A number of salient observations can be readily discerned from these results. First, of our three hypotheses regarding the vegetation surrounding the Palace and Temple reservoirs in ancient times, the third hypothesis, that the ancient Maya maintained native forest plants on the banks of the reservoirs, is strongly supported. If we look more closely at the eDNA results, however, there are more nuanced indications from the data recovered. To begin, there is no clear evidence for the presence of cultigens. From the Postclassic sample (WA08), there is a genetic sequence from the Cucurbitaceae, which includes the squashes and pumpkins, but there also are many wild species in the family that are native to the Neotropics, so results in this case were inconclusive. The same is true with the Poaceae sequences; maize is not clearly indicated in any of the reservoir samples and the only two grasses that were positively identified were Stipa sp. and Imperata sp., both wild genera native to the region. In summary, these observations support the validity of hypothesis three, i.e., the Maya did not plant cultigens around these two reservoirs, but allowed or encouraged native forest species to remain on the banks.
These findings suggest that in some cases trees and plant vegetation were near the banks of these reservoirs for other reasons, too. Given the central location of the two reservoirs at issue, they may have been used as recreational settings or aesthetic zones within the proximity of numerous towering architectural structures. The functionality of the reservoirs is clear—they were sources of potable water—but trees for shading, understory cooling and aesthetic and symbolic effects may have been an important concern, as well (25).
The earliest sample (WA07 from the Early Preclassic period) and most recent sample (WA08 from the Postclassic period) in a way represent bookends of the Maya occupation, with the earlier sample pre-dating the Maya occupation of the site and the latter sample representing a time period following the 9th century droughts and the abandonment of the site core of the city. They reflect similar environments with tall forest dominants, understory trees, vines and shrubs suggestive of well-structured, multi-tiered, semi-deciduous tropical forests, typical of the northern Petén region (34).
The two middle samples, however, present slightly different pictures. The Late Classic period sample (WA09) has some forest dominants, Brosimum alicastrum and Lonchocarpus sp. and some understory trees, but there are also some disturbance species appearing, such as Bidens alba, which is a common weed (29) and Trema sp. which is a genus of small trees generally found in second growth (29). The presence of these invaders suggest habitat disturbance and that the forest canopy around the reservoir may have been opening up.
A question arises in regard to these results from the Temple Reservoir. On close inspection of the lidar image (Fig. 1), there is really no appreciable space around the reservoir where an embankment may have been positioned and it is, for the most part, surrounded by pavement. There is, however, a large zone to the east and the southeast of Temple Reservoir, a low ridge where there appears to be an undisturbed land surface adjacent to what has been referred to as a “silting tank” (2) where the freshwater spring was found. Another, more cosmological interpretation (35), suggests that this area may have been a sacred well for the city, or ch’en, and the undisturbed land adjacent to it may have supported an ancestral grove of trees (cf. 24) associated with the sacred well.
During the Terminal Classic period, the results of sample WA01 offer no clear evidence for the presence of dominant forest tree species in the vicinity of the Palace Reservoir. There is the Meliaceae sequence, but it is unresolved whether this represents a large tree, e.g., Cedrela odorata, or one of the common understory species, such as Trichilia spp. Alternatively, the medium-sized trees clearly identified in this sample, Ficus tonduzii, Trophis racemosa, Tabernaemontana sp., Morus celtidifolia and the small palm, Cryosophila stauracantha, would have made an attractive and effective lining along the littoral zone of the south bank of the Palace Reservoir. Finally, Imperata sp., a perennial grass of wastelands and forest transition areas (36) makes an appearance during this time period. These plants probably took hold around the reservoir edges where there was more light available. The droughts of the mid-9th century may have opened up more habitat for this aggressive grass. Quite possibly, as the water levels in the reservoirs receded due to the droughts, the Imperata grass could have followed the waterline downward as the native trees, vines and shrubs suffered. Looking at the eDNA results as a whole, it appears evident that the Maya favored the maintenance of natural vegetation on the banks surrounding the reservoirs and their erosion control strategy worked well for many centuries.
The second noteworthy aspect of this research relates to the Uaxactun garden experiment. As mentioned above, there was a substantial overlap of 25 plants between those plants physically observed in the garden (total 49) and those detected by the genetic probe (total 93). Although the overlap was not complete, the 25 plants that were recorded by both methods were the trees, shrubs, herbs and crops of most economic value (32). Brosimum alicastrum, commonly referred to as ramón, detected by the DNA probe but not observed in the garden, is one of the most common forest dominants in the region (37,38). The leaves and fruits of ramón were likely brought into the compound. The fruits of ramón are edible and the leaves are commonly used for livestock fodder (31). Accordingly, the DNA probe detected not only the plants growing in the garden at the time, but also plants brought in from the surrounding forests and likely from their neighbor’s gardens, as well. The results from the Uaxactun home garden analysis demonstrated two important features of our methodology: 1) that our probe could readily detect the presence of domesticated plants and 2) the probe detected the presence of sufficient plant data to effectively characterize the environment from where the sample was collected.
The third important aspect of the information gained from the eDNA analysis was that the procedures followed in this study are an important complement to other paleoethnobotanical techniques such as pollen, macro-remain, phytolith and starch grain analysis. When we compare the results from the traditional paleoethnobotanical methods to those of the eDNA analysis (Fig. 3, Tables 1 and S1), we see a distinct overlap, yet both approaches produced some unique species undetected by the other. Note that our efforts to extract pollen from the Temple and Palace reservoirs were unsuccessful and this underscores the value of the eDNA data. Thus, it appears that eDNA analysis can add significantly to the paleoethnobotanical toolkit and our findings will be much richer and more detailed if, in the future, all of the techniques mentioned above are used in combination.
Finally, perhaps the most important result from this study was that we were able to extract DNA from ancient reservoir sediments, sequence the extracts and identify species-specific sequences. This achievement represents a huge technological leap forward for the field of archaeology. Some scholars have stated that the eDNA approach has enormous potential to efficiently detect human environmental impacts through time (39) while others have predicted that the technique will “revolutionize” the field of archaeology (40). This study, although limited in scope, places those predictions on a pathway to fruition.