A Modeling Approach to Predicting Cave Disturbance on Karst Landscapes: Implications on Cave Management in West-Central Florida

A-Modeling-ApproachA Modeling Approach to Predicting Cave Disturbance on Karst Landscapes: Implications on Cave Management in West-Central Florida

 

 

 

Introduction

 

            Karst is a landscape primarily shaped by the dissolution of carbonate bedrock or sediment and includes an aggregate of landforms and subsurface features (Quinlan, 2007). Within these landscapes, cave systems often exist that contain valuable resources used for scientific research (Kambesis, 2007). However, caves and their inherent resources must be actively managed to ensure their protection. The increased exploration and study of caves in the last 50 years heightened awareness of their importance as natural resources for groundwater, ecosystem biodiversity, and recorders of climate change (Palmer, 2007). Despite this increased attention and understanding, protection and management of cave resources are inadequate at many levels in the United States, especially in the private sector (Huppert, 1995). Grassroots efforts by burgeoning cave-oriented organizations, such as the National Speleological Society, increase not only the public perception of caves, but also amplify the number of people actively visiting caves, which creates a critical need for cave protection and management.

 

Combined with the already prevalent development and population growth issues impacting caves in both urban and rural karst areas (North et al., 2009), the destruction of inherent cave resources is an inevitable outcome without protection and management strategies in place (DuChene, 2006). Yet, before caves can be actively managed and protected, stakeholders (including public and private landowners, land managers, and policymakers) must understand the sensitivity and disturbance of resources found within caves.

 

Within the context of caves and cave environments, the definitions of disturbance and sensitivity are related to the amount of anthropogenic impact and the amount and condition of the resources found within a cave, respectively. Complexities in defining these terms arise from their existential meanings; for example, one cave’s location may make it more vulnerable than its counterpart containing more numerous biota or speleothems (cave formations), but it is located in a less populated or visited area, thereby creating the perception it is safer and less sensitive. The reality may be that popular knowledge of this cave, despite its remote location, could lead to increased disturbance of the existing features. Disturbances in a cave environment arise from many activities, including the cave inventory itself. Therefore, an objective, holistic approach, modified from environmental indicators of the Karst Disturbance Index (KDI; van Beynen and Townsend, 2005), is used here. By adapting the KDI indicators to the cave disturbance inventory system, and including several additional parameters, we attempt to more completely consider the amount of disturbance to a cave through a myriad of potentially intrusive in-cave activities.

 

Some attempts have been made to formalize the protection of caves at a national level (FCRPA, 1988); however, this only protects caves located on federal lands. Caves located on non-federal, public or private lands are not protected by the Federal Cave Resource Protection Act, and in most cases the severe lack of regional and local protection policies leads to their demise (Fleury, 2009). To better focus the attention of state and local stakeholders on how to manage and protect caves efficiently and effectively, their disturbance, sensitivity, and resources must first be inventoried and evaluated. In certain regions, such as west-central Florida, cave conservation ethics of the general public are lacking, and existing laws, regulations, or policies that address cave protection or require sound management of cave systems only provide minimal protection. One reason for this deficiency is the informational disconnect, in terms of the knowledge of cave resources, sensitivity, and disturbance, which exists between researchers, stakeholders, and general public. However, in west-central Florida, the use of a Geographic Information System (GIS) cave inventory method enabled qualitative analysis of data collected on air-filled caves in the Withlacoochee State Forest (Harley et al., 2010).

 

We built upon the Harley et al (2010) study using a cave inventory system to simplify the dissemination of resource knowledge to stakeholders so that cave management and protection policies can be implemented at the state and local level through data synthesis. In this study, a sample of 36 caves in west-central Florida, located on both public and private land, were inventoried, and cave resource data were analyzed to provide a quantifiable measure of cave sensitivity and disturbance using a standardized scoring system. Here, we present a model to evaluate these scoring systems and discuss their abilities to aid in cave management, as well as the complexities of their use based on cave ownership status.

 

Area of study

 

            This study was conducted in west-central Florida, which includes Marion, Citrus, Sumter, and Hernando counties. West-central Florida is a karst landscape conducive for researching cave sensitivity and disturbance because the area contains hundreds of air-filled caves, which are spatially dispersed throughout the landscape, and are located on both public and private lands. Additionally, there exist few laws or policies related to cave and karst protection. The lack of best management practices for caves is a serious problem in west-central Florida, as illustrated by their continued destruction and vandalism (Fleury, 2009). Florida is one of the fastest growing states in the nation, with a population growth of 12.2 % from 2004 through 2010 (Clouser and Cothran, 2009). With no comprehensive management and protection regulations in place, developers often build over previously undiscovered caves without acknowledging or addressing these unique features. Moreover, poor management decisions are continually made by varying agencies, not due to an unwillingness to protect caves and karst, but rather a lack of knowledge necessary to spur and validate alternative management choices.

 

Methods and Results

 

The cave disturbance model includes spatial parameters, including [1] distance to road, [2] population density, and [3] distance from cite center. Considering these parameters, spatial data sets were acquired from Marion, Citrus, Sumter, and Hernando Counties, Florida and input into a GIS. The predictive model will be presented, as well as a raster data set of land area associated with the most disturbed cave locations. We then calculate the associated model error by overlaying GPS locations from a cave inventory described in Harley et al. (2010, 2011). Implications on the management of caves and karst features in west-central Florida will be considered.

 

References

 

Clouser, R.L., Cothran, H., 2009. Issues at the Rural-Urban Fringe: Florida’s Population Growth, 2004–2010. Institute of Food and Agricultural Sciences publication FE567, University of Florida, Gainesville, Florida.

 

FCRPA, 1988. Federal Cave Resource Protection Act http://www.acave.us/ccms/federal caveprotectionact.htm (viewed 18.3.07.).

 

Fleury, S., 2009. Land Use Policy and Practice on Karst Terrains: Living on Limestone. Netherlands, Springer.

 

Harley, G.L., Reeder, P.P., Polk, J.S., van Beynen, P.E., 2010. Using a GIS-Based Inventory for Cave Management Implementation at Withlacoochee State Forest, Florida. J. Cave Karst Stud. 72(1), 40–48.

 

Kambesis, P., 2007. The importance of cave exploration to scientific research. J. Cave Karst Stud. 69(1), 46–58.

 

North, L.A., van Beynen, P.E., Parise, M., 2009. Interregional comparison of karst disturbance: West-central Florida and southeast Italy. J. Environ. Manage. 90, 1770–1781.

 

Quinlan, J.F., 1970. Central Kentucky karst. Reunion Internationale Karstologie en Langudoc-Provence, 1968, Actes: Mediterrane Etudes et Travaux, 7, 235–253.

 

van Beynen, P.E., Townsend, K., 2005. A disturbance index for karst environments. Environ. Manage. 36(1), 101–116.

 

Restoration Ecology and Tree-Ring Dating at Cumberland Homesteads Tower in Crossville, Tennessee, U.S.A.

RestorationEcologyand1ABSTRACT

During the New Deal era under Franklin D. Roosevelt, the Cumberland Homesteads were established as an approach to struggling communities and as an effort to combat raging unemployment issues facilitated by the Depression. Though the project was technically a failure, the community continued to thrive, with members purchasing the land and acclimating to the area. The rich history of this area and its inhabitants is of extreme importance to the Cumberland Homesteads Tower Association, and they strive to sustain the history of this period through the museum that is now established in the Homesteads Tower. Part of conserving this history is maintaining authenticity, and it was suspected that several of the trees blocking the Tower from spectators and travelers were planted after the initial Homesteads construction date. Some trees were also causing roof damage to the landmark, as well as threatening to split and harm the structure. All of the trees surrounding the Tower were cored (fourteen total). All but one tree were dated, and approximately half were proven to have been planted after the period of significance. These trees were subsequently removed and donated to schools to further education of dendrochronology.

 

BACKGROUND

The Cumberland Homesteads project began in 1934 holding a budget of 25 million dollars with the vision of relieving the Depression conflicts that had arisen with local farmers and blue-collar workers. Since most of these workers had been left unemployed and agriculture in Tennessee was struggling immensely, these stranded workers were relocated to sites where employment would be established through an autarchic and self-sustaining community. The project was part of the New Deal era, and was a fragment of the effort to end the Depression for struggling locals. Unfortunately, the workers were not familiar with management routines and purposes, and the program flitted between an array of organizations as it endeavored to succeed in its original and fleeting goal. However, in 1945, shortly after the close of World War II, the government officially ended the project. Though it was considered a nonsuccess, the community persevered and sojourned in the area, and brought an economic boost to the region. Most of the houses have been renovated and are still used as housing for current residents. However, by the 1970s, the Homesteads Tower, which was the original Administration office for the project, was suffering from neglect. The Tower was restored by the Cumberland Homesteads Historic District, qualified as a historic landmark in 1988, and listed as a non-profit organization by 1990. It is now used as an educational tool and is a major tourist attraction in the Cumberland region, and even hosts an Apple festival at the Tower every autumn.

 

STUDY AREA

Originally, the Administration office of the Cumberland Homesteads was located in the Homesteads Tower. After the government funding for the project had ceased, a museum was later established in the Tower. The Homesteads Tower is also the headquarters of the Tower Association, and it is essential to them to retain historical authenticity. Additionally, we cored three of eight trees that were situated across the street from the tower in a small triangle of field surrounded by two intersecting highways (127S and 68). The Tennessee Department of Transportation (TDOT) has threatened to widen this intersection, thus placing the trees in a precarious circumstance. The dating of these trees hopefully will aid in the preservation of the trees in the future.

 

RESULTS

Fourteen trees total that surrounded the Homesteads Tower Museum were cored: ten hemlock (Tsuga canadensis), two oak (Quercus alba), one hickory (Carya sp.) and one sugar maple (Acer saccharum) species. After employing COFECHA, only one tree could not be dated (an oak) because of the difficulty of removing the core from the trunk (the borer became stuck many times during the process, compromising the samples). The remaining thirteen trees were dated, and it was shown that six of the trees were planted post-World War II (1946 and after), six were planted in the 1930s during the construction and establishment of the Homesteads community, and the remaining tree sample was dated to 1926, prior to the formation of the project. Additionally, we cored three of the eight oaks (Quercus stellata) that lay across the street from the Tower. The oaks were found to have been planted in the mid-1800s (1825, 1863, 1887).

 

DISCUSSION

Following the results of the project, several trees have been removed according to the age of the tree not being conducive to the period of significance. Additionally, other trees were removed, not because of age, but because of the damage they were inflicting to the roof of the Tower and the imminent threat of splitting (a couple of the trees had attained split trunks and were in danger of falling and causing damage). The Homesteads Tower is now more visible to passerby and visitors, as well as retains the legitimacy of the historical landmark. The oak trees that were stationed across the highway from the Tower are still being protected by the Tower Association, encouraged by the discovered maturity of the trees; currently, TDOT has not moved further with plans to take down or dislodge any of the oak trees.

 

Comparing Remote Sensing-based Urban Tree Canopy Assessment Methods on the Virginia Tech Campus

Poster_SEDAAG-2013_WHHwangIntroduction

 

Urban Tree Canopy (UTC, ground area covered by tree crowns when viewed from above in urbanized areas) has been assessed in order to develop urban forest management plans and policies because assessments of UTCs and impervious surfaces provide data for modeling urban forest functions and developmental impacts (Nowak, et al., 2001; Nowak and Greenfield, 2010 and 2012). As remote sensing technologies have been developed, they are widely applied to assess UTC and other ground cover classes (Bridge, 2008; Myeong et al., 2001; MacFaden et al., 2012). Unlike field-based assessments, the UTC assessment using remote sensing technology has an ability to assess UTC within private residential properties where permission to access is often required.

 

Two primary methods using remotely sensed data are image classification and photo-interpretation. Image Classification (IC) takes a census of all pixels within an area of interest by assigning a ground cover class to each pixel using computer-automated algorithm(s). Typically, IC has been considered as an advanced method for identifying ground cover classes (Nowak and Greenfield, 2012). This requires specialized knowledge, software, and processing time to analyze remotely sensed data. As well, this method is significantly influenced by the spatial resolution of images (Huang et al. 2001; Walton, 2008; Nowak and Greenfield, 2012). Photo-interpretation (PI) is a survey approach that uses randomly sampled points within an area of interest to estimate proportions of ground cover classes. The PI method is efficient and effective method, however, it has an associated margin of error that reflects the uncertainty of its ground cover estimates. Therefore, sufficient sample points are necessary for lowering standard error and creating greater confidence in the UTC estimates (Nowak et al., 1996; Nowak and Greenfield, 2012).

 

 

 

Methodology

 

We compared two remote sensing-based methods to estimate UTC and other ground covers on approximately 885 acres of the central Virginia Tech (VT) campus located in Blacksburg, VA (Figure 1). We derived Virginia Tech UTC and other ground cover estimates from the Blacksburg UTC assessment (McGee et al., 2012). The Blacksburg UTC assessment started with ISODATA classification using the four-band National Agricultural Imagery Program (NAIP) imagery: clustering the imagery into 200 spectral classes, and then visually assigning them to one of seven land classes. After the initial classification, two land classes (shadow and mixed pixels) were reclassified and manually assigned into four final classes: water, impervious surface, non-tree vegetation, and tree canopy. Once image classification was complete, accuracy assessments were performed via cross-referencing with 100 random points using the Virginia Base Mapping Program (VBMP) imagery. Manual edits with misclassified points were performed until an overall accuracy of greater than 90% was achieved.

 

We used i-Tree Canopy as the PI method with sizes of random samples from 10 to 1,000. Random sample points, which were mutually exclusive and auto-generated by tool (i-Tree Canopy) across the area of interest, were manually interpreted for each sample point plotted onto Google Map. Then, i-Tree canopy automatically calculated proportions (%) and standard errors of each ground cover class based on an examiner’s visual interpretation. Once the primary photo-interpreter implemented visual classification of sample points using five initial classes (water, impervious surface, non-tree vegetation, tree canopy, and uninterpretable points), two other photo-interpreters assigned uninterpretable points to a land cover class. When the uninterpretable points were not agreed upon, the primary photo-interpreter made a final decision and classified points into final classes, which were the same as the IC method. For statistically examining differences between IC and PI, we produced ten replicated PI assessments of UTC and other ground covers.

 

 

 

Results and Discussion

 

The Blacksburg UTC assessment has achieved an overall accuracy of 95% and an overall kappa statistics = 0.95. As a result of this assessment, the IC method classified the central campus as approximately 1% water, 33% impervious surface (bare soil was considered as impervious surface for the purpose of the Virginia UTC Project), 46% non-tree vegetation, and 16% tree canopy (Figure 1. Table 1). The PI method indicated that estimated proportions were affected by sizes of random sample points (Table 2). Specifically, sample sizes greater than 250 points produced consistent results of estimate averages: water 0.9 to 1.1%, impervious surface 38.0 to 38.8%, non-tree vegetation 45.4 to 46.3%, and tree canopy 14.4 to 14.9%.

 

 

 

Table 1 Results of Image Classification

 

WA1

 

IS1

 

VA1

 

TC1

 

Proportion (%)

1.1

 

31.1

 

45.6

 

16.1

 

Overall Accuracy (%)

96.0

 

Producer’s Accuracy (%)

100.0

 

100.0

 

90.2

 

96.9

 

User’s Accuracy (%)

100.0

 

89.8

 

97.4

 

96.9

 

Overall Kappa Statistics

0.946

 

Kappa Statistics

1.000

 

0.871

 

0.961

 

0.958

 

1 WA (water); IS (impervious surface); VA (non-tree vegetation); and TC (tree canopy)

 

 

 

Compared to the IC method, i-Tree Canopy (PI method) under-estimated tree canopy (about 2%) and over-estimated impervious surface (about 5%). Even so, the statistical comparison revealed that overall results produced by the IC and the PI methods were not significantly different; the Chi (χ)-square test with a significance level of 0.05 presented that an overall closeness of agreement between the IC and the PI methods was achieved by using a sample size of at least 250 points (Table 2). In addition, the sample size of 250 points was the same size that the PI used as a starting point to stabilize mean and CV and to produce the constant results of estimates.  However, there were persistent differences between these estimation averages: UTC cover between IC (16.1%) and PI (14.4 to 14.9%) and impervious surface between IC (33.1%) and PI (38.4 to 38.7%).

 

 

 

Table 2 Results of i-Tree Canopy Assessments

 

10 points

 

12 points

 

25 points

 

50 points

 

WA1

 

IS1

 

VA1

 

TC1

 

WA

 

IS

 

VA

 

TC

 

WA

 

IS

 

VA

 

TC

 

WA

 

IS

 

VA

 

TC

 

Mean2 (%)

0.0

 

36.0

 

45.0

 

18.0

 

0.8

 

31.7

 

50.0

 

17.1

 

0.8

 

41.2

 

45.2

 

12.8

 

0.8

 

42.2

 

43.0

 

14.0

 

SD3

0.0

 

16.5

 

19.0

 

12.3

 

2.6

 

10.2

 

14.0

 

10.3

 

1.7

 

11.3

 

8.9

 

7.5

 

1.0

 

6.1

 

5.8

 

4.2

 

CV3 (%)

0.0

 

45.7

 

42.2

 

68.3

 

0.0

 

32.4

 

29.1

 

61.3

 

210.8

 

27.5

 

19.6

 

58.6

 

129.1

 

14.4

 

13.6

 

30.2

 

100 points

 

250 points

 

500 points

 

1,000 points

 

WA

 

IS

 

VA

 

TC

 

WA

 

IS

 

VA

 

TC

 

WA

 

IS

 

VA

 

TC

 

WA

 

IS

 

VA

 

TC

 

Mean2 (%)

0.0

 

36.0

 

48.0

 

14.0

 

1.0

 

38.8

 

45.8

 

14.4

 

0.9

 

38.4

 

46.3

 

14.5

 

1.1

 

38.0

 

45.4

 

14.9

 

SD3

1.0

 

6.1

 

7.6

 

2.5

 

0.5

 

2.4

 

1.7

 

1.8

 

0.4

 

2.0

 

1.8

 

1.3

 

0.3

 

1.3

 

1.2

 

0.9

 

CV3 (%)

161.0

 

16.9

 

15.5

 

17.8

 

54.2

 

6.2

 

3.7

 

12.6

 

51.5

 

5.2

 

3.9

 

9.0

 

29.2

 

3.4

 

2.7

 

6.0

 

1 WA (water); IS (impervious surface); VA (vegetation); and TC (tree canopy)
2 Mean of 10 replicated assessments at each point sample size
3 SD (standard deviation); and CV (coefficient of variation)

 

 

 

Findings concluded that the limitations of two methods will likely reflect differences in ground cover estimates. Due to the difficulties for handling the shadow and mixed pixels attained by remotely sensed data as well as distinguishing between trees, shrubs, and grass, there were some uncertainties in the ground cover estimates derived from the IC method (Myeong et al., 2001; MacFaden et al., 2012). In addition, photo-interpreters might inaccurately interpret points laid over Google Map, which is a mosaic of aerial photography collected from different sources at various dates and seasons (e.g. leaf-on and leaf-off images), different spatial resolution (typically, finer in urban areas, while coarser in suburban and rural areas upon data availability), as well as image quality (e.g. sun angle, cloud cover) (Taylor and Lovell, 2012).

 

According to the results of the study, the PI method achieved an overall closeness of agreement with the IC method in estimated proportions of UTC and other ground covers with sufficient sample sizes (at least 250 points in case of this project). The IC method provided a wall-to-wall classification map of the study area. This classification map can be used to identify potential tree planting sites and to analyze changes (via time-series data) in patterns of extent and distribution of UTCs (King and Locke, 2013). On the other hand, PI provided these advantages over the IC method: less remote sensing data required, efficient assessment, and an ability to distinguish trees from other vegetation. Therefore, different UTC assessment methods could be selected depending on the purposes and conditions. We recommend the IC method for municipalities who need a comprehensive and intensive UTC assessment and further analytical possibilities, including targeting canopy protection, potential tree planting site identification, urban landscape pattern analyses, and urban micro-climate research (Myeong et al., 2010; King and Locke, 2013). On the other hand, the PI method is recommended for urban foresters who want rapid and efficient estimates of UTCs.

 

 

 

Reference

 

    1. Huang, C., Yang, L., Wylie, B., Homer, C., 2001. A strategy for estimating tree canopy density using Landsat 7 ETM+ and high resolution images over large areas. Third International Conference on Geospatial Information in Agriculture and Forestry, Denver, CO. URL: http://landcover.usgs.gov/pdf/canopy_density.pdf (last accessed: 8 May, 2013).

 

    1. King, K.L., Locke, D.H., 2013. A comparison of three methods for measuring local urban tree canopy cover. Arboriculture & Urban Forestry, 39 (2), 62-67.

 

    1. MacFaden, S.W., O’Neil-Dunne, J.P.M., Royar, A.R., Lu, J.W.T., Rundel, A.G., 2012. High-resolution tree canopy mapping for New York City using LiDAR and object-based image analysis. Journal of Applied Remote Sensing, 1-23.

 

    1. McGee, J.A., Day, S.D., Wynne, R.H., White, M.B., 2012. Using geospatial tools to assess the urban tree canopy: decision support for local government. Journal of Forestry, 110 (5), 275-286.

 

    1. Myeong, S., Nowak, D.J., Hopkins, P.F., Brock, R.H., 2001. Urban cover mapping using digital high-spatial resolution aerial imagery. Urban Ecosystems 5, 243-256.

 

    1. Nowak, D.J., Noble, M.H., Sisinni, S.M., Dwyer, J.F., 2001. People and Trees: assess the US urban forest resource. Journal of Forestry, 99 (3), 37-42.

 

    1. Nowak, D.J. Greenfield, E.J., 2010. Evaluating the National Land Cover Database tree canopy and impervious cover estimates across the conterminous United States: a comparison with photo-interpreted estimates. Environmental Management, 46 (3), 378-390.

 

    1. Nowak, D.J., Greenfield, E.J., 2012. Tree and impervious cover change in U.S. cities. Urban Forestry & Urban Greening, 11 (1), 21-30.

 

    1. Walton, J.T., 2008. Difficulties with estimating city-wide urban forest cover change from national, remotely-sensed tree canopy maps. Urban Ecosystem, 11, 81-90.

 

    1. Taylor, J.R., Lovell, S.T., 2012. Mapping public and private spaces of urban agriculture in Chicago through the analysis of high-resolution aerial images in Google Earth. Landscape and Urban Planning, 108: 57-70.

 

 

THE POTENTIAL EFFECTS OF SNOWMELT PERCOLATION ON POLLEN RECORDS FROM FIRN AND GLACIER ICE

Slide1THE POTENTIAL EFFECTS OF SNOWMELT PERCOLATION ON

 

POLLEN RECORDS FROM FIRN AND GLACIER ICE

 

INTRODUCTION

 

Cryopalynology, or the analysis of palynomorphs found in snow, firn and ice, is emerging as an important and active subfield within traditional palynology. Fossil pollen found in snow and ice are direct biological links to past environments and, thus, have been used in a variety of research genres throughout the discipline’s maturation process, including historical and paleoecology. However, little attention has been given to understanding the modern pollen processes of deposition, entrainment, transport and other post-depositional processes on snowfields and how they affect pollen stratigraphy within snow and ice records.  Such a theoretical framework is essential to accurate environmental reconstructions with pollen archived in ice cores.  Therefore, the main objective of our study is to address one aspect of this modern-process question by investigating if meltwater percolation can effectively transport pollen within a snowpack.

 

 

 

Our study questions are as follows:

 

 

 

1. Can meltwater percolation effectively transport pollen from the surface downward into deeper layers of the snow?

 

 

 

2. Does surface slope, and meltwater, induce horizontal movement of pollen?

 

 

 

3. Do ice lens block pollen transport or change its direction?

 

 

 

4. Does pollen flow with sub-surface meltwater or become entrained in/on the ice beneath the firn?

 

 

 

 

 

MATERIALS AND METHODS

 

Sampling Procedure

 

We chose an experimental lab design because it was simply not cost or time efficient to observe a natural melting snow cover over a period of weeks or conduct a controlled experiment on a melting glacier as we could not adequately control the experiment’s variables (radiation, melt time, etc.).   This study was conducted from February 28-March 2, 2012 in Fairbanks, Alaska, before the start of winter snow melt.  The aim of the experiment was to maximize control and isolate the variables being observed, in order to focus on the fundamental relationship between meltwater percolation and pollen transport.  Fairbanks, Alaska also provided easy access to tools, supplies, electricity, and snow.

 

To simulate a natural snowpack, nine Styrofoam coolers of the same brand and model were set outside in October of 2011 and allowed to fill naturally with snow.  The dimensions of the coolers were 60X30X30cm.  Three of the coolers were filled with 5cm of liquid water and allowed to freeze before accumulation began. These three coolers, with ice lenses at the bottom were used in the ice-tilted (IT) experiment outlined below.

 

Immediately prior to use, individual coolers were dug out of the snowpack with a shovel and scraped level along the plane of the cooler rim to ensure consistent beginning volumes.  After the coolers were planed, five Lycopodium tablets (Batch #1031 from Lund University) were then pulverized using a porcelain mortar and pestle (20,848 spores/tablet; thus, 104,242 total spores/cooler {s=±3457, V=±3.3%}) (Maher 1981).  The five pulverized capsules, now in powder form, were then spread over a 20x20cm (400cm2) study area on the surface of each cooler prior to melting. The 20x20cm study area was located in the center of each cooler’s surface using a cardboard stencil to reduce edge effects. Then, using a radiant heat lamp (anchored approximately 25cm away from and parallel to the surface), each cooler’s snow contents were melted down to roughly two-thirds (~67%) of the original volume and allowed to percolate for 30 minutes post-heat.  All meltwater was collected from each cooler through a single small aperture (roughly 1.25cm in diameter) located at the lowest possible point.  To gain access to the snow profiles after melting, a saw was used to remove the Styrofoam sidewall of each cooler.

 

Three of the coolers were melted with the bottom of the cooler perfectly level (along a 0° plane) with the heat lamp directly overhead (perpendicular) to the snow surface.  These Snow/Flat (SF) coolers were used to simulate snow that was not on an angle, as might be found on a mountain summit, a plateau, or in certain areas of a col.  The remaining six coolers (3 completely filled with snow, 3 with snow that overlays a 5cm layer of ice) were melted on an 11-12° angle.  These six coolers (Snow/Tilted (ST) and Ice/Tilted (IT)) were used to simulate snow on an angled glacier surface that would be subjected to significant downhill forces that may transport pollen (via meltwater) more efficiently.

 

For the SF coolers (Figures 1-2) the surface samples represent the top 1-2 centimeters, consisting mostly of the penitente-like prisms of snow that formed at the surface during each melt.  The remaining snow profiles were then sampled in 5cm increments within the cooler to test for vertical spore transport.  To test if the spores were staying within the confines of the column of snow directly beneath the study area, the snow border (roughly 5cm) immediately surrounding this designated area was tested for spores as well and divided into two samples, Border and 16-20cm Border.  The Border samples contain the snow surrounding the Surface-15cm column, and the 16-20cm Border samples contain only the snow surrounding the bottom-most sample (16-20cm).  The same process was repeated for the ST coolers (excluding the border samples) (Figures 3-4) with the addition of the downhill column of snow, which was also sampled in 5cm increments to test for horizontal relocation of spores downslope of the study area.

 

The IT coolers were melted until at least 95% of the ice surface under the snow was revealed.  All meltwater and spores that ran off the surface were collected during this process.  The remaining ice block, from the study area to the drain hole, was also melted, collected and tested for spore concentrations to see if any spores remained in the ice surface or became entrained in the ice.

 

All samples were transferred and cased in 1-liter leak-proof polypropylene Nalgene bottles using a small metal shovel and plastic funnel, tagged, and shipped back to the Biogeography Laboratory at The University of Southern Mississippi for processing and counting.

 

RESULTS

 

Snow/Flat Stratum

 

Overall, spore concentrations within the Snow/Flat stratum (Figure 5) ranged from 1.2 to 16.8 spores/ml, exhibiting some within-group variability.  The highest spore concentration (16.8 spores/ml) occurred in cooler SF3’s 16-20cm and Meltwater samples, respectively.  The lowest concentration (1.2 spores/ml) occurred in SF3’s 6-10cm sample.  In terms of ranges and averages, surface samples ranged from 5.3 to 16.3 spores/ml (11.2 average), mid-level samples (1-15cm) contained 1.2 to 7.1 spores/ml (3.5 average), the bottom-most samples (16-20cm) contained 13 to 16.8 spores/ml (15 average), and lastly, the meltwater samples ranged from 5.3 to 16.8 spores/ml (9.9 average).  Samples outside of the 20x20cm study area (Border and 16-20cm Border) contained relatively low spore concentrations (1.4-4.5 spores/ml; 2.9 average) comparable to the mid-level samples.  With that said, the lower 16-20cm border snow was roughly two times more concentrated than the upper border samples in each cooler.

 

Peak spore concentrations within the Snow/Flat stratum occurred at the surface only in cooler SF1.  Coolers SF2 and SF3 both peaked in the two bottom-most samples (SF2: 16-20cm and SF3: 16-20cm and meltwater).

 

Snow/Tilted Stratum

 

Overall, the ST coolers (Figure 6) exhibited very little within-group variability. Concentrations within the Snow/Tilted stratum ranged from 0-22.9 spores/ml.  Among all three coolers, the highest concentrations were found in the downhill 16-20cm layers (16.2-22.9 spores/ml) while the lowest concentrations were located in the upper reaches of the downhill column of snow (Downhill Surface-10cm; 0-0.5 spores/ml).  The second highest concentrations were found in the meltwater samples (12.4-17.6 spores/ml) and third highest at the bottom of the study area snow column (16-20cm; 6-8.5 spores/ml).  In other words, the three most concentrated samples were found in the bottom-downhill portions and meltwater of each cooler.  Surface samples contained between 2-5.3 spores/ml.

 

 

 

Ice/Tilted Stratum

 

Overall, the IT coolers (Figure 7) exhibited only slight within-group variability.  Spore concentrations ranged from 2.1-15.1 spores/ml.  Among all three coolers, peak spore concentrations occurred in the meltwater samples (10.2-15.1 spores/ml), leaving the ice blocks with the lowest concentrations (2.1-5.2 spores/ml).  The average spore concentration for the IT meltwater samples was 12.3 spores/ml, while the IT ice samples averaged 3.3 spores/ml.  In other words, the meltwater samples were almost four times as concentrated as the ice samples.

 

Overall, peak spore concentrations occurred at the surface in only one of nine coolers (SF1).  In the other eight coolers, peak concentrations were found in either the bottom-most snow layers (16-20cm) or meltwater runoff, respectively.  As for the accounted spore percentages, by far the vast majority of Lycopodium spores introduced at the surface made their way downward (and downhill for ST coolers) into the 16-20cm snow layers and meltwater samples during each melt.  Tables 2-4 contain all of the raw data used to create the spore diagrams.

 

 

 

 

 

Conclusions

 

The results of this experiment clearly indicate that meltwater is capable of transporting pollen from the surface, downward.  Lycopodium grains are similar in size (25-40mm) to many pollen types found in glacial ice and exhibited significant transport during this experiment.  Contrary to Nakazawa and Suzuki’s findings (2008), pollen may not necessarily concentrate at the melting surface but instead can be transported downward/downhill and even be lost completely within meltwater runoff.

 

Slope angle also appears to aid in the downhill or horizontal transfer of pollen and, upon severe melting, may wash away the pollen signal completely with the meltwater runoff.  Godwin’s (1949) original ideas regarding horizontal pollen relocation at the surface are corroborated by these findings and may also help explain the sharp decreases in pollen concentrations witnessed by Nakazawa and Suzuki (2008) during their study in Japan.  These phenomena collectively have the potential to contaminate seasonal pollen stratigraphies in snow and firn zones and, ultimately, in glacial ice.

 

 

 

REFERENCES

 

Godwin, Harry. 1949. Pollen analysis of glaciers in special relation to the formation of various types of glacier bands. Journal of Glaciology 1, no. 6

 

 

 

Nakazawa, Fumio and Keisuke Suzuki. 2008. The alteration in the pollen concentration peak in a melting snow cover. Bulletin of Glaciological Research 25: 1-7.

 

 

 

Climate-Glacier Interaction in the Tropical Andes: Field Observations from the Cordillera Vilcanota, Peru

ClimateGlacierInteraction_SThe Cordillera Vilcanota in the southern Peruvian Andes houses the Quelccaya Icecap, the largest in the tropics, and represents the second largest glacierized area in the region. Six peaks reach elevations 6,000 m above sea level while surrounding areas are typically 3,000 m or higher. Primary precipitation delivery occurs as snowfall during the wet season, a critical water source for hydroelectricity, irrigation, and tourism of Andean population centers, though recent research uncovered in prior field campaigns has suggested a likelihood of more frequent rain events in context with overall warming. Significant deglaciation in the region has been the focus of research initiatives over the past several decades. Fieldwork was conducted 8 July – 4 August 2013 as part of an ongoing research initiative. The field experience was led by research P.I.s, Baker Perry, Appalachian State University, Anton Seimon, Wildlife Conservation Society, and carried out in conjunction with ongoing research objectives to further characterize precipitation patterns associated with past climate forcing (ENSO) using stable oxygen isotope ratio analyses. Research was conducted at high elevation sites (e.g. > 4,000 m) of the Cordillera Vilcanota to collect recent precipitation data from several installed meteorological stations on Nevado Osjollo Anante (5551 m) and from trained local weather observers in Murmurani, Pucarumi, and Lake Sibinacocha regions. These sites are often located above the freezing level where primary surface cover types are ice, rock, and snow. Backcountry travel through Pucarumi, Comercocha, and up to Condor Pass (> 5,100 m) was necessary to service the installed meteorological stations and to assist additional atmospheric and glaciological field research activities. One unique element of the proposed venue was the opportunity to work with local Quechua volunteer citizen scientists in collecting daily precipitation data from high elevation, local sites.

 

The field experience was part of ongoing annual research expeditions that began in 2008 to obtain data sets from snow pit samples yielding water isotope profiles. One field activity involved site visits to identify and train citizen-scientist observers surrounding the additional Coropuna, Huascarán, and Illimani ice core sites. Observers were trained to measure daily precipitation and collect water samples for isotopic analysis. In addition, one objective of the expedition was to obtain additional seasonal data from snow pits. High altitude alpine travel through glacier ice and snow was required to visit the primary area of interest on Nevado Jatunriti (6,106 m). Digging of snow pits to obtain water samples required the use of a backcountry shovel and measurements of snow depth, density, and water equivalent of each a layer were obtained using a ruler, scale, and tube kit. Maintenance was performed on previously installed research-grade meteorological stations and annual data from the past year was downloaded for use in this analysis. While at previous field sites, the preceding year’s daily precipitation observations were collected from already trained observers. Travel between field sites across the Cordillera Vilcanota region was used as a convenient opportunity to verify snow cover and glacier extent measured in satellite imagery obtained from the MODIS sensor. Ground truth points were collected using a GPS and digital camera at the Puca Glacier terminus location near Lake Sibinacocha, yielding valuable information that can be used in future analysis of precipitation patterns. Overall, the field experience required extensive backcountry travel over rugged terrain including ice, rock, and snow at high elevations (e.g. > 4,000 m).

 

To better understand the current state of the climate in the region, precipitation events were manually classified and maturation hours analyzed to detect overall and diurnal patterns in temperature, precipitation, and wind variables. Derived variables such as lapse rate and snowfall are also explored at the event scale. 281 individual precipitation events, each separated by six hours were identified. Results show that while diurnal trends vary significantly with regard to lapse rates, while other variables (e.g., wind direction, speed) were less directed by diurnal variation. A majority of events (193, 68%) were daytime with most events maturing at 1500 LST. A second maximum occurs during the night around 0300 LST. 281 Precipitation events were manually categorized for the 2012-2013 precipitation season for remote sites in the Cordillera Vilcanota in southeast Peru. Various frequency distributions show that while variables such as wind speed, direction, and precipitation type dominance are relatively unchanged during daytime and nighttime hours, distributions of thermodynamic variables such as lapse rates show a more dramatic difference (i.e., a more leptokurtic distribution). Additionally, snowfall is shown to be most frequent during the late/early portions of the year, with the sonic snow depth sensor serving as a poor analogue for the accurate derivation of snowfall.

 

Results from the expedition and subsequent analysis are of importance for providing new critical data sources to relate local climatological patterns with snowpack signatures at each field site. Data sets collected from a diverse range of locations and elevations were necessary to refine previous model analyses that combine broader circulation patterns with local precipitation variability and surface patterns. Several outcomes were produced from this data collection including the connection between local meteorological characteristics with stable oxygen isotopes found in water vapor of historical snow layers derived from ice cores. Furthermore, it is expected that interannual variability of water vapor occurring as snowfall on high Andean summits is predominately controlled by external forcings of ENSO. Climatological inferences derived from low latitude ice cores will be improved with the addition of Quelccaya cores and will make further contributions to understanding variability of precipitation processes occurring in tropical high mountains. Results of the ice core analysis are still pending. There are several critical issues regarding paleoclimate reconstructions from ice cores that inform problems of contemporary climate change. High mountains are often the most sensitive indicators of climate change, so it is necessary to resolve problems within proxy data records by incorporating additional data sources from geographically diverse sites. Results disseminated through the scientific community and through educational settings are merited based on the inclusion of a new highly resolved proxy data set of climatic variables. This research is significant in its efforts to evaluate current rates of change associated with past climatic variables and to determine the sensitivity of Earth’s climate to variations in external forcings.

 

References

 

Francou, B., M. Vuille, P. Wagnon, J. Mendoza, and J. E. Sicart, 2003: Tropical climate change recorded by a glacier in the central Andes during the last decades of the twentieth century: Chacaltaya, Bolivia, 16 degrees S. Journal of Geophysical Research-Atmospheres, 108.

 

Garreaud, R. D., M. Vuille, R. Compagnucci, and J. Marengo, 2009: Present-day South American climate. Palaeogeography Palaeoclimatology Palaeoecology, 281, 180-195.

 

Perry, L. B., Seimon, A., and Kelly, G., 2013: Precipitation delivery in the tropical high Andes of southern Peru: new findings and paleoclimatic implications. International Journal of Climatology, doi: 10.1002/joc.3679

 

Seimon, A., 2003: Improving climatic signal representation in tropical ice cores: A case study from the Quelccaya Ice Cap, Peru. Geophysical Research Letters, 30.

 

Thompson, L. G., E. Mosleythompson, J. F. Bolzan, and B. R. Koci, 1985: A 1500-YEAR RECORD OF TROPICAL PRECIPITATION IN ICE CORES FROM THE QUELCCAYA ICE CAP, PERU. Science, 229, 971-973.

 

Thompson, L. G., E. Mosley-Thompson, M. E. Davis, V. S. Zagorodnov, I. M. Howat, V. N. Mikhatenko, and P. N. Lin, 2013: Annually Resolved Ice Core Records of Tropical Climate Variability over the Past similar to 1800 Years. Science, 340, 945-950.

 

Vuille, M., B. Francou, P. Wagnon, I. Juen, G. Kaser, B. G. Mark, and R. S. Bradley, 2008: Climate change and tropical Andean glaciers: Past, present and future. Earth-Science Reviews, 89, 79-96.