The high degree of variability in growth rate and recruitment rate could also affect the ratio of supply and demand in the model. In an aggregation exhibiting anomalously low recruitment, the size of the rhizosphere would increase more rapidly than the biomass of the aggregation. This would lead to high rates of sulfide delivery and generation and low rates of sulfide uptake by tubeworm roots. When initial recruitment rate (a in equations 1 and 2) is decreased by 10%, the length of time that supply exceeds demand increases by 3.7%. This effect appears to be linear, with a 20% decrease in initial recruitment rate resulting in a 7.4% increase in persistence. If growth rate is increased, thereby increasing the rate of rhizosphere growth in terms of surface area for diffusion and advection, there appears to be little effect of the ratio of supply to demand (20% increase in growth—0% change in persistence time). In fact, increasing growth to the upper limits of the error term (equation 5) lowers the amount of time that the aggregation can be supported since biomass and sulfide demand increase more rapidly than increases in supply resulting from additional surface area. By decreasing growth rate, aggregations may be supported for longer periods of time, with a 20% decrease leading to a 6.3% increase in persistence time and a decrease of 88% leading to persistence for over 250 y. While an 88% lower growth rate lies outside of the range of existing growth data, this could be accomplished by ceasing growth for extended periods of time in a quiescent stage. This possibility remains to be investigated in L. luymesi. By utilizing a variable recruitment rate in the model, both between realized aggregations and between years within a model run, along with a growth error term encompassing the full range of observed growth data, the model is capable of generating aggregations within the range of the 10%–20% variability tested in this analysis. Even these outlying aggregations (presented as maxima and minima in Figure 1) support the qualitative conclusions drawn from model results.