Century-Scale Fire Dynamics in a Savanna Ecosystem
Round 1
Reviewer 1 Report
Lines 189-210 describe how charcoal in sediment records were used to reconstruct fire. It seems reasonable to reconstruct intervals between charcoal peaks and infer that these represents intervals between periods of burning (or relatively high burning relative to the background record). It seems reasonable to compare these peaks to a tree-ring based PDSI record. It is not reasonable to use SEA to compare fire years to annually resolved climate records because the fires inferred from charcoal are not annually resolved. You can use annually resolved fire events from cross dated fire scars recorded in tree rings, but not charcoal records from sediment cores. It is unclear exactly how the authors derived annually resolved fire dates from the charcoal records—I assume from line 341 that they picked the “median years,” presumably meaning the median year of a charcoal peak. But this clearly does not provide precise dates of fires because there is a suspicious fire gap between 1696 and 1807 and post-1800 dates do not match particularly well with the tree ring derived fire dates. I’m not even sure why the authors use SEA on these charcoal derived dates since there appears to be only one of two fire years derived from charcoal that are outside of the window of time during which fire years can be reconstructed from tree rings.
Please revise this sentence (lines 96-99): “We expected that because our study area was located within a matrix of the primarily fine fuels associated with prairies and savannas that extended dry conditions would have been associated with lower fire frequencies in theearly Holocene due to fuel limitation during droughts, and higher fire frequencies during moistconditions.” I don't understand the “the primarily fine fuels” construction… Please clarify if “lower fire frequencies” means shorter or longer intervals between fires. It is always confusing because a shorter(decrease) in fire intervals = more(increase) fire.
Line 102 has an awkward phrasing (“We expected that likely fine fuels was…”)
Figure 1 is way to blurry and low resolution to be interpreted. I assume that the final version of the paper will be at a higher resolution and easier on the eye.
Line 181: I don't undertand this phrasing: “with a smooth spline LOWESS.” Perhaps I’m mistaken, but I just haven’t seen the term “LOWESS” used like this before.
Line 201: I think “indicative for local fire” should be “indicative of.”
Line 203: I think “serie” should be “series.”
Line 206: “larger than 60 μm to be transported on long distances” needs to be re-written (“to be” is the wrong verb case).
Author Response
Lines 189-210 describe how charcoal in sediment records were used to reconstruct fire. It seems reasonable to reconstruct intervals between charcoal peaks and infer that these represents intervals between periods of burning (or relatively high burning relative to the background record). It seems reasonable to compare these peaks to a tree-ring based PDSI record. It is not reasonable to use SEA to compare fire years to annually resolved climate records because the fires inferred from charcoal are not annually resolved. You can use annually resolved fire events from cross dated fire scars recorded in tree rings, but not charcoal records from sediment cores. It is unclear exactly how the authors derived annually resolved fire dates from the charcoal records—I assume from line 341 that they picked the “median years,” presumably meaning the median year of a charcoal peak. But this clearly does not provide precise dates of fires because there is a suspicious fire gap between 1696 and 1807 and post-1800 dates do not match particularly well with the tree ring derived fire dates. I’m not even sure why the authors use SEA on these charcoal derived dates since there appears to be only one of two fire years derived from charcoal that are outside of the window of time during which fire years can be reconstructed from tree rings.
We understand that there is age uncertainty in the charcoal fire dates due to the lack of annual resolution in the sedimentary charcoal record. We take two approaches to handling this: (1) we clarify and we quantify the uncertainty in the analysis due to all sources, including age uncertainty (Fig. 3), and (2) we added a qualitative description of the PDSI conditions leading and lagging the charcoal peaks. We add Figure 4 with line graphs of 10-year low passed PDSI for the 21 years bracketing the charcoal identified fire events.
Beginning in Line 241, the methods section now reads “ To illustrate and qualitatively examine PDSI conditions around the time of charcoal determined fire events without precise calendar year dating, the annual PDSI values were fit with a 10-year cubic smoothing spline [50]. Figure 4 illustrates a composite plot of these smooth PDSI timeseries for the events before and after settlement at CE 1900."
Beginning around line 427, we added added brief description of the results from the qualitative analysis: “Considering the lack of annual resolution in the sedimentary charcoal record, we examine 10-year smooth PDSI values in the 21 year window bracketing the charcoal-based fire events (Fig. 4). Five of the seven pre settlement fire events exhibit 10-year PDSI spline values at or below zero. Three of the six post settlement fire events exhibit 10-year PDSI spline values at or below zero,"
Please revise this sentence (lines 96-99): “We expected that because our study area was located within a matrix of the primarily fine fuels associated with prairies and savannas that extended dry conditions would have been associated with lower fire frequencies in theearly Holocene due to fuel limitation during droughts, and higher fire frequencies during moist conditions.” I don't understand the “the primarily fine fuels” construction… Please clarify if “lower fire frequencies” means shorter or longer intervals between fires. It is always confusing because a shorter(decrease) in fire intervals = more(increase) fire.
We’ve revised this section for improved logic and clarity: “Because our study area was located within a landscape matrix of prairie and savanna, we expected that extended dry conditions would have driven a greater dominance of herbaceous plants, an increase in availability and dominance of fine fuels, and a higher frequency of lower intensity fire during in the early Holocene. Further, we expected drought and fire frequencies to become uncoupled because of Euro-American fire suppression."
Line 102 has an awkward phrasing (“We expected that likely fine fuels was…”)
We’ve revised this sentence for improved clarity: “We expected that fine fuels were most dominant prior to settlement, and that coarse fuels became more common following fire suppression reported in the 1930s and subsequent woody plant encroachment."
Figure 1 is way to blurry and low resolution to be interpreted. I assume that the final version of the paper will be at a higher resolution and easier on the eye.
Indeed, the final version will be provide in a resolution that exceeds the journal standards for print and digital media.
Line 181: I don't undertand this phrasing: “with a smooth spline LOWESS.” Perhaps I’m mistaken, but I just haven’t seen the term “LOWESS” used like this before.
We revised this phrasing for clarity, which now reads: “An age-depth model was established from the 14C dates and calibrated with Intcal09 calibration curve for the northern hemisphere [40] and the 210Pb dates with a locally weighted scatterplot smoothing (LOWESS) regression that is robust to outliers using the clam software package in R [41] (Fig. 1). “
We also revised our subsequent phrasing on the use of the LOWESS methodology in lines 220–222: “To decompose the CHAR series, we separated the CHARbackground and CHARpeak components by fitting the data with a LOWESS regression, and identifying outstanding events as those that surpass the locally defined threshold type."
Line 201: I think “indicative for local fire” should be “indicative of.”
We revised this phrasing accordingly: “indicative of local fire episodes"
Line 203: I think “serie” should be “series.”
We revised this spelling: “series”
Line 206: “larger than 60 μm to be transported on long distances” needs to be re-written (“to be” is the wrong verb case).
We revised this for improved grammar: “Although we cannot rule out the possibility of charcoal pieces larger than 60 µm being transported longer distances [46],"
Reviewer 2 Report
The authors have adequately addressed my comments/suggestions and I now recommend the manuscript for publication.
Author Response
The reviewer 2 did not ask for any further changes.
Round 2
Reviewer 1 Report
I am still uncomfortable using SEA on a fire history that is not annually resolved. However, the authors’ responses to my comments on this point are reasonable and I’m not prepared to argue the point further.
One last important comment about SEA: SEA is absolutely no appropriate when using a time series (e.g., climate reconstruction) that is auto-correlated at the same lag window as window used by SEA (I believe it is ten years in the case of this study). See Prager and Hoenig 1992 (“Can we determine the significance of key-event effects on a recruitment time series?—a power study of superposed epoch analysis”). A major benefit of most of the Cook PDSI reconstruction is that the resulting time series is typically not autocorrelated. The authors need to confirm that this is the case, and furthermore, confirm that smoothing the PDSI time series does not result in autocorrelation. I apologize for this late comment, but it just became clear to me that the PDSI time series was smoothed for use in SEA with the charcoal fire reconstructions.
Line 97: I would revise “would have driven a greater dominance of herbaceous plants…” (I think “driven” is the wrong verb).
Line 175: I would revise “During May 2016 CE, we established a short (101-cm) sediment…” (I think “established” is the wrong verb).
Lines 300-340 strike me as more appropriate for the discussion, but I don’t really care.
Revise line 394 (“in the that”).
The paragraph beginning line 460 is too long.
Author Response
I am still uncomfortable using SEA on a fire history that is not annually resolved. However, the authors’ responses to my comments on this point are reasonable and I’m not prepared to argue the point further.
We thank the reviewer 1 for their great implication in our study, and the time spent to provide useful insights. We would have prefer to completely remove the doubt about the choices we made in our analyses however we are grateful to read that the reviewer feels our effort as reasonable enough.
One last important comment about SEA: SEA is absolutely no appropriate when using a time series (e.g., climate reconstruction) that is auto-correlated at the same lag window as window used by SEA (I believe it is ten years in the case of this study). See Prager and Hoenig 1992 (“Can we determine the significance of key-event effects on a recruitment time series?—a power study of superposed epoch analysis”). A major benefit of most of the Cook PDSI reconstruction is that the resulting time series is typically not autocorrelated. The authors need to confirm that this is the case, and furthermore, confirm that smoothing the PDSI time series does not result in autocorrelation. I apologize for this late comment, but it just became clear to me that the PDSI time series was smoothed for use in SEA with the charcoal fire reconstructions.
We understand this point and we can reassure the reviewer there is absolutely no auto-correlation in our analyses, either coming from the time window used or from the proxy used to reconstruct the fire history. The PDSI was not smooth to use in the SEA, but the outputs of the SEA were averaged to discuss the 10 years prior and after fire episodes in the different periods considered. Furthermore, the fire episodes from the charcoal data have a time resolution varying from 2 to 8 years given the age-depth model. The smoothed PDSI is used on figure 4 to illustrate the tendency of dry or moist conditions before and after fire events, but the significance tests were not performed from this smoothed data.
Line 97: I would revise “would have driven a greater dominance of herbaceous plants…” (I think “driven” is the wrong verb).
Done.
Line 175: I would revise “During May 2016 CE, we established a short (101-cm) sediment…” (I think “established” is the wrong verb).
Done.
Lines 300-340 strike me as more appropriate for the discussion, but I don’t really care.
Since the results and discussion sections are merged in part 3, we think this is fine to let this paragraph here.
Revise line 394 (“in the that”).
Done
The paragraph beginning line 460 is too long.
We agree and separated in two paragraphs now.
