Discussion

The catastrophic effects of the Tiwanaku collapse on the Taraco Peninsula settlement system are conveyed admirably in Figure 9.5. The map is virtually a negative impression of the Tiwanaku period settlement density map (Figure 8.4), and effectively communicates the two profound settlement shifts which took place at the end of the Middle Horizon:

1. The shift from nucleated to dispersed settlement. With the collapse of the Tiwanaku polity, the old system of nucleated village habitation which had developed over more than a two millennia was abandoned altogether. We may imagine that a whole system of territoriality - borders, usufruct rights, landscape-related ceremonialism, and so on - was simultaneously abandoned.
2. General demographic collapse. The phase population index of the Early Pacajes phase is only about $\frac{1}{5}$ that of the preceding Tiwanaku period, compared to which it is most likely relatively exaggerated.

All settlement archaeologists working in the Titicaca Basin recognize the general shift from nucleated to dispersed habitation at the end of the Middle Horizon. This fact is entirely beyond controversy. It is one of the defining features of the Early Pacajes phase, together with the changes in ceramic style discussed above. What is more controversial, of course, is the reason for these changes. Approaches to explaining the dramatic changes at the end of the Middle Horizon can be classed in two principal groups.

On the one hand, there are those scholars who explain the differences between the Tiwanaku period and Early Pacajes inhabitants of the southern Titicaca Basin in terms of a population replacement. We may term this the ``discontinuity hypothesis''. These scholars, mainly linguists and ethnohistorians, posit a series of ethnic migrations - the famous ``Aymara invasions'' (cf. [Bouysse-Cassagne 1988,Gisbert et al. 1987,Torero 1985]). These invasions supposedly replaced the Tiwanaku period population of the southern Titicaca Basin - often identified as Pukina speakers - with a group of more nomadic Aymara-speaking pastoralists from the southern altiplano. These southern pastoralists had no tradition of village dwelling. In this way the discontinuity hypothesis is in fact able to explain the observed shift from nucleated to dispersed habitation.

Proponents of the ``continuity hypothesis'' - including most archaeologists (cf. [Browman 1981]: 417-418, [Rydén 1947]) - contend that the LIP inhabitants of the southern Titicaca Basin - and by extension the modern population - were in fact descendants of the builders of Tiwanaku, and that no invasion, migration or ethnic replacement took place in the LIP. They point to evidence of continuity, such as the strong similarity of domestic ceramic wares between the two phases. Most settlement archaeologists who have worked in the Titicaca Basin fall into this camp (cf. [Albarracín-Jordan 1996a]: 294-296, [Janusek 1994]: 380-383, [Mathews 1992]: 190-191, [Stanish et al. 1997]: 12-14, 118), though all admit that the evidence is not at all conclusive.

As we saw, the discontinuity hypothesis can explain the shift from nucleated to dispersed settlement by positing an invading Aymara horde unacquainted with village life. One problem with the continuity hypothesis that has yet to be addressed by its proponents is how exactly to explain this shift in settlement practice which is so crucial to understanding the Tiwanaku collapse and the Early Pacajes phase. The problem in not even recognized by most researchers, who seem to see the abandonment of the village system as a natural corollary of the collapse of the Tiwanaku administrative hierarchy. It is as if the village system was in some way simply a material manifestation - a shadow or reflection - of a bureaucratic structure, which naturally ceased to be when the institutions which generated it dissipated. Of course, nothing could be further from the truth. The present study has clearly demonstrated that the village system on the Taraco Peninsula originated in the Early Formative Period and persisted with only minimal modification for at least 1500 years prior to Tiwanaku state formation. Therefore, there is no reason to suppose that Tiwanaku state collapse would necessarily produce the wholesale abandonment of village life.

In considering the related problem of compact versus dispersed residential practices in Mesoamerican settlements, Drennan has observed that it is in some ways more difficult to explain dispersed habitation than nucleated habitation ([Drennan 1986]). His discussion is enlightening and well worth reading for anyone interested in these issues. He ends by explaining the dispersed settlement characteristic of Classic Maya cities - and incidentally of Postclassic Basin of Mexico settlement also - with reference to intensive forms of agriculture. His argument is that forms of agriculture characterized by ``not just large labor requirements, but the concentration of those requirements continually in a small area ... [make] it desirable for a household to locate its residence at its agricultural plot'' ([Drennan 1986]: 287). He contends that dispersed settlement occurs when the advantages to the household of proximity to its agricultural plot outweighs the advantages associated with proximity to other households. Nucleated settlement is produced when the balance of the advantage falls to proximity to other households. He specifically argues that ridged fields - chinampas - and terraces are agricultural techniques which produced dispersed settlement in Classic Maya and Late Postclassic Mexican highland contexts.

Given Drennan's argument, it is particularly interesting to observe that Graffam has constructed an argument that raised field agriculture did not end with the Tiwanaku collapse, but was in fact intensified further ([Graffam 1990]). He argues that in the wake of the Tiwanaku collapse individual households intensified production by incorporating raised field agriculture and expanded camelid pastoralism into their existing household subsistence strategies. Increasing importance of raised field agriculture certainly could account for a shift to dispersed settlement at the end of the Middle Horizon. Though most archaeologists interested in the Titicaca Basin seem to be skeptical of the empirical basis of Graffam's model, I think that it very easily explains the observed settlement shifts. For this reason, I would suggest that we consider Graffam's claims more carefully and systematically.

Another possibility is that the critical factor was not in fact the advantage associated with proximity to an agricultural plot, but rather concrete disadvantages associated with proximity to one's neighbors. Drennan does not make this argument, but the scenario follows naturally from his discussion. It is fairly easy to imagine that a large increase in the size of a household's herds might make life in a nucleated settlement problematic. Pastoral communities typically are characterized by dispersed settlement worldwide. An intensification of pastoralism could also account for the settlement shift.

Either of these two scenarios could also account for the dramatic decline of population on the Taraco Peninsula, which was much more pronounced than in adjacent regions. This population decline could be explained by households relocating to adjacent regions, such as the Pampa Koani and the Tiwanaku Valley, which are much better-suited for both raised field agriculture and pastoralism. The drop in the Taraco Peninsula population index could also be explained - but only in part - by a large number of households moving out onto the lake plain at this time. Remember that the lake level was much lower than today throughout the Early Pacajes phase. Many Early Pacajes pampa sites would be inundated today, and thereby excluded from the sample.

I should emphasize that I am not constructing a defense of the continuity hypothesis, though I do find that it explains the archaeological record more satisfactorily than does the invasion scenario. Rather, I have presented two hypothetical explanations for the observed shift from nucleated to dispersed habitation in the LIP, both related to changes in the domestic economy and both consistent with the continuity hypothesis. I consider these as hypotheses to be tested by future research. I have also attempted to stress that any account of the Early Pacajes phase must explain the observed shift from nucleated to dispersed modes of settlement.

The issue of a catastrophic demographic collapse is not so much contentious among archaeologists as it is avoided. The substantial increase in the number of sites in the Early Pacajes phase is sometimes taken to indicate not only that there was no demographic collapse, but also that population may, in fact, have increased relative to Tiwanaku period levels.

... the increase in number of sites by some 150 percent over the preceding Tiwanaku V period suggests not so much an overall population decrease in the [Middle Tiwanaku] valley with the collapse of Tiwanaku, but rather a population reorganization... [[Mathews 1992]: 188]

In general, it seems as though many Titicaca Basin archaeologists prefer to minimize the ``collapse'' aspect of the Tiwanaku/Early Pacajes transition, and to emphasize instead the ``reorganization'' or ``political fragmentation'' aspects. Janusek, for example, characterizes the situation in this way:

... Tiwanaku ``state collapse'' materialized as mutually supportive processes of political fragmentation and cultural revolution among local segmentary units, involving the assertion of political autonomy by local leaders and the gradual adoption of new ideals, attitudes, social identities, and material culture. [[Janusek 1994]: 383]

If we accept the continuity hypothesis, this is no doubt a fair picture of at least some of the cultural processes which were under way early in the Early Pacajes phase. However, Janusek's description serves to obscure the fact that the early LIP was a period of dramatic population decline and dislocation. Put plainly, it was a chaotic and dangerous time, probably involving considerable localized conflict. population relocation and even famine.

As I discussed in Chapter 4, using raw hectares as a population measure has the effect of exaggerating the importance of small and ephemeral sites relative to large and permanent villages or towns. In the context of the Tiwanaku/Early Pacajes transition, this means that the Early Pacajes phase population will appear relatively larger than it was, thus reducing the apparent severity of the population decline which took place.

In the case of the Taraco Peninsula, the difference between the two measures can be calculated directly, as shown in Table 9.2. The uncorrected growth rate is -0.37% annually, while the corrected rate is a more extreme -0.46%. The two values can also be compared in the case of the Lower Tiwanaku Valley, since Albarracín-Jordan presents his data in such a way as to make this possible. In this case the uncorrected growth rate appears to be higher than the 0.10% norm I have established in earlier chapters. This would suggest no population decline at all, but rather normal population growth. However, when sector sizes are corrected using the method described in Chapter 4, the corrected annual growth rate in the Early Pacajes phase is only 0.01%. This is essentially zero growth for a 350-year period.

Table 9.2: Corrected and uncorrected aggregate population measures by region: Middle Horizon and Late Intermediate Period

		Taraco	Lower Tiw	Middle Tiw	Catari Basin	Juli-Pomata
uncorrected	MH	131.5	86.3	642.7	229.0	62.3
(hectares)	LIP	35.7	122.9	269.9	162.4	73.9
	annual growth	-0.37	0.14	-0.25	-0.10	0.05
corrected	MH	6923	4627	[100]	[100]	[100]
(pop index)	LIP	1408	4871	[31.25]	[52.77]	[88.27]
	annual growth	-0.46	0.01	[-0.33]	[-0.18]	[-0.04]

Sources:

  Lower Tiwanaku Valley: [Albarracín-Jordan 1992]

  Middle Tiwanaku Valley: [Mathews 1992]

  Katari Basin: [Janusek and Kolata 2002]

  Juli-Pomata: [Stanish et al. 1997]

In the case of other surveyed Titicaca Basin regions, I have not been able to correct the occupation sizes in the same way that I have my own and Albarracín-Jordan's data. However, if we assume that the same relation pertains between average site sizes in the Tiwanaku and Early Pacajes phases, then it is possible to calculate corrected annual growth rates by another method.

The relation between corrected and uncorrected population values can be expressed in the following equation.

  
$\frac{Tiw_{c}}{EP_{c}}=k\frac{Tiw_{u}}{EP_{u}}$ 




where $Tiw_{c}$ = Tiwanaku period phase population index 


		 		 $Tiw_{u}$ = Tiwanaku period raw hectares 


		 		 $EP_{c}$ = Early Pacajes phase population index 


		 		 $EP_{u}$ = Early Pacajes raw hectares 


		 		 $k$ = a correction constant

It is a simple matter to derive the value of $k$ from this equation, as follows:

$k=\frac{Tiw_{c}EP_{u}}{Tiw_{u}EP_{c}}$

Solving for $k$ for the Taraco Peninsula and Lower Tiwanaku Valley data - the only two for which we have complete corrected settlement datasets - generates values of 1.33 (Taraco Peninsula) and 1.35 (Lower Tiwanaku Valley). These two values are very close to one another indeed, a fact which I hope inspires at least a degree of confidence in the method. I will assign a provisional value of 1.34 to $k$, obtained by averaging the two empirical values.

The corrected value for the Early Pacajes phase can therefore be obtained using the following equation:

$EP_{c}=\frac{Tiw_{c}EP_{u}}{kTiw_{u}}$

Inserting the value derived for $k$ above, and using an arbitrary population index value of 100 for the Tiwanaku period, a population index value for the Early Pacajes phase is derived as follows:

$EP_{c}=\frac{100}{1.34}\frac{EP_{u}}{Tiw_{u}}=74.63\frac{EP_{u}}{Tiw_{u}}$

Note that the value obtained in this way is not an absolute one. It is only a percentage of the corrected Tiwanaku period value, whatever that may be. However, it will suffice for our present purposes, since we care only to calculate an annual percentage growth rate for the Early Pacajes phase for the various regions in question. These growth rates are presented in Table 9.2 in brackets to indicate that they have been derived by the method above. The empirical growth rates from the Taraco Peninsula and Lower Tiwanaku Valley are presented without brackets.

When the ``splash zone'' effects of the shift from nucleated to dispersed habitation are subtracted, the population decline in the Early Pacajes phase appears much more severe than when uncorrected hectares are employed. Of the five regions sampled, not a single one has positive population growth in the LIP. The Middle Tiwanaku Valley and the Katari Basin experienced the most severe depopulations, excepting the Taraco Peninsula, as would be expected given the disintegration of the cities of Tiwanaku and Lukurmata. In the Juli-Pomata area, what Stanish characterizes as ``a slight leveling'' ([Stanish et al. 1997]: 60) in the growth curve in the LIP is actually, when corrected for site size effects, a decline in population, though a less severe one than in the Tiwanaku heartland. In fact, the general basin-wide decline is probably even more severe than I have indicated, since I have failed to correct for the effects of more ephemeral and generally shorter-term occupations in the LIP.

Since all of the areas in question are low and near the lakeshore we may perhaps postulate a generalized movement of population away from the lakeshore and into higher and more mountainous regions throughout the Late Intermediate Period in the entire Titicaca Basin. It is perhaps not necessary to imagine large-scale warfare and famine as the causes of the observed basin-wide depopulation. Whatever the case, though, the collapse of the Tiwanaku state was no simple rearticulation of its component segments, as has been suggested. Neither was it a matter only of a political or cultural revolution. The Tiwanaku collapse was not only a political collapse, but a demographic one as well.