macroecological factors shape local-scale spatial … › portal › files › 121439797 ›...
TRANSCRIPT
Macroecological factors shape local-scale spatial patterns in
agriculturalist settlements
Tingting Taoa#, Sebastián Abadesb#, Shuqing Tengc, Zheng Y.X. Huangd, Luís Reinoe,f,g, Bin J.W. Chenh, Yong Zhangh, Chi Xua,1, Jens-Christian Svenningc
a School of Life Sciences, Nanjing University, Nanjing 210023, China,b Center for Genomics and Bioinformatics, Faculty of Science, Universidad Mayor, Camino La Pirámide 5750, Huechuraba, Chile,c Section for Ecoinformatics & Biodiversity, Department of Bioscience, Aarhus University, Ny Munkegade 114, DK-8000 Aarhus C, Denmark,d College of Life Science, Nanjing Normal University, Nanjing 210046, China,e CIBIO/InBIO, Research Centre in Biodiversity and Genetic Resources, University of Porto, 4485-661 Vairão, Portugal,f CIBIO/InBIO, Research Centre in Biodiversity and Genetic Resources, Universidade de Évora, 7004-516 Évora, Portugal,g CEABN/InBIO-Centro de Estudos Ambientais ‘Prof. Baeta Neves’, Instituto Superior de Agronomia, Universidade de Lisboa, Tapada da Ajuda, 1349-017 Lisboa, Portugal,h College of Biology and the Environment, Nanjing Forestry University, Nanjing 210037, China
1 To whom correspondence should be addressed: Chi Xu, E-mail: [email protected] of Life Sciences, Nanjing University, Nanjing 210023, ChinaTel & Fax: +86-(0)25-89684396
#Authors contributed equally to this work
1
1
2
3456
789
101112131415161718192021
22232425
2627
Abstract
Macro-scale patterns of human systems ranging from population distribution to linguistic
diversity have attracted recent attention, giving rise to the suggestion that macroecological
rules shape the assembly of human societies. However, in which aspects the geography of
our own species is shaped by macroecological factors remains poorly understood. Here, we
provide a first demonstration that macroecological factors shape strong local-scale spatial
patterns in human settlement systems, through an analysis of spatial patterns in
agriculturalist settlements in eastern mainland China based on high-resolution Google Earth
images. We used spatial point pattern analysis to show that settlement spatial patterns are
characterised by over-dispersion at fine spatial scales (0.05-1.4 km), consistent with territory
segregation, and clumping at coarser spatial scales beyond the over-dispersion signals,
indicating territorial clustering. Statistical modelling shows that, at macroscales, potential
evapotranspiration and topographic heterogeneity have negative effects on territory size, but
positive effects on territorial clustering. These relationships are in line with predictions from
territory theory for hunter-gatherers as well as for many animal species. Our results help to
disentangle the complex interactions between intrinsic spatial processes in agriculturalist
societies and external forcing by macroecological factors. While one may speculate that
humans can escape ecological constraints because of unique abilities for environmental
modification and globalised resource transportation, our work highlights that universal
macroecological principles still shape the geography of current human agricultural societies.
Keywords: human macroecology, territory, scale, energy, topography, point pattern
2
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
1. Introduction
Among all organisms on Earth, our own species (Homo sapiens) is unique in many senses.
One remarkable characteristic of humans is the extraordinarily high sociocultural diversity,
despite comparatively little genetic variation [1]. Understanding sociocultural traits in
particular human communities has been a traditional focus in social sciences. However, recent
studies now start to put together the pieces for the ‘big picture’ of human societies, having led
to the discovery of a number of conspicuous large-scale patterns. Interestingly, a portion of
these anthropogenic patterns parallel classical patterns documented in the field of
macroecology [2], as seen e.g., in the congruence of latitudinal gradients in cultural and
species diversity [3-6]. It is increasingly suggested that human life histories are constrained
by ecological factors, and large-scale patterns of sociocultural traits may be explained by
macroecological processes that are analogous to those underlying biological patterns, giving
rise to the emerging field ‘human macroecology’, which aims to understand human-
environment relationships through quantifying statistical patterns of anthropogenic systems at
large spatiotemporal scales [7]. This emerging field has provided exciting findings showing
that macroecological rules underpin a wide range of anthropogenic patterns, e.g., population
distributions [8, 9], cultural distributions and dynamics [10-13], disease and public health [14,
15], as well as economy and sustainability [16, 17].
Macroecological approaches have provided valuable insights on many aspects of
ecological systems [18]. Spatial pattern is a fundamental characteristic of ecological systems;
importantly, it provides useful signatures for elucidating ecological interactions and processes
that structure the ecosystems [19-21]. However, the link between local spatial
patterns/processes and macroecology is not well quantified or understood. An important
example that illustrates such link are the so-called ‘Turing patterns’, a special class of patterns
periodically extended in space, resembling regular spots, stripes or labyrinths, that have been
found in a wide range of ecosystems such as dryland vegetation and coastal mussel beds [22,
23]. It has been suggested that Turing vegetation patterns are mainly governed by micro-scale
interactions, represented as coupled short-range positive feedbacks and long-range negative
3
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
feedbacks in general [24]. However, a recent worldwide analysis has shown that the
occurrence of these patterns are also well correlated with macroecological factors including
climate, topography and soil [25]. Such cross-scale correlations imply that at broad spatial
scales, relevant macroecological factors can substantially determine the local biotic and
abiotic factors underpinning spatial patterning. In this sense, linking local-scale spatial
patterns to large-scale determinants not only contributes to identifying the mechanisms
underlying pattern formation, but also help to scale up to better understand system
functioning and dynamics at macroscopic scales. So far, it remains unclear if and how
macroecological factors shape local spatial patterns in most systems.
Here, we connect the two emerging themes, human macroecology and cross-scale spatial
pattern formation, by assessing the role of macroecological factors in shaping local-scale
spatial patterns of anthropogenic systems. We are particularly interested in rural human
settlements, because they are among the most conspicuous imprints that humans created on
the Earth’s land surfaces, and are home to about half world’s populations. A glimpse at rural
settlements from airplanes or remotely sensed images may provide the impression that their
spatial patterns mimic those of other species’ colonies (termites for instance) at proportionate
scales (see Fig. 1 for some snapshots). This said, there is no doubt that spatial patterns of
humans are likely to differ from those of termites, because humans are arguably not a eusocial
species [26], and human settlements are generally shaped by more diverse and complex
displays of social behaviours. In social sciences, there have been extensive studies suggesting
the important role of diverse sociocultural factors in the formation and spatial patterning of
rural settlements, ranging from general theories of spatial location [27, 28] to specific
sociocultral processes [29-31]. However, irrespective of sociocultural context, settlement
formation and associated land use intensification involve emergent processes resulting from
interactions between human individuals, groups, and the environments [32]. If these human
interactions are governed by coarse-grained ecological constraints as in other species, one
would expect that human settlement patterning follows general macroecological rules.
Indeed, recent work has revealed macroecological underpinnings (mostly related to
water-energy patterns) of human collective behaviour and social structures, providing
4
78
79
80
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
106
compelling clues for exploring settlements patterning in this framework. In hunter-gatherer
societies, territory size of foraging groups has been found to be a function of energy (food)
supply and population density [33, 34]. A corollary is that the dependence on these factors
should also hold for the spacing distance between camps as a surrogate measure of territory
size. When foraging groups strongly compete for food resources, it is reasonable to expect
regularly distributed (i.e., over-dispersed) camps as a result of intensive territory segregation.
In contrast, if positive relationships (e.g., cooperative defense and food sharing) are prevalent,
group territories would tend to overlap [35, 36], likely resulting in clustered patterns.
Little macroecological work has been done on agrarian societies, despite their
importance. In parallel with hunter-gatherers, an individual agriculturalist settlement can be
seen as a group that mostly utilise resources (agricultural production) from surrounding lands
as its territory. An important transition from hunter-gatherer to agrarian societies is land use
intensification (farming), which makes resources more abundant and predictable [37]. This
could have fundamental consequences on collective behaviour and the settlement system.
First, denser resources would allow for higher population densities and smaller territories
[35]. Second, high-quality resources and more sedentary groups would intensify territory
defense, as predicted by the theory of economic defendability [37, 38]. Third, rich resources,
dense populations, and sedentary groups could favour the development of complex, long-
range social networks [37] between settlements. These characteristics have major implications
for spatial patterning of agriculturalist settlements, leading us to propose the following
hypotheses:
(1) Significant territory segregation occurs between neighbouring agriculturalist settlements,
resulting in over-dispersed settlement patterns at local scales. At coarse spatial scales,
agriculturalist settlements could have clustered territories and hence under-dispersed
settlement patterns due to environmental constraints or positive social interactions (e.g.,
within ethnic groups).
(2) At macroscopic scales, high resource availability (energy measured as PET, soil quality,
hydrological density) has a negative influence on the territory size of agriculturalist
5
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
settlements, and also favours territory clustering.
(3) In mountainous terrains, steep topography result in smaller territory sizes, while also
forcing territorial clustering.
(4) Resource availability and topographic forcing can produce indirect influences on
settlement patterning via mediating settlement size and population density.
To test these hypotheses, we used high-resolution (0.6 m) remotely sensed images to
scrutinize rural settlements across mainland China, where traditional agriculture has a long
history and has produced strong landscape imprints. We conducted spatial point pattern
analysis to study spatial associations between the settlements and thereafter multiple
regression and structural equation modelling (SEM) to examine the effects of
macroecological factors on the detected patterns.
2. Methods
(a) Study area and settlement data
China has a history of thousands of years as an agrarian society. Despite rapid urbanisation in
recent decades, over 700,000 rural settlements are home to c. 60% of the Chinese population
[39]. A clear national-scale differentiation has been observed across mainland China, with
most of the population and settlements distributed in the south-east part in contrast with the
north-west part where is characterised by drylands and alpine plateaus. This divide roughly
follows a straight line stretching diagonally across China, from the Heihe City to the
Tengchong County (known as the ‘Heihe-Tengchong Line’ or ‘Hu Huanyong Line’ [40]). We
restricted our study area to area southeast of this line, as this is where relatively dense
settlements accommodating c. 96% of total Chinese population are found (Fig. 1a).
We divided the study area into 490 grid cells sized of 100 km×100 km, within each of
which we systematically sampled a 10 km×10 km plot using high-resolution remote sensing.
In this way, the sampling plots were relatively evenly distributed across the study area. The
plots were selected following the criteria: 1) Each plot was capable of representing a
relatively homogeneous agricultural landscape, i.e., no apparent non-agricultural land use
6
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
features were found within the plot. 2) Areas that are in proximity to (< 10 km) or contain
towns/cities were avoided. 3) Grid cells intersected with metropolitan regions or urban
agglomerations were excluded to avoid areas under substantial urban land use. When multiple
candidate remote sensing plots existed in a given 100 km×100 km cell, we selected the one
closest to the cell centroid. We identified 320 sampling plots for the 490 grid cells (Fig. 1).
For each sampling plot we collected and geo-referenced (using the UTM projections)
high-resolution (0.6 m) QuickBird images covering 2005-2010, freely available from Google
Earth. The distribution of rural settlements was derived using a supervised classification (with
the maximum likelihood method) in combination with visual inspections (manual tracing was
used afterwards to correct classification errors). Each individual settlement was represented as
a vector polygon with its centroid representing the settlement’s spatial position. The centroids
with geographic coordinates were taken as input data for the subsequent spatial point pattern
analysis (Fig. S1). The image processing was done using ENVI 4.8 and ArcGIS 10.2.
(b) Spatial point pattern analysis
We applied the univariate point pattern analysis (PPA) based on the second-order pair-
correlation function g(r) to test if the settlements points present significant signals of under-
dispersion or over-dispersion in contrast with complete spatial randomness (CSR, as the null
model) over a range of observational spatial scales. A clear merit of the PPA approach for
testing our hypotheses is that it is capable of detecting nested hierarchical spatial structures at
multiple spatial scales with a single summary statistic (i.e., the g(r) statistic), facilitating
subsequent statistical modelling. To characterise any spatial association signal (over- or
under-dispersion) detected beyond the 99% CIs of CSR, we quantified 1) presence/absence of
the signal (as a binary variable), and 2) spatial scale range across which the signal arises (as a
continuous variable), calculated as the width of the scale domain presenting the non-CSR
signal (Fig. 2). For the detected signals we used the nomenclature indicating both association
properties and their relative scales, e.g., small-scale over-dispersion or large-scale under-
dispersion (see Results). The point pattern analysis was completed with the Programita
software [41]. See S1 for more detailed descriptions.
7
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
(c) Macroecological variables
We selected four environmental variables for testing our hypotheses. Mean annual potential
evapotranspiration (PET) was used to characterise energy availability. Elevation range
derived from Shuttle Radar Topography Mission (SRTM) data was used to characterise
topographic relief. Soil quality was characterised by total exchangeable bases (the sum of
HCl-soluble bases including Ca2+, Mg2+, K+ and Na+) as the major components of soil fertility
[42]. The length of rivers and major channels per unit area (river density for abbreviation)
derived from the China hydrological map was used to characterise the hydrological system. In
addition, our fourth hypothesis involves settlement size and regional population density. Mean
settlement size was calculated as the mean area of the settlement polygons derived from the
remotely sensed images (Fig. S1). Regional population density was calculated from a Chinese
gridded database on population density. The mean values of the abovementioned variables
were calculated for each sampling plot, except that the topographic variable was represented
by elevation range within the plot. In addition, as the broad geographic extent of China covers
a diversity of social/ethnic backgrounds, to corrected for this potential influence, we used a
categorical variable representing the major historical/cultural regions, i.e., Northern, Southern
and Western China. This region variable was derived from a national-scale regionalisation
map developed by cultural geographers to classify characteristics of traditional Chinese
settlements [43]. Data sources for the environmental variables are listed in Table S1.
(d) Statistical analyses
We used multiple regression to examine the relations between the macroecological variables
(as predictors) and the spatial association variables previously quantified from the point
pattern analysis (as response). For presence/absence of spatial association we fitted
generalised linear models (GLMs) with the binomial distribution and logit link (logistic
regression), while for the scale range of spatial association we fitted ordinary least square
(OLS) models. We checked for multicollinearity of the predictors based on the variation
inflation factor (VIF). The VIF value was less than 3 for all predictors, indicating no serious
multicollinearity. We started from a full model, then performed a stepwise backward model
8
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
selection to obtain a parsimonious model based on AIC. The resulting parsimonious models
are robust when all possible models are considered. We also checked the Moran’s I coefficient
(based on the inverse distance weight method) for the model residuals, and detected rather
low absolute values of Moran’s I (Table S2-S4), suggesting little potential for spatial
autocorrelation problems [44]. Before model fitting, elevation range and settlement size were
log-transformed and scale ranges of the pattern features were log(x+1)-transformed to avoid
highly skewed distributions. We used piecewise structural equation models (piecewise SEMs)
to test our fourth hypothesis, where population density and settlement size were included as
modulator variables. For the logistic regression, we used a random subset with equal presence
and absence data to avoid the data imbalance problem. The statistical analyses were
implemented in R 3.3.1 [45] with the ape [46], and piecewiseSEM [47] packages.
3. Results
Our results showed that local spatial associations between agriculturalist settlements in
eastern China are characterised by two classes of non-random signals, as indicated by the
summary statistic g(r) of the point pattern analysis. These two classes of signals are structured
within a spatial scale hierarchy (Fig. 2). One class of signals represents over-dispersion (as
indicated by g(r) values falling below the lower 99% confidence intervals of CSR) at the
lower end of the scale gradient (small-scale over-dispersion or SO hereafter). Within the scale
range of SO (SRSO for abbreviation; 0.05-1.4 km), there was a significantly lower probability
of finding a neighbouring pair of settlement points than expected by chance. Interpreted
intuitively, this scale range reflects the average distance at which the settlements are
‘repelling’ each other. SRSO can thus serve as a surrogate measure of territory size. The signal
of SO was found in all 320 sampling plots, suggesting that strong territory segregation among
neighbouring settlements is ubiquitous.
At larger spatial scales beyond SO (usually >0.2 km), we detected a second class of
signals representing under-dispersion (as indicated by g(r) values rising above the upper 99%
confidence intervals of CSR, large-scale under-dispersion or LU hereafter) in ~30% of the
plots, with the rest plots characterised by CSR (Fig. 1a). Note that for a given study plot,
9
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
over-dispersion was found to be always present at smaller spatial scales than under-
dispersion, but this is not necessarily the case when comparing between different plots. An
interpretation is that at any given distance within the scale range of LU (SRLU for
abbreviation, 0.1-2.6 km), there was a significantly higher probability of finding a pair of
settlement points than expected by chance, suggesting that settlements are ‘attracting’ each
other. SRLU can thus serve as a surrogate measure of cluster size.
Our results from the regression models showed that mean territory size (as measured by
SRSO) had significantly negative correlations with PET and elevation range (Fig. 3, S2, Table
S2), suggesting that higher energy availability and steeper topography favour smaller
territories, or put in another way, finer-grained territory segregation. We also found a
significant effect of the region variable, suggesting that territory size is influenced by large-
scale historical/cultural factors. Soil fertility and river density did not have significant effects.
Our SEM analysis further suggested several plausible pathways through which population
density and settlement size transmit the effects of the macroecological factors on territory size
(Fig. S4). The two major pathways are: 1) higher PET and lower elevation ranges can favour
higher population densities, in turn leading to smaller territories; 2) both PET and elevation
range have negative effects on settlement size, while settlement size has a positive effect.
Our statistical models also showed that occurrence of clustered patterns had significantly
positive correlations with PET and elevation range, suggesting that higher energy availability
and steeper topographic relief favored clustered territories at larger scales (Fig. 3, Table S3).
Again, our SEM analysis suggested population density and settlement size could play as
mediators of the influences from PET and elevation range (Fig. S4). Both population density
and settlement size had negative effects on the occurrence of clustered territories. The
abovementioned direct and indirect effects were similar, but weaker for cluster size (as
measured by SRLU, where present), except for a more pronounced effect of river density on
cluster size (Fig. 4, S3, S4, Table S4).
10
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
4. Discussion
The point pattern analyses revealed marked regularities in spatial pattern features in
agriculturalist settlements across eastern mainland China. These features could be
decomposed in terms of hierarchical patterns at two spatial scales: ubiquitous over-dispersed
patterns at small spatial scales, supporting the hypothesis of significant territory segregation;
and a moderate level of under-dispersed patterns at larger spatial scales, in line with the
hypothesis of territorial clustering.
Are these local-scale spatial pattern features governed by macroecological factors at
broad scales? Our statistical models indeed suggest so. One of our major findings points to a
significantly negative effect of PET on territory size of agriculturalist settlements, supporting
our second hypothesis. This energy effect coincides well with its counterparts for hunter-
gatherers [34], as well as group-living animals in general [48]. It has been commonly
observed that animal territory size expands when resource abundance decreases, with
individuals maintaining per-capita intake rates unchanged [48, 49]. When it comes to the
patterning of agriculturalist settlements, a similar mechanism may operate because individual
settlements would need more arable lands when per unit agricultural production is limited by
relatively low water-energy availability.
Further comparisons strengthen the suggestion that ‘universal’ ecological principles
shape territoriality of humans and animal species (here the distinction between humans and
animals is arbitrary from a biological perspective). The territory theory of animals predicts
that a relaxed degree of population crowding or a larger group (colony) size can result in
enlarged territory sizes [48]. In parallel, an analogous prediction for agriculturalist settlements
is that territory size should be positively correlated with settlement (group) size, because a
larger settlement with more residents would need a larger territory; while regional population
density should have a negative correlation with territory size, because increasingly crowded
populations at a regional level would compress individual territories if territory overlapping is
absent [36]. Our results from SEM confirmed that it is indeed possible that population density
and settlement size can have such mediation effects, as driven by energy availability (Fig S4).
11
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
The transmission pathway mediated via population density can be interpreted as high energy
availability providing support for higher population densities, which in turn compress
territories. Indeed, this mechanism has also been widely observed in group-living animals,
such as birds and mammals [48, 50, 51]. The other pathway represents the scenario that low
energy availability favours larger settlements (groups), which need larger territories. The
negative effect of PET on settlement size is plausibly explained by the sociality of humans: in
resource-poor areas, aggregating into (appropriately) larger settlements can have obvious
benefits for the residents as a collective, in the sense that they can ameliorate the harshness
via cooperation. Such enhanced positive interaction is a common strategy for plants in
stressful environments (the so-called stress gradient hypothesis) [52, 53], and has been
observed in some group-living animals, e.g., Octodon degu [54]. However, generally strong
intra-specific competition for food resources may be an important reason why this trend is not
prevalent in animal communities [55, 56].
Territorial clustering is not uncommon in nature [57-60]. For animals the prevailing
explanations for this pattern, regardless of specific mechanisms, address the benefits in
various forms gained through conspecific attraction [57]. A particular case is that in resource
abundant areas, territorial animals may settle closely next to each other, because in this way
they can identify and thereby access the high-quality resources more efficiently [58-60]. This
perhaps also partly explains our observed positive correlation between settlement clustering
and PET. Another benefit can be gained through intraspecific cooperation in terms of e.g.
intruder defense and predator protection [57, 61]. Likewise, rural settlement clustering could
be a sign of strong cooperation between the settlements. Indeed, it has been suggested that in
higher-resource settings, human societies tend to develop higher levels of cooperation and
more complex social networks [37]. In addition, settlement size and population density can
again play as significant mediators as suggested by SEM. A possible explanation is that
between-settlement connections would be weakened when the settlements are large and
relatively far away from their neighbours (because of their large territories), while dense
populations may enhance social connections that can interlink most of the settlements across
the whole region and therefore destruct formation of local clusters.
12
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
We did not observe the hypothesised effects of soil fertility and river density (Fig. S2,
S3). One possibility is that, humans can increasingly temper the limitation of soil quality and
surface water resource with the aid of technologies that have been developed through the
long-term agricultural practices during the past millennia, while rivers can act as another
environmental forcing that tends to obstruct the development of clustered patterning through
segregation. It is also possible that sometimes settlements developing along rivers form linear
patterns, but such river-settlement association cannot be well detected using the PPA. The
effects of strong macroecological forcing on settlement patterning might be dependent on the
environmental modification ability of our own species. A possible scenario in the future is that
advancement of technology would make humans escape macroecological constraints,
eventually reshaping settlements patterning and persistence. Still, it remains to be studied if,
when and how human settlements start to become sytematically unconstrained by
macroecological rules, pointing to an important area for future research on human
macroecology, i.e., focusing on regions most strongly characterised by the expected
Anthropocene societal development trends.
Topography as a strong environmental forcing usually poses marked limitations on
human behaviours [62, 63], increasing the difficulty for agricultural land use, settlement
establishment and human mobility. Our analysis suggested that steep topography restricts
territory size directly as well as indirectly via restricting settlement size. While steep
topography could also have an indirect positive effect via limiting population density, though
any such effect was outweighed by the overall negative effects. Another important
consequence of topographic relief is territorial clustering, in the sense that only some
(relatively flat) areas are suitable or attractive for settlements in complex terrains. This direct
forcing coincides with the indirect effects that small settlements and low population densities
in complex terrains also favour clustered patterns.
Quantifying sociocultural and historical effects is an important future challenge for
improving our understanding of settlement patterning [64]. The coarse-scale regionalisation
variable used in this study was not able to well reflect the LU signals (Table S3, S4),
suggesting that more complex processes are likely playing a role. This is especially the case
13
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
for understanding the reason why territorial clustering is mostly present in the southernmost
part of mainland China (Fig. 1a). A promising approach for future studies would be using
phylogenetic information (based on cultural lineages, rather than genetic lineages of humans)
to account for cultural/historic similarities, given that reliable cultural/historical phylogenetic
trees can be constructed.
Taken together, our results help to disentangle the complex interactions between intrinsic
spatial processes in agriculturalist societies and forcing by macroecological factors (see Fig. 4
for a synthesis of the direct and indirect effects). Settlement spatial patterns provide us an
opportunity to unravel the interesting similarities in the territoriality of agriculturists, hunter-
gatherers and animals, and to understand the ecological mechanisms that are plausibly
‘universal’ to humans and other species. Although territory theory has long been established
in ecology, successfully explaining territory patterning of many animal species, our work is,
to our knowledge, the first extension of territory theory to the geography of human settlement
patterns. Macroecological factors have played a strong role in shaping current local-scale
spatial patterns in agriculturalist settlement systems despite the fact that humans have the
ability to temper ecological constraints through environmental modification and globalised
resource transfer. Our work demonstrates the value of a macroecological perspective for
understanding human collective behaviour and settlement systems, highlighting the ecological
nature of human societies. While our focus is on ecological interactions, further consideration
of historical-geographic context would facilitate comprehensive understanding of variability
of settlement patterning. Importantly, macro-scale sociopolitical factors (e.g., governmental
planning) can play a substantial role in settlement patterning [65], leaving it an open question
when and how sociopolitical factors lead to settlement systems that deviate from those
generated by spontaneous intrinsic socio-ecological dynamics and extrinsic macroecological
forcing.
14
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
Funding
This work is supported by The National Key R&D Program of China (2017YFC0506001),
the National Natural Science Foundation of China (NSFC 31770512) and the Fundamental
Research Funds for the Central Universities (0208-14380068) awarded to C.X. J.-C.S.
considers this work a contribution to his VILLUM Investigator project “Biodiversity
Dynamics in a Changing World” funded by VILLUM FONDEN and the Danish National
Research Foundation Niels Bohr professorship project Aarhus University Research on the
Anthropocene (AURA). S.A. acknowledges support from FONDECYT 1170995. S.T.
acknowledges support from China Scholarship Council (201406190179) and the European
Research Council (ERC-2012-StG-310886-HISTFUNC). L.R. received support from
the Portuguese Ministry of Science and Higher Education and the
European Social Fund, through FCT, under POPH – QREN – Typology 4.1
(post-doc grant SFRH/BPD/93079/2013). Z.Y.X.H. acknowledges support from
NSFC (31500383). B.J.W.C. acknowledges support from NSFC (31600328) and Natural
Science Foundation of Jiangsu Province, China (BK20160924).
Acknowledgements
We thank Marten Scheffer and Guochun Shen for valuable discussion. We also thank
Melodie McGeoch and two anonymous reviewers for constructive comments on an earlier
version of this manuscript.
Author contributions
C.X. designed the study. T.T., S.A. and C.X. collected and carried out spatial analyses, S.A.
and C.X. carried out statistical analyses, C.X. wrote the first draft, all authors interpreted the
results and revised the manuscript. L.R., C.X. and J.-C.S. approved the final manuscript. T.T.
and S.A. contributed equally to this work.
Competing interests
The authors declare no competing interests.
15
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
References
[1] Laland, K.N., Odling-Smee, J. & Myles, S. 2010 How culture shaped the human genome: bringing genetics and the human sciences together. Nat Rev Genet 11, 137-148. (doi:10.1038/nrg2734).[2] Brown, J.H. & Maurer, B.A. 1989 Macroecology: the division of food and space among species on continents. Science 243, 1145. (doi:10.1126/science.243.4895.1145).[3] Collard, I.F. & Foley, R.A. 2002 Latitudinal patterns and environmental determinants of recent human cultural diversity: do humans follow biogeographical rules? Evol Ecol Res 4, 371-383.[4] Laitin, D.D., Moortgat, J. & Robinson, A.L. 2012 Geographic axes and the persistence of cultural diversity. Proc Natl Acad Sci USA 109, 10263-10268. (doi:10.1073/pnas.1205338109).[5] Pagel, M. & Mace, R. 2004 The cultural wealth of nations. Nature 428, 275-278. (doi:10.1038/428275a).[6] Sutherland, W.J. 2003 Parallel extinction risk and global distribution of languages and species. Nature 423, 276-279. (doi:10.1038/nature01607).[7] Burnside, W.R., Brown, J.H., Burger, O., Hamilton, M.J., Moses, M. & Bettencourt, L.M.A. 2012 Human macroecology: linking pattern and process in big-picture human ecology. Biol Rev 87, 194-208. (doi:10.1111/j.1469-185X.2011.00192.x).[8] Xu, C., Chen, B.J.W., Abades, S., Reino, L., Teng, S.Q., Ljungqvist, F.C., Huang, Z.Y.X. & Liu, M.S. 2015 Macroecological factors explain large-scale spatial population patterns of ancient agriculturalists. Global Ecol Biogeogr 24, 1030-1039. (doi:10.1111/geb.12343).[9] Benito, B.M., Svenning, J.-C., Kellberg-Nielsen, T., Riede, F., Gil-Romera, G., Mailund, T., Kjaergaard, P.C. & Sandel, B.S. 2016 The ecological niche and distribution of Neanderthals during the Last Interglacial. J Biogeogr. (doi:10.1111/jbi.12845).[10] Amano, T., Sandel, B., Eager, H., Bulteau, E., Svenning, J.-C., Dalsgaard, B., Rahbek, C., Davies, R.G. & Sutherland, W.J. 2014 Global distribution and drivers of language extinction risk. Proceedings of the Royal Society B: Biological Sciences 281. (doi:10.1098/rspb.2014.1574).[11] Botero, C.A., Gardner, B., Kirby, K.R., Bulbulia, J., Gavin, M.C. & Gray, R.D. 2014 The ecology of religious beliefs. Proc Natl Acad Sci USA 111, 16784-16789. (doi:10.1073/pnas.1408701111).[12] Capitán, J.A., Bock Axelsen, J. & Manrubia, S. 2015 New patterns in human biogeography revealed by networks of contacts between linguistic groups. Proceedings of the Royal Society B: Biological Sciences 282. (doi:10.1098/rspb.2014.2947).[13] Foley, R.A. & Mirazón Lahr, M. 2011 The evolution of the diversity of cultures. Philosophical Transactions of the Royal Society B: Biological Sciences 366, 1080. (doi:10.1098/rstb.2010.0370).[14] Guernier, V., Hochberg, M.E. & Guégan, J.-F. 2004 Ecology drives the worldwide distribution of human diseases. PLoS Biol 2, e141. (doi:10.1371/journal.pbio.0020141).[15] Stephens, P.R., Altizer, S., Smith, K.F., Alonso Aguirre, A., Brown, J.H., Budischak, S.A., Byers, J.E., Dallas, T.A., Jonathan Davies, T., Drake, J.M., et al. 2016 The macroecology of infectious diseases: a new perspective on global-scale drivers of pathogen distributions and impacts. Ecol Lett 19, 1159-1171. (doi:10.1111/ele.12644).[16] Burger, J.R., Allen, C.D., Brown, J.H., Burnside, W.R., Davidson, A.D., Fristoe, T.S., Hamilton, M.J., Mercado-Silva, N., Nekola, J.C., Okie, J.G., et al. 2012 The macroecology of sustainability. PLoS Biol 10, e1001345. (doi:10.1371/journal.pbio.1001345).[17] Brown, J.H., Burger, J.R., Burnside, W.R., Chang, M., Davidson, A.D., Fristoe, T.S., Hamilton, M.J., Hammond, S.T., Kodric-Brown, A., Mercado-Silva, N., et al. 2014 Macroecology meets
16
411
412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453
macroeconomics: Resource scarcity and global sustainability. Ecol Eng 65, 24-32. (doi:DOI 10.1016/j.ecoleng.2013.07.071).[18] Smith, F.A., Lyons, S.K., Ernest, S.M. & Brown, J.H. 2008 Macroecology: more than the division of food and space among species on continents. Prog Phys Geog 32, 115-138. (doi:10.1177/0309133308094425).[19] Levin, S.A. 1992 The problem of pattern and scale in ecology. Ecology 73, 1943-1967. (doi:10.2307/1941447).[20] McIntire, E.J. & Fajardo, A. 2014 Facilitation as a ubiquitous driver of biodiversity. New Phytol 201, 403-416. (doi:10.1111/nph.12478).[21] Xu, C., Holmgren, M., Van Nes, E.H., Maestre, F.T., Soliveres, S., Berdugo, M., Kefi, S., Marquet, P.A., Abades, S. & Scheffer, M. 2015 Can we infer plant facilitation from remote sensing? a test across global drylands. Ecol Appl 25, 1456-1462. (doi:10.1890/14-2358.1).[22] Liu, Q.X., Herman, P.M.J., Mooij, W.M., Huisman, J., Scheffer, M., Olff, H. & van de Koppel, J.V. 2014 Pattern formation at multiple spatial scales drives the resilience of mussel bed ecosystems. Nature Communications 5. (doi:10.1038/ncomms6234).[23] Rietkerk, M., Dekker, S.C., De Ruiter, P.C. & Van de Koppel, J. 2004 Self-organized patchiness and catastrophic shifts in ecosystems. Science 305, 1926-1929. (doi:10.1126/science.1101867).[24] Rietkerk, M. & Van de Koppel, J. 2008 Regular pattern formation in real ecosystems. Trends in Ecology and Evolution 23, 169-175. (doi:10.1016/j.tree.2007.10.013).[25] Deblauwe, V., Barbier, N., Couteron, P., Lejeune, O. & Bogaert, J. 2008 The global biogeography of semi-arid periodic vegetation patterns. Global Ecol Biogeogr 17, 715-723. (doi:10.1111/j.1466-8238.2008.00413.x).[26] Gintis, H. 2012 Clash of the Titans. Bioscience 62, 987-991. (doi:10.1525/bio.2012.62.11.8).[27] Christaller, W. 1966 Central places in southern Germany, Prentice-Hall.[28] Hudson, J.C. 1969 A location theory for rural settlement. Annals of the Association of American Geographers 59, 365-381.[29] Chisholm, M. 1973 Rural settlement and land use, Transaction Publishers.[30] Curtis, D.R. 2013 The emergence of concentrated settlements in medieval western Europe: explanatory frameworks in the historiography. Canadian Journal of History 48, 223-251. (doi:10.3138/cjh.48.2.223).[31] Lichter, D.T. & Johnson, K.M. 2006 Emerging rural settlement patterns and the geographic redistribution of America's new immigrants. Rural Sociology 71, 109-131. (doi:10.1526/003601106777789828).[32] Ellis, E.C. 2015 Ecology in an anthropogenic biosphere. Ecol Monogr 85, 287-331. (doi:10.1890/14-2274.1).[33] Freeman, J. & Anderies, J.M. 2015 The socioecology of hunter–gatherer territory size. Journal of Anthropological Archaeology 39, 110-123. (doi:10.1016/j.jaa.2015.03.002).[34] Hamilton, M.J., Milne, B.T., Walker, R.S. & Brown, J.H. 2007 Nonlinear scaling of space use in human hunter-gatherers. Proc Natl Acad Sci USA 104, 4765-4769. (doi:10.1073/pnas.0611197104).[35] Freeman, J. 2016 The socioecology of territory size and a "work-around" hypothesis for the adoption of farming. PLoS ONE 11, e0158743. (doi:10.1371/journal.pone.0158743).[36] Hamilton, M.J., Burger, O., DeLong, J.P., Walker, R.S., Moses, M.E. & Brown, J.H. 2009 Population stability, cooperation, and the invasibility of the human species. Proc Natl Acad Sci USA 106, 12255-12260. (doi:10.1073/pnas.0905708106).
17
454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497
[37] Marean, C.W. 2016 The transition to foraging for dense and predictable resources and its impact on the evolution of modern humans. Philosophical Transactions of the Royal Society B: Biological Sciences 371. (doi:10.1098/rstb.2015.0239).[38] Brown, J.L. 1964 The evolution of diversity in avian territorial systems. The Wilson Bulletin 76, 160-169.[39] Tian, G.J., Qiao, Z. & Zhang, Y.Q. 2012 The investigation of relationship between rural settlement density, size, spatial distribution and its geophysical parameters of China using Landsat TM images. Ecol Model 231, 25-36. (doi:10.1016/j.ecolmodel.2012.01.023).[40] Hu, Z., Wang, Y., Liu, Y., Long, H. & Peng, J. 2016 Spatio-Temporal Patterns of Urban-Rural Development and Transformation in East of the “Hu Huanyong Line”, China. ISPRS International Journal of Geo-Information 5, 24.[41] Wiegand, T. & Moloney, K.A. 2014 Handbook of spatial point-pattern analysis in ecology, Chapman and Hall.[42] Huston, M.A. 2012 Precipitation, soils, NPP, and biodiversity: resurrection of Albrecht's curve. Ecol Monogr 82, 277-296. (doi:10.1890/11-1927.1).[43] Liu, P., Liu, C., Deng, Y., Shen, X., Li, B. & Hu, Z. 2010 Landscape division of traditional settlement and effect elements of landscape gene in China. Acta Geographica Sinica 65, 1496-1506.[44] Dormann, C.F., McPherson, J.M., Araujo, M.B., Bivand, R., Bolliger, J., Carl, G., Davies, R.G., Hirzel, A., Jetz, W., Kissling, W.D., et al. 2007 Methods to account for spatial autocorrelation in the analysis of species distributional data: a review. Ecography 30, 609-628. (doi:DOI 10.1111/j.2007.0906-7590.05171.x).[45] R Core Team. 2016 R: a language and environment for statistical computing. Vienna, Austria., R Foundation for Statistical Computing.[46] Paradis, E., Claude, J. & Strimmer, K. 2004 APE: analyses of phylogenetics and evolution in R language. Bioinformatics 20, 289-290. (doi:10.1093/bioinformatics/btg412).[47] Lefcheck, J.S. 2016 PIECEWISESEM: Piecewise structural equation modelling in R for ecology, evolution, and systematics. Methods in Ecology and Evolution 7, 573-579. (doi:10.1111/2041-210x.12512).[48] Adams, E.S. 2001 Approaches to the study of territory size and shape. Annu Rev Ecol Syst 32, 277-303. (doi:10.1146/annurev.ecolsys.32.081501.114034).[49] Johnson, D.D.P., Kays, R., Blackwell, P.G. & Macdonald, D.W. 2002 Does the resource dispersion hypothesis explain group living? Trends Ecol Evol 17, 563-570. (doi:10.1016/s0169-5347(02)02619-8).[50] Enoksson, B. & Nilsson, S.G. 1983 Territory size and population density in relation to food supply in the nuthatch Sitta europaea (Aves). J Anim Ecol 52, 927-935. (doi:10.2307/4464).[51] Jedrzejewski, W., Schmidt, K., Theuerkauf, J., Jedrzejewska, B. & Kowalczyk, R. 2007 Territory size of wolves Canis lupus: linking local (Bialowieza Primeval Forest, Poland) and Holarctic-scale patterns. Ecography 30, 66-76. (doi:10.1111/j.0906-7590.2007.04826.x).[52] Callaway, R.M. 2007 Positive interactions and interdependence in plant communities, Springer, Dordrecht.[53] Holmgren, M. & Scheffer, M. 2010 Strong facilitation in mild environments: the stress gradient hypothesis revisited. Journal of Ecology 98, 1269-1275. (doi:10.1111/j.1365-2745.2010.01709.x).[54] Ebensperger, L.A., Villegas, Á., Abades, S. & Hayes, L.D. 2014 Mean ecological conditions modulate the effects of group living and communal rearing on offspring production and survival. Behav Ecol 25, 862-870. (doi:10.1093/beheco/aru061).
18
498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541
[55] Barrio, I.C., Hik, D.S., Bueno, C.G. & Cahill, J.F. 2013 Extending the stress-gradient hypothesis - is competition among animals less common in harsh environments? Oikos 122, 516-523. (doi:10.1111/j.1600-0706.2012.00355.x).[56] Kittle, A.M., Anderson, M., Avgar, T., Baker, J.A., Brown, G.S., Hagens, J., Iwachewski, E., Moffatt, S., Mosser, A., Patterson, B.R., et al. 2015 Wolves adapt territory size, not pack size to local habitat quality. J Anim Ecol 84, 1177-1186. (doi:10.1111/1365-2656.12366).[57] Stamps, J.A. 1988 Conspecific attraction and aggregation in territorial species. Am Nat 131, 329-347. (doi:10.1086/284793).[58] Dias, R.I., Kuhlmann, M., Lourenco, L.R. & Macedo, R.H. 2009 Territorial clustering in the blue-black grassquit: reproductive strategy in response to habitat and food requirements. Condor 111, 706-714. (doi:10.1525/cond.2009.090142).[59] Stamps, J.A. 1987 Conspecifics as cues to territory quality - a preference of juvenile lizards (Anolis aeneus) for previously used territories. Am Nat 129, 629-642. (doi:10.1086/284663).[60] Ward, M.P. & Schlossberg, S. 2004 Conspecific attraction and the conservation of territorial songbirds. Conserv Biol 18, 519-525. (doi:10.1111/j.1523-1739.2004.00494.x).[61] Port, M. & Johnstone, R.A. 2013 Facing the crowd: intruder pressure, within group competition, ‐and the resolution of conflicts over group membership. ‐ Ecology and evolution 3, 1209-1218. (doi:10.1002/ece3.533).[62] Odgaard, M.V., Bøcher, P.K., Dalgaard, T., Moeslund, J.E. & Svenning, J.-C. 2014 Human-driven topographic effects on the distribution of forest in a flat, lowland agricultural region. Journal of Geographical Sciences 24, 76-92. (doi:10.1007/s11442-014-1074-6).[63] Sandel, B. & Svenning, J.-C. 2013 Human impacts drive a global topographic signature in tree cover. Nature Communications 4. (doi:10.1038/ncomms3474).[64] Brewer, J., Gelfand, M., Jackson, J., MacDonald, I., Peregrine, P., Richerson, P., Turchin, P., Whitehouse, H. & Wilson, D. 2017 Grand challenges for the study of cultural evolution. Nature Ecology & Evolution 1, 0070.[65] Batistella, M., Robeson, S. & Moran, E.F. 2003 Settlement design, forest fragmentation, and landscape change in Rondônia, Amazônia. Photogrammetric Engineering & Remote Sensing 69, 805-812. (doi:10.14358/PERS.69.7.805).
19
542543544545546547548549550551552553554555556557558559560561562563564565566567568569570
Figure Legends
Figure 1. Distribution of the sampling plots and snapshots of typical rural settlement systems
from Google Earth images across mainland China. (a) The plots locations are represented by
squares (blue and red indicate presence and absence of large-scale under-dispersion of
settlement pattern, respectively); the dashed line is known as the ‘Hu Huanyong Line’ with
sparse human populations on the left side and dense populations on the right side. (b, c) The
red points represent the centroids of individual settlements in two typical rural landscapes as
examples. The high-resolution remotely sensed images were acquired from Google Earth (the
original source: DigitalGlobe).
Figure 2. Illustrations of settlement pattern features revealed by the spatial point pattern
analysis. The typical point maps in (a, c) present the two classes of pattern features detected
by the g(r) function (b, d). Small-scale over-dispersion is found ubiquitously, while large-
scale under-dispersion is found only in a portion of the settlement systems (e.g., present in the
map c whereas absent in the map a). The scale ranges of small-scale over-dispersion and
large-scale under-dispersion are calculated as the widths of the cyan and orange bars (scale
domains) in (b, d), as indicated by the g(r) values (red solid lines) falling below and above the
99% confidence intervals (dashed lines), respectively.
Figure 3. Effects of potential evapotranspiration (PET), elevation range (TOPO), soil fertility
(SOIL) and river density (RIVER) on the scale range of small-scale over-dispersion (a),
occurrence of large-scale under-dispersion (b), and scale range of large-scale under-dispersion
(c). Standardized coefficients with standard errors from the multiple regression models are
shown (***P<0.001, **P<0.01, *P<0.05, •P<0.1).
Figure 4. Synthesis of the direct and indirect effects of energy availability (measured by
potential evapotranspiration) and topographic relief as strong predictors on settlement spatial
pattern features characterised by (a) small-scale over-dispersion as a measure of territory size,
and (b) large-scale under-dispersion as a sign of territorial clustering. Blue: positive effects,
red: negative effects, solid: direct effects, dashed: indirect effects.
20
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597