally affected

(Supplementary

Table 3). Within that model ensemble, because of its strong response
to atmospheric CO

2

, LPJmL is on the more optimistic side, showing

yield changes of

−

7%, +18%, +24%, +15% for maize, rice, soybean, and

wheat, respectively, which are in line with those found in our study for
the same crops (

−

2%, +24%, +24%, +22%) (Supplementary Fig. 14). Our

ﬁ

nding that introducing adaptation might increase crop yields by 12%,

demonstrates the importance of considering farmers

’

decisions in

Fig. 3 | Geographic patterns of crop-yield bene

ﬁ

ts from sowing and cultivar

adaptation.

Yield bene

ﬁ

ts are computed as the relative difference (%) between the

timely adaptation

and

no adaptation

scenarios in the same climate period

(2080

–

2099, RCP6.0). No adaptation indicates a scenario in which crop sowing

dates and cultivars remain unchanged compared to the reference period
(1986

–

2005). Yield bene

ﬁ

ts are reported for (

a

) all

ﬁ

ve crops aggregated

(area-weighted mean of yields per grid cell) and for (

b

–

f

) each individual

crop (maize, rice, sorghum, soybean and wheat). Adaptation bene

ﬁ

ts are shown as

the mean across four GCMs. Maps for individual GCMs are reported in the Sup-
plementary Information. In grid cells where adaptation lead to either no bene

ﬁ

t or

maladaptation (negative yield change), adaptation is not considered.

Article

https://doi.org/10.1038/s41467-022-34411-5

Nature Communications

| (2022) 13:7079

5

global biophysical modeling under climate change. By conducting
global-scale assessments accounting for locally (and climate-scenario
speci

ﬁ

c) adapted growing seasons, which previously was only done for

individual case studies (location or region-speci

ﬁ

c; e.g., refs.

38

40

),

we are able to highlight the sizeable effect of growing-periods adap-
tation. In fact, adaptation not only avoids negative yield impacts on
individual crops (e.g., maize), but also further enhances yields of those
crops that bene

ﬁ

t from climate change (e.g., wheat), supporting the

necessary increase in food supply for a growing global population

41

Similar to the effect of CO

2

, adaptation should not be neglected in

future crop yield projections. Climate change scenarios can now be
supplemented with scenarios on crop growing season management in
model experiments such as AgMIP

,

and ISIMIP

42

. This constitutes an

important step in integrating dynamic management decision into
global models of agricultural systems, representing a signi

ﬁ

cant

improvement over assuming static historical management practices in
future climate change scenarios.

We here investigate farmers

’

decision-making that is only driven

by their experience of climate conditions. Some of this adaptation will
be implemented by the farmers directly (e.g., changing sowing dates

31

),

as they will be observing changing conditions and will respond to these
as good as they can. Bringing new cultivars into the

ﬁ

eld will, however,

require concerted action of farmers, breeders, and markets, as suitable
cultivars may not be directly transferable between regions,

,

Whilst in regions with functional markets, breeders work to provide
cultivars adapted to the farmers

’

needs

45

, in others, the predominance

of on-farm seed production systems can limit the introduction of new
cultivars

46

. As the impact of climate change in such more vulnerable

regions is largest, efforts are needed to exploit the full growing season
adaptation potential, along with other measures to improve crop
yields (e.g., improvement of soil fertility and water retention). These
will depend on functional, fair, and sustainable markets in less devel-
oped regions, including the infrastructure necessary for accessing
those markets

47

.

We identify two limitations that LPJmL has in common with sev-

eral other state-of-the-art global-scale crop models (e.g., ref.

48

). First,

the adaptation of sowing dates and cultivars will not only affect the
start and end of growing periods, but also the timing of intermediate
phenological phases, some of which are particularly susceptible to
high temperatures (e.g.,

ﬂ

owering). Although LPJmL simulates phe-

nology through a single phase from sowing to maturity, anthesis could
be identi

ﬁ

ed as the time when allocation to storage organs starts. For a

correct parameterization and evaluation of such additional phase and
to project changes in

ﬂ

owering and grain-

ﬁ

lling time under future

climate, global datasets on crop

ﬂ

owering dates would be needed.

Particularly, time series of

ﬂ

owering dates would also help the

improvement of the phenology models, including the effect of pho-
toperiod. The omission of photoperiod responses in the current and
previous global modeling assessment studies, might have led to the
oversensitivity of crop growing periods to increasing temperature (see
ref.

49

), which requires further investigation. Including intermediate

phenological phases would also allow to test the sensitivity of the
model to the use of phase-speci

ﬁ

c base temperatures

50

, instead of a

single base temperature as assumed by most global-scale models so
far

17

. Second, LPJmL is a global crop model that is restricted in its

representation of extreme-climate impacts on crop phenology and
growth, such as heat stress, frost damage and soil water excess.
Improving the representation of these impacts will be necessary to
increase con

ﬁ

dence in the estimated climate change impacts and

potentials of growing periods adaptation on global food yields. For
example, using air temperature instead of canopy temperature might
introduce some biases in the simulation of crop phenological progress
and other plant processes, particularly under water stress conditions.
On the other hand, the modeling of canopy temperature is still subject
to several challenges (e.g., less standardized measurements of canopy

temperature in the

ﬁ

eld to the data- and time-expensive calibration)

and has not been tested globally

51

, which makes it a longer-term

objective for model development.

Finally, future research on adaptation potentials will have to

advance on the integration of biophysical with socioeconomic aspects,
e.g., land-use change and technological progress, that can have even
larger effects than climate change and management

,

,

. It is plau-

sible to assume that farmers and farming systems will adapt growing
seasons to changing climate conditions. Here, we show how this can be
accounted for in global assessments of climate change impacts on
crop yields and that assessments not accounting for this could result
being overly pessimistic.

Methods

Rule-based mean sowing and maturity dates

Location- and climate-speci

ﬁ

c mean crop calendars are computed by

combining two rule-based approaches published by

and

to simulate

sowing and physiological maturity dates of grain crops, respectively.
The assumption is that farmers select growing seasons based on the
mean climatic characteristics of their speci

ﬁ

c location and on the

physiological limitations (base and optimum temperatures for repro-
ductive growth; sensitivity to terminal water stress) of the respective
crop species. Accordingly, they select sowing dates and cultivars with
phenologies that, on average, meet these adapted maturity dates.

The climate is classi

ﬁ

ed into (i) seasonality types, based on the

coef

ﬁ

cient of variation of monthly mean temperature and precipitation

and (ii) temperature levels, based on the temperature of the warmest
month as compared to the base and the optimum temperatures for the
crop reproductive growth. Optimal temperatures for sowing, optimal
temperature ranges for grain

ﬁ

lling, as well as indicators of soil moisture

conditions (based on precipitation/potential-evapotranspiration ratio
(P/PET)), are de

ﬁ

ned as global parameters for each crop (Supplemen-

tary Table 1) and used as thresholds to identify the best timing for
sowing and for the start or end of the crop grain-

ﬁ

lling phase. To cope

with

ﬂ

uctuations of daily values around these thresholds, mean daily

temperature, precipitation and potential evapotranspiration are
derived by linear interpolation between monthly values.

We distinguish between spring and winter crop types. Maize, rice,

sorghum, and soybean are simulated as spring crops only, for wheat
we simulate both types. For spring crops, farmers sow the crops at the
onset of the wet season (

ﬁ

rst day of the wettest 120 consecutive days),

in case of prevailing precipitation seasonality, or on the day of the year
when temperatures increase above crop-speci

ﬁ

c temperature

threshold

(Supplementary Table 1), in case of temperature-driven

seasonality.

For wheat, we distinguish three types:

winter wheat with vernali-

zation

is chosen if monthly temperatures fall below 0 °C, but winter is

neither too harsh (temperature of the coldest month is higher than

−

10 °C), nor too long (temperatures fall below the sowing temperature

threshold (12 °C) after 15th September (North hemisphere) or 31st
March (South hemisphere)

).

Winter wheat without vernalization

is

grown if winters are mild (the temperature of the coldest month is
higher than 0 °C) without dormancy. In this case, wheat is sown 75 days
before the coldest month of the year. This rule was arbitrarily chosen
based on observed wheat sowing dates in mild winter regions. If the
conditions for growing any of the winter-wheat types are not met
(winter too harsh and too long), then

spring wheat

(without vernali-

zation) is chosen. Note that the computed sowing dates do not differ
between rainfed and irrigated for any of the crops.

The mean maturity date is chosen so that the crop grain-

ﬁ

lling

phase, the most critical for yield formation, occurs under the least
stressful conditions possible in that location and climate as follows.
Under precipitation seasonality, grain

ﬁ

lling starts towards the end of

the rainy season, when a P/PET threshold is crossed. Under tempera-
ture seasonality, (a) grain

ﬁ

lling of spring crops starts in the warmest

Article

https://doi.org/10.1038/s41467-022-34411-5

Nature Communications

| (2022) 13:7079

6

month of the year (if summer temperatures are optimal), or right after
temperatures return within an optimal range; (b) grain

ﬁ

lling of winter

crops ends in the warmest month of the year (if summer temperatures
are optimal), or right before temperatures exceed the optimal range;
(c) eventually, maturity is advanced to escape terminal water stress.
Note that the grain-

ﬁ

lling phase has a static duration of 60 days for

maize and 40 days for all the other crops. This assumption is based on
empirical relationships between the total growth period and the post-

ﬂ

owering reproductive phase, showing that the partition between the

vegetative and reproductive phase of grain crops follows a saturation
curve that levels off after 90

–

100 days of total growth duration

54

.

Different crops are assumed to have only one crop cycle (sowing-to-
maturity) per year, therefore neither multi-cropping systems nor crop
rotations are accounted for in the decision-making rules. A detailed
description of the rules and parameterization can be found in
refs.

,

Simulated crop calendars re

ﬂ

ect current farmers

’

management

Simulated historical crop calendars, driven by the bias-corrected cli-
mate dataset WFDEI

, largely agree with observations

–

. We compare

results both at the country and grid-cell level because, although the
observed crop calendars used here are gridded datasets, their under-
lying sources are often reported per country. The country-level com-
parison highlights that the agreement is good for most countries,
importantly, including those with large cropland area. The area-
weighted Mean Absolute Error (MAE) is close or well below 30 days for
all considered crops (Fig.

4

). The simulated crop calendars compare

well with the observed data also at the grid-cell level. Large areas,
including major agricultural regions of importance for global yields,
show deviations within ±15 days for both sowing and maturity dates
(Supplementary Table 2 and Supplementary Figs. 21

–

24). However,

evaluating the accuracy below 30 days is limited by the time resolution
of the observations, which is either (i) monthly

and converted by us

into daily values, by taking the mid-day of the reported month, or (ii)
daily

12

,

, but resulting from averages over large time windows

(often > 1 month). Overall, the accuracy of the model is in line with the
original evaluations of this rule-base method

,

, as well as with other

studies simulating average growing periods across large regions

18

,

20

Simulation of daily crop phenology and yields with the LPJmL
crop model

We perform a modeling experiment across the global land grid at
0.5° × 0.5° resolution. We used the LPJmL5 crop model

24

25

to simulate

daily growth and phenological development of

ﬁ

ve crops, driven by

climate projections from four General Circulation Models (GCMs)
GFDL-ESM2M, HadGEM2-ES, IPSL-CM5A-LR and MIROC5 under the
Representative Concentration Pathways 6.0 (RCP6.0) as provided in
bias-adjusted form from the CMIP5 archive by the ISIMIP2b project

42

Irrigated and rainfed production systems are simulated separately on
their current harvested areas

, which is also used to compute total

crop yields at grid-cell and global scale, as the product of yield by crop-
speci

ﬁ

c area. A

ﬁ

rst 5000-year spin-up simulation is used to initialize all

model pools (e.g., soil carbon and nitrogen content). A second spin-up
simulation of 390 years is used to introduced effects of historical
human-driven land-use change on these pools. A change in cropping
area for the future scenarios is not considered in this study.

Phenological development is simulated based on the thermal-time

model, including the effect of vernalization. All crops are assumed to be
insensitive to photoperiod, due to a lack of parameters for multiple-
crops and global-scale simulations. Previous global studies

,

18

that

have focused on maize and wheat only, have found lower performances
in the growing-period simulations when using a photo-thermal
model, compared to a temperature-only driven approach and thus
recommend caution when using the photoperiodic response. State-of-
the-art global crop models

also typically do not consider sensitivity

to photoperiod or assume that the photoperiodic response of the cul-
tivars chosen in each location are perfectly tuned to the given
conditions.

Sowing dates are prescribed based on the external rule-based

algorithm. Crop cultivars are parametrized based on the phenological
units required to reach the corresponding maturity dates (TU

req

, °C

days). In line with

, TU

req

are derived consistently with the phenolo-

gical module of the crop model LPJmL for each grid cell, crop, and rule-
based computed growing period from the respective climate input.
They are calculated as the sum of daily mean air temperature incre-
ments above a crop-speci

ﬁ

c base temperature (TU) (Supplementary

Table 1) between rule-based sowing and maturity. In addition, winter-
wheat cultivars require effective vernalization days (VU

req

), that range

between 0 (mild winters) and 70 (cold winters), depending on the

Fig. 4 | Evaluation of simulated crop calendars.

Country-level comparison of

simulated and observed sowing (

A

) and maturity (

B

) dates (day of the year) for

ﬁ

ve

crops. Each circle refers to a country and a crop, the size of the circle is scaled
according to the cropland area per country. The area-weighted Mean Absolute
Error (MAE, days) is reported for each crop. Crop-calendar simulations are based on
WFDEI reanalysis climate forcing

for the period 1979

–

2012. The observed crop

calendar includes different sources

Article

https://doi.org/10.1038/s41467-022-34411-5

Nature Communications

| (2022) 13:7079

7

temperature of the 5 coldest months (Eq.

.

VU

req

=

70

5

× 1

T

m

3

10

3

ð

1

Þ

where

T

m

is the mean temperature of the month.

From the day of sowing, effective TU for phenological develop-

ment are accumulated daily, as the difference between the mean air
temperature on that day and the crop-speci

ﬁ

c base temperature for

phenological development (Eq.

(2

)). The vernalization effectiveness is

computed daily by a scaling factor (0

–

1), which is then multiplied to

the TU (Eq.

(2

)). For crops that are insensitive to vernalization, VU

d

is

set equal one.

TU

req

=

X

ndays

d

= 1

max 0,

T

d

T

base

×

X

d

0

VU

d

!

ð

2

Þ

where the scaling factor VU

d

is computed by a three-stage linear

response function with a range of optimal temperatures (Eq.

).

Temperature for effective vernalization range between

−

4 °C and

+17 °C, with an optimum range between 3 °C and 10 °C.

VU

d

=

T

d

4

ð

Þ

=

3

10

ð

Þ

if

4 <

T

d

< 3

1

if 3

≤

T

d

≤

10

17

T

d

=

17

10

ð

Þ

if 10 <

T

d

< 17

0

otherwise

8

>

<
>

>

:

9

>

=
>

>

;

ð

3

Þ

In this study, we have removed the effect of vernalization on slowing
down TU accumulation until 10% of the total vernalization requirements
is reached. In this way, the crop can accumulate both vernalization units
and heat units in fall, so that there is some leaf growth before winter (in
LPJmL, the LAI curve depends on accumulated heat units).

The LPJmL model simulates phenology as one single phase from

emergence to maturity. Although the

ﬂ

owering stage is not simulated

as an explicit break point, the fraction of above-ground biomass that is
allocated to the storage organs (fHI) depends on the phenological
progress (fTU

req

, fraction of TU

req

that have been ful

ﬁ

lled), with the

bulk of the storage organs start

ﬁ

lling up after 40% of TU

req

have been

reached (Eq.

(4

)). In line with this, the LAI curve reaches a plateau when

45% (wheat) or 50% (other crops) of the TU

req

are ful

ﬁ

lled, which could

be considered a proxy of the

ﬂ

owering stage.

fHI = 100 ×

fTU

req

100 × fTU

req

+ exp

11

:

1

10

:

0 × fTU

req

ð

4

Þ

Crop biomass growth is simulated by daily carbon accumulation and
allocation to different plant organs (roots, leaves, storage organs,
mobile reserves, and stem). The fraction of carbon allocated to each
pool is a function of the fraction of completed phenological pro-
gress. Water stress increases allocation to the roots and reduces
allocation to the leaves. The daily Net Primary Production (NPP) is
the result of the Gross Primary Production (daily gross photosynth-
esis) reduced by the respiration costs. Gross photosynthesis is
simulated as a function of absorbed photosynthetically active
radiation, CO

2

atmospheric mixing ratio, air temperature, day length,

and canopy conductance. Photosynthesis rate is given by the mini-
mum between light-limited and Rubisco-limited photosynthesis
rates, with distinguished pathways for C

3

and C

4

crops. Respiration

is tissue-speci

ﬁ

c and it is also driven by temperature. If accumulated

NPP is insuf

ﬁ

cient to satisfy all organ demands, allocation follows a

hierarchical order from roots, to leaves, to storage organs, and
consequently penalizing the harvest index. Crops are subject to yield
failure due to frost events (daily minimum temperature <

−

5 °C)

occurring during grain yield formation (50% <fPHU <95%).

All management inputs other than sowing dates and cultivars have

been provided according to global gridded datasets (0.5° × 0.5° spatial

resolution) of current and historical practices. Land-use datasets for
the period 1510

–

2015, of crop- and irrigation-speci

ﬁ

c area fraction and

of mineral nitrogen fertilizer annual application rates (kg N ha

−

1

year

−

1

),

were based on the LUH2v2

55

dataset. This has been dis-aggregated and

remapped using the MADRaT tool

56

, in order to match the crop clas-

si

ﬁ

cation to the Crop Functional Types simulated by LPJmL

24

. The crop

residue management input dataset (1850

–

2015) was obtained from

ref.

56

, which estimates crop residue removal rates from FAOSTAT

national statistics on residue-related management practices (e.g.,
burning on

ﬁ

eld, household fuel and livestock fodder). The crop-

speci

ﬁ

c spatial and temporal patterns of tillage and no-tillage systems

for the period 1973

–

2010 were derived from ref.

57

.

Crop-model skills are preserved when driven by simulated crop
calendars

We run LPJmL with both simulated and observed historical growing
periods, forced by the historical observation-based climate dataset
WFDEI

, and test model outputs against a standard benchmark for

global gridded crop models evaluation

58

. Forcing LPJmL with simu-

lated crop calendars preserves the model skills in simulating global
spatio-temporal patterns of crop yields, as indicated by the three
metrics to compare simulated and observed national yields (correla-
tion coef

ﬁ

cient, the root mean squared difference, and the variance)

(Supplementary Fig. 25).

Data processing and analysis

We assess the future adaptation potentials over the present-day
cropland

https://doi.org/10.5281/zenodo.7038163

. To quantify the performances of cropping systems, we

compute grid-cell level 20-years area-weighted average crop yields
(t ha

−

1

) in each simulation period and GCM. We quantify climate change

impacts with and without adaptation as the difference between future
and historical yields. Furthermore, we isolate the adaptation effect
from that of climate, by computing yield differences between con-
trasting management scenarios (e.g., with and without adaptation)
under the same future period (2080

–

2099) for each climate scenario

separately, where crops experience the same climate change scenario
but are grown in different growing periods. As crop yield interannual
variability metric, we use the coef

ﬁ

cient of variation computed as year-

to-year yield deviation normalized by the mean yield over the
same period (1986

–

2005 for the reference scenario and 2080

–

2099

for all other scenarios). We account for uncertainties from climate
scenarios by expressing the evaluation metrics at the aggregated level
(e.g., global yield losses) as the mean and range across the four GCMs.
In the main text, spatial patterns of growing period and yield changes
are shown as the mean change per grid cell across four GCMs. The
results for the individual GCMs are reported in the Supplementary
Figures.

We compute the area-weighted mean absolute error (MAE) at the

grid-cell and country levels as (Eq.

(5

)):

MAE =

P

n
i

= 1

∣

S

i

O

i

∣

A

i

P

n
i

= 1

A

i

ð

5

Þ

where

n

is the number of grid cells (countries),

A

i

is the crop-

speci

ﬁ

c area of grid cell (country)

i

,

S

i

, and

O

i

are the simulated and

observed average dates (sowing or maturity) of grid cell (country)

i

in

Julian days.

For data processing, we used R

59

and R-packages for handling

netcdf4

60

, performing computation

61

,

62

, and plotting results

63

. The

country borders displayed in all

ﬁ

gures that include maps were based

on the Natural Earth dataset, which is available at naturalearthdata.com.

Reporting summary

Further information on research design is available in the Nature
Portfolio Reporting Summary linked to this article.

Article

https://doi.org/10.1038/s41467-022-34411-5

Nature Communications

| (2022) 13:7079

8

Data availability

The raw outputs of the LPJmL crop model generated and used in this
study have been deposited in the Zenodo database under accession
code

(

https://zenodo.org/

record/7038163#.YyRHObRBwaE

). Climate reanalysis data (WFDEI)

and model projection of the four GCMs (HadGEM2-ES, GFDL-ESM2M,
IPSL-CM5A-LR, MIROC5) are available at

https://www.isimip.org/

. All

other input data necessary to run the LPJmL model were obtained from
external sources, as described in the Methods. They are archived in the
internal PIK publication meta-database with the entry ID number 3369
and can be provided by the corresponding author to individual users
who want to conduct the same analysis. Source data are provided with
this paper.

Code availability

The source code of the crop-calendar model, the source code of the
LPJmL model, as well as the scripts work

ﬂ

ow generated and used for

the analysis have been deposited in the Zenodo database under
accession code

https://doi.org/10.5281/zenodo.7038163

. An updated

version of the crop-calendar model in form of an R package is made
publicly available in the GitHub repository at

https://github.com/

AgMIP-GGCMI/cropCalendars

https://doi.org/10.2499/0896295354

References

1.

Menzel, A. et al. European phenological response to climate change
matches the warming pattern.

Glob. Change Biol.

12

,

1969

–

1976 (2006).

2.

Estrella, N., Sparks, T. H. & Menzel, A. Effects of temperature, phase
type and timing, location, and human density on plant phenological
responses in Europe.

Clim. Res.

39

, 235

–

248 (2009).

3.

Asseng, S. et al. Rising temperatures reduce global wheat pro-
duction.

Nat. Clim. Change

5

, 143

–

147 (2014).

4.

Zhao, C. et al. Temperature increase reduces global yields of major
crops in four independent estimates.

Proc. Natl Acad. Sci. USA

114

,

9326

–

9331 (2017).

5.

Craufurd, P. Q. & Wheeler, T. R. Climate change and the

ﬂ

owering

time of annual crops.

J. Exp. Bot.

60

, 2529

–

2539 (2009).

6.

Olesen, J. E. et al. Changes in time of sowing,

ﬂ

owering and

maturity of cereals in Europe under climate change.

Food Addit.

Contam.: Part A

29

, 1527

–

1542 (2012).

7.

Tao, F., Zhang, S., Zhang, Z. & Rötter, R. P. Maize growing duration
was prolonged across China in the past three decades under the
combined effects of temperature, agronomic management, and
cultivar shift.

Glob. Change Biol.

20

, 3686

–

3699 (2014).

8.

Roberts, E. H., Summer

ﬁ

eld, R. J. & Ellis, R. H. Reproductive

development and crop adaptation.

J. Biol. Educ.

31

, 97

–

105 (1997).

9.

Hill, C. B. & Li, C. Genetic architecture of

ﬂ

owering phenology in

cereals and opportunities for crop improvement.

Front. Plant Sci.

7

,

1

–

23 (2016).

10. Sacks, W. J. & Kucharik, C. J. Crop management and phenology

trends in the US Corn Belt: impacts on yields, evapotranspiration
and energy balance.

Agric. For. Meteorol.

151

, 882

–

894 (2011).

11.

Portmann, F. T., Siebert, S. & Döll, P. MIRCA2000-Global monthly
irrigated and rainfed crop areas around the year 2000: a new high-
resolution data set for agricultural and hydrological modeling.

Glob. Biogeochem. Cycles

24

, GB1011 (2010).

12. Sacks, W. J., Deryng, D., Foley, J. A. & Ramankutty, N. Crop planting

dates: an analysis of global patterns.

Glob. Ecol. Biogeogr.

19

,

607

–

620 (2010).

13. Jägermeyr, J. et al. Climate impacts on global agriculture emerges

earlier in new generation of climate and crop models.

Nat. Food

2

,

873

–

885 (2021).

14. Elliott, J. et al. Characterizing agricultural impacts of recent large-

scale US droughts and changing technology and management.

Agric. Syst.

159

, 275

–

281 (2018).

15. Jägermeyr, J. & Frieler, K. Spatial variations in crop growing seasons

pivotal to reproduce global

ﬂ

uctuations in maize and wheat yields.

Sci. Adv.

4

, eaat4517 (2018).

16. Rosenzweig, C. et al. Assessing agricultural risks of climate change

in the 21st century in a global gridded crop model intercomparison.

Proc. Natl Acad. Sci. USA

111

, 3268

–

3273 (2014).

17. Minoli, S. et al. Global response patterns of major rainfed crops to

adaptation by maintaining current growing periods and irrigation.

Earth

’

s Future

7

, 1464

–

1480 (2019).

18. Bussel, L. G. J., van, Stehfest, E., Siebert, S., Müller, C. & Ewert, F.

Simulation of the phenological development of wheat and maize at
the global scale.

Glob. Ecol. Biogeogr.

24

, 1018

–

1029 (2015).

19. Waha, K., Bussel, L. G. J., van, Müller, C. & Bondeau, A. Climate-

driven simulation of global crop sowing dates.

Glob. Ecol. Biogeogr.

21

, 247

–

259 (2012).

20. Mathison, C., Deva, C., Falloon, P. & Challinor, A. J. Estimating

sowing and harvest dates based on the Asian summer monsoon.

Earth Syst. Dyn.

9

, 563

–

592 (2018).

21. Iizumi, T., Kim, W. & Nishimori, M. Modeling the global sowing and

harvesting windows of major crops around the year 2000.

J. Adv.

Modeling Earth Syst.

11

, 99

–

112 (2019).

22. Minoli, S., Egli, D. B., Rolinski, S. & Müller, C. Modelling cropping

periods of grain crops at the global scale.

Glob. Planet. Change

174

,

35

–

46 (2019).

23. Weedon, G. P. et al. The WFDEI meteorological forcing data set:

WATCH Forcing Data methodology applied to ERA-Interim reana-
lysis data.

Water Resour. Res.

50

, 7505

–

7514 (2014).

24. Bloh, Wvon et al. Implementing the nitrogen cycle into the dynamic

global vegetation, hydrology, and crop growth model LPJmL (ver-
sion 5.0).

Geoscienti

ﬁ

c Model Dev.

11

, 2789

–

2812 (2018).

25. Lutz, F. et al. Simulating the effect of tillage practices with the

global ecosystem model LPJmL (version 5.0-tillage).

Geoscienti

ﬁ

c

Model Dev.

12

, 2419

–

2440 (2019).

26. Warszawski, L. et al. The inter-sectoral impact model inter-

comparison project (ISI

–

MIP): project framework.

Proc. Natl Acad.

Sci. USA

111

, 3228

–

3232 (2014).

27. Kimball, B. A. Crop responses to elevated CO

2

and interactions

with H

2

O, N, and temperature.

Curr. Opin. Plant Biol.

31

,

36

–

43 (2016).

28. Challinor, A. J. et al. A meta-analysis of crop yield under climate

change and adaptation.

Nat. Clim. Change

4

, 287

–

291 (2014).

29. Kahiluoto, H. et al. Decline in climate resilience of European wheat.

Proc. Natl Acad. Sci. USA

116

, 123

–

128 (2018).

30. Parent, B. et al. Maize yields over Europe may increase in spite of

climate change, with an appropriate use of the genetic variability
of

ﬂ

owering time.

Proc. Natl Acad. Sci. USA

115

, 10642

–

10647

(2018).

31. Karapinar, B. & Özertan, G. Yield implications of date and cultivar

adaptation to wheat phenological shifts: a survey of farmers in
Turkey.

Clim. Change

158

, 453

–

472 (2020).

32. Fatima, Z. et al. The

ﬁ

ngerprints of climate warming on cereal crops

phenology and adaptation options.

Sci. Rep.

10

, 1

–

21 (2020).

33. Rezaei, E. E., Siebert, S., Hüging, H. & Ewert, F. Climate change

effect on wheat phenology depends on cultivar change.

Sci. Rep.

8

,

1

–

10 (2018).

34. He, L. et al. Impacts of recent climate warming, cultivar changes,

and crop management on winter wheat phenology across the Loess
Plateau of China.

Agric. For. Meteorol.

200

, 135

–

143 (2015).

35. Butler, E. E., Mueller, N. D. & Huybers, P. Peculiarly pleasant weather

for US maize.

Proc. Natl Acad. Sci. USA

115

, 11935

–

11940 (2018).

36. Asseng, S. et al. Climate change impact and adaptation for wheat

protein.

Glob. Change Biol.

25

, 155

–

173 (2018).

37. Hunt, J. R. et al. Early sowing systems can boost Australian wheat

yields despite recent climate change.

Nat. Clim. Change

9

,

244

–

247 (2019).

Article

https://doi.org/10.1038/s41467-022-34411-5

Nature Communications

| (2022) 13:7079

9

38. Zimmermann, A. et al. Climate change impacts on crop yields, land

use and environment in response to crop sowing dates and thermal
time requirements.

Agric. Syst.

157

, 81

–

92 (2017).

39. Dobor, L. et al. Crop planting date matters: estimation methods and

effect on future yields.

Agric. For. Meteorol.

223

, 103

–

115 (2016).

40. Zhang, Y. et al. Adaptation in U.S. Corn Belt increases resistance to

soil carbon loss with climate change.

Sci. Rep.

10

, 1

–

7 (2020).

41. Bodirsky, B. et al. Food demand projections for the 21st century.

PLoS ONE

10

, 1

–

27 (2015).

42. Frieler, K. et al. Assessing the impacts of 1.5 °C global warming

—

simulation protocol of the Inter-Sectoral Impact Model Inter-
comparison Project (ISIMIP2b).

Geoscienti

ﬁ

c Model Dev.

10

,

4321

–

4345 (2017).

43. Pugh, T. A. M. et al. Climate analogues suggest limited potential for

intensi

ﬁ

cation of production on current croplands under climate

change.

Nat. Commun.

7

, 1

–

8 (2016).

44. Zabel, F. et al. Large potential for crop production adaptation

depends on available future varieties.

Glob. Change Biol.

27

,

3870

–

3882 (2021).

45. Voss-Fels, K. P. et al. Breeding improves wheat productivity under

contrasting agrochemical input levels.

Nat. Plants

5

, 706

–

714 (2019).

46. Singh, R. P., Prasad, P. V. V. & Reddy, K. R. Impacts of changing

climate and climate variability on seed production and seed industry.
In

Advances in Agronomy

(ed. Sparks, D. L.) 49

–

110 (Elsevier, 2013).

47. IFPRI. Climate change: impact on agriculture and costs of adapta-

tion.

(2009).

48. Li, Y., Guan, K., Schnitkey, G. D., DeLucia, E. & Peng, B. Excessive

rainfall leads to maize yield loss of a comparable magnitude to
extreme drought in the United States.

Glob. Change Biol.

25

,

2325

–

2337 (2019).

49. Zhu, P., Zhuang, Q., Archontoulis, S. V., Bernacchi, C. & Müller, C.

Dissecting the nonlinear response of maize yield to high tempera-
ture stress with model-data integration.

Glob. Change Biol.

25

,

2470

–

2484 (2019).

50. Sánchez, B., Rasmussen, A. & Porter, J. R. Temperatures and the

growth and development of maize and rice: a review.

Glob. Change

Biol.

20

, 408

–

417 (2014).

51. Webber, H. et al. Simulating canopy temperature for modelling

heat stress in cereals.

Environ. Model. Softw.

77

, 143

–

155 (2016).

52. Leng, G. & Huang, M. Crop yield response to climate change varies

with crop spatial distribution pattern.

Sci. Rep.

7

, 1

–

10 (2017).

53. King, M. et al. Northward shift of the agricultural climate zone under

21st-century global climate change.

Sci. Rep.

8

, 1

–

10 (2018).

54. Egli, D. B. Time and the productivity of agronomic crops and

cropping systems.

Agron. J.

103

, 743

–

750 (2011).

55. Hurtt, G. C. et al. Harmonization of global land-use change and

management for the period 850

–

2100 (LUH2) for CMIP6.

5425

–

5464 (2020).

56. Dietrich, J. P. et al.

mrland: MadRaT Land Data Package

. R package

version 0.40.2.

https://github.com/pik-piam/mrland

(2022).

57. Porwollik, V., Rolinski, S., Heinke, J. & Müller, C. Generating a rule-

based global gridded tillage dataset.

Earth Syst. Sci. Data

11

,

823

–

843 (2019).

58. Müller, C. et al. Global gridded crop model evaluation: bench-

marking, skills, de

ﬁ

ciencies and implications.

Geoscienti

ﬁ

c Model

Dev.

10

, 1403

–

1422 (2017).

59. R Core Team.

R: A Language and Environment for Statistical Com-

puting

(R Foundation for Statistical Computing, 2021).

60. Pierce, D.

ncdf4: Interface to Unidata netCDF (Version 4 or Earlier)

Format Data Files

.

https://cran.r-project.org/web/packages/ncdf4/

ncdf4.pdf

(2015).

61. Dowle, M. & Srinivasan, A.

data.table: Extension of

‘

data.frame

‘

.

https://cran.r-project.org/web/packages/data.table/data.table.
pdf

(2017).

62. Wickham, H. The split-apply-combine strategy for data analysis.

J.

Stat. Softw.

40

, 1

–

29 (2011).

63. Wickham, H.

ggplot2: Elegant Graphics for Data Analysis

(Springer-

Verlag New York, 2009).

Acknowledgements

We acknowledge

ﬁ

nancial support from the MAPPY project (01LS1903A,

S.M. and C.M.) funded through the German Federal Ministry of Education
and Research (BMBF Germany). J.J. received funding from the Open
Philanthropy Project and from the NASA GISS Climate Impacts Group. We
acknowledge

ﬁ

nancial support from the International Wheat Yield Part-

nership (IWYP, grant IWYP115, S.A.). For A.U., this research was conducted
under the USAID and Bill and Melinda Gates Foundation supported Cereal
Systems Initiative for South Asia (CSISA;

http://csisa.org/

) and with the

support of the CGIAR Regional Integrated Initiative Transforming Agri-
food Systems in South Asia, or TAFSSA (

https://www.cgiar.org/initiative/

20-transforming-agrifood-systems-insouth-asia-tafssa/

).

Author contributions

Conceptualization: S.M., C.M., and J.J. Methodology: S.M., C.M., J.J., and
A.U. Investigation: S.M. and J.J. Visualization: S.M. Writing

—

original

draft: S.M. Writing

—

review & editing: S.M., J.J., S.A., A.U., and C.M.

Funding

Open Access funding enabled and organized by Projekt DEAL.

Competing interests

The authors declare no competing interests.

Additional information

Supplementary information

The online version contains

supplementary material available at

https://doi.org/10.1038/s41467-022-34411-5