Sunday: 17 July 2011
We’ve had a pleasant change in the weather since Thursday night, when we returned from sweltering training to cool breezes and impending storms. On Friday, the temperatures dropped and the high was in the upper 70s, something that has not happened since May 18. Yesterday Saturday was a little warmer, and today might reach upper 80s. Then it’s back to above normal with 100F temperatures predicted for Wednesday and Thursday.
I’ve been getting all the box turtles in a row, or rather, multiple rows on excel, and dabbling a bit with mark and recapture analysis of the findings over the last six years. The goal is to estimate the size of the box turtle population on the property, although I’m violating somewhat some of the mark and recapture methods.
Mark and Recapture:
Mark and recapture experiments are fun to perform - biology student labs often attempt this with butterflies in the field or fish in a tank. The idea is to make two “visits” to a defined area. Each visit, as many animals of the species of interest are captured as possible. On the first visit, the individuals are captured, marked in some way, and released. The total count is M. On the second visit, a similar intensive capture is made. The total count is C, and of these some number R will have the markings of the first visit’s capture.
One simple model, the Lincoln-Petersen model, assumes that the proportion of marked individuals recovered among the total counted in the second visit is the same as that of all the marked individuals relative to the total population N.
So R/C = M/N , and N = M*C/R .
The mark and recapture experiments rely on some assumptions:
1. Marking does not perturb the animals, and markings are not lost. (In my case, “marking” is photography of the carapace patterns for matching).
2. Related to 1, marked individuals should not become easier or more difficult to recapture. Similarly animals in the second visit should not become harder or easier to capture, compared to the first visit. (Unless our turtles sound off “Oh god, hide. Here comes the creeper with the camera again,” we have no worries.)
3. There is no emigration or immigration of animals into or out of the study area during the study time. (In my case, there probably was movement into and out of the area, although box turtles tend to be territorial. Still, we know some of the turtles observed here during the “first visit” have been spending time over at the neighbors', and so it’s likely that some movement is violating this rule. Some box turtles take up a wandering lifestyle, staying in one place only a short time before moving on.)
4. Related to 3, the time between visits should be short enough so that movement of individuals into or out of the study area is minimized, and mortality does not occur. The time between visits should be long enough for individuals marked and released from the first visit to become interspersed with those not captured. (For box turtles, this time can be extended because of the territoriality and the low mortality rate of adults.)
There are other assumptions, but most are common sense.
Study Area:
Our property as a whole is relatively isolated. Goulding Creek bounds it on the north, and across from Goulding is more habitat of the same undisturbed sort. Black Snake Road uphill in that direction is a dirt and gravel road not presenting much in the way of traffic concerns. Goulding Creek and its attendent floodplain and hardwood forest is a major wildlife corridor into the study area. SBS Creek runs through the study area along a hollow from south to north, and is generally active year round, running dry during the summer for only two consecutive years since 1985. The forest upstream and downstream is hardwood - mostly white and northern red oaks, hickories, and beech, with hornbeams, buckeye, dogwood, and some redbud in the understory. Mayapples, woodvamp, muscadine, and poison ivy are common understory ephermerals or vines.
On the northeast side of the study area is an uphill slope, and this too represents a known box turtle corridor. To the east and southeast is mostly upland pine forest with some recent hardwood coming back, and this is true for much of the land to the west - not great box turtle habitat, but permeated by lowland hardwood forest. Its 300 acres represents a buffer disturbed on decade intervals by some logging. Wolfskin Road running along the south from west to east obviously represents a significant problem in traffic related mortality, and there is on either side coverage by human development - residences and/or pastures.
Predation and competition: predators of adults are probably mostly coyotes, perhaps the occasional raccoon or fox, although it’s been years since I’ve seen a fox on the property. Raccoons, opossums, canines, larger birds, and probably snakes can manage baby box turtles easily. The recent incursion of armadillos into the area may offer direct competition for resources.
SBS Turtle History:
I did not plan on doing a mark and capture - this analysis just came about after looking at all the data I had and wondering if I could find a way to rationalize using it to estimate a total population size. So I’ve abused the model somewhat, mainly by dividing a five year study period into two periods of 2 years (the first visit and “marking” period) and 3 years (the second visit and “recapture” period).
These are rather lengthy sampling periods, but for box turtles they may not be too long to yield results of some value. Probably the worst effect is that of turtles wandering into or leaving the area. This reduces R, the number rediscovered in the second visit, and therefore inflates the total population estimate. So our result is more likely to be an overestimate.
Our first “visit” covered Nov 2005-May 2008. During this time I came across and photographed 11 unique turtles (M) within the study area.
Our second “visit” began where the first one left off, May 2008, and is ongoing. During the past three years of the second visit, we’ve found 14 total turtles (C) and 4 of these are “marked” from the first visit (R).
Results:
Our estimate of the population size in the study area is therefore 11*14/4 = 38, of which we’ve seen 21. That’s somewhere around 4.8 turtles per hectare, or 1.9 turtles per acre. This is not an unusual population density in a forest location.
A few other results:
Over the last 5 years, I’ve found 3 dead adult turtles. That may be a rather high mortality rate for a population of 38 turtles: 7-10% over 5 years is 1.4-2% per year. In these three cases the cause of death was most likely disease, or weather related.
Of the 21 unique turtles observed, 10 were male and 11 were female.
Not included in the study were the turtles found in the 20 acres west of the study area. I’ve found 9 turtles (6 male and 3 female), but this area has only been observed for a couple of years, and not nearly so intensively.
Also not included were last and this year’s observations by our neighbors. They have found 5 previously unobserved turtles, and an additional 3 that had been previously marked by me. One of these was not included in my estimates, since I never actually came across Maggie a second time. Of the 8 unique turtles they’ve photographed, only one (a rediscovery, Ernest) was definitely male. Most of the females were in the process of digging nests.
The total number of unique (live) turtles found: 35, with 6 of these found at least twice.
I ran across an interesting PhD dissertation (pdf) by Nathan Nazdrowicz, surveying box turtles at four study sites in Delaware. He and his colleagues were, of course, far more meticulous, intensive, and consistent in their observations, and use more complex models to tease out a great deal more information on their populations.
I particularly liked the descriptions of the study areas - a little history of the area itself, the forest community, its degree of isolation and connections by wildlife corridors, and the encroachments by development around them. To give you an idea of the difference in scale, “From 16 April 2001 through 14 November 2002, we captured 268 turtles 892 times on the 4 study sites. We captured 16 turtles 77 times at the University of Delaware Woodlot...”
Besides the occasional documentation of discovered individuals, here are some links to relevant previous posts:
First turtle blog entry 11 Nov 2005
The comments that probably got me started on this: 28 Jul 2006.
Reflections on box turtles as property owners: 9 Sep 2008.
Yellow box turtle: 21 May 2008.
The turtles of 2009: 2009.
And of course, the babies:
Views of the babies: 18 Sep 2009.
Characterizing the babies: 19 Sep 2009.
Releasing the babies: 23 Sep 2009.
I’ve been getting all the box turtles in a row, or rather, multiple rows on excel, and dabbling a bit with mark and recapture analysis of the findings over the last six years. The goal is to estimate the size of the box turtle population on the property, although I’m violating somewhat some of the mark and recapture methods.
Mark and Recapture:
Mark and recapture experiments are fun to perform - biology student labs often attempt this with butterflies in the field or fish in a tank. The idea is to make two “visits” to a defined area. Each visit, as many animals of the species of interest are captured as possible. On the first visit, the individuals are captured, marked in some way, and released. The total count is M. On the second visit, a similar intensive capture is made. The total count is C, and of these some number R will have the markings of the first visit’s capture.
One simple model, the Lincoln-Petersen model, assumes that the proportion of marked individuals recovered among the total counted in the second visit is the same as that of all the marked individuals relative to the total population N.
So R/C = M/N , and N = M*C/R .
The mark and recapture experiments rely on some assumptions:
1. Marking does not perturb the animals, and markings are not lost. (In my case, “marking” is photography of the carapace patterns for matching).
2. Related to 1, marked individuals should not become easier or more difficult to recapture. Similarly animals in the second visit should not become harder or easier to capture, compared to the first visit. (Unless our turtles sound off “Oh god, hide. Here comes the creeper with the camera again,” we have no worries.)
3. There is no emigration or immigration of animals into or out of the study area during the study time. (In my case, there probably was movement into and out of the area, although box turtles tend to be territorial. Still, we know some of the turtles observed here during the “first visit” have been spending time over at the neighbors', and so it’s likely that some movement is violating this rule. Some box turtles take up a wandering lifestyle, staying in one place only a short time before moving on.)
4. Related to 3, the time between visits should be short enough so that movement of individuals into or out of the study area is minimized, and mortality does not occur. The time between visits should be long enough for individuals marked and released from the first visit to become interspersed with those not captured. (For box turtles, this time can be extended because of the territoriality and the low mortality rate of adults.)
There are other assumptions, but most are common sense.
Study Area:
| Although I’ve done turtle observations over most of our property, bounded in red below, I’ve confined the counts made here to the area I’ve most intensively covered, the 20 acre (8 hectare) area marked in green. You can ignore the blue dots and other unexplained markings - this is a map left over from some other purpose a couple of years ago. |  |
Our property as a whole is relatively isolated. Goulding Creek bounds it on the north, and across from Goulding is more habitat of the same undisturbed sort. Black Snake Road uphill in that direction is a dirt and gravel road not presenting much in the way of traffic concerns. Goulding Creek and its attendent floodplain and hardwood forest is a major wildlife corridor into the study area. SBS Creek runs through the study area along a hollow from south to north, and is generally active year round, running dry during the summer for only two consecutive years since 1985. The forest upstream and downstream is hardwood - mostly white and northern red oaks, hickories, and beech, with hornbeams, buckeye, dogwood, and some redbud in the understory. Mayapples, woodvamp, muscadine, and poison ivy are common understory ephermerals or vines.
On the northeast side of the study area is an uphill slope, and this too represents a known box turtle corridor. To the east and southeast is mostly upland pine forest with some recent hardwood coming back, and this is true for much of the land to the west - not great box turtle habitat, but permeated by lowland hardwood forest. Its 300 acres represents a buffer disturbed on decade intervals by some logging. Wolfskin Road running along the south from west to east obviously represents a significant problem in traffic related mortality, and there is on either side coverage by human development - residences and/or pastures.
Predation and competition: predators of adults are probably mostly coyotes, perhaps the occasional raccoon or fox, although it’s been years since I’ve seen a fox on the property. Raccoons, opossums, canines, larger birds, and probably snakes can manage baby box turtles easily. The recent incursion of armadillos into the area may offer direct competition for resources.
SBS Turtle History:
I did not plan on doing a mark and capture - this analysis just came about after looking at all the data I had and wondering if I could find a way to rationalize using it to estimate a total population size. So I’ve abused the model somewhat, mainly by dividing a five year study period into two periods of 2 years (the first visit and “marking” period) and 3 years (the second visit and “recapture” period).
These are rather lengthy sampling periods, but for box turtles they may not be too long to yield results of some value. Probably the worst effect is that of turtles wandering into or leaving the area. This reduces R, the number rediscovered in the second visit, and therefore inflates the total population estimate. So our result is more likely to be an overestimate.
Our first “visit” covered Nov 2005-May 2008. During this time I came across and photographed 11 unique turtles (M) within the study area.
Our second “visit” began where the first one left off, May 2008, and is ongoing. During the past three years of the second visit, we’ve found 14 total turtles (C) and 4 of these are “marked” from the first visit (R).
Results:
Our estimate of the population size in the study area is therefore 11*14/4 = 38, of which we’ve seen 21. That’s somewhere around 4.8 turtles per hectare, or 1.9 turtles per acre. This is not an unusual population density in a forest location.
A few other results:
Over the last 5 years, I’ve found 3 dead adult turtles. That may be a rather high mortality rate for a population of 38 turtles: 7-10% over 5 years is 1.4-2% per year. In these three cases the cause of death was most likely disease, or weather related.
Of the 21 unique turtles observed, 10 were male and 11 were female.
Not included in the study were the turtles found in the 20 acres west of the study area. I’ve found 9 turtles (6 male and 3 female), but this area has only been observed for a couple of years, and not nearly so intensively.
Also not included were last and this year’s observations by our neighbors. They have found 5 previously unobserved turtles, and an additional 3 that had been previously marked by me. One of these was not included in my estimates, since I never actually came across Maggie a second time. Of the 8 unique turtles they’ve photographed, only one (a rediscovery, Ernest) was definitely male. Most of the females were in the process of digging nests.
The total number of unique (live) turtles found: 35, with 6 of these found at least twice.
I ran across an interesting PhD dissertation (pdf) by Nathan Nazdrowicz, surveying box turtles at four study sites in Delaware. He and his colleagues were, of course, far more meticulous, intensive, and consistent in their observations, and use more complex models to tease out a great deal more information on their populations.
I particularly liked the descriptions of the study areas - a little history of the area itself, the forest community, its degree of isolation and connections by wildlife corridors, and the encroachments by development around them. To give you an idea of the difference in scale, “From 16 April 2001 through 14 November 2002, we captured 268 turtles 892 times on the 4 study sites. We captured 16 turtles 77 times at the University of Delaware Woodlot...”
Besides the occasional documentation of discovered individuals, here are some links to relevant previous posts:
First turtle blog entry 11 Nov 2005
The comments that probably got me started on this: 28 Jul 2006.
Reflections on box turtles as property owners: 9 Sep 2008.
Yellow box turtle: 21 May 2008.
The turtles of 2009: 2009.
And of course, the babies:
Views of the babies: 18 Sep 2009.
Characterizing the babies: 19 Sep 2009.
Releasing the babies: 23 Sep 2009.
