Geoscience Reference
In-Depth Information
communities. The lack of this kind of information on scaling patterns within
and among local communities represents a main limitation to assessing theo-
retical predictions in a robust manner (
Hayward et al., 2009, 2010
).
A. Study System
The study area is a system of temporal ponds, located in the Laguna de
Castillos basin in southeastern Uruguay (34
25
0
47
00
S, 53
98
0
10
00
W;
5-8 meters a.s.l.). Ponds are formed in land depressions which every autumn
and winter fill up with rainwater, drying up by the end of November. This
dynamic is repeated every year when the precipitation exceeds evaporation
rates. The system comprises more than 50 ponds, varying in abiotic attributes
(e.g. depth, spatial heterogeneity, content of organic matter), and exhibits
differences in area up to five orders of magnitude (minimum surface area:
1m
2
; maximum surface area: 25,000 m
2
;
Laufer et al., 2009
). This system is
characterized by its high diversity: more than 100 macrophytes species, 200
macroinvertebrates species (belonging to 22 orders), 5 species of fish (4 of
them annual fishes), and 4 species of amphibians have been identified in these
ponds. Annual fishes are the top predators of the system and exhibit a strong
positive association between body size and trophic position. All the afore-
mentioned taxa either complete their life cycles within these ponds, or are
adapted to these temporal dynamics (e.g. amphibian tadpoles and insects
with terrestrial adults). The samples used in this study were taken at the end
of one of the system's cycles and comprised 18 communities. Fishes, amphi-
bians, and macroinvertebrates were collected using a hand-net (15
20 cm,
1 mm mesh). The 6773 sampled individuals were measured in length, height,
and width to estimate their volume (
Arim et al., 2010; Laufer et al., 2009
).
B. Five DMRs in a Single System
For the analysis of the GSDR, all the estimation of population densities and
the estimated mean body sizes were used. The LSDR was analysed using
abundances and mean body sizes estimated in a single community. In these
cases, a linear regression was fitted to this dataset, and the occurrence of non-
linear association and linear or polynomic trends in quantile values were
evaluated (
Cade and Noon, 2003
). Cumulative distribution and ML values
for the scaling parameter were estimated for the ISD and SMSD following
the methodology introduced above. The existence of regimen shifts in scaling
value was evaluated with segmented regressions and non-linearity fitting
second- and third-order polynomics. The model with the lower Akaike
Information Criterion (AIC) was selected as the best model. Finally, the
