The evolution of latitudinal temperature gradients from the latest Cretaceous through the Present

Abstract

Latitudinal temperature gradients are a defining characteristic of the climate system. Using thermometric indicators, including δ¹⁸O, plant and animal fossils, glycerol dialkyl glycerol tetraether (GDGT) proxies, and clumped isotope estimates, we document marine and terrestrial temperature gradients for the latest Cretaceous, Late Paleocene-Early Eocene, Early Oligocene, Pliocene, and Recent. The changes in gradients reflect the transition greenhouse to icehouse conditions. The evolution of latitudinal temperature gradients in marine and terrestrial realms are similar but has some distinctive differences. Marine temperatures are generally warmer than those on land. Except for the Late Paleocene-Early Eocene, the marine records show distinct inflection points at ~30° and ~50° latitude indicating the existence of frontal systems in the ocean. Except for the Late Paleocene-Early Eocene, the marine records show an increasingly steeper trend, from latest Cretaceous through Recent, being most pronounced after the Early Oligocene greenhouse-icehouse transition. This trend reflects the increasing intensity of high-latitude and polar cooling as the icehouse state developed. During the Late Paleocene-Early Eocene the oceans were characterized by slightly warmer tropics and much warmer higher latitudes than at present. The continents have generally had tropical temperatures like those of today, varying by <5 °C. Higher latitude temperatures cooled during the latest Cretaceous, became much warmer during the Late Paleocene-Early Eocene, then cooled during the Early Oligocene and have become increasingly colder since then. The results suggest that there is a climate thermostat mechanism, probably related to greenhouse gas concentrations, that ameliorates tropical warming by redistributing warmth to the poles in the greenhouse world. That mechanism broke down as greenhouse gas concentrations declined resulting in the conversion from greenhouse to icehouse conditions.

Introduction

Latitudinal temperature gradients reflect the uneven distribution of heat at the Earth's surface. They record the dynamics of heat transfer from low to high latitudes, and define the distribution of the Earth's ecosystems (Peixoto and Oort, 1992; Frakes et al., 1994; Amiot et al., 2004; Barry and Chorley, 2009; Upchurch et al., 2015; Hay, 2016; O'Brien et al., 2017).

Much of our synoptic knowledge of past climate comes from numerical models. The models are verified using model-data comparisons. There is a wealth of sea surface temperature data for the later part of Earth history, but this decreases backwards in time as more and more of the ancient sea floor has been lost to subduction. Half of the sea floor that existed at 60 Ma has been lost, and only 25% of what existed at 95 Ma has been preserved (Hay et al., 1988; Müller et al., 2016).

On the contrary the land area has not changed so greatly. At present the total land area is about 148.3 × 10⁶ km². At the time of maximum sea level heights about 90 my ago, the land area has been estimated to have been only about 130 × 10⁶ km² (Ronov, 1994). For the time slices discussed here, the land areas were about 132 × 10⁶ km² for the latest Cretaceous, and 133 × 10⁶ km² for the Early Eocene, Early Oligocene, and Pliocene. The Quaternary emergence is due largely to the buildup of the Antarctic ice sheet, which lowered sea level by about 62 m. Liebrand et al. (2017) present an informative account of early Antarctic ice sheet development, but the data they use are based mostly on benthic deep-sea foraminifera and are not comparable with our analysis of surface temperature data. However, over the past 100 my there has been a significant change in the hemispheric distribution of land. Today about 67.4% of the land area is in the Northern Hemisphere, 32.6% in the Southern Hemisphere (Baumgartner and Reichel, 1975). In the Cretaceous the distribution between the hemispheres was much more even.

One of the most important aspects of understanding paleoclimates is the nature of the equatorial to polar temperature gradient. The locations of averaged Anthropocene gradients for the 1951–1980 interval (used as a baseline for studies of global warming) for six latitudinal transects over marine and terrestrial realms are shown in Fig. 1A.

Fig. 1B shows the gradients along the three transects over land (avoiding uplands, high plateaus and mountain ranges) and their average. Note that the temperatures are almost flat between about 15° N and S. Maximum temperatures are south of the equator in the transect through the Americas and north of the equator in the Europe-Africa transect. The Australia-Eastern Asia transect shows an equatorial depression, which corresponds to islands with temperatures moderated by the surrounding ocean. North and south of 15°, latitudinal temperatures decrease steadily at a rate of about 0.7 °C per degree of latitude. At about 60° N the slope becomes gentler as it approaches the Arctic Ocean. This corresponds to the average location of the atmosphere's Polar Front. The average gradient over all land areas is symmetrical, with a slight equatorial depression due to the maxima for the Americas and Europe-Africa being south and north of the equator.

Fig. 1C shows the temperature gradients over the Pacific, Atlantic, and Indian Oceans. The ocean's Subtropical and Polar Fronts, which separate the stratified low latitude oceans from the more vertically mixed subpolar-polar oceans are only a few hundred kilometers apart. The area between them is the outcrop on the ocean surface of the thermocline.

The gradient over the Pacific is almost symmetrical between the two hemispheres. It reaches maxima (28 °C) at about 10° N and S, and there is a slight depression by about 2 °C on the equator caused by upwelling of cooler waters. The depth of this depression depends on whether the Pacific is in an El Niño (restricted equatorial upwelling) or La Niña (extensive equatorial upwelling) condition.

In contrast, the Atlantic Ocean is asymmetrical. The South Atlantic is relatively cool and the North Atlantic warm. This is because of the existence of the global ocean thermohaline circulation system, with sinking in the Norwegian-Greenland-Iceland Sea pulls warmer waters across the equator.

The gradient in the Indian Ocean is similar to that of the South Pacific, but the warm equatorial waters extend northward to the Asian coast.

The average global latitudinal temperature gradient is shown in Fig. 1D. The flat region between 15° S and N is slightly asymmetrical, with the northern hemisphere slightly warmer. It is due to the greater areas of land in the northern hemisphere, although the Earth is presently closest to the Sun during Southern Hemisphere summer. Temperatures of the Polar Oceans are generally flat, about −2 °C whether exposed or covered by ice. Temperatures over the high Antarctic sheet are extremely cold, ranging down to −90 °C at the pole.

Over the past 540 million years, Earth's climate has oscillated between a globally warm ‘greenhouse state’ with no or minimal polar ice and an ‘icehouse state’ with ice which may extend to mid-latitudes (Frakes et al., 2005; Kidder and Worsley, 2010; Hay, 2011; Kidder and Worsley, 2012). The most recent transition between these climatic states began in the later Cretaceous. The equable mid-Cretaceous greenhouse climate, high CO₂ content, high temperature from low to high latitudes, and low equator-to-pole gradients with no polar ice (Hay et al., 2018), gradually drifted toward the fundamentally different icehouse condition during the Neogene (Zachos et al., 2001; Beerling and Royer, 2011; Pagani et al., 2011; Grossman, 2012b; Linnert et al., 2014; Anagnostou et al., 2016). Although there is evidence for sporadic ice sheet development during the Maastrichtian, no permanent ice sheets existed (Miller et al., 1999; Hunter et al., 2008; Thibault et al., 2010). During the Paleogene and Neogene, the Earth's climate has undergone a significant and complex evolution, with episodes of global warming and cooling, and expansion of areas covered by ice ultimately resulting in permanent Antarctic ice sheets and periodic Northern hemisphere ice sheet growth and decay (Zachos et al., 2001). Earth's ecosystems have also undergone a significant evolution, with a decline in diversity of a Mesozoic fauna culminating with many extinctions at the K-Pg boundary event. This was followed by worldwide diversification and development of a new Cenozoic fauna (Jablonski and Chaloner, 1994; Feduccia, 1995; MacLeod et al., 1997; Alroy, 1999; Schulte et al., 2010; Norris et al., 2013).

As a paradigm of greenhouse-icehouse state transition in Earth's history, the evolution of the latitudinal temperature gradients from the latest Cretaceous through the Present is of great interest. It is important to understand how this greenhouse-icehouse transition evolved, and what the main trigger mechanisms have been. These may serve as analogs for a near-future greenhouse world (Hay, 2011; Wang et al., 2014; C. Wang et al., 2017). Temperature records of this transition have been developed using microfaunal assemblages, δ¹⁸O, and glycerol dialkyl glycerol tetraether (GDGT) proxy data from marine sediments recovered by the Deep Sea Drilling Project (DSDP), the Ocean Drilling Program (ODP), and other sources (Zachos et al., 2001; Friedrich et al., 2012; Grossman, 2012b; Norris et al., 2013; Zhang et al., 2013; Tierney et al., 2017). However, knowledge of the terrestrial temperature history is substantially poorer (Wilf et al., 2003; Poole et al., 2005; Snell et al., 2013; Kemp et al., 2014; Tobin et al., 2014). Compared to the ocean system, stabilized by the high heat capacity of water, the continental environment is more sensitive to global paleoclimate change. Its record provides the possibility of resolving even relatively small variations and may, in some cases, yield information on seasonality (Poole et al., 2005; Snell et al., 2013; Wang et al., 2013b; Herman et al., 2016).

In the past, terrestrial temperature estimates were hampered by a lack of unambiguous proxies (Wang et al., 2013a; Wang et al., 2013b; Tobin et al., 2014). Terrestrial latitudinal temperature gradients were mainly reconstructed on the basis of the occurrence and distribution of fossil vertebrate and plant assemblages (Wolfe and Upchurch Jr, 1987; Frakes et al., 1994; Markwick, 1994). Paleotemperature analysis from plant fossils based on CLAMP (Climate-Leaf Analysis Multivariate Program) methodology have generally been considered to be the most reliable (Spicer et al., 2004). However, there have been some significant problems, such as the ‘cold continental interior paradox’ of the mid-Cretaceous (DeConto et al., 1999). Furthermore, comparison of methods for determining past atmospheric CO₂ levels (Royer et al., 2001) leaves open the question of which method is most reliable not only for CO₂ but also paleotemperatures.

The problems lie in the fact that we are not certain whether animals and plants in the past had the same temperature and precipitation requirements as those living today. Very major changes have taken place, such as the development and spread of angiosperms that replaced gymnosperms, cycads and other plants during the Cretaceous, and the spread of C₄ pants during the Pliocene. We are not sure whether dinosaurs were warm or cold blooded or when hibernation as an overwintering technique for mammals developed. We also do not know when the habit of trees dropping their leaves, originally thought to be an accommodation to polar night, became an adoption to cold winter temperatures.

Furthermore, there are problems with the fossils themselves. Preservation of plant and animal remains as fossils requires special conditions. Were these very special, or were they typical for the time? Many paleoclimate interpretations depend on fossil material from a single or few closely located fossil localities (Spicer et al., 2008; Tarduno et al., 1998).

Recently estimates of paleotemperatures have been revolutionized by the development of several new proxies such as the clumped isotope paleothermometer and lipid-based paleothermometers (Eiler, 2011; Kemp et al., 2014; Herman et al., 2016; Zhang et al., 2016a; Zhang et al., 2018). Here we compile data on global marine/terrestrial temperatures and establish latitudinal temperature gradients patterns during five periods: the latest Cretaceous, Late Paleocene-Early Eocene, Early Oligocene, Pliocene, and Recent. We evaluate the marine/terrestrial temperature changes during the greenhouse-icehouse transition on both temporal and spatial scales. We discuss two major issues: (1) how the latitudinal temperature gradients evolved across the Mesozoic-Cenozoic greenhouse-icehouse transition; and (2) the difference between the terrestrial and marine gradients during the transition.