**4. Discussion**

#### *4.1. Erosion Variability*

Variability exists in erosion statistics between the three geomorphic areas, such that channels had the highest variability and interfluves the lowest (Table 2), with sidewalls having intermediate variability. In particular, for both the overall annual dataset and for each seasonal partition, the mean and standard deviation were of similar magnitudes. Similar behavior was observed in a previous study in the same study area [13] and a study of gully erosion in the Karoo region of Africa [23]. Channels were dynamic and acted as both source and sink for sediment loads. Slugs of sediments gathered intermittently in the channel areas and were transported with channel flow following precipitation. Soil erosion was dominant in the gully sidewalls, however, the variability was moderate compared to channel erosion data, implying that sidewalls were less responsive with regard to erosion. In contrast, in the interfluve, the lesser amount of erosion and variability reflected the limited sediment yield, which may be due to the presence of vegetation that retarded erosion and lower gradient. Additionally, differences in soil cover thickness, soil types, moisture content, slope aspect and angle within the different geomorphic settings may explain the range of variability, however, that is beyond the scope of this paper and will be studied in the future.

#### *4.2. Erosion-Precipitation Relationships*

Seasonally, a comparison of erosion variables and precipitation parameters shows the same trend. Ordering seasonal precipitation parameters (Duration and TotAcc) and erosion variables from greatest to least, winter was greatest, followed by spring, summer and lastly, autumn. We see in Table 2 that winter months were the most dynamic, with the greatest mean erosion and the largest standard deviation of all seasons, and this pattern was consistent across channels, interfluves, and sidewalls for all erosion variables. This may be explained by the character of the winter precipitation: greater total accumulation and duration during these months associated with frontal precipitation events. Prior research has also demonstrated that freeze-thaw events are significant drivers of erosion in winter months at this site [14,19]. A similar pattern existed for spring, likely influenced by precipitation accumulation and duration as well as antecedent winter freeze-thaw activity [19]. Next, summer erosion and precipitation (Duration and TotAcc) ranked third, but interestingly, summer experienced the highest precipitation intensity of all seasons (both for AvgInt and MaxInt) (Table 1). This reflected the dominance of convectional precipitation events in summer. Autumn experienced the minimum erosion and precipitation accumulation and duration, but greater maximum precipitation intensity than the annual average. This suggests that autumn precipitation events were short duration, high-intensity events that did not produce much precipitation depth and had little erosive power.

During winter 2016–2017, precipitation variables were near normal levels for the winter season, however, erosion for all geomorphic areas was very low (Figure 3). We examined temperature during this time period to determine whether the reduced freeze-thaw activity may have played a part, but, while winter 2016–2017 had less intense freeze-thaw activity than other winters during the study period, freeze-thaw events occurred. The timing of the greatest precipitation accumulation and duration was late autumn/early winter, and because these events were coincident, they indicate a period of low-intensity precipitation that may have encouraged more infiltration and less runoff, leading potentially to less erosion during this period. Lower hydrostatic pressure in unsaturated soils increases cohesion [24] which may be a significant factor associated with reduced erosion in late autumn and early winter of that year.

Average Intensity and Maximum Intensity of precipitation were very different, with approximately three orders of magnitude between the generally low average precipitation intensities and maximum intensity for the full dataset and each seasonal partition (Table 1). Future research at this site should assess the soil's infiltration capacity and explore different metrics that may better capture the relation between precipitation intensity and erosion. For example, measuring the rainfall duration when the

rain rate exceeds the soil's infiltration capacity would generate a metric of the length of time during which there was a high probability of runoff generation.
