Ron Canter continues his study of Maya watersheds with an important article on the Grijalva River, which begins near the Guatemala border, winds through the States of Chiapas, and Tabasco, to join with the Usumacinta and empty into the Gulf of Mexico. It has long since been exploited for hydroelectric power, drowning innumerable and now unknowable ancient settlements. It is a hint of what could happen if the Usumacinta dams are ever built.

Ron's full article can be read by clicking "More". The photo above shows the Grijalva, at the site of San Isidro, in 1981 as the river is rising due to new dams.

The Grijalva River [Mezcalapa, Kandelumihi,Tabasco, Rio Grande de Chiapas]

The head of navigation on the Rio Grande de Chiapas [Rio Grijalva]
was at the junction of the Rios San Miguel and San Gregorio (Navarrete,
1978). Before the dams, the Rio Grijalva had long navigable segments
connected by portage trails past canyon rapids. For mountain trails, the
portages were not so bad – moderate climbs and descents. The river route
from the coastal lowlands to deep within the highlands would have been a
good alternative to the High Road farther east. Of the 360 km of river and
trail from the junction to Amaciote, where the river leaves the mountains,
only about 70 km was by land. 80% was navigable river, a low-cost route
from the middle of a semi-desert to the lush coastal plains. It began as
an Olmec trade route, possibly the first to penetrate the interior of the
Highlands of Chiapas, and was later rivaled for efficiency by only the
Usumacinta.
So much of the Grijalva trade system has been lost by the damming of
the river. Of the original 290 km navigable, only 70 km remain nearly as
they were. Four dams have flooded the rest. That the major features of
this river and portage route can be sketched at all is largely due to the
efforts of Lee, Navarrete, and Lowe in the 1960s and 70s. The Grijalva is
what the Usumacinta valley would look like today, if the planned dams had
been built - ancient cities gone, details of the river route forever
unclear, but lots of hydro power.
Travelers ascending the Rio Grijalva [Mezcalapa] would have traversed
the compound delta westward through the Chontalpa for approx 240 km from
Frontera to Amaciote at the foot of the Skinalel Tolja, the ”Watery
Mountains” of the Chiapas Higlands. After ascending the river to
Amacoite, there were two ways to enter the highlands in the Colonial
period. One was to continue upriver for 80 km through the ranges, against
current and occasional rapids, to foot of the Raudales de Malpaso, which
were unrunnable (Lowe, 1981). With the flow regime completely altered and
the riverbed now invisible, the locations of runnable rapids, their
difficulty, their relation to sites, and to seasonal navigation are all
lost. The Malpaso portage trail followed a dry valley, the site of the
modern town of Raudales, on the west side of the river to flatwater in the
Mal Paso Basin [Middle Grijalva Basin] above the present-day dam. In the
wet season the portage may have been as little as 4 km. The basin is now
flooded by the Embalsa de Netzahuatlcoyotl, and the river approach drowned
under the Presa Penitas.
In the lush Mal Paso Basin, the Rio Grijalva, formerly called the
Kandelumihi, was literally lined with sites east to the foot of the
Sumidero Canyon. There were over 60 along both the Rios La Venta and the
Grijalva, all now gone.
The Rio La Venta was probably navigable to the west for some
distance, where a trail from the headwaters of the Rio Tonala could have
joined through a pass 8 km north. The Rio Playas appears to be navigable,
with no rapids - only sandbars, from the Preclassic sites of Ceiba Grande
and Pueblo Viejo, which are 30 km by trail from the Mal Paso Basin. This
would have been a direct route from La Venta, a major Olmec site in the
coastal plain, into the Mal Paso Basin via the Rios Tonala and Playas.
The Grijalva itself was navigable for at least 35 km from the Rio La
Venta upstream to Quechula [Cachula], a river port in colonial times (Lee,
1978). About 6 km downstream of Quechula was the largest site in the
region, San Isidro, dating from the Middle Preclassic to the Late Classic.
Its excavation was a hectic salvage effort, from mid-March 1966 until the
waters of the reservoir spread over it forever in June. Its flooding was a
tremendous loss. Mound 20, a circular pyramid with attached structures
oriented northwest-southeast, yielded exceptional burials and caches with
clear Olmec connections. Two parallel “I”-shaped ball courts shared the
same end zones, a unique design (Lowe, 1981).
The other trail from the Rio Grijalva to Quechula was the Tecpatan
Shortcut, which started in the lowlands at Amacoite and followed the foot
of the mountains east before climbing south over several high ridges to
Tecpatan (Lee, 1978). A descent led back to the river at Quechula [Cachula
]. The route avoided all swiftwater and rapids, but at the expense of much
up and down on mountain trails.
To bypass the Sumidero Canyon upstream, with impassable rapids, and
falls up to 17 meters high pinched between 1000 meter cliffs, colonial
trails from Quechula crossed the river to the south shore. In the Classic
and earlier, the trail would have begun at San Isidro, the ancient port
city. Trails to the southeast were easier than ones through the higher,
more extended ranges north of the river. A trail did exist north of the
river by way of Chicoasen to Chiapa de Corzo, but was not a major one in
colonial times. Only 30 km of mountains separate Quechula and San
Fernando, at the head of a flat valley, versus 75 km of precipitous trails
between Tecpatan and Chiapa de Corzo. As late as the mid-1960s, the only
approaches to Los Altos de Chiapas were still a choice of either rapid
river or muddy trail.
On the south side of the river from Quechula, the colonial trail
split. One branch went over the mountain south to Ocozocuatla, east
through Tuxtla, and then to Chiapa de Corzo. The other branch headed
southeast to cross the Rio Achilote, top the ridge just before San
Fernando, and follow the Valley of Tuxtla down to Chiapa de Corzo. The
second branch is the line of least effort, and probably the ancient
portage. From San Isidro by trail to Chiapa de Corzo via San Fernando
would have been about 65 km.
At the time of Spanish contact at end of the Late Postclassic, the
trail from Quechula to Chiapas was not used. When Captain Marin invaded
Chiapas in 1524 they first went to a town called Tezpuztlan and then
“continued our route to another town called Cachula from whence we
proceeded, there being no passage previous to our expedition, from the fear
the other natives have of those of Chiapas” (Diaz, 1568). This is one
documented instance where a past major route was bypassed in the Late
Postclassic for non-geographic reasons. Trade avoids danger.
Where the river again becomes navigable was the ancient Olmec/Zoquean
city of Chiapa de Corzo [Chiapa, Chiapa de Los Indios], founded around 1400
BC. It flourished until 900 AD, stumbled in the general collapse, but
recovered. At its zenith it was a sprawling city of talud-tablero step
pyramids on massive platforms. The Chiapan Maya wrested it from the Zoque
in 1400 AD. At the time of Spanish contact “a city it might truly be
called, from the regularity of its streets and houses. It contained not
less than 4000 families” (Diaz, 1568). At a nexus of major routes between
both coasts and highlands, the city had an enviable trade location. From
here a colonial trail went north through a gap in the wall of Los Altos and
then climbed east, 1800 m in 40 km, to San Cristobal de Las Casas [Jovel].
Another went west via Cintalapa and the exposed, windswept crest trail on
Mount Maquilapa (Gage, 1648) to Tapanatepec on the Pacific littoral [
Soconosco].
The Grijalva, here called the Rio Grande de Chiapas [Kandelumihi], is
today canoeable from Chiapa de Corzo for another 70 km upstream to the
vicinity of Belisario Dominguez. “About midnight, ten chieftains of the
neighboring districts came down the river, which is very broad and deep, in
five canoes” (Diaz, 1568). Even though the Central Depression of Chiapas
is a semi-desert, the river receives its water from the highlands to the
north and south. The gradient between Chiapa de Corzo and Belisario is a
low 0.2 m/km (1 ft/mi). The river has no rapids, though there are sandbars
in the dry season. A little over halfway up the river, on the north shore,
is the Villa de Acala (Spangaya for “Village of Canoes”), with a Classic
Period site across the river. A few km upstream is the Preclassic Santa
Cruz site.
Either the Finca Amatl, or the Angosturo site close to Belisario
Dominguez, may have been a port. A modern trail climbs northeast up the
long slope of an old lava flow past Cerro Mispia to a break in the wall of
the highlands at San Isidro Chijilte. A trail east from Belisario would
have hugged the foot of the highlands before climbing to Comitan [Balun
Canan]. At intervals, streams tumble out of the mountains and furnish
drinking water.
Above Belisario the river exits Angostura Canyon, now plugged by the
Presa La Angostura, feeding CFE hydros. The 12 km winding gorge is
shadowed by cliffs. Navarrete makes no mention of serious rapids or
portages, but his information was second-hand. It seems likely that it was
navigable. This is far from certain, and probably will remain so.
Above the canyon, the gradient was low, and there was much more
navigable river, all the way upstream 100 km to the confluence of the Rios
San Miguel and San Gregorio, where the river was 26 meters wide and 5
meters deep in the channel (Navarrete, 1978). Northeast 45 km is the Quen
Santo pilgrimage site, dating from the Preclassic, and 30 km east-southeast
are the ruins of Classic Period Lagarteros, on an island in the Lagunas de
Colon. The head of navigation lies at the foot of the Guatemalan
highlands, so the Grijalva offered a direct, easy way to tap the resources
of the Cuchumatanes.
With the whole valley flooded by the Embalsa de Belisario Dominguez
[Angostura], all details of this upper reach are beyond recovery. Over 20
sites are underwater, including Preclassic Santa Rosa and Laguna Francesca.
The Argelia and San Felipe sites were very near the head of navigation, and
one of the two was probably the port (Witschey map). A rectangular
structure is visible in the shallows just north of the confluence of the
Rios San Miguel and San Gregorio. It may be ancient San Felipe, or it may
be Spanish.
The greatest loss along the upper Grijalva was the site of the
Salinas de La Concordia [Custepeques, San Pedro de Las Salinas], centered
10 km south of the river in the Valley of Custepeques. Since the area is
semi-desert in the rain shadow of the Chiapas Highlands, solar evaporation
worked well, at least in the dry season. The 27 springs, seven salt works,
and attendant structures (Andrews, 1983) are all drowned under the
reservoir. From the Middle Preclassic to the 20th century, the salinas
sent pink salt out in several directions. Historically, some went east
across the highlands into the Usumacinta basin (Andrews, 1983) but more
traveled down the Rio Grijalva to the coast and westward. It is unlikely
that any went south to the Pacific coast, which had its own salt
operations.

Connections from the head of the Grijalva:
North-northeast to Chinkultic, then east via the Rios Santo Domingo &
Jatate into the Miramar basin.
Northeast to Quen Santo and Chacula to either the Miramar basin or into the
Cuchumatanes.
Southeast via Lagarteros into the Guatemalan highlands.
South-southwest through a pass to Escuintla and the Soconusco Road along
the Pacific littoral.

From the Cuchumatanes to La Venta:
A chain of Preclassic sites begins at the western foot of the
Cuchumatanes and extends down the Rio Grijalva into the Mal Paso Basin.
The list includes: Chacula, Quen Santo, Laguna Francesca, Santa Rosa, Santa
Cruz, Chiapa, and San Isidro. A pass to the northwest connects the Mal
Paso Basin to the headwaters of the Rio Playas. At the head of navigation
on the Playas are two Preclassic sites, Ceiba Grande and Pueblo Viejo. The
Rio Playas leads to the Rio Tancochapa, which in turn joins the Rio Tonala
less than 20 km from the Olmec city of La Venta. The overall distance from
Chacula to La Venta was roughly 540 km, of which 20% (120 km) was by land
and 80% by river (and most of that flatwater). The route from Chacula to
La Venta is as close to a straight line as a river road can get, and is
interrupted by only two portages. It was both a very ancient and very
efficient route.

Ron Canter, 1-22-10