*2.3. Venom Differentiation Is Not Limited to Rare (Low Abundance) Proteins*

We also investigated whether the venom variability in *B. atrox* groups would be associated with the degree to which the venom proteins were expressed in the different habitats. RP-HPLC peaks for each habitat were classified into two sets: low or high abundance, based on the clr-transformed mean of each peak and a nonparametric MANOVA analysis, as previously described [23], where rare proteins = the mean of an individual peak < the geometric mean, and abundant proteins = mean of an individual peak > the geometric mean (Figure 4). Proteins showing deviations off the middle line are those in which the expression levels are different between the compared populations whereas those that are on the line are similar.

**Figure 4.** Comparisons of the degree of differentiation in low vs. high abundance proteins based on chromatographic peak areas from *B. atrox* individual venoms. The clr mean was plotted for each RP-HPLC peak across the different habitat (x-axis) and (y-axis) populations for *B. atrox* snakes. Bars indicate the SE, a solid line indicates a perfect agreement, dashed lines indicate the origin (i.e., the geometric mean), and proteins less than these values were considered low-expression proteins.

In our PERMANOVA analyses, only in the spatially proximate degraded and forest habitats the venom composition did not significantly differ (Table 1). Accordingly, minor differences in both rare and abundant proteins were observed between these populations, which showed only slight differences in the rare proteins. However, in the populations from the other habitats, the venom differentiation was not constrained to rare proteins. On the contrary, in some cases, these proteins showed more similarity across populations from different habitats (for example floodplain vs. forest; pasture vs. forest) than proteins

classified as abundant. In fact, except for peak 23 (rich in PIII-class SVMPs), abundant proteins showed differences across all habitats. Interestingly, the abundant proteins would be closely related to the functional activity of the venom toward their prey. As shown before [15], the conserved fraction 23 contains Batroxrhagin, a multifunctional PIII-class SVMP extremely conserved in different samples of *B. atrox* venom [2]. However, in Figure 4, we show that other abundant toxins are differentially expressed between the snakes from different habitats: fraction 21, which predominantly contains a PI-class SVMP, had lower expression in floodplain venoms; fraction 10, which contains mostly SVSPs, had higher expression in floodplain venoms; and fraction 20, which predominantly contains a PI-class SVMP and CTLs, was relatively over-expressed in venoms from pasture compared to the other areas.

#### *2.4. Specific-Level Differentiation among Venoms of Snakes Collected in the Same Environment*

Once the variability among the groups was defined, we quantified intrapopulation variability in venom composition, among specimens collected in close proximity in the same areas. As shown in Figure 1, HPLC chromatograms show variable profiles within venoms from snakes collected in the same environment (for more details see Supplementary Figure S1). Consistent with this pattern, we found great heterogeneity in the percentage area of each peak in the groups (Table 2).

In every habitat, individual venoms showed important differences in both their percent area and the presence/absence of some peaks. Venoms from pasture and floodplain snakes were again the most distinct among the individuals from the same group (Table 2; Supplementary Figure S1), with venoms from the floodplain being the most heterogeneous. From Table 2 we emphasize three regions: In the PLA2-eluting fractions (Peaks 3 and 5), Peak 3 was absent in several chromatograms, but its absence is offset by the homogeneous increase of Peak 5 in pasture specimens. Venoms from floodplain specimens showed the most heterogeneous distribution of both PLA2 fractions. CTL fractions (Peaks 15–17) are expressed at low but homogeneous levels in pasture venoms with increased variation in venoms from the forest and degraded area, reaching the highest heterogeneity in venoms from the floodplain. A notable observation is the high level of variation in the expression of the peaks containing SVMPs (peaks from 21 to 23) in floodplain specimens. These are the most abundant and homogeneous fractions in venoms from the other areas, but in venoms from the floodplain, fraction 21 is very low or even absent in venoms of five specimens while fraction 22, almost not present in venoms from other areas, is detectable in high levels in venoms from four specimens from the floodplain, indicating a high variety of SVMP isoforms venoms from snakes in this environment. It is important to note that the higher variability observed in venoms from floodplain snakes is not due to the two distant floodplain spots of snake collection. The differences appointed here can be observed within the snakes collected at Santarém (ATXV 5, 7, 8, 9, and 16) or Oriximiná (ATXV 10, 11, 12, and 13), which are approximately 300 km apart.

#### *2.5. Differences in the Composition of Individual Venoms Resulting in Functional Variability*

The most variable components among the venoms included SVMPs, SVSPs, and PLA2s. Our next step was to evaluate some of the main biological activities related to these protein families. With the goal of reducing the number of experimental animals for toxicity tests for ethical reasons and due to the limited amounts of some individual venom samples, venoms from only 16 specimens were evaluated per functional test, comprising four from each habitat.



As shown in Figure 5, individual variation was observed in the catalytic activities of the major enzymatic components (SVMPs, SVSPs, and PLA2s) among the venom samples from each habitat (Figure 5A,C,E), except in the venoms from the floodplain, in which the SVMP catalytic activity was similar and low. There were significant differences among the venoms collected at the same habitat in hemorrhagic (Figure 5B) and myotoxic (Figure 5F) activities. The four venoms from the floodplain snakes (V5, V8, V13, and V16) induced hemorrhagic spots comparable to those induced by snake venoms from the other habitats, with an emphasis on the V8 snake, whose venom had the highest hemorrhagic activity in the floodplain group. For myotoxic activity, the greatest variation was found among the venoms from floodplain snakes: this group was the only one in which all tested venoms showed statistically significant differences in activity (Figure 5F).

**Figure 5.** Functional assays: (**A**) SVMP catalytic activity: evaluated by hydrolysis of FRET substrate (Abz-AGLA-EDDnp), expressed as RFU/min/μg of venom. (**B**) Hemorrhagic activity: evaluated by the size of the lesions observed 3 h after venom injection (10 μg) into the dorsal skin of mice, and expressed in cm2. (**C**) SVSPs catalytic activity: evaluated by hydrolysis of the chromogenic synthetic substrate (L-BAPNA), and expressed in Abs/min/μg of venom. (**D**) Pro-coagulant activity: evaluated the clotting times measured by thromboelastography of recalcified plasma from chickens, and the results

were expressed in terms of Coagulation Dose 50% (CD50). (**E**) PLA2s activity: evaluated by hydrolysis of the chromogenic substrate (NOBA), and the results were expressed in Abs/min/μg of venom. (**F**) Myotoxic activity: evaluated by the creatine kinase activity in mice serum 3 h after venom injection, and the results expressed in U/L. The data shown represent the mean + SD of three independent experiments. Controls: PBS and venom pools from *Bothrops jararaca*—Jar; *Bothrops jararacussu*—Jssu; *Bothrops insularis*—Ins. (\*) Asterisks indicate significant variations among venoms from a same habitat.

> The greatest within-group difference was found in procoagulant activity among the floodplain snakes, in which the DC50 values varied from 0.0035 to 1.965 μg for V8 and V10 snakes, respectively (Figure 5D). No other group showed such a large variation in DC50 values. However, the wide range of activity in this group was mostly due to the very high DC50 value of only one specimen (V10). On the other hand, pasture venoms were similar in procoagulant activity with only a small amount of variation observed in the venoms from the forest and degraded areas.
