Pasture, grazing, and meat production in Kazakhstan

Hankerson, Brett R. Doctoral Thesis Pasture, grazing, and meat production in Kazakhstan Suggested Citation: Hankerson, Brett R. (2025) : Pasture, grazing, and meat production in Kazakhstan, Humboldt-Universität zu Berlin, Berlin, https://doi.org/10.18452/34649 , http://edoc.hu-berlin.de/18452/35289 This Version is available at: https://hdl.handle.net/10419/328055 Standard-Nutzungsbedingungen: Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen Zwecken und zum Privatgebrauch gespeichert und kopiert werden. Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich machen, vertreiben oder anderweitig nutzen. Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen (insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten, gelten abweichend von diesen Nutzungsbedingungen die in der dort genannten Lizenz gewährten Nutzungsrechte. Terms of use: Documents in EconStor may be saved and copied for your personal and scholarly purposes. You are not to copy documents for public or commercial purposes, to exhibit the documents publicly, to make them publicly available on the internet, or to distribute or otherwise use the documents in public. If the documents have been made available under an Open Content Licence (especially Creative Commons Licences), you may exercise further usage rights as specified in the indicated licence. https://creativecommons.org/licenses/by/4.0/ Humboldt-Universität zu Berlin – Geographisches Institut Pasture, grazing, and meat production in Kazakhstan – current state and future opportunities DISSERTATION zur Erlangung des akademischen Grades Doctor rerum naturalium (Dr. rer. nat.) im Fach Geographie eingerichtet an der Mathematisch-Naturwissenschaftlichen Fakultät der Humboldt-Universität zu Berlin von Brett Robert Hankerson, M.Sc. Präsidentin der Humboldt-Universität zu Berlin Prof. Dr. Julia von Blumenthal Dekan der Mathematisch-Naturwissenschaftlichen Fakultät Prof. Dr. Emil List-Kratochvil Gutachter: Prof. Dr. Patrick Hostert Prof. Dr. Dr. h.c. Norbert Hölzel Prof. Dr. Friedhelm Taube Eingereicht am: 25. März 2025 Datum der Promotion: 8. September 2025 ii iii Acknowledgments I have had the pleasure and the privilege of working with fantastic people at the Leibniz Institute of Agricultural Development in Transition Economies (IAMO). My supervisor, Daniel Müller, gave me the opportunity to check all the boxes of my graduate research goals: study the geography of agriculture using remote sensing in a former eastern bloc country at a top European university. Daniel provided me with the tools I needed to succeed, and the freedom to develop and adapt my research to the most relevant and interesting topics. I am forever grateful for the many field trips that allowed me to dive deep into Kazakh agriculture and understand on the ground level the questions I was trying to answer with satellites. Thank you, Daniel, for sticking with me through the lean years, when it was not at all clear if I would ever finish. Alfons Balmann, I thank you for the opportunities I received through my scholarship at IAMO and the freedom you granted me in pursuing my research interests, it was a great pleasure to be a part of the Department of Structural Change. Florian Schierhorn, often my travel companion and co-conspirator in Kazakh agricultural research, thank you for your many contributions and thoughts that helped shape both my research questions and their answers. Similarly, Sasha Prishchepov, thank you for helping me in my early days, honing my scientific focus and enhancing our field trips to Kazakhstan with your local knowledge and contacts. In Kazakhstan, I want to thank Dauren Oshakbaev, often the organizer of our activities, research partner, translator, and tour guide all rolled into one outstanding colleague. Also from Kazakhstan, but who quickly became a fellow student and fast friend, I want to thank Andrey Dara, first for your help and enthusiasm on our field trips, and later for your friendship and collaboration as we developed our dissertations around very similar topics. Fellow coauthors: I want to thank Changxing Dong, without whose coding expertise I wouldn’t have been able to develop the final model for our first paper. Also from that paper, I want to thank Christina Eisfelder for doing all the hard work in determining grassland productivity in Kazakhstan. From my second paper, I would’ve been lost without the natural processes knowledge of Tobias Kümmerle and the local grassland expertise of Johannes Kamp, thank you. I also want to thank my first student assistant, Alina Drokina, whose tireless efforts digitizing handwritten statistics and compiling annual bulletins allowed the breadth and depth of Kazakh data utilized in my research. Also instrumental in data collection, I want to thank Max Hofmann for your expertise in R and similar research interests, helping to keep me invested during the lean years. I managed to keep the IT department pretty busy with my numerous and occasionally impudent requests, thank you especially to Tim Illner for always keeping me connected and updated. My time in Halle was a wonderful period of my life. Besides the incredible travel opportunities that my research afforded me, I was also blessed with the opportunity to travel to my new friends’ home countries. I will never forget those trips, and those I took them with. It wasn’t all just traveling, either: in Halle lifelong iv friends were made during weekend grill-outs, game nights, movie nights, Super Bowl parties, Thanksgiving dinners, and all manner of social get-togethers. If I begin naming everyone now, I won’t know where to stop, and so I’m going to do a very annoying thing and say: you know who you are, thank you! I do need to name Lena Kuhn, Vasyl Kvartiuk, and Boris Rajković, exceptional friends who have graciously lent me their couches while I finished this dissertation. The most recent acknowledgments belong to Patrick Hostert, Norbert Hölzel, and Friedhelm Taube, who have kindly agreed to review this dissertation, thank you. v Abstract The dramatic increase of demand meat and animal products has placed a strain on land resources and created a livestock sector that is responsible for one fifth of all anthropogenic greenhouse gas emissions and is associated with desertification, deforestation, water contamination, and the loss of wild grazers, biodiversity, and ecosystem services. However, quantifying this strain on land resources remains difficult, as livestock are mobile and grazing is a spectrum. This dissertation focuses on a country with a long history of pastoralism that has undergone dramatic changes in the last century. Kazakhstan has one of the world’s largest grassland areas and supports a large livestock population that has fluctuated greatly from early transhumant nomadism, to sedentary pasturing of the Soviet Era, to a partial return of migratory pasturing post-Soviet. Livestock numbers peaked near the end of the Soviet period, plummeted in the 1990s, but have seen steady increase since. Ample evidence of overgrazing during the Soviet Era exists, however with livestock movement severely restricted, many distant pastures fell out of use, and the actual carrying capacity of Kazakh grasslands is poorly understood. With the current expansion of the livestock industry— especially beef—it is crucial that (re)utilization of Kazakhstan’s immense grassland reserves is sustainable. A key factor to maintain biodiversity and ecological functions on grasslands is recurrent fire events. Grazing is an important determinant of fire regimes, but the relative magnitude in comparison to climatic factors is poorly understood. The historical context of livestock in Kazakhstan provides an excellent setting to study the effect of grazing on fire occurrence over time. In the first part of this dissertation, I present a spatial model developed to assess the extent and the intensity of grazing demand by livestock. In Kazakhstan 2015, about half of available pasture was utilized, but most at low intensities. A 31% increase in beef production could be possible without improved infrastructure or intensification practices. In the second part, I correlated the annual change in grazing intensity of the grasslands utilized by livestock with annual burned area over the period 2001-2019, accounting for changes in temperature, precipitation, growing degree days, and relative humidity. There was a clear and negative relationship between grazing demand and burned area. Grazing demand also had more explanatory power in burned area than any of the climate variables tested. In the third part, I explored the potential of horsemeat as an alternative to beef. Nutritionally, horsemeat is the best red meat in almost every way, with a similar taste and texture to beef, and compares favorably to the major white meats (pork and chicken). Additionally, raising horses instead of cattle for meat production would drastically reduce the amount of methane released per kilogram of meat produced. However, horses require more land area, and a full conversion would likely coincide with a reduction in meat consumption. In summation, Kazakhstan has a large potential for pasture-based livestock expansion, a coordinated expansion of grazing livestock could be an excellent strategy for mitigating fire risk, and the most appropriate livestock species to fulfill both roles are horses, due to their mobility, hardiness, meat characteristics, and environmental footprint. vi Zusammenfassung Der dramatische Anstieg der Nachfrage nach Fleisch und tierischen Produkten hat den Druck auf die Landressourcen verstärkt und dazu beigetragen, dass die Tierproduktion für ein Fünftel aller anthropogenen Treibhausgasemissionen verantwortlich ist und zu Wüstenbildung, Abholzung, Wasserverschmutzung, und dem Verlust von Wildweidetieren, Artenvielfalt und Ökosystemdienstleistungen beigetragen hat. Allerdings ist es weiterhin schwierig, den Fußabdruck der Tierproduktion zu quantifizieren, da Weidetiere meist mobil sind und in unterschiedlicher Intensität beweiden. Diese Dissertation konzentriert sich auf Kasachstan, ein Land mit einer langen Geschichte des Pastoralismus, das im letzten Jahrhundert dramatische Veränderungen durchgemacht hat. Kasachstan verfügt über eine der größten Grünlandflächen der Welt und beherbergt einen großen Viehbestand, der sich vom frühen Transhumant-Nomadentum über die sesshafte Weidehaltung in der Sowjetzeit bis hin zu einer teilweisen Rückkehr der wandernden Weidewirtschaft nach der Sowjetzeit wandelte. Der Viehbestand erreichte gegen Ende der Sowjetzeit seinen Höhepunkt, ging in den 1990er Jahre stark zurück und stieg seitdem wieder an. Es gibt zahlreiche Beweise für Überweidung während der Sowjetzeit, aber da die Bewegung des Viehs stark eingeschränkt war, wurden weiter entfernte Weiden nicht genutzt. Außerdem ist die tatsächliche Tragfähigkeit des kasachischen Graslandes kaum bekannt. Angesichts der derzeitigen Expansion der Viehwirtschaft in Kasachstan – insbesondere des Rindfleischsektors – ist eine nachhaltige (Wieder-)Nutzung der immensen Grünlandreserven Kasachstans von entscheidender Bedeutung. Ein Schlüsselfaktor für die Erhaltung der Artenvielfalt und der ökologischen Funktionen von Grünland sind wiederkehrende Brände. Die Beweidung ist eine wichtige Determinante des Brandregimes, aber das relative Ausmaß im Vergleich zu klimatischen Faktoren ist nur unzureichend verstanden. Der historische Kontext der Viehwirtschaft in Kasachstan ermöglichen es die Auswirkungen der Beweidung auf das Auftreten von Bränden im Laufe der Zeit zu untersuchen. Im ersten Teil dieser Dissertation stelle ich ein räumliches Modell vor, das entwickelt wurde, um das Ausmaß und die Intensität der Weidenachfrage durch das Vieh zu erfassen. In Kasachstan wurde 2015 etwa die Hälfte der verfügbaren Weideflächen genutzt, meist aber mit geringer Intensität. Eine 31% Steigerung der Rindfleischproduktion wäre ohne verbesserte Infrastruktur oder Intensivierungspraktiken möglich. Im zweiten Teil korreliere ich die jährliche Veränderung der Weideintensität der von Tieren genutzten Grünlandflächen im Zeitraum 2001-2019 mit der jährlichen verbrannten Fläche unter Berücksichtigung von Änderungen in Temperatur, Niederschlag, Wachstumsgradtage und relativer Luftfeuchtigkeit. Es bestand ein klarer und negativer Zusammenhang zwischen der Weidenachfrage und der verbrannten Fläche. Ich stellte außerdem fest, dass die Nachfrage nach Weideflächen eine größere Erklärungskraft für das Ausmaß der verbrannten Fläche hat als jede der getesteten Klimavariablen. Im dritten Teil untersuchte ich das Potenzial von Pferdefleisch als Alternative zu Rindfleisch untersucht. Aus ernährungsphysiologischer Sicht ist Pferdefleisch in fast jeder Hinsicht das beste rote Fleisch, es hat einen ähnlichen Geschmack und eine vii ähnliche Textur wie Rindfleisch und schneidet im Vergleich zu den gängigen weißen Fleischsorten (Schwein und Huhn) gut ab. Darüber hinaus würde die Haltung von Pferden anstelle von Rindern zur Fleischproduktion die Menge an Methan, die pro Kilogramm produziertem Fleisch freigesetzt wird, drastisch reduzieren. Pferde benötigen jedoch mehr Landfläche, und eine vollständige Umstellung würde wahrscheinlich mit einer Reduzierung des Fleischkonsums einhergehen. Zusammenfassend lässt sich sagen, dass Kasachstan über ein großes Potenzial für die Ausweitung der Weideviehhaltung verfügt. Eine koordinierte Ausweitung der Weideviehhaltung könnte eine hervorragende Strategie zur Minderung des Brandrisikos sein. Die am besten geeignete Nutztierart zur Erfüllung dieser beiden Aufgaben sind Pferde aufgrund ihrer hohen Mobilität und Widerstandsfähigkeit, der ausgezeichneten Fleischeigenschaften und ihres geringen ökologischen Fußabdrucks. viii Contents ACKNOWLEDGMENTS .......................................................................................................................................................... III ABSTRACT .......................................................................................................................................................................... V ZUSAMMENFASSUNG ........................................................................................................................................................... VI CONTENTS ....................................................................................................................................................................... VIII LIST OF FIGURES .................................................................................................................................................................. X LIST OF TABLES .................................................................................................................................................................... X LIST OF SUPPLEMENTARY EQUATIONS ..................................................................................................................................... XI LIST OF SUPPLEMENTARY FIGURES .......................................................................................................................................... XI LIST OF SUPPLEMENTARY TABLES ........................................................................................................................................... XI CHAPTER 1: INTRODUCTION ....................................................................................................................................... 1 1.1 SCIENTIFIC BACKGROUND ................................................................................................................................................ 2 1.1.1 Grasslands and pasture .................................................................................................................................... 2 1.1.2 Livestock ........................................................................................................................................................... 3 1.1.3 Study region ...................................................................................................................................................... 5 1.1.4 Livestock in Kazakhstan .................................................................................................................................... 7 1.1.5 Grazing and grassland fires .............................................................................................................................. 8 1.2 CONCEPTUAL FRAMEWORK ............................................................................................................................................. 8 1.2.1 Research questions and objectives ................................................................................................................... 9 1.2.1.1 Question I: Where, and to what extent, could increases in Kazakh livestock production take place?......................... 9 1.2.1.2 Question II: To what extent are livestock in Kazakhstan actors in fire regimes? ......................................................... 9 1.2.1.3 Question III: is there a path to sustainable livestock development in Kazakhstan? ................................................... 10 1.3 STRUCTURE OF THE THESIS ............................................................................................................................................ 10 CHAPTER 2: MODELING THE SPATIAL DISTRIBUTION OF GRAZING INTENSITY IN KAZAKHSTAN ................................ 11 ABSTRACT ....................................................................................................................................................................... 12 2.1 INTRODUCTION ........................................................................................................................................................... 13 2.2 MATERIALS AND METHODS ........................................................................................................................................... 15 2.2.1 Study area ....................................................................................................................................................... 15 2.2.2 Kazakh farm structure .................................................................................................................................... 16 2.2.3 Model development ........................................................................................................................................ 16 2.2.4 Grazing supply ................................................................................................................................................ 18 2.2.5 Grazing demand ............................................................................................................................................. 19 2.2.6 Production potentials of meat and milk ......................................................................................................... 22 2.3 RESULTS .................................................................................................................................................................... 23 2.3.1 Grazing gap and demand distribution ............................................................................................................ 23 2.3.2 Off-take rate ................................................................................................................................................... 24 2.3.3 Grazing distances and pasture extent ............................................................................................................ 25 2.3.4 Production potentials of meat and milk ......................................................................................................... 27 2.4 DISCUSSION ............................................................................................................................................................... 29 2.4.1 Comparison to global products ....................................................................................................................... 29 2.4.2 Validation ....................................................................................................................................................... 31 2.4.3 Targeting areas for livestock expansion ......................................................................................................... 31 2.4.4 Notes on model inputs and assumptions ........................................................................................................ 32 2.4.5 Livestock productivity on private farms .......................................................................................................... 34 2.5 CONCLUSION .............................................................................................................................................................. 34 2.6 ACKNOWLEDGMENTS ................................................................................................................................................... 35 2.7 SUPPLEMENTARY MATERIAL .......................................................................................................................................... 37 CHAPTER 3: CHANGES IN LIVESTOCK SYSTEMS EXPLAIN POST-SOVIET FIRE TRENDS ON THE EURASIAN STEPPE BETTER THAN CLIMATE ............................................................................................................................................. 45 ABSTRACT ....................................................................................................................................................................... 46 3.1 INTRODUCTION ........................................................................................................................................................... 47 3.2 MATERIALS AND METHODS ........................................................................................................................................... 50 3.2.1 Study area ....................................................................................................................................................... 50 3.2.2 Livestock grazing demand .............................................................................................................................. 50 3.2.3 Climate indicators ........................................................................................................................................... 51 3 considered grazed (Sayre et al., 2017). Grassland is considered a land cover, while pastureand grazing land are land uses. Rangeland is typically considered a land cover where grazing could occur, but is also used as a term for grazed native grasslands to differentiate from sown pastureland. Unlike grassand pastureland, rangeand grazing land can encompass non-grasslands—shrubland, wetland, woodland, tundra, and semidesert (Allen et al., 2011). Regardless of the terminology used, determining the extent and intensity of the utilization of land by livestock is difficult (Ramankutty et al., 2008). 1.1.2 Livestock While defining and delineating pasture is difficult, keeping track of the number of animals utilizing pasture is relatively easier (Figure 1-1). Cattle, goats, and, on smaller scales, buffalo and camels, have all seen steady increases in numbers over the last 60 years. Sheep numbers have fluctuated, but are now higher than ever. Of the grazing livestock included in FAO statistics, only horses and horse crosses (mules and hinnies) have decreased in number since 1961. This can mostly be attributed to the decrease in demand for draft animals (Pearson, 1998). Figure 1-1. Global population of livestock species that obtain energy primarily through forage and fodder, 1961-2023 (FAO, 2025). (a) The three main grazing species. (b) Minor grazing livestock species. Other camelids include llamas and alpacas. Note the scale. As the global human population has increased, the distribution of wealth has shifted, with some developing countries becoming more affluent and demanding a larger proportion of meat in their diets (FAO, 2023b). This has driven the increase in meat production over the last 60 years (Figure 1-2). In the 1960s and 1970s, beef and pork were produced at nearly equal quantities at the global scale. Beef production increases began to tail off in the 1980s, with pork taking the clear lead in production quantity until the mid-2010s, when African Swine Fever devastated populations in Asia and Europe (Penrith, 2020). Over the last 60 years, chicken has seen the most dramatic upswing in production, and recently overtook pork as the largest meat production sector in the world. Duck, turkey, and goose are often combined with chicken as the 4 poultry sector. Frequently combined as closely related caprines, sheep and goat often co-occur. Cattle and buffalo are closely related bovines that can also be combined, though they do not often co-occur. Though very important in the few regions where they are raised for meat, equids and camelids (mainly horses and camels, respectively) are not major players at the global scale. Though not expressly defined, game meat likely includes a high proportion of grazing animals, as many deer species are popular hunting targets and are even farmed as a specialty meat in many regions (Serrano et al., 2019). Figure 1-2. Global production of meat, 1961-2023 (FAO, 2025). (a) The three main sources of meat. (b) Secondary sources of meat. (c) Minor sources of meat. Other meat includes rodent and snail. Equid includes horse, ass, and mule. Camelid includes camel, llama, and alpaca. Other birds include but are not limited to parrots, pheasants, and pigeons. Note the scale. Despite falling to a distant third, steady growth in the beef sector has kept beef production at the forefront of sustainability awareness. In a multitude of metrics, beef is the most inefficient of the major meats: poultry, pork, beef, and mutton/goatmeat. Cattle are much less efficient than poultry or pigs at turning 5 feed energy into meat (Hou et al., 2016), and account for three fourths of all non-CO2 greenhouse gas emissions from livestock production, mainly due to methane from their ruminant digestive system (Dunkley & Dunkley, 2013; Herrero, Havlík, et al., 2013). However, cattle and other grazers have one major advantage over poultry and pigs: they obtain the majority of their energy from forage, which can be grown on land that is otherwise unsuitable for agriculture. Thus, grazing livestock provide human-edible food from land otherwise incapable of being a food or energy source (van Zanten et al., 2018). 1.1.3 Study region Stretching 8000 km from central Europe to northeast China, the Eurasian steppe is a vast grassland biome that is vital to the agricultural economies of the countries it spans. Dryland agriculture supports primarily small grains and grazing livestock. The steppe has been home to transhumant pastoralism for millennia, with significant and established herds of sheep, goats, cattle, horses, camels, and yaks. Wild grazers are also a major feature, with important populations of Przewalski’s horse (Equus ferus przewalskii), wild ass (Equus hemionus), saiga antelope (Saiga tatarica), goitered gazelle (Gazella subgutturosa), Mongolian gazelle (Procapra gutturosa), wild camel (Camelus bactrianus), and mouflon (Ovis orientalis) (Wesche et al., 2016). Many ecoregions subdivide the Eurasian steppe, with one of the largest falling mostly within the borders of Kazakhstan and, predictably, is called the Kazakh steppe, where several factors make it a very interesting region to study. First, the Kazakh steppe is huge and largely uninterrupted, covering 808,000 km2, which places it among three largest temperate grassland ecoregions in the world along with the Mongolian-Manchurian grassland (889,000 km2) and the Pontic steppe (997,000 km2)—both also members of the Eurasian steppe (Dinerstein et al., 2017). Second, in recent history livestock populations within Kazakhstan have fluctuated greatly. While part of the Soviet Union, Kazakhstan underwent substantial landuse conversion to crop cultivation, and livestock production became largely sedentary. During this period livestock numbers skyrocketed, and overgrazing caused degradation and desertification of grasslands (Robinson et al., 2003). The breakup of the Soviet Union in 1991 precipitated a collapse in the Kazakh livestock sector, with livestock numbers decreasing by 70% within a decade (KazStat, 2019). There has been steady growth in livestock numbers since then, providing a natural experiment of huge swings in livestock numbers and dynamic shifts in grazing practices over a relatively short time period. Given the substantial land resources, and the governmental focus on livestock development (The World Bank, 2020), it is an ideal time to identify areas of interest and gaps in the knowledge record. The eastern and southeastern border of Kazakhstan runs along the edge of the Altai (at the convergence of China, Kazakhstan, Mongolia, and Russia) and Tien Shan (running through Kyrgyzstan and into China) mountain ranges (Figure 1-3). Ecosystems along these borders differ greatly from the majority of Kazakhstan, receiving the highest annual precipitation amounts that support forests and alpine meadows (Figure 1-4). In the south, the climate quickly turns dry traveling west. The Chu, Syr Darya, and Amu Darya rivers support extensive irrigated agriculture producing a variety of fruits, vegetables, and nuts. However, 6 while Almaty and Shymkent receive upwards of 600 mm/year precipitation, nearby Taraz and Turkistan receive only 350 mm/year and 200 mm/year, respectively, and the montane climate quickly transitions to semi-desert and eventually desert to the west and southwest, with less than 100 mm/year in the Kyzylkum Desert, located between the Amu Darya and Syr Darya. Moving north, precipitation increases gradually across the country (excepting the aforementioned eastern and southeastern border) to over 350 mm/year in Petropavl (Climate Research Unit, 2019). Northcentral Kazakhstan is home to conventional rainfed agriculture, predominately wheat and fodder crops. Livestock are found throughout the country (FAO classifies nearly 70% of the country as “permanent meadows and pastures”) (FAO, 2024), with the highest concentration in the fertile foothills of the Tien Shan. Cattle, sheep, goats, and horses are all ubiquitous and grown for meat, dairy, and other products. Camels are also an important food source in the drier regions of the south and southwest (KazStat, 2019). Figure 1-3. Topography of Kazakhstan in Central Asia with major water features and cities with over 100,000 residents (European Space Agency & Sinergise, 2021; GADM, 2022; Geofabrik, 2024; KazStat, 2011). 7 Figure 1-4. Average annual precipitation of Kazakhstan (de Pauw, 2008; GADM, 2022). 1.1.4 Livestock in Kazakhstan While cattle, sheep, goats, and horses are found in nearly every district of Kazakhstan, and often share pasture, there are distinct niches that the different species occupy at more local scales. Cattle are the breadwinners that receive the choicest grass, either near human settlements or shepherded to seasonal pastures. Cattle almost exclusively graze on grass, but are relatively unconcerned about the quality of grass. They use their tongues to rip easily accessibly grasses, often leaving behind short, young, more nutritious grass. Sheep and goats are not as mobile as cattle, but are sure-footed and occupy more marginal pastures, either degraded lands surrounding settlements or more montane pastures inaccessible to cattle. Sheep are picky grazers in the sense that they seek out highly nutritious grasses and can nibble them to the roots. When an area is overgrazed by sheep this can be devastating, but in a healthy rotation sheep are able to utilize quality grass that cattle leave behind. Goats are the least picky grazers, willing to munch on just about any type of plant. Thus, they can be destructive when left in a private yard or garden, but in a pasture, goats are able to utilize foliage that is snubbed by the other grazers. Horses are the most mobile and often range great distances from their point of origin, accessing a variety of pastures beyond the range of cattle, sheep, and goats. Like sheep, horses are picky grazers that seek high-quality grasses and can chew grasses down to their roots. Unlike sheep, horses are highly mobile, so unless they are confined overgrazing is less of a threat (Goodwin, 2007; Koshkina et al., 2022; Osoro et al., 2015). The different species also occupy distinct economical niches. Products from cattle are considered the most valuable, both beef and dairy, and receive the majority of governmental stimuli (Robinson et al., 2021). Sheep meanwhile are physically smaller, relatively cheaper, and are the most numerous of Kazakhstan’s 8 livestock. Lamb is a common meat, and mutton is also found from the rural villages to the city supermarkets. Sheep milk is drunk in the countryside and their cheese is popular everywhere. Sheep also have the added value of wool production. The main economic benefit of goats is that they can produce human food out of plants eschewed by other grazers. Goatmeat is not particularly prized, though it can be found in most meat markets. Goat milk and especially goat cheese, however, is quite popular. Horses are treated somewhat like goats in the economic sense, they are able to turn otherwise inaccessible biomass into human food (Taylor et al., 2020). However, horses have the added benefit of being a culturally significant and historically traditional food source (Malacarne et al., 2002). Mare milk is fermented into the traditional drink kumys, and horsemeat is found in every butcher shop and in most restaurants. 1.1.5 Grazing and grassland fires Fire is a natural phenomenon that is integral to the health and functioning of grasslands (Bhagwat et al., 2023). It can also be devastating to livestock, consuming entire herds if there are no means of escape or shelter. Kazakhstan has experienced several periods of unusually high fire occurrence since the breakup of the Soviet Union. Climate variables are always an enabling factor with naturally-occurring fires, and climate change has led to an increase in extreme weather events worldwide (Newman & Noy, 2023), but the sudden and repeated increases in frequency and extent in Kazakhstan warrant investigation. The biophysical characteristics of grazing—biomass removal, plant damage, soil compaction, riverbank erosion, manure fertilization—change the community structure of a grassland ecosystem (Milchunas et al., 1988). Livestock grazing is generally understood to be a mitigating factor in fire occurrence—their consumption of fuel overshadows any potentially exacerbating feature of animal husbandry. However, the degree to which livestock grazing contributes to fire mitigation has historically been difficult to determine, partially due to the fact that in order to determine the magnitude, a very large, intact grazing area that experiences regular fires would have to be subjected to increases and decreases in livestock numbers that vary over space and time. The large-scale, widespread, and most importantly, well-documented changes in livestock numbers and grazing practices in Kazakhstan over the last several decades—the first at such a scale—provide ample data and opportunity to explore the extent to which livestock grazing can promote or stifle fire occurrence and burned area extent. 1.2 Conceptual framework The overarching goal of this dissertation is to advance the knowledge and understanding of land use, especially pastureland use, in Kazakhstan. Extent, grazing intensity, productivity, accessibility, and species distribution are all vital aspects of determining the current and future prospects for livestock production and grazing in Kazakhstan. These aspects are explored by 1) developing a fine-scale, spatiotemporal model for livestock distribution and 2) assessing the environmental externalities of livestock production with regard to changes in grazing intensity and species distribution. 9 1.2.1 Research questions and objectives 1.2.1.1 Question I: Where, and to what extent, could increases in Kazakh livestock production take place? Determining the land use footprint of livestock is historically difficult. Satellite imagery has been used for decades to accurately map cropland use, however pastures lack the clearly defined boundaries and changes in spectral reflectance that make cropland easy to detect (Shelestov et al., 2017; Yan & Roy, 2014). Additionally, “use” is a matter of degree much more in pastureland than in cropland, as frequency, length, and intensity can all very greatly on land classified as pasture (Kuemmerle et al., 2013). Moreover, to allocate livestock on pasture provides an additional quandary, as livestock may be housed in feedlots, fenced pastures, or nomadically grazed across hundreds of square kilometers. In Kazakhstan it is typical for all three styles of grazing practices to exist in the same area. 1.2.1.1.1 Objective I: develop a spatial model using fine-scale inputs to allocate livestock in Kazakhstan based on the predominating local grazing practices. Before the spatial distribution of livestock in Kazakhstan can take place, several inputs are needed. Livestock numbers, herd dynamics, energy requirements, and energy provided by feed and fodder were used to determine grazing demand. Land cover, net primary productivity, and plant species energy content were used to determine grazing supply. Settlement locations were used as a starting point in a search algorithm designed with knowledge of the local grazing practices to estimate the extent and off-take rate of pasture in Kazakhstan. 1.2.1.2 Question II: To what extent are livestock in Kazakhstan actors in fire regimes? Fire is an integral part of grassland ecosystems, and human activities have altered the natural fire regimes with diverse and diverging effects (Bowman et al., 2011). In Kazakhstan, fire occurrence is widespread, but spatially and temporally varied, and does not appear to be well-explained by climate patterns (Y. Xu et al., 2021). Additionally, livestock numbers have fluctuated greatly over time and space, providing a unique natural experiment to study the links between fire, grazing, and climate variables. 1.2.1.2.1 Objective II: analyze the correlation between livestock grazing demand and burned area extent compared to the traditional climate variables of precipitation, temperature, relative humidity, and growing degree days. A binomial generalized linear mixed-effects model was used to regress the predictors against the observed values of burned area (Bolker et al., 2009; Warton & Hui, 2011). Akaike Information Criterion was calculated to determine model strength, as well as conditional and marginal R2. Since none of the climate or livestock variables occur in a vacuum, interrelations were explored as well, namely with additive, interactive, and quadratic models. 10 1.2.1.3 Question III: is there a path to sustainable livestock development in Kazakhstan? In Kazakhstan, like most countries, beef is king. Despite beef currently coming in a distant third in global consumption to chicken and pork, cattle are the standard of wealth for livestock owners in developing countries (Herrero, Grace, et al., 2013). The meat itself is also a driving force behind beef’s proliferation, none of the other major meats can easily replace beef’s taste and texture. Unlike chicken and pork, beef can be produced on land otherwise unsuitable for human food production, but by many metrics beef is an extraordinarily inefficient source of protein (Hou et al., 2016). What is needed is a more environmentallyfriendly source of meat that can still be produced by grazing the earth’s vast grassland resources. 1.2.1.3.1 Objective III: explore the potential for the return of a historically popular and gastronomically similar meat in order to at least partially replace beef as the red meat of choice. Horses were one of mankind’s earliest known sources of meat (Taylor, 2024). The current taboo surrounding hippophagy needs to be reevaluated. The environmental externalities of greenhouse gas emissions and land use were evaluated, especially in comparison to beef production. However, all the environmental arguments mean nothing if horsemeat is not desirable as a dish, thus the nutritional characteristics were also evaluated in comparison to beef and other red meats. 1.3 Structure of the thesis This thesis consists of five chapters. Chapter 1 introduces the core three Chapters 2-4, which address the aforementioned three objectives, and Chapter 5 synthesizes the results and conclusions presented and looks toward the future landscape of livestock, meat production, land use, and grassland ecosystems. Chapters 2-4 were written as independent manuscripts for peer-reviewed journals, and as such have only minor formatting alterations from their published or submitted versions: Chapter 2 Hankerson, Brett R., Florian Schierhorn, Alexander V. Prishchepov, Changxing Dong, Christina Eisfelder, Daniel Müller (2019). Modeling the spatial distribution of grazing intensity in Kazakhstan. PLOS ONE, 14 (1), e0210051. Chapter 3 Hankerson, Brett R., Florian Schierhorn, Johannes Kamp, Tobias Kuemmerle, Daniel Müller (in review). Changes in livestock systems explain post-Soviet fire trends on the Eurasian steppe better than climate. Regional Environmental Change. Chapter 4 Hankerson, Brett R., Daniel Müller (in review). The other red meat: environmental and nutritional advantages of horsemeat over beef and other red meats. Nature Food. 11 Chapter 2: Modeling the spatial distribution of grazing intensity in Kazakhstan PLOS ONE, 2019, Volume 14, Issue 1 Brett R. Hankerson, Florian Schierhorn, Alexander V. Prishchepov, Changxing Dong, Christina Eisfelder, Daniel Müller 12 Abstract With increasing affluence in many developing countries, the demand for livestock products is rising and the increasing feed requirement contributes to pressure on land resources for food and energy production. However, there is currently a knowledge gap in our ability to assess the extent and intensity of the utilization of land by livestock, which is the single largest land use in the world. We developed a spatial model that combines fine-scale livestock numbers with their associated energy requirements to distribute livestock grazing demand onto a map of energy supply, with the aim of estimating where and to what degree pasture is being utilized. We applied our model to Kazakhstan, which contains large grassland areas that historically have been used for extensive livestock production but for which the current extent, and thus the potential for increasing livestock production, is unknown. We measured the grazing demand of Kazakh livestock in 2015 at 286 Petajoules, which was 25% of the estimated maximum sustainable energy supply that is available to livestock for grazing. The model resulted in a grazed area of 1.22 million km2, or 48% of the area theoretically available for grazing in Kazakhstan, with most utilized land grazed at low intensities (average off-take rate was 13% of total biomass energy production). Under a conservative scenario, our estimations showed a production potential of 0.13 million tons of beef additional to 2015 production (31% increase), and much more with utilization of distant pastures. This model is an important step forward in evaluating pasture use and available land resources, and can be adapted at any spatial scale for any region in the world. 19 production (grams dry matter/square meter) using the conversion coefficient of 0.47 grams carbon/grams dry matter (IPCC, 2006). On the grassland map, we converted biomass to available energy using a value of 8.6 Megajoules (MJ) per kilogram dry matter (kgDM), based on literature from similar regions and climates (Figure 2-) (Grebennikov & Shipilov, 2012; Nasiev et al., 2014; Safin et al., 2011). The total energy available from grasslands was calculated to be 3537 PJ. To estimate the biomass available from the foraging of crop residues, we applied a harvest index of 0.48 (using wheat as the base reference) to the annual NPP (Chen et al., 2014; Sommer et al., 2013). The harvest index is the mass ratio of crop yield (grain) to the crop’s total aboveground biomass (Smil, 1999). Wheat is the dominant crop grown in Kazakhstan and thus was used for the calculation of crop residues (KazStat, 2016). At harvest, around 90% of wheat biomass is aboveground (Baret et al., 1992). The energy contained in crop residues is generally less than that of pasture, and a value of 6 MJ/kgDM (using wheat as the base reference) was used for croplands (Ørskov et al., 1988; Smil, 1999). The total energy available for grazing from croplands was calculated to be 293 PJ. The grassland map and cropland map were then merged to produce a map of grazing supply (Figure 2-2). Figure 2-2: Map of annual available grazing supply (MJ/m2) derived in this study. Based on total NPP measured by Eisfelder et al. (2014). The land-cover classification used for masking is from Klein et al. (2012), and the protected area mask is from Kamp et al. (2015). Low available NPP for foraging on croplands can easily be seen in the northcentral. White areas are unavailable for grazing. 2.2.5 Grazing demand To calculate grazing demand, we gathered data on livestock numbers, fodder yield and fodder consumption at the district level (2nd level administrative division) for 2015 from the Kazakh National Statistics Agency (Figure 2-3) (KazStat, 2016). In Kazakhstan, there are 200 district-level units, consisting of districts (rayons) and city administrative units (gorodskie administratsii). Livestock nutritive requirements were taken from recommended values published in a livestock nutrition handbook by KazAgroInnovation (2008), a 20 subordinate of the Kazakh Ministry of Agriculture. These values are specific to Kazakh livestock at different stages of growth and for different animal functions (e.g., beef heifers for breeding vs. beef heifers for finishing), as well as for different desired growth rates. Because sheep far outnumber goats in Kazakhstan, and because of their similar grazing characteristics and energy requirements, all goats were treated as sheep. Figure 2-3: Livestock density (head/km2) at the district level for the three farm types in Kazakhstan for the year 2015 (KazStat, 2016). Livestock in Kazakhstan are not distributed evenly across space, nor across the three farm types. To calculate the total nutritive demand (Figure 2-), we used information on the age group and animal function of the Kazakh livestock herds. The proportions of the different age groups and animal functions were available at the province (oblast) level for the different livestock types from the 2006 agricultural census (KazStat, 2008). As no newer or more detailed data exist, these proportions were applied to the 2015 numbers in our disaggregation equation (Equation 2-S2). Animal productivity differs depending on living conditions, and living conditions in Kazakhstan can broadly be defined based on the farm type. The differences in animal productivity, and thus energy demand, on the different farm types was estimated using different animal growth rates (g/day) as indicated in the handbook (Table 2-S2) (Zhazylbekov et al., 2008). A description of how the handbook values were used to calculate the different animal age and function groups can be found in Table 2-S3. 21 The fraction of total energy demand that is not met by fodder and therefore must be met by grazing (i.e., grazing demand divided by total demand) is called the grazing gap, and was obtained by subtracting the amount of energy that is consumed as fodder from the total demand (Fetzel, Havlík, Herrero, Kaplan, et al., 2017). The estimation of fodder consumption was not straightforward, as such statistics are reported consistently only for agricultural enterprises. Instead, gross yield of harvested fodder crops at the district level was used (KazStat, 2016), and fodder consumption statistics were used in an equation (Equation 2-S3) to allocate fodder to the different livestock types using their relative proportions (i.e., proportion of total fodder consumption allocated to cattle, sheep, goats, pigs, poultry, horses, and camels). This was the only possible way to estimate the total amount of fodder consumed by grazing livestock in Kazakhstan. Fodder crop yields were converted to energy values using the Soviet system of “fodder units” (Forage On-Line, 2009), due to its easy conversion to MJ and its widespread use in Kazakh agricultural literature and statistical reporting (Table 2-S4 shows the conversion rates used). Table 2-1 details the inputs used to distribute the demand onto the supply as shown in Figure 2-. When distributing the supply, it is important to acknowledge that only a fraction of the total NPP can be consumed by livestock. The Eisfelder et al. (2014) map is a measurement of total NPP, which includes the portion that is belowground and unavailable to the livestock. We used the work of Propastin et al. (2011), who found aboveground NPP in central Kazakhstan to be on average 77% of total NPP. In addition, a considerable portion of the aboveground NPP must be left to allow regrowth. Published values of recommended stocking rates and pasture utilization in similar climatic conditions suggest a maximum offtake rate of 40% of aboveground NPP (Hall et al., 1998; Holechek, 1988; Wirsenius, 2000), and thus the maximum sustainable off-take rate was estimated at 40% of 77%, or 30% of total NPP. Table 2-1: Summary of parameters used. Net primary production was taken as an average of the annual products from 2003 to 2011. Input parameter (units in parentheses) Spatial resolution Reference period Source Livestock numbers District 2015 KazStat (2016) Fodder production (Joules) District 2015 KazStat (2016) Fodder consumption (Joules) District/Province 2015 KazStat (2016) Nutritive requirements (Joules) 2008 KazAgroInnovation (2008) Herd age structure Province 2006 KazStat (2008) Human population Settlement 2009 KazStat (2011) Settlement location Settlement 2016 Geofabrik (2016) District and municipal areas District 2015 GADM (2015), GIS-Lab (2013) Net primary production (gC/m2) 1 km 2003-2011 Eisfelder et al. (2014) Land cover 250 m 2009 Klein et al. (2012) Obviously, not all pasture is grazed at the maximum sustainable off-take rate. Mapping actual grazing intensity requires estimating the variation in off-take rate on a spatial scale. To map variation in off-take rate using our model, we first ran the model under a range of eleven different off-take rate assumptions (5% increments from 10%-60%). We then calculated the maximum distance from each settlement that each 22 livestock type needed to fulfill their grazing demand under each off-take rate. To determine an accurate off-take rate for each settlement, we used the maximum grazing distances for cattle in households as the defining variable, as they, along with sheep and goats (which are distributed before cattle), are the most restricted by distance from settlement. We chose 10 km as the maximum distance for cattle in households based on the findings of Kamp et al. (2012). We grouped settlements into their districts, and for each district, we selected the lowest off-take rate that corresponded to the median of maximum grazing distances for cattle in households being less than 10 km. In cases where districts had no reported cattle in households, sheep and goats in households instead were used (districts without either of these had no grazing livestock of any kind). The model was then run again, with settlements maintaining their determined off-take rate. 2.2.6 Production potentials of meat and milk We estimated the potential to increase production of meat and milk in Kazakhstan based on the efficiency with which pasture is being used. To estimate pasture use efficiency, we took meat and milk production in 2015 from the national statistics (KazStat, 2016). We calculated pasture requirement from the model results and estimated the yield of meat and milk (tons per km2 utilized) for the different livestock types. Our model results do not differentiate between beef and dairy cattle, so we made an adjustment based on the relative proportions of beef and dairy cattle. The fraction of cattle classified as dairy in 2015 for agricultural enterprises, private farms, and households was 0.40, 0.50, and 0.85, respectively (KazStat, 2016). We multiplied the land requirement by this fraction as a rough estimate for the area used by dairy cattle. We then divided milk production by the adjusted land requirement to estimate milk productivity. We used the calculated land use efficiencies to estimate production potential. First, we made a conservative assumption that all land within 10 km of a settlement could currently be utilized. Therefore, unutilized land within 10 km of a settlement was considered for potential expansion. The modeled off-take rates were used to calculate the number of additional livestock that could be supported. Second, we proposed a scenario where pasture was grazed at its maximum sustainable intensity (30% off-take rate), and calculated the resulting unutilized area within 10 km of a settlement. Using a less conservative assumption that all land within 20 km of a settlement could be utilized, we repeated the previous two calculations. Potential increase in beef production was calculated with the assumption that all additional livestock were beef cattle. Similarly, for potential increase in milk production, we assumed that all additional livestock were dairy cattle. Therefore, the results presented are “either-or”, and the reality likely falls somewhere in between. 23 2.3 Results 2.3.1 Grazing gap and demand distribution The grazing demand was calculated for each animal and farm type combination (Figure 2-4). The total energy demand by all livestock types in 2015 (the sum of all bars in Figure 2-4) was 368 Petajoules (PJ). The grazing gap is displayed above each bar as the fraction of total energy demand obtained through grazing. Of the three livestock types, the grazing gap is lowest for cattle, and of the three farm types, the grazing gap is lowest for agricultural enterprises, with the lowest being cattle on agricultural enterprises. This is due to cattle on agricultural enterprises receiving more and higher-quality fodder than other livestock and on other farm types. The total amount of energy supplied by fodder was 82 PJ (sum of all darker portions of the bars), leaving 286 PJ to be obtained through grazing. Figure 2-4: Energy balance for Kazakh grazing livestock in 2015. The darker bottom portion of each bar is the fodder supply, and the lighter top portion is the remaining demand that must be acquired from grazing. The total demand is represented by the full bar height. Fractions above each bar show the grazing gap (grazing demand divided by total demand). Nutritive demand information is from KazAgroInnovation (2008) and supply statistics from KazStat (2016). Figure 2-5 shows the total grazing demand of all livestock types for each settlement in Kazakhstan. Grazing demand is not distributed evenly across the country. In the north, the demand is large, but dispersed across many settlements, whereas in the south it is also large, but concentrated in relatively fewer settlements. The center and southwest have both few settlements and little grazing demand. 24 Figure 2-5: Grazing demand (in Terajoules, TJ) in 2015 disaggregated to settlements (all farm types combined). 2.3.2 Off-take rate Off-take rate is not uniform across Kazakhstan. The grazing demand for 2015 was 7.5% of the total biomass supply (when converted to energy)—i.e., if the off-take rate were 7.5%, all available land would be utilized. We tested the sensitivity to off-take rate by running the model with eleven different off-take rates, from 10% to 60% (5% increments) (Figure 2-6). As the off-take rate decreases, the area required for grazing increases exponentially (Figure 2-7). In our results, all off-take values used are as a percent of total available NPP. 25 Figure 2-6: Grazing extent by all livestock under varying off-take rates. The image is a superimposition of the eleven model runs. The map of each individual off-take rate includes the area of all higher off-take rates. Figure 2-7: Area required by grazing livestock depending on the percent of biomass off-take. 2.3.3 Grazing distances and pasture extent The maximum distances traveled by household cattle under each off-take rate assumption were analyzed at the district level to determine the average off-take rate in each district. The model was re-run with variable off-take rates to derive the maximum grazing distances. Figure 2-8 shows the median and quartiles of these distances by animal and farm type. A smaller quartile range on the left-hand side of the median for every livestock type is a result of the distances being skewed by relatively few settlements with a large livestock population located close to one another, most notably in southcentral Kazakhstan (Figure 2-5). Most settlements had much shorter maximum grazing distances, within 6 km for cattle, sheep, and goats in households, and within 15 km for cattle, sheep, and goats on private farms. Despite being distributed later, cattle on agricultural enterprises were found to have lower maximum grazing distances than sheep and goats on agricultural enterprises. This was due to agricultural enterprises specializing in cattle production 26 being located mainly in the north in small settlements, whereas agricultural enterprises with sheep and goats were located mainly in the south and southeast in or near large settlements. Figure 2-8: Maximum grazing distances of the variable off-take rate map. Box shows 50% of settlements (with median), whiskers are ½ inter-quartile range. AE: agricultural enterprises, PF: private farms, HH: households. Figure 2-9 shows the land-use footprint of grazing livestock in 2015, using variable off-take rates at the district level. The area required was 1.22 million km2, 48% of the area theoretically available for grazing. While the off-take rate was determined at the district level, the result shows that off-take rates did not strictly adhere to district boundaries, as individual settlements are not obliged to graze within district boundaries. In the north, a relatively higher number of livestock are kept in private farms and agricultural enterprises, which are not as restricted as household livestock to the immediate vicinity of settlements. Thus, they can utilize distant pastures at lower off-take rates, and almost all of the north and northeast was utilized to some extent. The south and southeast showed less land being utilized, however at a much higher off-take rate. In the east, high NPP allows for lower off-take rates, and high numbers of private farm livestock can search out distant pastures. Two riparian pasture regions are clearly visible due to their course running through otherwise arid and semi-arid regions: the Ural in the far west and the Syr Darya flowing northwest out of the southern tip. The Chu River (to the east of the Syr Darya) is a historically important river that used to flow into the Syr Darya, but for many years has been diverted for irrigation and now disappears before reaching the Syr Darya. The Ili River in the southeast flows from the mountains of Tian Shan into Lake Balkhash, where it forms a large delta, providing grazing opportunities in an otherwise arid landscape. 27 Figure 2-9: Distribution of grazing intensity in Kazakhstan for the year 2015. Off-take rate is the percent of total available biomass that is consumed. I.e. on croplands, it is the percentage consumed of the biomass that remained after harvest. Major rivers are shown in blue with names. 2.3.4 Production potentials of meat and milk The summarization of utilized pasture made it possible to estimate the associated productivity of livestock production with regard to pasture use. Table 2-2 shows the area of pasture utilized and the respective productivity of meat and milk (production per km2 utilized). Meat productivity is highest for cattle on agricultural enterprises, but not by a lot. With the much smaller grazing gap for cattle on agricultural enterprises (Figure 2-4), one would expect the meat productivity (which doesn’t account for fodder) to be much higher. This is not the case because most cattle on agricultural enterprises are in northcentral Kazakhstan (Figure 2-3), where the off-take is low (Figure 2-9) and a lot of grazing on cropland (Figure 2-2) occurs. Both factors increase the land utilized by cattle on agricultural enterprises compared to other livestock and on other farm types, and thus decrease the relative meat productivity. Total beef production in 2015 was 417 thousand tons (kt), and total dairy milk production was 5.1 million tons (Mt). Table 2-2: Pasture use, production, and productivity of meat and milk in 2015. Meat and milk production statistics from KazStat (2016). These numbers do not account for land used for fodder production. *Adjusted for relative proportion of cattle in dairy production. AE: agricultural enterprises, PF: private farms, HH: households. Livestock type Farm type Pasture utilized (mil. km2) Meat production (kt) Meat productivity (t/km2) Milk production (kt) Milk productivity (t/km2) Cattle AE 0.30 28.44 96.09 263.01 2213.05* PF 1.47 77.80 52.90 777.55 1048.46* HH 4.44 310.57 69.98 4101.06 1091.96* Sheep AE 0.10 3.16 32.52 0.04 0.38 and goats PF 1.14 38.74 34.11 0.33 0.29 HH 1.82 123.19 67.60 1.25 0.68 28 Horses AE 0.17 2.17 12.94 0.69 4.14 PF 1.35 25.01 18.47 9.76 7.21 HH 1.44 74.26 51.44 15.42 10.68 For sheep, goats, and horses, meat productivity is highest in households. For sheep and goats, this is due to sheep on agricultural enterprises and private farms primarily being raised for wool, with meat only a byproduct. Similarly, for horses, most meat production is done at the household level. Regarding milk production, cattle on agricultural enterprises are clearly the most land productive (when adjusted for the proportion of dairy production). Milk productivity is very low for sheep and goats, with almost all production coming from the very few goats in the country. Horse milk productivity is somewhat higher, because of the demand for the traditional horse-milk drink kumys, which is produced mainly at the household level. We produced a conservative estimate for increased production potential by implementing the scenario where all land within 10 km of a settlement is utilized. Assuming the estimated off-take rates shown in Figure 2-9 as business-as-usual (BAU), the additional pasture utilized was 0.14 million km2, with an associated energy of 29.9 PJ. Assuming a proportional increase in fodder production (i.e., that the grazing gap remains the same), if the additional 0.14 million km2 of pasture was used entirely for cattle on agricultural enterprises, beef production could be increased by 0.13 Mt, an increase of 31% (Figure 2-10). Conversely, if all expansion was used for dairy on agricultural enterprises, dairy milk production could be increased by 3.11 Mt (above 2015 level) under the business-as-usual scenario, an increase of 60%. These are conservative estimates, as off-take rates were very low for most settlements (Figure 2-9). If all land within 10 km was used at its maximum sustainable off-take rate (30%), and if all additional livestock on pasture within 10 km were cattle on agricultural enterprises, beef production could be increased 1.91 Mt (above 2015 level), an increase of 457%. By comparison, Brazilian beef exports in 2013 totaled 1.25 Mt (Schierhorn et al., 2016). Hence, Kazakhstan has the potential to become one of the leading beef exporters in the world. If the radius of land around a settlement that can be utilized was increased to 20 km, with business-as-usual off-take rates, and if all additional pasture was utilized by cattle on agricultural enterprises, beef production could be increased by 0.41 Mt (98%). Assuming maximum sustainable off-take rates within 20 km, this estimate increases to 3.96 Mt. 35 The model enabled a gridded estimate of the utilized pasture in Kazakhstan, which is a prime example of a country well suited to grazing livestock production. Kazakhstan’s dry continental climate also reduces the suitability of livestock production’s main competitor for land, crop production, making it a suitable target area for development of range-based livestock production. Our results show that despite relatively low natural productivity, ample capacity exists to increase livestock production in Kazakhstan because large areas are characterized by low pasture utilization and off-take rate, and available biomass resources could support many more grazing animals, especially in the east and the northwest (Figure 2-9). Under conservative estimates of grazing range constraints and with 2015 productivity levels, beef production could be increased by 0.13 Mt (31%) or milk production by 3.11 Mt (60%), or some combination. However, harnessing even a fraction of these potentials would necessitate infrastructure development measures, such as more, improved processing facilities and improved road networks and market access. Repaired wells and outposts would allow the rejuvenation of old migration patterns and would open up distant pastures for even more potential production increases. This research is an important step forward in the field of livestock mapping. Our model uses much finerscale inputs than other global-scale products, and a direct measurement of biomass production, enabling us to make a gridded estimate of pasture distribution based on the energy demand of the livestock. The search algorithm we created is easily transferable to other regions where livestock are restricted to a central point, but can be adapted to any region where grazing patterns can be defined. The result of our research can be used to find patterns in livestock distribution, and to target areas where the supply is underutilized. Moreover, our results help the spatial targeting of possible investments for expanding the production of grazing livestock, including assessing the tradeoffs of production expansion with greenhouse gas emissions and biodiversity conservation. 2.6 Acknowledgments We want to thank several colleagues who provided invaluable assistance and advice. We thank Anne Jungandreas for help with coding the early versions of the search algorithm, Alina Drokina for compiling the agricultural statistics from KazStat, Johannes Kamp for his expertise in grazing patterns, Norbert Hölzel for his expertise in plant species distribution, Matthias Baumann for his insights into land cover dynamics, Martin Petrick for his insights into household socioeconomics, Andrey Dara, Natalya Tsychuyeva, and Alyona Chukhatina for their guidance and expertise during field trips, and Yerlan Syzdykov for his efforts in data procurement and his insights on large farm management in Kazakhstan. We are particularly grateful to Dauren Oshakbaev for his assistance, expertise, and coordinative efforts, and without whom this research could not have been completed. Additionally, we are thankful to Ruslan Urazaliyev, Albert Salemgareyev, and the colleagues working for the Altyn Dala Conservation Initiative for their knowledge and information on wild grazers in Kazakhstan. We 36 are also grateful to Tamara Fetzel for sharing her map of global grazing intensity, allowing us to perform a qualitative comparison of similar research. 37 2.7 Supplementary material Equation 2-S1: The piecewise linear function used to estimate the number of livestock owners in a settlement based on the total settlement population. 𝑓(𝑃)= { 0.95∗𝑃, 𝑃≤1000 (0.95−0.15 5000−1000∗(𝑃−1000))∗𝑃, 1000<𝑃≤5000 (0.8− 0.5 10000−5000∗(𝑃−5000))∗𝑃, 5000<𝑃≤10000 (0.3− 0.2 50000−10000∗(𝑃−10000))∗𝑃, 10000<𝑃≤50000 (0.1− 0.095 1000000−50000∗(𝑃−50000))∗𝑃, 50000<𝑃≤1000000 0.005∗𝑃, 𝑃>1000000 where: 𝑃=𝑠𝑒𝑡𝑡𝑙𝑒𝑚𝑒𝑛𝑡 𝑝𝑜𝑝𝑢𝑙𝑎𝑡𝑖𝑜𝑛 (2009 𝑐𝑒𝑛𝑠𝑢𝑠) 𝑓(𝑃)=𝑠𝑒𝑡𝑡𝑙𝑒𝑚𝑒𝑛𝑡 𝑙𝑖𝑣𝑒𝑠𝑡𝑜𝑐𝑘 𝑜𝑤𝑛𝑒𝑟𝑠 Figure 2-S1: Piecewise function for the distribution of livestock to settlements based on human population. The steps are 95% from 0-1,000, 80% at 5,000, 30% at 10,000, 10% at 50,000, and 0.5% at 1 million residents (not shown), with linear interpolation. Values were determined from expert opinion and personal observation. Note that this does not affect the total number of livestock, only their distribution within a district. Table 2-S1: Assumptions made in the distribution model. Assumption Source Impact Livestock are located in settlements (Alimaev & Behnke, 2008; Coughenour et al., 2008; Ellis & Lee, 2003; Kamp et al., 2012; Nomadism is precluded, as are potential outposts and wells that are located far away from any settlement. Thus the areas furthest from settlements are the least likely to be distributed in this model 38 Robinson, 2000) Livestock seek pastures with the highest productivity (Deli et al., 2005; Kerven et al., 2016) “Optimal forager” activity is severely truncated due to the introduction of search radii. Affects the distribution of individual livestock/farm types. The overall distribution is affected only at the edge of each settlement’s grazing area, where distributed area is clustered due to the priority of high-NPP values Livestock graze on croplands (Coughenour et al., 2008) The total energy available for grazing increased by 293 PJ, or 8.3%. Most of this energy was located in the northcentral, where cropland dominates. Figure 2-S2: A magnified view showing the result of the distribution model at the settlement level. Each unique color represents the land distributed to a unique settlement (black dots). This illustrates the operation of the search algorithm when settlements are in close proximity and their pasture requirements conflict. Equation 2-S2: The equation used to estimate the number of livestock in each age group in each district in 2015 using the more detailed 2006 agricultural census. 𝐿𝑚𝑟𝑙𝑎 =𝐿𝑚𝑟𝑙 ∗𝐿𝑜𝑎 𝐶 𝐿𝑜 𝐶 where: 𝐿=𝑛𝑢𝑚𝑏𝑒𝑟 𝑜𝑓 𝑙𝑖𝑣𝑒𝑠𝑡𝑜𝑐𝑘 𝑚=𝑓𝑎𝑟𝑚 𝑡𝑦𝑝𝑒 𝑟=𝑑𝑖𝑠𝑡𝑟𝑖𝑐𝑡 𝑙=𝑙𝑖𝑣𝑒𝑠𝑡𝑜𝑐𝑘 𝑡𝑦𝑝𝑒 𝑎=𝑎𝑔𝑒 𝑔𝑟𝑜𝑢𝑝 𝑜=𝑟𝑒𝑔𝑖𝑜𝑛 𝑤ℎ𝑒𝑟𝑒 𝑜∋𝑟 𝐿𝐶=𝑛𝑢𝑚𝑏𝑒𝑟 𝑜𝑓 𝑙𝑖𝑣𝑒𝑠𝑡𝑜𝑐𝑘 𝑖𝑛 2006 𝑐𝑒𝑛𝑠𝑢𝑠 Table 2-S2: Values used in the calculation of total energy demand (Zhazylbekov et al., 2008). The number of each livestock species are recorded at the district level for each farm type. Livestock numbers for each age group were recorded in the 2006 agricultural 39 census at the regional level (in which cattle were further divided into beef and dairy) (KazStat, 2008). The number of livestock in each age group of each livestock species in each district was estimated by multiplying the number of each livestock species by the number in each species’ age group (in 2006) and dividing by the number of each livestock species (in 2006). Sheep and goat numbers were combined and sheep nutritive requirements were used, due to the low number of goats and their similar nutritive requirements. Type Breed Age/Sex Period Mass (kg) Gain (g/d) Milk (kg/d) ME (MJ/d) Cattle Beef Cows 2 mos pre calving 400 91 Cattle Beef Cows 2 mos pre calving 450 98 Cattle Beef Cows 2 mos pre calving 500 105 Cattle Beef Cows 2 mos pre calving 550 112 Cattle Beef Cows 3-4 mos post calving 400 113 Cattle Beef Cows 3-4 mos post calving 450 117 Cattle Beef Cows 3-4 mos post calving 500 122 Cattle Beef Cows 3-4 mos post calving 550 127 Cattle Beef Cows 2nd half lactation 400 92 Cattle Beef Cows 2nd half lactation 450 101 Cattle Beef Cows 2nd half lactation 500 109 Cattle Beef Cows 2nd half lactation 550 117 Cattle Beef Bulls Not coupling 800 83.2 Cattle Beef Bulls Avg load 800 90 Cattle Beef Bulls Inc load 800 110 Cattle Beef Heifers 9-10 mos 244 575 69 Cattle Beef Heifers 9-10 mos 260 675 78 Cattle Beef Heifers 11-12 mos 277 575 76 Cattle Beef Heifers 11-12 mos 300 675 85 Cattle Beef Heifers 13-14 mos 311 575 83 Cattle Beef Heifers 13-14 mos 340 675 94 Cattle Beef Heifers 15-16 mos 345 575 90 Cattle Beef Heifers 15-16 mos 380 675 104 Cattle Beef Heifers 17-20 mos 413 575 105 Cattle Beef Heifers 17-20 mos 420 675 112 Cattle Beef Bull calves 9-10 mos 279 825 72 Cattle Beef Bull calves 9-10 mos 285 925 79 Cattle Beef Bull calves 11-12 mos 330 825 80 Cattle Beef Bull calves 11-12 mos 340 925 88 Cattle Beef Bull calves 13-14 mos 379 825 86 Cattle Beef Bull calves 13-14 mos 396 925 94 Cattle Beef Bull calves 15-16 mos 428 825 92 Cattle Beef Bull calves 15-16 mos 451 925 102 Cattle Beef Calves 0-6 mos 32 Cattle Beef Calves 6-9 mos 54.6 Cattle Beef Calves 10-12 mos 78.5 Cattle Beef Calves 13-15 mos 93.5 Cattle Dairy Cows Milking 1st half 400 6 72.5 Cattle Dairy Cows Milking 1st half 400 10 106 Cattle Dairy Cows Milking 1st half 400 14 127.5 Cattle Dairy Cows Milking 1st half 400 18 149 Cattle Dairy Cows Milking 1st half 400 22 172.5 Cattle Dairy Cows Milking 1st half 400 26 197.5 Cattle Dairy Cows Milking 1st half 400 30 227.5 Cattle Dairy Cows Milking 2nd half 400 5 61 Cattle Dairy Cows Milking 2nd half 400 7 83.5 40 Cattle Dairy Cows Milking 2nd half 400 9 100.5 Cattle Dairy Cows Milking 1st half 500 6 79.5 Cattle Dairy Cows Milking 1st half 500 10 115 Cattle Dairy Cows Milking 1st half 500 14 137 Cattle Dairy Cows Milking 1st half 500 18 158 Cattle Dairy Cows Milking 1st half 500 22 180.5 Cattle Dairy Cows Milking 1st half 500 26 205.5 Cattle Dairy Cows Milking 1st half 500 30 230.5 Cattle Dairy Cows Milking 2nd half 500 5 65 Cattle Dairy Cows Milking 2nd half 500 7 94.5 Cattle Dairy Cows Milking 2nd half 500 9 109.5 Cattle Dairy Cows Dry or sterile 400 0 66 Cattle Dairy Cows Dry or sterile 500 0 78 Cattle Dairy Cows Dry or sterile 400 0 78 Cattle Dairy Cows Dry or sterile 500 0 86 Cattle Dairy Bulls Not coupling 600 70 Cattle Dairy Bulls Not coupling 700 78 Cattle Dairy Bulls Not coupling 800 84 Cattle Dairy Bulls Not coupling 900 91 Cattle Dairy Bulls Not coupling 1000 97 Cattle Dairy Bulls Not coupling 1100 102 Cattle Dairy Bulls Not coupling 1200 108 Cattle Dairy Bulls Avg load 600 76 Cattle Dairy Bulls Avg load 700 83 Cattle Dairy Bulls Avg load 800 90 Cattle Dairy Bulls Avg load 900 97 Cattle Dairy Bulls Avg load 1000 104 Cattle Dairy Bulls Avg load 1100 110 Cattle Dairy Bulls Avg load 1200 117 Cattle Dairy Bulls Inc load 600 92 Cattle Dairy Bulls Inc load 700 102 Cattle Dairy Bulls Inc load 800 110 Cattle Dairy Bulls Inc load 900 119 Cattle Dairy Bulls Inc load 1000 127 Cattle Dairy Bulls Inc load 1100 134 Cattle Dairy Bulls Inc load 1200 141 Sheep Wool Ewes Preg 1st half 50 12.5 Sheep Wool Ewes Preg 1st half 60 13.5 Sheep Wool Ewes Preg 2nd half 50 14.5 Sheep Wool Ewes Preg 2nd half 60 16.5 Sheep Meat/wool Ewes Preg 1st half 60 12.1 Sheep Meat/wool Ewes Preg 1st half 70 13 Sheep Meat/wool Ewes Preg 2nd half 60 16 Sheep Meat/wool Ewes Preg 2nd half 70 17.2 Sheep Meat/tallow Ewes Preg 1st half 60 13.5 Sheep Meat/tallow Ewes Preg 1st half 70 14.5 Sheep Meat/tallow Ewes Preg 2nd half 60 17.5 Sheep Meat/tallow Ewes Preg 2nd half 70 18.5 Sheep Wool Ewes Suckling 1st half 50 20 Sheep Wool Ewes Suckling 1st half 60 23 Sheep Wool Ewes Suckling 2nd half 50 15.5 Sheep Wool Ewes Suckling 2nd half 60 17 41 Sheep Meat/wool Ewes Suckling 1st half 60 22 Sheep Meat/wool Ewes Suckling 1st half 70 23 Sheep Meat/wool Ewes Suckling 2nd half 60 18.4 Sheep Meat/wool Ewes Suckling 2nd half 70 19.2 Sheep Meat/tallow Ewes Suckling 1st half 60 21 Sheep Meat/tallow Ewes Suckling 1st half 70 22 Sheep Meat/tallow Ewes Suckling 2nd half 60 18.5 Sheep Meat/tallow Ewes Suckling 2nd half 70 19.5 Sheep Wool & meat/wool Rams Not coupling 90 19 Sheep Wool & meat/wool Rams Not coupling 100 20 Sheep Wool & meat/wool Rams Not coupling 110 21 Sheep Meat/tallow Rams Not coupling 80 19 Sheep Meat/tallow Rams Not coupling 90 20 Sheep Meat/tallow Rams Not coupling 100 21 Sheep Wool & meat/wool Rams Coupling 90 24 Sheep Wool & meat/wool Rams Coupling 100 25 Sheep Wool & meat/wool Rams Coupling 110 26 Sheep Meat/tallow Rams Coupling 80 24 Sheep Meat/tallow Rams Coupling 90 25 Sheep Meat/tallow Rams Coupling 100 26 Sheep Wool Ewe lambs 4-6 mos 27.5 8.525 Sheep Wool Ewe lambs 6-8 mos 33 9.625 Sheep Wool Ewe lambs 8-10 mos 38 10.725 Sheep Wool Ewe lambs 10-12 mos 41 11 Sheep Wool Ewe lambs 12-18 mos 46 11.275 Sheep Meat/wool Ewe lambs 4-6 mos 38.5 11 Sheep Meat/wool Ewe lambs 6-8 mos 38.5 12.1 Sheep Meat/wool Ewe lambs 8-10 mos 43 13.2 Sheep Meat/wool Ewe lambs 10-12 mos 47.5 14.025 Sheep Meat/wool Ewe lambs 12-18 mos 51.5 14.025 Sheep Wool Ram lambs 4-6 mos 32 11.275 Sheep Wool Ram lambs 6-8 mos 39.5 12.43 Sheep Wool Ram lambs 8-10 mos 45 13.75 Sheep Wool Ram lambs 10-12 mos 49.5 15.18 Sheep Wool Ram lambs 12-18 mos 61.5 15.4 Sheep Meat/wool Ram lambs 4-6 mos 36.5 13.2 Sheep Meat/wool Ram lambs 6-8 mos 44.5 14.3 Sheep Meat/wool Ram lambs 8-10 mos 52.5 15.4 Sheep Meat/wool Ram lambs 10-12 mos 60 16.775 Sheep Meat/wool Ram lambs 12-18 mos 70 17.325 Horses Meat Stallions 350 1000 93.2 Horses Meat Stallions 400 1000 97.4 Horses Meat Stallions 450 1000 101 Horses Meat Stallions 500 1000 108.8 Horses Meat Stallions 550 1000 112 Horses Meat Stallions 600 1000 122.4 Horses Meat Foals 0-1 mos 70 1300 42 Horses Meat Foals 1-2 mos 105 1000 45 Horses Meat Foals 2-3 mos 133.5 900 47 Horses Meat Foals 3-4 mos 160.5 900 52 Horses Meat Foals 4-5 mos 187.5 900 56 Horses Meat Foals 5-6 mos 214.5 900 60 42 Horses Meat Foals 6-7 mos 235.5 900 60 Horses Meat Foals 7-8 mos 258 1000 71 Horses Meat Foals 8-9 mos 291 1300 88 Horses Milk Mares 400 10 84.8 Horses Milk Mares 400 12 92.1 Horses Milk Mares 400 14 98.4 Horses Milk Mares 500 14 105.7 Horses Milk Mares 500 16 113.1 Horses Milk Mares 500 18 120.4 Horses Milk Mares 500 20 126.5 Table 2-S3: Age groups used in the estimation of livestock energy demand. Restrictions are noted when the age group is a subset of the Age/Sex in Table 2-S2. Nomenclature of the age group is as follows (square brackets enclose string variables): [*Beef/Dairy*][Function]_[*age*]_[*reproduction stage*]. * – if applicable. Function – common name considering age, sex, and castration. Age group Age/Sex in Table 2-S2 Period/weight restriction BeefCows Beef Cows BeefBulls_breeding Beef Bulls BeefHeifers Beef Heifers BeefHeifers_1to2yrs Beef Heifers 13-16 mos. BeefHeifers_1to2yrs_inseminated Beef Heifers 17-20 mos. BeefHeifers_2yrs Beef Heifers 17-20 mos. BeefCalves_heifers_0to1yr Beef Calves 0-12 mos. BeefCalves_bulls_0to1yr Beef Calves 0-12 mos. BeefCalves_bulls_1yr Beef Bull calves 13-16 mos. BeefSteers Beef Bull calves BeefOxen Beef Bulls Not coupling BeefCattle_finishing Beef Bull calves 15-16 mos. BeefBuffaloes Beef Bulls Not coupling DairyCows Dairy Cows DairyBulls_breeding Dairy Bulls DairyHeifers Beef Heifers DairyHeifers_1to2yrs Beef Heifers 13-16 mos. DairyHeifers_1to2yrs_inseminated Beef Heifers 17-20 mos. DairyHeifers_2yrs Beef Heifers 17-20 mos. DairyCalves_heifers_0to1yr Beef Calves 0-12 mos. DairyCalves_bulls_0to1yr Beef Calves 0-12 mos. DairyCalves_bulls_1yr Beef Bull calves 13-16 mos. DairySteers Beef Bull calves DairyOxen Beef Bulls Not coupling DairyCattle_finishing Beef Bull calves 15-16 mos. DairyBuffaloes Beef Bulls Not coupling EweDoe_1yr Ewes RamBuck_breeding Rams LambKid_0to1yr Lambs Wethers_1yr Rams Not coupling Mares_3yrs Mares Stallions_breeding Stallions Foals_0to1yr Foals Fillies_1to3yrs Stallions 350 kg, 400 kg, 450 kg Colts_1to3yrs Stallions 350 kg, 400 kg, 450 kg Colts_3yrs Stallions 500 kg, 550 kg, 600 kg Geldings Stallions Equation 2-S3: The equation used to estimate the amount of energy supplied by fodder to each livestock species in each district in 2015. 43 𝐶𝑚𝑟𝑓𝑙 =𝑃𝑚𝑟𝑓 ∗𝐶𝑜𝑓𝑙 𝐶𝑜𝑓 where: 𝐶=𝑐𝑜𝑛𝑠𝑢𝑚𝑝𝑡𝑖𝑜𝑛 (𝑀𝐽) 𝑃=𝑝𝑟𝑜𝑑𝑢𝑐𝑡𝑖𝑜𝑛 (𝑀𝐽) 𝑚=𝑓𝑎𝑟𝑚 𝑡𝑦𝑝𝑒 𝑟=𝑑𝑖𝑠𝑡𝑟𝑖𝑐𝑡 𝑓=𝑓𝑜𝑑𝑑𝑒𝑟 𝑡𝑦𝑝𝑒 𝑙=𝑙𝑖𝑣𝑒𝑠𝑡𝑜𝑐𝑘 𝑡𝑦𝑝𝑒 𝑜=𝑟𝑒𝑔𝑖𝑜𝑛 𝑤ℎ𝑒𝑟𝑒 𝑜∋𝑟 Table 2-S4: Conversion ratios used to convert kg to MJ (Forage On-Line, 2009). Production of each fodder type is recorded at the district level for each farm type. Consumption of each fodder type by each livestock species is recorded at the regional level (all farm types combined). The consumption of each fodder type by each livestock species at the district level for each farm type was estimated by multiplying the production of each fodder type by the consumption of each fodder type by each livestock species and dividing by the total consumption of each fodder type by each livestock species. Feed class ME (MJ/kg) FeedGrain 12.98 FeedLegume 13.2 SilageNonCorn 1.98 FeedRoot 1.65 FeedMelon 0.88 FeedCorn 2.2 CornGreenFodder 2.2 CornSilage 2.2 Hay 5.72 HayPasture 5.72 HayPastureCultivated 5.72 HayPastureSeeded 5.72 HayPastureNatural 5.72 HayAnnualGrass 5.61 HayPerennialGrass 5.61 GrassFodder 2.75 GrassFodderSeeded 2.53 GrassFodderNatural 2.86 GrassFodderAnnual 1.87 GrassFodderPerennial 2.75 SeedGrassAnnual 3.41 SeedGrassPerennial 3.41 GrazingGrassAnnual 2.53 GrazingGrassPerennial 2.75 44 Figure 2-S3: Side-by-side comparison with the results of a) Fetzel et al. (2017) and b) our model. Note that the units are not directly convertible. However, using very different methods and inputs, both maps show similar distributions of relative grazing activity. White areas in both maps represent unutilized area. a) Reproduced with permission of the authors. 51 the study region. We obtained annual livestock numbers at the district level from the Kazakh National Statistics Agency (KazStat, 2019). To determine the energy demand of the livestock, we used the 2006 agricultural census to divide the herds into age and function groups (e.g. beef heifers) (KazStat, 2008). We then applied recommended energy intake values from the Ministry of Agriculture (Zhazylbekov et al., 2008). We summed these values to create a total livestock energy requirement for each district. We used district-level gross yields (KazStat, 2019) to determine annual fodder production, which we converted to energy and then subtracted from total livestock energy demand, resulting in the total energy demand of livestock from grazing. We then converted the annual grazing demand to units of energy per area grassland for each district. We introduced a lag of one year to account for the “end of year” value of grazing demand (Table 3-). 3.2.3 Climate indicators Monthly precipitation and average temperature values were obtained from the Climate Research Unit (CRU) (Harris et al., 2014). Annual growing degree days (base 5°C) (Wouters, 2021) and monthly relative humidity (Hersbach et al., 2018) were taken from ERA5 reanalyses. The gridded data were averaged to produce a single value for each district. Here we were interested in annual and seasonal values—spring (March-May), summer (June-August), fall (September-November), and winter (December-February), as well as fire season (May-October), and fire off-season (November-April). For simplicity we refer to these variables as climate, for both annual and long-term analyses. For precipitation, the sum of the given period was calculated. For temperature and relative humidity, the mean of the given period was calculated. There were 124 districts and 19 years, of which we used 18 after excluding the lag year, hence 2,232 observations of each variable. Lagging the climatic variables by one year did not improve the model, therefore the current-year climatic variables were combined with the previous year’s grazing demand. 3.2.4 Fire indictors We used the Terra and Aqua combined MCD64A1 v6 Burned Area product to estimate the annual area burnt in the study area (Giglio et al., 2015). We stacked the monthly MODIS 500m resolution images to create an annual binary time series of burned or unburned areas. We used the land-cover map of Klein and others (2012) to mask cultivated areas, artificial surfaces, forests, bare areas, ice and snow, and waterbodies. The resultant grassland map, along with an administrative district map (GADM, 2018), allowed us to calculate the proportion of grassland burnt in each district for each year. 3.2.5 Statistical analysis We used a binomial generalized linear mixed-effects model (GLMM) to regress the predictors against the observed values of burned area (Bolker et al., 2009; Warton & Hui, 2011). Fixed-effects variables in our models were annual grazing demand (MJ/km2), annual cumulative growing degree days, and seven aggregations of total precipitation (mm), average temperature (°C), and relative humidity (%): annual, fire 52 season, off-season, spring, summer, fall, and winter. We included administrative district identifiers as random effects to account for the non-independence of repeated sampling of burnt areas within the same district in different years. We assume topography doesn’t change over the study period, and thus slope— an important driver of fire—is accounted for at the district level as a random effect. All variables were normalized to compare different units and different scales of magnitude. The binomial regression followed the form 𝑝= 𝑒𝛽0+𝛽1𝑥+𝛽2𝑥2 1+𝑒𝛽0+𝛽1𝑥+𝛽2𝑥2 where 𝑝=𝑝𝑟𝑜𝑝𝑜𝑟𝑡𝑖𝑜𝑛 𝑏𝑢𝑟𝑛𝑡 and 𝛽0+𝛽1𝑥+𝛽2𝑥2 is the quadratic (or linear if reduced to 𝛽0+𝛽1𝑥) regression expression (Crawley, 2013). We determined the goodness-of-fit for the aforementioned combinations of variables using conditional and marginal R2. Marginal R2 measures variance explained by fixed factors, while conditional R2 measures variance explained by both fixed and random factors (Nakagawa & Schielzeth, 2013). Akaike Information Criterion (AIC) was calculated to determine model strength (Mazerolle, 2006). The calculations were performed using the “arm” and “lme4” packages within R (Gelman et al., 2022). In addition to single-variable models, we explored multi-variable models to determine if combining variables produced better models. 3.3 Results 3.3.1 Temporal trends in burned area Between 2001 and 2019, the annual burned area exceeded 5% (about 74,000 km2) of the total grassland area in seven years (Figure 3-2a), of which six years occurred between 2001 and 2010. The Kazakh steppe has experienced a general downward trend in burned area of 3,670 km2 per year over these 19 years, though with high variation between years. Precipitation experienced a general decrease during the 2000s, followed by large inter-annual variation throughout the 2010s (Figure 3-2b). Temperature was relatively stable and relatively high during the 2000s, before also entering a period of larger inter-annual variability in the 2010s. Grazing demand has risen steadily (mirroring the rebounding livestock numbers since circa 2000) (Figure 3-2c). The observable break in 2011 is an artifact of a referendum in statistical counting (Nazarbayev, 2010), and is accounted for by the GLMM. 53 Figure 3-2. (a) Annual burned area extent on grassland in our study region (Giglio et al., 2015). The linear trendline is for visualization purposes only. (b) Average annual precipitation (blue) and average annual relative humidity (green) across the study region (Harris et al., 2014; Hersbach et al., 2018). (c) Annual grazing demand of grazing livestock in the study region. The energy equivalent was calculated using annual national statistics bulletins (KazStat, 2019), the 2006 agricultural census (KazStat, 2008), and recommended intake values (Zhazylbekov et al., 2008). See Hankerson and others (2019) for calculations. (d) Average annual temperature and average annual growing degree days (Harris et al., 2014; Wouters, 2021). 3.3.2 Spatial trends in burned area Across the Kazakh steppe, the most frequently burned areas occurred across the middle and northeast of the study area, bordering the predominately crop-producing region in the northcentral (Figure 3-3). Of the grassland in the study area over the 19-year study period, 42% burned at least once, 21% at least twice, 10% at least three times, and 4% four or more times. 54 Figure 3-3. Frequency of fires (measured as the number of years with burned area) across the study area from 2001-2019. Only burns on grasslands are shown. Gray area shows the rest of Kazakhstan, with neighboring countries (GADM, 2018). 3.3.3 Analyses of determinants in fire patterns and trends We first tested our assumption of restricting the area of interest to grasslands available for pasture. We tested three different scenarios, first where grazing demand can be spread across all land types, second where grazing demand can only be spread across grasslands and croplands (grazing is known to occur on croplands (Coughenour et al., 2008)), and third where grazing demand can only be spread across grasslands. Using grasslands only for the distribution of grazing demand had the highest R2 values (Table 31. Marginal and conditional R2 values for model results of three different land use masks with and without a one-year lag of burned area with respect to grazing demand (Nakagawa & Schielzeth, 2013).). We then examined the lag effect of the variables, to determine whether burned area is more a result of the conditions in the current year or in the previous year. Applying a one-year lag of burned area with respect to precipitation, temperature, relative humidity, growing degree days or combinations of them did not increase the models’ R2 value. However, applying a one-year lag of burned area with respect to grazing demand increased the R2 value by approximately 0.2. This makes sense considering that the livestock statistics were reported “at end of year”. Therefore, we chose to use the climatic variables of the current year in combination with the previous year’s grazing demand, which has precedent in similar regions (Dubinin et al., 2011). Table 3-1. Marginal and conditional R2 values for model results of three different land use masks with and without a one-year lag of burned area with respect to grazing demand (Nakagawa & Schielzeth, 2013). Set of variables Marginal R2 Conditional R2 Grazing on all land types 0.157 0.369 Grazing on grasslands and croplands 0.215 0.424 Grazing on grasslands 0.295 0.522 Grazing on all land types with lag 0.327 0.526 Grazing on grasslands and croplands with lag 0.404 0.612 Grazing on grasslands with lag 0.505 0.749 55 We then fitted the GLMMs with all possible combinations of grazing demand and climate variable combinations as predictors, and the proportion burned area per grid cell as the response variable (over 16000 models were analyzed, a complete list can be provided upon request). An analysis of R2 values produced the highest values when burned area was a function of the previous year’s grazing demand. Figure 3-4 plots the normalized values of precipitation, temperature, relative humidity, growing degree days, and grazing demand (previous year) against the proportion burned in a given district and year. For each set of climate variables, we chose the best-performing single-variable model according to AIC. Interestingly, longer time periods explained burned area better, either the annual average for temperature, or the fire season values for precipitation and relative humidity. Figure 3-4. Burned area of a district-year with fitted binomial regression for normalized values of (a) fire season precipitation (P), (b) average fall relative humidity (RH), (c) average fall temperature (T), (d) annual growing degree days (DD), and (e) annual grazing demand (G). We found that any model that included grazing demand greatly increased both marginal and conditional R2 values compared to models without that variable (Figure 3-5). The best combination of solely climatic variables was the interaction between summer precipitation and fall relative humidity combined with annual temperature interacted with growing degree days (Figure 3-5k). However, whenever grazing demand was combined with at least two other variables, the R2 values were routinely better than any model with solely climatic variables. 56 57 Figure 3-5. Marginal and conditional R2 for increasingly complex models. Asterisks are applied to the best model of its type for each climate and grazing variable. For example (h) has only two graphs, the first is the best-performing four-variable additive model that has relative humidity in it, the second graph is shared among every other variable as their best-performing four-variable additive model. The overall best-performing model (by AIC value) is the five-variable model including interacting summer precipitation & fall relative humidity, interacting annual temperature & growing degree days, and grazing demand (k). ann=annual, ssn=fire season, off=fire off-season, spr=spring, smr=summer, fal=fall, win=winter. 3.3.4 Regional analyses We evaluated each of the eight ecoregions in the study area separately to determine possible variations in burned area determinants (two ecoregions, the Junggar Basin semi-desert and the Emin Valley steppe, encompass only a single district, and were therefore not analyzed). Per ecoregion, both the best set of predictive variables (determined by AIC) and the model with the highest conditional R2 was determined (Table 3-2). The R2 values varied substantially between ecoregions (cf. Figure 3-). The Kazakh steppe ecoregion has the largest number of districts, but only moderate R2 values, with a quadratic model produced the highest R2 values. The Kazakh upland steppe covers only three districts and encompasses predominately cropproducing regions, where the burned area is small. Here an interactive model produced the highest R2 values, though conditional R2 is the lowest of any ecoregion. The Kazakh semi-desert is the second-largest ecoregion and exhibited a high degree of variation and uncertainty. However, accounting for spatial distribution through inclusion of random effects (districts) greatly increased the conditional R2 with a quadratic model. The Pontic steppe was the only ecoregion without temperature in its best R2 model. The Caspian lowland desert was the only ecoregion without precipitation in its best R2 model, and had the 58 highest R2 values of any ecoregion. The Altai steppe and semi-desert is the second smallest ecoregion and did not have grazing demand in its best R2 model. Relative humidity was the only variable found in every ecoregion’s best R2 model. Notably, the best predictive power (valuated by AIC) was obtained by including all variables in all ecoregions, with the two most well-represented ecoregions (Kazakh steppe and Kazakh semi-desert) having the best associated R2 values. The ecoregions can be split roughly into two groups, the “wet” group—Kazakh steppe, Kazakh upland steppe, and Altai steppe and semi-desert—and the “dry” group—Kazakh semi-desert, Pontic steppe, and Caspian lowland desert. The “dry” group are situated in the south and west of the study area, and share a number of characteristics, including many of the lowest precipitation and highest temperature values. In these, burned area shows a stronger relationship with spring, summer, and fire season climate variables. The “wet” group are in the north and east of the study area and have the highest precipitation and lowest temperature values. In these, winter and spring climate variables are more common in the best-performing R2 models. These groups also share the type of predictive model (AIC), with the “wet” group preferring interactive models and the “dry” group preferring quadratic. Perhaps most notably, when analyzing on the scale of ecoregions, the seasonality of climate variables are revealed as many more of the shorter time periods (spring, summer, fall, and winter) appear in the best-performing models. Table 3-2. Models with the lowest AIC value and the highest conditional R2 for each ecoregion. Ecoregion № districts Value Best-performing model R2m, R2c Kazakh steppe 73 AIC R2c Pssn*RHfal, Toff*DDann, Gann Pwin2, Tspr2, RHsmr2, DDann, Gann 0.39, 0.60 0.54, 0.75 Kazakh upland steppe 3 AIC R2c Pwin*RHssn, Tann*DDann, Gann Pwin*DDann, Tsmr*RHfal, Gann 0.42, 0.46 0.47, 0.64 Altai steppe & semi-desert 6 AIC R2c Pspr*DDann, Twin*RHoff, Gann Pspr2, Twin2, RHann 0.28, 0.37 0.47, 0.81 Kazakh semi-desert 24 AIC R2c Pssn2, Toff2, RHssn2, DDann2, Gann Pssn2, Tann, RHssn2, Gann 0.23, 0.67 0.26, 0.74 Pontic steppe 8 AIC R2c Psmr2, Tsmr2, RHfal2, DDann2, Gann Psmr*RHspr, Gann 0.22, 0.52 0.60, 0.80 Caspian lowland desert 8 AIC R2c Psmr2, Toff2, RHfal2, DDann2, Gann Tsmr2, RHspr2, DDann, Gann 0.26, 0.53 0.50, 0.97 3.4 Discussion Climate factors and grazing both have been shown to be important determinants of fire patterns and trends, but their relative importance has often been elusive. Understanding their roles is key as climate change progresses. Making use of the natural experiment of rapid, widespread, and spatially heterogeneous declines in livestock numbers in the steppes of post-Soviet Kazakhstan, a global fire hotspot (Archibald et al., 2013; Hantson et al., 2013), we assessed the relative importance of changes in grazing demand and climate variables on burned area at unprecedented spatial and temporal detail. Our analyses yield three major insights. First, fire regimes changed markedly, with exceptionally high frequency and extent of fires in the 2000s followed by a decline in the 2010s, corroborating earlier findings for smaller 59 subregions and from other parts of the post-Soviet sphere (Dara, Baumann, Hölzel, et al., 2020; Dubinin et al., 2010; Hao et al., 2021). Second, there was a strong negative association between grazing demand and burned area, suggesting that restoring grazing regimes—either with domestic livestock or natural grazers— can suppress burning in Eurasian steppes. Third, changing livestock grazing demand after the breakdown of the Soviet Union explained fire dynamics better than both broad-scale climate patterns and inter-annual climate variation, highlighting the importance of grazing and, more generally, the need for better data on grazing pressure in grasslands. Answering our first research question, we found that burned area extent varied greatly from year to year (Figure 3-2a). Linear regression showed an average decrease in burned area of 3,670 km2/year from 20012019. As percent change (about -5.4%), this is well above the global trend of about -1.6% over the same time period (Zheng et al., 2021). We also found that the most frequently burned areas are located predominantly along the border between the Kazakh steppe and the Kazakh semi-desert, as well as in the Pontic steppe (cf. Figure 3- & Figure 3-3). This agrees with the median fire return interval for this region found by Archibald and others (2013), which extends westward across the Southern Federal District of Russia (Dubinin et al., 2010). The negative association between grazing and fire in our study area is in line with Dara and others (2020), who found fire occurrence to be correlated with livestock presence in a subset of our study area. Similarly Dubinin and others (2011) found livestock population to be negatively associated with burned area in a region adjacent to our study area. We also find agreement with Hao and others (2021), who looked at a similar time period across a large area in northern Eurasia and found grazing demand to be strongly correlated with declining burned area. Importantly, we go beyond prior work in that the relationship between fire and grazing has neither been studied across a broad range of environmental conditions and natural vegetation zones while using high-resolution data on grazing demand and fire activities, nor across such a length of time while using sub-annual data. Our most important finding was that fire was more strongly related to grazing than climate factors. We have shown that grazing demand is strongly, and negatively, correlated with burned area (Figure 3-4), and can be used alone or in cohort with climate variables to produce a robust model (Figure 3-5). Indeed, as grazing demand has steadily increased over the period 2001-2019, burned area has become less variable and generally smaller. While we did not measure forage quantity or quality, our findings corroborate the linkages between large grazers and fire occurrence (Reid et al., 2022). We identified trade-offs between conservation, fire suppression, and livestock production, which can inform policy makers and local governments looking to optimize the allocation of resources assigned to fire prevention and control. The results corroborate that livestock grazing can be used to mitigate fire occurrence on temperate grasslands, which more broadly can support mitigation of greenhouse gas 60 emissions from fire and promote production from grazing livestock in regions with large grassland resources at lower environmental costs. This is timely for Kazakhstan and the former Soviet Union in general as they start to release carbon that had been sequestered in the years immediately following the dissolution of the Soviet Union (Schierhorn et al., 2019). A few limitations need mentioning. First, there were several variables that may have added explanatory value. Topography, wind patterns, fuel continuity, and access to ignition sources are all unaccounted-for variables that also determine fire frequency and extent (IAFC & NFPA, 2019). Our intent was not to create a definitive model for ranking fire drivers, but rather to investigate the rank of grazing demand in comparison to some of the most important climatic variables (Oliveira et al., 2012). Second, the MODIS burned area product may not capture all burned areas, especially small burns. However, small burns are not characteristic of our study area, and indeed, temperate grasslands and xeric shrublands (the biomes of our study area) are among the most highly accurate biomes in terms of burned area estimates (Boschetti et al., 2019). Third, despite our use of district-level statistics, our spatial representation of grazing demand remains a coarse aggregation, with no way to definitively allocate livestock within the district (Hankerson et al., 2019). Finally, small sample sizes reduced the goodness-of-fit for some ecoregions, even when including all variables and quadratic terms (Table 3-2). Some ecoregions, especially in eastern Kazakhstan (cf. Figure 3-), are small, isolated, and unique, not lending themselves well to analyses at the district level. Larger ecoregions, however, produced robust results. Our work also has major policy implications. As much of the steppe is too arid for profitable crop production, livestock grazing has little competition from other land uses for Kazakhstan’s grassland resources (Fetzel, Havlík, Herrero, & Erb, 2017), a bias which may strengthen under continuing climate change (Weindl et al., 2015). Additionally, Kazakhstan is not considered a priority area for climate change mitigation and adaptation (Bonilla-Cedrez et al., 2023), suggesting that it is a suitable candidate for livestock expansion. As a predominately pastoralist society, the mitigation of fire occurrence on the Kazakh steppe could at least in part be achieved through the management of grazing practices, providing that livestock numbers continue to grow in order for large-scale reduction of fuel loads, possibly with the help of targeted governmental programs. Any program implemented by the government would have to take into account the vast differences in grazing practices between large agricultural enterprises with tens or hundreds of thousands of animals, medium-sized private farms—which are the fastest growing farm type— and small households with a handful of animals, which is where the majority of grazing livestock in Kazakhstan are still raised, but whom are rarely able to take advantage of any of the government’s current support programs (Kerven et al., 2021). Despite a potential revival of livestock husbandry in the steppes of Kazakhstan, inevitably there will also be large areas where livestock grazing remains infeasible (e.g., too remote, too marginal) or prohibited (e.g., inside protected areas) (Dara, Baumann, Freitag, et al., 2020; Kamp et al., 2016; Kerven et al., 2016). These 67 Figure 4-2: Selected properties of beef and its less popular substitutes. All data from Paleari et al. (2003) except lamb, from Paleari et al. (2006). (a) Physical composition of the meat. (b) Fatty acids composition, SFA—saturated fatty acids, MUFA— monounsaturated fatty acids, PUFA—polyunsaturated fatty acids. (c) Essential amino acids (not calculated for lamb) and cholesterol levels, with standard error. Meats were cured, fermented, and dried in a rigorously consistent manner following the traditional preparation of ‘bresaola’, an Italian dried-meat product. 4.3 Land use Feed-use efficiency is an important indicator of the environmental impact of meat production. A number of indicators can be used, which highlight different footprints. Feed conversion ratio measures all feed intake (as dry matter) to mass of product (meat), energy conversion ratio measures energy intake to energy in product, protein conversion ratio measures protein intake to protein in product, and nitrogen use efficiency measures the percent of nitrogen intake retained in the product (Hou et al., 2016). For conversion ratios, lower is better, while for use efficiencies, higher is better. In the relatively intensive production systems of the EU, non-dairy cattle (i.e. beef cattle) have conversion ratios 4-7 times higher than pork or chicken, while having a nitrogen use efficiency 3-4 times less (Hou et al., 2016). Similar differences in feed conversion ratios were found in the US (Mekonnen et al., 2019). However, beef cattle obtain the majority of their intake requirement through grazing and forage (around 80% in the EU & US), energy sources that are not in direct competition with human-edible feed, whereas pork and chicken receive almost exclusively humanedible feed (Hou et al., 2016; Mekonnen et al., 2019). Thus, there is a niche that beef fills, providing human- 68 edible food (meat) from land otherwise not capable of producing human-edible food (or at economically infeasible levels) (van Zanten et al., 2018). Figure 4-3 shows the 2015 distribution of cattle and horses, along with the shaded outline of four biomes that are commonly grazed (Dinerstein et al., 2017): Temperate broadleaf and mixed forests (eastern US, Europe, and East Asia), tropical and subtropical grasslands, savannas, and shrublands (Brazil, Africa, and northern Australia), temperate grasslands, savannas, and shrublands (central US, Argentina and across Eurasia), and montane grasslands and shrublands (the Andes, East Africa and the Tibetan Plateau). Obviously, this is not a perfect representation of areas to target for expansion, for example there are heavily wooded areas in the temperate broadleaf and mixed forests not suited for grazing, and montane grasslands range from the lush Ethiopian grasslands to the frigid and barren eastern Himalayan alpine shrub. Likewise, some eligible regions are excluded from this map: The Aravalli west thorn scrub forests in Pakistan are very heavily grazed, but many other regions in the deserts and xeric shrublands biome are devoid of livestock. The shaded area covers nearly 50 million km2, or about one third of the land surface. It is estimated that livestock production currently utilizes about 33 million km2 for pasture (X. Xu et al., 2021). 69 Figure 4-3: The Gridded Livestock of the World for cattle and horses (Gilbert et al., 2022a, 2022b). Shaded areas are broad, variegated biomes that may have desirable characteristics for horse grazing (Dinerstein et al., 2017). Some of the potential grazing land in Figure 4-3 is currently being used as cropland. Livestock production uses about 20% of the 16 million km2 of global croplands for feed (FAO, 2024; X. Xu et al., 2021). Additionally, feed production consumes about 20% of agricultural blue water use (mainly irrigation) (Heinke et al., 2020). Horses can graze on the same land that cattle do, as well as more marginal and distant pastures due to their greater mobility, sturdiness, and the higher proportion of grass in their diet (Cymbaluk, 1994; National Research Council, 1981; Willekes, 2013). Moreover, studies have shown that introducing human-edible cereals during finishing can even be wasteful economically and harmful to horses’ welfare (Raspa et al., 2021). Their ability to dig through snow and break ice with their hooves are skills that cattle lack and allow horses to graze year-round at latitudes and altitudes where cattle need to be sheltered and fed entirely with feed and fodder (Gudmundsson & Dyrmundsson, 1994). 70 4.4 Greenhouse gas emissions Livestock produce approximately 21% of human-induced carbon dioxide-equivalent (CO2e) GHG emissions (X. Xu et al., 2021). Methane (CH4) is by far the largest source of livestock GHG emissions, and cattle are by far the largest contributors to methane production, accounting for 77% of all non-CO2 emissions from livestock production. The amount of methane released to produce a kilogram of beef varies greatly depending on the production system and region, with intensive systems in developed nations producing the fewest GHGs per kg protein and extensive systems in developing countries producing the most, especially in arid regions (Herrero, Havlík, et al., 2013). Dairy beef emits about 25% less CO2e GHG than traditional beef cow/calf systems (Nguyen et al., 2010). Greenhouse gas emissions from livestock depend on management practices. Horses are also hindgut fermenters, meaning that they produce substantially less methane than foregut fermenters (such as cattle, sheep, and goats) (Hristov et al., 2013). Horses emit an estimated 18 kg methane per head per year through enteric fermentation (Crutzen et al., 1986). Cattle are treated to more regionalized precision, with dairy cattle ranging from 46 kg/head/year in Africa and the Middle East to 128 kg/head/year in North America. Other cattle (mainly beef cattle) range from 27 kg/head/year in the Indian Subcontinent to 60 kg/head/year in Oceania (Dong et al., 2006). In 2021, average global methane emissions (enteric fermentation plus manure management) from dairy cattle were 74.7 kg/head/year, other cattle 45.4 kg/head/year, and horses 19.5 kg/head/year (FAO, 2023a). While both are predominately grazing species, cattle and horses differ in important meat-producing aspects such as average daily gain (ADG), live weight at slaughter (LWS), and dressing percentage (DP%), which is a measure of meat yield as a percent of LWS. Calves up to 16 months typically have higher ADG than foals (López et al., 2019). After 16 months, however, horses overtake cattle in ADG, and by three years are more massive than cattle (Sarriés & Beriain, 2005). In addition to the increased LWS, DP% is also higher for horses, around 65%, vs. around 55% for cattle (Coyne et al., 2019; He et al., 2005). Moreover, DP% in horses increases with age, while for cattle it decreases (Janiak et al., 2017; Mantovani et al., 2014). This leads the difference in emissions per kg meat to be even greater than per head. In Kazakhstan, horses and cattle raised for meat are slaughtered at similar ages, with horses slightly outweighing cattle. In 2022, the average cattle LWS was 339 kg, while the average horse was 347 kg (KazStat, 2023). With an average age of 18 months, methane emissions per kg meat were almost three times higher for beef (0.37 kg/kg) compared to horsemeat (0.13 kg/kg). 4.5 Secondary sources of horsemeat The practice of slaughtering horses that have reached the end of their useful life in a sector other than meat production are often unfit for human consumption. Racehorses, especially, have been given many substances that are illegal to administer to animals meant for human consumption (Anderson, 2015). 71 Horses are classified as companion animals by the American Federal Drug Administration, meaning all drugs approved for pets can be administered. Several of these are banned for food-producing animals, and poor oversight of the horse slaughter industry has led to many horses entering the human food market that shouldn’t have (Weber et al., 2023). Nevertheless, the market for horses from the non-food sector could be structured and expanded such that properly documented horsemeat therefrom could safely enter the human food chain (Saastamoinen, 2015). 4.6 Other horse products A discussion of horse production would not be complete without mentioning kumys (airag in Mongolian), a drink made from fermented mare milk. Historically produced from Hungary to China, kumys remains a popular dairy product in Mongolia, Central Asia, and neighboring Russian provinces (Langlois, 2011). Fresh mare milk is similar to human milk and is an option for infants allergic to cow milk (Malacarne et al., 2002). However, its very similarity to human milk, low fat content, and inability to make cheese severely limits mare milk’s ability to replace cow milk in the general public (Langlois, 2011). Kumys, on the other hand, is often produced concomitantly with horsemeat, and competes with kefir (fermented cow, goat, or sheep milk) in the regions where it is popular. Considered an “acquired taste”, kumys nevertheless has been gaining in popularity outside its traditional home due to its purported health benefits (Singh & Shah, 2017), and could accompany expanding horsemeat production (Askarov et al., 2020). 4.7 Outlook Besides overcoming the tremendous taboo surrounding horsemeat consumption, there remains another, more tangible hurdle. Horses require more grazing land per kg meat because they are less efficient digesters than cattle, especially of higher-fiber diets (Chenost et al., 1985). While access to more pastures for longer periods helps to mitigate the additional land demand, the transition from beef to horsemeat would necessarily coincide with a decrease in red meat consumption. This is a trend that is already being realized and is encouraged from both a health and an environmental standpoint (Smil, 2014). Indeed, reduction in meat—particularly bovine meat—consumption is useful and important to avoid crossing several planetary boundaries (Parlasca & Qaim, 2022). Cropland freed from cattle feed production can be used to further mitigate the additional land demand through high-quality hay and pasture, and further land sparing could occur through upcycling of agricultural byproducts in animal feed (Govoni et al., 2023). There is currently a paucity of scientific literature concerning horses raised for meat. Several knowledge gaps need to be explored and given the same rigorous treatment that beef production has received over the last century. How to optimize horses’ diet for meat production—what, if at all, would feedlot production look like? How to reorganize cropland production, as a shift to horsemeat production would free up a lot of land previously used for cattle feed—should we use it to feed horses, other livestock, grow food, biofuel, or increase natural conservation? How different production systems affect GHG emissions— 72 could horse emissions be reduced even further with higher-energy grain supplements? How to conserve and preserve biodiversity and ecosystem services on marginal pastures—what fraction of Figure 4-3 is acceptable for expansion? Animal welfare has been an increasingly important aspect of livestock production, with cattle, poultry, and pork receiving the lion’s share of attention. For example, customers are willing to pay premiums for “cage-free” eggs or “grass-fed” beef. Horses raised for meat have recently also been treated with studies on welfare indicators (Raspa et al., 2020). While we have elaborated on the environmental benefits of converting beef production to horsemeat, for many consumers, the battle will be fought on the dinner table. Horsemeat is substantially more healthy than beef, with a taste and texture similar, if not even superior to beef (Jaskari et al., 2015). On top of extolling the health benefits, premium-quality horsemeat could also improve its market share by better presentation and advertising. Besides the well-known sausages and dried meats in hippophagic regions, raw horsemeat could be graded by cut in the same way that beef is, increasing the value and desirability of better cuts. Restaurants that introduce and prepare horsemeat as gourmet could do much to help overcome the negative stigma that horsemeat has in many countries. We have explored the potentials and externalities of expanding horsemeat production while reducing beef production. In a novel approach, we have combined social aspects (public perception, religious taboos), environmental aspects (land and water use, GHG emissions), production aspects (feed requirements, grazing patterns), nutritional aspects, and animal welfare. Our goal was to create a holistic and realistic analysis of horsemeat production, to define the state of the art and provide a possible, partial solution to the problem of livestock’s long shadow (Steinfeld et al., 2006). 73 Chapter 5: Synthesis 74 5.1 Summary The overarching goal of this thesis was to advance the knowledge and understanding of land use— especially pastureland use—and livestock production in Kazakhstan. Several facets were explored that contribute to knowledge gaps in the relationship between livestock productivity, grazing patterns, ecological functions of grasslands, and environmental footprints of meat production. Kazakhstan was chosen as the study area for a number of reasons: 1) it contains huge tracts of uninterrupted grazing land; 2) it has significant herds of four different grazing livestock species; 3) those herds have seen dramatic rises and falls in populations over a short and recent time period; 4) is at risk of desertification and other climate change-related effects; and 5) has potential to increase meat production without taking land away from crop production. 5.1.1 Research questions and answers 5.1.1.1 Question I: Where, and to what extent, could increases in Kazakh livestock production take place? Chapter 2 developed a model to distribute grazing demand across Kazakhstan’s vast grassland resources. Much of Kazakhstan was found to be ungrazed or grazed at very low intensities. However, there is ample evidence for localized overgrazing around settlements across the country, and extensive, proliferant overgrazing likely occurs in the south and southeast, corresponding to the highest livestock densities and despite high grassland productivity. In the areas of low-intensity grazing, large increases in livestock production could take place without changing the current state of pastoralism. With repaired and improved infrastructure, however, Kazakhstan could increase meat and dairy production by an order of magnitude. Improvement projects are well underway in the northcentral, where fodder crop production supports large agricultural enterprises and feedlot-style beef and especially dairy. An area of interest for potential expansion that has not begun to be tapped is in the northwest, West Kazakhstan and Aktobe Oblasts. Beef and dairy cattle understandably attract the most attention, but Kazakhstan has the capacity and pathways to further vitalize the sectors of other grazers: sheep, goats, horses, and camels are all economically significant and each could step up if global demand for cattle products were to wane due to concerns over the environmental footprint of cattle. 5.1.1.2 Question II: To what extent are livestock in Kazakhstan actors in fire regimes? Chapter 3 utilized binomial generalized linear mixed-effects models to analyze the correlation that grazing demand has on burned area extent. The magnitude of this correlation was then compared to other common predictors of burned area extent: precipitation, humidity, temperature, and growing degree days. The climate variables were further stratified into seasonal values, as fire is ephemeral and often driven by temporary, extreme conditions, and all possible combinations of these variables were analyzed to determine a best fit. Additive, interactive, and quadratic models were considered, however the best 75 performing models always included grazing demand. Grazing demand is strongly, and negatively, correlated to burned area extent. While this is logical and not in itself earth-shattering, the magnitude of correlation was surprising, a small increase in grazing demand corresponded to a large decrease in burned area extent. Besides the obvious conclusion that increasing livestock numbers can mitigate fire occurrence, this research suggests the efficacy of rotational grazing as a way to drastically reduce the risk of wildfire without needing to increase herd size. Improved infrastructure and incentivizing programs could be used in local mitigation schemes. As far as targeting livestock for expansion, horses are most mobile and would be best-suited to forming traveling fire-prevention bands. 5.1.1.3 Question III: Is there a path to sustainable livestock development in Kazakhstan? Chapter 4 concerns the large, growing, and detrimental impact beef and dairy cattle have on the environment. The biological processes that allow cattle to be energy-efficient digesters of grasses also emit more greenhouse gases (GHGs) per kilogram of product than any other mainstream meat or dairy source. The potential of one of mankind’s very first domesticated meat sources that has only relatively recently fallen out of favor was explored as an alternative to beef: horsemeat. Nutritionally, horsemeat is superior to beef in every measureable way: more protein, less fat, better fat, and more essential amino acids. Texturally, horsemeat is more similar to beef than any other red meat, and when raised for meat is sweeter and tenderer. Environmentally, horsemeat production is more of a mixed bag compared to beef production. Horses are hindgut fermenters, and therefore by nature emit fewer GHGs per kilogram of product than cattle, but they are less efficient digesters, requiring far more energy input in the form of grass than cattle do, which means a larger land use footprint. On the positive side, this land requirement can be satisfied using marginal grasslands not suitable for crop production, but the more marginal the land the less productive it is and the more land is required. Simply based on the land requirement, horsemeat is unable to replace beef kilogram-for-kilogram. A reduction in meat consumption or a shift to other meats is also required. 5.2 Insights Together, Chapters 2-4 offer several cross-cutting insights into the recent history and modern state of livestock production in Kazakhstan, environmental interactions, and potential for further development. First, as discovered in Chapter 2, much of Kazakhstan is ungrazed or grazed at low intensities. Livestock numbers plummeted after the Soviet Union dissolved in the early 1990s, leaving distant pastures and stations abandoned. Since the launch of the MODIS satellites in 1999 and 2002 there has been a continuous record of burned area extent, and Chapter 3 shows a strong correlation between livestock grazing and burned area extent. The high variability of burned area in Kazakhstan in the 2000s coincides with the decrease in livestock numbers, suggesting that decreased grazing allowed variation in climatic factors to play a larger role in fire occurrence. Given the current low grazing intensities in the hotbed of Kazakhstan’s 76 grassland fires, targeted increases in pasture use, either by increasing livestock numbers or by scheduled rotational grazing, can drastically reduce the threat of wildfire by removing a relatively small amount of fuel. Second, based on the grazing map from Chapter 2, much grassland near human populations is at risk of overgrazing. Moreover, there are huge swaths of unor underutilized pasture far from settlements, especially in western and central Kazakhstan. Utilizing this pasture is a difficult proposition under the current circumstances of decaying Soviet infrastructure: crumbling roads, dilapidated livestock stations, and failed wells. However, this barrier to expansion has made one assumption that need not be true: that cattle are the species to propagate. Cattle are dependent on human infrastructure, short distances to food and water, and are largely defenseless against apex predators and the bitter cold of Kazakh winters. Introduced in Chapter 4, the idea of expanding horsemeat production has many positive attributes. In relation to Chapter 2, horses could be an immediate answer to utilizing distant pastures, as they can range much further than cattle, can defend themselves from predators, and are hardy enough to survive even the dzhuts—various extreme weather conditions causing mass die-offs—that repeatedly threaten grazers on the Eurasian steppe. Third, in a grand cross-cut of Chapters 2, 3, and 4, employing a highly mobile, hardy species that requires very little cut-and-carry fodder to roam the fire-prone grasslands of the Kazakh steppe, dramatically reducing the risk of uncontrolled wildfire while utilizing grassland otherwise inaccessible for human food production is a powerful argument for the expansion of horse husbandry. There is strong emotion in many countries against the eating of horses, and in many others, horse is viewed as a lesser meat. However, Chapter 4 reveals that horsemeat is in fact superior to beef in almost every way: nutritionally, environmentally, and—perhaps most importantly—gastronomically. The potential for horsemeat operations to reduce or even supplant beef as the red meat of choice relies on the ability to (a) efficiently and sustainably utilize marginal grasslands and (b) change the majority of mindsets with tender, tasty slabs of high-quality “I can’t believe it’s not beef”. 5.3 Implications There are two overarching, universal questions associated with changes in livestock production, agricultural production in general and indeed, with any kind of production: 1) how will the economy be affected, and 2) how will the environment be affected? Economic impacts of agricultural policies are extensively studied and forecasted before a policy is implemented, continuously analyzed during its effective period, and thoroughly evaluated after its conclusion. Comprehensive assessment always remains elusive, however, as real-world policy never occurs in a vacuum, and there is no “control variable”. Nevertheless, analysis has come a long way in the last half-century, as the framework for assessing agricultural policy developed (Josling, 1974), as computational power increased dramatically (Just & Antle, 1990), as statistical databases 83 Chang, C., Chang, Y., Xiong, Z., Ping, X., Zhang, H., Guo, M. & Hu, Y. (2023). Predicting Grassland FireOccurrence Probability in Inner Mongolia Autonomous Region, China. Remote Sensing, 15(12), 2999. https://doi.org/10.3390/rs15122999 Chang, J., Viovy, N., Vuichard, N., Ciais, P., Campioli, M., Klumpp, K., Martin, R., Leip, A. & Soussana, J.-F. (2015). Modeled Changes in Potential Grassland Productivity and in Grass-Fed Ruminant Livestock Density in Europe over 1961–2010. PLoS ONE, 10(5), e0127554. https://doi.org/10.1371/journal.pone.0127554 Chen, X., Cui, Z., Fan, M., Vitousek, P., Zhao, M., Ma, W., Wang, Z., Zhang, W., Yan, X., Yang, J., Deng, X., Gao, Q., Zhang, Q., Guo, S., Ren, J., Li, S., Ye, Y., Wang, Z., Huang, J., . . . & Zhang, F. (2014). Producing more grain with lower environmental costs [Letter]. Nature, 514(7523), 486-489. https://doi.org/10.1038/nature13609 Chenost, M., Martin-Rosset, W., Boissau, J.M., Dudilieu, M., Jailler, M., L'Hotelier, L., Dubroeucq, H., Jailler, R. & Moins, G. (1985). Comparaison entre espèces (mouton, cheval, bovin) de la digestibilité et des quantités ingérées des fourrages verts. Annales de Zootechnie, 34(3), 291-312. https://hal.science/hal00888375v1 Climate Research Unit. (2019). High-Resolution Datasets Version 4.02) U. o. E. Anglia. https://crudata.uea.ac.uk/cru/data/hrg/cru_ts_4.02/ Clutton-Brock, J. (1992). Horse power: a history of the horse and the donkey in human societies. London, UK: Natural History Museum Publications. Conant, R.T. (2010). Challenges and opportunities for carbon sequestration in grassland systems: A technical report on grassland management and climate change mitigation Vol. 9. Rome, Italy: FAO. Coughenour, M., Behnke, R., Lomas, J. & Price, K. (2008). Forage Distributions, Range Condition, and the Importance of Pastoral Movement in Central Asia - A Remote Sensing Study. In R. Behnke (Ed.), The Socio-Economic Causes and Consequences of Desertification in Central Asia (pp. 45-80). Dordrecht, The Netherlands: Springer. https://doi.org/10.1007/978-1-4020-8544-4_4 Coyne, J.M., Evans, R.D. & Berry, D.P. (2019). Dressing percentage and the differential between live weight and carcass weight in cattle are influenced by both genetic and non-genetic factors1. Journal of Animal Science, 97(4), 1501-1512. https://doi.org/10.1093/jas/skz056 Crawley, M.J. (2013). Proportion Data. In The R Book (2nd ed., pp. 628-649). Imperial College London at Silwood Park, UK: John Wiley & Sons. https://doi.org/10.1002/9781118448908.ch16 Crutzen, P.J., Aselmann, I. & Seiler, W. (1986). Methane production by domestic animals, wild ruminants, other herbivorous fauna, and humans. In Tellus B: Chemical and Physical Meteorology (Vol. 38, pp. 271284). https://doi.org/10.3402/tellusb.v38i3-4.15135 Cymbaluk, N.F. (1994). Thermoregulation of horses in cold, winter weather: a review. Livestock Production Science, 40(1), 65-71. https://doi.org/10.1016/0301-6226(94)90266-6 Dara, A., Baumann, M., Freitag, M., Hölzel, N., Hostert, P., Kamp, J., Müller, D., Prishchepov, A.V. & Kuemmerle, T. (2020). Annual Landsat time series reveal post-Soviet changes in grazing pressure. Remote Sensing of Environment, 239, 111667. https://doi.org/10.1016/j.rse.2020.111667 Dara, A., Baumann, M., Hölzel, N., Hostert, P., Kamp, J., Müller, D., Ullrich, B. & Kuemmerle, T. (2020). PostSoviet Land-Use Change Affected Fire Regimes on the Eurasian Steppes. Ecosystems, 23, 943–956. https://doi.org/10.1007/s10021-019-00447-w de Pauw, E. (2008). Central Asia: mean annual precipitation. GIS Unit Technical Bulletin. Aleppo, Syria: I. C. f. A. R. i. t. D. A. (ICARDA). DeBoe, G. (2020). Impacts of agricultural policies on productivity and sustainability performance in agriculture: A literature review. (OECD Food, Agriculture and Fisheries Papers, No. 141). Paris, France. Del Bò, C., Simonetti, P., Gardana, C., Riso, P., Lucchini, G. & Ciappellano, S. (2013). Horse meat consumption affects iron status, lipid profile and fatty acid composition of red blood cells in healthy volunteers. International Journal of Food Sciences and Nutrition, 64(2), 147-154. https://doi.org/10.3109/09637486.2012.728198 Deli, W., Guodong, H., Yuguang, B. & McGilloway, D.A. (2005). Interactions between foraging behaviour of herbivores and grassland resources in the eastern Eurasian steppes. Grassland: a global resource. Plenary and invited papers from the XX International Grassland Congress, Dublin, Ireland, 26 June-1 July, 2005., Dublin, Ireland. 84 Díaz, S., Lavorel, S., McIntyre, S., Falczuk, V., Casanoves, F., Milchunas, D.G., Skarpe, C., Rusch, G., Sternberg, M. & Noy‐Meir, I. (2007). Plant trait responses to grazing – a global synthesis. Global Change Biology, 13(2), 313-341. https://doi.org/10.1111/j.1365-2486.2006.01288.x Dinerstein, E., Olson, D., Joshi, A., Vynne, C., Burgess, N.D., Wikramanayake, E., Hahn, N., Palminteri, S., Hedao, P., Noss, R., Hansen, M., Locke, H., Ellis, E.C., Jones, B., Barber, C.V., Hayes, R., Kormos, C., Martin, V., Crist, E., . . . & Saleem, M. (2017). An Ecoregion-Based Approach to Protecting Half the Terrestrial Realm. BioScience, 67(6), 534-545. https://doi.org/10.1093/biosci/bix014 Dobranić, V., Njari, B., Večkovec, A. & Kadivc, M. (2008). Horsemeat and hippophagia. Meso, 10(4), 288292. https://hrcak.srce.hr/35575 Dong, H., Mangino, J., McCallister, T.A., Hatfield, J.L., Johnson, D.E., Lassey, K.R., Aparecida de Lima, M. & Romanovskaya, A. (2006). Agriculture, Forestry and Other Land Use. Chapter 10: Emissions from Livestock and Manure Management. (IPCC Guidelines for National Greenhouse Gas Inventories). Hayama, Japan: IGES. Dubinin, M., Luschekina, A. & Radeloff, V.C. (2011). Climate, Livestock, and Vegetation: What Drives Fire Increase in the Arid Ecosystems of Southern Russia? Ecosystems, 14(4), 547-562. https://doi.org/10.1007/s10021-011-9427-9 Dubinin, M., Potapov, P., Lushchekina, A. & Radeloff, V.C. (2010). Reconstructing long time series of burned areas in arid grasslands of southern Russia by satellite remote sensing. Remote Sensing of Environment, 114(8), 1638-1648. https://doi.org/10.1016/j.rse.2010.02.010 Dunkley, C.S. & Dunkley, K.D. (2013). Greenhouse gas emissions from livestock and poultry. Agriculture, Food & Analytical Bacteriology, 3(1), 17-29. Eisfelder, C., Klein, I., Bekkuliyeva, A., Kuenzer, C., Buchroithner, M.F. & Dech, S. (2017). Above-ground biomass estimation based on NPP time-series − A novel approach for biomass estimation in semi-arid Kazakhstan. Ecological Indicators, 72, 13-22. https://doi.org/10.1016/j.ecolind.2016.07.042 Eisfelder, C., Klein, I., Niklaus, M. & Kuenzer, C. (2014). Net primary productivity in Kazakhstan, its spatiotemporal patterns and relation to meteorological variables. Journal of Arid Environments, 103, 17-30. https://doi.org/10.1016/j.jaridenv.2013.12.005 Elliott, C. (2014). Elliott Review into the integrity and assurance of food supply networks—Final report: A national food crime prevention framework). London, UK: Food Standards Agency. Ellis, J. & Lee, R.-Y. (2003). Collapse of the Kazakhstan livestock sector: A catastrophic convergence of ecological degradation, economic transition and climate change. In C. Kerven (Ed.), Prospects for Pastoralism in Kazakhstan and Turkmenistan: From State Farms to Private Flocks (pp. 52-74). London, UK; New York, New York: RoutledgeCurzon. Erb, K.-H., Fetzel, T., Kastner, T., Kroisleitner, C., Lauk, C., Mayer, A. & Niedertscheider, M. (2016). Livestock Grazing, the Neglected Land Use. In Social Ecology (pp. 295-313). Switzerland: Springer. Erb, K.-H., Gaube, V., Krausmann, F., Plutzar, C., Bondeau, A. & Haberl, H. (2007). A comprehensive global 5 min resolution land-use data set for the year 2000 consistent with national census data. Journal of Land Use Science, 2(3), 191-224. https://doi.org/10.1080/17474230701622981 European Space Agency & Sinergise. (2021). Copernicus Global Digital Elevation Model. https://doi.org/10.5069/G9028PQB FAO. (2017). FAOSTAT. http://faostat3.fao.org FAO. (2022). Food balances 2010-2019: Global, regional and country trends. (FAOSTAT Analytical Brief, No. 40). Rome, Italy: Food and Agriculture Organization of the United Nations. FAO. (2023a). FAOSTAT. Emissions from Livestock. http://www.fao.org/faostat/en/#data/GLE FAO. (2023b). Pathways towards lower emissions – A global assessment of the greenhouse gas emissions and mitigation options from livestock agrifood systems. Rome, Italy: Food and Agriculture Organization of the United Nations. https://doi.org/10.4060/cc9029en FAO. (2024). FAOSTAT. Land Use. https://www.fao.org/faostat/en/#data/RL FAO. (2025). FAOSTAT. Emissions intensities. https://www.fao.org/faostat/en/#data/EI Farrell, D. (2013). The nutritional benefits of chicken meat compared with other meats. In Poultry Development Review (pp. 4). Rome, Italy: FAO. Fernández, P.D., Baumann, M., Blanco, L., Murray, F., Nasca, J., Piipponen, J., Tasquer, M. & Kuemmerle, T. (2025). Improving the estimation of grazing pressure in tropical rangelands. Environmental Research Letters, 20, 034036. https://doi.org/10.1088/1748-9326/adb445 85 Fetzel, T., Havlík, P., Herrero, M. & Erb, K.-H. (2017). Seasonality constraints to livestock grazing intensity. Global Change Biology(23), 1636-1647. https://doi.org/10.1111/gcb.13591 Fetzel, T., Havlík, P., Herrero, M., Kaplan, J.O., Kastner, T., Kroisleitner, C., Rolinski, S., Searchinger, T., van Bodegom, P.M., Wirsenius, S. & Erb, K.-H. (2017). Quantification of uncertainties in global grazing systems assessments. Global Biogeochemical Cycles. https://doi.org/10.1002/2016GB005601 Flannigan, M.D. & Wotton, B.M. (2001). Climate, weather, and area burned. In Forest Fires (pp. 351-373). Elsevier. https://doi.org/10.1016/B978-012386660-8/50012-X Foley, J.A., Ramankutty, N., Brauman, K.A., Cassidy, E.S., Gerber, J.S., Johnston, M., Mueller, N.D., O'Connell, C., Ray, D.K., West, P.C., Balzer, C., Bennett, E.M., Carpenter, S.R., Hill, J., Monfreda, C., Polasky, S., Rockstrom, J., Sheehan, J., Siebert, S., . . . & Zaks, D.P.M. (2011). Solutions for a cultivated planet [10.1038/nature10452]. Nature, 478(7369), 337-342. https://doi.org/10.1038/nature10452 Forage On-Line. (2009, 18 June 2009). Nutritive value of forage. http://www.furazh.ru/dictoth/?data=18453 Forrest, S. (2017). The age of the horse: An equine journey through human history. New York, New York: Grove Atlantic. Freitag, M., Kamp, J., Dara, A., Kuemmerle, T., Sidorova, T.V., Stirnemann, I.A., Velbert, F. & Hölzel, N. (2021). Post‐Soviet shifts in grazing and fire regimes changed the functional plant community composition on the Eurasian steppe. Global Change Biology, 27(2), 388-401. https://doi.org/10.1111/gcb.15411 Fuhlendorf, S.D. & Engle, D.M. (2004). Application of the fire–grazing interaction to restore a shifting mosaic on tallgrass prairie. Journal of Applied Ecology, 41(4), 604-614. https://doi.org/10.1111/j.00218901.2004.00937.x Fuhlendorf, S.D., Engle, D.M., Kerby, J. & Hamilton, R. (2009). Pyric Herbivory: Rewilding Landscapes through the Recoupling of Fire and Grazing. Conservation Biology, 23(3), 588-598. https://doi.org/10.1111/j.1523-1739.2008.01139.x Gade, D.W. (2000). Horses. In K. F. Kiple & K. C. Ornelas (Eds.), The Cambridge World History of Food (Vol. 1, pp. 542-544). Cambridge, UK: Cambridge University Press. GADM. (2015). Global Administrative Areas v2.8. http://www.gadm.org GADM. (2018). Global Administrative Areas v3.6. http://www.gadm.org GADM. (2022). Global Administrative Areas v4.1. http://www.gadm.org Garnett, T., Godde, C., Muller, A., Röös, E., Smith, P., De Boer, I., zu Ermgassen, E.K.H.J., Herrero, M., van Middelaar, C.E., Schader, C. & van Zanten, H.H.E. (2017). Grazed and confused? Ruminating on cattle, grazing systems, methane, nitrous oxide, the soil carbon sequestration question - and what it all means for greenhouse gas emissions). Oxford, UK: Food Climate Research Network. Gelman, A., Su, Y.-S., Yajima, M., Hill, J., Pittau, M.G., Kerman, J., Zheng, T. & Dorie, V. (2022). Data Analysis Using Regression and Multilevel/Hierarchical Models. In (Version 1.13-1) https://cran.rproject.org/package=arm Geofabrik. (2016). OpenStreetMap Data Extracts. http://download.geofabrik.de/ Geofabrik. (2024). OpenStreetMap Data Extracts. http://download.geofabrik.de/ Geyer, L.L. & Lawler, D. (2013). Yea or Neigh? The Economics, Ethics, and Utility of the Horsemeat Filet. Journal of Food Law & Policy, 9(2), 247-274. https://scholarworks.uark.edu/jflp/vol9/iss2/8 Giglio, L., Justice, C., Boschetti, L. & Roy, D.P. (2015). MCD64A1 MODIS/Terra+Aqua Burned Area Monthly L3 Global 500m SIN Grid v006. https://doi.org/10.5067/MODIS/MCD64A1.006 Gilbert, M., Cinardi, G., Da Re, D., Wint, W.G.R., Wisser, D. & Robinson, T.P. (2022a). Global cattle distribution in 2015 (5 minutes of arc). https://doi.org/10.7910/DVN/LHBICE Gilbert, M., Cinardi, G., Da Re, D., Wint, W.G.R., Wisser, D. & Robinson, T.P. (2022b). Global horses distribution in 2015 (5 minutes of arc). https://doi.org/10.7910/DVN/JJGCTX Gilbert, M., Nicolas, G., Cinardi, G., Van Boeckel, T.P., Vanwambeke, S.O., Wint, G.R.W. & Robinson, T.P. (2018). Global distribution data for cattle, buffaloes, horses, sheep, goats, pigs, chickens and ducks in 2010 [Data Descriptor]. Scientific Data, 5, 180227. https://doi.org/10.1038/sdata.2018.227 Giri, C., Zhu, Z. & Reed, B. (2005). A comparative analysis of the Global Land Cover 2000 and MODIS land cover data sets. Remote Sensing of Environment, 94(1), 123-132. https://doi.org/10.1016/j.rse.2004.09.005 GIS-Lab. (2013). OpenStreetMap data in shape-file format. http://beryllium.gis-lab.info/project/osmshp/ 86 Godde, C.M., Boone, R.B., Ash, A.J., Waha, K., Sloat, L.L., Thornton, P.K. & Herrero, M. (2020). Global rangeland production systems and livelihoods at threat under climate change and variability. Environmental Research Letters, 15(4), 044021. https://doi.org/10.1088/1748-9326/ab7395 Godde, C.M., Garnett, T., Thornton, P.K., Ash, A.J. & Herrero, M. (2018). Grazing systems expansion and intensification: Drivers, dynamics, and trade-offs. Global Food Security, 16, 93-105. https://doi.org/10.1016/j.gfs.2017.11.003 Godfray, H.C.J., Beddington, J.R., Crute, I.R., Haddad, L., Lawrence, D., Muir, J.F., Pretty, J., Robinson, S., Thomas, S.M. & Toulmin, C. (2010). Food Security: The Challenge of Feeding 9 Billion People. Science, 327(5967), 812-818. https://doi.org/10.1126/science.1185383 Gong, P., Wang, J., Yu, L., Zhao, Y., Zhao, Y., Liang, L., Niu, Z., Huang, X., Fu, H. & Liu, S. (2013). Finer resolution observation and monitoring of global land cover: First mapping results with Landsat TM and ETM+ data. International Journal of Remote Sensing, 34(7), 2607-2654. https://doi.org/10.1080/01431161.2012.748992 Goodwin, D. (2007). Horse behaviour: evolution, domestication and feralisation. In N. Waran (Ed.), The Welfare of Horses (pp. 1-18). Dordrecht, The Netherlands: Kluwer Academic Publishers. Govoni, C., D’Odorico, P., Pinotti, L. & Rulli, M.C. (2023). Preserving global land and water resources through the replacement of livestock feed crops with agricultural by-products. Nature Food, 4(12), 10471057. https://doi.org/10.1038/s43016-023-00884-w Grau, H.R., Torres, R., Gasparri, N.I., Blendinger, P.G., Marinaro, S. & Macchi, L. (2015). Natural grasslands in the Chaco. A neglected ecosystem under threat by agriculture expansion and forest-oriented conservation policies. Journal of Arid Environments, 123, 40-46. https://doi.org/10.1016/j.jaridenv.2014.12.006 Grebennikov, V.G. & Shipilov, I.A. (2012). Perennial grasses in pasture rotations (0372-3054), No. 1). Stavropol, Russia: Stavropol Scientific Research Institute of Livestock and Fodder Production. Gudmundsson, O. & Dyrmundsson, O.R. (1994). Horse grazing under cold and wet conditions: a review. Livestock Production Science, 40(1), 57-63. https://doi.org/10.1016/0301-6226(94)90265-8 Haberl, H., Erb, K.H., Krausmann, F., Gaube, V., Bondeau, A., Plutzar, C., Gingrich, S., Lucht, W. & FischerKowalski, M. (2007). Quantifying and mapping the human appropriation of net primary production in earth's terrestrial ecosystems. Proceedings of the National Academy of Sciences, 104(31), 12942-12947. https://doi.org/10.1073/pnas.0704243104 Haddad, N.M., Brudvig, L.A., Clobert, J., Davies, K.F., Gonzalez, A., Holt, R.D., Lovejoy, T.E., Sexton, J.O., Austin, M.P. & Collins, C.D. (2015). Habitat fragmentation and its lasting impact on Earth’s ecosystems. Science Advances, 1(2), e1500052. https://doi.org/10.1126/sciadv.1500052 Hall, W.B., McKeon, G.M., Carter, J.O., Day, K.A., Howden, S.M., Scanlan, J.C., Johnston, P.W. & Burrows, W.H. (1998). Climate change in Queensland's grazing lands: II. An assessment of the impact on animal production from native pastures. The Rangeland Journal, 20(2), 177-205. https://doi.org/10.1071/RJ9980177 Hankerson, B.R., Schierhorn, F., Prishchepov, A.V., Dong, C., Eisfelder, C. & Müller, D. (2019). Modeling the spatial distribution of grazing intensity in Kazakhstan. PLoS ONE, 14(1), e0210051. https://doi.org/10.1371/journal.pone.0210051 Hantson, S., Padilla, M., Corti, D. & Chuvieco, E. (2013). Strengths and weaknesses of MODIS hotspots to characterize global fire occurrence. Remote Sensing of Environment, 131, 152-159. https://doi.org/10.1016/j.rse.2012.12.004 Hao, W.M., Reeves, M.C., Baggett, L.S., Balkanski, Y., Ciais, P., Nordgren, B.L., Petkov, A., Corley, R.E., Mouillot, F. & Urbanski, S.P. (2021). Wetter environment and increased grazing reduced the area burned in northern Eurasia from 2002 to 2016. Biogeosciences, 18(8), 2559-2572. https://doi.org/10.5194/bg18-2559-2021 Harris, I., Jones, P.D., Osborn, T.J. & Lister, D.H. (2014). Updated high-resolution grids of monthly climatic observations - the CRU TS3.10 Dataset. International Journal of Climatology, 34(3), 623-642. https://doi.org/10.1002/joc.3711 He, M.L., Ishikawa, S. & Hidari, H. (2005). Fatty Acid Profiles of Various Muscles and Adipose Tissues from Fattening Horses in Comparison with Beef Cattle and Pigs. Asian-Australas J Anim Sci, 18(11), 1655-1661. https://doi.org/10.5713/ajas.2005.1655 87 Heinke, J., Lannerstad, M., Gerten, D., Havlík, P., Herrero, M., Notenbaert, A.M.O., Hoff, H. & Müller, C. (2020). Water use in global livestock production—opportunities and constraints for increasing water productivity. Water Resources Research, 56(12), e2019WR026995. https://doi.org/10.1029/2019WR026995 Herrero, M., Grace, D., Njuki, J., Johnson, N., Enahoro, D., Silvestri, S. & Rufino, M.C. (2013). The roles of livestock in developing countries. Animal, 7(s1), 3-18. https://doi.org/10.1071/RJ9980177 Herrero, M., Havlík, P., Valin, H., Notenbaert, A., Rufino, M.C., Thornton, P.K., Blümmel, M., Weiss, F., Grace, D. & Obersteiner, M. (2013). Biomass use, production, feed efficiencies, and greenhouse gas emissions from global livestock systems. Proceedings of the National Academy of Sciences, 110(52), 20888-20893. https://doi.org/10.1073/pnas.1308149110 Herrero, M., Thornton, P.K., Gerber, P. & Reid, R.S. (2009). Livestock, livelihoods and the environment: understanding the trade-offs. Current Opinion in Environmental Sustainability, 1(2), 111-120. https://doi.org/10.1016/j.cosust.2009.10.003 Hersbach, H., Muñoz-Sabater, J., Nicolas, J., Rozum, I., Simmons, A., Vamborg, F., Bell, B., Berrisford, P., Biavati, G., Buontempo, C., Horányi, A., Peubey, C., Radu, R., Schepers, D., Soci, C., Dee, D. & Thépaut, J.- N. (2018). Essential climate variables for assessment of climate variability from 1979 to present. https://doi.org/10.24381/7470b643 Hobbs, N.T., Galvin, K.A., Stokes, C.J., Lackett, J.M., Ash, A.J., Boone, R.B., Reid, R.S. & Thornton, P.K. (2008). Fragmentation of rangelands: implications for humans, animals, and landscapes. Global Environmental Change, 18(4), 776-785. https://doi.org/10.1016/j.gloenvcha.2008.07.011 Hoekstra, J.M., Boucher, T.M., Ricketts, T.H. & Roberts, C. (2005). Confronting a biome crisis: global disparities of habitat loss and protection. Ecology Letters, 8(1), 23-29. https://doi.org/10.1111/j.14610248.2004.00686.x Holechek, J.L. (1988). An approach for setting the stocking rate. Rangelands, 10-14. Hölzel, N., Haub, C., Ingelfinger, M.P., Otte, A. & Pilipenko, V.N. (2002). The return of the steppe large-scale restoration of degraded land in southern Russia during the post-Soviet era. Journal for Nature Conservation, 10(2), 75-85. https://doi.org/10.1078/1617-1381-00009 Hou, Y., Bai, Z., Lesschen, J.P., Staritsky, I.G., Sikirica, N., Ma, L., Velthof, G.L. & Oenema, O. (2016). Feed use and nitrogen excretion of livestock in EU-27. Agriculture, Ecosystems & Environment, 218, 232-244. https://doi.org/10.1016/j.agee.2015.11.025 Hristov, A.N., Oh, J., Lee, C., Meinen, R., Montes, F., Ott, T., Firkins, J., Rotz, A., Dell, C., Adesogan, A., Yang, W., Tricarico, J., Kebreab, E., Waghorn, G., Dijkstra, J. & Oosting, S. (2013). Mitigation of Greenhouse Gas Emission in Livestock Production – A review of technical options for non-CO₂ emissions. Rome, Italy: FAO. IAFC & NFPA. (2019). Fundamentals of fire fighter skills and hazardous materials response 4th ed. Burlington, Massachusetts: Jones & Bartlett Learning. Ibrahim, Y. & Howarth, A. (2016). Constructing the Eastern European other: The horsemeat scandal and the migrant other. Journal of Contemporary European Studies, 24(3), 397-413. https://doi.org/10.1080/14782804.2015.1135108 IIASA & FAO. (2012). Global Agro‐ecological Zones (GAEZ v3.0)). Laxenburg, Austria; Rome, Italy. http://gaez.fao.org Insausti, K., Beldarrain, L.R., Lavín, M.P., Aldai, N., Mantecón, Á.R., Sáez, J.L. & Canals, R.M. (2021). Horse meat production in northern Spain: Ecosystem services and sustainability in High Nature Value farmland. Animal Frontiers, 11(2), 47-54. https://doi.org/10.1093/af/vfab003 IPBES. (2018). The regional assessment report on biodiversity and ecosystem services for Europe and Central Asia M. Rounsevell, M. Fischer, A. Torre-Marin Rando, & A. Mader, Eds. Bonn, Germany: Secretariat of the Intergovernmental Science-Policy Platform on Biodiversity and Ecosystem Services. IPBES. (2019). Global assessment report on biodiversity and ecosystem services of the Intergovernmental Science-Policy Platform on Biodiversity and Ecosystem Services). Bonn, Germany. IPCC. (2006). IPCC guidelines for national greenhouse gas inventories). Hayama, Japan: Institute for Global Environmental Strategies. Janiak, M., Sawa, A. & Bogucki, M. (2017). Effect of age and sex of slaughtered cattle on dressing percentage and EUROP classification results. Acta Scientiarum Polonorum Zootechnica, 15(4), 41-54. https://doi.org/10.21005/asp.2016.15.4.04 88 Jaskari, M.-M., Leipämaa-Leskinen, H. & Syrjälä, H. (2015). Revealing the paradoxes of horsemeat–The challenges of marketing horsemeat in Finland. NJB, 64(2), 86-102. https://urn.fi/URN:NBN:fife2022063050737 Jastrzębska, E., Daszkiewicz, T., Górecka-Bruzda, A. & Feliś, D. (2019). Current situation and prospects for the horse meat market in Poland and the world. Medycyna Weterynaryjna, 75(4), 196-202. https://doi.org/10.21521/mw.6203 Johnson, C.N., Prior, L.D., Archibald, S., Poulos, H.M., Barton, A.M., Williamson, G.J. & Bowman, D.M.J.S. (2018). Can trophic rewilding reduce the impact of fire in a more flammable world? Philosophical Transactions of the Royal Society B: Biological Sciences, 373(1761), 20170443. https://doi.org/10.1098/rstb.2017.0443 Josling, T.E. (1974). Agricultural Policies in Developed Countries: A Review. Journal of Agricultural Economics, 25(3), 229-264. https://doi.org/10.1111/j.1477-9552.1974.tb00547.x Just, R.E. & Antle, J.M. (1990). Interactions between agricultural and environmental policies: a conceptual framework. The American Economic Review, 80(2), 197-202. https://www.jstor.org/stable/2006569 Kaduyu, I., Tsheko, R., Chepete, J.H. & Kgosiesele, E. (2022). Remotely sensed dry matter productivity and soil moisture content as potential predictors of arid rangeland wildfires: A case study of Kgalagadi District, Botswana. World Journal of Advanced Engineering Technology and Sciences, 7(2), 143-156. https://doi.org/10.30574/wjaets.2022.7.2.0143 Kamp, J., Koshkin, M.A., Bragina, T.M., Katzner, T.E., Milner-Gulland, E.J., Schreiber, D., Sheldon, R., Shmalenko, A., Smelansky, I., Terraube, J. & Urazaliev, R. (2016). Persistent and novel threats to the biodiversity of Kazakhstan's steppes and semi-deserts [journal article]. Biodiversity and Conservation, 25(12), 2521-2541. https://doi.org/10.1007/s10531-016-1083-0 Kamp, J., Sheldon, R.D., Koshkin, M.A., Donald, P.F. & Biedermann, R. (2009). Post‐Soviet steppe management causes pronounced synanthropy in the globally threatened Sociable Lapwing Vanellus gregarius. Ibis, 151(3), 452-463. https://doi.org/10.1111/j.1474-919X.2009.00938.x Kamp, J., Siderova, T.V., Salemgareev, A.R., Urazaliev, R.S., Donald, P.F. & Hölzel, N. (2012). Niche separation of larks (Alaudidae) and agricultural change on the drylands of the former Soviet Union. Agriculture, Ecosystems & Environment, 155, 41-49. https://doi.org/10.1016/j.agee.2012.03.023 Kamp, J., Urazaliev, R., Balmford, A., Donald, P.F., Green, R.E., Lamb, A.J. & Phalan, B. (2015). Agricultural development and the conservation of avian biodiversity on the Eurasian steppes: a comparison of landsparing and land-sharing approaches. Journal of Applied Ecology, 52(6), 1578-1587. https://doi.org/10.1111/1365-2664.12527 Kamruzzaman, M., Makino, Y., Oshita, S. & Liu, S. (2015). Assessment of visible near-infrared hyperspectral imaging as a tool for detection of horsemeat adulteration in minced beef. Food and Bioprocess Technology, 8(5), 1054-1062. https://doi.org/10.1007/s11947-015-1470-7 KazStat. (2008). Results of the first national agricultural census. Astana, Kazakhstan Retrieved from https://stat.gov.kz/ru/national/agriculture2006_2007/digest/ KazStat. (2011). Results of the national population census. Astana, Kazakhstan KazStat. (2016). Agriculture, Forestry and Fisheries. http://www.stat.gov.kz KazStat. (2019). Agriculture, Forestry and Fisheries. Main indicators of livestock development. http://www.stat.gov.kz KazStat. (2023). Agriculture, Forestry and Fisheries. Main indicators of livestock development. http://www.stat.gov.kz Kelly, L.T., Giljohann, K.M., Duane, A., Aquilué, N., Archibald, S., Batllori, E., Bennett, A.F., Buckland, S.T., Canelles, Q., Clarke, M.F., Fortin, M.-J., Hermoso, V., Herrando, S., Keane, R.E., Lake, F.K., McCarthy, M.A., Morán-Ordóñez, A., Parr, C.L., Pausas, J.G., . . . & Brotons, L. (2020). Fire and biodiversity in the Anthropocene. Science, 370(6519), eabb0355. https://doi.org/10.1126/science.abb0355 Kennedy, O.B., Stewart‐Knox, B.J., Mitchell, P.C. & Thurnham, D.I. (2004). Consumer perceptions of poultry meat: a qualitative analysis. Nutrition & Food Science, 34(3), 122-129. https://doi.org/10.1108/00346650410536746 Kerven, C., Robinson, S. & Behnke, R. (2021). Pastoralism at Scale on the Kazakh Rangelands: From Clans to Workers to Ranchers [Review]. Frontiers in Sustainable Food Systems, 4. https://doi.org/10.3389/fsufs.2020.590401 89 Kerven, C., Robinson, S., Behnke, R., Kushenov, K. & Milner-Gulland, E.J. (2016). Horseflies, wolves and wells: biophysical and socio-economic factors influencing livestock distribution in Kazakhstan’s rangelands. Land Use Policy, 52, 392-409. https://doi.org/10.1016/j.landusepol.2015.12.030 Kirchner, M., Schmidt, J., Kindermann, G., Kulmer, V., Mitter, H., Prettenthaler, F., Rüdisser, J., Schauppenlehner, T., Schönhart, M., Strauss, F., Tappeiner, U., Tasser, E. & Schmid, E. (2015). Ecosystem services and economic development in Austrian agricultural landscapes — The impact of policy and climate change scenarios on trade-offs and synergies. Ecological Economics, 109, 161-174. https://doi.org/10.1016/j.ecolecon.2014.11.005 Klein Goldewijk, K., Van Drecht, G. & Bouwman, A.F. (2007). Mapping contemporary global cropland and grassland distributions on a 5 × 5 minute resolution. Journal of Land Use Science, 2(3), 167-190. https://doi.org/10.1080/17474230701622940 Klein, I., Gessner, U. & Kuenzer, C. (2012). Regional land cover mapping and change detection in Central Asia using MODIS time-series. Applied Geography, 35(1–2), 219-234. https://doi.org/10.1016/j.apgeog.2012.06.016 Kock, R.A., Orynbayev, M., Robinson, S., Zuther, S., Singh, N.J., Beauvais, W., Morgan, E.R., Kerimbayev, A., Khomenko, S., Martineau, H.M., Rystaeva, R., Omarova, Z., Wolfs, S., Hawotte, F., Radoux, J. & MilnerGulland, E.J. (2018). Saigas on the brink: Multidisciplinary analysis of the factors influencing mass mortality events. Science Advances, 4(1). https://doi.org/10.1126/sciadv.aao2314 Koshkina, A., Freitag, M., Grigoryeva, I., Hölzel, N., Stirnemann, I., Velbert, F. & Kamp, J. (2022). Post-Soviet fire and grazing regimes govern the abundance of a key ecosystem engineer on the Eurasian steppe, the yellow ground squirrel Spermophilus fulvus. Diversity and Distributions, 00, 1-14. https://doi.org/10.1111/ddi.13668 Kraemer, R., Prishchepov, A.V., Dara, A., Kuemmerle, T., Müller, D., Radeloff, V., Terekhov, A. & Frühauf, M. (2015). Long-term agricultural land-cover change and potential for cropland expansion in the former Virgin Lands area of Kazakhstan. Environmental Research Letters, 10(5), 054012. https://doi.org/10.1088/1748-9326/10/5/054012 Krausmann, F., Erb, K.-H., Gingrich, S., Lauk, C. & Haberl, H. (2008). Global patterns of socioeconomic biomass flows in the year 2000: A comprehensive assessment of supply, consumption and constraints. Ecological Economics, 65(3), 471-487. https://doi.org/10.1016/j.ecolecon.2007.07.012 Kuemmerle, T., Erb, K.-H., Meyfroidt, P., Müller, D., Verburg, P.H., Estel, S., Haberl, H., Hostert, P., Jepsen, M.R., Kastner, T., Levers, C., Lindner, M., Plutzar, C., Verkerk, P.J., van der Zanden, E.H. & Reenberg, A. (2013). Challenges and opportunities in mapping land use intensity globally. Current Opinion in Environmental Sustainability, 5(5), 484–493. https://doi.org/10.1016/j.cosust.2013.06.002 Küster, H. & Keenleyside, C. (2009). The origin and use of agricultural grasslands in Europe. In P. Veen, R. Jefferson, J. de Schmidt, & J. van der Straaten (Eds.), Grasslands in Europe – of high nature value (pp. 814). Leiden, The Netherlands: KNNV. https://doi.org/10.1163/9789004278103_002 Lambin, E.F. & Meyfroidt, P. (2011). Global land use change, economic globalization, and the looming land scarcity. Proceedings of the National Academy of Sciences, 108(9), 3465-3472. https://doi.org/10.1073/pnas.1100480108 Lamy, A., Costa, S., Vial, C., Badji, I., Carrère, M., Rollet, P. & Amiot, M.J. (2023). Horsemeat consumption in France: Determinants and sustainable market perspectives. Meat Science, 198, 109083. https://doi.org/10.1016/j.meatsci.2022.109083 Langlois, B. (2011). The History, Ethnology and Social Importance of Mare's Milk Consumption in Central Asia. Journal of Life Sciences, 5(10), 863-872. https://hal.science/hal-01001395 Lee, C.-E., Seong, P.-N., Oh, W.-Y., Ko, M.-S., Kim, K.-I. & Jeong, J.-H. (2007). Nutritional characteristics of horsemeat in comparison with those of beef and pork. Nutrition Research and Practice, 1(1), 70-73. https://doi.org/10.4162/nrp.2007.1.1.70 Leteux, S. (2012). Is hippophagy a taboo in constant evolution? Menu: Journal of Food and Hospitality, 1-13. https://shs.hal.science/halshs-00796033 Levine, M.A. (2005). Domestication and early history of the horse. In D. S. Mills & S. M. McDonnell (Eds.), The domestic horse: the origins, development and management of its behaviour (pp. 5-22). Cambridge, UK: Cambridge University Press. Librado, P., Khan, N., Fages, A., Kusliy, M.A., Suchan, T., Tonasso-Calvière, L., Schiavinato, S., Alioglu, D., Fromentier, A., Perdereau, A., Aury, J.-M., Gaunitz, C., Chauvey, L., Seguin-Orlando, A., Der Sarkissian, C., 90 Southon, J., Shapiro, B., Tishkin, A.A., Kovalev, A.A., . . . & Orlando, L. (2021). The origins and spread of domestic horses from the Western Eurasian steppes. Nature, 598(7882), 634-640. https://doi.org/10.1038/s41586-021-04018-9 Lopatin, J., Fassnacht, F.E., Kattenborn, T. & Schmidtlein, S. (2017). Mapping plant species in mixed grassland communities using close range imaging spectroscopy. Remote Sensing of Environment, 201, 12-23. https://doi.org/10.1016/j.rse.2017.08.031 López, C.L., Celaya, R., Ferreira, L.M.M., García, U., Rodrigues, M.A.M. & Osoro, K. (2019). Comparative foraging behaviour and performance between cattle and horses grazing in heathlands with different proportions of improved pasture area. Journal of Applied Animal Research, 47(1), 377-385. https://doi.org/10.1080/09712119.2019.1649679 Lorenzo, J.M., Munekata, P.E.S., Campagnol, P.C.B., Zhu, Z., Alpas, H., Barba, F.J. & Tomasevic, I. (2017). Technological aspects of horse meat products – A review. Food Research International, 102, 176-183. https://doi.org/10.1016/j.foodres.2017.09.094 Lorenzo, J.M., Sarriés, M.V., Tateo, A., Polidori, P., Franco, D. & Lanza, M. (2014). Carcass characteristics, meat quality and nutritional value of horsemeat: A review. Meat Science, 96(4), 1478-1488. https://doi.org/10.1016/j.meatsci.2013.12.006 Löw, F., Fliemann, E., Abdullaev, I., Conrad, C. & Lamers, J.P.A. (2015). Mapping abandoned agricultural land in Kyzyl-Orda, Kazakhstan using satellite remote sensing. Applied Geography, 62(0), 377-390. https://doi.org/10.1016/j.apgeog.2015.05.009 Madichie, N.O. & Yamoah, F.A. (2017). Revisiting the European horsemeat scandal: the role of power asymmetry in the food supply chain crisis. Thunderbird International Business Review, 59(6), 663-675. https://doi.org/10.1002/tie.21841 Malacarne, M., Martuzzi, F., Summer, A. & Mariani, P. (2002). Protein and fat composition of mare's milk: some nutritional remarks with reference to human and cow's milk. International Dairy Journal, 12(11), 869-877. https://doi.org/10.1016/S0958-6946(02)00120-6 Manceron, S., Ben-Ari, T. & Dumas, P. (2014). Feeding proteins to livestock: Global land use and food vs. feed competition. OCL, 21(4), D408. https://doi.org/10.1051/ocl/2014020 Mann, H.K. (1914). Gregory III. In The Lives of the Popes in the Early Middle Ages (Vol. 1, pp. 203-224). Kegan Paul, Trench, Trubner & Company. Mantovani, R., Guzzo, N., Sartori, C. & Bailoni, L. (2014). In vivo performance of Italian Heavy Draft Horse weanlings fed two protein levels and slaughtered at two ages. Journal of Animal Science, 92(11), 49985008. https://doi.org/10.2527/jas.2014-7875 Marino, R., della Malva, A., Maggiolino, A., De Palo, P., d’Angelo, F., Lorenzo, J.M., Sevi, A. & Albenzio, M. (2022). Nutritional Profile of Donkey and Horse Meat: Effect of Muscle and Aging Time. Animals, 12(6), 746. https://doi.org/10.3390/ani12060746 Marlon, J.R., Bartlein, P.J., Carcaillet, C., Gavin, D.G., Harrison, S.P., Higuera, P.E., Joos, F., Power, M.J. & Prentice, I.C. (2008). Climate and human influences on global biomass burning over the past two millennia. Nature Geoscience, 1(10), 697-702. https://doi.org/10.1038/ngeo313 Mazerolle, M. (2006). Improving data analysis in herpetology: using Akaike's Information Criterion (AIC) to assess the strength of biological hypotheses. Amphibia-Reptilia, 27(2), 169-180. https://doi.org/10.1163/156853806777239922 McNaughton, S.J., Milchunas, D.G. & Frank, D.A. (1996). How can net Primary Productivity be Measured in Grazing Ecosystems? Ecology, 77(3), 974-977. https://doi.org/10.2307/2265518 Mekonnen, M.M., Neale, C.M.U., Ray, C., Erickson, G.E. & Hoekstra, A.Y. (2019). Water productivity in meat and milk production in the US from 1960 to 2016. Environment international, 132, 105084. https://doi.org/10.1016/j.envint.2019.105084 Meyfroidt, P., Schierhorn, F., Prishchepov, A.V., Müller, D. & Kuemmerle, T. (2016). Drivers, constraints and trade-offs associated with recultivating abandoned cropland in Russia, Ukraine and Kazakhstan. Global Environmental Change, 37, 1-15. https://doi.org/10.1016/j.gloenvcha.2016.01.003 Milchunas, D.G., Sala, O.E. & Lauenroth, W.K. (1988). A Generalized Model of the Effects of Grazing by Large Herbivores on Grassland Community Structure. The American Naturalist, 132(1), 87-106. https://doi.org/10.1086/284839 91 Milner-Gulland, E.J., Kholodova, M.V., Bekenov, A., Bukreeva, O.M., Grachev, I.O., Amgalan, L. & Lushchekina, A.A. (2001). Dramatic declines in saiga antelope populations. Oryx, 35(4), 340-345. https://doi.org/10.1046/j.1365-3008.2001.00202.x Mirzabaev, A., Ahmed, M., Werner, J., Pender, J. & Louhaichi, M. (2016). Rangelands of Central Asia: challenges and opportunities. Journal of Arid Land, 8(1), 93-108. https://doi.org/10.1007/s40333-0150057-5 Moreno, M.V., Conedera, M., Chuvieco, E. & Pezzatti, G.B. (2014). Fire regime changes and major driving forces in Spain from 1968 to 2010. Environmental Science & Policy, 37, 11-22. https://doi.org/10.1016/j.envsci.2013.08.005 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. https://doi.org/10.1111/j.2041210x.2012.00261.x Nasiev, B.N., Gabdulov, M.A. & Zhanatalapov, N.Z. (2014). Selection of forage crops for the semi-desert zone of western Kazakhstan. Proceedings of the Orenburg State Agrarian University(2), 54-56. National Research Council. (1981). Effect of environment on nutrient requirements of domestic animals. Washington, DC: The National Academies Press. Nazarbayev, N.A. (2010). On State Statistics. (257-IV). Astana, Kazakhstan: Parliament of the Republic of Kazakhstan Nazarbayev, N.A. (2012). Strategy Kazakhstan 2050. Astana, Kazakhstan Neumann, K., Elbersen, B., Verburg, P., Staritsky, I., Pérez-Soba, M., de Vries, W. & Rienks, W. (2009). Modelling the spatial distribution of livestock in Europe. Landscape Ecology, 24(9), 1207-1222. https://doi.org/10.1007/s10980-009-9357-5 Newman, R. & Noy, I. (2023). The global costs of extreme weather that are attributable to climate change. Nature Communications, 14(1), 6103. https://doi.org/10.1038/s41467-023-41888-1 Nguyen, T.L.T., Hermansen, J.E. & Mogensen, L. (2010). Environmental consequences of different beef production systems in the EU. Journal of Cleaner Production, 18(8), 756-766. https://doi.org/10.1016/j.jclepro.2009.12.023 Nicolas, G., Robinson, T.P., Wint, G.R.W., Conchedda, G., Cinardi, G. & Gilbert, M. (2016). Using Random Forest to Improve the Downscaling of Global Livestock Census Data. PLoS ONE, 11(3), e0150424. https://doi.org/10.1371/journal.pone.0150424 O'Mara, F.P. (2011). The significance of livestock as a contributor to global greenhouse gas emissions today and in the near future. Animal Feed Science and Technology, 166, 7-15. https://doi.org/10.1016/j.anifeedsci.2011.04.074 Olcott, M.B. (1995). The Kazakhs. Stanford, California: Hoover Institution Press; Stanford University Press. http://books.google.de/books?id=fnB9QgAACAAJ Oliveira, S., Oehler, F., San-Miguel-Ayanz, J., Camia, A. & Pereira, J.M.C. (2012). Modeling spatial patterns of fire occurrence in Mediterranean Europe using Multiple Regression and Random Forest. Forest Ecology and Management, 275, 117-129. https://doi.org/10.1016/j.foreco.2012.03.003 Ørskov, E.R., Reid, G.W. & Kay, M. (1988). Prediction of intake by cattle from degradation characteristics of roughages. Animal Science, 46(1), 29-34. https://doi.org/10.1017/S000335610000307X Osoro, K., Ferreira, L.M.M., García, U., Martínez, A. & Celaya, R. (2015). Forage intake, digestibility and performance of cattle, horses, sheep and goats grazing together on an improved heathland. Animal Production Science, 57(1), 102-109. https://doi.org/10.1071/AN15153 Paleari, M.A., Moretti, V.M., Beretta, G. & Caprino, F. (2006). Characterization of a lamb ham: Fatty acids and volatile compounds composition. Journal of Muscle Foods, 17(4), 398-412. https://doi.org/10.1111/j.1745-4573.2006.00059.x Paleari, M.A., Moretti, V.M., Beretta, G., Mentasti, T. & Bersani, C. (2003). Cured products from different animal species. Meat Science, 63(4), 485-489. https://doi.org/10.1016/S0309-1740(02)00108-0 Parlasca, M.C. & Qaim, M. (2022). Meat Consumption and Sustainability. Annual Review of Resource Economics, 14(1), 17-41. https://doi.org/10.1146/annurev-resource-111820-032340 Parr, C.L., Lehmann, C.E.R., Bond, W.J., Hoffmann, W.A. & Andersen, A.N. (2014). Tropical grassy biomes: misunderstood, neglected, and under threat. Trends in Ecology & Evolution, 29(4), 205-213. https://doi.org/10.1016/j.tree.2014.02.004 92 Pärtel, M., Bruun, H.H. & Sammul, M. (2005, 29-31 August 2005). Biodiversity in temperate European grasslands: origin and conservation. In R. Lillak, R. Viiralt, A. Linke, & V. Geherman, Integrating Efficient Grassland Farming and Biodiversity 13th International Occasional Symposium of the European Grassland Federation, Tartu, Estonia. Pearson, R.A. (1998). Draught Animals and their Management: The Future in Rain-fed Agriculture. Annals of Arid Zone, 37(3), 233-251. Penrith, M.L. (2020). Current status of African swine fever. CABI Agriculture and Bioscience, 1(1), 11. https://doi.org/10.1186/s43170-020-00011-w Perino, A., Pereira, H.M., Navarro, L.M., Fernández, N., Bullock, J.M., Ceaușu, S., Cortés-Avizanda, A., van Klink, R., Kuemmerle, T. & Lomba, A. (2019). Rewilding complex ecosystems. Science, 364(6438), eaav5570. https://doi.org/10.1126/science.aav5570 Petrick, M., Wandel, J. & Karsten, K. (2013). Rediscovering the Virgin Lands: Agricultural Investment and Rural Livelihoods in a Eurasian Frontier Area. World Development, 43, 164-179. https://doi.org/10.1016/j.worlddev.2012.09.015 Pflugmacher, D., Krankina, O.N., Cohen, W.B., Friedl, M.A., Sulla-Menashe, D., Kennedy, R.E., Nelson, P., Loboda, T.V., Kuemmerle, T., Dyukarev, E., Elsakov, V. & Kharuk, V.I. (2011). Comparison and assessment of coarse resolution land cover maps for Northern Eurasia. Remote Sensing of Environment, 115(12), 3539-3553. https://doi.org/10.1016/j.rse.2011.08.016 Phelps, L.N. & Kaplan, J.O. (2017). Land use for animal production in global change studies: Defining and characterizing a framework. Global Change Biology, 23(11), 4457-4471. https://doi.org/10.1111/gcb.13732 Piipponen, J., Jalava, M., de Leeuw, J., Rizayeva, A., Godde, C., Cramer, G., Herrero, M. & Kummu, M. (2022). Global trends in grassland carrying capacity and relative stocking density of livestock. Global Change Biology, 28(12), 3902-3919. https://doi.org/10.1111/gcb.16174 Prishchepov, A.V., Myachina, K.V., Kamp, J., Smelansky, I., Dubrovskaya, S., Ryakhov, R., Grudinin, D., Yakovlev, I. & Urazaliyev, R. (2021). Multiple trajectories of grassland fragmentation, degradation, and recovery in Russia's steppes. Land Degradation and Development, 32(11), 3220-3235. https://doi.org/10.1002/ldr.3976 Propastin, P.A., Kappas, M.W., Herrmann, S.M. & Tucker, C.J. (2011). Modified light use efficiency model for assessment of carbon sequestration in grasslands of Kazakhstan: combining ground biomass data and remote-sensing. International Journal of Remote Sensing, 33(5), 1465-1487. https://doi.org/10.1080/01431161.2011.577105 QGIS Development Team. (2017). QGIS Geographic Information System. In (Version 2.18.3) Open Source Geospatial Foundation Project. http://www.qgis.org Qi, J., Xin, X., John, R., Groisman, P. & Chen, J. (2017). Understanding livestock production and sustainability of grassland ecosystems in the Asian Dryland Belt. Ecological Processes, 6(1), 22. https://doi.org/10.1186/s13717-017-0087-3 R Core Team. (2017). R: A Language and Environment for Statistical Computing. In (Version 3.3.2) R Foundation for Statistical Computing. https://www.R-project.org Rabin, S.S., Magi, B.I., Shevliakova, E. & Pacala, S.W. (2015). Quantifying regional, time-varying effects of cropland and pasture on vegetation fire. Biogeosciences, 12(22), 6591-6604. https://doi.org/10.5194/bg-12-6591-2015 Ramankutty, N., Evan, A.T., Monfreda, C. & Foley, J.A. (2008). Farming the planet: 1. Geographic distribution of global agricultural lands in the year 2000. Global Biogeochemical Cycles, 22(GB1003). https://doi.org/10.1029/2007GB002952 Ramankutty, N. & Foley, J.A. (1999). Estimating historical changes in global land cover: Croplands from 1700 to 1992. Global Biogeochemical Cycles, 13(4), 997–1027. https://doi.org/10.1029/1999GB900046 Raspa, F., Dinardo, F.R., Vervuert, I., Bergero, D., Bottero, M.T., Pattono, D., Dalmasso, A., Vinassa, M., Valvassori, E. & Bruno, E. (2021). A Fibre‐vs. cereal grain‐based diet: Which is better for horse welfare? Effects on intestinal permeability, muscle characteristics and oxidative status in horses reared for meat production. Journal of Animal Physiology and Animal Nutrition, 106(2), 313-326. https://doi.org/10.1111/jpn.13643