Arxiv:2103.06330V1 [Astro-Ph.EP] 10 Mar 2021 the Impact of Civilization on the Planet (Solomon Et Al the Point Is Crossed in State Space, Subsequent Evolu- 2007)

Draft version March 12, 2021 Typeset using LATEX twocolumn style in AASTeX63

Triggering A Climate Change Dominated ”Anthropocene”: Is It Common Among Exocivilizations?

Ethan Savitch,1 Adam Frank,1 Jonathan Carroll-Nellenback,1 Jacob Haqq-Misra,2 Axel Kleidon,3 and Marina Alberti4

1Department of Physics and Astronomy, University of Rochester, Rochester, New York, 14620 2Blue Marble Space Institute of Science, 600 1st Avenue, 1st Floor, Seattle, Washington 98104 3Max-Planck-Institut fuer Biogeochemie, Jena, Germany 4Department of Urban Design and Planning, University of Washington, Seattle, WA 98195, USA

Submitted to AJ

ABSTRACT We seek to model the coupled evolution of a planet and a civilization through the era when energy harvesting by the civilization drives the planet into new and adverse climate states. In this way we ask if triggering ”anthropocenes” of the kind humanity is experiencing now might be a generic feature of planet-civilization evolution. In this study we focus on the eﬀects of energy harvesting via combustion and vary the planet’s initial atmospheric chemistry and orbital radius. In our model, energy harvesting increases the civilization’s population growth rate while also, eventually, leading to a degradation of the planetary climate state (relative to the civilization’s habitability.) We also assume the existence of a Complex Life Habitable Zone in which very high levels of CO2 are detrimental to multi-cellular animal life such as those creating technological civilizations. Our models show that the civilization’s growth is truncated by planetary feedback (a ”climate dominated anthropocene”) for a signiﬁcant region of the initial parameter space.

Keywords: Anthropocene; Astrobiology; Civilization; Dynamical System Theory; Exoplanets; Popula- tion dynamics

1. INTRODUCTION man civilization is difficult to predict. These impacts It has now become clear that human activity has are, however, accepted to have negative consequences altered the state of the coupled Earth systems (at- with assessments ranging from a difficult adaptation mosphere, hydrosphere, cryosphere, lithosphere, bio- to full-scale collapse. Also unknown are the require- sphere). There are multiple measures of human impact ments needed to successfully manage our entry into the on these systems including the transport of key com- Anthropocene and then create a long-term sustainable pounds and materials (Lenton et al 2015); the coloniza- version of civilization. One can even ask if such long- tion of surface area (Hooke et al 2013); human appropri- term sustainable versions of civilization are even possi- ation of the terrestrial productivity (Vitousek et al 1986) ble. It is possible that the Anthropocene may represent and energy (Kleidon 2012). Global Warming driven by a “tipping point” in the coupled dynamical system rep- resenting both planet and civilization such that once CO2 emissions represents the most dramatic example of arXiv:2103.06330v1 [astro-ph.EP] 10 Mar 2021 the impact of civilization on the planet (Solomon et al the point is crossed in state space, subsequent evolu- 2007). tion proves detrimental to the civilization. (Lenton et Taken as a whole these changes in the state/behavior al 2008; Kuehn 2011). We note that in Frank Alberti & of Earth’s coupled planetary systems have been de- Kleidon(2017) final sustainable planetary states were scribed as a new planetary/geologic epoch termed the explored. Anthropocene (Crutzen 2002). In fact, recent studies In Frank et al.(2018) the Earth’s entry into the An- have shown that 2020 marks the moment when human- thropocene was examined from an astrobiological per- made ’anthropogenic’ mass has exceeded all of Earth’s spective. That study asked if the situation currently living biomass (Elhacham et al. 2020). The specifics encountered on Earth was unique. In particular, given of the long-term impact of the Anthropocene on hu- its global scale, might the transition represented by the 2 Savitch et al.

Anthropocene be a generic feature of any planet evolv- the coupled planetary systems, including ”population ing a species that intensively harvests resources for the growth advantages” gained via the harvesting of the re- development of a technological civilization (Haqq-Misra sources. The models showed three distinct classes of exo- and Baum 2009; Frank Alberti & Kleidon 2017; Mul- civilization trajectories. The first of these were smooth lan and Haqq-Misra 2019)? This question has direct entries into long-term, “sustainable” steady-states. The consequences for both the study of astrobiology and the second class were population booms followed by various sustainability of human civilization. levels of “die-off”. Finally were rapid “collapse” trajec- Relevant to astrobiology, it is now apparent that most tories for which the population (N) approaches N = 0. stars harbor families of planets (Seager 2013). In- In this work we seek to take a step up in complexity deed, many of those planets will be in the star’s hab- and realism compared to Frank et al.(2018). In partic- itable zones (Howard 2013). Tremendous effort has ular, we represent the evolution of the planetary state gone into the study of biosignatures, i.e. imprints a bio- via an explicit energy balance climate model (EBM) and sphere leaves on detectable light from the planet. Re- take the global temperature to be representative of that cently it has been recognized that imprints from technol- state (North & Kim 2017). The interaction between the ogy created by an intelligent civilization might be just civilization and the planetary coupled systems is medi- as, or more easily detectable (Lingam and Loeb 2019) ated by the civilization’s CO2 production. This means than ”traditional” biosignatures. If Anthropocenes are we are explicitly considering civilizations whose energy a common consequence of a civilization developing on a generation comes through some form of combustion. As given inhabited world, then this co-evolutionary period in the first paper, the use of this energy allows the civ- between planet and civilization may effect the nature, ilization to increase its population (via increases in the and even existence, of technosignatures. In addition, birth rate of the population). At the same time, the if anthropocenes prove fatal for some civilizations then feedback of the energy use on the planetary state, now they can be considered as one form of a ”Great Filter” via CO2 emissions, alters that state. Thus planetary and are therefore relevant to discussions of the Fermi conditions can be driven beyond what is tolerable for Paradox (Carroll-Nellenback et al. 2019) the functioning of the civilization. This is reflected in The possibility that anthropocenes are common is an increase in the mortality (the death rate) of the pop- equally of interest to the pressing concerns about our ulation. own immediate future. We are, essentially, without a We explore the effect of changing two key parameters playbook in dealing with the planetary transition we in the models: the orbital distance of the planet from now face (though however Frank Alberti & Kleidon its host star and the initial atmospheric chemistry of (2017)). Any understanding of generic features in the the planet in terms of CO2. In essence we are asking if co-evolution of planetary systems and civilizations could we moved Earth to different orbits and/or changed its be of use in charting out the possible futures for our own initial CO2 concentration, would we still have triggered efforts to navigate our own version of the Anthropocene. the climate change we are experiencing now. Even purely modeling/theoretical perspectives on how Most animal life on Earth can not tolerate high levels techno-spheres co-evolve with the other geospheres (the of CO2 (Wittmann & Portner 2013). Thus a key as- biosphere in particular) may help us understand the sumption of our models is that complex life of the kind range and efficacy of viable options. expected to build a technological civilization will emerge The modeling framework presented in Frank et al. from a ”Complex Life Habitable Zone” (CLHZ) where (2018) was meant as a first step in studying generic be- initial CO2 concentrations are below a threshold (Schwi- haviors in the interaction between a resource-harvesting eterman et al. 2019; Ramirez 2020; Catling et al. 2005). technological civilization (an exo-civilization) and the We will discuss the consequences of this assumption in planetary environment in which it evolves. Using meth- the discussion section. ods from dynamical systems theory, a suite of sim- Finally we emphasize that this paper focuses on the ple equations was introduced for modeling a popula- triggering of a anthropocene which we will define as tion which consumes resources (for the purpose of run- planetary-systems change created by the civilization ning a technological civilization) and the feedback those which truncates the civilization’s population growth. resources drive on the state of the host planet. The When the main driver of the end of population growth feedbacks drive the planet away from the initial state is climate change for brevity we call this a ”climate- that gave birth to the civilization. The simple models dominated anthropocene”. This is in contrast to asking in Frank et al.(2018) conceptualized the problem pri- what a civilization can do to manage such an anthro- marily in terms of feedbacks from the resource use onto pocene once it occurs. To deal with this second ques- Triggering A Climate Change Dominated ”Anthropocene”: Is It Common Among Exocivilizations?3 tion requires including the civilization’s response (and the timing of that response) in the model. This is a topic which should be addressed in future work. The plan of the paper is as follows. In section 2 we describe the model and use the Earth’s recent history to tune and test it. In section 3 we provide an analysis of a linearized version of the model to extract key features of its behavior in terms of dimensionless parameters. In section 4 we present the results of the full non-linear model. We first show two sets of experiments run with either constant initial temperature or constant initial Figure 1: Surface plot of Sun’s Habitable Zone, calculated with atmospheric CO2 concentration (pCO2). We then show the EBM given by equation (1). The blue dot marks the location the results from a sweep of two parameters: orbital dis- of Earth. tance (a) and initial (pCO2). Finally we run a suite of models in which the civilization tolerance for global tem- Note that variations of the inner edge (ai) of the hab- perature changes (∆T ) is varied. In section 6 we discuss itable zone with P0 are due to fits in the absorption the consequences of these results for emerging studies of coefficients used in the EBM. While these could be rec- the ”astrobiology of the anthropocene” and present our onciled with more detailed models, the small variation conclusions and summary. imposed on ai did not effect the conclusions of the study. The dynamics of the civilization’s population, N, is gov- 2. THE MODEL erned by the per-capita net growth rate R, which we let We take a dynamical systems approach to the cou- vary with time (R = R(t)) and define as the balance be- pled evolution of the planet and civilization. The planet tween the per-capita birth (A) and death (B) rates. In is described in terms of its atmospheric state given by our simulations we assume a ”pre-technological” growth its average temperature T . This state depends on the rate. influx of stellar radiation and the atmospheric chemi- 1 dN R0 ≡ A0 − B0 = (2) cal composition which will change due to the activity N dt t=t0 of the civilization. In our model all the civilization’s As the civilization becomes more proficient at energy energy harvesting occurs via combustion. Thus we fol- harvesting, it’s ability to produce more offspring in- low the emission of CO by the civilization. Changes 2 creases. In our model we assume the civilization’s tech- in its partial pressure, P (t) = pCO (t), represent the 2 nological capacity (and hence its ability to harvest en- principle evolutionary driver occurring in the planet’s ergy) tracks with the production of combustion byprod- atmospheric composition. ucts. In other words, growth rates depend on technol- We use a ”1-D” energy balance model (EBM) to cal- ogy, and we take the rise in P to be a measure of techno- culate the temperature in latitudinal (θ) bands. logical advance. Thus we define an enhanced growth co- dT (θ, P ) ψ(1 − A) − I + ∇ · (κ∇T (θ)) efficient to be a function of P relative to the initial value = (1) (P0) the civilization found the planet in when it began dt Cv its technological evolution. For our enhanced growth where A is the planetary albedo, κ is the latitudinal coefficient we choose the form, heat transport and C is the heat capacity at con- v stant volume. Our version of the model was origi- P − P0 R+ = R0 1 + (3) nally developed by Darren Williams in Williams & Kast- ∆P ing(1997). It was then modified by Jacob Haqq- Where ∆P is a normalization constant that roughly Misra, who used it most recently in Fairen et al. corresponds to the pCO2 required to double the growth (2012). This version of the model is publicly available rate of the civilization. In this way equation (3) captures on GitHub at https://github.com/BlueMarbleSpace/ in a simple way how increases in technology (measured hextor/releases/tag/1.2.2. In our implementation of the by combustion products released into the atmosphere) model we average across latitudinal bands to obtain a increase the birth rate of the civilization. single globally averaged temperature. As technology produces higher P and more births In Figure1 we show the domains of our model in there will, eventually, be a corresponding feedback on (a, P0) space where P0 = pCO2,0 is the initial CO2 com- the planet, dictated by equation (1), and hence on the position of the planet before the civilization appears. 4 Savitch et al. population. We model this feedback via a term we de- and its detrimental impacts. In terms of our equations note the diminished growth rate, which we take to be this means there is no equilibrium for the temperature temperature dependent. except for the trivial one of the absence of a technological civilization (N = 0). T − T 2 R = R 0 (4) − 0 ∆T 2.1. Modeling Anthropocene Earth

Where T0 is the average planetary temperature when In order to provide both a test and a calibration of our the civilization began (t = t0), and ∆T describes the model, we apply it to the recent co-evolution of Earth range of temperatures amenable to the civilization’s and its human inhabitants into the Anthropocene. It is health. Thus we call ∆T the civilizations temperature worth noting a few points about the initial parameters. tolerance. This term can refer to both the biology of The model was begun at t = t0 = 1820 CE. The individuals or the functioning of the civilization as a global world population was taken to be N = N0 = 9 whole. Thus while individual members of the civilization 1.29 × 10 with an initial CO2 partial pressure of P = may be able to survive at T > T0 + ∆T the civilizations P0 = 284 ppm, approximately equal to the values on functioning as a complex system may be compromised. Earth prior to the industrial revolution. We chose the The final governing equation for N is, civilizations temperature tolerance of ∆T = 5K as this is representative of the range of temperatures considered in IPCC models and acts to quantify our civilizations dN = min [NR+,R0(Nmax − N)] − NR− (5) ”fragility” (Solomon et al 2007). Also, our choice for dt the technology birth benefit (∆P = 30ppm) was chosen The use of the min function in equation (5) introduces as an order of magnitude approximation to the levels of a carrying capacity (Nmax) into the systems dynamics. pCO2 humans thrive in (Schwieterman et al. 2019). The Carrying capacity is a foundational principle in popula- CO2 generation coefficient (C) was taken from current tion dynamics (Kot 2001). Without it the civilization’s global conditions while the initial net growth rate (R0) population can grow to levels that are unrealistic based was tuned to reproduce a best fit to the data. purely on food production capacities. For example, for The results are shown in Figure2, which shows the Earth, N > 100 billion appears unrealistic under even evolution of population N(t), growth rate, and global the most optimistic scenarios (Cohen 1995). In the mean temperature T (t). As can be seen, the model does classic logistic growth model, an excellent job of tracking both the rise in temperature and population during the last two centuries. B0 is the dN N parameter we used to tune the population part of our = NR 1 − (6) dt Nmax model. A0 was fixed by our assumption that the average the carrying capacity appears in the second term time between births was approximately 25 years. We which functions as the death rate. In our model we then adjusted B0 in order to have our model fit the chose to impose the carrying capacity through the min population data we have (shown as the top-most plot function to avoid the arbitrary non-linear dependence of Figure2). Finally, we adjusted the per-capita carbon on population which occurs in the logistic equation. We footprint (C) in order to match our climates response to will discuss the behavior this produces in the results sec- population growth, thus matching our global trends in tion. temperature (shown as the bottom-most plot of Figure 2). Finally, we model the production of CO2 via the simple equation, 3. ANALYTIC MODELING dP = CN (7) Before we begin numerical integration of our equa- dt tions, we first explore aspects of the solutions that can We do not include any means of reducing the CO2 in be extracted from a semi-analytic approach. We begin the atmosphere. While this can occur through natural by noting that its possible to approximate the climate means via weather and carbonate cycles, the relevant response to increased P = pCO2 via a simplified loga- timescales are much longer than we are interested in rithmic dependence (Huang & Bani Shahabadi 2014). here (t ∼ 106 y). We are also not attempting to model P the possible responses of a civilization to the climate T ≈ T0 + ∆TF log (8) change they generate. Here we only wish to know how P0 broad are the conditions that can lead to such change Triggering A Climate Change Dominated ”Anthropocene”: Is It Common Among Exocivilizations?5

Table 1: Dimensionless Model Quantities Var Deﬁnition Description

1 η N/Nmax Normalized population N

1 τ R0t Normalized time

1( (T − T0)/∆T Normalized temperature

dt θ ∆T/(D∆P ) Normalized Birth rate

log N acceleration d

1 γ (DCNmax)/(R0∆T ) Normalized forcing 1 α ∆T/∆TF Ratio of temperature change to aﬀect biology 1 K

to temperature change T to aﬀect climate sensi-

tivity 1 1 1 1 1 1 1 β (CNmax) / (R0∆P ) Increase in birth rate after burning sufficient Figure 2: Model Output for Earth’s recent history. By using CO2 to change temper- established parameters and tuning the growth rate models we recover population (N), its first derivative and temperature (T). ature by ∆T Note model results are the solid black line and global data is represented by the dotted blue line. enhanced term. Our dimensionless time (τ), is calculated with units of the population growth timescale, where ∆T is the change in temperature required for F t = 1/R , as discussed in Appendix A. The other three the climate sensitivity to drop by a factor of e and is G 0 parameters θ, α, and γ determine the behavior of the ≈ 4K for present day Earth (IPCC 2014). system and are defined in Table1. Also note that with the above approximation, In what follows we will define a climate anthropocene

dT ∆TF − T −T0 − T −T0 as the trajectory in which the population growth is = e ∆TF ≡ De ∆TF (9) dP P0 strongly truncated by the increase in planetary temperature. This means the population never gets close to This allows us to eliminate the CO partial pressure 2 the natural carrying capacity of the planet. The most as an independent variable and the model reduces to important parameter with regards to whether a civilization goes through a climate anthropocene is the nor- dN P0 T −T0 malized climate forcing γ. It represents how quickly the = min R N 1 + e ∆TF − 1 ,R (N − N) dt 0 ∆P 0 max climate would change if the population were to reach 2 the carrying capacity. The parameter θ represents the T − T0 − R0N (10) expected change in population growth rate due to the ∆T consumption of fossil fuels as the temperature changes dT − T −T0 = CNDe ∆TF (11) by ∆T . The parameter α represents the drop in the dt climate sensitivity (as a number of e-foldings) as the We can make the equations dimensionless by dividing temperature changes by ∆T . One additional parame- the first equation by R0Nmax and the second by R0∆T ter, β ≡ γθ, is also important as it is independent of the initial climate sensitivity (D) and population temperature tolerance (∆T ). It reflects the degree to which dη θ α 2 = min η 1 + (e − 1) , 1 − η − η (12) CO2 consumption increases the birth rate per natural dτ α growth time assuming the population was at the carry- d = γηe−α (13) ing capacity. dτ Our best fit Earth model had a γ = 1.94, θ = 14.45, The dimensionless population (η) represents how close and α = 1.52. See Table1 for a summary of the dimen- the civilization is to its carrying capacity. The dimensionless parameters. sionless temperature anomaly () represents how much 3.1. Low α limit the temperature has changed from its initial value, relative to the temperature change required for the di- It is first worth considering the role played by α in the minished growth term to counteract the benefits of the analysis. The parameter α is the ratio of the tempera- 6 Savitch et al. ture change required to affect population growth (∆T ) 1 1.8 relative to the temperature change needed for the cli- 1.6 mate sensitivity to decrease (∆TF ). For present day 0.8 1.4 conditions on Earth, ∆TF ≈ 4 K (IPCC 2014). This means that if 4 K of warming is sufficient to impact the 1.2 0.6 rate of human population growth (∆T < 4), then α . 1. 1 In the limit where α 1 we can take only the highest 0.8 0.4 order terms and our equations become 0.6 dη 0.4 = min [η (1 + θ) , 1 − η] − η2 0.2 dτ 0.2 d = γη 0 0 dτ 0 50 100 150 200 Figure 3: Evolution of population η and temperature for cases with low forcing γ << 1 and constant climate sensitivity α = 0. The population reaches carrying capacity quickly followed by a 3.1.1. Low Climate Forcing (γ 1) slower rise in temperature over a timescale ≈ γ−1 If γ is small, the climate forcing d/dτ remains small even as the population reaches the carrying capacity then grow exponentially (or faster if aided by technology (η → 1). After the population reaches the carrying ca- when θ & 1). The temperature anomaly () will also 1 pacity, the climate will be slowly forced on a timescale grow exponentially (or faster) until η ' γ and ' 1. γ−1, causing the population to also decline on the same Figure4 shows trajectories for cases with θ = 0 and time scale. Figure3 shows trajectories for the semi- γ = 10 and 50. In both cases the population rises ex- analytic model for γ = .01, γ = .05 and θ = 0 as well as a ponentially and then begins to turn over when ≈ 1 1 model with γ = .05 and θ = 1. Note the exponential rise and η ≈ γ . The case with the larger γ = 50 forces the of the population (η) to the carrying capacity followed temperature faster and as a result has a smaller peak by a slower, linear rise in the temperature () over time population which it achieves earlier. In both cases the scales ≈ γ−1. The change in slope during the decline timescale for the population to decline is of order τ = 1. is due to our birth rate (regardless of population and Figure4 also shows the trajectory for the case with technology) being capped at 1 − η as a means of imple- γ = 50 but with a technology benefit θ = 5. In that menting a carrying capacity. As the population begins case the rise is accelerated due to a technology assisted to decline, for a while it is able to maintain the maxi- enhanced birth rate resulting in not only an earlier peak, mum birth rate (which itself is increasing). This helps but one that also has a higher population. This results to counter the increased death rate due to the changing in faster environmental forcing which then leads to a climate. Eventually, however, the population drops far shorter collapse time. For very large θ, the min function enough that the technology enhanced birth rate is less acts to ensure that technology does not accelerate birth than the enhanced death rate at which point the popula- rates to an unrealistic value by enforcing a maximum tion drops off more precipitously. For θ = 0 this occurs growth rate given by 1−η. Thus, in these cases, initially at η = .5 while for the case with θ = 1 this transition θ accelerates births so much that the maximum rate is occurs later as technology is able to assist in increasing exceeded, at which point this maximum rate takes over, the birth rate for a longer time. Also note in this case allowing the civilization to smoothly enter their collapse. that θ has little affect on the initial growth or early de- As they begin to collapse, η begins to decrease so that cline. For this model β ≡ γθ 1 so CO2 consumption 1 − η starts to increase. Eventually we return to the is not significantly altering the growth rate until after initial state where 1 − η > η(1 + θ). At this point the the population peaks. Also, θ does not affect the initial min function switches to using the technology enhanced decline because the birth rate is at the carrying capacity birth term, which has the effect of softening the decline. limit, 1 − η, and is independent of the amount of CO2 This can be seen during the collapse of the trajectory consumed. with γ = 50 & θ = 5 in Figure4 as a slight deviation in the slope of η (i.e. a ’shoulder’). 3.1.2. High Climate Forcing (γ 1) Now we can estimate the collapse time by looking at In the limit of high climate forcing, the population is the second derivative at the peak population. able to force (i.e. change) the climate while being well below the carrying capacity (η < 1). The population will Triggering A Climate Change Dominated ”Anthropocene”: Is It Common Among Exocivilizations?7

0.16 0.5 7

0.14 6 6 0.4 0.12 5 5 0.1 0.3 4 4 0.08 3 3 0.2 0.06 2 2 0.04 0.1 0.02 1 1

0 0 0 0 246810 0 5 10 15 20 25 30 35 40

Figure 4: Evolution of population η and temperature for cases Figure 5: Evolution of population η and temperature for cases with high forcing γ 1 and constant climate sensitivity α = 0. with high forcing γ = 10, a modest CO2 birth beneﬁt θ = 1 and The population pushes the climate well before reaching the diﬀerent values for α. Cases with larger α have shallower carrying capacity followed by a faster collapse on the order of the declines because the climate sensitivity is decreasing as the growth time - or even faster if the growth rate going into the temperature increases. Anthropocene is accelerated by technology. amount of increased death due to changing temperatures, resulting in a similar tendency.

s η τ = − 4. RESULTS OF FULLY COUPLED MODEL coll d2η 2 dτ dη It is important to understand the meaning of the so- dτ =0 lution domains we have just explicated in terms of the goal of the study. We are interested in the ubiquity of To simplify things we will assume that γ is large climate dominated anthropocenes. Thus in this study enough so that the technology assisted growth rate never we ask, given different planetary initial conditions, how exceeds the carrying capacity limit (1 − η). In that case often will a civilization’s energy harvesting (in this case we find that (see Appendix B) via combustion) lead to population growth which then leads to rapid climate change which, finally, leads to 1 τcoll = √ (γ 1) (14) rapid population decline. Addressing this question is the max( 2, θ) specific intent of our modeling. As demonstrated by the Combining this with the decline times for the γ 1 analysis above, however, some solutions exist in which case, we have the population rises to the planet’s carrying capacity be- 1 1 fore the changing climate produces adverse effects. This τ = max , √ (15) would not be a climate dominated anthropocene in our coll γ max( 2, θ) definition. 3.2. Role of α It is, however, worth noting that a rapid (exponen- Finally we turn to the role of α, which determines how tial) population rise to the host world’s carrying capac- much the climate sensitivity will decrease as the temper- ity (Nmax) will bring its own potentially existential chal- ature changes. Figure5 shows the results of runs with lenges. The very definition of carrying capacity implies γ = 10, θ = 1, and α = 0, 1, & 2. In the cases with that at when N = Nmax, the civilization is at the edge α > 0, increasing temperatures will reduce the climate of what the planet can provide in terms of ”ecosystem sensitivity before reducing net growth rates, allowing services”. Thus, while these classes of systems will not civilizations to reach higher peak populations while also fall under our definition of climate-dominated anthro- delaying the time for them to be reached. Furthermore, pocenes, they should not be considered to be cases that the initial decline begins to taper off as both the de- have escaped the possibility of rapid population declines creasing climate sensitivity and increasing birth rate act or even collapse. It is simply that our models do not to reduce the decline in population. It may be unrea- include the processes (i.e. biospheric feedbacks) which sonable to expect the birth rate to continually increase could produce them. over many generations due to CO2 consumption, however CO2 consumption might also act to mitigate the 8 Savitch et al.

Portner 2013). For a lower value we choose 10 ppm (we

could not go to zero due to limitations imposed by our EBM). 2 We begin with two sets of experiments that illustrate the basic behavior of the civilization-planet system. The 2

AU first of these are ”constant composition” models. These keep initial pCO2 constant and allow the initial (equilibrium) planetary temperature, T0, to vary as we change Distance the orbital distance (a). Using T0 = 287.1K as our 2 ppm fiducial value, we ran four additional models with tem- ppm ppm peratures evenly spaced above and below T0 in steps of 6K. This spacing was chosen in order to have all models 2 2 2 Initial Temperature K safely within the habitable zone (273K < T0 < 373K). In Figure6 we show the location of the models in the Figure 6: Parameter space and model locations for two initial experimental sets. Experiment #1: Constant Composition was (a, T0) plane. This representation is important because defined by constant levels of initial pCO2 = 284 ppm, and is we will later overlay contours of various quantities such denoted by the pluses (+). Experiment #2: Constant as γ and the collapse time τ when we run a larger Temperature was defined by constant initial temperature coll array of models that sweep across (a, T0) space. T0 = 287.09K, and is denoted by the crosses (×). The two experiments intersect at Earth when initial pCO2 = 284 ppm The second set of experiments are ”constant tempera- and T = 287.09. 0 ture” models. These keep initial temperatures (T0) constant and allow the initial pCO2 to vary as we change 4.1. Results: Constant Temperature and Composition the orbital distance (a). Five models were run centered Models. on Earth’s current pCO2 concentration with two below and three above, each spaced by log(pCO ) = 0.7. The We now return to the full non-linear model described 2 highest, P = pCO ≈ 28, 000 ppm was meant to il- in section 2. We have carried forward a large suite of 0 2,0 lustrate the evolution a system with CO concentration numerical experiments using this model with the goal 2 beyond what animal life on Earth can tolerate. of investigating how the trajectory of coupled planet- We now explore the trajectories occurring along our civilization systems depend on various initial conditions. lines of constant T and a. Figure7 shows four runs To recap we set up our initial conditions with two key 0 from our two sets of experiments. Consider first the assumptions. (1) The biology of the organisms building model in the upper left of the figure which corresponds the civilization requires liquid water, so the host planet to a constant initial composition case with a = 1.021 must be within the star’s habitable zone. (2) The organ- AU. This model begins with log(pCO ) = 2.28. Using isms have temperature and pCO limits beyond which 2 0 2 the climate model to determine the planet’s climate sen- they can not survive. Taken together, these two con- sitivity dT/dP , along with the other model parameters, ditions define a ”Complex Life Habitable Zone” (CLHZ yields a γ = 2.663. Since γ > 1 we expect this model Schwieterman et al.(2019)), as discussed in the intro- to experience a climate anthropocene. This is, in fact, duction. For illustrative purposes, we will begin our what occurs as we see a steep rise in population begin- study by using values similar to Earth and human life ning at approximately t = 2700 y. The population then for these limits, but will always consider them to be free peaks at N = 10.5 billion individuals about two cen- parameters. peak turies later. After the peak, the population declines by We focus on two initial planetary parameters: the or- half over the following two centuries. The cause of the bital distance from the planet to its host star (a), and peak and decline can be see in the rising pCO levels. the initial chemical composition of the atmosphere quan- 2 At N we find log(pCO ) ∼ 2.9. The increased plan- tified in terms of P = pCO . The effect of these param- peak 2 0 2 etary greenhouse effect has driven temperatures almost eters on the models are not fully independent. Planets past the civilization’s tolerance ∆T . Note that in this on the inner edge of the habitable zone (a < 1) have model the population never comes close to the planet’s climates that are less stable against small increases in carrying capacity of N = 20 billion. CO when compared to similar perturbations on dense max 2 In the lower left panel of Figure7 we show a second CO planets near the outer edge (a > 1). We chose our 2 constant initial composition case, though in this model boundaries in pCO to have an outer edge correspond- 2 a = 0.9795 AU. Since log(pCO ) is the same as the ing to 5, 000 ppm as this represents the upper limit of 2,0 model above, it also has γ = 2.392 and we expect a cli- pCO2 amenable for animal life on Earth (Wittmann & Triggering A Climate Change Dominated ”Anthropocene”: Is It Common Among Exocivilizations?9

Distance: 1.021 AU, Carrying Capacity: 20 billion ppl, : 2.663 Distance: 1.0287 AU, Carrying Capacity: 20 billion ppl, : 0.252 Population vs Time Population vs Time 10.78 17.50 Npeak = 10.5 billion

N1/2 = 5.2 billion 7.19 11.69 Nfinal = 5.5 billion N = 17.0 billion NA = 7.5 billion peak N = 8.5 billion 3.60 5.89 1/2 Nfinal = 12.8 billion

NA = 79.5 billion Population (billions) 0.01 Population (billions) 0.08 Temperature vs Time Temperature vs Time 281.89 3.24 289.47 3.97 ) ) m m p p

278.83 2.92 p 287.94 p

( 3.89 ( 2 2

tpeak = 2923 O tpeak = 2744 O C C

275.76 2.60 p 286.40 3.82 p Teq = 273 K Teq = 285 K g g o o l l Temperature (K) Teq + T = 278 K Temperature (K) Teq + T = 290 K 272.70 2.28 284.86 3.74 2400 2600 2800 3000 3200 3400 2000 2200 2400 2600 2800 3000 3200 Time (years) Time (years)

Distance: 0.9795 AU, Carrying Capacity: 20 billion ppl, : 2.392 Distance: 0.974 AU, Carrying Capacity: 20 billion ppl, : 181.096 Population vs Time Population vs Time 10.8 0.09854 Npeak = 0.1 billion

N1/2 = 0.0 billion 7.2 Npeak = 10.5 billion 0.06603 Nfinal = 0.0 billion N1/2 = 5.2 billion NA = 0.1 billion Nfinal = 4.8 billion

3.6 NA = 8.4 billion 0.03351

Population (billions) 0.0 Population (billions) 0.00100 Temperature vs Time Temperature vs Time 306.69 293.64 3.20 1.43 ) ) m m p p

303.32 2.86 p 290.70 1.28 p ( ( 2 2

tpeak = 2905 O tpeak = 2927 O C C

299.96 2.52 p 287.77 1.13 p Teq = 297 K Teq = 285 K g g o o l l Temperature (K) Teq + T = 302 K Temperature (K) Teq + T = 290 K 296.59 2.18 284.83 0.97 2400 2600 2800 3000 3200 3400 2000 2200 2400 2600 2800 3000 3200 3400 Time (years) Time (years)

Figure 7: Selected trajectories. The left column shows the highest and lowest distance from experiment #1 (top and bottom respectively). Similarly, the right column shows the same for experiment #2. Note that on the left column, the value of γ changes marginally, this is due to the marginal difference between values of dT dP at these two locations. In contrast, on the right column, the change in gamma is drastic, corresponding to the drastic change in dT dP resulting from different levels of initial pCO2. mate anthropocene. Because the γ values for these two as such, a significantly lower initial pCO2 is required runs are relatively similar we find similar trajectories in (log(pCO2,0) = 0.97). Because this world is closer to terms the rise, peak and decline in population N along it’s star and begins with a lower CO2 concentration it is with a relatively steady increase in log(pCO2) and T . more sensitive to changes in pCO2. This is reflected in In the upper right panels of Figure7 we show a con- its value of γ = 181. Examining the trajectory we once stant T0 model with a = 1.029 AU. This world is further again see the rapid rise and fall of the population. In this from its (solar type) star so it requires a higher pCO2 model however the extreme climate sensitivity which −3 to maintain T0 = 287 K. Thus with log(pCO2)0 = 3.74 drives γ 1 manifests itself in Npeak ∼ 10 Nmax. we find a γ = 0.252. In this case γ < 1 and we ex- The population never comes close to the planet’s carry pect this model to avoid a climate anthropocene. The capacity before the climate is driven into detrimental results show this to almost be the case. The population states (T ∼ T0 + ∆T ). Note that this behavior was peaks at Npeak/Nmax = .89. Thus the initial exponen- captured in the value of the anthropogenic population tial growth phase of the population is able to carry the defined in Appendix A, which, for this model, takes on civilization close to the planets’ carrying capacity before the value NA = 0.1 billion individuals. a decline sets in. Note also the longer period between It is also of interest to consider the trajectories for the onset of initial exponential growth and time for the planets with initial pCO2 greater than that which Earth population to reach Npeak. life can tolerate. These worlds would orbit at larger Finally in the lower right we show the trajectory for radii (a) and would require higher greenhouse gas con- second case with a constant initial temperature. This centrations to maintain habitable environments. Fig- planet is closer to its sun with a = .974 AU and, ure8 shows such a model with a = 1.06 AU and 10 Savitch et al.

Distance: 1.0596 AU, Carrying Capacity: 20 billion ppl, : 0.091 In the top left we present contours and color mapping Population vs Time 19.30 of the initial atmospheric composition, log(pCO2,0), for all the runs. This was calculated using only the uncou- 12.94

Npeak = 18.7 billion pled energy balance model. Note that we exclude mod- N = 9.4 billion 6.58 1/2 Nfinal = 17.0 billion els with log(pCO2,0) > 3.7 as being outside the CLHZ

NA = 219.5 billion

Population (billions) (Schwieterman et al. 2019). 0.22 Temperature vs Time The top right panel in Figure 10 presents initial γ, 287.74 4.47

) deﬁned in Table1. Recall that γ is dependent principally m p

286.84 4.44 p ( on dT/dP |P =P0 , which itself is dependent principally on 2

tpeak = 2797 O C initial pCO2. Thus the contour lines of initial pCO2 also 285.93 4.41 p Teq = 285 K g o l Temperature (K) Teq + T = 290 K correspond to contour lines for γ. This plot shows that 285.03 4.38 2000 2200 2400 2600 2800 3000 3200 only the outermost layers of orbits have γ < 1, indicative Time (years) of worlds not at risk of a climate Anthropocene. Recall Figure 8: This model results in the civilization reaching their that even models with with γ slightly less than 1 can still carrying capacity, thus overpopulating their planet. The initial have their growth truncated by climate effects. Thus 4.38 pCO2 is 10 ≈ 24, 000 ppm, which puts it in the ”danger we find that most civilizations arising in CLHZ will be zone” for habitability. susceptible to having their growth truncated by rising temperatures and changing climate. log(pCO2,0) = 4.38, yielding a γ = 0.09. The trajectory The middle row describes the population dynamics for this model shows the population climbing to within for the civilizations and focuses on parameters associ- 1.5% of the planet’s carrying capacity or (N ∼ Nmax). ated with growth. The left column presents the numeri- When this peak occurs the global mean temperature is cally measured percentage of the carrying capacity each still well below the threshold for a climate dominated civilization reached (Npeak/Nmax). The right column Anthropocene (T < T0 + ∆T ). Thus for this case the shows the analytic predictions of the quantity NA, de- population grows until it reaches Nmax without driving fined in Appendix A as the number of people required the climate into a significant new state that is detrimen- to force the climate out of equilibrium in a single growth tal to the functioning of the civilization. The trajectory timescale. In Npeak/Nmax we see only the outermost or- shown in this figure is typical for worlds with very high bits at each initial temperature are able to rise to their log(pCO2)0. The addition of more CO2 via combustion carrying capacity before increasing temperatures signifi- does not alter the climate significantly before the pop- cantly increase death rates and halt population growth. ulation rises to levels beyond what the planet, via its Note all models in the parameter sweep began with a carrying capacity, can sustain. carrying capacity of Nmax = 20 billion. Consideration In Figure9 we show all trajectories for both exper- of the NA contour plots demonstrates how the linear iments. The upper panel shows results from constant model accurately identifies the peak population possible initial composition models while the lower panel shows in climate Anthropocene worlds. those for constant initial temperature models. The sim- Finally, the last row of plots considers what happens ilarities and differences both within and between the after N = Npeak. On the left we show the time for popu- models offers insight into the dynamics and its relation- lation to decline by 20%. Here we see most of the models ship to our dimensionless parameters. For example, the experience a decline on timescales of a few centuries at value of γ can be seen to uniquely determine the re- most, while initially hotter worlds on inner orbits can sulting evolution of population. Furthermore, the upper have declines over just decades. The collapse time pa- panel shows that models with constant pCO2 yet differ- rameter τcoll is shown on the lower right. Once again we ent orbital radii and initial temperatures will share the see timescales of order decades to a few centuries associ- same γ. ated with significant population decline. Note however 4.2. Results: 2D Parameter Sweeps that τcoll shows a ”valley” feature at intermediate orbital distances where τ falls and then rises again as To explore the broad dependence on initial conditions coll one moves outward in orbital distance along a line of we next choose 100 different distances and temperatures constant T . We return to this valley which reflects the for the models. The results of this parameter sweep are 0 dependence of τ on 1/γ in the next section. shown as contour plots in Figure 10. The left column of coll the plots show quantities taken from the full numerical 4.3. Dependence of Civilization Temperature Tolerance models, while the right column shows the corresponding analytical quantities. Triggering A Climate Change Dominated ”Anthropocene”: Is It Common Among Exocivilizations?11

1.021 AU ( = 2.66)

) 10

9 1.011 AU ( = 2.09) 0

1 8 1.000 AU ( = 2.08) × (

0.989 AU ( = 2.25)

n 6 o

i 0.980 AU ( = 2.39) t a

l 4 u p

o 2 P

0.974 AU ( = 181.10) )

9 0.991 AU ( = 14.99)

0 15

1 1.000 AU ( = 2.08) × (

1.010 AU ( = 0.67) n 10 o

i 1.029 AU ( = 0.25) t a l 1.060 AU ( = 0.09) u

p 5 o P

0 1800 2000 2200 2400 2600 2800 3000 3200 3400 yr

Figure 9: The top plot shows the trajectories for Experiment #1: Constant Composition, defined by constant levels of initial pCO2 = 284 ppm. The bottom plot shows the trajectories for Experiment #2, defined by constant initial global temperatures (T0 = 287.09K). The most influential quantity for the co-evolution of intelligent civilizations and their planets is the quantity dT dP , which is a function of pCO2. The value of pCO2 was found in order to make any given distance have any given temperature. Thus, the value of initial pCO2 principally determines the resulting co-evolution. This is why the trajectories for experiment #1 are very similar, as they all have approximately equal initial levels of pCO2. In contrast, the trajectories in the bottom plot all have the same temperature, but different distance, thus also have different levels of pCO2 resulting in very different values for dT dP . This is why the trajectories in this plot are much more diverse. All the models discussed in the last section used a dividual. We consider such a situation as likely to pose constant value of the civilization’s temperature toler- the greatest risk for civilizations. ance, ∆T = 5K. As discussed earlier this parameter is We also show data for the models as a scatter plot of intended to capture biological, ecological and cultural ef- gamma versus decline time in Figure 12. On the x and fects of the civilization’s response to its climate forcing. y axis we show corresponding marginal probability dis- Depending on the fragility of both the social organiza- tributions for γ and the decline time. These are shown tion and ecosystems on which it depends, a smaller ∆T as a kernel density estimation graph such that the area may be enough to trigger detrimental consequences for under each curve is normalized to one. the civilization. Thus, in this section we vary ∆T to Shown as the solid black line is our analytically de- asses its role in model outcomes. rived collapse time, given by equation (15). It can be To see the effect of ∆T on the models, Figure 11 shows seen that our analytic approximation reproduces the the marginal probability distributions for both popula- necessary features that arise in our numerical runs. The tion decline times and for γ. These are represented as deviations are explained in Figure 13 in Appendix A. ”violin plots” showing the range of values for both for There are three distinct regions for τcoll. Starting from runs with a given ∆T . These plots also show the average the right hand side of Figure 12, we are in the region of and first moment of the distributions. high γ and low θ. We say that civilizations in this re- Note that the median values for the collapse times in- gion experience a climate-dominated anthropocene. The crease with ∆T . This is to be expected as the sooner the high γ indicates that the timescale for the climate to planetary temperature rises beyond T0 + ∆T the sooner change is much shorter than the timescale for popula- the population growth is truncated by climate effects. tion growth. The low θ indicates that the timescale for The values of γ extracted from the models reflect this, climate to change is also much shorter than the timescale showing a decrease with increasing temperature toler- for technology to evolve. Thus, in this region, civiliza- ance. Note that for ∆T < 5K, the decline times are less tions experience climate change before they experience than, or of order, a century. Given that the timescale any significant growth benefits due to their technologi- for a generation in our models is ∼ 25 y, this means cal capacities. Thus these civilizations reach only a tiny significant population loss across the lifetime of an in- fraction of their carrying capacity before they begin to decline. Since they do not have much time to increase 12 Savitch et al.

log Initial pCO (ppm) log 2 eff 1.0

3.5 1.04 1.04 0.8

1.02 3.0 1.02 0.6 ) m p ) ) p ( U 1 U 0.4 A A 0 1 2 f

( . (

6 f 1.00 0 2.5 1.00 O e e e

p C

c p c

m p g n n l

0.2 o a a l a t t i t s s i i i n D D 0.98 2.0 I 0.98 g

o 0.0 l 10 2. 40 1 p 1 0 1 p 0 1 . m .0 60 0.96 p 1.5 0.96 0.2 p 1 3 m 0 .20 10 0.0 ppm 10 0.5 0.4 0.94 1.0 0.94 280 290 300 310 320 330 340 280 290 300 310 320 330 340 Initial Temperature (K) Initial Temperature (K)

Percent of Carrying Capacity Achieved log N (ppl) A 1.0

1.04 1.04 0.8

80 0.6 1.02 1.02

0.4

) 60 ) ) U U l A A p ( (

1.00 1.00 0.2 p ( e e c c A n n N a a t t g

s s 0.0 o i i 40 l D 0.98 D 0.98 0.2

10 0. 00 25% 75% 20 bi 0.4 0.96 50% 0.96 ll 1 io 10 .00 Percent of Carrying Capacity Achieved n bi llion 0.6

0.94 0.94 280 290 300 310 320 330 340 280 290 300 310 320 330 340 Initial Temperature (K) Initial Temperature (K)

log Time to Decline by 20% of Peak Population (yrs) log coll/R0 (yrs)

1.04 1.04 3.5 4.0

1.02 1.02 3.0 3.5 ) ) ) s r U U y A A (

( 1.00 ( 1.00 0 e e R / c c 3.0 l l n n o c a a

t 2.5 t s s i i g D D o 0.98 0.98 l 2.5

10 2 .50 2.0 1 2 yr 0 .00 0.96 yr 0.96 2 2.0 10 2.00 10 .50 yr yr 1 0 2. 00 y

log Time to Decline by 20% of Peak Population (yrs) r 0.94 1.5 0.94 280 290 300 310 320 330 340 280 290 300 310 320 330 340 Initial Temperature (K) Initial Temperature (K)

Figure 10: The plots on the left hand side show numerically calculated quantities, while the plots on the right hand side show analytically derived quantities. The top row shows values calculated with the uncoupled energy balance model. The middle row shows quantities related to the populations growth. The bottom row shows timescales related to the populations collapse. The bottom left of the plots are grey because of limitations imposed by the EBM. The top right part of the plots are grey because the value of pCO2 required there was greater than 5, 000 ppm, a level deemed uninhabitable for long-term habitability by intelligent civilizations (Schwieterman et al. 2019). The black arrow points to the location of the experiment that we ran in the ’danger zone’. Triggering A Climate Change Dominated ”Anthropocene”: Is It Common Among Exocivilizations?13 their overall growth rate above the ”natural” value, their time, and the marginal distribution is a projection onto collapse rate ends up being approximately equivalent in the y-axis. This can also be seen as the large shoul- magnitude to their natural growth rate. der protruding in the distribution of collapse times for As we move leftward in Figure 12 to lower γ, we see ∆T = 0.5K. a dip in the collapse times (this is the valley seen in The most important takeaway from these results is the contour plots discussed in the last section). This that an ever increasing share of the models experience region, between the two black dotted lines, corresponds climate dominated anthropocenes as ∆T decreases. For to high γ and high θ. We say civilizations in this re- ∆T < 5K, most models experience rapid population gion experience a technology-dominated Anthropocene. declines. Even for ∆T = 10K, the average decline time The high γ indicates that the timescale for climate to was 384 years and 22.2% of models had decline times change is much shorter than the timescale for popula- less than 200 years. tion growth. The high θ indicates that the timescale for climate to change is now longer than the timescale for 5. DISCUSSION AND CONCLUSIONS technology to evolve. Thus civilizations in this range In this paper we have modeled the coupled evolution of experience a birth benefit due to their technological ca- a planet and a civilization through the era when energy pabilities before they experience adverse effects stem- harvesting by the civilization drives the planet into new ming from climate change. This is the region Earth and potentially adverse climate states. Our goal, contin- currently inhabits. The technological birth benefit al- uing from the work of Frank et al.(2018), is to determine lows the civilization to begin approaching their planet’s if ”anthropocenes” of the kind humanity is experienc- carrying capacity. However the high γ value prohibits ing now might be a generic feature of planet-civilization them from reaching this limit. As we move towards the evolution (Haqq-Misra and Baum 2009; Frank Alberti leftmost vertical dotted black line in the plot, civiliza- & Kleidon 2017; Mullan and Haqq-Misra 2019). To tions peak closer to their carrying capacity. Since the this end we introduced and analyzed a set of coupled decline time for civilizations correlates to their growth ODE’s that track the trajectory of the civilization’s pop- rate, civilizations whose populations begin falling closer ulation (N), the generation of CO2 via the civilization’s to their carrying capacity experience the fastest decline energy harvesting activities and, finally, the mean plan- times. In essence, in this region the more technology etary temperature T . The principle innovation in this props a civilization up, the harder they end up falling. paper over (Frank et al. 2018) is the use of an energy Continuing leftward in Figure 12 to lower γ, we begin balance model to track the change in climate state as to leave the region of the Anthropocene and approach the atmospheric composition, P = pCO2, changes due the limit of overpopulation. In this region, θ is much to the civilization’s growth. A different form for the greater than one, while γ is now decreasing < 1. Low population growth equation compared to (Frank et al. γ means that the timescale for population to grow is 2018) was also used. now much shorter than the timescale for the climate to Using both direct simulation of the non-linear model change. High θ again indicates that the timescale for cli- and an analysis of a linearized, non-dimensional ver- mate to change is also much longer than the timescale sion of the equations we have found that most planets for technology to evolve. Thus, in this region, civiliza- in the Complex Life Habitable Zone (CLHZ Schwieter- tions experience a birth benefit due to their growing man et al.(2019)) undergo a climate-dominated anthro- technological capacities, allowing them to rise in popu- pocene. We define this to be a trajectory of the coupled lation towards their carrying capacity. As γ drops below planet-civilization system in which population growth is 1 civilizations are able to reach their carrying capacities truncated by changes in the climate state driven by the (η → 1). In this region, the timescale for these civiliza- civilization. Our analysis further shows such climate- tions to decline is then dictated by the timescale for cli- dominated anthropocenes can occur in different ways mate to change, thus resulting in a longer collapse time depending on the planets initial atmospheric composi- τcoll = 1/γ = tC /tG. Moving further leftward yields ever tion, orbital location and the technological capacities of lower values of γ < 1 and the timescale for climate to the civilization. On planets with high climate sensitivity change becomes larger resulting in a steadily increasing (dT/dP ), even modest technological innovation in terms collapse time. of energy harvesting capacities can trigger detrimental Finally, we note that in Figure 12 the marginal distri- climate change (Figure5). This situation is likely to oc- bution of the decline time for ∆T = 1K seems to peak cur at the inner edge of the habitable zone and would be higher than ∆T = 0.5K. This occurs because many of akin to climate change occurring during the early era of the ∆T = 0.5K models are in the valley of our collapse industrialization on Earth (i.e. the Victorian period in 14 Savitch et al.

Log Time to Decline by 20% of Peak Population (yrs) 3.5

2.8

2.2

1.5 T = 0.5K T = 1K T = 2.5K T = 5K T = 10K

log( ) 4.5 eff

2.5

0.5

1.5 T = 0.5K T = 1K T = 2.5K T = 5K T = 10K

Figure 11: This ﬁgure shows a box plot of the marginal distributions shown in ﬁgure 12. The white dot represents the median value. The box’s lower bound corresponds to the median of the lower half of the data-set, while the box’s upper bound is the median of the upper half of the data-set. It is of interest to note how increasing the population-temperature sensitivity parameter (∆T ) results in a steadily decreasing γ and a steadily increasing time for the population to decline.

3.0 log( coll/R0) yrs T = 0.5K T = 1K 2.8 T = 2.5K T = 5K 2.6 T = 10K

2.4

2.2

2.0

1.8

1.6

Log Time to Decline by 20% of Peak Population (yrs) 1.0 0.5 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 log( eff)

Figure 12: Shown above is a scatter plot of gamma versus the time for civilizations to decline by 20% of the their peak population. The solid black line shows our analytically derived collapse time, given by equation (15). The colors represent diﬀerent values of our population-temperature tolerance parameter ∆T . The black dotted lines divide the graph into three regions. The leftmost region corresponds to low γ and high θ. This region results largely in overpopulation, which is the reason behind the long decline times. The middle region corresponds to both high γ and high θ, where technology is able to accelerate birth rates, but eventually ends up contributing to an increased death rate and short decline time. The rightmost region corresponds to a high γ and a low θ, where climate changes on a faster timescale then both technology and population growth. Thus, in this region, civilizations experience climate change before they experience any growth beneﬁts due to technology. This means that these civilizations reach only a tiny fraction of their carrying capacity before they begin to decline. Triggering A Climate Change Dominated ”Anthropocene”: Is It Common Among Exocivilizations?15

England). Planets on the outer edge of the host star’s based on combustion on high pCO2 worlds? If civi- habitable zone require higher values of CO2 concentra- lizations could occur on such planets then our results tions to be temperate. Such worlds have lower climate indicate that these worlds would not undergo a climate- sensitivity and civilizations there can drive higher frac- dominated anthropocene. Instead, in our models, their tional increases in pCO2 before temperatures rise signif- populations climb until they reach the planet’s carry- icantly. The technological capacities of the civilization ing capacity. This situation, however, presents its own effect the trajectories when they allow for population difficulties and could represent a different form of neg- growth rapid enough that it can compete with increasing ative feedback on the planet-civilization system that is death rates from an adversely changing climate. Finally, not modeled in our equations. It is also possible that the tolerance of the civilizations to rising temperatures without rapid detrimental feedbacks from adverse cli- ∆T represents another important input condition and, mate change a civilization would reduce its population as could be expected, lower values of ∆T led to stronger growth on its own before carrying capacity is reached. climate dominated anthropocenes (Figure 11). Finally we note that Howell(2019) criticized the use of The models presented in this paper represent an ad- the CLHZ as being too narrow a a view of evolutionary vance over our initial study (Frank et al. 2018) because possibilities. We view certainly has merit we however they assumed a specific form of energy harvesting (com- that the criticism was based mainly an arguments sur- bustion) and included an explicit physical models for its rounding the biochemistry of CO and not CO2. climate impact via a combustion dependent (i.e. CO2) Future modeling efforts could explore different repre- Energy Balance Model. In this way we added a higher sentations of the population growth and its coupling to level of physio-chemical realism to the planet-civilization climate state and energy harvesting modality. In addi- model system. With the new model we were able to vary tion, other energy harvesting modalities such as wind or parameters such as orbital distance and atmospheric solar could also be explored. Future work should also in- composition for a specific class of worlds (i.e. Earth- vestigate the impact of the stellar spectral type, as this like planets orbiting Sun-like stars). will impact the climate of the planet and may also im- One significant question to arise from our studies is pact the ability of a civilization to utilize stellar energy the applicability of the Complex Life Habitable Zone (as M-dwarfs have a different SED than G-dwarfs, etc.). (Schwieterman et al. 2019; Ramirez 2020). For evo- Finally models that include changes in the civilizations lutionary reasons it is generally believed that a tech- behavior, such as simply switching from one harvesting nological civilization would only arise from complex mode to another could also be included in the modeling. multi-cellular ”animal” life (Carter 2008; Watson Each of these steps would represent a further articula- 2008). Schwieterman et al.(2019) emphasized that tion of an astrobiology of the anthropocene by exploring the oxidation of organic matter via free O2 is the best issues associated with the biospheric dimensions of cre- means of producing significant free energy via respira- ating a long-term sustainable civilization. tion. Given that O2 is the only high-potential oxidant sufficiently stable to accumulate within planetary at- APPENDIX A: INTRINSIC TIMESCALES AND mospheres (Catling et al. 2005) it comprises a neces- DIMENSIONLESS QUANTITIES sary condition for intelligence species. Based on Earth’s In analyzing the model we see that it contains three evolutionary history it is clear that complex aerobic intrinsic timescales. life can also be strongly impacted by CO with lim- 2 1 its for animal life appearing at atmospheric conditions t ≡ = Timescale for Population Growth G R of pCO > 5 × 104 ppm (humans pCO lethality limit 0 2 2 ∆P may be 10 times lower). Thus, if we take Earth’s his- tT ≡ = Timescale for Technological Advancements tory as a guide, atmospheres with high oxygen and low CNmax ∆T CO2 may be necessary for the emergence of intelligent tC ≡ = Timescale for Climate Change civilization-building species. If this is the case then CNmaxD climate-dominated anthropocenes would appear to be a Where generic feature of coupled planet-civilization evolution.

∆TF dT The difficulty however is knowing how universal the con- D ≡ = = Initial Climate Sensitivity P dP straints on pCO2 are for life elsewhere. Is it possible for 0 T =T0 a planet with 105 ppm of CO to evolve complex life 2 as defined in equation (9). Also, R is the initial ”natu- that goes on to create a technological civilization? In 0 ral” growth rate, ∆P is the technological birth benefit, addition, what constraints exist for energy harvesting ∆T is the civilizations temperature tolerance, Nmax is 16 Savitch et al. the carrying capacity and C is the annual, per-capita We can setη ˙ = 0 and solve for to determine what carbon footprint. We can use these timescales to define the environmental state is at the point when population our dimensionless quantities shown in Table1. begins to collapse (ie. the moment whenη ˙ drops below zero). We call this value of its critical value, which we tG DCNmax tC ∆T tG CNmax γ ≡ = θ ≡ = β ≡ = find to be tC R0∆T tT D∆P tT R0∆P s θ θ 2 = + 1 + (B2) Note that β = θγ. Since we have defined γ to be approx- c 2 2 imately constant, based on the initial climate sensitivity, we can define an ’effective’ γ to be based on the value We can also find the time scale for the population to of the climate sensitivity when the population reaches decline after it reaches this peak. Since these civiliza- its peak value, which occurs approximately when global tions do not reach their carrying capacity, their growth temperatures have deviated from their initial values by rate as they enter the Anthropocene dictates what their ∆T . collapse rate will be directly after. Thus, we use the free CNmax dT fall time for civilizations as their collapse rate. γeff ≡ (A1) R0∆T dP T ≈T +∆T r 0 η As defined by equation9, this means that τcoll = − (γ 1) η¨ η˙=0 CN ∆T max F −∆T/∆TF γeff = e We can use our population growth rate equation to find R0∆T P0 this second derivative DCNmax −α −α ≡ e = γe (A2) η¨ = η(θ˙ − 2˙) R0∆T

Where we have defined α ≡ ∆TF /∆T . Thus, as tem- Where we let ˙ ≡ d/dτ. When population begins to de- peratures increase, the civilizations effective γ will drop cline, global temperatures continue to rise exponentially. from its fiducial value. See Figure 13 to visualize this Thus, at this point, the rate of increasing temperatures effect. Furthermore, the value of the carrying capacity is approximately equal to its critical value ˙c ≈ c. Thus, that makes γ = 1 is indicative of the number of people it follows that at this point, directly afterη ˙ = 0... required to force the climate out of equilibrium in a sin- 2 −η¨ ≈ η(2c − θc) gle growth timescale (tG). This quantity is called the civilizations ”anthropogenic population” Thus, it follows that the collapse timescale for γ 1 is

γ R0∆T given by NA ≡ = = Anthropogenic Population Nmax DC 2 −1/2 τcoll = 2c − θc (γ 1) APPENDIX B: DERIVATION OF COLLAPSE TIME This can further be broken down into two cases, depen- FOR HIGH CLIMATE FORCING dent on θ. For high climate forcing (γ 1) we ﬁnd that the collapse time is given by (1) Low Technological Growth Acceleration  √ If θ 1, then c → 1 and 1 1/ 2 θ 1 √ τ = √ = (B1) coll τcoll = 1/ 2 max( 2, θ) 1/θ θ 1

We can derive this from our linearized model as follows. In this case, the civilizations technological abilities are not advanced enough to increase their growth If γ 1, then tC < tG, so the climate changes on a much faster rate than the population does. This is the rates. Also, the fastest timescale is that for cli- region of the climate-driven Anthropocene, which means mate change. Since their rate of technological ad- that global temperatures will begin to increase death vancements is negligible and their rate of climate rates before civilizations reach their carrying capacity. change exceeds that for population growth, these A consequence of this is that civilizations in this region civilizations will have negligible population growth never reach their maximum growth rate, η 1, which before entering their Anthropocene. As a result, implies that 1−η > η(1+θ), so their population growth their growth rate as they enter the Anthropocene equation will have the form. will be approximately equal to its initial ”natural” value. Thus, all civilizations in this area will also η˙ = η(1 + θ) − η2 = η(1 + θ − 2) ≡ dη/dτ collapse with a constant ”natural” collapse rate. Triggering A Climate Change Dominated ”Anthropocene”: Is It Common Among Exocivilizations?17

T=10K T=5K T=2.5K T=1K T=0.5K log( coll/R0) yrs 1.0 1.040 3.0 ) U A )

1.015 ( s r e y c ( 2.5 n

a n

0.8 t o s

i 0.990 i t D a l l u

2.0 a p t i o

0.965 b P r

k O

a 0.6 e 1.5 P

0.940 f

o 1 0 1 2 3 1 0 1 2 3

0 log( ) log( eff) 2

b 330.0

e 0.4 ) n 3.0 i l K ( c e 317.5 e r D

u t o a t 2.5

r e e

305.0 p m

i 0.2 m T Density e

T g 2.0 o l L a

292.5 i t i n I 1.5 0.0 280.0 0.01 0 0.21 2 0.4 3 1 0.60 1 20.8 3 1.0 log( eff) log( eff)

Figure 13: The top left plot shows our numerically calculated values of γ, as defined in table1 and derived in appendix A, plotted versus the numerically calculated times for our models populations to decline by 20% from their peak values. The black dotted line shows our analytical expression for collapse times, derived in appendix B. For contrast, the bottom left plot shows the same things yet instead for γeff, as given by equation A1. The biggest difference as compared to the plot above is a net decrease in γ, which is as expected from equation A2. This also has the effect of greatly reducing the deviations from our analytical predicitoins, that is, adheres much more to our check-mark shaped prediction. Although, not all deviations have been solved by using γeff. The right column shows plots of γeff colored by orbital distance (top), and initial global temperature (bottom). It can be seen that the models that deviate greatest from our analytical prediction or those that have large orbital radii and low initial global temperatures. As shown in figure 10, the contour lines for pCO2 travel diagonally across the parameter space of a/T0. Thus, as a result, civilizations with high orbital distance and low initial global temperature could have the same value of initial pCO2 as the civilizations with low orbital distance and high initial global temperature. Since γ is principally dependent on the value of the initial pCO2, this means that these two classes of civilizations will have the same value of γ and hence the same value of τcoll. Although, in reality, the civilizations with the higher orbital radii end up taking longer to fall, hence have longer decline times. (2) High Technological Growth Acceleration ties. This accelerated growth leads to an accelerated decline, which means that these civilizations collapse If θ 1, then c → θ and at a faster than natural rate. τcoll = 1/θ

In this case the civilizations technological abilities are able to accelerate their growth rates. Thus, these ACKNOWLEDGMENTS civilizations enter their Anthropocene with an accelerated growth rate due to their technological abili- This work was supported by NASA grant #80NSSC20K0622. REFERENCES

Carroll-Nellenback, J., Frank, A., Wright, J., Scharf, C. Carter, B. (2008). Five- or six-step scenario for evolution? 2019. The Fermi Paradox and the Aurora Eﬀect: International Journal of Astrobiology, 7(2), 177-182. Exo-civilization Settlement, Expansion, and Steady doi:10.1017/S1473550408004023 States. The Astronomical Journal 158. doi:10.3847/1538-3881/ab31a3 18 Savitch et al.

Catling, D. C., Glein, C. R., Zahnle, K. J., McKay, C. P. Howell, J. E., 2019 ”Comment on ”A Limited Habitable

2005. Why O2 Is Required by Complex Life on Habitable Zone for Complex Life”, Res. Notes AAS 3 85 Planets and the Concept of Planetary “Oxygenation Jones, E. M. 1976. Colonization of the Galaxy. Icarus 28, Time”. Astrobiology 5, 415–438. 421. doi:10.1089/ast.2005.5.415 Kot, M., 2011, Elements of Mathematical Ecology, Cohen JE. Population growth and earth’s human carrying Cambridge. capacity. Science. 1995 Jul 21;269(5222):341-6. doi: Kleidon, A. (2010) Life, hierarchy, and the thermodynamic 10.1126/science.7618100. machinery of planet Earth. Physics of Life Reviews, 7, Crutzen, P. J. 2002. The “anthropocene”. Journal de 424-460. Physique IV 12, 1–5. doi:10.1051/jp4:20020447 Kleidon, A., 2012. How does the Earth system generate and Elhacham, E., Ben-Uri, L., Grozovski, J. et al. 2020. Global maintain thermodynamic disequilibrium and what does it human-made mass exceeds all living biomass Nature imply for the future of the planet?. Phil. Trans. R. Soc. 588, 442–444, https://doi.org/10.1038/s41586-020-3010-5 A 370: 1012-1040. A. G. Fairen and J. D. Haqq-Misra and C. P. McKay 2012 Kennett D.,& Beach, T. P. (2013) Archeological and Reduced albedo on early Mars does not solve the climate environmental lessons for the Anthropocene from the paradox under a faint young Sun EDP Sciences 540, Classic Maya collapse Anthropocene 4 (2013) 88–100 A13, https://doi.org/10.1051/0004-6361/201118527 Korpela, E. J., Sallmen, S. M., & Leystra Greene, D. 2015, Frank, A., Alberti, M., & Keliedon, A.,. 2017, ”Earth as a ApJ Hybrid Planet, The Anthropocene in an Evolutionary Landis, G. A. 1998. The Fermi paradox: an approach based Astrobiological Context, Anthropocene, (in press). on percolation theory.. Journal of the British Frank, A., Carroll-Nellenback, J., Alberti, M., Kleidon, A. Interplanetary Society 51, 163-166. 2018. The Anthropocene Generalized: Evolution of Lenton, T, Held, H., Kriegler, E., Hall, H., Lucht, W., Exo-Civilizations and Their Planetary Feedback. Rahmstorf, S., Schellnhuber, H., 2008, PNAS, 105, Astrobiology 18, 503–518. doi:10.1089/ast.2017.1671 1786-1793, “Tipping elements in the Earth’s climate Gaidos, E., Knoll, A.H., 2012. Our evolving planet: from system” the dark ages to an evolutionary renaissance. In: Impey, Lin H. W., Gonzalez Abad G., Loeb A., 2014, ApJL, 792, C., Lunine, J., Funes, J. (Eds.), Frontiers in Astrobiology. L7 Cambridge University Press, Cambridge. Lingam, M., Loeb, A. 2017. Natural and artiﬁcial spectral Haqq-Misra, J. D., Baum, S. D. 2009. The Sustainability edges in exoplanets. Monthly Notices of the Royal Solution To The Fermi Paradox. Journal of the British Astronomical Society 470, L82-L86. Interplanetary Society 62, 47-51. Kuehn,C., Physica D, 2011, 240 1020–1035 “A Huang, Y., Bani Shahabadi, M. 2014. Why logarithmic? A mathematical framework for critical transitions: note on the dependence of radiative forcing on gas Bifurcations, fast–slow systems and stochastic dynamics” concentration J. Geophys. Res. Atmos., 119, 13,683– Lenton, T., 2011, Nature Climate Change 1, 201–209, 13,689, doi:10.1002/2014JD022466 “Early warning of climate tipping points” Climate Change 2014: Synthesis Report. Contribution of Lenton et al. 2015 Planetary boundaries: Guiding human Working Groups I, II and III to the Fifth Assessment development on a changing planet. Science, 347, 736 Report of the Intergovernmental Panel on Climate Lenton, T.M., Pichler, P., & Weisz H.,Earth Syst. Dynam., Change [Core Writing Team, R.K. Pachauri and L.A. 7, 353-370, 2016 Meyer (eds.)]. IPCC, Geneva, Switzerland, page 43 Lingam, M., Loeb, A. 2019. Relative Likelihood of Success Mullan, B., Haqq-Misra, J. 2019. Population growth, in the Search for Primitive versus Intelligent energy use, and the implications for the search for Extraterrestrial Life. Astrobiology 19, 28–39. extraterrestrial intelligence. Futures 106, 4–17. doi:10.1089/ast.2018.1936 doi:10.1016/j.futures.2018.06.009 D. H. Meadows and D. L Meadows and J. Randers and W. Hooke, R., Mart´ın-Duque,J., Pedraza, J. (2012) Land W. Behrens III, The limits to growth, 1972, Universe transformation by humans: A review GSA Today, 22(12), Books 5- 10 Howard, A.W. (2013) “Observed Properties of Extrasolar Planets”. Science 340, 572-576. Triggering A Climate Change Dominated ”Anthropocene”: Is It Common Among Exocivilizations?19

Matthews, B., A. Narwani, S. Hausch, E. Nonaka, H. Peter, Solomon, S., D. Qin, M. Manning, Z. Chen, M. Marquis, M. Yamamichi, K. Sullam, K. Bird, M. Thomas, T. K.B. Averyt, M. Tignor and H.L. Miller (eds.) Hanley, and C. Turner. (2011). Toward an integration of Contribution of Working Group I to the Fourth evolutionary biology and ecosystem science. Ecology Assessment Report of the Intergovernmental Panel on Letters 14(7): 690-701. 1 Climate Change, 2007, Cambridge University Press, Miller1 L., Gans, F., Kleidon, A., 2011, Earth Syst. Cambridge, United Kingdom and New York, NY, USA. Dynam., 2, 1–12, “Estimating maximum global land surface wind power extractability and associated climatic Vitousek, P. M. Ehrlich, P. R., Ehrlich, A. H., Matson, P. consequences” A., 1986. Human appropriation of the products of North, G & Kim, K., 2017, Energy Balance Climate photosynthesis. Bioscience 36: 368-373. Models, Wiley, New York. A. J. Watson and J. E. Lovelock, 1983, Biological Newman, W. I., Sagan, C. 1981. Galactic civilizations - homeostasis of the global environment: the parable of Population dynamics and interstellar diffusion. Icarus 46, Daisyworld, Tellus, 35B,284-289 293-327. Perretto, P., & Valente S., 2015, J. Econ Growth, Way, M. J., and 10 colleagues 2017. Resolving Orbital and 20,305-331 Climate Keys of Earth and Extraterrestrial Pimentel, D. (1961) Animal population regulation by the Environments with Dynamics 1.0: A General Circulation genetic feedback mechanism. Am. Nat. 95, 65–79 Model for Simulating the Climates of Rocky Planets. Ramirez, R. M. 2020. A Complex Life Habitable Zone ArXiv e-prints arXiv:1701.02360. Based On Lipid Solubility Theory. Scientific Reports 10. Williams, Darren M. and James F. Kasting (1997) doi:10.1038/s41598-020-64436-z “Habitable Planets with High Obliquities”. Reuveny, R., 2012, Annu. Rev. Resour. Econ. 2012. In:Icarus129.1, pp. 254 –267. ISSN: 0019-1035. 4:303–264 Rockstrom, J. et al. (2009) Planetary boundaries: exploring Williamson, M., Bathiany, S.,Lenton, T, 2016, Earth Syst. the safe operating space for humanity. Ecol. Soc. 14, 32 Dynam., 7, 313–326, 2016, “Early warning signals of Rull, V., 2016, Quaternary Science Reviews 150 (2016) tipping points in periodically forced systems” 31e41 Wright, J. T., Cartier, K. M. S., Zhao, M., Jontof-Hutter, Seager, S., 2013. ExoPlanet habitability. Science 340, D., Ford, E. B. 2016. The Search for Extraterrestrial 577–581. Civilizations with Large Energy Supplies. IV. The Schellnhuber, H.J., Wenzel, V, Earth System Analysis, Integrating Science for Sustainability, 1998, Editors Signatures and Information Content of Transiting Schellnhuber, H.J., Wenzel, V., Springer Megastructures. The Astrophysical Journal 816, 17. Solomon, S., Qin, D., Manning, M., Chen, Z., Marquis, M., Carter, B. (2008) Watson, A, 2008, Implications of an Averyt, K.B., Tignor, M., Miller, H.L. (Eds.), 2007. Anthropic Model of Evolution for Emergence of Complex Climate Change 2007: The Physical Science Basis. Life and Intelligence, ASTROBIOLOGY Volume 8, Cambridge University Press, Cambridge, United Number 1, 2008 Kingdom and New York, NY, USA. Wittmann, A. C., & Pörtner,H.-O. 2013, Sensitivities of Safuan, H.,Sidhu, H., Jovanoski, Z., Towers, I., 2014, ANZIAM J., 54, pp.C768–C787 extant animal taxa to ocean acidification, NatCC, 3, 995 Schwieterman, E. W., Reinhard, C. T., Olson, S. L., Zhang, X., Christian Kuehn, C., Sarah Hallerberg, S. 2015, Harman, C. E., Lyons, T. W. 2019. A Limited Habitable PHYSICAL REVIEW E 92, 052905, “Predictability of Zone for Complex Life. The Astrophysical Journal 878. critical transitions” doi:10.3847/1538-4357/ab1d52