Content uploaded by Pramit Verma
Author content
All content in this area was uploaded by Pramit Verma on Sep 10, 2022
Content may be subject to copyright.
Title: Effect of rainfall variability on tree phenology in moist tropical deciduous forests
Authors:
Pramit Verma1,, Priyanshi Tiwari2, Rishikesh Singh3, Akhilesh Singh Raghubanshi1*
1. Pramit Verma
Integrative Ecology Laboratory (IEL), Institute of Environment and Sustainable development (IESD),
Banaras Hindu University (BHU), Varanasi, India
Email: coda.zeppelin@gmail.com ; pverma@v.umk.pl
2. Priyanshi Tiwari
Department of Botany, Institute of Science, Banaras Hindu University, Varanasi, India
Email: tpriyatiwari9@gmail.com
3. Rishikesh Singh
Department of Botany, Panjab University (PU), Chandigarh 160014, India
E-mail: rishikesh.iesd@gmail.com; rishikesh.singh@bhu.ac.in
* Corresponding Author
1,*Akhilesh Singh Raghubanshi
Integrative Ecology Laboratory (IEL), Institute of Environment and Sustainable development (IESD),
Banaras Hindu University (BHU), Varanasi, India
Email: asr.iesd.bhu@gmail.com
Acknowledgements:
We acknowledge the funding provided by the University Grants Commission, New Delhi, India as a
fellowship for the first author.
Author Contributions:
ASR conceptualised the idea for conducting this work. PV and PT analysed the data and PV, PT and
RS wrote the manuscript. The final revision was done by PV and RS.
Data Availability:
All the data, including raw files, will be made available on request from the first author
1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
Title: Effect of rainfall variability on tree phenology in moist tropical deciduous forests
Abstract
Plants in their life cycle go through a series of life processes. These phenological changes are
influenced by different climatic conditions. Abiotic factors like temperature, precipitation,
and photoperiodism affect the onset and offset of particular phenophase in the plant periodic
cycle. In this study, we tested the influence of precipitation on the forest phenology of
Dudhwa National Park (DNP), Uttar Pradesh and Simlipal National Park (SNP), Odisha,
India. DNP and SNP receive an annual average rainfall of 1093.5 mm and 1500 mm,
respectively, of which most rainfall (~90%) occurs during June–September. Normalized
Difference Vegetation Index (NDVI) was measured for two years of 2015 and 2018, with
2015 being a drought year and 2018 being a normal rainfall year. NDVI was analyzed at
different temporal scales of months, season, and years using the T-test (Welch’s two-tailed)
and General Linear Mixed Model (GLMM). Effect of drought (2015) and normal (2018)
rainfall year was not evident at both the sites, whereas season, year*season interaction,
season*rainfall interaction, and year*season*rainfall interaction were found significant at
DNP (P<0.05, ICC=0.68, marginal R2=0.81; conditional R2=0.94). At SNP, rainfall, year,
season, and their interaction were non-significant whereas several months showed a
significant effect on the NDVI values for both sites. Winter and monsoon season in DNP and
post-monsoon season in SNP showed a significant effect on the NDVI patterns. Thus, the
effect of precipitation stress in the deciduous forests was evident at small intervals of
observation. Tree phenology compensated for differences when observed from a higher
temporal scale of a year. There existed a mechanism in trees to tide over adverse conditions
and maintain the phenology over longer intervals of time. The resilience and vulnerability of
such forest ecosystems against abiotic factors and extreme events would be instrumental in
2
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
climate change adaptation strategies. Tree phenology can be used as an indicator of forest
health and resilience.
Keywords: drought; general linear mixed model (GLMM); Normalized Difference
Vegetation Index (NDVI); rainfall; phenology indicator; remote sensing
1. Introduction
Forests are an integral part of the terrestrial ecosystem accounting for the removal of about 3
Pg C year–1 of anthropogenic carbon (Canadell and Raupach 2008). Forestry is one of the
most important sectors which determines the world’s biodiversity and nutrient cycles to a
large extent (Kok et al. 2018). However, evaluation of the Aichi Convention on Biodiversity
targets for 2020 reported that biodiversity has declined in recent years (Ibid.). Natural
vegetation is dependent on climate to a large extent and any changes in the global carbon
balance result in changes in the forest ecology (Emanuel et al. 2008). These changes become
more evident in the case of extreme events, which are projected to increase simultaneously in
the future (Sharma and Majumdar 2017). Phenology is one such indicator for forest
ecosystems that can be used to assess the impact of climatic factors on plant responses
(Chmielewski & Rötzer 2001). Phenology describes the changes in plant biology due to
temporal variability (seasons) and the impact of biotic and abiotic factors. Phenological
changes within plants and animals can influence human activities, such as public health,
fisheries, hunting, forestry, nature management, gardening, tourism and recreation,
transportation, and water management (Vliet et al. 2003). Phenology formed a large part of
the impact of climate change on forests according to an IPCC report (Rosenzweig et al.,
2008).
Tropical forests have been shown to have a significant effect on the carbon cycle,
biomass accumulation, ecosystem services, and climate regulation, including precipitation
3
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
distribution (Singh et al. 2020). However, the recent changes in climate and the occurrence of
extreme events, like droughts and floods, have harmed the structure and functioning and
forest systems (Chen et al. 2014). Extensive studies have recognized the cumulative effects of
climatic factors (e.g., temperature and precipitation) on phenological events (Liu et al., 2016).
This is exacerbated by the rampant deforestation activities like logging, clearing of forests for
infrastructure projects, and removal of the forest-dependent tribal communities. This
changing relationship between the forests and climate has impaired the structure and
functioning of tropical ecosystems. However, it has been reported that tropical regions,
especially the deciduous forests, have a quick adaptability potential to changing
environmental conditions (Singh et al. 2020), whereby, they display periodic functions like
flowering, leaf emergence, and leaf fall, fruiting, and other vegetative functions. The leaf fall
phenomenon takes place throughout the year according to local meteorological conditions;
however, it is much more pronounced in the dry season (Singh et al. 2020). Thus, changes in
leaf fall pattern (deciduousness) or phenology can be used as an important metric for the tree
response to climate change as it is critically related to precipitation and temperature (Deka et
al. 2019).
A large body of literature has dealt with the impact of climatic variables on the tree
phenology, and its response in temperate areas. It was found that in temperate forests,
warming resulted in shifting in seasonal dynamics e.g., early spring and late autumn (Menzel
et al., 2006), whereas cooling delayed the timing of spring and stretching in autumn (Piao et
al., 2019). The temperature has been reported to activate plant stress response and cause
endo-dormancy (low-temperature) or ecto-dormancy (high-temperature) (Delpierre et al.,
2016; Hanninen, 2016).
Water and nutrient availability were less critical in temperate and boreal forests as
compared to temperature and photoperiod (Jaworski & Hilzczanski, 2013). Decreased water
4
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
availability influenced the nitrogen availability to the plant growth in arid and semiarid
regions, and spring advancement (Estiarte & Penuelas, 2015). Tropical forest phenology,
which has more complex biodiversity than temperate and boreal regions, is majorly governed
by precipitation patterns when compared to temperature variability (Pau et al., 2011). The
extent of stress experienced by tropical forests varies greatly during dry months as drought
condition is overcome in many species of tropical forest by using stored trunk water, the
presence of deeper roots and soil moisture (Borchert, 1994). However, more research is
needed to know the mechanism by which water and nutrients interact with other abiotic
factors such as temperature and photoperiod in determining plant phenophases.
It is extremely difficult to deconvolute the impacts of the different covarying drivers
of phenology and to gain insight into the mechanisms underlying observed patterns and
changes therein. Advancements in remote sensing technology have made possible the study
of landscape-scale plant phenology by detecting the timing of phenological events in the
temporal profile of greenness-related vegetation indices, such as the normalized difference
vegetation index (NDVI) and the enhanced vegetation index (EVI) (Piao et al., 2019). NDVI
is widely used by researchers to calculate the vegetation structure and function (Prasad et al.,
2007). NDVI is more sensitive to phenological changes (Roy & Ravan, 1996). Recent
investigations into the phenology of Xylocarpos (Robertson et al. 2020), dry tropical
deciduous forests (Huechacona-Ruiz et al. 2020), threatened species of family Cactaceae
(Toledo et al. 2021), and canopy phenology (Valderrama-Landeros et al. 2021), highlight the
increasing importance phenology has acquired and the application of remotely sensed data
and image processing in this field. In this respect, the NDVI as an indicator of leaf health has
become very important and a strong relationship of NDVI has been reported with rainfall
(Singh et al. 2020) and temperature (Deka et al. 2019). Reich (1995) reported that in tropical
ecosystems, rainfall plays a dominant role, whereas, in temperate regions, temperature shows
5
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
a marked influence on the phenology. In a humid subtropical region, Deka et al. (2019),
reported that the temperature NDVI correlation was stronger than the rainfall-NDVI
correlation, and temperature may act as a critical factor especially in the climate change
scenario.
Despite the large body of literature, exploring the phenotypical responses to changes
in the environment and utilizing the NDVI as an indicator of forest resilience, has not been
undertaken yet. Moreover, in tropical forests, such studies have yet to exploit the advantages
of remote sensing analysis. With this background, we hypothesized that the drought, as an
environmental stress factor, would have an influence on the tree leaf phenology in a tropical
forest and the tree response would resist such a change over a certain temporal scale. Our
objectives for this work were to examine whether there was any influence of drought on the
tree leaf phenology in the tropical deciduous forest of Dudhwa National Park (DNP) and
Simlipal National Park (SNP) at the temporal scale of months, seasons, and year. The two
sites were chosen to get comprehensive knowledge about the phenological responses in two
similar forest types at two geomorphological locations. The results indicated that forest
ecosystems resisted changes in the long yearly phenological patterns but the impact of
rainfall, season and their interaction worked at several levels to influence the monthly and
seasonal phenological phases in several instances.
2. Methodology
2.1. Study Area
The study was conducted in moist deciduous forests to observe the impact of rainfall
variability on tree phenology. Two sites in India were selected for this study, Dudhwa
National Park (DNP) and Simlipal National Park (SNP). Both the sites represented large core
areas of protected forests and also with the least direct anthropological influence. Dudhwa
6
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
National Park is present in the Lakhimpur Kheri district in the sub-Himalayan belt on the
Indo-Nepal border (Fig. 1). Lakhimpur Kheri is primarily drained by the Ghaghra and
Gomati rivers. The normal rainfall is 1093.5 mm, of which about 86% is received during the
monsoon period (Ground Water Scenario Lakhimpur District, 2013). Generally, the depth to
the water level in the district varied from 2.95 to 9.66 mbgl (meters below ground level) and
1.48 to 7.26 mbgl, during the pre-monsoon and post-monsoon seasons, respectively. Simlipal
National Park is present in the Mayurbhanj district of Odisha (Fig. 1). Mayurbhanj lies
between the 21ᵒ17’ and 22ᵒ34’N and 85ᵒ40’ and 87ᵒ10’E. The district is mainly drained by
the Budhabalanga, Kharkai, Jamira, and several other tributaries that originate from the
Similipal hills. The district has a tropical to the sub-subtropical type of climate with well-
distributed rainfall. The average rainfall in the district is 1500 mm. The pre-monsoon and
post-monsoon depth to water level ranged from 3.54 to 14.50 mbgl and 1.39 to 8.20 mbgl,
respectively (Ground Water Resources, Mayurbhanj 2013).
Dudhwa National Park (DNP) had a protected forest cover of about 490 km2 and a
buffer area of 190 km2. SNP covers a vast area of about 2750 km2, with 303 km2 of the
protected forest. It holds a critical place in providing ecosystem services and a habitat for the
endemic flora and fauna. Further, since the objective was to study the impact of drought on
forest phenology, we selected a drought year (2015) and a normal year (2018) and collected
as much NDVI data as possible for each month. These years were chosen because of two
important reasons. First, the Landsat 8 program started distributing satellite data in 2013 and
is considered superior and more accurate compared to other Landsat series (Verma et al.
2020). Secondly, deficient rainfall was observed at both places in the 2015 monsoon. The
selection of the year was made based on India Meteorological Department seasonal data
weather reports published in Mausam for the years 2016 and 2019. Rainfall and departure
7
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
Three well-defined seasons were experienced in DNP and SNP, summer (March/April
to June), rainy (July to September), and winter (December to February). The average day and
night temperatures during July and August are 29°C and 22°C. The annual rainfall ranged
from 5 to 15 cm in winter, 5 to 57 cm in summer, and 135 to 225 cm in rainy seasons
(Naithani et al., 2018). In this study, we considered December to February as the winter
season, March to May as pre-monsoon (summer season), June to September as monsoon, and
October to November as the post-monsoon season (Singh and Mal 2014).
The dominant tree species of DNP constitute Sal (Shorea robusta), Asna (Terminalia
tomentosa), Shisham (Dalbergia sissoo), Jamun (Syzygium cumini), Gular (Ficus glomerata),
Sehore (Streblus asper), Neora (Baringtonia acutangula), and Putranjiva roxburghii, Bahera
(Terminalia bellerica). Sal also dominates the forest diversity in SNP covering about 55.5%
of the forested area (Murthy et al. 2007).
2.2. Satellite image processing
Satellite images were used to extract NDVI data of DNP and SNP. Landsat 8 satellite images
were used for our study. Only 30-m spatial resolution bands were included for analysis i.e.,
bands 1–7 and band 9 of Landsat 8. Landsat 8 ensures continuous monitoring and availability
of Landsat data utilizing two-sensor payloads, the Operational Land Imager (OLI) and the
Thermal InfraRed Sensor (TIRS). Satellite images of DNP and SNP were downloaded from
the USGS (United States Geological Survey) Library for the available months for years 2015
and 2018. The actual dates were selected based on the availability of cloud-free images.
Downloaded files were extracted and ENVI software was used for further processing of
images.
9
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
Fig. 2. Flowchart of the methodology applied (for Dudhwa National Park)
The detailed methodology is given in Fig. 2 with DNP as an example. A similar
methodology was followed for SNP. NDVI of the subset file was calculated using the
Spectral Delineation tool of software. 15 sample locations were identified randomly from the
core zone of the national park to evaluate the NDVI changes. These sample locations were
considered representative of the whole vegetated area. Each of the locations consisted of 9
pixels covering a total area of 8100 m2. The reason for selecting only these locations was to
10
221
222
223
224
225
226
227
228
reduce the data processing time as well as to avoid data loss due to cloud cover during
monsoon season. The obtained NDVI file was a subset using the ROI files of the 15 sites.
Each sample site was used to derive the NDVI value, and the result was saved as a text file.
This was followed by a statistical analysis of the NDVI data.
2.3. Statistical Modelling
We used paired Welch’s two-tailed T-test for monthly, seasonal and annual datasets for the
years 2015 and 2018 to test our hypothesis. The paired t-test was applied for several
combinations of drought and normal years (2015 and 2018), seasons (winter, pre-monsoon,
monsoon, and post-monsoon), and months. In the case of DNP, NDVI could be extracted for
the February, March, April, May, July, August, September, October, November, and
December months, whereas, in SNP, NDVI was extracted for January, February, March,
April (2015 only), May (2018 only), July, August, September, October, November, and
December months. To further establish the impact of drought and normal year, season, and
month, a hierarchical General Linear Mixed Model (GLMM) was applied with the month as a
random factor and year and season as fixed effects. The GLMM was applied using the
“lme4” package in R (ver. 4.2.0) (R Core team 2013; Bates et al. 2014). To establish the
significance of the terms in the GLMM, Wald’s (type 2) Chi-square test was performed using
the “car” package in R (Chang et al. 2019; Bhadouria et al. 2019). Marginal and conditional
R2 was also noted to assess the robustness of the model, along with the Interclass Correlation
Coefficient (ICC) to support the GLMM. A high ICC indicates the necessity and validity of
the GLMM. Marginal R2 accounts for random effects and should not be close to zero,
whereas conditional R2 accounts for both random and fixed effects (Nakagawa & Schielzeth
2013; Sotirchos et al. 2019).
11
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
3. Results
The NDVI results for SNP and DNP are shown in Fig. 3 for March 2018. The pooled NDVI
for DNP had a positive correlation with rainfall (0.45, P<0.05) whereas no such relation was
found for the SNP. Rainfall was not considered in the GLMM for SNP.
Fig. 3. Vegetation delineation results show NDVI for the (a) Simlipal National Park and (b)
Dudhwa National Park. The red dots denote the actual NDVI sampling points.
12
252
253
254
255
257
258
259
260
261
262
263
264
Fig. 4. Monthly variation of NDVI and rainfall for 2015 (drought) and 2018 (normal) in (a) Dudhwa
National Park and (b) Simlipal National Park
3.1 Monthly variation of NDVI for normal and drought years
In DNP, the months of February, April, May, July, August, September, October, and
December showed significant differences (P<0.05; Table S1), however, the t-test null
hypothesis was rejected for November and March. For February, May, July, October, and
December, mean NDVI were higher in the drought year than the normal year (Fig. 4).
However, for April, August and September, the NDVI was higher in the normal year.
In Simlipal, January, March, September, October, and November showed a significant
difference between the normal and drought years (P<0.05; Table S1). In 2018, the month of
January (0.88), October (0.92), and November (0.88) had a significantly higher NDVI than
their counterparts in 2015, a drought year, whereas, in the same year, the NDVI was higher in
March (0.86) and September (0.89; Fig. 4).
13
b
a
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
3.2 Seasonal variation of NDVI for the drought and normal years
Fig. 5. Seasonal NDVI trend in Dudhwa National Park in 2015 (drought) and 2018 (normal)
years
Fig. 6. Seasonal NDVI trend in Simlipal National Park in 2015 (drought) and 2018 (normal)
years
14
287
288
289
290
291
292
293
In DNP, seasonal variation was found significant (P<0.05) in winter and monsoon,
where the mean NDVI was higher in the winter season (0.56) of 2015 and monsoon season of
2018 (0.77; Table S1). NDVI in the pre-monsoon ranged from 0.46 to 0.76 and 0.45 to 0.78,
in 2015 and 2018, respectively. In monsoon, NDVI ranged from 0.40 to 0.86 and 0.61 to
0.84, in 2015 and 2018, respectively (Fig. 5). Post-monsoon and pre-monsoon did not have a
significant difference in the NDVI for the drought and normal years. At SNP, only the post-
monsoon season showed a significant difference, with NDVI in 2018 (0.90) greater than that
in 2015 (0.85; Fig. 6).
3.3 Yearly variation of NDVI for drought and normal years
For DNP, a t-test was applied on the average data of each month. The mean value of NDVI in
the years 2015 and 2018 was 0.66 and 0.65, respectively. The two years did not have any
significant difference in their NDVI values (Table S1). Similar was the case for the SNP
where no significant difference was observed in the NDVI data between the two years.
3.4. General Linear Mixed effect model
The GLMM was performed to assess the impact of yearly and seasonal variation on the
monthly NDVI for each site. The month was considered as a random factor with year, season,
and rainfall as fixed effects. At DNP, the impact of year, rainfall and year*season*rainfall
interaction were not significant at P<0.05 (Table 1). Season, year*season interaction, and
year*rainfall interaction were found significant (Table 1). The model performed well with
ICC=0.68, a marginal R2 of 0.81 and a conditional R2 of 0.94 (Table S2). A high marginal R2
denoted that fixed effects explained most of the variance in the model. At Simlipal, rainfall
was not included in the model as it did not have a significant correlation with NDVI. The
GLMM performed well with ICC=0.64 (Table S3). Year, season and year*season interaction
did not have any significant effect on the NDVI of the two years under consideration in
15
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
Simlipal (P>0.05). Marginal R2 was also small (0.20) indicating a small variance was
explained by the fixed effects, whereas, the conditional R2 was 0.7, indicating the overall
variance of the fixed and random effects.
Table 1. Type II Wald chi-square tests for the General Linear Mixed Model results of NDVI data
from DNP and Simlipal for drought (2015) and normal (2018) years ( Significance codes: 0 ‘***’
0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1)
Chi. Sq. Df Pr (>Chi. Sq.)
Dudhwa National Park
Year 3.04 1 .
Season 14.2 3 **
Rainfall 0.31 1
Year:season 17.85 3 ***
Year:rainfall 13.03 3 **
Year:season:rainfall 9.03 3 *
Simlipal National Forest
Year 0.99 1
Season 0.85 3
Year:Season 7.15 3 .
4. Discussion
Both the study sites viz., DNP and Simlipal showed contrasting NDVI relations with the
drought and normal rainfall years at different temporal scales. The yearly variation did not
have a significant effect on the NDVI at both sites, whereas, seasonal changes had a
significant impact at DNP. In Simlipal, only in the post-monsoon season, a significant
difference between the two years was found. Overall, the effect of the season in Simlipal did
16
318
319
320
321
322
323
324
325
326
327
328
329
330
not have a significant effect. This indicated the resilience of the Simlipal forest and the
possible availability of water resources to tide over the drought condition. In the next
sections, we have discussed the results according to the monthly, seasonal, and yearly
variations.
4.1 Impact of monthly precipitation on the leaf phenology during drought and normal
years
In DNP, the NDVI in December and February, April, May, July, August, September and
October differed significantly between the drought and normal rainfall years, whereas in
March and November, there was no significant difference in the NDVI. Only two out of ten
months showed no significant difference between the NDVI values. In 2015, West Uttar
Pradesh received less rainfall than normal for the months of February (-59%), May (-37%),
July (-25%), August (-50%), September (-82%), October (-84%), November (-56%) and
December (-63%). Whereas, January, June, March and April received more than the normal
rainfall. In 2018, January, February, March, May, Jun, October, November, December
received -89%, -68%, -87%, -10%, -48%, -84%, -84% and -94% less rainfall than normal.
April, July, August, September received 193%, 24%, 10% and 6% higher rainfall than normal
in 2018. In general, rainfall had a lag effect of one to two months on the NDVI which has
also been reported by Rousta et al. (2020) and Rita et al. (2020). The exploration of this lag
effect is beyond the scope of this paper, however, the monthly variation was translated to a
lesser degree into seasonal differences, indicating a coping mechanism for the phenological
cycles.
In SNP, January (39%), October (-78%) and November (-78%) experienced rainfall
departure from normal rainfall by 39%, -78% and -78% respectively, in 2015. They had
significantly less NDVI in 2015 than in 2018, whereas the summer month of March and
17
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
monsoon month of September, had significantly higher NDVI in 2015, even though March
(2015) experienced a rainfall departure of -61% in 2015. The rainfall departure for March and
September in 2018 was -96% and 21%. September showed a higher NDVI in 2015 than in
2018, even after experiencing a rainfall departure of 0% (2015) as compared to the rainfall
departure of 21% in 2018. From July, the monthly rainfall increased in 2015 reaching its peak
in September, whereas, in 2018, the monthly rainfall had already reached its peak in July and
decreased over September, October and November. A similar effect of the previous months
was observed in January with 39% more than normal rainfall in 2015 and 99% less rainfall in
2018. This lagged effect needs to be investigated further with shorter intervals and larger
temporal scales.
This implied that there was a direct correlation between drought as an environmental
stress factor and NDVI when observed monthly for several months. Since the forest type was
predominantly deciduous, it implied that such vegetation showed significant phenological
changes due to drought stress at shorter durations.
4.2 Impact of seasonal precipitation on tree leaf phenology in drought and normal years
The impact of the season was significant on the NDVI in DNP (Table 1). Season’s interaction
with rainfall and year and rainfall was also significant indicating that in the two years under
consideration, the seasonal changes and their interaction with rainfall patterns across months
and years, determined the phenological changes at DNP. Monsoon and winter seasons
showed a significant difference in NDVI for the years 2015 and 2018 in DNP. The study area
received deficient rainfall during the winter season in 2018 (-80%) than in 2015 (16%).
Higher winter NDVI in 2015 could indicate that leaf fall was prolonged in the winter season
when rainfall was normal or higher than normal. Similarly, the monsoon season in the year
2018 received normal rainfall (1% more than normal) whereas 2015 monsoon rainfall
18
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
received deficient rainfall (-43%). The monsoon NDVI of DNP was significantly higher in
2018 compared to 2015.
Western Uttar Pradesh received higher rainfall (230%) in 2015 than in 2018 (-8%)
during the summer season (pre-monsoon). The departure from normal rainfall in 2018 was
very small and there was no significant difference between the pre-monsoon NDVI values for
the two years. Leaf flushing in a moist deciduous forest in Simplipal occurred in April
(Mishra et al. 2006). The number of species of trees flushing leaves on a given date is strictly
related to rainfall over the preceding fortnight (Lieberman & Lieberman, 1984). However, the
pre-monsoon season did not show any significant difference in NDVI for both DNP and SNP.
In DNP, the post-monsoon rainfall was less than normal in both 2015 (-79%) and 2018 (-
86%). However, there was no significant difference in the NDVI in the post-monsoon season
for the two years.
In SNP, pre-monsoon, monsoon and winter seasons did not show a significant
difference. In post-monsoon, the 2018 NDVI was higher than in 2015. The October to
December 2018 months experienced a rainfall departure of 23% in 2018, whereas, the
departure from normal rainfall was -68% for the same months. The rainfall departure was -
10% and 12% for monsoon, -18% and 20% for summer, and -42% and -99% for winter, for
2015 and 2018, respectively.
Thus, rainfall affected the phenology of trees at the seasonal interval too. However,
the effect of drought was more pronounced during the monsoon and winter season in DNP
and post-monsoon season in SNP. In considering other drivers of change for phenology,
much of the evidence in plants comes from changes observed during the spring season
(Rosenzwei et al., 2008). Many studies have demonstrated that spring phenological events
(e.g., leaf unfolding and flowering) and autumn phenological events (e.g., leaf-fall) have been
19
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
advanced and delayed, respectively, in recent decades, and that both events tend to extend the
length of the growing season (Menzel et al., 2006). The effect of higher rainfall during the
pre-monsoon season did not have any significant effect on the tree leaf phenology compared
to the normal rainfall year. The effect of rainfall was more pronounced in determining leaf-
fall for both the sites, whereas in DNP, leaf health after their emergence was better in the case
of higher precipitation.
4.3 Impact of drought and normal precipitation on the tree leaf phenology
The drought and normal year rainfall did not show any significant difference in the NDVI for
both DNP and SNP. The GLMM results further showed that year and rainfall did not have an
impact on NDVI for the two years, but their interaction with seasonal changes had a
significant influence on the phenological changes in DNP. In SNP, rainfall, year and season,
as well as their interaction did not influence the overall phenological changes in the two
years. However, on the monthly scale, certain differences were found as mentioned in the
results section. This result, when coupled with the monthly and seasonal variations of NDVI
for the drought and normal years, revealed that there was a more pronounced effect of
drought on the tree leaf phenology at shorter intervals which gradually diminished as the
temporal scope of the study increased from months to season and then year. The effect of
drought was apparent in some months, like April and May (pre-monsoon), October and
December (post-monsoon) in DNP, but their effect on the seasonal NDVI was not significant.
Similarly, the winter and monsoon season in DNP and post-monsoon season in SNP showed
significant phenological variations in the NDVI, whereas such a result was not reflected in
the yearly NDVI results. This opens up possibilities whereby trees might be compensating for
the loss of phenological functions or disturbance in phenological function due to
environmental stresses. Such a mechanism would aim to neutralise the effect of
environmental stress, like drought, to maintain a state of homeostasis in plant phenological
20
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
functions. The effect of such environmental stress would be apparent at small intervals of
observation, like weeks and months, but diminished as the observation window increased in
time. The aim would be to achieve a near-normal phenological function in a seasonal cycle of
the four seasons.
4.4 Drought and tree leaf phenology
The geomorphology of the area would play a significant role in maintaining the long-term
phenology of tree species, helping the trees to tide over adverse conditions. Several studies
have shown that long-term precipitation which exceeded the demand of plants (i.e., when
precipitation was more than evapotranspiration) in wet tropical forests led to a negative
impact on forest growth because high water input is negatively correlated to the availability
of other essential plant resources such as nutrients or light (Posada & Schuur, 2011). On the
other hand, water stress due to regional drought may be the dominant contributor to a
widespread increase in tree mortality rates across tree species, sizes, elevations, longitudes
and latitudes (Peng et al., 2011).
Dudhwa National Park is present in the Terai region, near two famous rivers, Suheli
and Mohana, which are called as lifeline of this park. These rivers abound with many nalas
(small streams) which may fulfil the requirement of water for trees, especially in times of
drought. The degree of drought in an area varies widely for different tree species according to
the temperature and availability of soil water (Kushwaha and Singh 2005). The groundwater
availability also varied from 1.48 to 9.66 mbgl from post- to pre-monsoon in DNP which was
less than that in SNP (1.39 to 14.50 mbgl from post- to pre-monsoon). However, the SNP was
fed by Budhabalanga, Kharkai, Jamira and several other tributaries and it had a normal
rainfall of about 1500 mm compared to the 1093 mm in DNP. Because of relatively easier
water availability in SNP, the effect of seasonal or monthly drought was less prominent,
21
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
especially at yearly scales as compared to the phenological changes at seasonal and monthly
scales in DNP. Further, due to the availability of groundwater resources, SNP may be
considered more resilient against environmental stresses, like drought, as root systems also
play an important part. Most flora in these forests, like Sal (Shorea robusta), Asna
(Terminalia tomentosa), Shisham (Dalbergia sisoo), Jamun (Syzygium cumini), Gular (Ficus
glomerata), Sehore (Streblus asper) and Neora (Baringtonia acutangula), have well-defined
root system hence leaf phenology can be revived in time of drought (Kumar, 2009). The
cumulative and time-lag effect of major climatic driving forces on phenology could help in a
better prediction and evaluation of vegetation response (Anderson et al., 2010).
4.5 Limitations of the study
Despite a sound methodology and analysis, our study had some limitations which
arose because of data availability and logistical constraints. The study period was chosen
based on the computation available and the remote sensing data was lessened due to the cloud
cover in some images. However, long-term changes in the phenology of tree species due to
the prolonged effect of drought, or other extreme climate events, which also affect the soil
geomorphology, would be interesting to study from the perspective of climate change. To
determine a reasonable response to phenological change and to improve the adaptation
potential, there is a need to continue and enhance the monitoring of phenological changes.
5. Conclusions
The phenological response of deciduous forest trees towards drought as environmental stress
was analyzed using remote sensing data. The hypothesis was that precipitation patterns
influence the tree leaf phenology. We obtained the NDVI data for a normal (2018) and
drought (2015) year for the DNP and SNP in western Uttar Pradesh and Odisha, respectively.
Both had moist deciduous forest types. We found that there was a significant difference for
22
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
several monthly NDVI values at both sites. The winter and monsoon season showed
significantly different phenological responses to rainfall patterns at DNP, with less NDVI in a
drought year. At SNP, only the post-monsoon season had different NDVI values for the
normal and drought years. At the year level, there was no significant difference between the
phenological responses of the forest at both sites. This made our null hypothesis correct,
according to which there was no variation in phenology of tree leaves due to rainfall stress
(drought) in the tropical deciduous forests. We could draw some interesting conclusions from
this analysis.
First, the effect of drought, as an abiotic stress factor, influenced tree leaf phenology.
However, such influence depended on the scale of phenological observation. It was apparent
at several months, and winter and monsoon seasons (at DNP), and post-monsoon season (at
SNP). The response of the tree to such abiotic stress factors would depend on their
ecophysiology and geomorphology of the region, where a more resilient system would resist
changes in phenological patterns at even lesser intervals of seasons and even months, as seen
in SNP. The resilience of the forest ecosystem would be strengthened by the presence of
groundwater resources. There is a possibility that the tree strives to maintain to near-normal
phenological function over a seasonal cycle.
Further studies in this direction would require higher spatial resolution satellite data,
longer duration of observations, and ground observations to verify the satellite data. Satellite
sensor data might be accompanied by automatic low-cost cameras complemented by
spectroradiometers for research, which can continuously record phenology more accurately
and should be installed in networked sites for long-term monitoring of phenology. Because
these observation systems give us data of phenological observations across multiple scales,
from the canopy, landscape, regional to the global and help to collect long-term human-
collected data. Apart from NDVI, other vegetation indices such as Enhanced Vegetation
23
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
Index (EVI) and Green Red Vegetation Index (GRVI) may also be used in phenological
studies. It has now become important to develop a large monitoring network for forest
phenology across the tropical and sub-tropical countries, which will help in continuous
monitoring and better analysis of the influence of biotic and abiotic factors on phenological
changes.
Funding statement:
We acknowledge the funding provided by the University Grants Commission, New Delhi,
India as a fellowship for the first author.
Declaration of Interest Statement:
The authors of the manuscript titled, “Effect of rainfall variability on tree phenology in moist
tropical deciduous forests”, declare that there exists no conflict of interest between the
authors.
References:
Anderson, J. T., Inouye, D. W., McKinney, A. M., Colautti, R. I., & Mitchell-Olds, T. (2012).
Phenotypic plasticity and adaptive evolution contribute to advancing flowering
phenology in response to climate change. Proceedings of the Royal Society B:
Biological Sciences, 279(1743), 3843-3852.
Bates, D., Maechler, M., Bolker, B., & Walker, S. (2014). lme4: Linear mixed-effects models
using Eigen and S4. R package version 1.1-7.
Borchert, R., Rivera, G., & Hagnauer, W. (2002). Modification of Vegetative Phenology in a
Tropical Semi‐deciduous Forest by Abnormal Drought and Rain 1. Biotropica, 34(1),
27-39.
Canadell, J. G., & Raupach, M. R. (2008). Managing forests for climate change mitigation.
science, 320(5882), 1456-1457.
24
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
Chang, T., Wei, J., Wang, X., Miao, J., Xu, L., Zhang, L., ... & Gao, H. (2019). A rapid and
efficient linear mixed model approach using the score test and its application to GWAS.
Livestock Science, 220, 37-45.
Chmielewski, F. M., & Rötzer, T. (2001). Response of tree phenology to climate change
across Europe. Agricultural and Forest Meteorology, 108(2), 101-112.
Delpierre, N., Vitasse, Y., Chuine, I., Guillemot, J., Bazot, S., & Rathgeber, C. B. (2016).
Temperate and boreal forest tree phenology: from organ-scale processes to terrestrial
ecosystem models. Annals of Forest Science, 73(1), 5-25.
Emanuel, W. R., Shugart, H. H., & Stevenson, M. P. (1985). Climatic change and the broad-
scale distribution of terrestrial ecosystem complexes. Climatic change, 7(1), 29-43.
Estiarte, M., & Peñuelas, J. (2015). Alteration of the phenology of leaf senescence and fall in
winter deciduous species by climate change: effects on nutrient proficiency. Global
change biology, 21(3), 1005-1017.
Fu, Y. H., Piao, S., Zhao, H., Jeong, S. J., Wang, X., Vitasse, Y., ... & Janssens, I. A. (2014).
Unexpected role of winter precipitation in determining heat requirement for spring
vegetation green‐up at northern middle and high latitudes. Global change biology,
20(12), 3743-3755.
Ground Water Resources, Mayurbhanj 2013, available at: https://www.google.com/url?
sa=t&rct=j&q=&esrc=s&source=web&cd=&cad=rja&uact=8&ved=2ahUKEwjIrt7mqr
fvAhWQ63MBHS84DrcQFjABegQIARAD&url=http%3A%2F%2Fcgwb.gov.in
%2FDistrict_Profile%2FOrissa
%2FMayurbhanj.pdf&usg=AOvVaw2gT8a2jRIZ7aMc1H19KnkG
Ground Water Scenario Lakhimpur District, 2013, available at : https://www.google.com/url?
sa=t&rct=j&q=&esrc=s&source=web&cd=&cad=rja&uact=8&ved=2ahUKEwin9Je_qr
fvAhXS7HMBHfgNCPgQFjAAegQIARAD&url=http%3A%2F%2Fcgwb.gov.in
%2FDistrict_Profile%2FUP%2FLakhimpur-
kheri.pdf&usg=AOvVaw1cXrXgE5bNPiUzidR3pE_d
Huechacona-Ruiz, A. H., Dupuy, J. M., Schwartz, N. B., Powers, J. S., Reyes-García, C.,
Tun-Dzul, F., & Hernández-Stefanoni, J. L. (2020). Mapping Tree Species
25
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
Deciduousness of Tropical Dry Forests Combining Reflectance, Spectral Unmixing,
and Texture Data from High-Resolution Imagery. Forests, 11(11), 1234.
Jaworski, T., & Hilszczański, J. (2013). The effect of temperature and humidity changes on
insects development their impact on forest ecosystems in the expected climate
change. Forest Research Papers, 74(4), 345-355.
Kok, Marcel TJ, Rob Alkemade, Michel Bakkenes, Martha van Eerdt, Jan Janse, Maryia
Mandryk, Tom Kram et al. "Pathways for agriculture and forestry to contribute to
terrestrial biodiversity conservation: a global scenario-study." Biological Conservation
221 (2018): 137-150.
Kushwaha, C. P., & Singh, K. P. (2005). Diversity of leaf phenology in a tropical deciduous
forest in India. Journal of Tropical Ecology, 21(1), 47-56.
Mausam (2016a). https://metnet.imd.gov.in/imdmausam/, 67, (1), 311-322
Mausam (2016b). https://metnet.imd.gov.in/imdmausam/, 67, (2), 513-528
Mausam (2016c). https://metnet.imd.gov.in/imdmausam/, 67, (3), 723-744
Mausam (2016d). https://metnet.imd.gov.in/imdmausam/, 67, (4), 955-972
Mausam (2019a). https://metnet.imd.gov.in/imdmausam/, 70, (1), 181-194
Mausam (2019b). https://metnet.imd.gov.in/imdmausam/, 70, (2), 371-396
Mausam (2019c). https://metnet.imd.gov.in/imdmausam/, 70, (3), 593-624
Mausam (2019c). https://metnet.imd.gov.in/imdmausam/, 70, (4), 853-876
Menzel, A., Sparks, T. H., Estrella, N., Koch, E., Aasa, A., Ahas, R., ... & Chmielewski, F.
M. (2006). European phenological response to climate change matches the warming
pattern. Global change biology, 12(10), 1969-1976.
Mishra, R. K., Upadhyay, V. P., Bal, S., Mohapatra, P. K., & Mohanty, R. C. (2006).
Phenology of species of moist deciduous forest sites of Similipal biosphere reserve.
Lyonia, 11(1), 5-17.
26
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
Murthy, M. S. R., Sudhakar, S., Jha, C. S., Reddy, C. S., Pujar, G. S., Roy, A., ... & Roy, P.
S. (2007). Biodiversity Characterisation at Landscape Level using Satellite Remote
Sensing and Geographic Information System in Eastern Ghats, India. Proc. Natl. Sem.
Conserv. Eastern Ghats. ENVIS Centre, EPTRI, Hyderabad, India, 394-405.
Nakagawa, S., & Schielzeth, H. (2013). A general and simple method for obtaining R2 from
generalized linear mixed‐effects models. Methods in ecology and evolution, 4(2), 133-
142.
Pau, S., Wolkovich, E. M., Cook, B. I., Davies, T. J., Kraft, N. J., Bolmgren, K., ... &
Cleland, E. E. (2011). Predicting phenology by integrating ecology, evolution and
climate science. Global Change Biology, 17(12), 3633-3643.
Piao, S., Liu, Q., Chen, A., Janssens, I. A., Fu, Y., Dai, J., ... & Zhu, X. (2019). Plant
phenology and global climate change: Current progresses and challenges. Global
change biology, 25(6), 1922-1940.
Posada, J. M., & Schuur, E. A. (2011). Relationships among precipitation regime, nutrient
availability, and carbon turnover in tropical rain forests. Oecologia, 165(3), 783-795.
Prasad, V. K., Badarinath, K. V. S., & Eaturu, A. (2007). Spatial patterns of vegetation
phenology metrics and related climatic controls of eight contrasting forest types in
India–analysis from remote sensing datasets. Theoretical and Applied
Climatology, 89(1-2), 95.
Reed, B. C., Schwartz, M. D., & Xiao, X. (2009). Remote sensing phenology. In Phenology
of ecosystem processes (pp. 231-246). Springer, New York, NY.
Reich, P. B. (1995). Phenology of tropical forests: patterns, causes, and consequences.
Canadian Journal of Botany, 73(2), 164-174.
Rita, A., Camarero, J. J., Nolè, A., Borghetti, M., Brunetti, M., Pergola, N., ... & Ripullone,
F. (2020). The impact of drought spells on forests depends on site conditions: The case
of 2017 summer heat wave in southern Europe. Global change biology, 26(2), 851-863.
27
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
Robertson, A. I., Dixon, P., & Zagorskis, I. (2020). Phenology and litter production in the
mangrove genus Xylocarpus along rainfall and temperature gradients in tropical
Australia. Marine and Freshwater Research.
Rousta, I., Olafsson, H., Moniruzzaman, M., Zhang, H., Liou, Y. A., Mushore, T. D., &
Gupta, A. (2020). Impacts of drought on vegetation assessed by vegetation indices and
meteorological factors in Afghanistan. Remote Sensing, 12(15), 2433.
Roy, P. S., & Ravan, S. A. (1996). Biomass estimation using satellite remote sensing data—
an investigation on possible approaches for natural forest. Journal of
biosciences, 21(4), 535-561.
Sharma, S., & Mujumdar, P. (2017). Increasing frequency and spatial extent of concurrent
meteorological droughts and heatwaves in India. Scientific reports, 7(1), 1-9.
Singh, B., Jeganathan, C., & Rathore, V. S. (2020). Improved NDVI based proxy leaf-fall
indicator to assess rainfall sensitivity of deciduousness in the central Indian forests
through remote sensing. Scientific reports, 10(1), 1-16.
Singh, R. B., & Mal, S. (2014). Trends and variability of monsoon and other rainfall seasons
in Western Himalaya, India. Atmospheric Science Letters, 15(3), 218-226.
Sotirchos, E. S., Fitzgerald, K. C., & Crainiceanu, C. M. (2019). Reporting of R2 statistics for
mixed-effects regression Models. JAMA neurology, 76(4), 507-507.
Team, R. C. (2013). R: A language and environment for statistical computing.
Toledo, S., Barrios, D., Angel Garcia-Beltran, J., & Roberto Gonzalez-Torres, L. (2021).
Phenology of the threatened species Leptocereus scopulophilus (Cactaceae) in a semi-
deciduous forest of western Cuba. ACTA BOTANICA MEXICANA, 128.
Valderrama-Landeros, L., Flores-Verdugo, F., Rodríguez-Sobreyra, R., Kovacs, J. M., &
Flores-de-Santiago, F. (2021). Extrapolating canopy phenology information using
Sentinel-2 data and the Google Earth Engine platform to identify the optimal dates for
remotely sensed image acquisition of semiarid mangroves. Journal of Environmental
Management, 279, 111617.
28
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
Verma, P., Raghubanshi, A., Srivastava, P. K., & Raghubanshi, A. S. (2020). Appraisal of
kappa-based metrics and disagreement indices of accuracy assessment for parametric
and nonparametric techniques used in LULC classification and change detection.
Modeling Earth Systems and Environment, 1-15.
29
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
Supplementary Informaon
Table S1. Welch’s T-Test results of NDVI for Dudhwa National Park (DNP) and Simlipal
National Park (SNP)
DNP
T_Statistic df P_value
Year -0.6293 149 0.5301
Winter 10.793 29 1.137E-11 ***
pre_monsoon -1.1183 44 0.2695
monsoon -3.9183 44 0.0003074 ***
post_monsoon 0.55289 29 0.5846
February 27.641 14 1.29E-13 ***
March -1.9145 14 0.07622
April -11.509 14 1.595E-08 ***
May 7.1028 14 0.000005314 ***
July 4.207 14 0.0008785 ***
August -6.7006 14 0.0000101 ***
September -3.4219 14 0.004129 **
October 2.1567 14 0.04891 *
November -1.8396 14 0.08712
December 5.7373 14 0.00005137 ***
SNP
Year -2.1355 278.59 0.03359
Winter -1.7064 82.163 0.09172
pre_monsoon -0.19607 70.669 0.8451
monsoon 1.5985 55.519 0.1156
post_monsoon -3.7637 41.489 0.0005195 ***
January -3.3813 27.464 0.00218 **
February 0.48625 23.852 0.6312
March 2.9806 20.82 0.007175 **
May -1.5538 16.082 0.1397
July 1.0364 22.894 0.3108
September 3.2381 23.22 0.003602 **
October -2.2526 23.528 0.0339 *
November -8.1153 20.61 7.50E-08 ***
30
655
656
657
December -1.6637 24.75 0.1088
Table S2. GLM results for NDVI of DNP for 2015 (drought) and 2018 (normal) years
NDVI
Predictors Estimates CI p
(Intercept) 0.69 0.57 – 0.81 <0.001
Year [n] 0.14 -0.04 – 0.31 0.126
Season [postmonsoon] 0.01 -0.19 – 0.21 0.942
Season [premonsoon] 0.05 -0.10 – 0.20 0.519
Season [winter] 0.13 -0.11 – 0.37 0.294
rainfall 0.00 -0.00 – 0.00 0.699
Year [n] * Season[postmonsoon] -0.14 -0.35 – 0.08 0.207
Year [n] * Season [premonsoon] -0.37 -0.61 – -
0.13
0.002
Year [n] * Season[winter] -0.33 -0.57 – -
0.09
0.007
Year [n] * rainfall -0.00 -0.00 – 0.00 0.261
Season [postmonsoon] *rainfall -0.01 -0.04 – 0.03 0.727
Season [premonsoon] *rainfall -0.00 -0.01 – -
0.00
0.005
Season [winter] *rainfall -0.06 -0.10 – -
0.01
0.009
(Year [n] * Season[postmonsoon]) * rainfall -0.00 -0.03 – 0.02 0.920
(Year [n] * Season[premonsoon]) * rainfall 0.02 0.01 – 0.03 0.003
(Year [n] * Season[winter]) * rainfall 0.00 -0.03 – 0.03 0.991
Random Effects
σ20.00
31
658
659
τ00 Month 0.00
ICC 0.68
N Month 10
Observations 20
Marginal R2 / Conditional R20.808 / 0.939
Table S3. GLM results for NDVI of Simlipal forest for 2015 (drought) and 2018 (normal)
years
NDVI
Predictors Estimates CI p
(Intercept) 0.86 0.82 – 0.91 <0.001
Year [n] -0.01 -0.05 – 0.02 0.436
Season [postmonsoon] -0.01 -0.08 – 0.05 0.648
Season [premonsoon] 0.02 -0.04 – 0.08 0.478
Season [winter] -0.00 -0.06 – 0.05 0.919
Year [n] * Season [postmonsoon] 0.07 0.01 – 0.12 0.015
Year [n] * Season [premonsoon] 0.01 -0.05 – 0.06 0.796
Year [n] * Season [winter] 0.02 -0.02 – 0.07 0.332
Random Effects
σ20.00
τ00 Month 0.00
ICC 0.64
N Month 10
Observations 19
32
660
661
662
Marginal R2 / Conditional R20.199 / 0.711
33
663
664
