Can political complexity evolve from small-scale cooperative herding groups?

Last week I was informed that my project proposal “From small-scale cooperative herding groups to nomadic empires – a cross-cultural approach (COMPLEXITY)” was funded through the ERC Consolidator Grant scheme.

Had to wait to announce it since it was not official until 12:00 17.03.2022 when ERC announces their press release with all the successful grants. The press release can be found here.

The overall aim of the Consolidator Grant is to “… support mid-career researchers and will help them consolidate their teams and conduct pioneering research on topics and with methods of their choosing” (from press release).

Central thesis

COMPLEXITY is situated at the intersection of anthropology and ecology and deals with the evolution of political complexity.

The prevalent view of the evolution of complex societies favours agriculture as the main factor.

How do we then explain the rise of nomadic empires?

One common explanation refers to conflict, and large-scale conflict with China has been presented as the central element in the rise of, for example, the Mongol Empire.

In 1242, Europe stood on the precipice of destruction. Based in Hungary and Serbia, the Mongol armies were poised for conquering the rest of Europe. Only the death of the Great Khan halted the Mongol advance, sparing Europe from the fate of an inevitable conquest. Twenty-five years after the withdrawal, the Mongol Empire reached its peak with the establishment of the Yuan dynasty – making it the largest land empire in history, stretching from the Sea of Japan to the Mediterranean Sea and the Carpathian Mountains.

Thus, pastoralists could only develop complex levels of organisation when facing strong agricultural neighbours.

But this cannot explain how pastoralists transitioned from small, kin-based groups to complex stratified societies.

COMPLEXITY’s central thesis is that before large-scale conflict is even possible, a level of within-group cooperation must be present.

Noteworthy, it is almost impossible for pastoralists to survive without cooperative labour investment and help from other households

By viewing cooperative herding groups as the building blocks of nomadic societies, COMPLEXITY aims to increase our understanding of the evolution of political complexity based on a new theoretical explanation of pastoral political organisation.

Structure

COMPLEXITY adds to state of the art through three steps.

While cooperative herding has been documented, previous studies have been based on single case studies.

The preliminary extent of cooperative herding groups.

Evidence is also fragmented, and little systematic attempts have been made to understand general patterns of pastoral cooperation.

The first step of COMPLEXITY is thus to cross-culturally analyse and document the prevalence of cooperative herding groups by using the existing ethnographic literature and a cross-cultural database

This will be used to select four field sites in Africa and Inner Asia: 

Study design where the overall starting point is to select two communities at two sites within each region. Arrows indicate levels of comparison undertaken in the project: between regions; between sites; and between communities. Coloured areas on the map indicate the already documented presence of herding groups while animal figures indicate the traditional Old World nomadic pastoral zones defined by the key cultural animal

Cooperation, performance and the rise of pastoral inequality

Understanding cross-cultural diversity and patterns in behaviour is a central goal of human behavioural ecology.

Nevertheless, the predominant view of cooperation is shaped by studies focusing on food sharing among foragers.

A conceptual overview of the domain, focus, problem, mechanisms and research load in evolutionary aspects of cooperation in anthropology. Food sharing has been a focus because it carries a cost for the giver: the giver must share parts of their food without knowing if the action will be reciprocated. Thus, sharing is a collective action dilemma, i.e., a situation where free-riders can thwart cooperation. Sharing labour is not riddled with the same dilemma: it is mutually beneficial and, thus, represent a coordination problem. Since individuals who share labour have common interests and share preferences, they always benefit from cooperation. Also referred to as mutualism, coordination has been argued to be a better representation for many situations of human cooperation. Nevertheless, they have been viewed as less interesting and trivial than collective action dilemmas: when everyone benefits from collective action, the cooperative solution should be obvious.

In contrast, less focus has been placed on cooperative production, the primary form of cooperation among pastoralists.

Consequently, COMPLEXITY’s second step is to use field studies to comparatively investigate to what degree pastoral cooperation is structured by evolutionary factors and investigate how cooperation affects pastoral performance.

The evolution of political complexity: from small-scale cooperative group to empires?

There is also a view that livestock, as the primary source of wealth, limits the development of inequalities, making pastoralism unable to support complex organisations.

However, the Gini coefficient for reindeer in Norway indicates that wealth in livestock is more unevenly distributed than for Norway in general (see this preprint https://doi.org/10.31235/osf.io/zv92t).   

Temporal trends in wealth inequality measured as reindeer numbers for (A) the Saami reindeer husbandry in Norway and (B) the Saami reindeer husbandry in the North and South (Fig 1.). The Gini coefficient ranges from 0 (perfect equality; everyone owns equally) to 1 (perfect inequality; one individual owns everything). Data for Gross Domestic Product (GDP) for Norway downloaded from Statistics of Norway (https://www.ssb.no/). See preprint https://doi.org/10.31235/osf.io/zv92t for details.

Since we cannot observe the history of nomadic empires, COMPLEXITY’ will model if, for example, livestock as wealth can generate inequalities resulting in hierarchical power structures.

The third step is thus to combine empirical data with Agent-Based Modelling, to investigate whether cooperative herding groups can be considered prototypes for more complex organisations.

The funding makes it possible to hire 2 postdocs and 2 Phds working alongside me in Tromsø!

So stay tuned for job openings!

Using Python to summarize articles

Earlier this year I got an article published in Acta Borealia.

The paper, The Sami cooperative herding group: the siida system from past to present, is open access.

I usually publish a short summary on this blog, but recently I’ve been learning to analyze text using Python so I thought I should try to leverage Python to help me summarize my own paper.

The result? Have a look (I’ve only removed citations and reorganized the sentences for flow):

Background

The Sami – both pastoralists and hunters – in Norway had a larger unit than the family, i.e., the siida.

History

Historically, it has been characterized as a relatively small group based on kinship.

The siida could refer to both the territory, its resources and the people that use it.

The core institutions are the baiki (household) and the siida (band).

Names of siidas were, in other words, local.

Moreover, it was informally led by a wealthy and skillful person whose authority was primarily related to herding.

One of these groups’ critical aspects is that they are dynamic: composition and size change according to the season, and members are free to join and leave groups as they see fit.

Results

Only two herders reported to have changed summer and winter siida since 2000.

Furthermore, while the siida continues to be family-based, leadership is becoming more formal.

Nevertheless, decision-making continues to be influenced by concerns of equality.

Code

The code is shown below. Lacks a bit in comments, but should work for documents. I’ve load the text used from a docx file.

Imports

import numpy as np
import os
import sys
import nltk
from nltk.corpus import stopwords
import re
import textacy # have installed spacy==3.0 and textacy==0.11
import textacy.preprocessing as tprep
import docx
import networkx as nx
from sklearn.feature_extraction.text import TfidfVectorizer
import nltk
import os
import warnings
warnings.filterwarnings("ignore")

stop_words = nltk.corpus.stopwords.words('english')
wtk = nltk.tokenize.RegexpTokenizer(r'\w+')
wnl = nltk.stem.wordnet.WordNetLemmatizer()

Loading file

doc = docx.Document('your_filename.docx')

Functions for processing and summarizing text

def extract_text_doc(doc):
    paras = [p.text for p in doc.paragraphs if p.text]
    revised_paras = [p for p in paras if len(p.split('.')) >1]
    text = " ".join(revised_paras)
    return text

def normalize_document(paper):
    '''
    Tokenize ++
    '''
    paper = paper.lower()
    paper_tokens = [token.strip() for token in wtk.tokenize(paper)]
    paper_tokens = [wnl.lemmatize(token) for token in paper_tokens if not token.isnumeric()]
    paper_tokens = [token for token in paper_tokens if len(token) >2]
    paper_tokens = [token for token in paper_tokens if token not in stop_words]

    doc = ' '.join(paper_tokens)
    return doc

def normalize(text):
    '''
    Normalizes text, string as input, returns normalized string
    '''
    text = tprep.normalize.hyphenated_words(text)
    text = tprep.normalize.quotation_marks(text)
    text = tprep.normalize.unicode(text)
    text = tprep.remove.accents(text)
    return text

def text_rank_summarizer(norm_sentences, original_sentences, num_sent):
    if len(sentences) < num_sent:
        num_sent -=1
    else:
        num_sent=num_sent

    tv_p = TfidfVectorizer(min_df=1, max_df=1, ngram_range=(1,1), use_idf=True)
    dt_matrix = tv_p.fit_transform(norm_sentences)
    dt_matrix = dt_matrix.toarray()

    vocab = tv_p.get_feature_names()
    td_matrix = dt_matrix.T

    similarity_matrix = np.matmul(dt_matrix, dt_matrix.T)


    similarity_graph = nx.from_numpy_array(similarity_matrix)

    scores = nx.pagerank(similarity_graph)

    ranked_sentences = sorted(((score,index) for index, score in scores.items()))

    top_sentence_indices = [ranked_sentences[index][1] for index in range(num_sent)]
    top_sentence_indices.sort()
    summary = "\n".join(np.array(original_sentences)[top_sentence_indices])
    return summary

Processing text

normalize_corpus = np.vectorize(normalize_document)
text = extract_text_doc(doc)
text1 = normalize(text)
sentences = nltk.sent_tokenize(text1)
norm_sentences = normalize_corpus(sentences)
summary = text_rank_summarizer(norm_sentences, sentences, 10)
print(summary)

Historically, it has been characterized as a relatively small group based on kinship.
Moreover, it was informally led by a wealthy and skilful person whose authority was primarily related to herding.
Only two herders reported to have changed summer and winter siida since 2000.
Furthermore, while the siida continues to be family-based, leadership is becoming more formal.
Nevertheless, decision-making continues to be influenced by concerns of equality.
One of these groups' critical aspects is that they are dynamic: composition and size changes according to the season, and members are free to join and leave groups as they see fit.
Lowie (1945) writes that the Sami – both pastoralists and hunters – in Norway had a larger unit than the family, i.e., the siida.
Names of siidas were, in other words, local.
The siida could refer to both the territory, its resources and the people that use it (see also Riseth 2000, 120).
The core institutions are the baiki (household) and the siida (band).

For more options and better ways of summarizing text, check out https://pypi.org/project/sumy/ with its many summarizer classes. For example, TextRankSummarizer is similar (but way better) than the approach taken here.

But similarly, it conceptualizes the relationship between sentences as a graph: each sentence is considered as vertex and each vertex is linked to the other vertex. But, rather than using PageRank from networkx for similarity, it uses Jaccard Similarity.

Collaborative foundations of herding: The formation of cooperative groups among Tibetan pastoralists

Just got a paper published in Journal of Arid Environments on cooperation among Tibetan pastoralists.

You can read the paper here, it’s open access.

Luckily, the paper got published on the same day as I got grant applications rejected from the Research Council of Norway.

Why bring that up? Because the paper and one of the grant applications have some obvious parallels.

The paper documents why cooperation is vital for Tibetan pastoralists: it increases control of herds, reduces individual household’s labour demand and increases the potential for economic diversification.

Tibetan herder bring a subset of the herd back to camp for the night. Photo (C) Marius Warg Næss

The grant application extends this aspect: it wants to comparatively investigate pastoral cooperation by taking as its starting point that cooperative herding groups are a necessary component of pastoral life.

The paradox is that while it is almost impossible for pastoral households to maintain production without cooperative labour investment and mutual help from other households, a comparative perspective of cooperative herding group formation is currently lacking.

Moreover, while cooperation is widely documented, it is treated in a somewhat ad hoc fashion. For example, it has been used as an explanation for why there is no relationship between household labour investment and pastoral production (see this paper).

Furthermore, herding groups have been described as fluid: changing composition from year to year and between seasons. Thus, they have been viewed as less important than more extensive and more political, groupings.

My personal point of view that despite the instability of herding units, they represent an essential building block of nomadic societies because they are concerned with daily cooperation.

More to the point, we know relatively little about them: what evidence we have represent a snapshot in time. We lack longitudinal data as well as data concerning how cooperation is changing as a consequence of changes in land tenure.

While getting grant rejections sucks, I still think this is a worthwhile project that I’ll pursue.

Anyone interested in collaborating on such a project is welcome to contact me.

Density or climate: Is that the question?

A recent paper argues that climate is more important than density in the reindeer husbandry in Norway. Using the same analysis, I find that reindeer density is essential: In high-density environments, average varit (1.5-year-old bucks) carcass weight is 8 kg lower, and calf carcass weight is 4 kg lower compared to low-density environments.

A recent paper, ‘Productivity beyond density: A critique of management models for reindeer pastoralism in Norway’, published in Pastoralism: Research, Policy and Practice sets out to investigate the validity of the premise that there is a strong relationship between density and carcass weights in the reindeer husbandry in Norway.

In short, the paper aims to challenge the official view of overstocking and reframe reindeer herding in terms of non-equilibrium ecology.

Their focus of attack is the Røros model which, according to the authors, hinges on

“… classic ecological equilibrium models where there is a clear unequivocal relationship between animal densities, production, and carcass weights”

p. 15

As such the article fits nicely in a growing trend: rather than investigating problems currently facing pastoralists, the main point is to establish systems as non-equilibrium, and thus all issues are assumed resolved, or at least externally caused (for an excellent example from the reindeer husbandry in Norway, check out ‘Conceptualising resilience in Norwegian Sámi reindeer pastoralism‘).

In another paper, some of the same authors have, for example, argued that reindeer herding in Norway is better characterised as a non-equilibrium system

“…where herbivore populations fluctuate randomly according to external influences, [and] the concepts of carrying capacity and overgrazing have no discernible meaning”.

Misreading the Arctic landscape: A political ecology of reindeer, carrying capacities, and overstocking in Finnmark, Norway’, p. 223

Productivity beyond density’ goes at least further in attempting to quantify the relative importance of non-equilibrium factors (such as climate) and equilibrium factors (such as density).

While the paper is well-written and exciting, I find it a bit strange that in the only quantitative analysis they present the sole focus is on statistically significant effects of precipitation and temperature for the carcass weights of reindeer:

Source: Table 1 in publication.

While the analysis shows indeed that climate factors (precipitation: all the daily observation in the stated period and growing degree days [GDD]) are significant, the discussion of the table completely fails to address two critical factors:

They never discuss whether the variables measuring climate is correlated or not (as they are monthly based, it wouldn’t be a huge surprise if they are).

High or even moderate, collinearity is problematic when effects are weak (as the climate effect sizes indicate). If collinearity is ignored, it is possible to end up with a statistical analysis where nothing is significant, but were dropping one predictor may make others significant, or even change the sign of estimated parameters.

The point is technical; it would be interesting to see how these potential problems were accounted for.

Concerning effect size, the most substantial effect by far is that of density: -0.16 kg for calves and -0.32 kg for varit (1.5-year-old bucks).

In effect, this has a considerable impact on the carcass weights in high-density vs low-density environments

Keep in mind that they do not provide information concerning variable transformation, so I take it for granted that the intercept represents average carcass weights when every other variable is at zero. I also take for granted that all variables are continuous. Moreover, not all of the data was in the supplemental material so I couldn’t re-analyse the data properly.

In short, at density 0 average calf carcass weight is 18.52 kg and average varit carcass weight is 25.34 kg.

The paper does not indicate the range of density utilised in this analysis, but Fig. 5 presents the range for mainland districts (which are the same districts used in table) to be from 0 to 25.

Disregarding the climate parameters (since there are no interactions and the range of the climate parameters are not presented) density has a significant effect in high-density environments:

Calves: 18.52 – 0.16 X 25 = 14.52 kg

Varit: 25.34 – 0.32 X 25 = 17.34 kg

In short, the model in Table 1 predicts that the difference in varit carcass weight between a low-density environment and a high-density environment is 8 kg. For calf carcass weight, the model predicts a difference of 4 kg.

While I fully agree with the authors that an over-emphasis on density and herd size is too simplistic when modelling pastoral production, it is bizarre that the above is not communicated at all.

Part of the problem, I think, stems from the simplified representation of non-equilibrium ecology. Concerning Africa and Asia, for example, they write “…a wholesale paradigm shift from equilibrium to non-equilibrium modelling took place from the early 1990s” (p. 9).

This is in fact, only partially true.

In the chapter ‘Why are there so many animals? Cattle population dynamics in the communal areas of Zimbabwe’, Ian Scoones, for example, investigated factors affecting herd growth among Zimbabwean pastoralists.

He mainly focused on periodic events such as droughts (a density-independent factor) and more persistent factors such as herd size (a density-dependent factor)

In other words, he investigated the degree to which density-dependent and density-independent factors explained herd size fluctuations (data for 60 years).

In short, he found:

  • In years with high precipitation, the population of cattle approaches a ceiling, which he terms the carrying capacity. As density increases, the birth rate drops, and mortality rates increases (although they never reach equilibrium and the cattle population never reaches its theoretical maximum).
  • The cattle population never reaches a maximum because stochastic events such as droughts occur and kill off large parts. Noteworthy, the number of animals killed by these events was more substantial than what can be predicted from density-dependent factors alone.

In the long term, it thus looks like non-equilibrium factors have the most significant impacts on cattle populations. Still, equilibrium factors are essential in years without stochastic climatic effects and when the population is high.

Scoones’ investigation show what seems now to be forgotten:

It is unlikely that any system is characterised by either equilibrium or non-equilibrium factors alone, but rather that they both operate on a continuum.

This supports the predominant ecological perspective that at high population sizes, herbivores are sensitive to a combination of density-dependent and -independent factors, which has been shown for reindeer in Norway.

As I argued in the paper ‘Climate Change, Risk Management and the End of Nomadic Pastoralism’.

“To understand the effects of climate change on nomadic pastoralists, it is thus necessary to move beyond the simplistic dichotomy of characterising pastoral system as equilibrial (density dependence: livestock and pastures are regulated by grazing pressure) or non-equilibrial (density independence: livestock and pastures are limited by external factors such as climate) and look at the interplay between density dependent and density independent factors”

p. 131

Cultural group selection and the evolution of reindeer herding in Norway

The debate about reindeer husbandry in Norway is characterised by two contrasting views.

On one hand is the prevailing view of overstocking and rangeland degradation.

On the other hand, is the view that overstocking and overuse represents a misreading of the Arctic landscape that perpetuates a dominant crisis narrative that functions as “… an enduring ‘social fact’, whose narrative reality is in large part decoupled from its supposed scientific basis” (Benjaminsen et al. 2015:228).

While the overstocking perspective is based on a presumed ‘Tragedy of the Commons’, the other perspective argue that reindeer herding is characterised as a non-equilibrium system

“…where herbivore populations fluctuate randomly according to external influences, [and] the concepts of carrying capacity and overgrazing have no discernible meaning” (ibid.:223).

In my new paper, Cultural Group Selection and the Evolution of Reindeer Herding in Norway, I argue differently.

Through a comparative historical analysis, I argue that herding is better viewed as an assurance game with two different strategies for minimising risk:

  1. maximising quantity (i.e., increasing livestock numbers or herd size)
  2. maximising livestock quality (i.e., increasing livestock body mass)

I demonstrate that intra-group competition has led to the
adoption of (1) in the Northern parts of Norway, while inter-group competition has led to the adoption of (2) in the Southern parts.

Read the full paper here.

New research paper about cooperation in groups of Saami reindeer herders

The Tangled Woof of Fact

People rely on one another in fundamental ways, but cooperation in groups can be fragile. Every day, we face tensions between acting in a socially responsible manner and following our own self-interest. These situations are called social dilemmas and they come in varying shades of subtlety, from littering and eBay to overpopulation and climate change. Overcoming these dilemmas can make all the difference, especially for marginalised groups such as pastoralists – people who make their living from herding animals.

Pastoralists use about a quarter of the world’s land for grazing their herds. Nowadays, all over the world, governments are privatising many of their pastures, and so herders must work together in increasingly fragmented places.

We wanted to learn how groups of Saami reindeer herders living in Norway’s Arctic Circle worked together. Our study, just published in the journal Human Ecology, found that cooperation pivoted around the ‘siida’: a…

View original post 420 more words

Tibetan lives: Hunting

I’ve just got a paper accepted in Land Use Policy about nomadic pastoralists in Tibet and hunting. As we all know, space is limited in scientific journals, so here is additional text as well as pictures. Continue reading “Tibetan lives: Hunting”

Reindeer Husbandry in a Globalizing North – resilience, adaptations and pathways for Actions (ReiGN)

It’s the time of the year when we eagerly await the results from the year’s (many) research proposals.

Continue reading “Reindeer Husbandry in a Globalizing North – resilience, adaptations and pathways for Actions (ReiGN)”

Predatory or prey – the rise of nomadic empires

In 1227 Genghis Khan died leaving behind a legacy of conquest and the largest land empire in history, only fully realized by his Grandson Kubhlai Khan with the establishment of the Yuan Dynasty in 1267 (Chaliand 2004). Continue reading “Predatory or prey – the rise of nomadic empires”