Climatic variability and human impact during the last 2000 years in western Mesoamerica : evidences of late Classic and Little Ice Age drought events

Introduction Conclusions References


Introduction
The knowledge of past climate and environmental events is critical for understanding a wide variety of modern environmental processes in order to better estimate causes and effects of future environmental changes.Records from many parts of the world have led to the recognition of climate variability in the late Holocene.However, its duration and the possible mechanisms of climate change involved at each location are not clear, especially at tropical latitudes where there are fewer paleoclimatic continental records than at higher latitudes (Maasch et al., 2005;Mann et al., 1999).
In Mesoamerica, the last two thousand years appear to have been a time of significant climatic change with a complex signal, mostly linked to moisture availability.At present, the influence of the Intertropical Convergence Zonce (ITCZ) and Subtropical High-Pressure Cells (SHPCs) varies seasonally, showing latitudinal shifts associated with the tropospheric equator-pole temperature gradient in each hemisphere, and the thermal contrast between both hemispheres (Haugh et al., 2001;Nyberg et al., 2002).The seasonal shift in insolation and latitudinal location of the ITCZ and SHPCs brings shifts in moisture availability that affect most of Mesoamerica, with an intense summer rainfall season associated with the northerly location of the ITCZ and the onset of the North American Monsoon (NAM).Past changes in the amplitude, intensity and location of the ITCZ and SHPCs and in the intensity of the NAM are therefore likely explanations of the changes documented in the Mesoamerican geological record (Hodell et al., 2005a;Metcalfe et al., 2010;Haug et al., 2003;Barron et al., 2012).Long term changes in these climatic systems can be recorded in the sedimentary archives of the lakes in the region given that lakes are natural systems that are sensitive to climatic variations which can lead to changes in their stratification patterns, their water chemistry and/or their biological associations.Specifically, western Mexico is an area which precipitation is modulated by the position of the ITCZ and the intensity of the NAM, however, most of the available paleolimnological records for Mesoamerica are concentrated in the highlands of central Mexico or in the Yucatan region (Metcalfe et al., 2000).In this paper we present a multiproxy, high resolution (∼ 20 yr per sample) study of a laminated sedimentary sequence from a closed lake in western Mexico (Santa María del Oro, Nayarit, Fig. 1) which aims to reconstruct the past climatic events that affected western Mesoamerica during the past 2000 years.This lake currently sits at the intersection between the arid climates of northern Mexico, mostly under the influence of the north Pacific SHPC, and the temperate, sub-humid to humid climates of central Mexico (García-Oliva et al., 2002), under the influence of the ITCZ and the NAM systems.This record gives clear evidence of two intense drought periods, the late Classic (AD 500 to 1000) and the onset of the Little Ice Age (AD 1400 to 1550) and a third, less intense, around AD 1690 to 1770.

Site description
Santa Maria del Oro (SMO, 21 • 23 N, 104 • 35 W, 750 m a.s.l.) is a topographically closed freshwater crater lake at the western end of the Transmexican Volcanic Belt and only 65 km from the Pacific coast (Fig. 1).The catchment rocks are dominated by dacite-rhyolite and basaltic-andesite.Minor exposures of basaltic lava flows are present on the northern wall of the crater.There are no carbonate rock outcrops in the catchment or in the surrounding areas (Ferrari et al., 2003;Vázquez-Castro et al., 2008).
The climate at SMO is warm and sub-humid, with monthly average temperatures that range from 16.5 to 25 • C (Station 18005-Cerro Blanco -Servicio Meteorológico Nacional).This region receives an average annual precipitation of ∼ 1200 mm mainly during the warm, summer months (June to September) when the ITCZ is at its most northerly location and the NAM is active (Liebmann et al., 2008;Berbery, 2001).During summer tropical storms and hurricanes can also bring moisture from the Pacific (Castro, 2010;García-Oliva et al., 2002).The rest of the year (October to May) is relatively dry, particularly during spring.
SMO is a nearly round, ∼ 2 km diameter lake.It has a maximum depth of 65.5 m and a mean depth of 46 m (Serrano et al., 2002).The crater has steep slopes but there is a small flat and shallow bay (12 m deep) where the core for this study was retrieved (Fig. 1).The lake receives inflows directly from precipitation, seasonal runoff and ground water, and loses water to evaporation and seepage.The lowest point of the crater is on the north-eastern side, where the natural sill has been anthropogenically modified and today it lies ∼ 5 m above the lake level (755 m a.s.l.).The closed nature of the basin and relatively unpolluted conditions make of this lake an ideal paleoenvironmental research site, where lake chemistry depends mostly on the precipitation -evaporation balance.At present SMO is a freshwater, [HCO , oligotrophic, monomictic lake, with a winter mixing period (Caballero et al., 2013); the oxycline lies between 13 to 20 m depth the rest of the year.Currently the lake ostracode fauna is dominated by Potamocypris variegata Brady and Norman, Cypridopsis vidua Müller and Darwinula stevensoni Brady and Robertson and the planktonic diatom flora by Aulacoseira granulata (Ehrenberg) Simonsen in association with Nitzschia amphibia Grunow.

Methods
Four parallel cores between 4 and 9 m long were recovered from the shallow southwestern bay (12 m depth) of SMO in 2002 (Fig. 1c) with a Usinger coring system (Mingram et al., 2007).All cores were split, described, photographed and sampled at the Instituto de Geofísica, Universidad Nacional Autónoma de México (UNAM) and except otherwise specified all analyses were performed there.The longest core (SMO02-V, 8.8 m) was sampled on average at 5 cm intervals for magnetic susceptibility (MS), biological (ostracodes, diatoms, pollen) and geochemical analyses (total inorganic carbon, total organic carbon, X-Ray Fluorescence elemental analysis).Seven samples for international system units (SI) and TIC expressed as its CaCO 3 % equivalent, together with new diatom, ostracode, Ca and Ti analysis from the same (SMO02-V) core.

X-Ray Fluorecence
Thirty nine freeze-dried sediment samples were chosen, on average every 20 cm, for X-Ray Fluorescence (XRF) elemental analysis.Samples were homogenized and ground to 200 mesh using an agate mortar.Elemental concentrations were measured in a Siemens SRS 3000 wavelength dispersive XRF spectrometer with a precision of 5 % at the Institute of Geology, UNAM.Titnanium and calcium concentrations are expressed as percentages, Ca was normalized against titanium and expressed as Ca/Ti ratio.

Diatoms
For diatom analysis 108 samples were chosen, on average at every 10 cm interval, of which 78 contained diatoms.Dry sediment (0.5 g) was successively cleaned with HCl (10 %), H 2 O 2 and HNO 3 .Permanent slides were made with 200 µL of clean material, with Naphrax ® as mounting medium.Minimum counts of 300 valves were undertaken in most samples, but 22 had low diatom abundance and only 100 valves were counted.Diatom counts were done with an Olympus BX50 microscope with Interdifferential Phase Contrast at 1000 × .Species composition is reported as relative abundance (%), and total abundance is reported as valves per gram of dry sediment (v gds −1 ).

Ostracodes
For ostracodes the same 108 samples were analyzed, but only 57 levels contained ostracodes.Two cubic centimeters from each sample were freeze dried, from which 1 g of dry sediment was disaggregated by standard methods (Delorme, 1990;Forester, 1988) and slowly wet-sieved (63 µm).Ostracode valves were picked under an Olympus SZX12 stereo microscope, counted, determined to species taxonomic level and stored in micropalaeontological slides.Care was taken to pick whole adult valves to allow species identification, juvenile valves were counted separately.Ostracode counts are expressed as valves per gram of dry sediment (v gds −1 ).

Core description
The SMO02-V sequence consists of sand/silt laminations with intercalated woody peat layers.Individual silt layers can be of ∼ 1-3 to 60 mm in thickness and either light-brown calcareous, greenish, reddish or dark-brown (Vázquez- Castro et al., 2008).A tephra layer at 4.4 m depth was identified as the Toba Jala (AD ∼ 860), from nearby Ceboruco volcano.The light-brown calcareous silt layers contain either authigenic carbonates or ostracodes, and are more frequent towards the top of the core, while woody peat layers are more frequent towards the bottom.There is no evidence for hiatuses, erosion features or slumping structures in the cores.

Chronology
The seven dates (Table 1) were calibrated to calendar years with Calib 5.0 (Stuvier et al., 2005) and the IntCal04 data set (Reimer et al., 2004).The age model (Fig. 2) assumed uniform sediment accumulation rates between the dates and anchored the top sediment to the year of coring ( 2002) (Vázquez-Castro et al., 2008).Sedimentation rates vary between 0.21 and 1.15 cm yr

Non biological proxies
Magnetic susceptibility (MS) (Fig. 3) fluctuates along the core between 24 and 520 × 10 −6 SI, with the lowest values (< 200 × 10 −6 SI) observed between 530 and 490 cm, and between 200 and 150 cm.The highest MS values (> 300 × 10 −6 SI) are mainly present from 600 to 530 cm, from 490 to 420 cm.Titanium content ranges between 0.3 and 1.5 % with lower than average (< 0.8 %) values concentrated between 530 and 370 and 200 to 150 cm.The highest values (> 1.3 %) in the record are present between 450 and 420 cm.TIC as its CaCO 3 equivalent varies along the core from 0 to 59 % (Fig. 3), in a nearly opposite way as magnetic susceptibility, showing in general terms higher values (> 20 %) between 530 and 370, 320 and 240, 200 to 150 and 120 to 100 cm.Ca/Ti ratio ranges between 1.4 and 175 and shows above average values (> 15) between 530 and 370 cm and at 200 cm.

Diatoms
Diatom abundance shows several intervals of lower than average abundance values (< 40 × 10 6 v gds −1 , Fig. 4), but the lowest abundances are recorded between 740 to 690 cm and from 380 to 330 cm.Sixty four species are recorded in the sequence, 9 with relative abundances higher than 20 % in more than one sample (Fig. 3).Aulacoseira granulata, Nitzschia amphibia and small species of Fragilaria sensu lato (Staurosira construens, Staurosirella pinnata, Pseudostaurosira parasitica, P. brevistriata) are the dominant species along the record, Achnanthidium minutissimum (Kützing) Czarnecki has a preferential distribution between 750 and 650 cm and from 400 to 250 cm, Amphora lybica Ehrenberg and Hippodonta luneburgensis (Grunow) Lange-Bertalot, Metzeltin and Witkowski have higher abundances on the top 500 cm of the core and Eolimna minima (Grunow) Lange-Bertalot and W. Schiller shows a distinctive distribution with maxima between 530 and 420 cm and two less important peaks between 200 and 110 and 75 to 60 cm.), the first between 530 and 420 cm and the second between 200 and 110 cm.A smaller peak at 75 to 60 cm is also present and the rest of the core has very low (< 100 adult v gds −1 ) ostracode concentration (Fig. 3).Six species are recorded in the sequence; Candona patzcuaro Tressler is the most abundant and constant along the core in association with Potamocypris variegata Brady and Norman and minor numbers of Chlamydotheca arcuata Sars.Cypridopsis vidua Müller and Darwinula stevensoni Brady and Robertson are also present along the core, but show a preferential distribution between the 530 and 420 cm interval.Limnocythere verrucosa Hoff is only present as juvenile valves, and shows maximum concentrations between 570 and 530 cm.

Proxy interpretation
MS generally indicates the abundance of magnetic minerals that reach a lake mainly by surface runoff from the catchment area or, in occasions, from volcanic activity (Evans and Heller, 2003).Titanium is an element which originates from minerals in the catchment rocks and its abundance in lacustrine sediments is also indicative of surface runoff to the lake (Metcalfe et al., 2010;Sosa-Nájera et al., 2010).Both proxies can therefore be used to identify times of higher or lower precipitation and surface transport of sediments to the lake.On the other hand, given that in this lake there are no carbonate rocks in its catchment, high values of TIC are indicative of times of higher authigenic carbonate precipitation.This is a process that currently occurs in SMO during the warmer part of the year (Caballero et al., 2013) and that is in general favored by high evaporation rates leading to high electric conductivity, alkalinity and pH (Eugster and Hardie, 1978;Gierlowski-Kordesch, 2010).Calcium is an element which can originate from the minerals in the catchment but is also incorporated in authigenic carbonates.Sosa-Nájera  2010) demonstrated that in this lake higher Ca concentrations correlate with historical droughts during the last 700 years, supporting the interpretation that in SMO high calcium values are mostly related with carbonate deposition during times of higher evaporation.Nevertheless, Ca was normalized against Ti to down weight the detrital signal in this element.High TIC and high Ca/Ti values in this record can therefore be taken as indicators of increased carbonate deposition during times of lower precipitation and higher evaporation rates leading to relatively high lake water electric conductivity, pH and alkalinity.Given the opposite interpretation of the MS and Ti records compared to TIC and Ca, it is not surprising that they show nearly opposite patterns of downcore variations.Together these records allow to identify two intervals of intense evaporation and reduced surface runoff to the lake, the first from 530 to 490 cm (AD 600 to 800) and the second from 200 to 150 cm (AD 1400 to 1450).With respect to the biological proxies, in the diatom record three ecological groups can be identified: (1) a planktonic/tychoplanktonic association dominated by Aulacoseira granulata, Nitzschia amphibia and small Fragilaria spp., which is abundant along most of the sequence (Fig. 3); this association includes the two main diatom species that are present in the modern lake plankton (Caballero et al., 2013).(2) A benthic/periphytic association where Amphora lybica, Hippodonta luneburgensis and Eolimna minima are the main species.These periphytic genera (Amphora) and species (H.luneburgensis, E. minima) are currently distributed in the shallower environments of the lake (Caballero et al., 2013).E. minima in particular is an aerophilic species (Wolfe and Härtling, 1996;Van Dam et al., 1994) that shows a distinctive distribution along the record (shaded areas in Fig. 3) that suggests intervals of particularly shallow lake levels.(3) An eutrophication/human impact association defined by the presence of Achnanthidium minutissimun, which is a species that has been considered by some authors as a colonizer in disturbed environments (Peterson and Stevenson, 1992;Hodgson et al., 1997) and that in Mexico has been recorded in the plankton of eutrophic and hypereutrophic tropical lakes which basins have been importantly altered by human activities (Caballero et al., 2006).This interpretation is further supported by the presence of Zea mays pollen grains in the same segments of the core (S.Sosa-Nájera, personal communication, 2013).
The ostracode record can be interpreted under the light of the study undertaken in this lake by Caballero et al. (2013) where ostracodes were found to be most abundant in shallow water conditions with littoral vegetation and a low sediment supply.
Given that the dominant species in the SMO record are the same that currently live in this lake (P. variegata, D. stevensoni, C. vidua) high ostracode abundances are taken to be indicative of these conditions, which are consistent with a lower surface runoff.Following the ostracode abundance record and considering the distribution of E. minima, three zones of lower than average lake levels can be identified from the biological record (Fig. 3): 560 to 410 cm (AD 500 to 1000), 200 to 100 cm (AD 1400 to 1550) and 75 to 56 cm (AD 1690 to 1770).The first two of these periods include the two events of intense evaporation identified in the geochemical record.

Paleoenvironmental history of the SMO record
Together the MS, Ti, Ca/Ti, ostracodes and diatom records (Fig. 3) allow to reconstruct the environmental history of lake SMO during the last 2000 years.Low values in Ca/Ti, TIC and ostracod abundance together with the absence of benthic/periphytic diatoms in the record prior to AD 500 indicate that the conditions in SMO were generally deeper and more diluted from 100 BC to AD 500.An early phase of human impact is identified by the abundance of A. minutissim from 100 BC to AD 300 possibly related with the local expression of the Shaft and Chamber Tom Tradition (200 BC to AD 600) characteristic of western Mexico (Beekman, 2010;Barrera-Rodríguez, 2006).
After AD 500 the lake experienced a first low level stage defined mainly by an increase in E. minima and in ostracode concentrations and also by its magnetic mineralogy (Vázquez-Castro et al., 2008).This low lake level period extended from AD 500 to 1000, with the interval from AD 600 to 800 representing the driest conditions (lowest MS and Ti values and highest TIC and Ca/Ti ratios).This first low lake level phase in  (Manzanilla, 2011) and by a high development of the Maya cities in the Yucatan region, that ends abruptly between AD 850 and 950.In western Mexico in general and specifically in the state of Nayarit (Barrera-Rodríguez, 2006), this is a time of cultural transition as the Shaft and Chamber Tomb tradition ended by AD 600 (Barrera-Rodríguez, 2006).Not surprisingly in the SMO record, there is no evidence of human impact during this dry interval.This guaranties that the trend to dry climates during this period is a true climatic signal, not obscured by anthropogenic activities.
This late Classic drought has been consistently identified in several records from Mexico, particularly from Yucatan (Hodell et al., 2005a;Metcalfe et al., 2010;Caballero et al., 2002;Lozano-García et al., 2010;Curtis et al., 1996) (Fig. 2), to the extent that it is referred to as the "Great Maya Drought" (Gill, 2001) and has been put forward as an important factor in the collapse of the May culture at the end of the Classic.The exact timing of this late Classic drought varies slightly in each region but in general it is centered in a similar time band as in SMO (AD 500 to 1000).In some records it shows a recurrent or cyclic nature (Haug et al., 2003;Hodell et al., 2001;Curtis et al., 1996) which in the SMO sequence could explain the erratic values of the geochemical records (alternating high and low TIC, Ti and MS) from AD 800 to 1000 and even until AD 1200.
After AD 1000 the low abundances of E. minima and ostracode valves suggest that the lake level in SMO recovered, however TIC, Ca/Ti and Ti still fluctuate between high and low values until AD ∼ 1200.After this date, the age model (Fig. 2) also suggest that sedimentation rates increased (from ≤ 0.3 to > 1 cm yr −1 ), in accordance with a higher surface runoff to the lake.This moister conditions after AD 1000 and especially from AD 1200 to 1400 correlate with moist conditions inferred from the Ti record from lake Juanacatlán, also in western Mexico (Metcalfe et al., 2010).Between 1000 and AD 1400, however, the presence of A. minutissimum points to a time of human disturbance in the system leading to higher turbidity and nutrient level.This trend in  (Gálvez-Rosales, 2006) and is also in agreement with the late Postclassic cultural peak reached in western Mesoamerica by groups like the Purepecha or Tarascans (Pollard, 2005).
From AD 1400 to 1550 the increase in TIC, E. minima and ostracode abundance as well as the magnetic mineralogy (Vázquez-Castro et al., 2008) suggest that the lake level in SMO dropped again.The lowest MS and Ti values and highest TIC suggest that the reduction in moisture was most intense from AD 1400 to 1450.This second dry interval in the SMO record corresponds with the onset of the Little Ice Age.This period has also been recorded as a time of reduced moisture and lower lake levels in other Mesoamerican records (Metcalfe et al., 2010;Hodell et al., 2005b;Cuna et al., 2014) and represents the climatic scenario for the Spanish conquest of Mesoamerica in 1521.Lake levels recovered in SMO between AD 1550 and 1690, but a third, if however less intense period of reduced moisture availability is evident from a slight increase in E. minima and ostracode concentrations between AD 1690 and 1770.This lighter reduction in moisture is coeval with the end of the Little Ice Age and shows a dual nature of the Little Ice Age cooling, also evident in the record from Lago Verde, eastern Mexico (Lozano-García et al., 2007).

Paleoclimatic implications
The data from SMO allows to identify two periods of human impact which follow closely the findings from recent archaeological research in the area (Barrera-Rodríguez, 2006;Gálvez-Rosales, 2006) the first related with the Shaft and Chamber Tomb tradition (100 BC to AD 300) and the second during the late Postclassic (AD 1100 to 1400).It is interesting to note that these periods correspond with times of moister conditions while there are no evidences of human impact in this lake during the dryer late Classic or in recent times.The biological and non-biological proxies from SMO give solid evidence of three dryer than average intervals, the most intense during the late Classic (AD 500-1000) and the other two related with the onset (1400-1550) and the late (1690-1770) Little Ice Age.There is also evidence of a wetter late Postclassic period (AD 1200 to 1400).In this region, in which climate is controlled by the seasonal shift of the north Pacific SHPC, the ITCZ and the onset of the NAM, dryer than average conditions can be related with an expansion of the north Pacific SHPC, a southward migration of the ITCZ and a less active NAM.These are mechanisms that have also been used by other authors to explain drought conditions over Mesoamerica (Haug et al., 2003;Metcalfe et al., 2010).Other mechanisms that could bring wetter/dryer conditions in this region are changes in the frequency or strength of "La Niña"/"El Niño", as today "La Niña" is related with wetter than average summers (Magaña et al., 2003;Metcalfe et al., 2010).Changes in the paths or frequency of Pacific tropical storms and hurricanes could also be elements controlling the moisture balance in this area.
The data presented in this paper give further evidence that the late Classic drought (AD 500 to 1000) was the most important one in Mesoamerica during the last 2000 years, affecting the whole region, from Yucatan to the Pacific coast and not only a circum-Caribbean feature.In SMO the most intense phase is recorded from AD 600 to 800.The stochastic nature of the SMO record from AD 800 to 1200 seems consistent with the proposed recurrent nature of droughts, which has been suggested to follow cycles in solar activity (mainly 50 and 200 yr), (Hodell et al., 2001;Haug et al., 2003;Hodell et al., 2005a).This time of strong fluctuations in the SMO record also correlates with the onset of warmer than average conditions over the Northern Hemisphere from AD 830 to 1100 during the Medieval Climatic Anomaly (PAGES-2k-Consortium, 2013) that locally correlated with warmer sea surface temperatures in the Gulf of California (Barron et al., 2002).the Maunder solar minimum.This dual nature of the Little Ice Age cooling, following the two solar minima, is also present in the record from Lago Verde, eastern Mexico (Lozano-García et al., 2007).In lakes SMO and Juanacatlán (both in western Mexico, Fig. 1) the Spörer minimum shows the most intense signal, however in other records in central or eastern Mexico the Maunder minimum represents the driest and/or coldest period (Haug et al., 2003;Cuna et al., 2014).Furthermore, in others records the two periods seem to merge into a single, longer Little Ice Age drought (Hodell et al., 2005b).These Little Ice Age droughts are undoubtedly related with the Spörer and Maunder solar minima which could be the ultimate control on the intensity of the NAM system.

1891
Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper |

1893
Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper |

− 1 (
Fig. 2).According to this model the base of the core (882 cm) is ca.2600 years -old, with an average resolution of 23 years per sample.Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper |

1895
Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | 4.5 Ostracodes Ostracode total abundance shows two distinct intervals with high values (> 1000 adult v gds −1

Discussion
Paper | Discussion Paper | Discussion Paper | Discussion Paper | et al. (

1897
Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | SMO correlates with what is known as the late Classic period in the development of Mesoamerican cultures, characterized by a decline of the Teotihuacan culture in central Mexico after AD 650

1899
Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | human occupation in SMO agrees with recent archaeological findings in Nayarit that show late Postlassic (AD 1100-1350) occupations Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | The dry interval from 1400 to 1550 which corresponds with the onset of the Little Ice Age, correlates closely with the Spörer minimum in solar activity.The less intense dry period towards the end of the Little Ice Age (AD 1690 to 1770) correlates with 1901 Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper |