How to translate text using browser tools
1 September 2013 Predicting morbidity and mortality for a saguaro cactus (Carnegiea gigantea) population
Lance S. Evans
Author Affiliations +
Abstract

Scale and bark injuries (a condition called epidermal browning) have been shown to occur on stem surfaces of 21 species of tall, long-lived cactus species throughout the Americas. The amounts of surface injuries are highly correlated with amounts of sunlight exposures. Epidermal browning causes cactus morbidity and mortality of saguaro cacti. For this study, rates of morbidity and mortality of saguaro cacti (Carnegiea gigantea (Engelm.) Britt and Rose) were determined for 1061 plants from 50 desert field plots. From 2002 to 2010, 16.5% of all saguaros died (increased mortality). During the same period, the percentage of cacti with few epidermal injuries decreased from 25 to 9%, with a 16% increase in initial morbidity. Over this same period, about 10% of the population had a rapid increase of epidermal browning (rapid morbidity). Thus, marked increases in morbidity and mortality occurred over the eight-year period. WEKA, a standard machine-learning program, was used to determine the characteristics of saguaros that predispose them to (1) die, (2) initiate morbidity (start of epidermal browning symptoms) and (3) initiate rapid morbidity (a rapid increase in epidermal browning). Decision trees were created using the C4.5 algorithm, a top-down method that repeatedly chooses the question that produces a maximal information gain. Amounts of epidermal browning on north-facing troughs alone predicted cactus mortality with 89.1% accuracy. Initial morbidity was predicted with 80.2% accuracy with data of south-facing troughs and crests only. Rapid morbidity (less than 20% epidermal browning in 2002 and more than 50% by 2010) was predicted with a very large tree with 79.6% accuracy. For example, rapid morbidity is predicted by more than 3.1% epidermal browning on left troughs on south-facing surfaces in 2002 with 80% accuracy. Moreover, rapid morbidity was also predicted if east-facing crests had more than 3.9% epidermal browning in 2002. Overall, decision tree analysis was very effective at predicting changes in mortality and morbidity for the population of 1061 saguaros from 2002 to 2010. Data from trough surfaces were more effective than data from crests for predicting mortality and morbidity.

Torrey Botanical Club
Lance S. Evans "Predicting morbidity and mortality for a saguaro cactus (Carnegiea gigantea) population," The Journal of the Torrey Botanical Society 140(3), 247-255, (1 September 2013). https://doi.org/10.3159/TORREY-D-12-00079.1
Received: 30 November 2012; Published: 1 September 2013
KEYWORDS
Carnegiea gigantea
Data mining
decision trees
mortality and morbidity rates
saguaros
scale and bark injuries
RIGHTS & PERMISSIONS
Get copyright permission
Back to Top